Competition in the European banking sector

COMPETITION IN THE EUROPEAN BANKING SECTOR

by

Ioannis Samantas

A thesis submitted in partial fulfilment of the requirements for the degree

of Doctor of Philosophy (DPhil)

National and Kapodistrian University of Athens

Department of Economics

2013

I Ioannis Samantas declare that the work presented in this PhD thesis is my own.

ii

Acknowledgments

This thesis is dedicated to Grigoris and Mary, my parents, Kalliopi, my grandmother

and Thanassis, my brother. Had there not been their patience as well as their

emotional and financial support, that thesis would not reach its completion.

I wish to thank my principal supervisor, Prof. Panayotis Alexakis for his

encouragement, guidance and support during my thesis-writing phase. His availability

upon my request for any issues popped up, and sincere interest in my progress are

deeply appreciated. I also owe sincere thanks to the other two members of my thesis

scientific committee, Prof. Yannis Bilias and Prof. Manolis Xanthakis for their

advice.

I also wish to express my gratitude to Bank of Greece and, especially, to Prof. Ioannis

Papadakis, Deputy Governor of Bank of Greece, and Dr. Heather Gibson, Director-

Advisor in the Economic Research department of Bank of Greece, for granting me

with access to bank-specific data of the Bankscope database. Without their assistance,

I could not possibly move on to the econometric analysis of my thesis.

I onw special thanks to the participants of the XXI European Workshop on Efficiency

and Productivity Analysis (EWEPA), the 18th Annual Conference of Multinational

Finance Society, the conference on Advances in Financial and Insurance Risk

Management, the International Conference on Applied Business and Economics and

the 10th Annual Conference of the Hellenic Finance and Accounting Association for

their valuable comments.

Last but not least, I am also indebted to my close friends in Greece and abroad for

being patient and supportive with me, and for making my life emotionally ample and

consistent.

iii

Table of contents

CHAPTER 1 PROLEGOMENA ........................................................................................... 1

1.1. General standpoint .......................................................................................................... 3

1.2. Research questions........................................................................................................... 5

1.3. Contribution of the thesis ................................................................................................ 8

1.4. Structure of the thesis .................................................................................................... 12

CHAPTER 2 COMPETITIVE ISSUES .............................................................................. 17

2.1. Introduction .................................................................................................................... 19

2.2 Institutional perspective .................................................................................................. 20

2.2.1 Efficiency ................................................................................................................... 20

2.2.2 Consolidation .............................................................................................................. 22

2.2.3 Relationship banking .................................................................................................. 24

2.3 Macro finance perspective .............................................................................................. 26

2.3.1 Regulatory developments ........................................................................................... 26

2.3.1.1 International initiatives ....................................................................................... 28

2.3.1.2 Towards a single EU financial market ................................................................ 36

2.3.1.3. Basel III – framework in progress ...................................................................... 42

2.3.1.4. Free banking ....................................................................................................... 46

2.3.2. Integration ................................................................................................................. 47

2.3.3. Financial stability ...................................................................................................... 49

1.4. Conclusion ....................................................................................................................... 52

CHAPTER 3 EMPIRICAL MODELS OF COMPETITION IN THE EUROPEAN

BANKING .............................................................................................................................. 53

3.1. Introduction .................................................................................................................... 55

3.2. Non-formal structural methods ..................................................................................... 55

3.2.1. The Structure – Conduct – Performance (SCP) paradigm - The relative efficiency

(RE) hypothesis ................................................................................................................... 55

3.2.2. Formal structural approaches .................................................................................... 60

3.2.2.1. The HHI in a S – P model ................................................................................... 62

3.2.2.2. The CRk in a S - P model ................................................................................... 65

iv

3.2.3. Concentration ratios................................................................................................... 66

3.2.3.1. The k bank concentration ratio ........................................................................... 68

3.2.3.2. Herfindah - Hirschman Index (HHI) .................................................................. 69

3.2.3.3. The Hall - Tideman Index (HTI) and Rosenbluth Index (RI) .............................. 71

3.2.3.4. The comprehensive industrial concentration measure ....................................... 72

3.2.3.5 The Hannah and Kay index ................................................................................. 73

3.2.3.6. The U index......................................................................................................... 74

3.2.3.7. The Hause indices ............................................................................................... 75

3.2.3.8 Entropy measure .................................................................................................. 77

3.3. Non-structural methods ................................................................................................. 78

3.3.1. The New Empirical Industrial Organisation (NEIO) approach ................................. 78

3.4. Alternative models .......................................................................................................... 83

3.5. Review of empirical analysis .......................................................................................... 91

3.5.1. Panzar – Rosse methodology ..................................................................................... 91

3.5.1.1. Cross-country comparisons ................................................................................ 91

3.5.1.2. Country-specific studies ..................................................................................... 94

3.5.2. Bresnahan-Lau methodology ..................................................................................... 97

3.5.2.1. Cross-country applications................................................................................. 97

3.5.2.2. Country-specific studies ..................................................................................... 97

3.5.3. Lerner index .............................................................................................................. 99

3.6. Conclusion ..................................................................................................................... 100

Chapter 3 Appendix ............................................................................................................ 101

CHAPTER 4 A REVIEW OF EFFICIENCY ANALYSIS ............................................. 111

4.1. Introduction .................................................................................................................. 113

4.2. Definition of bank ......................................................................................................... 113

4.3. Measurement of X – efficiency .................................................................................... 114

4.3.1. Parametric approaches ............................................................................................. 115

4.3.1.1. The Stochastic Frontier Analysis ...................................................................... 115

4.3.1.2. Distribution-free approach ............................................................................... 116

4.3.1.3. Thick frontier approach .................................................................................... 117

4.3.2. Non-parametric approaches ..................................................................................... 120

4.3.2.1. Data Envelopment Analysis .............................................................................. 120

4.3.2.2. The free disposal Hull approach ...................................................................... 122

4.4. Review of empirical applications ................................................................................ 125

4.4.1. International comparisons ....................................................................................... 125

4.4.2. European cross-country comparisons ...................................................................... 129

v

4.4.3. Developed economies .............................................................................................. 133

4.4.4. Transition economies............................................................................................... 137

4.4.5. Country – specific studies ....................................................................................... 141

4.5. Conclusion ..................................................................................................................... 148


CHAPTER 5 THE NEXUS BETWEEN BANK COMPETITION AND FINANCIAL

STABILITY ......................................................................................................................... 155

5.1. Introduction .................................................................................................................. 157

5.2. Theory............................................................................................................................ 157

5.3. Empirical review ........................................................................................................... 162

5.4. Conclusion ..................................................................................................................... 167

Chapter appendix..................................................................................................................157

CHAPTER 6 INCOME-SPECIFIC ESTIMATES OF COMPETITION IN EUROPEAN

BANKING ............................................................................................................................ 177

6.1. Introduction .................................................................................................................. 179

6.2. Methodology .................................................................................................................. 180

6.3. Competition determinants ........................................................................................... 183

6.4. Data ................................................................................................................................ 185

6.5. Results............................................................................................................................ 189

6.6. Conclusion ..................................................................................................................... 213

CHAPTER 7 COST AND PROFIT EFFICIENCY IN EUROPEAN BANKING:

COMPARISON OF PARAMETRIC METHODOLOGIES AND CONVERGENCE

DYNAMICS ......................................................................................................................... 217

7.1. Introduction .................................................................................................................. 219

7.2. Methodological issues ................................................................................................... 222

7.3. The model ...................................................................................................................... 225

7.4. Sample - Evidence ........................................................................................................ 228

7.5. Scale economies ............................................................................................................. 240

vi

7.6. Convergence .................................................................................................................. 245

7.7. Other stylized facts ....................................................................................................... 250

7.8. Conclusion ..................................................................................................................... 254

Chapter 7 appendix ............................................................................................................. 256

CHAPTER 8 BANK COMPETITION AND FINANCIAL (IN)STABILITY IN

EUROPE: A SENSITIVITY ANALYSIS ......................................................................... 263

8.1. Introduction .................................................................................................................. 265

8.2. Methodology .................................................................................................................. 266

8.3. The model ...................................................................................................................... 269

8.4. Determinants of stability .............................................................................................. 272

8.4.1. Bank-specific variables ........................................................................................... 272

8.4.2. Macroeconomic variables ........................................................................................ 273

8.4.3. Regulatory environment .......................................................................................... 273

8.5. Data ................................................................................................................................ 275

8.6. Main results ................................................................................................................... 281

8.7. Does the effect of market structure variables alter with the interplay between

regulation and ownership? ................................................................................................. 292

8.8. Are there other governance indicators that enter non-linearly? .............................. 295

8.9. Are the results robust to alternative measures of risk? ............................................ 298

8.10. How market structure variables behave per specialization group? ...................... 303

8.11. Other robustness checks ............................................................................................ 305

8.12. Conclusion ................................................................................................................... 308


CHAPTER 9 SYNOPSIS .................................................................................................... 317

9.1. Major findings of the research .................................................................................... 319

9.2. Policy implications of the research ............................................................................. 325

9.3. Limitations of research ................................................................................................ 329

9.4. Suggestions for further research ................................................................................. 331

REFERENCES .................................................................................................................... 333

THESIS APPENDIX ........................................................................................................... 390

vii

List of tables

TABLE 1: NUMBER OF BANKS .............................................................................................................. 186

TABLE 2: SUMMARY STATISTICS OF VARIABLES IN MODEL 6.1 .............................................................. 187

TABLE 3: SUMMARY STATISTICS OF VARIABLES IN MODELS 6.4 AND 6.5 ............................................. 188

TABLE 4: MARGINAL COSTS AND LERNER INDEXES ................................................................................. 190

TABLE 5: REGRESSION OUTPUT (WHOLE SAMPLE) ................................................................................... 191

TABLE 6: REGRESSION OUTPUT PER BANK TYPE ....................................................................................... 196

TABLE 7: SUR FOR COMMERCIAL BANKS ................................................................................................... 200

TABLE 8: SUR FOR COOPERATIVE BANKS .................................................................................................. 204

TABLE 9: SUR FOR SAVINGS BANKS ............................................................................................................. 207

TABLE 10: SUR FOR 'OTHER' BANKS ........................................................................................................... 210

TABLE 11: NUMBER OF BANKS ...................................................................................................................... 229

TABLE 12: COST EFFICIENCY (INTERMEDIATION APPROACH) ............................................................... 231

TABLE 13: COST EFFICIENCY (VALUE-ADDED APPROACH) ..................................................................... 233

TABLE 14: PROFIT EFFICIENCY (INTERMEDIATION APPROACH) ........................................................... 235

TABLE 15: PROFIT EFFICIENCY (VALUE-ADDED APPROACH) ................................................................. 237

TABLE 16: COST/PROFIT EFFICIENCY BY PRODUCTIVE SPECIALISATION AND ASSET CLASS ............ 239

TABLE 17: SCALE ECONOMIES BY COUNTRY .............................................................................................. 241

TABLE 18: SCALE ECONOMIES BY BANK TYPE AND ASSET CLASS .......................................................... 244

TABLE 19: Β-CONVERGENCE AND Σ-CONVERGENCE OF COST EFFICIENCY .......................................... 248

TABLE 20: Β-CONVERGENCE AND Σ-CONVERGENCE OF PROFIT EFFICIENCY ...................................... 249

TABLE 21: DISTRIBUTION OF COST/PROFIT EFFICIENCY ......................................................................... 253

TABLE 22: DESCRITIVE STATISTICS OF KEY VARIABLES ......................................................................... 277

TABLE 23: CORRELATION BETWEEN COUNTRY-LEVEL VARIABLES ...................................................... 280

TABLE 24: REGRESSION OUTPUT OF MODEL 8.1 ........................................................................................ 283

TABLE 25: EXTREME BOUNDS OF MODEL 8.1 ............................................................................................. 286

TABLE 26: REGRESSION OUTPUT OF MODEL 8.2 ........................................................................................ 288

TABLE 27: EXTREME BOUNDS OF MODEL 8.2 ............................................................................................. 291

TABLE 28: SENSITIVITY ANALYSIS OF MODEL 8.7 ..................................................................................... 294



TABLE 31: SENSITIVITY ANALYSIS WITH ALTERNATIVE MEASURES OF STABILITY ............................ 302

TABLE 32: MODEL 8.1 OUTPUT PER BANK TYPE ........................................................................................ 304

TABLE 33: FURTHER EBA OF MODEL 8.1 ................................................................................................... 306

TABLE 34: FURTHER EBA OF MODEL 8.2 ................................................................................................... 308

viii

Abstract

The aim of the thesis is to investigate the structure of banking sectors in the European

Union during the period 2002-2010. It encompasses the theoretical premises of bank

competition and how it is interrelated to contemporary issues in European banking.

For this purpose, the thesis comprises two parts; the former refers to the literature

review of theoretical and empirical analysis and, the latter, to the econometric

analysis.

The first aspect of this thesis is the measurement of the degree of competition in the

developed sub-group of European Union since the advent of Euro. We, therefore,

estimate the marginal cost out of a translog cost function and then investigate

potential correlates of income-specific estimates of market power for commercial,

cooperative and savings banks as well as other financial institutions.

We next compare parametric methodologies widely employed in the literature in

order to estimate cost and profit efficiency scores of financial institutions. In so doing,

we juxtapose the results from different definitions of bank outputs/inputs, namely the

intermediation and value-added approach. Other section reports time-varying

estimates of scale economies per banking market, bank type and asset class, while the

analysis concludes with the investigation of beta and sigma convergence of cost/profit

efficiency along with equality and other distribution issues.

The final part of the thesis deals with the competition-stability nexus; that is, whether

and to what extent bank competition and concentration drive to systemic stability or

fragility. We conduct sensitivity analysis of the underlying relationship subject to

different informational sets so as to construct the extreme bounds of their beta

coefficients and draw remarks on their statistical and sign persistence. The analysis

allows for non-linearities between competition and controls of regulation, ownership,

various governance indicators for robustness-checking reasons.

CHAPTER 1

Prolegomena

3

1.1. General standpoint

The operation of banks in the wake of a single currency system has had a considerable

impetus to the economic growth and financial stability of the European banking

sector. Social welfare is theoretically accomplished by the competitive conduct and

high economic efficiency of profit-maximising financial institutions. All firms and

consumers take advantage of the broad supply of banking services at low prices

within a context of fast-developing financial markets that eventually foster economic

development.

From a policy perspective, there exists a number of banking issues that have

embarked the academic debate and a series of international initiatives in order to

„unfreeze‟ the potential of financial system. To this end, structural reforms have taken

place since the „80s culminating in the advent of Euro and until the onset of current

financial crisis. In particular, banks increased the range of their products in response

to free trade of foreign-owned banks in domestic markets bringing about the

classification of the financial institutions with respect to their productive

specialisation. Hence, high competitive pressures have made banks engage in lower

costs, efficient risk management, new corporate governance structures and alternative

non-traditional income activities.

That dynamic tendency of integration comes under the Financial Service Action Plan,

which seeks to alleviate centrifugal forces that undermine the harmonisation in retail

and wholesale markets; European Commission (2002) identifies heterogeneous

consumer protection schemes, regulation, costly asset allocation and payment in

cross–border banking, among others. The premises of Basel Accord constitute a new

regime that imposes requirements on capital reserves with asymmetric implications on

competition and efficiency. In fact, capital requirements enable banks in the form of

better risk management to price competitively and thereby eschewing from cross-

subsidising practices followed in an unsystematic and unsophisticated way. In

contrast, it might be the case of banks aiming not at market shares but implementing

high prices to preserve their profit margins. Although pillar II sets out some degree of

supervisory discretion, which dismantles a level-playing field for all banking markets,

4

the estimation of capital requirements pertains to a portion of non-banking firms

through standard as well as sophisticated methods of quantifying risk.

The motivation towards greater consolidation is traced to the importance of the

aforementioned techniques. Standard methodology is preferable to smaller sized

banks, as it is less cumbersome and expensive than more advanced techniques, albeit

the estimated capital requirements in the former case constitute a greater burden than

that of the latter. Thus, mergers and acquisitions are motivated by the superior

structure of large banks to afford more expensive but effectively economising capital

reserves.

The ongoing financial crisis since 2008 has rendered the mandate of further reforms,

since in perfect hindsight we are able to cast doubt on traditional tenets in finance. For

example, the risk-free asset allocations on Treasury bills and other bonds turn out to

withhold a considerable risk once the exacerbation of country spreads (due to

speculation in fixed-income markets), institutional shortcomings and investment

shortage have put additional risks on banks‟ solvency. By the time politics and

banking go one step further to restructure the Euro zone, the institutional reform

seems to be advancing previous policies that have already been proved inadequate to

armour financial institutions in times of bank crises. Accordingly, the most

contemporary act of Basel Accord III expected to take place in 2013 builds upon the

premises of previous initiatives to strengthen banks by requiring more fraction of

common equity, Tier 1 capital and additional buffers in periods of high credit growth.

In a nutshell, competition constitutes a multidimensional concept of bank behaviour

and operation in a context of efficiently regulated financial system. If regulatory

initiatives do not effectively account of the importance of shadow banking system

along with financial innovation (derivatives) and off-balance sheet practices, negative

externalities are bound to expose the rickety financial apparatus vis-à-vis potential

shocks.

5

1.2. Research questions

The general purpose of this thesis is to shed some light on banking issues related to

competition. To this end, we assess the level of competition, efficiency and financial

stability and their interconnections amongst them as well as with contemporary

concepts of regulation and supervision, bank-level specificities and other

macroeconomic conditions. We classify the research questions into three groups

according to the three chapters (6, 7, 8) of the empirical part, which are theoretically

and empirically reviewed in chapters 3, 4 and 5. In particular, we provide answers to

the following research questions.

1. What is the level of competition in Europe? Are there any determinants of

competition with respect to specific sources of banking income? What does

the analysis tell us about the pricing conduct of differently specialized

financial institutions?

2. What is the level of cost and profit efficiency in European banking according

to different parametric methodologies and definitions of bank outputs/inputs?

Are relative rankings of banking markets, bank types and asset classes stable

across all specifications? What pattern has the evolution of scale economies

followed since 2003? Do we experience beta and sigma convergence of cost

and profit efficiency even after allowing for dynamic effects?

3. Does market structure variables drive to financial stability and to what extent?

Which are the bounds that their respective coefficients lie within? Is this

tendency persistent and robustly significant? Are there non-linearities among

competition, regulation, ownership and other governance indicators that

determine bank soundness? Are alternative measures of stability explained by

similar model specifications? Which banks are affected the most by market

structure?

6

Research question 1

This question highlights a new endeavour of investigating competitive conditions in

the European banking industry. Since the vast literature of competition modeling has

produced mixed results, the proposed methodology goes one step further De Guevara

et al. (2005), Bolt and Humphrey (2010) and Carbo et al. (2009) in order to produce

income-specific competition indicators providing the intensity of their key effects on

bank competition breaking further down the subsets of bank activities.

We therefore employ the translog cost function of Ariss (2010b) that omits the price

of borrowed funds as input price, since it may capture some degree of monopoly

power of incumbent banks in the deposit markets. Next, we construct Lerner indexes

with respect to total income as well as income on loans, other interest income, fees

and commissions and other non-interest income. As such indexes reflect unobserved

bank pricing on relative products, we employ Seemingly Unrelated Regression (SUR)

framework to run regressions concurrently and account thereby of the within error

correlation across the models.

The sample comprises nine of the most developed banking markets in the European

region during the period 2002-2010. In so doing, we draw remarks over the

explanatory power of traditional collusion, relative market power, efficiency and other

key controls on bank pricing conduct.

Research question 2

We aim to shed light on the comparison of different parametric methodologies in

measuring cost and profit efficiency in European banking. We therefore employ

REM, DEA their truncated versions at 1%, 5% and 10% of each distribution tail,

along with the TFA producing averaged efficiency scores at the bank level and its

time-varying specification. The analysis follows alternative definitions of bank

inputs/outputs, which seem to be widely applied in the literature, namely under the

intermediation and value-added approach. We report time-varying country averages to

conclude on relative rankings and their persistence across different specifications. We

next estimate economies of scale to see whether and to what extent they are evolving

through time per banking market, bank type and asset size class.

7

We also employ OLS and ADL specifications to verify the significance of „catching-

up effect‟ and dispersion of cost and profit efficiency according to IA and VA

approaches. We conclude on how the efficiency distribution shifted in terms of

skewness and kurtosis from 2003 to 2008, apart from the mean and standard deviation

of EU-15 and EU-12 European regions. We also observe whether the evolution of

cost and profit efficiency scores make banks more equal in the distribution of the

employed European sample.

Research question 3

We attempt to address the question of causality of bank soundness from market

structure variables in a sample of banks headquartering within the enlarged European

Union for the period 2002-2008. We utilize concentration and competition in a

sensitivity analysis that determine bank risk subject to different information sets of

regulatory, macroeconomic and bank specific factors. Due to our heterogeneous

sample, we employ Random Effects (RE) methodology to the baseline models that

allow for (non-) linearities between regulatory/supervisory factors and bank

competition. Thus, we implement extreme bound analysis of the values that the

Lerner and Herfindahl indexes can take across all the possible combinations of key

deterministic factors. Furthermore, we implement ad hoc regressions employing the

quadratic term of competition to verify whether it is the case of U-shaped relationship.

We also extend the analysis allowing for non-linear relationships between regulatory

and foreign ownership variables to see if market structure variables lie in the

estimated bounds. As a next step, interactions of competition and governance

indicators retrieved from the World Bank come into play for more robust results. And

last, we decompose the explanatory power of independent variables on alternative

measures of financial stability and systemic risk.

Last section of chapter 8 replicates the analysis for commercial, cooperative, savings

and „other‟ banks of the sample and provides support to the model fit that is

representatively attributed to certain bank specialisation groups.

8

1.3. Contribution of the thesis

The thesis pertains to the field of studying the level of bank-level competition and its

potential correlates in European banking. It investigates sample of 9 out of the most

developed banking markets since the advent of Euro whereas other papers deal with

the previous decade since the first years of Euro zone. Hence, it might be the case of

structural break around the year 2002 and even if it is econometrically feasible to

account of such institutional events, it does not make sense when comparing periods

before and after 2002, and from a policy perspective to verify how competition

evolves within the Euro system in conjunction with other competitive issues.

The per se fact that there is no extensive literature for the period 2002-2010 over a

heterogeneous European sample is attributed to the fact that Bankscope database has

no available data for European countries especially after 2005. Our attempt is to adapt

the methodology of cost and profit function to include the most available variables not

at the expense of model fit. We rely on the work of Ariss (2010b), who omits the

price of borrowed funds as input price in a version of the cost function as it could

capture some degree of monopoly power of large banks in the deposit markets. That

enables us to come up with competition-adjusted measures of marginal cost that is

subsequently employed to measure competition at the bank level (Lerner index).

Paper 1 updates the analysis of De Guevara et al. (2005) adding some other potential

correlates of competition, like credit risk, income diversification, equity capital,

liquidity, efficacy of regulatory laws and cyclicality with respect to price and

population effects. What seems totally new to the empirical literature is the estimation

of income-specific estimates of market power although it has no merit of cross-

country comparison. Our aim is to investigate how the aforementioned effects

determine simultaneously the pricing behaviour of certain banking products and to

what extent. In so doing, we replicate the analysis for every productive specialisation

group across different specifications allowing for sample asymmetries and

unobserved bank specificities. Thus, it is the first time that we can explain, at least

partially, differences on relative market power at the income level although the

mechanisms lying underneath are not a priori known.

9

Paper 2 contributes more on the literature that studies cross-country comparison of

efficiency in the European region, with a particular focus on various parametric

approaches. Non-parametric methodologies have produced quite divergent results and

given that data shortage brings about bias in the analysis, we opt to report even more

detailed evidence with truncated versions of random effects methodology (REM),

distribution free approach (DFA) and thich frontier approach (TFA). The problematic

application of SFA, failing to accomplish curvature conditions in the maximum

likelihood technique, advocates to our employed methoology.

We also extend the literature that applies both profit and cost functions and compares

the results across various regions or countries. The underlying juxtaposition of

efficiency scores is facilitated by the employment of different definitions of bank

outputs, which is addressed scarcely in a handful of studies. Scale economies come

out of the employed cost models and draw remarks on their evolution since 2002; in

contrast, almost all relevant studies seem outdated published before 2003 and

generally in the „90s. Last, convergence analysis for bank efficiency is a new area of

research and our application aims at contributing to this accumulating knowledge.

Paper 3 contributes to the literature by extending the limited number of studies

investigating the competition-stability nexus over a European sample rather than on

specific countries or sub-regions. We also provide support of the U-shaped

relationship between competition and bank soundness, as proposed by Martinez-

Miera and Repullo (2010), on top of the applications of Berger et al. (2009) and Beck

et al. (2012).

By means of extreme bound analysis, which has not been applied in the banking

literature so far, we observe how the effects of competition and concentration take

shape along with different informational sets and specifications. The latter consists in

regulation and supervision variables, governance indicators and ideas about non-linear

effects of competition and bank ownership, among others. We, thus, produce various

robustness checks coming up with remarks on collateral banking issues.

The findings of the above research are presented on the synopsis chapter and mainly

refer to the significant:

10

1) negative relationship between asset size and market power in cooperative,

savings and „other‟ banks. In more detail, higher size drives to low prices on

loans (cooperative, „other‟ banks), other interest products (commercial,

cooperative banks) and other non-interest bearing activities (commercial,

cooperative, „other‟ banks); however, higher scale operation comes in line with

higher (lower) fees in cooperative (savings and „other‟) banks. Non-linearities

imply positive bearing of size on market power switching to negative in savings

and „other‟ banks (of higher levels of total assets). We also see this inverse U-

shaped (positive-negative) pattern in loans (commercial, cooperative, „other‟

banks), fees and commissions (cooperative, savings, „other‟ banks) and other

interest income (savings banks) whereas a U-shaped (negative-positive) path is

traced in loans of savings banks, other interest-bearing activities in „other‟ banks

and other non-interest products in commercial, cooperative and „other‟ banks.

2) SCP doctrine with the persistent effect of market concentration on banks‟

monopolistic pricing (especially in the case of savings banks). In particular,

commercial (savings) banks seem to enjoy higher fees (other interest income)

whereas cooperative banks exploit collusion in the deposits (loans) market

through lower (higher) loan rates. „Other‟ banks respond to the monopolistic

structure of deposit markets through competitive loan rates and high prices on

other interest activities.

3) relative market power in „other‟ banks and cooperative banks of relatively large

market share. Higher market share is associated with lower fees (in commercial

and cooperative banks), other interest charges (in cooperative and savings banks)

and other non-interest rates (savings banks) while „other‟ banks tend to compete

in loans and other non-interest products and act monopolistically for the

remainder. Higher shares in the deposit markets induce banks to compete each

other for loan rates and „other‟ banks to charge higher fees. In contrast, higher

loan prices (of commercial and cooperative banks) and lower fees (for

commercial and cooperative banks) are evident once share is considerable in

loans markets.

4) relative efficiency hypothesis, according to which lower costs enable banks to

impose low prices on their products systematically across all specifications.

11

5) differences in cost and profit efficiency across various parametric methodologies

and definitions of bank outputs. However, they converge in levels if we assess

their truncated levels (TFA though is much lower due to negative efficiency

scores) and seem to produce almost identical country rankings.

6) discrepancies in yearly scale economies with the intermediation approach

demonstrating diminishing pattern from diseconomies of scale whereas the

production approach comes from below unity towards their full exploitation.

7) catching-up effect and converging dispersion of cost and profit efficiency insofar

as dynamic effects are employed. Above unity coefficient (in cost efficiency

under VA) has no economic sense due to the oscillation of negative values to

positive and back, albeit the lagged dependent variable operates in a

counterbalancing manner.

8) inequality amongst the cost efficiency scores within their frequency distribution,

while IA (VA) approach shows increasing (decreasing) path until 2006 and

exacerbates (ameliorates) thereafter.

9) fragile relationship between market power and systemic stability even after

accounting for interaction terms with country-specific factors of regulation,

supervision and macroeconomic conditions.

10) U-shaped relationship between competition and stability since the quadratic term

appears significant in the extreme bounds of the Lerner and Herfindahl-

Hirschman index. The devastating turn of market power effect coming at the

expense of stability lie at the point of 28.5%.

11) positive - albeit of marginal fragility in persistence - impact of concentration on

bank soundness and even higher slopes in markets of monopolistic conduct.

12) positive association of stability with capital regulation and market discipline as

well as negative one with low activity restrictions, supervisory intervention and

foreign ownership especially in cases of well-developed governance.

13) risk-taking tendency in markets with high degree of foreign ownership if coupled

with official intervention, considerable capital regulation, granted market entry

applications and information requirements.

12

1.4. Structure of the thesis

To fulfil the aim of providing answers to the aforementioned research questions, the

structure of my thesis takes the following form:

Chapter 2 refers to contemporary banking issues that are closely related to bank

competition. In fact, by „competition‟ we do not imply several models that have been

proposed in the literature for the measurement of its level, but particularly delineate

policy-relevant concepts in an institutional and macro finance perspective. The former

comprises efficiency (section 2.2.1), - some of its standard constituents, namely X-

efficiency, allocative efficiency, economies of scale and scope, synergies and

technological change -consolidation (section 2.2.2) - expressed by means of mergers

and acquisitions, the motivation behind such practices and cross-border activity – and

relationship banking (section 2.2.3) along with its definition, motivation, benefits and

primary costs.

On the other hand, the latter points out the importance of macroprudential regulation

(section 2.3.1) and the evolution of international initiatives (Basel I, II, III) since the

„80s. Integration (section 2.3.2) follows, which is literally a broader concept that

consolidation since it is also expressed by the organic growth of branches and

subsidiaries and the cross-border provision of financial services. Last, we investigate

the theoretical underpinnings of financial stability (section 2.3.3) and the implications

of potential empirical evidence on bank competitiveness. Section 2.4 summarizes the

analysis on competitive issues.

Chapter 3 deals with the empirical models of competition proposed in the literature

and the applications based upon their premises. It covers the structural approach that

engulfs the whole range of concentration ratios and their empirical importance on

bank performance. That link is firmly grounded on the Relative efficiency (RE) and

Structure-Conduct-Performance (SCP) paradigms (section 3.2) and the empirical

models implemented in the literature to verify their significance.

The New Empirical Industrial Organisation (NEIO) approach (section 3.3) is then

analysed as the alternative empirical literature, which defines market competition in

terms of the pricing conduct of banks. We therefore adduce the non-structural

13

methods of Iwata (1974), Bresnahan (1982), Lau (1982) and Panzar and Rosse

(1987), coupled with the notion of contestability, the practice of incumbent banks to

impose competitive prices in fear of policies against abnormal profits through „hit-

and-run‟ market entries.

We last report some alternative models (section 3.4) that propose a non-systematic

methodology of measuring competition intensity. In particular, some examples are the

generalized linear pricing model of Hefernan (2002), the so-called Boone indicator of

Leuvensteijn et al. (2011), the error-specification model of Carbo et al. (2009), the

empirical model of Delis and Tsionas (2009) that simultaneously estimate market

power and operating efficiency, and the attempt of Bolt and Humphrey (2010) to

measure competition efficiency along the lines of efficiency-related standard

methodology (DFA). Section 3.5 sums up the theoretical and empirical review on

competition.

Chapter 4 lays down alternative definitions of bank output/inputs (section 4.2),

namely the production and intermediation approach (among others), before it comes

to the measurement of X-efficiency based on various parametric (section 4.3.1) and

non-parametric approaches (section 4.3.2). The former comprises the Stochastic

Frontier Analysis (SFA) of Aigner et al. (1977) and Meeusen and Van de Broeck

(1977), the Distribution-free approach of Bauer et al. (1992) and the Thick frontier

approach of Berger and Humphrey (1991). On the other hand, the latter includes the

linear programming method of Data Envelope Analysis (DEA) first proposed by

Charnes et al. (1978) and the free disposal Hull approach (FDH) of Deprins et al.

(1984).

Section 4.4 presents a thorough review of empirical applications conducting

international comparisons of efficiency estimates, European cross-country

comparisons, developed economies, transition economies and country-specific

studies. Conclusion is offered in the last section (section 4.5).

Chapter 5 makes reference to the relationship between bank competition and

financial stability. In other words, there are two strands of the literature (section 5.3)

that endorse whether competitive conduct of banks is a conducive channel-through of

systemic risk in international and European samples. On the one hand, we report the

so-called „franchise (charter) value‟ paradigm of Keeley (1990), which appears to

14

trace in a competitive market banks eager to recoup their squeezed profit margins by

engaging in risky projects. On the other hand, Stiglitz and Weiss (1981) put ‟blame‟

on incumbent banks for imposing high prices and exacerbating thereby their risk

profile due to potential defaults. Along the lines of both theories, we review the whole

argumentation stemming from theoretical papers to establish the underlying

relationship (section 5.2).

We also document the proposition of Martinez-Miera and Repullo (2010), who concur

that the two contending theories need not be mutually exclusive as a U-shaped

relationship may unify them if occurring at markets of different bank competition.

Section 5.3 offers some concluding remarks.

Chapter 6 initiates the empirical analysis of the thesis investigating the intensity of

competitive conditions in nine of the most developed European banking industries.

Section 6.2 refers to the methodology followed with respect to the translog cost

function in order to estimate the marginal cost. We then end up with estimates of

competition through the Lerner index referring to the overall banking income and to

its decomposition into loans, other interest income, fees and commission and other

non-interest operating income. We also propose a model of potential correlates of

market power for commercial, cooperative, savings and „other‟ banks as well as to the

whole sample.

The analysis in section 6.3 makes a brief description of competition determinants that

proxy for bank-specific, industry-specific and macroeconomic variables along with

time and country fixed effects. In section 6.4 we delineate descriptive statistics of the

employed variables and the way each banking market represents the sample

represents over the whole period. Section 6.5 shows the results of empirical modeling

the (income-specific) estimates of competition, and section 6.6 summarizes the

analysis.

Chapter 7 documents some methodological issues (section 7.2) regarding alternative

definitions of bank outputs/inputs, specifications of the cost function and the reasons

beneath our decision to apply specific methodologies. Section 7.3 presents the

employed model the results from the comparison of various parametric methodologies

(REM, DFA, TFA) that estimate cost and profit efficiency scores under alternative

definitions of bank outputs/inputs. We come up with relative rankings of country

15

rankings as well as rankings of different productive specialisations and asset size

classes. We next measure economies of scale (section 7.4) through the partial

derivative of total costs with respect to total assets and observe how they have

evolved since 2003 under both IA and VA approaches.

Section 7.5 investigates the beta and sigma convergence dynamics of cost and profit

efficiency scores. We apply the models for all cases along the lines of standard linear

modeling and dynamic specifications. Last, section 7.6 draws remarks on the

evidence.

Chapter 8 shows a sensitivity analysis of the causal effect of market structure

variables, namely concentration and market power, on bank soundness. In so doing, in

section 8.2 we present the methodology of extreme bound analysis as set out by

Leamer (1983; 1985) and Leamer and Leonard (1983) as well as the methodological

issues of modeling the underlying relationship. Section 8.3 lays down the employed

cost function and the formulas to estimate the marginal cost and the Lerner index in

the first place. Next, in section 8.4 we make reference to the determinants of

competition with regard to regulation and supervision alongside other controls of

bank specificities and economic development. Section 8.5 displays descriptive

statistics correlation matrices between key variables and section 8.6 provides full

account of the empirical evidence stemming from the baseline models.

We also investigate in section 8.7 how the impact of competition takes shape in

response to the interplay between regulation and foreign ownership. The analysis goes

one step further allowing for non-linear relationship between competition and

indicators of institutional development (section 8.8), such as corruption, government

effectiveness, dissemination of credit information, among others. Section 8.9 allows

for the constituents of Z-score to be used as dependent variables in order to verify

which aspects of bank soundness are explained by the model. Section 8.10

investigates how market structure variables behave for commercial, cooperative,

savings and „other‟ banks of the market, and section 8.11 makes a synopsis of the

results.

Chapter 9 provides the concluding remarks (section 9.1) of the empirical analysis

and pertinent policy implications (section 9.2), coupled with methodological

limitations (section 9.3) and some suggestions for further research (section 9.4).

17

CHAPTER 2

Competitive issues

19

2.1. Introduction

European banking has undergone a state of structural reforming either in an

institutional or macro finance perspective. The evolving financial engineering along

with regulatory initiatives and advanced information technology have contributed to

bank disintermediation as more and more funds are transferred away to insurance

corporations, pensions or investments funds. Firms often use capital markets as a

means of portfolio diversification and debt financing while households are keen on

equity investments to reap high returns in buoyant periods.

Advocate to this trend has been the impact of information technology on the

internationalisation of capital and money markets. The integration of European

financial markets was intensified in the aftermath of the Euro adoption fostering

higher levels of competitive conduct. In addition, bank assets are reported roughly

over 15% to be in hands of foreign owners, especially subsidiaries (branches

secondarily) in Southeastern union to procyclically anticipate and, therefore, affect

economic growth in this region. Further evidence is derived in the significant external

position of European banks in terms of cross-border asset/liability holding and income

geography.

However, the by-product of nonbank competition does not imply a modest role in the

provision of financial services. The driving force of large-scale R&D and allocative

efficiency, potential market entries or market structure impinges on interest rate

margins and bank profitability over time and across different sectors and size classes.

As such continuous process is unfolding, the implications on stability is a matter of

economic theorising and empirical application. On the one hand, Stiglitz and Weiss

(1981) allege that monopolistic markets tend to adopt higher interest rate margins and

thereby inducing potential defaults, whereas Keeley (1990) attributes financial

instability to competition when banks in an attempt to recoup returns carry on riskier

projects.

Before moving onto the literature review of empirical competition models, it is

necessary to define every single issue involved and the way they are related to each

other.

20

2.2 Institutional perspective

2.2.1 Efficiency

Considerable empirical research has been conducted on the X – efficiency of European

Banks. It is measured in terms of costs and profits to highlight the ability of specific

sectors to control costs or revenues with better management and/or technological

advancement. A profit-maximising bank is intended to be close to or right on the

efficiency frontier and, therefore, to eliminate expense preference behaviour of

managers when not acting to the best interests of shareholders. Above and beyond that

threshold, only technological changes can bring about further efficiency levels.

Two measures here are related to X – efficiency: scale and scope economies; that is,

the marginal (negative) change of unit costs in response to the employment of the

total volume of production or range of products, respectively. Possible allocative

inefficiencies are attributed to administrative costs or insufficient management and

staff productivity. In other words, efficiency can be accomplished by the optimal

allocation – and the efficient utilisation, therefore - of the input mix.

However, the fact of whether there is such thing in banking as scale economies has

occupied an extensive part of banking literature. Strictly speaking, it underlines the

possibility of altering the production level on the grounds of varying all factor inputs.

Some problems may arise here; it is quite hard to allege that all inputs are totally

variable and even if we do so, the risk profile of a bank may be at odds with the

maximisation of the shareholders utility. For example, how is it possible to triple the

deposit accounts? And is that action worth taking given that the prospective provision

of loans implies higher portfolio risk? Besides, the problem of indivisibilities rules out

any change of an input unit by a fraction; there is no way of adding one-third of an

ATM machine or so. Last, the range of financial services worldwide makes it hard to

quantify multiple outputs and empirically verify the extent to which scale economies

do exist.

On the other hand, the concept of scope economies stands out in cost minimisation

when it comes to evaluate the positive effects of a merger or acquisition. As a

causatum, it is not always obvious, if at all possible, that a joint production is bound

21

to plunge costs through intermediation. They can apply when supposing, for instance,

the core banking activity of intermediation, whereby a predetermined fraction of

deposits is loaned out to borrowers. However, the payments service system in Europe

is quite fragmented, comparatively to USA, and that constitutes a hindrance to exploit

cost complementarities. Correspondent and custody services as well as payments

through debit/credit charges are not sufficiently spread out in European countries to

be conducive to cost reductions.

The business term of economies of scope is synergy, through capturing a more

dynamic process. It consists in the blend-up of skilful employees and exertion of a

considerable influence on suppliers for further reduction in input prices, among

others. In banking, this concept often proves to be costly in cases where innovation

forces fall short of mixing different cultures or inspiring the entrepreneurial spirit of

small financial boutiques. Consequently, the extent to which scale/scope economies

and synergies exist in European banking is indispensable for strategic decision-

making. However tough it is to come up with such measures, empirical evidence

gives mixed results whether large and multiproduct banks have a comparative cost

advantage over smaller types.

Technological change is the other important component of the efficiency compound.

Irrespectively whether it is applied in either cost or profit equations, it is estimated by

taking the partial derivative of the already cost/profit function with respect to the time

trend. It seems quite possible to draw some lines regarding the constituents of

technological innovation (Baltagi and Griffin, 1988); that is, the pure technological

progress – the effect of time trend depicting knowledge advancement - the non-

neutral change - the elasticity of cost/profit to variation in input prices - the scale

augmenting change - the sensitivity of cost/profit to changes in bank outputs - and

quasi-fixed factor augmenting change – cost reduction attributed to changes in fixed

factors. Empirical evidence has shown that technological change improves profits to

make up for the foregone technology investment up to a point where everything is

diffused and decreasing revenues are offset by greater costs.

It appears to be a hidden trade-off between reductions in costs and profitability,

though not verified on the whole. In countries where big banks endeavoured to taper

costs by increasing ATMs or closing branches faced a much poorer quality in services

22

thereby incurring a considerable profit plummeting. Only commercial banks have

accomplished greater profitability as they utilized innovation for better services and

risk management purposes. However, smaller commercial banks have been subject to

diminishing revenues and smaller cost reductions, probably sticking to reap the

benefits of innovative forces over non-cost channels.

2.2.2 Consolidation

In the European landscape, the consolidation of banks and financial firms is expressed

either by the merger of two independent banks or by the acquisition; the former turns

out to establish a single unit and the latter refers to the control of the acquired bank

through holding the majority of its shares. In such a case, the assets are not integrated

and by no means constitute any form of combination of the two firms. Strategic

alliances can also lead to monopolistic power when tacit agreements come into play

thereby exerting a considerable influence on the other banks‟ competitive stance.

In reality, academics may be up against consultants when they are called to articulate

a word about an imminent merger. There is not always a clear-cut answer whether the

deal is worthwhile as no one is able to back it up with convincing evidence. After all,

both can be right if, for example, bankers are focusing on the cost cuts in the post-

merger phase and at the same time economists observe no efficiency gains in cases

where revenues are moving downwards in line with costs (Rhoades, 1994).

M&As could actually take place for several reasons. In case of expecting efficiency

gains, either stemming from scale/scope economies or cost/profit X – efficiencies

from better organisation and management, the resulting market power of the new

entity may end up to higher pricing and value-added for the shareholders. Moreover,

it might be the case of managerial selfish ego the intimate motive of a merger in a

way to maintain and even enhance their personal power. In addition, structural

changes in the European banking sector have rendered M&As to be imperative and

intensified their operation in off-balance sheet activities (bancassurance) and other

non-interest income sources (commission, fees). By and large, the latter occupies a

large proportion of the income of European banks and keeps on escalating up to 50%

particularly in larger diversified banks (Laeven and Levine, 2007).

23

The empirical literature has been testing the effects of M&As by applying the event

study method – whether and to what extent abnormal returns of stock market shares

are created when announcements about an imminent merger make agents form

relative expectations – and comparing efficiency and performance ratios before and

after the time of a merger/acquisition.

In general, European bank merging has experienced great abnormal returns when the

announcements of deals involved domestic banks and insurance companies for

product differentiation purposes (Cybo-Ottone and Murgia, 2000). In fact, Campa and

Hernando (2006) conclude that only the target banks enjoyed positive market

expectations around the time of the announcement and, on average, improvements in

performance and efficiency; the latter remains put, according to Vennet (1996), but

especially in cases of equal-sized domestic banks and cross-border acquisitions. For a

strategic standpoint, Figueira and Nellis (2007) found consolidation to have incurred

more efficient gains than alternative development strategies from other large banking

entities. However, higher performance has been evidently feasible by domestic banks

of equal balance sheet structure, and cross-border deals that drive similar loan

portfolios, risk management strategies, and capital - cost structures (Altunbas and

Marques, 2008).

Cross-border activity is supposed to be conducive to the integration of European

banking sector. Barros et al. (2005) point out that the tendency towards the creation of

subsidiaries and not branches constitute a hindrance that deters full integration or even

geographic diversification and deregulation enforcement. In the case of horizontal

product differentiation and exogenously determined sunk costs, the banking sector

becomes more fragmented and deconcentrated the greater the market size (Sutton,

1991). If customers acknowledge different quality in financial products (vertical

differentiation), and incumbent banks are able to endogenously affect sunk cost

expenditures, they are highly motivated to operate in a higher efficient scale by

financially and technologically innovating and branch networking. Discretionary sunk

cost competition is, therefore, focused on fixed costs and take several forms, most

inclined with propagating effects to advertising and, primarily, to information

technology (Vives, 2001). On the other hand, competition on variable costs is

profoundly the typical strategy if concentration is negatively associated to market

24

size; in the „80s, Gual (1999) empirically came up with an ongoing deconcentrated

landscape in European banking with the same patterns.

2.2.3 Relationship banking

The role of relationship banking lies in the definition of the bank‟s raison d‟être. In a

typical balance sheet, bank liabilities (e.g. deposits) are liquid since anyone can

withdraw on demand while assets (e.g. loans) are not marketable due to informational

frictions. The latter includes both asymmetric and proprietary information that needs

to be developed when the provision and pricing of loans requires ruthless screening of

the potential borrowers. The qualitative transformation of assets in the contemporary

financial intermediation theory consists in the way banks are able to serve long-term

loans with short-term deposits, guaranteeing at the same time the liquidity of savings

and inevitably undertaking credit and liquidity risks.

Hence, relationship banking is called to provide specific (proprietary) information for

every customer and evaluate whether the repeated provision of financial services

across time is profitable (Elyasiani and Goldberg (2004). The key thing here is that

banks obtain information when screening and monitoring services in the lending case

(Allen, 1990; Winton, 1995). As Berger (1999) puts it, relationship banking holds if

there is something more to be obtained than the readily available information, it is

confidential and a product out of many repetitive interactions with the customer over

a range of financial products. However, relationship banking is present in nonbank

intermediation and involves other financial services like deposits, check clearing and

so forth.

Now securitization comes into play, being an ongoing funding development in the

contemporary banking finance. Although it seconds more transaction-oriented rather

than relationship-oriented interactions, the so-called credit enhancement is facilitated

by lots of collaterals and letters of credit. The bank, on the other hand, is now more

cautious to price its assets credibly and keep on monitoring collaterals so as to acquire

the necessary proprietary information that abridges default risk; even so, the bank

should be able to preserve accountability in reports and internalise a portion of the

incurred cost.

25

The benefits of relationship banking seem quite obvious once we cerebrate that

bilateral exchangeability is maximized between borrowers and banks when they are

both more willing to reveal exclusive information. The social welfare is positively

affected by long-term agreements that allow for more discretion, – flexibility and

negotiation of contract terms - contractual covenants, - adaptive to the underlying

information set after negotiation takes place as being more effective in loans (Berlin

and Mester, 1992) – collaterals, - mitigating adverse selection and moral hazard

problems (Stiglitz and Weiss, 1981) on the grounds it is monitored closely as far as

relationship banking makes it possible - and time – bearing losses in the short term but

then recouped as relationship has been established. These losses may include

subsidies to new borrowers without which, they would be condemned due to adverse

selection and moral hazard problems. Before moving to the costs of relationship

banking it should not be overlooked that bank provision of loans constitutes a

credential of financial soundness and the quality of the borrower when being

monitored seems to complement capital market funding.

The primary costs of relationship banking are the soft-budget problem and the hold-up

constraint. The former pertains to the lack of bank credible threat (Dewatripont and

Maskin, 1995) to enforce the credit terms since the bank is bound to extend the loaned

money to avert borrower‟s default. It seems quite obvious that the borrower may

demonstrate inadequate endeavour to survive on its own unless the seniority of bank‟s

debt structure renders the threat credible and the overall value insensitive to the claim.

The hold-up problem pertains to the monopoly power a bank has gained from

proprietary information. Borrowers run the risk of getting charged higher interest rates

and trammelled to carry on thus forfeiting valuable investments (Rajan, 1992).

However, multiple bank relationships could disperse the burden of credit at the

expense of less valuable information to be obtained and greater competition, which is

reluctant to lend to young firms. The only way out, as put forward by Von Thadden

(1995), is the long-term contracting and continuation of it upon performance. In other

words, poor projects may be terminated or get compensated depending on some

parameters or alternatively, refinanced at prespecified terms. Therefore, ex post

bargaining power of banks is circumscribed and firms are induced to keep up valuable

projects with single bank relationships.

26

It is quite ambiguous whether competition is negatively correlated with relationship

banking. One strand of argumentation falls in the competitive environment that

enables customers to switch to other banks; that is deemed to imply negative

repercussions relative to reduced and shorter relationship-intense investments

inhibiting costly proprietary information and so forth. However, Degryse and Ongena

(2008) have emphasised the importance of switching costs – e.g. search and shoe-

leather costs, that is time and effort to open/close accounts, transfer funds, etc. – that

amplifies the development of relationship banking. In such case, screening becomes

an integral part of the relationship though it may end up inefficient for borrowers,

whose assets are dominantly intangible (Houston and James, 2001). It could be much

worse for new firms that need to be subsidised in the first place as loan rates may

escalate intertemporally to offset the prospect of no retrievalibility of bank rents.

On the other hand, Boot and Thakor (2000) emphasise that relationship banking may

be part of a bank‟s competitive advantage increasing the welfare of the some. Tailor-

made services render the bank unique in the eyes of borrowers even amid fierce

interbank competition. However, capital market competition is expected to downplay

relationships and the per se lending orientation. In contradistinction, Dell‟ Ariccia

(2000) argues exactly the opposite, that there is a potential trade-off between market

share and relationship banking. As competition heats up, banks are aware of fewer

borrowers (adverse selection problem) and have therefore an incentive to screen.

However, the more banks are competing each other, the higher is the motive to switch

into transaction-oriented mode and grasp greater market share. Empirical evidence

has showed that single bank relationship has resulted in more value-added than

multiple schemes, the less the duration of such relationship, the greater is the distance

between bank and firm and less is that between bank and its competitors (Farinha and

Santos, 2002; Degryse and Ongena, 2001; 2005).

2.3 Macro finance perspective

2.3.1 Regulatory developments

Before delving into the regulatory directives put forth to the unification of the

European countries, we should have a look at the rationale of the purpose of

27

regulating banking markets. To the best interests of any investor, regulation applies

for his protection by providing sufficient information for hidden risks of traded

financial products. Even in cases of law breach, illegal activities, like tax evasion,

money laundering, find their way to judicial prosecution and investors are alarmed for

not engaging in. In addition, the abuse of potential oligopolistic markets make

authorities keep a vigilant eye on mergers and acquisitions to verify whether such

restructurings will have a considerable impact on market shares and product pricing.

Along the lines, negative market externalities might otherwise undermine financial

stability and the viability of intermediation.

Prudential regulation at the microeconomic level is indispensable once the

repercussions of a bank run can have a detrimental effect at the macroeconomic level;

that emanates from the banks operation in a homogenous product market and thus

their exposure to the same kind of risk. However, the reputation of a bank, if exists,

compensates for the prohibitive costs of acquiring information, the transparent

activities as depicted in the balance sheets, among others. A bank failure is

susceptible to the variable confidence of investors, which may then metastasize to

other banks as investors herd to withdraw their deposits before their bank liquidate

their assets. The underlying contagion is exacerbated by the systemic risk possibly

disturbing wholesale and retail markets and, in the extreme, the collapse of bank

intermediation, shortage of liquidity services and problematic resource allocation.

Therefore, social costs are quite severe to be overlooked given that the aggregate

consequences of potential bank runs in the European union render the

macroprundential regulation of primary concern (Borio, 2003, 2006; Acharya, 2009).

On the other hand, if banks are secured in view of upcoming runs on deposits, moral

hazard problems may arise. Banks may be willing to undertake risky investments if

deposits are protected by some insurance scheme or the state treat banks as being too-

big-to-fail institutions. The looting hypothesis as first articulated by Akerlof and

Romer (1993) provides some answers to the managers‟ gambling attitude. If managers

are well informed about the imminent collapse of their bank, they tend to engage in

risky projects to increase short-term profits and therefore share prices to exercize their

put options. Even shareholders overlook the underlying gambling when enjoying

share overvalues whereas well-informed shareholders have the opportunity to sell at

high prices. In the end, if gambling has averted the imminent collapse, everyone is

28

better off; if not, managers have managed to flee with profits abandoning the bank

with major losses. The key argument here is that managers are expected to be

uninterested in deposit losses while shareholders are falling on the realisation of short-

term profits even if deposit insurance is capped at certain level, if there is co-

insurance scheme or if the deposit type is not qualified for it (Heffernan, 2005).

2.3.1.1 International initiatives

In response to the harsh bank failures in 1980s, the Basel committee was established

as a permanent secretariat at the Bank of International Settlements. It consists of the

supervisory authorities of G-10 countries, Luxembourg and Switzerland, meeting up

once every three months.

In 1975, the Basel concordat was the first agreement upon the way responsibilities are

split into home and host country supervisors. Host countries were charged with the

solvency and liquidity of foreign bank subsidiaries while home countries were

responsible only for the liquidity of foreign branches. An amendment was put forth in

1983 (Revised Basel concordat) after the panic triggered by the suicide of Roberto

Calvo, the chairman of the Banco Ambrosiano bank. That incident made depositors

withdraw all their money and Bank of Italy provide a rescue plan, though in vain. The

bank failure underlined the gaps in regulation when a bank‟s subsidiary located in

Luxembourg declared bankrupt after the refusal of both host and home country

authorities to inject cash. Ever since, liquidity problems lie in the joint jurisdiction of

host and home authorities, either occurring in branches or subsidiaries. Solvency

responsibilities are split to the home state‟s central bank for branches whereas to both

home and host countries for subsidiaries.

However, until then there had been no provisions of the lender of last resort in many

territories in the EU. Although rescue packages were quite frequent the last fifty

years, the coordination of authorities on the rescue of branches and subsidiaries as

precipitated by the upcoming failure of their headquarters. On the top of that, the

Basel Committee addressed the issue of applying deposit insurance on other bank

liabilities. However, in interbank and wholesale activities there is limited information

asymmetry to require insurance; even currency deposits are preferred to be unsecured

29

since the concern of shifting away from the foreign bank to its headquarters cannot be

overlooked.

The Basel Accord was passed in 1988 to set out the capital required to safeguard the

financial health of banks with reference to their assets and liabilities. Stemming from

the Cooke ratio, the risk assets ratio is defined to be the capital over assets and credit

risk equivalents. Specifically,

Capital (tier 1&2) / risk-weighted assets + credit risk equivalents

Where,

Tier 1 (core capital): equity capital, including common shares, non-cumulative

preferred stock and other hybrid instruments, disclosed reserves,

Tier 2 (supplementary capital): undisclosed reserves, cumulative preferred stock,

revaluation reserves, hybrid debt instruments, general provisions, loan loss

allowances, subordinated debt.

The risk weights of all assets are categorised by credit type:

0%: cash, gold and bonds issued by OECD countries and U.S. treasuries

20%: bonds issued by OECD banks, municipal governments, securities issued by U.S.

agencies and insured mortgages

50%: uninsured mortgages

100%: corporate bonds, claims on non-OECD banks and under-developed countries,

real estate, plant and equipment.

The other component of the denominator, namely credit risk equivalents, refers to the

(non-)trading off-balance sheet instruments (futures, swaps, letters of credit, etc),

which are assigned the weight of the counterparty‟s type in tandem with the given

claim. The Basel Accord sets out the minimum of 8% of capital for banks to set aside

vis-à-vis imminent risks (credit, counterparty). Hence the capital requirement for

every single asset category of a bank is calculated by the risk-weighted asset times the

8% minimum capital. The off-balance sheet charges are supposed to be the product of

the credit exposure of derivatives according to the Basel accord. Last, the credit risk

30

charge for off-balance sheet items falls under the same formula for on-balance sheet

items, though imposing a 50% weighting. The latter is justified to address a good

quality of credit, since only large and sophisticated banks are able to engage in such

behaviour.

Some objections have been raised against the effectiveness of Basel Accord to

accomplish a level playing field for international banking. First, there is a debate on

whether the use of book-market values of equity in capital ratios is appropriate.

However, in times of interest rate and stock price volatility, the use of book values is

expected to mitigate the fluctuations of capital assets ratios, thus without having to act

correctively to abide by the rule of 8%. Second, tier 1 and 2 capital is measured

differently across countries in line with the discrepancies of national accounting,

regulatory and tax systems. Thus, it seems hard to quantify risks and compare capital

ratios amongst European states. Third, the credit risk equivalents of off-balance sheet

instruments do not account for any risks involved other than the credit risk. The

operational and market risk may jeopardize the default on certain instruments, when

no capital is set aside to guarantee against volatility in interest rates, liquidity and

currency risks, among others. Forth, the weightings assigned to every asset category

follow an one-size-fits-all approach.

The classification of assets according to its credit risk applies falsely to every

instrument within any category. For example, a bond issued by an OECD bank which

receives AA rating, gets lower weight (20%) that that of a loan to an enterprise with

AAA rating (100%). Last, regulatory capital arbitrage can exploit gaps in regulation

to retain lower capital than that corresponding to a bank‟s risk exposure. In case, for

example, a bank has unsecured mortgages, it can pass them (sell) on to a special-

purpose vehicle (SPV; subsidiary), to use them as a collateral for the issuance of new

bonds. In that way, the bond investors are classified to priority/seniority tranches, and

the stream of mortgage payments is used to pay off the coupons on a rollover basis,

first serving the most senior tranches followed by the next senior ones, etc. The key

thing is that senior tranches have comparatively high credit ratings, allowing for low

capital requirements outweighing – totally in absolute numbers - the higher

guarantees for the least safe tranche.

31

The imperative of filling the loopholes in regulations came along with the Basel

amendment in 1996. In two years time, its provisions were introduced internationally

to capture the changes in market risk; that is perceived to include the market and

specific equity price risk, interest rate risk, currency risk and commodity risk. In that

way, the off-balance sheet items are treated in a more sophisticated way, rendering the

approach of credit risk equivalents as inefficient and lumpy. Besides, in the nominator

of the Basel ratio, the tier 3 capital, marking the capital charges in response to the

underlying market risk, is added up to define the short-term subordinated debt on the

grounds that the minimum capital requirement is fulfilled at all costs (interest,

principal).

The use of a particular internal model to quantify the market risk was left up to the

discretion and the approval of national regulators. The internal model involves the

average market VaR and the stressed value-at-risk numbers over a period of sixty

days. In each case, bank models of VaR require:

daily computation

one-tailed confidence interval set out to be 99%

minimum „holding period‟ of 10 days prior to asset liquidation, unless

otherwise sufficiently justified for employing shorter periods

the time series of sample observations not less than one year

data update at least once a month, unless otherwise called for vis-a-vis great

price volatility

no particular model prescription but contingent on the nature of portfolio

whatsoever. In addition, the employed model could be linear (covariance

matrices, factor modeling, diagonal modeling), nonlinear (gamma,

convexities), historical or Monte Carlo simulations, coupled with the

necessary stress-testing of scenarios, correlations and policy responses that

may infer great bank losses.

32

though banks are required to empirically model all the risk factors influencing

rates and prices, including non-linearities in options (-like) positions.

The alternative model of market risk follows the standardized approach which

determines the capital charge as accrued by the involved risks and the conversion of

off-balance sheet iterms into spot equivalents. Portfolio diversification is dismissed in

the way price changes in equity prices, commodity prices, interest rates and exchange

rates do not comprise covariance factors.

Concerning the risk in gold and exchage rates, the capital charge of derivatives is

computed on the 8% basis of the sum of long or short positions – whichever the

greater in absolute values is – along with the net gold position. The equity risk

computation involves the capital charge to be applied on both specific and market

risks. In particular, the capital required to be set aside is the sum of the factor specific

risk – net position times 8% - and the market risk factor – gross position times 4%,

assuming that the systematic risk is the same and diversification is left off to the

discretion of national authorities for approval. Commodity risk‟s source engulfs a

great pool of commodities, such as metals, gas, grains among others, and appertains

price changes (exchange rates worldwide, etc) and other risks inherent even in the

production process (namely, input cost, quantities, political instability).

However, the underlying approach is reminiscent of the aforementioned methods, but

nonetheless unpopular. The reason is that there is no way to diversify risks by

accounting for correlations and the compliance to such risk management system is

bound to incur huge costs. Last, interest rate risks are accounted for by classifying all

debt instruments in time bands conditional on their maturity. A specific risk factor is

attributed to each time band that takes values between 0 (short maturity period) and

12.5 (long) and thereby reflects sensitivity of certain positions to conjectural interest

rate changes. For the valuation, in turn, of the net position in each time band, vertical

disallowance (10% capital charge levied on the smaller of long/short position) is

applied to offset possible duration mismatches within each band (namely, basis and

gap risk). Then, the underlying time bands are classified further into three zones - 0-1

year, 1-4, 4-over 20 years – and the net position is calculated for every single zone.

That is so to account for correlation of interest rate movements in different time

33

bands. Thus, a risk factor is implemented on the net position of each zone (horizontal

disallowance), 40%, 30% and 30% respectively. Thereafter, banks should come up

with the net positions between zones in pairs and set aside a 40% capital for adjacent

time zones (1-2, 2-3) and 100% capital for the zones 1 and 3. Finally, the net positions

are rolling over the aforementioned steps as soon as the calculation of the capital

charge is rounded off.

On May 2004, the new Basel Accord (II) was passed much to the following

commentary reports of banks and the impact studies of members of the Basel

committee to propose further changes in the initial proposal (2001). Though a

between-season was provided for G-10 countries to abide by the standardized

approach by 2006 and the advanced ones by 2007, capital charges were calculated on

the basis of both Basel I and II rules. Only a small number of US banks and the

majority of broker dealers were supposed to adhere to the new rules, while the rest

would keep on operating to the letter of Basel I. Contrary to such subversive attitude,

provisions of the new Basel accord became part of European directives and ratified by

the European parliament. In addition, countries with no explicit representation on the

Committee were eager to abide by the new rules in the aftermath of the affiliation of

developing countries in the membership body of the Bank of International

Settlements.

It comprises a 3-pillar approach setting out the measurement of all risks, the role of

supervisory authorities and the establishment of market laws regarding soundness and

transparency. In particular, credit risk is calculated by the new standardized approach

that applies quite different pro-rata risk weightings. Alternatively, the bank may opt

to utilize the credit scores published by qualified Export Credit Agencies (ECAs) or

those participating in the „Arrangement on officially supported export credits‟ (OECD

trade directorate). Banks assign a risk weight lying in the next less favourable rating

to the respective rating of a country or alternatively, a risk weight based on the

external assessments of themselves. See in detail the comprehensive version, Basel II

: June 2006.

In reference to the above assigned risk weightings, debt instruments are likely to be

allocated to higher-rated firms since otherwise banks may be required to set aside

34

more capital as counterbalance for potential losses. In absolute terms, that is the 8%

capital charge of total risk assets that remains put from Basel I.

The alternative internal ratings based approaches (IRB henceforth) have been devised

in order to allow banks to use sophisticated risk-weighting techniques. By using

internal ratings and information, the capital to be set aside covers unexpected losses

along with potential default on credit. The opportunity for banks to do so is granted by

the discretionary decision of the Committee upon the fulfilment of explicit criteria:

risk differentiation, internal validation, probability of default estimated for each group

of borrowers with data covering a period of at least 2 years since Basel 2 kicked off,

loss given default (LGD) requiring 7 years of data for advanced IRB banks, approved

risk components by which risk weights are determined, internal ratings and VaR

methods to be employed in an integrated system of risk management for the definition

of economic capital as well as the establishment of disclosure standards.

Banks are asked to classify their assets to classes of several exposures according to

the definition of risk features underlying therein. Such classes are corporate,

sovereign, equity, bank and retail with further decomposition to sub-classes in some

cases. Even though banks are not required to apply what the Committee is asking for

risk management, they do have to comply with the treatment specified to each

exposure relative to the minimum capital requirements. The foundation and the

advanced approaches are founded now in terms of their risk components, the

functions that the latter is transformed into risk-weighted assets and the minimum

capital requirements for each asset class.

As far as the risk components are concerned, banks are asked to place their own

estimates of probability of default (PD), loss given default (LGD), exposure at default

(EAD) and maturity (M). All the functions applied to derive the risk-weighted assets

are laid down in paragraphs 270 to 370 whereas the minimum requirements for first

entry to IRB methods and on-going use of it – e.g. leasing, disclosure, corporate

governance, among others - are presented in paragraphs 388 to 537.

Apart from that, the Basel Committee recognizes securitisations by means of

collaterals, guarantees and derivatives as risk mitigants. If a bank is originating

securitisation, then an amount of capital requirement should be calculated as if that

asset was not securitized at all. In fact, the definition of transactions that are treated as

35

securitisations are explicitly provided in the conditions part of paragraphs 543 to 552.

However, cases when securitisations are not accounted for the calculation of risk-

weighted assets are boiled down to the operational requirements of paragraphs 554 to

559.

For the collateral treatment, the standardized approach of credit risk is applied

utilising specific risk weights. The key types of recognized collaterals are cash, gold,

government bonds of at least BB- rating or BBB-rated bonds and other unrated

securities when issued banks or securities firms, among others. If now banks are

qualified to use the IRB approach by their national authorities, the components

involved are again the aforementioned previously plus receivables and several types

of residential property. The maximum requirement is the equivalent to the „sheer‟ IRB

result had there not been securitisation; deduction from tier 1 and 2 (50% each) is also

provided, unless it is not the case of an equity increase, which should be deducted

from tier 1, or an implicit support. For an extensive analysis of IRB approaches, see

„Treatment of securitisation exposures, section 4 (BIS, 2003a).

Guarantees are recognized on the grounds of the guarantor‟s ability to pay off all

„non-payments‟ of the counterparty. Credit derivatives, especially credit default

swaps, are more preferable since they provide insurance in cases of imminent

defaults.

The last component of pillar 1 deals with operational risk. It is defined as ‘…the risk

of losses resulting from inadequate or failed internal processes, people and systems, or

external events.‟ There are three approaches of methodology, namely the basic

indicator, the standardized and the advanced measurement approach(es). The first

amounts to the average gross income over a three-year period times an alpha factor

that highlights the level of capital charge. All negative values are to be deducted from

both numerator and denominator.

As for the standardized approach, all banks services are classified into eight classes

and assigned specific beta factors. Thus, the capital requirement is the average capital

charges across all eight classes over a period of three years. Last, the advanced

measurement approaches (AMAs) are available to greater-sized banks, which are able

to purchase insurance in order to reduce the capital charge by at most 20%. However,

the subordination of banks to this option is contingent on certain conditions and

36

approval of host country authorities. Large banks may opt to include diversification

benefits within their operation and not those of the whole spectrum of their group

activities.

Pillar 2 sets out the responsibilities of national supervisors. More specifically, they are

expected to produce Basel ratios according to the provisions of the aforementioned

methodology and evaluate the risk management of banks guaranteeing the

preservation of an appropriate operational level. Moreover, they have to keep a close

eye on the fulfilment of minimum capital requirements and go through immediate

action in the advent of capital deviations. However, the framework of supervisory

conducting is not provided to the letter of any endorsed directive, turning out thereby

to play a back-up role in the execution of pillar 1. Pillar 3 last reinforces banks to

adhere to the provisions of the other two pillars. They should abide by the definitions

of risk exposures, the computation of respective capital charges through the

appropriate methodology and the disclosure of transparent information to the market.

2.3.1.2 Towards a single EU financial market

Initiatives to the harmonisation of the regulations in the Euro area are traced back to

the treaty of Rome in 1957, when free trading was to be accomplished with the

convergence of national regulative directives. The subsequent Single European act in

1986 proved to be an accelerating mechanism to second integration of markets

transforming the whilom perception of collective convergence into the principle of

mutual recognition of a minimum set of rules. European Commission has put forward

the minimum requirements, upon which the so-called European passport can be

applied; that are the provision of financial services and the establishment of bank

branches across the borders. All countries and banks headquartering in the euro area,

should abide by the rules of financial supervision and controlling.

Amongst the acts to be applied, banking markets could achieve higher degrees of

integration according to the first banking directive in 1977. The notion of credit

institution was defined to be a bank entity that loans out its deposits and other

repayable funds. The overall supervision of all credit institutions is in the jurisdiction

37

of the member state where the head office of a bank is located in order to secure

savings and ensure competitive conditions.

The second banking directive followed in 1989 to define the measures with which the

European passport would be accomplished. Before this promising general banking

framework, banking business was hindered by and subject to the specific

requirements of the host countries laws; e.g. authorisation or endowment capital on

entry. But ever since, the pursuit of business has been prescribed by the notification of

a bank, intending to establish a branch, to the home country authorities. Unless they

have reason to cast doubt on the endeavour, the host country‟s authorities should be

informed within three-months period. The endorsement of consolidated supervision

came along with rendering the host country authorities responsible for the supervision

of the liquidity of banks and their branches in cooperation with the home country,

though the final decision of whether a bank is to be liquidated lies in the hands of the

home country supervisor. Full responsibility of the host country applies in the

supervision of risk management and implementation of monetary policy. For

example, if a European country hosts a subsidiary of an outsider, then it is subject to

the monetary policy of the European Central Bank.

On the contrary, prudential supervision of any credit institution, reflecting the

soundness of administration and accounting procedures as well as the provision of

internal audit mechanisms, is the responsibility of the home country competent

authorities. In general, the principle of equal treatment applies everywhere. Disputes

that may occur when European banks find themselves to be unequally treated with the

subsidiaries of third country banks, since different sets of regulation apply, are

resolved using the right of suspending banking licences, negotiating bilateral terms or

taking refuge in the World Trade Organisation.

Hence, it seems to be the most important directive that gives palpable guidelines and

sets out a great range of bank activities like deposits and other repayable funds,

lending, factoring, forfaiting and financial leasing, money transmission mechanisms

(cheques, ATMs, credit/debit cards), derivatives, money market and exchange rate

instruments, securities, portfolio management and consulting on capital structure,

strategy, M&As and personal finance, safekeeping and administration of securities,

credit reference and custody services.

38

The underlying directive set out the minimum equity capital requirement to be 8%

and any deviation of it, must be restored by all means as soon as possible. In addition,

competent authorities must be notified in cases when more than 10% of the equity is

acquired by single shareholders, and the bank stock holdings exceed the threshold of

10% of non-financial firm‟s value and/or 60% of the bank‟s capital. Apart from

those provisions, some other directives have been adopted to determine a level

playing field of the minimum requirements.

The Large exposures directive was passed in 1992 and put into effect in 1994. Credit

institutions are required to be exposed at most by 15 per cent of their equity capital to

an individual borrower or 40 per cent of bank‟s funds. Banks are subject to the

threshold of 25 per cent exposure to one borrower (or group of), which by no means

can exceed eight times their own funds.

The Investment Services directive was passed in 1993 and implemented in 1995. It is

reminiscent of the second bank directive, that is, the mutual recognition on some

minimum requirements set out to second investment services. The objective is to fill

the loops of national regulations in order to ease the provision of investment services

by banks and securities firms. Such services have granted cross-border access to

trading systems and include derivatives, money market instruments, transferrable

securities and unit trusts.

The directive on deposit guarantee schemes (1994) and its amending acts (2005;

2009) as precipitated partly by the financial crisis, have been established to avert bank

failures and at the same time protect small depositors/investors. Banks are bound to

finance a minimum fund of 50,000 Euros (20,000 Euros in the beginning) to cover up

deposits on Euro or foreign currency. This amount is fixed up to 100,000 Euros by the

31st December 2010. There is also the option – or the mandate in states like UK - for

co-insurance, according to which, depositors contribute as much as 10 per cent of

their deposits to the insurance fund as an incentive to monitor bank‟s activities.

However, a member state‟s branch located in another European country is covered by

the host country whereas out-of-the-union banks with branches located in the

European territory is required to provide an equivalent cover or otherwise be

subjugated to insurance schemes already operating in the host country along with the

dissemination of the underlying information to actual and potential investors.

39

Passed in 1999, Financial Services Action Plan was an ambitious endeavour aspiring

the integration of wholesale and retail markets, the financial stability and the

optimality in the operation of the single financial market. In doing so, it establishes a

set of conditions, rules and supervisory directives as a framework of action and

introduces the methodology, that is, a group of measures to assess the economic

impact of FSAP on banking, insurance and securities markets.

The directives of consolidated supervision, the first passed in 1982 but replaced by

another in 1992 and implemented in 1993, referred to the consolidation of accounting

reporting of financial groups. According to it, parent institutions are required to own

at least 20 per cent of the capital of their own financial subsidiaries.

The Reorganisation and winding-up of credit institutions directive was passed in

2001, at a time when financial services were at a glooming path. It sets out that in the

case of a credit institution with branches in other member countries going bankrupt,

the underlying procedure of the country (home state) which hosts the headquarters is

to be followed. Besides, at short notice the competent authorities of the host country

should be informed about the winding-up proceedings. Otherwise, there might have

been conflicts of jurisdiction and unequal treatment between creditors. The same

applies in the reorganisation case of an institution, where the consumer is protected

vis-à-vis potential conflicts in employed approaches. However, if the head office is

located outside the EU, the branch, which is going to be wound up, lies in the

jurisdiction of the member country.

The Statute for the European Company (known in Latin as Societas Europeae or SE)

came into force in 2004 after being passed in 2001. The European company can be the

product of four possible ways of formation: merger of public limited companies from

different member states, holding company available to both private and public limited

companies headquartering in different member states or having branches/subsidiaries

in countries different from those where their offices are registered, joint subsidiary –

the same laws apply - and public limited company previously pertained to the national

law of a member state. The registered office may change location within European

union without being dissolved and established again. Moreover, it is indicative that

the initial capital for a SE is 120,000 Euros unless otherwise required by a member

state for certain kind of activities.

40

The directive of financial conglomerates was signed in 2002 to consolidate the

supervision of them in a way to stabilize the international financial system. They are

defined as entities engaged in at least two out of the sectors of banking, insurance and

investment. Their main activity (usually banking) should contribute at least 40 per

cent to their balance sheet and the secondary one at least 10 per cent. The benefits

stem from their ability to diversify risks and exploit economies of scale through

moving assets in different activities. Still the lack of harmonized legislation

encourages potential fraud and capital arbitrage to different divisions within the

conglomerate to avoid great capital charges. Maybe the most discouraging drawback

is the fact that there is no provision of the way total risk can be quantified since

financial products are hardly sorted out by sector along with the different risk

exposures of banking, insurance and investment to financial products.

As for the EU takeover directive, it was agreed in 2003 after long negotiations. There

has been no provision of a level playing field and member states are not obliged to

abide by every single article of it. Its general principles are the protection of all

holders of the securities of the offeree company in terms of equal treatment, time and

information to decide on the takeover bid and its effects on employment among

others. There should not exist any false markets to leap at the chance of the bid,

benefiting from the temporary destabilation of any company‟s prices. The bid should

be able to be fulfilled by the offeror and by no means delay considerably the way

offeree company‟s affairs have been conducted so far. Moreover, the supervisory

authority of the bid is the member country at which the securities of the offeree

company have been accepted for trading, notwithstanding the usual case being the

home supervisory authority that is where its office is registered. Last, the mandatory

bid is required to be included in the takeover rules if that confers control of the offeree

company. The latter is to be determined by the percentage of voting rights, that is a

threshold that has neither maximum limits nor the same values across companies or

states. Rather, the bid is used to being made at an equitable price, namely the highest

of those during the last period of six to twelve months. Of course, discretion is

provided to the authorities to adjust it upwards or downwards.

The leitmotif of the European monetary Union was the initiatives to strengthen the

regulatory framework and supervision in the financial sector. The Lamfalussy report,

launched in 2001 comprised a four-level action in banking, insurance and securities

41

markets; namely, the adoption of the legal framework and the proper implementation

of those measures provided by Community directives, the coordinated action of

supervisory authorities and the diffusion of such policies in national laws. The advice

of the underlying directive rules lies in the jurisdiction of Committee of European

Securities Regulators (ESRC), which passes their propositions onto the European

Securities Committee (ESC). The ESC is charged to make further legislative changes

to keep up with rapid changes in financial markets, before they are endorsed by the

EU parliament. Hence, the new legislative process has been established to overcome

thorny problems at a higher pace, though in some crucial cases (e.g. Basel II)

European Commission opted for the traditional ratification process, which is bound to

creep up for at least 3 years until passing through the EU parliament.

The aims that the Financial Services Action Plan (FSAP) had envisioned for,

followed the White Paper on Financial Services Policy for the period 2005-2010.

Apart from that, the integration of financial markets is deemed to be accomplished by

transpositing the progress across all European territories, perfection of the employed

measures, harmonisation of the supervisory action and the removal of any barriers

involved. Priorities are identified especially in retail services, long-term savings

products and venture capital markets through more effective lawmaking, ex-post

evaluation of the measures and coherent compaction between European and national

body laws. Thus, by that time, the untapped potential for further employment and

economic growth is now accompanied - and to a degree backed up – by running

projects on retail banking, sound and transparent shareholdings, regulatory framework

in insurance market and efficient clearing and settlements in cross-border

transactions.

The European Parliament ratified the directive on Markets in Financial Instruments

and Investment Services (MiFID) in 2004. Along with the aforementioned action

plans, efficiency and integration in financial markets, investor protection and the

establishment of requirements that authorize bank activities were put forth in the

amendment of legislative framework in investment services. Incumbent upon the

Member states, the operation of banks, stock exchanges and other firms across the

borders is subject to the same conditions and procedures while prospective

acquisitions are evaluated according to harmonized assessment criteria, which

explicitly lay down notification thresholds, qualifying shareholdings, among others. In

42

addition, investors are protected by competent member authorities, which constantly

monitor potential breaches of law. On the other hand, market credibility is ensured by

a pre-trading transparency rule, according to which firms („internalisers‟) are obliged

to make quotes publicly for preferable transactions, but having the right to alter or

withdraw them in view of high risks jeopardising their position.

2.3.1.3. Basel III – framework in progress

Basel III is a new development of the Basel Committee to reform bank-level

regulation for the resilience of financial institutions during bank crises, and

macroprudential regulation that mitigates the impact of systemic risk and its

procyclical impetus over time. In particular, it pertains to two main banking variables

that are presumed to play a vital role in banks‟ solvency: equity capital and liquidity.

First, the classification of the issue of capital into three pillars takes the following

form:

Pillar 1: level of capital buffers, risk coverage and leverage ratio

Pillar 2: Risk management and supervision

Pillar 3: Market discipline

According to Pillar 1, banks are required to withhold 4.5% of risk-weighted assets as

common equity (2% in Basel II) and an extra capital conservation buffer (2.5% of

risk-weighted assets). There is also a countercyclical buffer within the range of (0-

2.5%) imposed at the discretion of the supervisory authorities on banks if they deem

that the pace of credit growth can result in high systemic risk. Furthermore, Tier 1

capital includes predominantly common shares and retained earnings and less by

other subordinated instruments with non-accumulative dividends and coupons, no

maturity dates or redeeming incentive. The instruments of Tier 2 will be harmonized

and those of Tier 3 totally phased out. There is also the alternative in cases of bank

insolvency for the private sector to resolve banking crises by contracting for common

shares in lieu of write-offs or conversions; a practice that reduces moral hazard if

financial support is provided „externally‟.

43

Risk coverage renders the need to reform against risk potentially related to (on- and

off-) balance sheet items and derivatives. The committee strengthens with more

rigorous analysis the capital requirements for securisation exposures that are

externally rated. We also have the introduction of stressing value-at-risk capital

requirements over a 12-month period of banking stress for the bank‟s trading book

(trading, other derivatives) coupled with a capital charge that accounts for risks

(default, migration) inherent in unsecuritized credit products. Pillar 2 extends its

coverage over counterparty credit risk by the use of stressed inputs in a

countercyclical context. It is also binding for banks to adapt their capital buffers in

line with mark-to-market losses that illustrate essentially how the creditworthiness of

the counterparty is fading out. This credit valuation adjustment (CVA) builds upon

the premise of Basel II that only outright defaults are addressed in the analysis.

For derivatives exposures, the committee asks for more effective collateral

management and longer margining periods, makes banks be subject to low risk weight

(2%) if they use central counterparties (clearing houses - CCP); default fund

exposures will be capitalized in CCPs according to a risk-based method. Hence, by

imposing more capital requirements for OTC derivative exposures, banks are

motivated to utilize the CCP facility. Moreover, exposures to the financial sector are

deemed more correlated than those to non-bank sector, and thus require higher risk

weights for the sake of systemic risk.

Containing leverage is the last constituent of the Pilar 1, which is not grounded on risk

exposure. Rather it puts a restriction on bank leverage in order to mitigate the

tendency of deleveraging during bank crises. In such times, the banking sector is

compelled to lower leverage driving to diminishing asset prices, high losses and credit

crunch. For measurement purposes, the leverage ratio is easily estimated in an easily

comparable way allowing for adjustments in differences of accounting standards.

Pillar 2 provides further framework on risk management and supervision. It addresses

a set of measures that mitigate procyclicality and banking sector vulnerability in times

of economic growth. In particular, the failure of Basel II was the procyclical

mismatch of minimum capital requirements against exposures on trading,

securitisations and off-balance-sheet activities. Such capital coverage fosters effective

risk sensitivity across financial institutions albeit at the expense of introducing some

44

degree of cyclicality on risk-weighted requirements. Hence, the Committee allows the

use of data over longer time horizons in order to estimate default probabilities, losses

given default (LGD) that are then expressed in appropriate capital requirements. Apart

from assessing the impact of cyclicality of the minimum requirement as set out by

Basel II, the Committee may reconcile risk sensitivity and stability in capital

requirements by utilising probabilities of default in internal ratings-based models for

the status of banks‟ portfolios during deteriorating economic conditions.

The Committee also promotes less procyclical provisioning practices based on

forward looking. In particular, the use of the expected loss (EL) approach instead of

the old „incurring cost‟ approach stems from different accounting standards that

highlight actual losses in a more transparent way. We also have incentives for greater

provisioning along with explicit guidance to the supervisory authorities about the

effective employment of the EL approach.

As far as the capital conservation buffer is concerned, the Committee enabled

supervisors to effectively introduce it under a framework that prescribes a

harmonisation of conservation standards internationally. The banking sector becomes

more resilient during economic downturns as banks are subject to a range of possible

actions consistent with the standard. That alleviates procyclicality given that market

fails to adapt accordingly to deteriorating economic conditions in the fear of sending

negative signals and thus banks, even if their profitability recovers, fall short of

engaging in adequate lending due to low capital buffers.

This new framework of strengthening equity capital comes into play in times of

excess credit growth. The Committee introduces a range of buffer level contingent on

the point of the business cycle we are standing at. We therefore make up for

unexpected losses through the cyclicality of the minimum capital requirements and

other additional buffers. There might be cases when extra buffers are necessary above

the cyclical minimum even if the latter is zero; supervisory authorities may require

that in anticipation of exogenous shocks.

We also have the issue of systemic risk and interconnectedness in Pillar 2. Since

procyclicality is amplified by the degree of interconnectedness among the banks of

high systemic risk, the Committee promoted an integrated approach of quantitative

and qualitative features that defines which bank seems important. Then, these banks

45

are required to have a higher capacity of loss absorbency by means of more capital

requirements, liquidity charges and ruthless supervision. We have also mentioned

other supplementary measures including higher requirements for exposures on the

financial sector, derivatives, securitisations and off-balance sheet activities, incentives

to resort to central counterparties for OTC derivatives, and higher liquidity

requirements if banks tend to fend their long-term assets by interbank (short term)

funding.

Pillar 3 refers to the transparency of capital and market discipline, according to which

banks are forced 1) to reconcile all capital regulatory elements in the audited financial

statements, 2) to disclose any items not deducted from Tier 1 equity capital, 3) to

encompass all relevant limits and minima set out by the committee, 4) to describe

features of the issued capital instruments and 5) to disclose explicit explanation on

how certain ratios of regulatory capital are estimated. All requirements are

accompanied with terms and conditions of the whole set of instruments of the

regulatory capital to be published on banks‟ websites.

Apart from the three-pillars approach extending that of Basel II, the Committee deems

strong liquidity base of equal importance with capital. The financial crisis brought out

how management overlooked the resilience of banks when liquidity was fading out

and this shortage has been long lasting. On these grounds, alongside principles of

better risk management and supervisory guidance on their effective implementation,

the Committee introduced complementary objectives. First, the liquidity coverage

ratio (LCR) aims at making banks resilient to short-term liquidity risk and second, net

stable fund ratio (NSFR) employs a longer horizon (one year) in order to incentivize

banks to use stable fund resources for their activities. For harmonisation purposes, the

Committee has introduced a minimum set of information for supervisors so as to

monitor contractual maturity mismatch, funding concentration, unencumbered assets,

LCR by currency and market-based tools like asset prices, spreads of credit default

swaps (CDS) and ability for wholesale funding.

46

2.3.1.4. Free banking

An alternative thought would be the dismantling of bank regulation, the laisser-faire

of bank conduct. As already experienced in Canada, Scotland, Sweden and

Switzerland with success, high degree of competition drives innovation and profits by

means of establishing branches and subsidiaries, loaning-out facilities, among others.

Free banking is an ill-defined notion, devised to depict the operation of a competitive

banking market without the regulatory counterpart of a central bank. Deposit

insurance and lender-of-last-resort schemes could be provided by private institutions

(banks, clearing houses) to secure themselves from bank-runs. Under such conditions,

market concentration is favoured when bigger banks are free to impose great charges

on payment and settlement services. However, moral hazard problems are still

possible in that case, since emergency way-outs undermine monitoring of bank‟s

activities and managers‟ gambling behaviour. From a macroeconomic perspective, the

excessive issuance of bank notes may contaminate the stability of the financial

system.

In fact, if those notes are – or perceived to be – perfect substitutes, smaller banks have

an incentive to issue demand liabilities (wildcat banking). Given that informational

asymmetries and mistrust do exist concerning banks‟ solvency and assets,

noteholders, in fear of not redeeming at par, are likely to redeem in specie at larger

banks and affect a run on their fractional reserves. Nelder (2003) supported the over-

issuance of notes by small banks since it is highly likely to be redeemable by bigger

Swiss banks. They all turned out to depreciate franc and approve the establishment of

central bank, as the institution with the jurisdiction of notes issue. An alternative

explanation is given by Rockoff (1975): If it is easy for everyone to set up a bank, just

by depositing a government bond or mortgage in order to issue notes, the way these

notes are valued in conjunction with bonds determines the probability of financial

fragility. In particular, in case of having bonds valued at par and notes depicting the

same value, when the market value is (temporarily) below face value, the „banker‟ has

an incentive to flee with a „loan‟ to himself and bank closure. In addition, Rolnick and

Webber (1984) give credit to the declining value of bonds rather than the wildcatting

47

hypothesis1 as the general constituent of bank failures in free-banking periods of US

states.

However, considerable failures in free banking era should drive further research to

address potential endogeneity in banks‟ operations, failure of bank regulation to cope

with various bank specificities and policy implications of those failures to the recent

deregulating trend (Rolnick and Webber, 1983). Notwithstanding the importance of

such an issue, modern practice is directed to the advancement of international

regulation, setting such an „experiment‟ out of the westernized frame.

2.3.2. Integration

By and large, financial markets have become integrated in the aftermath of the

adoption of the euro. The fast converging bond and equity markets seem to have

outreached the banking industry, which is falling behind with considerable time lag

(Gropp and Kashyap, 2009). Cross-border integration can be expressed in three forms

according to Walkner and Raes (2005), the organic growth of branches and

subsidiaries, M&As and the provision of banking services.

The advent of Euro as a single currency amongst the European countries has already

eliminated the exchange rate risk and transaction costs induced by currency

conversion in arbitrage practices. Cultural and language differences as well as

consumer preferences may also jeopardize the law of one price, or alternatively, the

complete levelling of the prices of assets and financial services (interest rates). Under

free capital flows, a complete convergence may stick at informational asymmetries in

cases where incumbent national banks have a comparative advantage over foreign

entries about clients; the latter is mainly present in relationship-based operations.

Impediments to this ongoing process may be judicial systems that encourage banks to

price loans heavily to customers, who run the risk of going default, to make up for

potential recovery costs. Besides, the fiscal conditions - that is taxes and subsidies -

1 During the period 1816-1863 in US, the only form of banking was degenerated to some dubiously

sound banks being subject to national and not federal regulation and operating in distant places where

only wild cats can thrive (Luckett, 1980). These banks issued private notes circulated at par and backed

by dubious securities (bonds, mortgages, etc). The fact that these notes were redeemable only on those

remote banks caused abuses and swindles by bankers insofar as the National bank act came in 1863 to

forbid „wildcat‟ currency.

48

are likely to directly affect the cost of capital and the relative corporate finance

decisions. Barro and Sala-i-Martin (1992) have put forward the notion of conditional

convergence which simply highlights the tendency of each economy (and interest

rates, too) to reach a different steady state, on the grounds there exist such

macroeconomic conditions that enable special microeconomic fundamentals to have

an effect on market prices.

Local too-big-to-fail champions are also favoured by political interference along with

labour market rigidities and lack of economic diversification hurdling cross-border

mergers in many cases. Integration seems to be high in wholesale banking

(underwriting, rating, trading, brokerage, M&A) and low in retail banking since long-

term relationship business has rendered the proximity to clients and informational

advantage a strong competitive edge (Barros et al., 2005; Fernandez et al., 2007).

Cross-border activity can be intensified by the deployment of branches and/or

subsidiaries in other European countries. Alternatively, banks may join strategic

partnerships with institutions or place particular operations in other states. Structural

break is placed on the date of adopting Euro as a single currency, while developed

countries like France, Germany, Spain and Netherlands have witnessed an increase in

their equity market interdependence, followed closely by United Kingdom and

Sweden. Gropp and Kashyap (2009) found that unlisted, cooperative and savings

banks seem to have no convergence point, constituting thereby a detrimental

impediment of integration.

However, integration might be problematic in terms of diversification. Assuming a

fully integrated area, where business cycles are totally synchronized, there is no room

for portfolio diversification. The per se fact that cross-border activity is hindered by

various barriers in euro area, eliminates any systemic risk potentially emanated by

bank failures and crises. In addition, bank-lending behaviour has demonstrated pro-

cyclical behaviour, in that during economic upturns credit is provided excessively,

followed by stringent loan granting in recessions. In any case, the change in bank

lending is proportionally greater than that of the business cycle exacerbating the

economic conditions either way. Some possible explanations that have been tested in

the empirical literature are informational asymmetries, herding behaviour,

institutional memory, agency problems, etc.

49

Therefore, regulation is deemed to nationalize credit when growth is escalating in

order to alleviate potential loan losses, and boost the economy when it is creeping at

the rock bottom. It is actually debatable whether the regulator of the interbank activity

should be either the home or the host country. On the one hand, the home country

might drive to conflicts of interest to shut down a branch or a subsidiary in another

country as the latter is bound to suffer the economic repercussions of such an event,

whereas in some cases a small country cannot afford to bail out problematic

international bank groups. Fragmentation especially in retail banking defers the

necessity of a supervisory and regulatory body whereby European interbank activity

comes under; what is flickering in the nigh future is a consolidated entity that will

take over in cases of inefficient national actions.

2.3.3. Financial stability

Banks may be a significant channel-through of instability in the economy once

considering liquidity and credit pause drying out payment services and interbank

lending. From a policy perspective, a significant part of the banking literature

investigates the fact whether competition drives to either financial stability or

fragility.

Keeley (1990) showed the franchise value paradigm, that if the market power of

banks, along with profit margins, is diminishing in response to greater competitive

pressures, banks in order to recoup increased returns tend to take on risky projects.

Bank failures are likely to occur when adverse selection and moral hazard make loan

portfolios comprise marginal applicants and thus exacerbate the risk exposure.

However, another severe source of instability is traced at deposit competition; thanks

to the ongoing deregulation of financial markets, interest rates have slumped on the

sly ever since the endeavour of removing entry barriers and expanding restrictions

flourished. In such a situation, banks strive to curb low franchise value and

profitability engaging in riskier asset allocation given that in hard times of insolvency

and banks runs, deposit insurance schemes are stand by to intervene. Hence, it is

deemed to be essential for the authorities to impose restrictions on deposit

competition to discourage „gambling for resurrection‟ (Cole et al., 1995).

50

The other strand of literature contending the competition-stability nexus, emanates

from Stiglitz and Weiss (1981) who showed that monopolistic market structures are to

be blamed for great charges on loans and thereby upcoming defaults. Safe borrowers

are repelled by high borrowing costs and information asymmetries render a significant

part of loans nonperforming and hence constituent of financial instability. Boyd and

De Nikolo (2005) employed the loan market channel in their analysis to conclude that

the positive relationship between risk and competition is fragile. As monopolistic

structures increase loan rates, borrowers surrender to riskier projects. Thus, the

probability of default rates is conditional on banks‟ pricing conduct in the loan

markets. Almost the same applies to Boyd et al. (2006), in which a positive

relationship between competition and financial stability as well as, ceteris paribus,

bank willingness to lend was found after running different model specifications.

Matutes and Vives (1996) put forward a model of unregulated bank competition.

They consider the self-fulfilling expectations of depositors to endogenously affect the

quality – or the failure probability - of a bank. In other words, the bank, which enjoys

depositors‟ trust, will enjoy higher margins and greater market share as perceived to

be diversified and hence safer in their eyes. In addition, they examine the welfare

implications of deposit insurance equilibria: notwithstanding the positive impact of

insurance in preventing crises, mitigating transport costs and extending the market,

deposit insurance guarantee that all banks are credible. Hence, in the absence of no

expected diversification gains to exploit, all banks are discounted at the same rate and

the resulting higher competition hits high failure probabilities. If decisive regulatory

authorities allow for takeovers of the failed banks, banks‟ assets remain put

contributing further to financial stability (Perotti and Suarez, 2000; Nagarajan and

Sealey, 1995).

That is not the case for the model of mean shifting investment technologies by

Koskela and Stenbacka (2000), since higher competition diminishes the loan rates

without necessarily triggering default risk in equilibrium out of the increased

demanded volume of investments. Caminal and Matutes (2002) found the same

pattern, that monopolistic markets bearing the costs of monitoring tend to be more

susceptible to risky loans and thereby subsequent failures. Even Schaeck et al. (2009)

after applying different sets of samples, econometric methodologies and periods, and

51

controlling for macroeconomic factors and concentration, the competition measure

(H-statistic) was persistently related to bank stability.

On the other hand, Allen and Gale (2004) put forward a number of theoretical models

of competition and financial stability trying to shed light on the multifaceted

underlying relationship. They include aspects like spatial and Schumpeterian

competition, agency costs, financial intermediaries and contagion to highlight the

efficient levels of potential trade-off between competition and stability. It is Pareto

optimal, though socially undesirable, to have instability in cases of a) perfect

competition and complete markets, b) present agency problems due to the incentive to

acquire greater market share and „last bank standing effect‟ (Perotti and Suarez,

2002)2 and c) many banks occupying the same locations and lack of innovation

(Schumpeterian competition). Contagion might well be an outcome triggered by the

systemic risk of an aggregate shock on liquidity only where competitive interbank

market leaves price takers off to liquidate their assets.

Along the lines, Boyd et al. (2004) underscore inflation as another key determinant of

bank failure irrespectively to the underlying banking system. When the nominal rate

of interest (inflation) is below a certain threshold, a relatively higher probability of

bank failure is present in monopolies on the grounds that the incentive of loaning out

cash reserves dominates that of paying low rates on deposit accounts. Secondly, asset

loss is greater in competitive structures in times of a crisis, as monopolies tend to

make profits upon the liquidation of assets (e.g. deposits) except cash, for they are

able to provide inter-temporally much lower deposit insurance.

Matutes and Vives (2000), in an imperfect competition model, show that deposit

limits can produce some welfare gains in an uninsured competitive market when the

social cost of failure is prohibitive. In the presence of risk-based insurance on

deposits, the prudent behaviour of banks can be restored improving further social

welfare. In the same vein, Hellmann et al. (2000) allege that the imposition of capital

requirements may have adverse effects on decreasing franchise values inducing banks

to embark on gambling behaviour. However, deposit-rate controls are more Pareto-

efficient instruments that promote prudency in investments even off the equilibrium

2 Supporting the competition-stability literature, it concedes the prudent bank behavior as precipitated

by the probability of the other market players be hit by random solvency shock. At the end, duopolies

turn out to produce great monopoly rents.

52

path. Even when there exist only large institutions in the market, supervisory

authorities should keep a vigilant eye on their prudent operation while the limited role

of deposit insurance schemes should not undermine their per se existence along with

government safety net in times of emergency (Mishkin, 1999).

Martinez-Miera and Repullo (2010) concur that there exist a U-shaped relationship

between competition and bank failure risk. In particular, monopolistic markets

experience the risk-shifting effect, that is more competition with low loan rates

stabilize banks as they run less risk of default whereas the margin effect – lower

revenues of total non-defaulting loans may jeopardize banks in view of potential

entries – occurs usually in competitive markets. Empirical studies of the UK banking

market show less competitive market structures to be stable during the period 1840

and 1940 (Capie, 1995) and sustainably profitable in the past few decades as

compared to the variable performance of German banks. Last, Laeven and Levine

(2009) empirically investigate how risk taking is interconnected with ownership

structures and national regulations. In fact, more powerful owners and equity holders

are willing to engage in riskier investments than debt holders and non-shareholding

managers.

1.4. Conclusion

We reviewed the theoretical underpinnings of competitive issues in European

banking. Academic research has had a particular interest in the structural analysis of

competition along with the pricing conduct of financial institutions since the „80s in

order to provide evidence with fruitful implications for antitrust policies and

institutional reforming of the whole financial system. Hence, to the best interest of

social welfare, bank competition constitutes an indispensable issue closely related to

economic development and financial stability.

53

CHAPTER 3

Empirical models of competition in the European banking

55

3.1. Introduction

The review is going to cover the structural approach, whose focal point is the

description of all the extensively applied concentration ratios, discussing their

theoretical characteristics and relative strengths or weaknesses when it comes to apply

and assess them empirically. Next step is to build up a link between structural changes

and bank performance, mainly on the grounds of the Relative Efficiency (RE) and the

Structure-Conduct-Performance paradigm (SCP). Non-structural measures developed

in response to the deficiencies of structural models to quantify bank competition

based on monopoly power measures - the New Empirical Industrial Organization

(NEIO) approach. Another case may be the degree of contestability in a bank market;

that is, few banks or only one implement competitive pricing in order to second its

monopoly power.

Due to the limited capacity of SCP and relative efficiency, Heffernan (2002) made a

step further by proposing a general linear model of competitive pricing of the

important retail banking products: deposits, loans, credit cards and mortgages. Further

insights will be discussed regarding the circumstances under which competition

measures can be significantly differentiated in cross-country and within-country

comparisons. Accounting for all possible reasons hidden thereof, a new indicator of

competition is proposed when employing particular error-correction specifications

(Carbo et al. 2009). Last, Leuvensteijn et al. (2011) first applied the Boone indicator

in the banking industry to see whether efficiency drives profitability in terms of

market shares.

3.2. Non-formal structural methods

3.2.1. The Structure – Conduct – Performance (SCP) paradigm - The relative

efficiency (RE) hypothesis

The Structure-Conduct-Performance model is an empirical model developed after the

Second World War for industrial economics. It is based on empirical studies of

manufacturing industries during 1940s and „50s when monopolistic conditions were

56

quite evident (high fixed costs, firm entry and few competitors). However, this

method lacks any theoretical gratification capturing the idea that a change in the

market structure affects the behaviour and performance of banks. They are willing to

operate inefficiently by not minimising their relative costs as the market is getting

more concentrated.

The so-called collusion hypothesis relates high concentration with reduced costs of

collusion resulting in high prices and profits. Sketching out the model in more

concrete way, structure is determined by the interaction of supply and demand in a

particular industry (Reid, 1987; Scherer and Ross, 1990), conduct is the pricing

behaviour of banks as determined by the cost structure, barriers to entry and the

number of buyers-sellers and performance is equivalent to the profitability measured

by the bank‟s conduct.

The causal relationship that may replicate the reality goes from structure through

prices to profitability proxies like higher profits. The underlying links address the

common practice of oligopolies where banks achieve higher margins by offering

lower deposit rates and higher loan rates. Unlike the monopolistic framework of

Cournot and Dixit-Stiglitz, there is an implicit assumption that monopolistic

competition is exogenously formulated to bar potential entries. Since there exist high

barriers out there, it is wise to give emphasis on the SCP paradigm.

However, it has been critisized by (e.g) Bos (2002), Gilbert (1984) and Vesala (1995)

especially for the one-way causality hidden thereof and the failure of capturing the

recent development of industrial organisation. The component of „conduct‟ has been

overlooked in the empirical literature showing explicit interest in the S – P

relationship.

The Relative Efficiency Hypothesis was developed by Demsetz (1973) and Peltzman

(1979) to give another interpretation of the profitability of banks; high profits are

attributed to the efficient operation, which predicates the higher produced output or

higher capabilities of undergoing low prices. The latter is exactly the opposite of what

SCP surmise about the positive correlation between concentration and prices. Hence,

the degree of efficiency determines the profit maximising conduct (high output/low

prices – Molyneux and Forbes, 1995) and through the increased market share to high

performance (profits).

57

Conjectural sources of efficiency may include a production technology or a better

management much to the X-efficiency of lower costs. High profit margins per unit of

output materialize in the presence of product diversification and/or scale economies,

which are transmuted into higher, market shares (concentration). In this regard,

Vesala (1995) opines that the market structure is endogenously determined by the

performance of more efficient banks.

It is evident how important the hypothesis employed to conduct policymaking is. The

SCM model signifies the imperative of intervening to reduce the market power by

giving incentives to new firms to enter the market. Alternatively, regulatory measures

may be taken to bound the prices such as setting out explicitly how much deposit

(loan) rates can go below (above) the central bank rate. Confirmation of the relative

efficiency gives way to bank consolidation through mergers and acquisitions.

However, particular emphasis should be placed on whether a merger promotes

relative efficiency or just amplifies concentration; the latter case must be forbidden by

all means.

To test the validity of both SCP and relative efficiency hypotheses, a general function

of estimated equations is presented to best capture the vast published work, e.g.

Molyneux et al. (1996), Berger (1995) and Smirlock (1985). Schematically,

Pi,t a0 a1C j,t a2MSi,t ak2Xi, j,tk

k

(3.1)

where

Pi,t : measure of performance of bank i at time t (profits or price)

Cj,t : market structure of the market j at time t, proxied by a concentration measure

MSi,t : market share of bank i at time t

Xi, j,tk : vector of control variables including bank-specific and country-specific

features. Some examples are market demand (per capita income, immigration,

population density), variables to reflect cost and size-reflected differentials (scale

58

economies), various market risks (Molyneux et al., 1996) and state ownership of

banks (Lloeyd-Williams et al., 1994).

The SCP model, reminiscent of Molyneux and Forbes (1996), is verified once we see

only a statistically significant α1 coefficient (α2 = 0), while for the relative efficiency

the exactly opposite must hold. Berger and Hannan (1992) regressed the deposit

interest rate on concentration ratios and various control variables to verify directly

whether SCP and relative hypothesis hold. In this case, the negative sign of the

relative beta coefficient of concentration endorses SCP power. In contrast, if the

dependent variable were the loan rate, SCP hypothesis would be in line with positive

coefficient of the concentration variable.

Alternatively, Berger (1995) elaborated the previous model with measures of X-

efficiency and economies of scale, testing four hypotheses: the market power (MP)

hypotheses, those of SCP and relative market power, and efficient structure (ES)

hypotheses, that is the relative X-efficiency and relative scale efficiency. Hence, SCP

hypothesis would require positive relationship between profits and concentration

ratios, the relative market power positive correlation between profits and market

share, whereas any measure of ES hypotheses prerequisites not only their statistical

significance with respect to profits in the initial model but also vis-à-vis both

concentration and market share variables in separate reduced form equations. That

idea was quite interesting since in previous models, simultaneity problems due to the

relationship of concentration and market share variables biased the respective

coefficients.

Corvoisier and Gropp (2002), employing a Cournot model, regressed the differences

of money market rates from those of retail interest rates for each bank and country

(margins) on concentration (HHI), probability of borrower‟s default, operating costs

(proxied by average cost to income ratio), aggregate demand for loans (proxy:

confidence indices), market capitalisation in each country (proxy: assets to GDP ratio)

and indicator dummies to underline whether the HHI measure is used or not. The

relevant equation was tested for a pool of loan and deposit products.

In general, special attention should be drawn to the different measures of

performance, namely price and profitability proxies. Heffernan (2005) illustrates a

summary of them as occurring in a bulk of published work:

59

Price proxies

Loan rates – interest rates on (e.g.) personal/business loans or residential

mortgages.

Deposit rates – interest rates on term/deposit savings or money market

accounts

Bank service charges levied on a current or standard account

Profitability measures

Return on assets (ROA) – net income/total assets

Return on capital (ROC) – net income/capital

Return on equity (ROE) – net income/stockholder‟s equity

The first group of proxies employs either the average price or the price of particular

products and services (business loans). Even if data on defined product markets is

employed for good, it is usually the case of potential cross-subsidization in multi-

product kind of banks that blur the validity of the overall performance results.

Relevant product markets constitute an interesting issue, as well. The second group of

measures is extensively used in the most recent literature since it groups the profits

and losses of multi-product activities. Some criticisms have been raised here by

Molyneux et al. (1996) and Vesala (1995), who opined for the shortcoming of

profitability measurement upon the basis of both stock and flow variables and for the

likely negative relationship of market power and profitability when banks suffer from

operational inefficiencies. Moreover, the aggregation of profit measures takes no

account of the market power in individual markets.

60

Before delving into the reservations of the underlying models, a brief summary of the

indicators usually met to proxy the market structure should be considered. In various

empirical studies, there are four categories of proxies according to Bikker and Haaf

(2002); Concentration – HHI, k ratio, number of banks, other measures - Market

share – though mostly used to capture bank efficiency - Entry barriers3- Regulatory

barriers: bank charters/branches with the conditions involved, Non-regulatory

barriers: minimum efficiency size, technology, scale economies, product

differentiation – and Number of branches.

It has been argued against the aforementioned models that bank performance is ill-

proxied or poorly explained by a great number of industry and bank-specific

variables; that explains the vague results in favour of either hypothesis (Cetorelli

(2004); Cetorelli and Gamberra (2001) and Beck et al. (2003). The US literature finds

persistence evidence lately that the relative efficiency holds, whereas in European

banking structural factors tend to determine significantly competitive behaviour

(Goddard et al., 2001).

Carletti (2006) argues that financial stability is a new issue closely related to the

collusion hypothesis. Stability is more at stake if monopolistic conditions do not lead

to higher spreads and profitability and banks do not take on risky investments to

imperil their positive market valuations (Hellman et al., 2000). Boyd and De Nikolo

(2005) alternatively argue that borrowers may bear great risks to cope with the higher

interest payment service. Thus, the default risk of borrowers jeopardize the banks‟

loan portfolio; that underlines a disadvantage of SCP paradigm when empirical

studies give more emphasis on structure and much less on conduct.

3.2.2. Formal structural approaches

Martin (1993) has summarized the studies that attempted to formalize the underlying

analysis with the use of certain profitability measures; in most cases, they generalized

the Lerner index of monopoly power. The following framework relies on the premises

3 Vesala (1995) attributes the correlation between concentration and profitability to the dual positive

correlation of regulation with both concentration and profitability.

61

of industrial organizational theory, replicating the derivations of profit maximization

problems made by Cowling (1976) and Cowling and Waterson (1976). Suppose now

there are n banks of different size in the market producing a homogenous product. To

capture this inequality, banks have different cost functions ci xi included in the

profit function:

i pxi ci xi FCi (3.2)

where

πi : profit

xi : the volume of output

ci : the variable costs

FCi : the fixed costs of bank i and

p p X p x1 x2 ... xn (3.3)

The profit maximization of bank i yields the first order condition:

ixi

p p X dX / dxi xi ci xi 0,i 1,2,...,n (3.4)

or

p p X 1 i xi ci xi 0 (3.5)

given that i d x j / dxij i

n

is defined to be the conjectural variation of bank i with

respect to the other banks j i in the market. In other words, it shows how the

market output will adjust in response to a change of the bank i. Cowling argues that

this concept incorporates expectations on potential entries and the behaviour of

62

market rivals if the banks‟ reaction functions are continuous to allow for the

parameterisation of any static/dynamic equilibrium though differentiation in various

market forms.

In case of perfect competition, no change in the market output and price will be

resulted by any marginal change of a bank, so dX / dxi 0 1 i , and thus i 1 .

In a Cournot oligopoly, a bank operates independently since it expects no retaliation

from other banks; that is, market share will increase by the same amount:

dX / dxi 11 i , with i 0 . In perfect collusion, all banks intend to preserve

their market share in view of any potential change of bank i‟s output. Hence, banks

will increase their output such that the ratio X / x will remain put for any unit change

of bank i. The conjectural variation will become i X xi / xi x j xij i

n

.

3.2.2.1. The HHI in a S – P model

Multiplying the equation (3.5) with xi and X2 X 2

, as well as summing all the banks‟

yields,

pxi p X dX / dxi i1

n

i1

n

xi

2

X2

X2 c xi

i1

n

xi 0 (3.6)

Then, we divide both sides by pX,

pxi c xi xi pX xi

2

X 2

p X X 2

pX dX / dxi

i1

n

i1

n

and multiply with dxi / dxi the right-hand side:

63

pxi c xi xi pXi1

n

si2 dpX2

dXdxi pX

i1

n

dXdxi dxi

where si2

xi2

X 2

pxi c xi xi i1

n

/ pX si2

i1

n

1 d x j / dxiji

n

dpX 2dxi

dXdxi pX

where dX / dxi 1 d x j dxiji

n

pxi c xi xi / pX si2

i1

n

i1

n

dxi

dxi d x j / dxi

ji

n

dpX

dXp

pxi c xi xi i1

n

/ pX si2

i1

n

dxi

dxi d x j / dxi

ji

n

n

D

as nD dXp / dpX

pxi c xi xi / pX 1 i1

n

HHI / nD (3.7)

where

64

ixi2 / xi

2

i1

n

i1

n

. The last equation expresses the average price-cost margins in

terms of the elasticity of demand (nD), the Herfindahl index (HHI) and conjectural

variation term (γ). It is in line with the SCP framework, for the high degree of

concentration will ensue great price-cost margins; HHI seems to be appropriate in the

underlying case so long as the γ term is known and the same for all banks.

Dickson (1981) proposed the final expression of the average price-cost margin to be

reparameterized in terms of the conjectural variation elasticity (ti), namely the

proportional change of market output as expected by bank i:

ti dX

dxi

xi

X (3.8)

This new insight is truly welcome since expectations on the degree of banks‟

retaliation illuminate the market structure, let alone it is theoretically derived that

concentration measures can determine price-cost margins. Alternatively, it is

redefined as:

A sitii1

n

, with ti si 1 i (3.9)

Hence, siti si2 1 i

i1

n

i1

n

HHI HHI i 1 HHIi1

n

(3.10)

Suchlike the parameter μi of conjectural variation, the elasticity term ti take values

according to specific market structure. In a Cournot oligopoly,

ti dX

dxi

xi

X 1

xi

X si , while ti

dX

dxi

xi

X 0

xi

X 0 in perfect competition. Last, the

perfect collusion case requires ti dX

dxi

xi

XdX

XX

dX 1 .

65

3.2.2.2. The CRk in a S - P model

Suppose in a market there are n banks, out of which k banks are involved in a cartel.

Hence, n-k banks constitute the competitive fringe of price takers and maximize their

profits when operating at levels where prices are equal to the marginal cost

[competitive condition: p MCi , k 1 ,...,n ]. The supply of the n-k banks is

defined to be the reciprocal of their cost function, i.e. ci1 p ; the aggregate supply is

therefore Snk p Ci1 p

ik1

n

. The market demand is DN p with DN p 0

but what portion of it is faced by the cartel is derived from:

Dk p DN p Snk p , where Dk p 0 and Sn k p 0 . Taking the derivatives

of both sides with respect to p and dividing them with Dk p / p , we have:

Dk p p

Dk p DN p

p

Dk p Sn k p

p

Dk p (3.11)

nDk nDNDN p Dk p

nSnkSn k p Dk p

(3.12)

The k-bank cartel maximizes its profits according to the letter of Lerner index. In

particular, it sets the P-MC margin equal to the reciprocal of the elasticity of its

demand.

p c j

p

1

nDk

1

nDNDN p Dk p

nSnkSnk p Dk p

Ck

nDN nSnk 1Ck

where

66

DN p Dk p

1

Ck and

Sn k p Dk p

DN p Dk p

Ck

1

Ck1

1 Ck

Ck (3.13)

It is finally justified to include the CRk concentration ratio in the empirical

applications of S – P model, given that the market is dominated by a cartel of k banks.

3.2.3. Concentration ratios

Concentration ratios capture the structural features of a market. They are used in

models or interpreted in conjunction with other performance measures to explain the

competitive behaviour of a specific industry without sufficing stand-alone to

extrapolate competitive conditions. Even changes in concentration can be deduced

regarding market entries and exits, a feature widely used in U.S for anti-trust

purposes.

The literature of industrial concentration has proposed a bulk of measures, all

showing a final convergence towards the inclusion of two elements: number of banks

and the size of them. The general form (concentration indices) of all the measures to

be presented hereinafter is captured by:

CI SiWi

i1

n

(3.14)

where

Si: the market share of each bank i

Wi: the weight attached to each share

n: the number of banks

67

Marfels (1971) and Dickson (1981) made a theoretical classification of ten

concentration measures according to their weighting scheme and structure; in some

cases, some ratios are consistent with more than one weighting schemes or with none.

The weighting scheme refers to the sensitivity of each measure to changes in the tails

of a bank distribution.

I. To an arbitrary k number of banks, a weight of unity (i k) is attached,

while to the rest a weight of zero (i > k) . Example: K concentration ratio.

II. The market shares are properly weighted by the per se shares (Wi Si ,i) .

All banks are employed in this case with larger banks getting larger shares.

Example: Herfindahl-Hirschman Index (HHI).

III. The weights of all banks take the form of the ranking number of banks either

in ascending or descending order (Wi i,i) . Examples: Rosenbluth index

and Hall-Tideman index.

IV. Although Marfels used as weights those of a weighted geometric mean, Bikker

and Haaf (2002) proposed the negative of the market share logarithm

(Wi logSi ,i) . Thus, larger banks get a smaller absolute weight;

examples: Entropy index.

The relative structure of its concentration ratio is discrete or cumulative. The former

refers to the level of the concentration curve at a conjectural point (recall k

concentration ratio). The concentration curve measures the cumulative market share

(vertical axis) against the bank rankings in the market (horizontal axis). The curve

increases from left to right with a diminishing rate reaching the level of 100 per cent

of the market share at the point (x) where all banks of the market are included. In the

following graph, 85% of the market is held by 10 banks while the other 15% by the

remaining 10 banks.

68

Graph 1: The concentration curve

100

75

50

0

10 20

Advantages of this category are the simple limited data needed, as those markets

dominated by few large banks will change marginally by all the rest institutions.

Concerns against it have been raised stressing a severe disadvantage of discrete

indexes: they cannot capture structural changes in the number of banks which are out

of the index, overlooking the possibility that the competitive behaviour of small banks

may force the dominant ones to act appropriately.

Cumulative measures of concentration examine the whole bank distribution, capturing

all the structural changes occurred in every part of it. Some examples are HHI, CCI,

RI and E. Juxtaposed to them, discrete measures cannot easily identify the changes

precipitated either by the changing number or by the size disparity.

3.2.3.1. The k bank concentration ratio

It is the most frequently used in the empirical literature. The following general form

simply sums the market shares of an arbitrary number of large banks, treating equally

the k leading banks. Small banks are out of the frame, as the ratio is one-dimensional

measure taking values from zero to unity. In the case of zero, we have an infinite

number of equally sized banks whereas unity suggests that the chosen banks make up

for the whole market.

69

CRk Sii1

k

(3.15)

Assuming that a market comprises n banks of equal size then,

CRk 1n kn ne

kCRki1

k

(3.16)

Ne denotes the Numbers-equivalent as proposed by Adelman (1951) and White (1982)

later on. It simply transforms the concentration ratio to a number of equally sized

banks, which represent a level of concentration similar to the ratio.

3.2.3.2. Herfindahl - Hirschman Index (HHI)

It is widely known in the theoretic literature and in politics as a usual benchmark of

antitrust enforcement laws. Cetorelli (1999) gives some guidelines for the evaluation

of an imminent merger; it is approved if the post-merger market has a HHI ratio lower

than 0.18 so long as the change of the ratio compared with the pre-merger market is

0.02 at most. Alternatively called full-information index as incorporating all features

of the entire distribution, it has the following form:

HHI Si2

i1

n

(3.17)

The interesting feature of the index is that a greater weight is assigned to large banks

(squared market shares) and all small banks are included in the estimation; it avoids

arbitrary cut-offs and inertia to market share changes, as opposed to the CRk ratio that

70

finds it hard to resolve. It takes values ranging from 1/n and 1; the former illustrates a

market of equally sized banks and the latter monopolistic conditions. Davies (1979)

concludes that the more sensitive the index becomes the less the number of banks in

the market. Adelman (1969) and Kwoka (1985) rewrote the HHI in terms of the mean

𝑛 , the index ends up with the following form:

HHI s Si s 2

i1

n

or HHI 1n n 2 (3.18)

Observing the equation (2.5), the relationship of HHI with its both constituents is not

straightforward. Holding the number of banks constant, Adelman (1969) asserts that

the HHI increases with the variance, but the latter is a function of the per se number of

banks. On the top of this, HHI can take the same values with different combinations

of variance and bank number (Kwoka, 1985), without constituting necessarily a

blemish. In the same vein, Rhoades (1995) foresees the same HHI derived by great

inequality in market shares across different bank markets.

To derive the numbers equivalent of HHI, we again consider equal size in all banks:

HHI 1

n

i1

n

2

n1

n

2

1

n (3.19)

Recall that the value 1/n is the minimum value in the space of the standard HHI; thus,

HHI can be a measure of at least two different size distributions.

Hart (1975) followed a slightly different method for cases where the number of banks

was not exactly known. Surmising that the only available information is the banking

market size and its classification, he proposes to split up the distribution of bank sizes

into classes and estimate the parameters of the original distribution by parameter

estimations of the second moment distribution (the distribution of bank size among

size classes); to do so, a relationship between the distributions is given. Hart (1975)

71

concludes that HHI is derived by n0

2 1

N, where n0

2 is the coefficient variation. The

latter is generally defined as:

c V / x (3.20)

where x the mean bank size and

V xi x 2

i1

n

n (3.21)

3.2.3.3. The Hall - Tideman Index (HTI) and Rosenbluth Index (RI)

The HTI and RI indices were proposed by Hall and Tideman (1967) and Rosenbluth

(1955), respectively. Hall and Tideman (1967) wrote down a set of rules that all

concentration ratios should satisfy, and only under such conditions they could accept

HHI. Their index assigns great absolute weights to the small banks, rendering it very

sensitive to small bank entries. It takes the following form:

HTI 1 2 isi 1i1

n

(3.22)

The market share of each bank is weighted by its order ranking, ranging from i =1 for

the largest banks and higher values (on) the more banks are listed. The index takes

value from zero to unity, namely from perfect competition case of infinitely equal-

sized banks, to monopoly. Alike HHI, in the case of equally sized banks, HTI gets the

value of 1/n. Thus, the numbers equivalent is derived to be ne 1 / HTI .

The Rosenbluth index is highly sensitive to small banks in the distribution, as it uses

the same weights with the HTI measure. It essentially depicts the distance (C) from

72

above the concentration curve to the horizontal line where the curve becomes

marginally tangent at 100 per cent market share. It takes the form:

HTI 1 2C (3.23)

Hause (1977) criticizes the usefulness of the RI measure in highly concentrated

markets, as it fails down to capture the real change in competition when displaying

high sensitivity in entries of small banks.

3.2.3.4. The comprehensive industrial concentration measure

This measure has been the product of a hot debate over the appropriability of

measures when capturing either concentration or dispersion in the banking sector. The

former takes account (only) the structural changes occurring in the largest banks while

the latter (e.g. the Lorenz curve and the Gini coefficient) estimates the dispersion of

the market undervaluing the share of the large firms. Horvath (1970) constructed a

new measure that captures and alleviates the deficiencies of earlier studies. It has the

following form, reminiscent a bit of the HHI:

CCI s1 si2 1 1 si

i2

n

(3.24)

where

s1: the market share of the leading bank and

si: the market share of the rest.

The most interesting thing about this measure is the absolute accentuation of the

leading bank and the relative dispersion of effect due to the smaller banks. It gets

73

values of unity if monopoly is the dominant market structure and values greater than

the absolute value of s1 the larger is the number of the remaining banks.

3.2.3.5 The Hannah and Kay index

The general form of the index as proposed by Hannah and Kay (1977) is the

following:

HKI sia

i1

n

1 1a

, a 0 and a 1 (3.25)

where the parameter a is the elasticity of concentration with respect to entries/exits of

the market and sales shifting through banks. The choice of what value this parameter

should take reflects the alternative and contradictious beliefs of the weighting scheme;

that is to be placed either on large banks (upper segment) or on smaller ones (low

segment of the distribution).

If, for instance, α tends to zero (0) the index gets closer to the number of banks in the

market, and if α goes to infinity the index amounts to the reciprocal of the largest

bank share. The numbers equivalent can be easily derived by solving for ne:

HKI ne 1 ne a

1 1a (3.26)

The effective average size of the banks within a market is defined to be the market

size divided by the numbers equivalent. The reduction in the concentration index will

be the greatest if a bank, whose size is equal to the effective average size (EAS), steps

in the market. If a bank of greater EAS enters, concentration will be less affected to

diminish; it might be the case of being reversed if the size effect outweighs the

number(s) (equivalent) effect. However, in cases of a bank expanding its size more

(less) than the EAS, the HKI of a given market will aggravate (abate).

74

3.2.3.6. The U index

Davies (1979) developed the index in response to the usual shortcoming of many

indices, which place overweight either on market inequality or the number of banks in

it. It takes the form:

U I an1 , where a 0 (3.27)

I: measure of inequality

n: number of banks and

α: the elasticity parameter which can be estimated after running the model i Ci

where πi and Ci are the price-cost margin and concentration in the market I,

respectively. Before doing so, U replaces Ci and logarithms transform the model into

the following form:

log i loga a log Ii logni ui loga 1 log Ii 2 logni ui (3.28)

where β1 = αβ and β2 = - β. Combining β1, β2 the parameter α can be derived by

1 / 2 . According to Davies‟ cogitation, U index is theoretically approached by

plugging in the Equation (3.27) the inequality defined to be a simple transformation of

the coefficient of variation (c2), 1 c2

. Given that the latter is estimated by

xi x 2

/ ni1

n

x 2

, it is straightforward to extrapolate the final expression of

inequality:

I xi

2

i1

n

nx 2

xi

2

i1

n

n

xii1

n

2

n2 n si

2

i1

n

(3.29)

75

since sii1

n

2

1

U n si2

i1

n

a

n1 na1 a si

2

i1

n

a

si sina1 a

i1

n

a

(3.30)

Davies attempted to investigate the sensitivity of U index to mergers and new bank

entries. As anti-trust policies base decision rules upon the well-known Herfindahl

index, the average effective size should equal to the industry size over the numbers

equivalent calculated from the HHI. The elasticity parameter a is proposed to be

bounded (∴≤ 1) allowing for the empirical studies to provide plausible results. In this

regard, the U index is increasing with a merger indicating greater concentration,

whereby the proportionate change is directly linked to the size of merging banks.

However, the sensitivity of the U index with respect to different kind of mergers is

prescribed by the high values of α apart from the large size of the merging banks.

If a bank enters the market, concentration will shrink, provided that its size in no more

than twice as much as the EASHHI; the recorded reduction can be maximized if the

bank size hits the exact average effective size. Small banks, on the other hand, will

result in minimal effects depending on the „sensitivity‟ values of α. The practical

justification of the restriction of α is nonetheless over against the theoretical

gratification of oligopoly. In particular oligopolistic incentives of mergers will not be

realized as long as α is sufficiently greater than unity. Hence, the only case of

delivering plausible and theoretically justifiable results is when the elasticity

parameter takes the value of one (1) – U index coincides with HHI.

3.2.3.7. The Hause indices

Hause (1977) proposed two industrial (Cournot) concentration indices that abide by

the six rules he thought important for a measure to be theoretically grounded. Inspired

by the Cournot model, his concern is over collusion effects (parameter α) in an

76

oligopolistic environment. He made numerical applications to conclude finally in the

advent of new banks n in the market, low values of α (high collusion) make the

transmission „creep‟ into higher degrees of competition. It takes the form:

Hm a, si si

2 si HHI si2

a

i1

n

(3.31)

where

HHI: the Herfindahl index

α: inversely related to the degree of collusion

The parameter α is then restricted for to secure the index to be a decreasing

convex function of the numbers equivalent. As α is getting closer to infinity, the index

approaches HHI. Hause went on identifying the range of the power of si given the

underlying restriction; the component si HHI si2

a

of the si exponent is always

less than unity, implying that the overall power exceeds unity.

However, the extent to which the index is rendered to be valid relies on the per se

values of α. If, say, the parameter α tends to infinity, the weights attached to every

single bank‟s market share collapse to those of HHI. The values of the Hause index

ranges from unity to 1 n 1 n1 n3

0 for monopoly and infinite number of equally

sized banks; the latter case seems to converge to Herfindahl‟s index minimum

(1 n 0 , if n ). Alternatively, Hause proposed the adjusted Cournot measure,

defined as:

Ha , si si2 si HHI si

2

a i1

n

, 1 (3.32)

77

The restriction guarantees the convergence of the index to the HHI as the number of

banks is escalating. If the market comprises banks of equal size, the Hα index gets

values according to the relationship: n1 n12 1 n1

.

3.2.3.8 Entropy measure

This measure lays its theoretical underpinnings in the ex-ante expectations of a

distribution‟s information content. Its typical form is:

si log2 sii1

n

(3.33)

It takes values from 0 to over and above unity since its maximum is log2 n . Lower

values of it indicate high monopolistic conditions, whereas values closer to its

maximum imply equal market shares and low concentration. In addition, White

(1982) observed the index falling as inequality exacerbates amongst a fixed number of

banks, and the weights attached to market shares decrease in absolute terms as these

shares become larger. Last, the numbers equivalent of the index gets the form

ne 2E .

Apart from the CRk and HHI, all the other measures have been rarely applied in

empirical studies. These two measures are used as proxies for the market

concentration in the so-called structural approaches of competition – the SCP

paradigm and the efficiency hypothesis. Not surprisingly, when it comes to use all the

aforementioned methods for the same market, the results are patchy and diverging

reflecting thereby the different weighting schemes. Policymakers assess the relative

impact of small/big banks each time, the size distribution or the number of them on

the market concentration so as to use certain indices that fit in the special features of

the market.

78

3.3. Non-structural methods

An alternative analysis of market competition can be directed towards the observation

of the pricing behaviour and the degree of market contestability, a notion developed

initially by Baumol (1982) and Baumol et al. (1982). To assess the predictive power

of SCP paradigm, non-structural indicators should be employed as the case under

examination is the „competing oligopolists‟. Lerner‟s (1934) measures of competition

constitute the ground upon which all the non-structural indicators have been

developed. In particular, the measures are Iwata (1974), Bresnahan (1982), Lau

(1982) and Panzar and Rosse (1987). The underlying empirical literature is known as

the New Empirical Industrial Organization (NEIO) approach, which sets out

behavioural equations concerning the price/output specification.

The measures have a common factor: they all stem from equilibrium states of a static

framework and signify price deviations from an ad hoc competitive benchmark.

Unobserved bank data cannot endorse bank conduct at the point where marginal

revenue equals marginal cost. However, an indirect measure could be to observe the

patterns of variation in the employed variables and then see how the pricing rule is

treated to justify them.

3.3.1. The New Empirical Industrial Organisation (NEIO) approach

The empirical literature of the New Empirical Industrial Organisation approach has

extensively applied the Bresnahan and Panzar and Rosse models. The Lerner index, as

applied in the Italian industry by Angelini and Cetorelli (2003), is defined as „the

relative mark-up of price over marginal cost‟. Schematically,

Lc n (3.34)

where

n: number of firms

79

ε: the elasticity of demand

ν: the conjectural variation of output (a bank‟s expectations about the industry output

to be responsive to a change of its own output).

Competition is escalating the lower the index. In that study, it was regressed on a pool

of independent variables (e.g. HHI, number of firms) to investigate sources of

potential mark-ups. Positive sign was found in the number coefficient and negative in

the concentration one. That finding implied the rejection of the SCP hypothesis and

was attributed to the increased efficiency gains once banks, facing market

contestability, were to go about strategic consolidation. However, contestability is at

odds with the positive and significant coefficient of number of firms whereas

simultaneity problems that pop up when HHI measure is constructed by the number of

firms, jeopardize the interpretation of the underlying results.

Bresnahan (1989) constructed an empirical framework of the market power

determination of the average bank, on the basis of the previous work of Breshanan

(1982) and Lau (1982). It turns out to be a straightforward transformation of Lerner

index, the mark-up of price/average revenue over marginal cost. The method includes

a parameter of conjectural variation: d x j / dxij i

n

/ n,0 1 , which is

determined by the simultaneous equations of market demand and supply. Recall that

dX / dxi 1 d x j / dxiji

n

is a different thing since under bank equality assumption,

μ is unique for the average bank. Shaffer (1993) argues that for profit maximization

purposes, banks set marginal cost equal to a perceived marginal revenue which

coincides both with the competitive demand price and the marginal revenue of the

market collusive extreme.

Tracking the extrapolation of the μ values in the previous section, in the perfect

competition case, 𝜇 is equal to zero so as to equate 𝑝 with the marginal cost. In the

Cournot oligopoly, dX / dxi 1 d x j / dxij i

n

1 , thus d x j / dxij i

n

0 and

80

k 1 n . Perfect collusion requires d x j / dxiji

n

X xi xi and in a final analysis

k 1 X xi / xi / n xi / nxi 1,i .

Banking markets may be contestable in cases where few banks or only one

implements competitive pricing in order to second its monopoly power. By setting

prices close/equal to marginal cost, there are no barriers to entry or exit for

newcomers. Hence, if incumbent firms raise their prices to earn abnormal profits, new

firms will hasten to reap the high margins. That highlights the „hit and run‟ entry and

exit at the time the former banks decide to lower the prices. In this manner, oligopoly

or even monopoly in banking market demonstrates a persistent competitive behaviour.

The potential of new coming banks to hit the market is also prescribed by further

market features such as whether the time lag of bank repricing is greater than that of

costumers to switch suppliers, whether the access to the employed technology or

factors is similar for all participants along with no sunk costs at all; the latter consists

in fixed costs that become irrevocable since a bank cannot sell it in a secondary

market when forced to step out of the market. Under such conditions, new banks entry

the market when (e.g.) incumbent firms price deposit (loan) rates lower (higher) than

in a perfectly competition state, and thereby capture a market share with lower prices

insofar as incumbent banks react by narrowing down the margins.

If a particular market is found to be contestable, the fear of entry implies the

maximisation of consumer surplus on the grounds that competitive pricing of

incumbent firms is present to discourage policy measures against abnormal profits.

Shaffer (1982) first argues as well as Nathan and Neave (1989) later on that Panzar –

Rosse (1982; 1987) statistic (PR) constructed until then for competition tests in

newspaper industries could be applied in the US and Canadian banking markets.

The aforementioned models of SCP, relative efficiency have been set aside as

suffering from many difficulties. Berger (1995) adduces their limited capacity of

explaining bank profits although a bulk of empirical studies employed them during

the last two decades and provided patchy results. Even the PR statistic has its

limitations stopping short at testing the degree of competition in a bank market.

81

Panzar and Rosse statistic, on the other hand, presupposes that banks will maximize

their profits at the point where marginal revenues are equal to marginal costs:

R xi ,n,ri Ci xi ,vi , zi 0 (3.35)

where

xi : output of bank i

n : number of banks

ri : a vector of exogenous factors that shifts the revenue function

vi : a vector of k factor input prices of bank i

zi : a vector of exogenous variables that shift the cost function.

In the non-profit state of market equilibrium, we end up with the equilibrium values

of specific variables (denoted by *):

Ri* x*,n*,ri Ci* x*,vi ,zi 0

The method measures the change in equilibrium revenues in response to a change

in factor input prices. PR constructed a measure of competition as being the sum of

elasticities of the reduced form of revenues with respect to the factor input prices.

H Ri

*

vkii1

n

vki

Ri*

(3.36)

82

For a given increase in factor prices, competitive markets are expected to experience

an equiproportionate H = 1 change in total revenue, whereas monopoly should

realize reduction in revenues − ∞ < H < 0 as higher costs increase price in the

elastic part of the market demand curve. Monopolistic competition should be the case

along the space 0,1.

Before making reference to the alternative models of competition, we should note the

Hall-Roeger methodology even if it is popular in applications of the manufacturing

industry. It is essentially developed by Hall (1988) and extended by Roeger (1995) in

the US industry. Hall (1988) argued that „the variation of the standard Solow residual

must be unaffected by the log-change of output in the change of monopoly‟ (Rezitis,

2010). Roeger (1995) showed the way we can get rid of unobservable productivity

shocks through the use of differences of Solow residuals that stem from production

and cost. If we assume that the banking market produces output Q through the

utilisation of labour l and capital k, the production function has the following form:

Qt t f (lt ,kt ) (3.37)

where θ denotes the term of Hicks-neutral productivity. The production-based Solow

residual (SR) is defined as the difference between the growth of output and input,

with the latter weighted by their contribution to the total value added. That is

theoretically grounded on the presumption of constant returns of scale, perfect

competition in the inputs market as well as imperfect competition in the products

market. Hence, SR is estimated by the following equation:

SR Qi,t

Qi,t i,t

li,t

li,t 1 i,t

ki,t

ki,t Li,t

Qi,t

Qi,tki,t

ki,t

1 Li,t

i,t

i,t (3.38)

where γ defines the cost of labour over total value added wtlt / ptQt , where wage

amounts to wt, and pt the output price. The L coefficient highlights the degree of

market power, which is effectively the price mark-up over marginal cost (the Lerner

index). However, when it comes to estimate the above equation, the productivity term

83

θ is correlated with the error term producing biased results. Roeger (1995) proposed a

resolution to the problem in the underlying estimation by obtaining from a cost

function the difference between the changes in weighted factor inputs and that in

prices:

DSR twi,t

wi,t 1 i,t

ci,t

ci,tpi,t

pi,t Li,t

pi,t

pi,tci,t

ci,t

1 Li,t

i,t

i,t (3.39)

where ct is the (user) cost of capital. We subtract equation (2.27) from equation (2.26)

to get rid of the productivity shock:

Qi,t

Qi,tpi,t

pi,t

i,t

li,t

li,twi,t

wi,y

1 i,t

ci,t

ci,twi,t

wi,t

Li,tQi,t

Qi,tpi,t

pi,t

ci,t

ci,twi,t

wi,t

(3.40)

3.4. Alternative models

Heffernan’s generalized linear pricing model proposed an alternative approach of

competitive pricing of the important retail banking products in UK market: deposits,

loans, credit cards and mortgages. The intuition behind is focusing on the factors

affecting the pricing setting behaviour in the retail banking sector.

Heffernan (2002) then tried to investigate how UK retail banking market price loans,

mortgages and credit cards. The proposed model is the following:

Rdi,t a0 jLibort j t iDi nt i,tj (3.41)

where

84

Rdit: gross domestic rate paid by firm i at time t

Libort-j, j=0, 1, 2, 3: monthly lags used on the London interbank offer rate

n: number of banks offering the same product

t: time trend

Di: dummy variable for each financial firm i (1,0)

For loans and mortgages, the dependent variable is Rlit, namely the annual percentage

rate imposed by firms i at time t:

Rli,t a0 jLibort j t iDi nt i,tj (3.42)

For the credit cards, the Rlit variable remains put but on the right-hand side of the

equation (5.2) the fit variable is added up to capture the fee charged by firms i at t:

Rli,t a0 jLibort j t iDi nt t fi,t i,tj (3.43)

The data4 of all variables were on monthly basis over a period of 1993-99 and the

underlying equations were estimated by running ordinary least squares. This method

was appropriate for pooled data set of banks but utterly inefficient when more data

available for each bank produced high orders of autocorrelation. To resolve the

problem, an autorgressive error regression model was employed to compute the

maximum likelihood (ML) estimators with more than 95% fit (R2) and no

autocorrelation evidence from the relevant Durbin – Watson test (DW).

4 For the complete explanation of how the representative account was determined along with the rate

quoted, see Heffernan (2002).

85

After running the econometric models and coming up with the results, a benchmark is

needed as a proxy of the competitive rate to compare the involved rates. Heffernan

proposed the rate quoted by each bank for its overnight transactions, that is the Libor

rate. It represents the opportunity cost of all bank‟s assets, the basis upon which the

marginal cost of all liabilities as well as the marginal revenue of all assets are

estimated. It is the best proxy of an international competitive level of all rates –

including deposit, loan, mortgage and credit card - especially when all banks have

access to it. In this regard, some pricing intertia with respect to current Libor leads to

the employment of a lagged rate by up to three months, as explicitly determined by

the monthly average of the daily three-month Libor.

The n variable is chosen to test for Cournot behaviour in case its coefficient is

positive (negative) in the deposit (loan) equations. It is also an implicit test for market

contestability since a significant coefficient of n variable throughout all equations

rejects the null hypothesis of contestability; that is the price setting is not sensitive to

the number of incumbent banks.

The dummy variable of each firm relies on the model of monopolistic competition as

developed by Salop and Stiglitz (1977). When market conditions are such (lower

fixed costs, increased demand) that favour an increase in the number of firms, it is

matter of information to actualize a good buy. Information costs are unforeseen while

some costumers know how prices are distributed across banks and some customers do

not. In the former case, low prices are accomplished (bargains) otherwise banks will

step out of the market; the latter case simply includes banks charging higher prices

(rip offs) exploiting the market ignorance of ill-informed customers. Hence, a bank

market incorporates firms, which make profits either by offering expensive products

as long as ignorant customers purchase them, or by selling a higher volume of cheaper

products.

In order to test the competitive behaviour of a bank, the dummy variable gets values

„1‟ for the rate values of a particular bank and „0‟ otherwise (for other banks). For a

certain product, the respective coefficient of all banks is then displayed in terms of a

default bank; in other words, the coefficient of the default bank is subtracted from all

banks‟ coefficients to construct a margin of deviations around which the relative best

bargain and the worst rip off can be spotted. It should be noted that the selection of a

86

default bank is firmly grounded on its available set of data of all under-investigation

products allowing for the major players as well as new players to be ranked. Had there

been another comparator as the default bank, neither the range of deviations nor the

relative rankings would have altered.

An interesting insight would be if the coefficient on the number of firms is juxtaposed

to that on the dummy variable of the Salop-Stiglitz model. The Cournot model is

compatible with the practice of bargains/rip-offs under certain circumstances; that is,

a rise in the number of firms might result in adopting the practice of lower prices as

determined by the relative effect of bargains and rip offs. In any case, the SS model

measures the monopolistic competition irrespectively of the number of firms that

formulates it. For example, the markets could be characterized as contestable if the

margins are low in tandem with the insignificant coefficient on the number of firms

(in the case of savings).

Hence, net interest margin seems to be appropriate to measure competition in

traditional banking services (deposit/loan) in Europe. In contrast, Lerner index and

ROA capture competition better in the general banking activity since they include

both non-interest returns (fee and off-balance sheet returns) and non-interest

(operating) cost.

Leuvensteijn et al. (2011) proposed the application of the Boone indicator as a new

measure of competition on the loan markets of the major European countries, Japan,

US and UK. It is differentiated, hence, to the other measures, in that its application

may lie in particular segments of the banking market and for several types of banks

without even requiring a great pool of data. It is the first measure that is used in the

banking sector, though firstly proposed and elaborated by Boone (2000; 2004) and

Boone et al. (2004). It measures the extent to which efficiency affects performance in

terms of market shares or profits, thus to be verifying the relative efficiency

hypothesis. It assumes that part of the efficiency gains are passed on to the clients

while its shortcoming is that the model does not control for qualitative features like

design and attractiveness of innovation.

Leuvensteijn et. al (2011) build upon one of the preceding theoretical models of

Boone (2000; 2001; 2004) and Boone et al. (2004). According to the latter, each bank

i produces output qi facing the following demand function:

87

p qi ,q j i a bqi d qiji

(3.44)

Each bank is facing constant marginal costs mci and we assume that a mci and

a d b . The first-order condition of the profit maximisation problem

pi mci qi is simply the following Cournot-Nash equilibrium:

a 2bqi d qi mci 0i j

(3.45)

In case N banks are producing positive outputs, the N first-order derivations end up

with the following form:

qi ci 2b / d 1 a 2b / d N 1 mci mc jj

/ 2b d N 1 2b / d 1

(3.46)

Entry costs ε are excluded in this case and, in equilibrium, banks are willing to entry

the market as far as i . In the underlying market, competition is likely to heat up

the closer substitutes are the interbank services and the lower is the entry cost (ε).

Then the following model is constructed after computing the market share by

si qi / q jj

:

ln si a lnmci (3.47)

Any bank facing lower marginal costs is becoming more efficient in the market. The

greater is the competition the stronger will be the effect on obtained market shares.

The parameter β, the Boone indicator, is therefore expected to have a negative sign

and obtain high absolute values amid competitive pressures. Although the model was

formerly specified in log-linear form to eliminate heteroskedasticity, this form enables

88

interpretation across equations, as the parameter β is elasticity. As being time

dependent it is calculated separately for each year to verify whether there exists a

significant change over time in the same industry. Such interpretations are less robust

if the parameter β is cross-sectionally compared across different sectors or countries at

the same date.

Efficiency in banks may be expressed either by increased profits through cost

minimisation or by low prices to gain market share. Boon indicator is intuitively

constructed to explain bank behaviour in between these two extrema. Therefore, it

presupposes that banks pass part of their efficient gains to their clients. That is quite

expected in the passing-on activity of all banks, whereby convergence to such pattern

is justified.

The advantage of this model is that marginal costs are focusing certain banking

sectors where no information are available and the use of market share, rather than

profits, which can take negative values. In line with the theory, by employing translog

cost function (TCF) we derive the marginal costs so as to avoid averaging variable

costs. It is actually a second-order Taylor expansion around the mean value of the

dual cost function satisfying the rules of linear homogeneity and cost exhaustion. For

every national banking sector, it incorporates bank outputs (securities, loans, other

services), factor inputs (wages, other expenses) and a control variable (equity ratio) to

correct for loan portfolio risks as proposed by Berger and Mester (1997).

Carbo et al. (2009) constructed an alternative metric of bank pricing power,

accounting for differences amongst the aforementioned competition measures. What

might explain those cross-country differences is boiled down to cases when TR is

shared by fees and off-balance sheet revenues (non-traditional banking services),

operating cost is diminishing at a different rate (ATMs, etc), banks of unequal average

size imply different scale economies across countries and taxes - loan losses are

different as well.

Hence, as country-specific indicators may distort competition measures, Carbo et al.

(2009) proposed a linear error-specification model of the employed competition

measures with the following independent variables:

89

Operating cost to asset value (OCTA) – the intuition hidden here is that influences in

profitability and rate of return explain cross-country discrepancies in cost efficiency.

The lower the ratio, the greater the cost efficiency is. Annual rate of real GDP growth

(GDPGR) – bank profitability is expected to rise in the peak of the business cycle due

to extensive loan provision. Cost of living index (COL) – high inflation is directly

linked to the previous case. Fee income to asset value (FEEINC) – as already

mentioned, banks deal with traditional and non-traditional activities; the former

engulfs the deposit and loan services, which are less sophisticated but, nonetheless,

more efficiently priced in the market. The latter comprises fund management,

derivatives, insurance services, underwriting and so on rendering the control of them

an imperative to verify whether it is the case of affecting competition measures. Type

of the bank (TYPEBANK) – commercial, savings, cooperative.

The regression model with all the presented variables is schematically the following:

CM i a0,i a1,iOCTA a2,iGDPGR a3,iCOL a4,iFEEINC a5,iTYPEBANK ui i

(3.48)

where

i = NIMTA, Lerner index, ROA, and H-statistic

The expression ui i is supposed to include the random error ui and the

unexplained part εi of each competition measure (CMi). Carbo et al. (2009)

interestingly propose this part to be a proxy for the bank pricing power after being

subjected to adjustment for the effects of the independent variables, i.e. cost

efficiency, non-traditional banking activities and changes due to inflationary pressures

and the business cycle. The process of doing so is to take averages of the whole

expression concerning all banks over time and add the estimated intercept of each

equation to the error term so as only the real effects of the independent variables are

not „accounted‟.

90

The adjusted measure highlights the cross-country bank average control over prices

and increase significantly pairwise correlations between the new concentration

measures. That is because NIMTA, Lerner index and ROA are influenced by the

operating cost or the fee income irrespectively whether they are expressed either in

the quantity or the weighted average price component of the change in the

concentration ratios (or both). However, the H-statistic is affected only by the

association of changes in the input prices and not at all by changes in the operating

input quantities as it assumes fixed bank output and composition. Carbo et al. (2009)

observed that H-statistic did not perform well, possibly due to relatively greater

changes in the output component.

The most recent development in literature regarding the measurement of competition

is that of Bolt and Humphrey (2010). The latter has inspired the idea from efficiency

analysis applying frontier methodology for the ranking of countries in terms of

competition intensity instead of estimating absolute measures of market structure. The

motivation here is that the existing literature has focused on single measures that

provide inconsistent results when compared in cross-country and time-varying

comparative studies, while they fall short of covering non-interest activities other than

traditional ones. Such insights are addressed by applying the distribution-free

approach on a SUR framework and surmising that the key determinants of bank

revenues are boiled down to the effects of unit operating costs, factor input

productivity and competition in the banking market. The general form of employed

models is schematically the following:

ln(R /OC) f (X,P) lne lnu (3.49)

where R denotes R1: non-interest income over total operating cost, and R2: revenues

from the spread of deposit-loan rate. P is a vector of the average price of labour, the

market interest rate that expresses the opportunity cost of investment in bank physical

capital. X expresses the vector of unit costs of processing payment transactions, and

ATMs, indicators of labour productivity and substitution of ATMs with branch

offices, GDP output gap (in model R1) and the risk proxy of capital (in model R2).

What is left off, after the standard error (lnu) fades out over the time, is the

component (lne), out of which we estimate competition efficiency according to

91

e / emin 1. The underlying evidence, when juxtaposed to the results of standard

competition measures (H-statistic, CR-3, among others), demonstrates weak levels of

consistency. Hence, the choice upon the use of specific measure is likely to produce

different, if not erroneous, results and remarks should be drawn with caution.

Delis and Tsionas (2009) proposed the theoretical gratification of an empirical model

that estimates simultaneously market power and operating efficiency scores of

individual banks. The proposed methodology is econometrically challenged by local

maximum likelihood technique (LML).

3.5. Review of empirical analysis

3.5.1. Panzar – Rosse methodology

3.5.1.1. Cross-country comparisons

There exist plenty of studies conducting cross-country comparisons of competitive

conditions across different continents as well as within certain regions like US, Latin

America, Europe, Middle East, Africa, Asia and Indonesia. The most widely applied

model in the banking literature is the Panzar – Rosse statistic for comparison reasons

along with its relationship with other competitive issues.

In particular, we have the studies of Claessens and Laeven (2004) and Bikker et al.

(2012) over a sample of 50 and 63 countries (respectively) around the world to

conclude that monopolistic competition is commonplace to the world banking market

during the period 1994-2004; a result that becomes evident in Bikker and Haaf (2002)

using a narrower sample of 23 industrialized countries within 1988-1998. However,

Sun (2011) investigated US, UK and EU from 1995 until the onset of the financial

crisis and found monopolistic competition for foreign and domestic commercial,

cooperative and savings banks in contrast with smaller banks of Netherlands before

EMU that demonstrated perfectly competitive conduct. More evidence between 1998-

2004 on G7 countries identifies Italy in perfect competition, Germany and France in

monopolistic competition and Japan, UK and US in perfect monopoly (Goddard and

Wilson, 2009).

92

Other studies on Islamic banking are those of Ariss (2010a) and Weill (2011)

investigating 13 and 17 Islamic countries, respectively, and drawing remarks on

persistent monopolistic competition. The same pattern is evident by Murjan and Ruza

(2002) in Arab Midlle East from 1993 to 1997, whereas Al-Mugarrami et al. (2006)

extending the sample until 2002 for 6 GCC banking markets found Qatar and Bahrain

with monopolistic competition, Oman with monopoly and perfect competition in

Kuwait, Saudi Arabia and UAE. Jordan has also demonstrated low competition with a

diminishing trend during the period 1994-2006 as other peer economies like Israel,

Lebanon, Morocco and Tunisia (Dermiguc-Kunt and Peria, 2010).

Almost similar samples constitute the MENA5 region, which appears with

considerable market power according to the studies of Al-Fayoumi and Abuzayed

(2010), Anzoategni et al. (2010) and Ariss (2009). For the CEMAC6 region, evidence

remains the same and in long-run equilibrium within the period 1993-2004 (Saab and

Vacher, 2007) as in EAC7 territory, particularly in Kenya, Tanzania, Rwanda and

Uganda from 2001 to 2008 (Gaertner and Sanya, 2012).

Laeven (2005) came up with colluding behaviour on 7 East Asian countries for the

period 1994-2004 in line with Perera et al. (2006) for Bangladesh, India, Pakistan, Sri

Lanka, Nguyen et al. (2012) for ASEAN countries and Anzoategui et al. (2010) for

Russia, China, India and Brazil. In addition, Olivero et al. (2011) compare a sample of

Asian and Latin American countries highlighting India, Korea, China and Venezuela

as countries lying off the equilibrium in the short-run or under monopolistic conduct.

The Latin America region appears to be monopolistically competitive except the

perfect competitive markets of Peru and Paraguay during 1999-2006 (Chortareas et

al., 2012). In the same vein, Yeyati and Micco (2003) found monopolistic competition

in 8 Latin American countries between 1994 and 2002 while Gelos and Roldos (2004)

for the same period picked up Argentina as an exception and Hungary out of the

Central Eastern European region.

European samples may be classified in transition and developed economies along

with other versions like EU-10, EU-15, European Monetary Union (EMU), Central

5 Middle East and North Africa region

6 It stands for Central African Economic and Monetary Community comprising the countries of

Cameroon, Chad, the Central African Republic, Equatorial Guinea, Gabon and the Republic of Congo.

7 East African community: Burundi, Kenya, Rwanda, Tanzania and Uganda.

93

and Eastern Europe (CEE) or other sub-regions. Monopolistic competition is quite

common in large European samples in Hussan and Mustapha (2008), Kasman (2010),

Staikouras and Koutsomanoli-Fillipaki (2006) as well as for EU-15 according to Casu

and Girardone (2006), Bikker and Groenveld (2000). Other papers dealing with

similar samples are those of Weill (2004b) on 12 EU countries during 1994-1999,

Schaeck (2006) on 10 countries over 1999-2004, Mamatzakis et al. (2005) on

Southeastern European economies during the five years before Euro. Moreover,

smaller samples on Europe boil down to Drummond et al. (2007) ending up with

monopolistic competition for France, Germany, Italy, Spain UK and US between

1998 and 2004 as well as De Bandt and Davis (2000), Boutillier et al. (2005) and

Casu and Girardone (2009) analysing the same sample for the „90s without UK/Spain,

US/UK and US, respectively. Delis et al. (2008) similarly found the PR statistic

between 0 and 1 for Greece, Latvia and Spain during the period 1993-2004.

Cases where the outcome does not advocate monopolistic competition are traced in

Delis (2010), who came up with monopoly in Bosnia, Estonia, FYROM and

Kazakhstan during the decade 1996-2006. Brissimis and Delis (2011) investigated 20

transition economies between 1999-2006, out of which Bulgaria, Hungary, Latvia,

Lithuania and Slovenia experienced monopolistic conditions; however, when they

employed country-specific countries in the revenue function monopoly became

prevalent in Croatia, FYROM, Hungary, Latvia and Lithuania. Other pick-ups off the

standard pattern lie in the prevalent monopolistic tendency in Latvia and FYROM

over the period 1993-2000 (Yildirim and Philippatos, 2007a), in Latvia according to

Drakos and Konstantinou (2005), and Italy during the subperiod 1987-1989

(Molyneux et al., 1994).

On the other hand, Bikker and Haaf (2000) through the PR methodology endorsed

high competitive conditions in 9 developed European countries between 1971 and

1998 apart from the deposits market of Portugal, which experienced colluding

practices. Last, Delis et al. (2008) identified higher market power in dynamic

specifications relative to the static reduced-form revenue function.

94

3.5.1.2. Country-specific studies

We now delve into country-specific applications of Panzar-Rosse statistic. In

particular, different US samples stand within the studies of Shaffer (1982; 2002;

2004) between 1979 and 1999 extrapolating monopolistic competition in Texas and

New York. In contrast, Chang et al. (2012) argued about perfectly competitive

conditions in a broader US sample within the period 1990-2005. We also deem

Canada as being under perfect competition after 1982 until 1994, except for the period

1983-‟84 when higher market power is explicitly evident (Nathan and Neave, 1989).

Jamaica turns out in monopolistic competition between 1989 and 2002 (Duncan,

2003) whereas Daley and Matthews (2012) identify perfect competition in

commercial banks and monopoly in the wholesale banking sector since 1998.

Relatively high market power is also existent in Colombia within the time period

1985-1998 (Barajas et al., 2000).

As for the Latin America region, monopolistic competition is significant in Brazil

during the period 1997-2011 according to Belaisch (2003) and Tabak et al. (2012),

whereas Lucinda (2010) provide inconclusive evidence and long-run disequilibrium

during 2000-2005. Likewise, the banking market of Chile comes with significant

market power for 10 years after the mid-nineties (Karasulu, 2007) although the

private pension system appears under monopoly since 1996. Uruguay is the last sole-

studied country by Gelfand and Spiller (1987) in the region with monopolistically

competitive propensity during 1977-1980.

In the Asia market, Chen and Wuong (2012) found China experiencing high market

power during the period 1998-2007 corroborating the results of Masood and Sergi

(2011) and Ziyi et al. (2005). However, Ou and Tan (2011) argued that the intensity

of competitive conditions is fluctuating periodically in the region. Furthermore, we

observe monopolistic competition in Armenia (Mkrtchyan, 2005), India (Prasad and

Ghosh, 2007), Malaysia (Sufian, 2011; Sufian and Habibullah, 2012), Indonesia

except the monopolistic conduct of large banks (Rachmaningtyas, 2006) and Nepal

(Gazurel and Pradhan, 2012). There are also cases of perfect competition in Ηong

Kong during the „90s, in Philippines (Pasadilla ad Millo, 2005) and Korea since 2007

(Simpere and Hall, 2012). On the contrary, before 2007 until 1992 Korea had been

under significant market power according to Lee and Lee (2005) and Park (2009); in

95

the sub-period between 1997 and 2001, Korean banking market seems to be operating

under perfect monopoly Chun and Kim, 2004). Along the lines, Malaysia within the

period 1996-2008 turns out in long-run equilibrium under monopolistic competitions

as Sufian (2011) and Sufian and Habibullah (2012) argue about. Moreover, Molyneux

et al. (1996) verify the evidence of Lloyd-Williams et al. (1991), who traced

monopoly conditions in 1986 and monopolistic competition in 1988.

In the Africa region, monopolistic competition prevails in Egypt during the period

1992-2007 (Pashakwale and Qian, 2011), Ghana since 1998 (Buchs and Mathisen,

2005; Biekpe, 2011), Kenya (Mweha, 2011), Tanzania (Simpasa, 2011), Tunisia

(Hamza, 2011; Mensi, 2012) and Uganda (Hauner and Peiris, 2005). For South

Africa, we have mixed evidence as Mlambo and Ncumbe (2011) alongside

Simbanegavi and Greenberg (2012) found relative market power opposing to

Greenberg and Simbanegavi (2009), who decomposed the PR estimation into

different asset sizes. Indeed, the while sample as well as the larger banks operative in

perfect competitive conditions whereas smaller banks engage in monopolistic

competition.

In the Islamic world, Kuwait tends to have operated under perfect competition over

the period 1993-2002 while Lebanon since 2000 turns out with PR statistic between 0

and 1 highlighting monopolistic competition, which expends even in Qatar (Al-

Muharrami, 2009a) and Saudi Arabia (Al-Muharrami, 2009b). Turkey, which appears

more popular in the banking literature, maintains a pattern of monopolistic

competition between 1990 and 2010 according to the recent papers of Abbasoglu et

al. (2007), Gunalp and Celik (2006), Macit (2012) and Masood and Aktan (2010).

However, in nich markets like the credit card and non-life insurance market

monopolistic conditions do prevail during 2002-2008 and 1996-2001, respectively;

nevertheless, the latter demonstrates monopolistic competition for the period 2002-

2004.

Over the European region monopolistic competition, which covers the whole range

between the two extrema, is addressed significantly on the grounds of fulfilling the

long-run equilibrium prerequisite. In the developed world, Germany follows similar

pattern during 1988-2002 across different samples employed by Gischer and Stiele

(2009), Hempell (2002) and Lang (1997). Same results are traced in the studies of

96

Hahn (2008) analysing the Austrian market within the period 1995-2002, De Rozas

(2012) for Spain during 1986-2005, Habte (2012) for Sweden since 2003 and Rime

(1999) for Switzerland over 1987-1994. Vesala (1995) investigated the period 1985-

1992 in the Finnish banking market with PR statistic indicating monopolistic

competition during the periods 1985-1988 and 1991-1992, as well as perfect

competition over 1989-1990.

Hondroyiannis et al. (1999) argue about significant market power in Greece between

1993 and 1995, which is contradistinction with Rezitis (2010), who remarked on

perfect competitive tendency for the whole period 1995- 2004 and the subperiod

1995-1998; however, in 1999-2004 monopolistic competition dominated the market.

Furthermore, studies on Italy (e.g. Coccorese, 2004; Trivieri, 2007) highlight

monopolistic competition in the late 90‟s even over samples of single-branched banks

up until 2005 (Coccorese, 2009). Evidence to the contrary is traced in the insurance

market experiencing monopoly and some degree of collusive oligopoly (Coccorese,

2012) during 1998-2003 and perfect competitive conditions in 1992 and 1994.

Moreover, Portugal since 1991 has undergone institutional restructuring that brought

about a significant bearing on the level of competition. Indeed, Boucinha and Ribeiro

(2009) found monopolistic competition between 1991 and 2004, with weaker

competition during 1991-1996, an unstable period of 4 years when restructuring took

place, and stronger competition over 2001-2004.

Amongst the developing European economies, monopolistic competition does hold in

Croatia during the period 1994-2004 (Kraft, 2006), Estonia within 1995-2001 (Kalle,

2002), Georgia since 1999 (Mercan, 2012), Lithuania between 2000 and 2006

(Vaškelaitis and Deltuvaitė, 2007) and Romania over 2003-2009 (Marius and Bogdan,

2011). Czech Republic experiences equivalent conditions between 2005 and 2009,

although five years before perfect competition becomes prevalent according to

Stavarek and Pepkova (2011). Last, Poland comes with PR statistic lying between the

range (0,1) when total interest income is employed as the dependent variable.

However, when the latter is subject to deflation with respect to total assets the PR

methodology shows perfect monopoly over the period 1997-2009 (Pawlowska, 2012).

97

3.5.2. Bresnahan-Lau methodology

3.5.2.1. Cross-country applications

The underlying non-structural methodology has met a handful of cross-country

applications relatively to the popularity of PR statistic. In particular, international

samples are only traced in Shaffer (2001) dealing with North America, Europe and

Asia during the period 1979-1991. They provide evidence over contestability or,

alternatively, Cournot-like oligopoly in Belgium, Denmark, France, Japan and US

whereas competitive structures are evident in Canada, Germany, Italy, Norway,

Sweden and Finland.

Neven and Röller (1999) argued about colluding behaviour in the banking markets of

Belgium, France, Spain, Germany, Denmark and UK over the period 1981-1999,

albeit with a considerable diminishing trend over time. Last, Delis et al. (2008)

applied additional dynamic error-correction specifications along the lines of Steen and

Salvanes (1999) and concluded about potential bias of static models when they fail to

capture short-run dynamics. Indeed, monopolistic competition is generally evident but

in dynamic models the market power is relatively higher. Bikker (2003) reported on a

European sample within the period 1987-1997 with respect to deposits and loans

markets. Especially, the deposit markets of the entire European region as well as those

of Germany and Spain operate under monopolistic competition, while at the same

time the same holds for the lending markets of Germany, Portugal, Spain, Sweden and

UK.

3.5.2.2. Country-specific studies

To our knowledge, there are three studies on US banking market; Shaffer (1989)

examined the periods 1941-1975 and 1941-1993 and remarked on the perfect

competitive conduct of US banks. The latter seems to corroborate the period right

after, namely 1990-2005, when high competition is significant albeit coupled with

some degree of market power in the long run (Chang et al., 2012). In addition,

Hannan and Liang (1993) came up with considerable market power in the money

market deposits between 1983-1989. Likewise, Shaffer and DiSalvo deemed bank

98

behaviour as imperfectly competitive rather than collusive between 1970 and 1986 in

South Central Pennsylvania.

In Colombia, Barajas et al. (1999) allege about stable average spread over time

although the per se market power diminished significantly during almost two decades

after 1974. Furthermore, Uruguay in the „70s experienced oligopolistic rivalry when

legal barriers on potential market entries made banks more willing to engage in

anticompetitive behaviour (Gelfand and Spillerm 1987). However, as soon as

permissions were granted to a specific banking segment (US dollar-denominated

lending), incumbent banks had little air for colluding practices. Along the lines,

Spiller and Favaro (1984) encompass the banking market as Stackelberg-like

structure, in which incumbent banks are supposed to have commitment power and

rationally avoid colluding interactions due to the per se relaxation of entry barriers.

Gruben and Koo (1997) verified competitiveness in Argentina during the period

1991-1997.

Over the European territory, Greece experienced a significant shift towards

anticompetitive practices after 1998 up until 2004 (Rezitis, 2010). Similarly, the

deposits and loans markets turn out as oligopolistic in Netherlands with the exclusion

of mortgage market, where competition intensified significantly over the period 1957-

1990 (Swank, 1995). On the other hand, Toolsema (2002) concluded on perfectly

competitive conditions between 1993 and 1996 in the underlying banking industry.

Berg and Kim (1994) argued that the Cournot and cost-minimisation models do not fit

the data of Norwegian banking market in the „80s, endorsing the explanatory power of

conjectural variation, namely the situation in which banks operate expecting

retaliation in response to changes in their output. For the next three years, Norway

maintained its oligopolistic tendency in retail and corporate banking according to

Berg and Kim (1996). We also observe monopolistic competition in Portugal during

1990-1995 (Canhoto, 2004) as in the case of Sweden with significant market power in

the deposits market during 1989-1997 (Oxenstierna, 2000) and over the period 1996-

2002 (Sjoberg, 2004).

Last, a recent paper of Simbanegavi and Greenberg (2012) infers monopolistic

competition in the South African region.

99

3.5.3. Lerner index

The underlying index has been widely applied in the banking literature especially

investigating correlations of the degree of competition with other contemporary

competitive isssues. Hence, it is rarely juxtaposed to the aforementioned

methodologies so as to compare and thereby challenge the robust persistence of

market power levels. Exception to this rule is the study of Hussain and Hassan (2012),

who dealt with the US market through the application of Bresnahan-Lau model and

Lerner index over the period 2003-2007; the former illustrated perfect competition

when the negative values of the latter implied predatory pricing through pricing below

marginal cost.

Additionally, Dermiguc-Kunt and Peria (2010) utilized structural (concentration

indices) and non-structural measures (PR-Lerner) of bank competition in Jordan and

the analysis produced evidence of mismatch between lower concentration and low

levels of competition, which seems to be deteriorating during 1994-2006. Through the

same measures, Poshakwale and Qian (2011) report V-shaped temporal changes of

competition and average monopolistic competition in Egypt within the period 1992-

2007 by means of conjectural variation, PR statistic and Lerner index.

Other studies used the paradigm of price mark-up over marginal cost in order to shed

light on the relationship between competition and issues like regulation, efficiency,

liquidity, economic growth, financial stability, among others. For example, Skully and

Perera (2012) found a U-shaped relationship between market power and liquidity

across 113 countries, that is banks with low and rising market power tend to lend in

the interbank market and withhold liquid assets up to a point where they allocate less

on liquid assets and become net borrowers in the interbank market. Bramer et al.

(2012) analysed the level of domestic competition in the European Monetary Union

since 2003 and documented its tendency to determine significantly the lending rates.

Khiabani and Hamidisahneh (2012) conducted analysis on the Iranian banking sector

and observed how the estimated market power increased over the period 1996-2006 in

response to the deregulation of the market in the form of market entries. In particular,

bank competition remained stable (until 2000) as soon as entry derestrictions took

place and made them act more competitively. Moreover, Coccorese and Pellechia

100

(2010) came up with not remarkable bearing of market power on efficiency for the

Italian industry during 1992-2007.

Other issues in banking closely related to bank competition are traced to the economic

growth and financial stability. Soedarmono (2010) verified for the Asian region that

in countries with relatively higher economic freedom, market power tends to

undermine potential growth capabilities. As far as the issue of competition-stability

nexus is concerned, we reserve our right to analyse it in the following chapters (5,8)

reviewing the pertinent literature and empirically testing it within the enlarged

European Union. Last, all the aforementioned issues that presuppose the estimation of

the Lerner index to be used at a second stage of analysis, are not exhaustive of the

literature as such work by no means constitutes the scope of any paper.

3.6. Conclusion

We analysed the state of knowledge with respect to the relevant models proposed in

the literature as well as the studies that applied them over various samples and time

periods. That signifies the importance of estimating the true level of competition in

order for researchers to investigate a wide set of objectives like operational

performance, access to services and financial stability, among others. Apart from the

theoretical sophistication in some of the aforementioned models, the empirical

literature is vast and to a certain degree converging to monopolistic competition in

various regions and continents. That induced financial institutions to promote

investment strategies, product differentiation and considerable low costs that

intensified further access to banking services.

It is therefore of utmost importance for incessant progress in the measurement

techniques of competition in order to adapt relevant institutional policies to

contemporary challenges mandated by adverse economic conditions and exogenous

shocks.

101

Chapter 3 Appendix

Review of empirical analysis

Author(s) Period Countries Results

PANZAR AND ROSSE METHODOLOGY

International studies

Al-Fayoumi and

Abuzayed (2010) 1998-2007

MENA MC

Al-Muharrami et al.

(2006) 1993-2002 6 Arab GCC countries

MC (Bahrain, Qatar)

M (Oman)

PC (Kuwait, Saudi

Arabia, UAE)

Anzoategni et al.

(2010) 1994-2008 MENA MC

Anzoategui et al.

(2010) 2002-2008

Russia

Comparison to Brazil,

China, India

MC

Bikker and Haaf

(2002) 1988-1998

23 Industrialized

countries MC

Bikker et al. (2012) 1994-2004 63 countries MC

Chortareas et al.

(2012) 1999-2006 Latin America

MC

C (Peru and Paraguay)

Claessens and Laeven

(2004) 1994-2001 50 countries MC

Dermiguc-Kunt and

Peria (2010) 1994-2006

Jordan (comparison to

Israel, Lebanon, Morocco

and Tunisia)

Doku et al. (2012) 1999-2008 Africa MC

Gaertner and Sanya

(2012) 2001-2008

EAC (Kenya, Tanzania,

Rwanda, Uganda) MC

Gelos and Roldos

(2004) 1994-1999

8 Latin American and

CEE countries

MC (except Argentina

and Hungary)

Goddard and Wilson

(2009) 1998-2004 G7

Italy: C

Germany, France: MC

Japan, UK and US: M

Laeven (2005) 1994-2004 7 East Asian countries MC

102

Murjan and Ruza

(2002) 1993-1997 Arab Middle East MC

Olivero et al. (2011) 1996-2006 Asia and Latin America

MC

M (India, Korea,

China and Venezuela)

or short-run

disequilibrium

Perera et al. (2006) 1995-2003 Bangladesh, India,

Pakistan, Sri Lanka MC

Saab and Vacher

(2007) 1993-2004 CEMAC countries MC

Skully and Perera

(2012) 1996-2010 113 countries

U-shaped relationship

between market power

and liquidity.

Skully and Pereva

(2012) 1998-2008

ASEAN (Indonesia,

Malaysia, Philippines,

Thailand, Vietnam)

MC

Soedarmono (2010) 1999-2007 Asia

Market power in

countries with higher

economic freedom

undermine economic

growth

Soedarmono et al.

(2011) 2001-2007 12 Asian countries

Market power drives

to financial instability

Sun (2011) 1995-2009 EU, UK, US

MC (all countries,

foreign and domestic

banks, commercial,

savings and

cooperative banks)

C (Netherlands for

small banks before

EMU)

Ariss (2009) 2000-2006 MENA MC

M (North Africa)

Ariss (2010a) 2000-2006 13 Islamic countries MC

Ariss (2010b) 1999-2005

Africa, Asia, Pacific,

Europe, Latin America,

Middle East

MC

Weill (2011) 2000-2007 17 Islamic countries MC

Yeyati and Micco

(2003) 1993-2002

8 Latin American

countries MC

European studies

103

Delis (2010) 1996-2006 Central and Eastern EU

Bosnia, Estonia,

FYROM, Kazakhstan

(M)

Brissimis and Delis

(2011) 1999-2006 20 transition countries

MC

M (Bulgaria, Hungary,

Latvia, Lithuania,

Slovenia)

M (Croatia, Fyrom,

Hungary, Latvia,

Lithuania by

employing country-

specific variables)

Bramer et al. (2012) 2003-2009 EMU

Lending rates are

affecting the level of

domestic competition

Casu and Girardone

(2009) 2000-2005

France, Germany, Italy,

Spain, UK MC

Hussan and Mustapha

(2008) 2000-2008 EU MC

Coccorese (2009) 1988-2005 Italy (single-branched

banks) MC

Yildirim and

Philippatos (2007a) 1993-2000 EU (CEE)

MC

M (FYROM,

Slovakia)

Kasman (2010) 1995-2007 EU MC

Delis et al. (2008) 1993-2004 Greece, Latvia, Spain MC

Bikker and

Groeneveld (2000) 1989-1996 15 EU countries MC

Boutillier et al. (2005) 1993-2000 Germany, France, Italy,

Spain MC

Casu and Girardone

(2006) 1997-2003 EU-15 MC

De Bandt and Davis

(2000) 1992-1996


US MC

Drakos and

Konstantinou (2005) 1992-2000 10 CEE countries

MC

M (Latvia)

Drummond et al.

(2007) 1998-2004


Spain, UK, US MC

Mamatzakis et al.

(2005) 1998-2002

Southeastern European

countries MC

104

Molyneux et al.

(1994) 1986-1989


Spain, UK

MC

M (Italy: 1987-1989)

Schaeck (2006) 1999-2004 10 European countries MC

Staikouras and

Koutsomanoli (2006) 1998-2002 EU-10, EU-15 MC

Weill (2004b) 1994-1999 12 EU countries MC

Yildirim and

Philippatos (2002) 1993-2000 14 CEE countries

MC (except FYROM

and Slovakia)

Bikker and Haaf

(2000) 1971-1998

Belgium, France,

Germany, Italy, Spain,

Netherlands, UK,

Portugal, Sweden

High competition;

oligopoly in the

deposits market of

Portugal

Delis et al. (2008) 1993-2004 Greece, Latvia, Spain

MC (Higher market

power in dynamic

models)

Country-specific studies

Mkrtchyan (2005) 1998-2002 Armenia MC

Hahn (2008) 1995-2002 Austria MC

Belaisch (2003) 1997-2000 Brazil MC

Lucinda (2010) 2000-2005 Brazil

In long-run

disequilibrium

Inconclusive evidence

Tabak et al. (2012) 2001-2011 Brazil MC

Nathan and Neave

(1989) 1982-1994 Canada

MC (1983-‟84)

PC

Karasulu (2007) 1995-2004 Chile MC

Sepulveda (2012) 1996-2008 Chile (private pension

system) M

Chen and Wuong

(2012) 1998-2008 China MC

Masood and Sergi

(2011) 2004-2007 China MC

Ou and Tan (2011) 1993-2007 China Competition fluctuates

periodically

Ziyi et al. (2005) 1993-2003 China MC

Barajas et al. (2000) 1985-1998 Colombia MC

105

Kraft (2006) 1994-2004 Croatia MC

Stavarek and Pepkova

(2011) 2001-2009 Czech Republic

C (2001-2005)

MC (2005-2009)

Poshakwale and Qian

(2011) 1992-2007 Egypt MC

Kalle (2002) 1995-2001 Estonia MC

Vesala (1995) 1985-1992 Finland

MC (1985-1988,

1991-1992)

PC (1989-1990)

Giustiniani and Ross

(2008) 1997-2005 FYROM M or perfect cartel

Mercan (2012) 1999-2010 Georgia MC

Gischer and Stiele

(2004) 1993-2002 Germany MC

Gischer and Stiele

(2009) 1993-2002 Germany MC

Hempell (2002) 1993-1998 Germany MC

Lang (1997) 1988-1992 Germany MC

Buchs and Mathisen

(2005) 1998-2003 Ghana MC

Biekpe (2011) 2000-2007 Ghana (commercial) MC

Hondroyiannis et al.

(1999) 1993-1995 Greece MC

Rezitis (2010) 1995-2004 Greece

C (for the whole

period and 1995-1998)

MC (1999-2004)

Jiang et al. (2004) 1992-2002 Hong Kong PC

Prasad and Ghosh

(2005) 1996-2004 India MC

Prasad and Ghosh

(2007) 1996-2004 India MC

Rachmaningtyas

(2006) 1999-2004 Indonesia

MC

M (large banks)

Khiabani and

Hamidisahneh (2012) 1996-2006 Iran

Entry regulation on

competition

106

Coccorese (2002) 1988-1996 Italy MC

PC (1992, 1994)

Coccorese (2004) 1997-1999 Italy MC

Coccorese and

Pellechia (2010) 1992-2007 Italy

„The impact of market

power on efficiency in

not remarkable‟

Trivieri (2007) 1996-2000 Italy MC

Coccorese (2012) 1998-2003 Italy (insurance) M (and collusive

oligopoly)

Coccorese (2009) 1988-2005 Italy (single-branched

banks) MC

Daley and Matthews

(2012) 1998-2009 Jamaica

M (merchant banking

sector)

C (commercial)

Duncan (2003) 1989-2002 Jamaica MC

Lloyd-Williams et al.

(1991) 1986,1988 Japan

M (1986)

MC (1988)

Molyneux et al.

(1996) 1986-1988 Japan M (1986), MC (1988)

Al-Jarrah and Dirasat

(2010) 2001-2005 Jordan MC

Mwega (2011) 1998-2007 Kenya MC

Chun and Kim (2004) 1994-2001 Korea MC before 1997

M after 1997

Lee and Lee (2005) 1992-2002 Korea MC

Park (2009) 1992-2004 Korea MC

Shin-Kim (2012) 1992-2007 Korea MC

Simpere and Hall

(2012) 2007-2011 Korea C

Al-Muharrami (2008) 1993-2002 Kuwait C

Moussawi and Saad

(2012) 2000-2010 Lebanon MC

Vaškelaitis and

Deltuvaitė, 2007 2000-2006 Lithuania MC

Sufian (2011) 1996-2008 Malaysia MC

107

Sufian and Habibullah

(2012) 1996-2008 Malaysia MC

Gazurel and Pradhan

(2012) 2001-2009 Nepal MC

Pasadilla and Milo

(2005) 1990-2002 Philippines PC

Pawlowska (2005) 1997-2002 Poland MC

Pawlowska (2012) 1997-2009 Poland

MC (interesting

income as a dependent

variable)

M (interesting income

over total assets as a

dependent variable)

Boucinha and Ribeiro

(2009) 1991-2004 Portugal

MC (1991-1996 weak

competition; 1997-

2000 restructuring in

disequilibrium; 2011-

2004 stronger

competition)

Al-Muharrami (2009) 1993-2002 Qatar MC

Marius and Bogdan

(2011) 2003-2009 Romania MC

Al-Muharrami (2009) 1993-2006 Saudi Arabia MC

Greenberg and

Simbanegavi (2009) 1998-2007 South Africa

C (larger and all

banks)

MC (small banks)

Mlambo and Ncube

(2011) 1999-2008 South Africa MC

Simbanegavi and

Greenberg (2012) 1992-2008 South Africa MC

De Rozas (2012) 1986-2005 Spain MC

Habte (2012) 2003-2010 Sweden MC

Rime (1999) 1987-1994 Switzerland MC

Simpasa (2011) 2004-2008 Tanzania MC

Hamza (2011) 1999-2008 Tunisia MC

Mensi (2012) 1990-2007 Tunisia MC

Abbasoglu et al.

(2007) 2001-2005 Turkey MC

108

Gunalp and Celik

(2006) 1990-2000 Turkey MC

Macit (2012) 2005-2010 Turkey MC

Masood and Aktan

(2010) 1998-2008 Turkey MC

Akin et al. (2012) 2002-2008 Turkey (credit card

market) M

Kasman and Turgutlu

(2008) 1996-2004

Turkey (non-life

insurance)

M (1996-2001)

MC (2002-2004)

Hauner and Peiris

(2005) 1999-2004 Uganda MC

Matthews and Zhao

(2007) 1980-2004 UK MC

Gelfand and Spiller

(1987) 1977-1980 Uruguay MC

Chang et al. (2012) 1990-2005 US C

Hussain and Hassa

(2012) 2003-2007 US

Lerner: Predatory

pricing

Bresnahan-Lau:

Perfect competition

Shaffer (2002) 1984-1999 US (Jayton, Texas) MC

Shaffer (1982) 1979 US (New York) MC

Shaffer (2004) 1988-1994 US (Texas Kentucky) MC

BRESNAHAN-LAU METHODOLOGY

Cross-country studies

Neven and Roller

(1999) 1981-1990

Belgium, France, Spain,

Germany, Denmark, UK,

Colluding behaviour

Diminishing trend

over time

Shaffer (2001) 1979-1991 North America, Europe,

Asia

Contestability or

Cournot oligopoly

(Belgium, Denmark,

France, Japan, US)

Competitive (Canada,

Germany, Italy,

Norway, Sweden,

Finland)

Bikker (2003) 1987-1997 EU MC (deposit markets

109

of the entire EU,

Germany and Spain;

lending markets of

Germany, Portugal,

Spain, Sweden and

UK).


Oxenstierna (1998) „Asymmetric game-theoretic static Bertrand-type oligopoly model‟

Gruben and Koo

(1997) 1991-1997 Argentina Competitiveness

Shaffer (1993)

Barajas et al. (1999) 1974-1996 Colombia

Decreased market

power although

average spread

remained put

Suominen (1994) Finland

Rezitis (2010) 1995-2004 Greece

„Shift to non-

competitive market

structure after 1998‟

Uchida and Tsutsui

(2005) 1974-2000 Japan

„Cournot oligopoly

cannot be rejected

Swank (1995) 1957-1990 Netherlands

Oligopolistic

conditions in the

deposits and loans

market

Intensified competition

in the mortgage market

Toolsema (2002) 1993-1996 Netherlands C

Berg and Kim (1994) 1980-89 Norway

„Banks behave as if

they expect their

competitors to retaliate

in response to output

change‟.

Berg and Kim (1998) 1990-1992 Norway

Oligopolistic

behaviour in retail and

corporate banking

Canhoto (2004) 1990-1995 Portugal MC

Simbanegavi and

Greenberg (2012) 1998-2008 South Africa MC

Shaffer and DiSalvo

(1994) 1970-1986

South Central

Pennsylvania

„The conduct appears

imperfectly

competitive, but far

110

from collusive.

Oxenstierna (2000) 1989-1997 Sweden Market power in the

deposit market

Sjoberg (2004) 1996-2002 Sweden Intermediate

competition

Gelfand and Spiller

(1987) 1977-1980 Uruguay

Oligopolistic rivalry

Entry barriers

conducive to non-

competitive behaviour

Spiller and Favaro

(1984) Late „70s Uruguay

Stackelberg-like

industry (entry barriers

diminish oligopolistic

interactions)

Chang et al. (2012) 1990-2005 US

High competition

Some degree of market

power in the long run.

Hannan and Liang

(1993) 1983-1989 US

Market power in

money market deposit

Shaffer (1989) 1941-1975/1993 US C

111

CHAPTER 4

A review of efficiency analysis

113

4.1. Introduction

Market structure has undergone remarkable changes the recent years by the

deregulation in the provision of financial services, the designation of monetary policy

in the discretion of European central bank and waves of large-scale consolidation. To

investigate the extent to which all these developments have vitally affected the

European banking sector, it is necessary to review the methodological foundations of

market concentration and efficiency in the competition-related framework.

First, it is of utmost importance to lay down the alternative definitions of bank output,

namely the intermediation and production approach. The treatment is expressed in

terms of capital, labour or values of loan and alternative investments followed by

shortcomings explicitly articulated in the literature. Second, concentration ratios

utilising the number of banks and their market shares under a different weighting

scheme, capture market structure and facilitate an indirect competition assessment

along the lines of SCP paradigm. Last, efficiency scores across banking sectors and

countries are retrievable through stochastic or DEA modeling as summarized

hereinafter.

4.2. Definition of bank

The production approach treats banks as firms and measures their output taking into

account the employment of capital and labour for the production of different kinds of

deposit and loan accounts. The quantitative measurement is based on the number of

accounts or the total transactions per account thereby constituting a flow of output per

unit of time independently of inflation biases (Heffernan, 2005).

There are some problems with this approach; first, there is not an explicit way of

computing the output and weighting every bank service and second, interest rate costs

are disregarded as of minor importance without accounting for the negative

relationship of the deposit rates and the number of branches. During the recent years,

almost all countries are on their way to converge towards International Financial

Reporting. Until then, different accounting systems make bank data of different

countries difficult to derive relative efficiency gains.

114

The intermediation approach simply considers a bank as an intermediary, not the

producer of accounts and loans. Bank output is treated as stock, namely, the value of

loans and investments. Total costs are supposed to be the sum of operating costs (the

factor cost of labour and capital) plus interest costs. In addition, some argue that

deposits should be treated as output, as earning assets (e.g. loans, securities, etc),

being in total contradistinction with Sealey and Lindley (1977). It is nonetheless

impossible to account for the whole range of the financial products since the unit cost

of an otherwise traditional intermediary bank will increase (unless weights indices are

used).

The contemporary practice favours the view of having earning assets produced by

specific factor inputs. The former includes the multiproduct of loans, securities,

interbank assets, deposits and non-interest income whereas the latter the labour cost,

the cost of physical capital (proxy: non-interest expenses/fixed assets) and that of

financial capital (proxy: interest paid/purchased funds).

When it comes to the empirical work, many problems arise from the way output is

measured. For example, some studies use the number of deposit accounts for which

banks incur costs to serve intermediation and liquidity but cannot loan out all the

deposit value whatsoever. On top of that, maturity structure and any risks associated

with and attached to each loan are generally ignored. If interest margins are getting

lower due to the fierce competition, output will decrease with reference to the national

accounts; the other approaches record output fluctuations in case margins influence

the volume of loans.

4.3. Measurement of X – efficiency

The empirical application of the X – efficiency methodology has been based on the

long-standing parametric and non-parametric approaches. Their striking difference is

that the former prerequisites an explicit model of the key determinants of inefficiency;

that is, the case of factor inputs being sub-optimally used (technical) or that of

inefficient allocation of resources due to ineffective management or expense

preference behaviour (Berger and Hannan, 1998).

115

4.3.1. Parametric approaches

4.3.1.1. The Stochastic Frontier Analysis

The development of the Stochastic Frontier Analysis (SFA) is traced back at the

pioneering studies of Aigner et al. (1977) and Meeusen and Van de Broeck (1977).

The idea behind the model specification lies in the definition of the composite error

component in terms of the classical random error and an inefficient term; that is why

Fried et al. (1993) pointed out that the production lag (inefficiency) of banks might

not be ascribed to the banks themselves, but to the unpredicted (random) error effects

whatsoever. The underlying model is schematically the following:

(4.1)

The last two components are supposed to be independently distributed along with the

X vector of variables. The random error is the symmetric disturbance that follows

independently the normal distribution [~ N (0, )] while for the inefficient term u,

various one-sided distributions have been proposed including, among others, the

normal distribution truncated above at zero (half normal)8 and the exponential one.

Such error specifications have been suggested when previous mathematical

programming methods lack any economic gratification and difficulties in statistical

inference. Furthermore, by postulating a non-positive error term (u), the problem of

having observations above the maximum frontier is effectively alleviated.

Thus, the derivation of the distribution function of the composite error is

straightforward (see Weinstein, 1964) in order to produce the conditional density of u

given ε. The inefficient measures may, thus, be obtained by the mode and mean value

of the per se conditional distribution, or alternatively the maximum likelihood

estimators from the maximisation of the joint density of u and ε, subject to the

contraint .

8 It refers to the inefficient term taking the absolute values of a N (0,

2

) variable.

116

4.3.1.2. Distribution-free approach

The underlying approach was developed by the studies of Bauer et al. (1992), Berger

(1993) and Berger and Humphrey (1992) following Schmidt and Sickles (1984). It

postulates a different modeling in error specification, as the inefficiency component is

disentangled by relaxing the assumptions of other models. In fact, the efficiency

follows no particular distribution and relative differences across firms tend to be

stable over time leaving the random error to be averaged out. The formalisation of the

process that sorts out the two components of the composite error consists of three

steps. For a start, the cost – or profit – function of the European banking sector is

estimated for each period by employing unbalanced panel data analysis9, secondly

the predicted series of cost (profit) is deducted from the observed cost to give out the

persistent residuals. Finally, the inefficiency measures are calculated by normalising

the underlying indicators accordingly:

(4.2)

where is the vector of averaged residuals and highlights the most

efficient bank. What really distinguishes this method pertains to the fact that the result

show how efficient a bank is over time relatively to market conditions, rather than the

degree of relative efficiency. That consigns greater explanatory power amid various

conditions in the economy, e.g. interest rates, business cycle, technology, or in the

environment that may impose a considerable burden upon bank swifts in operational

performance.

The shortcomings of the method are quite sensitive to the length of the period. For the

assumptions to be satisfied, random errors should cancel each other out and the

inefficiency component to be stable over time. In case of too short period, the error

term needs more time to vanish asymptotically, thus whatever left off depicts the

inefficiency component, which has not been disentangled, yet. On the other hand, a

longer period tends to strangle the stabilising average efficiency through the

aggregation and intensity of market conditions. Mester (2003) has proposed a 6-year

period, the best solution that counterbalances the aforementioned centrifugal forces.

9 Otherwise, biases may pop up and alter their relative rankings (Akhavein et al., 1997).

117

4.3.1.3. Thick frontier approach

As developed by Berger and Humphrey (1991), this approach estimates a cost

function for the quartiles of the lowest and highest average cost. The thick frontier

refers to the former, by which banks are classified as being of higher than average

efficiency. It assumes that the error terms within the cost functions of each quartile

are random reflecting potential measurement errors and luck. Moreover, the

differences in the predicted costs are further decomposed into the component captured

by market factors and the one explained by inefficiency. In other words, the cost

function is estimated to ex post predict the average cost of the quantiles. Then, the

quantile of the highest AC is estimated by employing the most efficient technology of

the lowest AC quantile in a way to account of all the exogenous market factors of an

economy where banks operate. The difference of the predicted costs of the highest

AC quantile depict the per se inefficiencies, which may be further decomposed to

verify what degree of inefficiency is attributed to market factors and operating costs

(allocative, technical), interest on deposits, among others.

This methodology treats data and in particular the outliers in a way that its impact

poses little, if any, bias on the overall ineffciency level in banking. Allowing for no

specific distribution of the random errors, a potential cost is hidden thereof: surmising

that the random errors follow thick-tailed distribution and get high absolute values,

then the predicted cost differences are not attributed to inefficiencies, which may

follow a flat and thin-tailed distribution, but to the per se random errors.

The most noteworthy extended strand of conventional stochastic frontier modeling

has been the development of Bayesian analysis from Van de Broeck et al. (1994) and

Koop et al. (1994). The former compared alternative models attaching prior

probabilities to the same general model with respect to the distribution of the

composed error term. Thus, Monte Carlo simulations enable posterior inference to

spot the resulting differences in efficiency parameters. The latter utilized Gibbs

sampling10

techniques for flexible stochastic forms of cost function in order to draw

posterior inferences under variable returns to scale and assymptotically ideal

modeling.

10

An algorithm produces several samples from the joint probability distribution of random variables.

118

Koop et al. (1997) extended the framework by allowing for time component in a cost

frontier model. In other words, fixed-random effects modeling, the n effects are

supposed to independently follow marginal or conditional distribution across time or

firms, respectively. Unike the fixed-effect analysis of Schmidt and Sickles (1984),

inefficiency is not derived from the occuring deviations from the most efficient firm,

but interestingly from the relative firm-specific efficiency, which is a function of the

number of firms (n). When accounting for random effects, the postulate of

hierarchical structure is developed on the way firm efficiencies are functionally

formed by exogenous variables or by a common distribution. A similar approach is

ascribed to Osiewalski and Steel (1998) and Fernandez et al. (1997), who extended

the panel analysis of efficiency to construct a priori dependence upon firms‟

charasteristics.

A unifying study that shed some light on the interdependence of various Bayesian

models and empirical results is attributed to Kim and Schmidt (2000). They compared

classical - (e.g.) univariate version, bootstrapping, maxim likelihood - and Bayesian

approaches to the same data set to infer technical efficiency scores in SFM with panel

data and, eventually, verify little differences in the underlying estimators. Further

insights are drawn by Koop (2001) and Fernandez et al. (2000; 2002) to account of

multiple outputs and undesired outputs, respectively. Tsionas (2002) and Huang

(2004), among others, formulated stochastic frontier modeling with gamma

distribution to extrapolate better statistical inference for panel-data applications.

Atkinson and Dorfman (2005) further developed the models of Kim (2002) and

Zellner and Tobias (2001) proposing input distance function specifications.

Toward the end of relaxing the restrictive assumptions of particular functional forms

of the composed error term, Fan et al. (1996) and Kumbhakar et al. (2007) extended

Aigner et al‟s (1977) linear SF model to a semiparametric frontier model that

estimates nonparametric inefficiency scores and pseudo-likelihood variances of the

composed errors utilising a kernel-based technique of the conditional mean function.

Deng and Huang (2008) and Tran and Tsionas (2009) extended former models (viz

Kumbhakar et al., 1991; Batesse and Coelli, 1995), of technical efficiency to be

contingent upon time and other variables within the context of SF production

technology. In a more generalized perspective, Kumbhakar and Tsionas (2008) build

a linking framework between nonparametric frontier analysis and SF modelliing by

119

the use of local ML methodology that drops the adoption of a unique production

technology and provides returns to scale measures.

Park and Simar (1994) proposed alternative estimators of linear semiparametric

frontier modeling, followed by Park et al. (1998; 2007) to include lagged dependent

variable and dependency structures between random errors and regressors,

respectively. Adams et al. (1999) built upon the latter framework to develop

multioutput stochastic distance technologies that mitigate endogeneity and

mispecification problems. Sickles (2005) provided a thoughtful review on the

microfoundations and properties of all estimators that the whole spectrum of

theoretical panel models produce in the literature, and compare them with respect to

various mispecified settings.

Shifting the attention to the employment of multioutput technologies, Fare et al.

(2010) used Monte Carlo simulations to investigate the performance of distance

functions so as to realize that quadratic forms of technology produce better results

than translog representations. In the same vein, Zhang and Garvey (2008) found that

there exist correlations between the results obtained by stochastic distance function,

ray and DEA frontier models whether inefficiency effects are accounted of or not.

However, the criterion according to which input or output dintance function is

applicable is to some extent given by Kumbhakar et al. (2007), who utilize latent class

modeling to identify the factors of efficiency orientation with the necessary indices to

cross-estimate technical inefficiencies.

Sickles et al. (2002) also compared empirically various semiparametric and

nonparametric estimates of a stochastic (multiple input/output) distance frontier

model - introducing multivariate kernel measures and Malmquist index numbers,

respectively – finding poor convergence of technical efficiency between them. Coelli

and Perelman (2000) analysed various distance functions recommending the

robustness of multilateral Tornqvist output index rather than revenue as a proxy of

aggregate ouput in production functions. On the contrary, Hailu and Veeman (2000)

incorporated (un)desirable outputs in an input distance frontier model to produce

efficiency scores that are friendly to the environment.

120

4.3.2. Non-parametric approaches

4.3.2.1. Data Envelopment Analysis

It is the most frequently used method in the empirical literature as far as non

parametric approaches are concerned. Data envelope analysis (DEA) was firstly

developed to compute the „best practice‟ of non-profit-making organisations due to

the lack of accounting measures (Charnes et al., 1978). Since then, theoretical

developments and empirical applications have been published covering up an endless

literature, which is beyond the scope of any paper.

Three assumptions are crucial for the construction of DEA model: constant returns of

scale, convex production function and free disposability of inputs. Charnes et al.

(1978) proposed a non-linear (fractional) programming problem to estimate the

efficiency upon a ratio of weighted outputs to weighted inputs. Given that these

weights are not known, out of a pool of all banks‟ data, the maximized (best-practice)

ratio constitutes the benchmark, which the efficiency of every single bank is

compared to. According to Charnes and Cooper (1962), the problem can be expressed

in linear terms allowing for the comparison to be based upon the horizontal (vertical)

projection of every bank‟s point (input, output) to the estimated frontier for input

(output)-oriented approaches.

However, Banker et al. (1984) dropped the assumption of constant returns of scale.

They added an additional unrestricted variable (convexity condition) to deploy

comparisons of inefficient firms to efficient ones of similar size, thereby eliminating

any scale effect on technical efficiency. Such scale classification, within which

efficiency scores are comparable, has been proposed by applying the SE index

method (Fare et al., 1994) or Banker‟s (1984) idea adopting the scale size of greatest

productivity. In the case of multiple optima, where scale classification is not unique

but subject to the specific solution, Zhu and Shen (1995), Seiford and Zhu (1997),

among others, have proposed ways to reconcile multiple optimal solutions with

returns-to-scale classification (RTS).

Another interesting insight on the development of DEA model comes from Charnes et

al. (1985a). They constructed the additive (Pareto-Koopmans) model, which

combines the output and input orientation into a single linear optimisation problem.

121

Since the model does not provide a measure of (in)efficiency, Charnes et al. (1985b)

proposed the Q measure to be subject to the model‟s restrictions; Consistency to the

aforementioned baseline models was restored by Sueyoshi (1990), who made the

efficiency measure (1-Q) take values between [0,1]. The possibility of the latter

measure to take negative values urged Tone (2001) to construct a slack-based

measure (SBM) that gets over the non-linearities of Green et al.‟s (1997) model and

produces efficiency scores within the interval [0,1] with the following two properties:

measurement units invariance and decreasing monotonicity. Along the lines, Cooper

et al. (2006) alleged that the Russell model of Fare and Lovell (1978), or even the

enhanced version of Pastor et al. (1999), turns out to be equivalent with SBM in the

case of taking the underlying constraints as equalities.

Multiplicity in DEA modeling constitutes another area of the literature, investigating

the evolution of efficiency measurement in reference various situations. The studies

of Fare and Grosskopf (1996) with network models, Zhu (2003) and Seiford and Zhu

(1999) with supply chain settings, are typical in examining the structure of production

process. Furthermore, research has been done for the restrictions imposed on the

multipliers of DEA models. Pertinent methodologies comprise absolute bounds (Roll

et al. (1991), cone ratio restrictions (Charnes et al., 1990), assurance regions to mould

the differences in multipliers (Thompson et al., 1986; 1990; Cook and Zhu, 2008),

facet models to overcome zero weights on multipliers (Bessent et al., 1988; Greene et

al., 1996) and unobserved decision making units (DMU) to extend observed facets

(Thanassoulis and Allen, 1998).

Several extensions of DEA models also include the use of non-discretionary

variables. However, the first model, proposed by Banker and Morey (1986), may

inhere upward bias in technical efficiency due to existing production impossibilities

(Ruggiero, 1996). The model of the latter has excluded such impossibilities and found

empirically to produce good efficiency scores according to Syrjanen (2004) and

Muniz et al. (2006). Provisions for categorical variables (Banker and Morey, 1986),

ordinal variables (Cooper et al. 1999; Cook et al., 1993; 1996;) and flexible variables

(Cook and Greene, 2005; Cook and Zhu, 2007) are also suggested in DEA

applications (Cook and Seiford, 2009).

122

DMUs may exhibit different efficiency scores when it comes to account for several

alterations in the employed parameters or data variability. Sensitivity analysis refers

to whether, and to what degree, the change of the number of DMUs (Wilson, 1995) or

the shocks in input and output values (Neralic, 1997; 2004) may have an impact on

the relative efficiency scores. In the case of super efficiency model, the problem of

ranking DMUs with unity efficiency is approached by Andersen and Petersen (1993),

Lovell and Rouse (2003) and Cook et al. (2008), who proposed a way to resolve the

issue of infeasible solutions that may come up under certain conditions.

Thore (1987), Cooper et al. (1996, 2004) and Land et al. (1992, 1994) have modelled

technical efficiency assuming that input and output variables are random, but

nonetheless following known distributions. Such data variability, though in a time-

series context, is allowed by means of window analysis (Charnes et al., 1985a) in

order to produce robust efficiency scores. In so doing, window is referring to the m

portions that observations of each DMU are divided by and, thus, m scores are

produced for the DMU itself; then, the analysis is rolling over the sample period on a

portion-by-portion basis ending up with m x n scores. A shortcoming of this

approach, articulated by Cooper et al. (2006), is that the very beginning and ending

periods are not used as often as all the others in between. In panel data methodology,

along with the window analysis, the Malmquist index was developed by Fare et al.

(1994) to measure how productivity is evolving across time in terms of its two

components: technology and technical efficiency. Last, a robust treatment of data

variability is provided by the studies of Banker and Maindratta (1992), Banker and

Natarasan (2004), which proposed an estimator on the grounds of stochastic

deviations from an arbitrary concave production function.

4.3.2.2. The free disposal Hull approach

The model was developed by Deprins et al. (1984) and subsequently by Tulkens

(1993) to generate the frontier of the observed production activities of DMUs. The

convexity assumption of the production function is dropped while keeping fixed the

variable returns of scale and the free disposability of inputs and outputs as the

nomination of the approach is explicitly highlighting. The study of Banker et al.

(1984) constitutes the reference baseline model, upon which convexity is relaxed

123

through the inclusion of an additional restriction. The production technology depicts

the reference frontier whereby the horizontal projections from the input-output data

points are utilized to produce the efficiency scores.

Given that the underlying frontier is a feasible set of input-output vectors constructed

on a linear piecewise technology of dominating combinations of observed (only) data,

the production set is a priori unknown, albeit right on or inside the DEA frontier. That

implies, according to Deprins et al. (1984), that inefficient DEA combinations are not

always inefficient in FDH models, whereas the opposite does hold. Average

efficiency scores are larger in this case due to the higher percentage of efficient

observations, albeit performing better fit for the observed data than DEA (Tulkens,

1993). However, this approach befalls in limited empirical studies since any statistical

inference in finite samples, or even bootstrapping, is rendered to be cumbersome, if

not unreliable, from discontinuous FDH estimators. Thus, Jeong and Simar (2006)

proposed a linear version of FDH model without losing the non-convexity assumption

in the production set.

Other attempts that fall within the nonparametric literature are quite a few. Contrary

to the postulates and intuition of the additive models, Frei and Harker (1999), Portela

et al. (2003) and Aparicio et al. (2007), among others, proposed models that estimate

the closest distance projection to the Pareto-efficient frontier, rather than all the facets

of it. In addition, another interesting insight is to rank DMUs with respect to their

respective operating performance without betaking to restricting weights schemes on

multipliers (Sexton et al. 1986). The impossibility of acquiring the optimal weights,

cross efficiency is jeopardized to encompass no useful gratification according to

Doyle and Greene (1994). From a game theoretic perspective, Liang et al. (2008) ends

up with the whole set of efficiency scores by setting out DMUs as players, who are

trying to maximize their efficiency in view of given scores of the others, and this

process is rolling over a stream of updated scores till the point of a perfect

competitive equilibrium.

The most recent contributions have managed to shed some light on the shortcomings

of DEA models. The determinacy of non-statistical methodology, or in other words

the lack of statistical inference has been challenged through pinning down maximum

likelihood properties for DEA estimators. Grosskopf (1996) conducted a selective

124

review on regularity tests and sensitivity analysis of univariate DEA frameworks

along with Kneip et al. (1998) and Park et al. (2000). The limitations here pop up in

small samples where asymptotic properties are off the frame and extra noise

jeopardizes the consistency of the constructed confidence intervals.

Bootstrapping methods have been devised to allow for efficiency sensitivity to

sampling variations introducing a stochastic component into efficiency estimators.

Ferrier and Hirschberg (1997) estimated the confidence intervals from the computed

empirical distributions of the efficiency scores, though failing down to eliminate the

bias of DEA estimators through bootstrapping. Simar and Wilson (1998; 2000a) made

a step further by establishing a resampling procedure of generating data that do

correct the bootstrap distribution; Simar and Wilson (2000b) extended the initial

model by refining the correcting bias mechanism and allowing for efficiency

heterogeneity.

The sensitivity of such deterministic methodology to outliers is revisited by Cazals et

al. (2002), who introduced a trimming parameter to get rid of the effect of outliers and

extreme values; in so doing, partial frontiers produce efficiency estimators that do not

envelope all the data points. Aragon et al. (2005) enhanced the methodology by

proposing the concept of conditional quantile-based efficiency scores on the grounds

of continuous partial frontiers of a univariate framework. Daouia and Simar (2007)

expressed further the asymptotic properties of Aragon et al.‟s (2005) estimators in a

multi-dimensional vector of inputs and outputs as well as providing some numerical

illustrations to verify the robustness of estimators.

Other advances of dynamic and stochastic DEA models that account of noise

inclusion are traced in the pioneering work of Sengupta (1990) along with other

relative programming formulations. Olesen and Petersen (1995), Cooper et al. (1996,

2002), applied the theory of chance constrained programming to model the

confidence regions for observed multiple combinations of outputs and inputs;

sensitivity of efficiency scores to changes in probability levels of data variability

highlight the indirect effect of data noise. Kneip and Simar (1996) articulated the

general framework of DEA modeling with panel data along with a (semi)parametric

firm-effect model that captures random effects. The rate convergence of estimators is

also comparatively investigated in each model specification. However, in the case of

125

(ratio) bounded or ordinal data, the Imprecise DEA (IDEA) has been developed by

Cooper et al. (1999) and Kim et al. (1999) to make linear the problem of variable

variability in terms of either scale or variable mix. The resulting nonlinearities in the

presence of weight restrictions made Chen et al. (2002) and Zhu (2002) estimate a

nonlinear IDEA in the context of the basic CCR model obtaining valid results.

4.4. Review of empirical applications

4.4.1. International comparisons

Allen and Rai (1996) investigated the operational efficiency of international

commercial banks of fifteen European countries applying stochastic cost frontier

model with translog and distribution-free specification. In doing so, the required

analysis consists in the examination of technical and allocative X-efficiencies as well

as the contribution of scale and scope economies to bank‟s output. During 1988-1992,

input inefficiencies were present and relatively greater than output inefficiencies in

large banks that operate in non-integrated banking countries; however, the most

efficient banks seem to operate in Australia, Austria, Canada, Denmark, Germany and

Sweden and the worst performers in France, Italy, UΚ and US.

Fitzpatrick and McQuinn (2008) utilize the framework of Batesse and Coelli (1995) –

that of simultaneously computing the inefficiency scores and regressing the latter on a

set of independent variables – in a stochastic profit model. Their contribution lies in

the use of managerial effects as regressors on inefficiency to capture and verify the

hypotheses of bad luck and management practices. An interesting result is that, during

1996-2002, the average profit loss due to inefficiency is as much as 31% essentially

exacerbated by the worst efficient player, UK, among Ireland, Austria and Canada.

In the same period, Gonzalez (2009) examined the relative effects of efficiency and

on market structure after controlling for political economy variables across the

different environments of Europe, Asia, US, Caribbean and Latin America. More

specifically, he estimated a different frontier for each country to allow for comparison

of efficiency between banks within national borders and verify whether that may

affect on concentration and market share. Along the same line, such an effect is

126

tracked whether and to what extent varies across countries of all regions under

constant and variable returns to scale. Applying the two-stage least squares random

effects estimator of Baltagi (2001) (EC2SLS), the results give credit to the indirect

effects of instrumented political economy variables (e.g. infrastructure, supervision

and regulation) on the statistically significant efficiency-structure hypothesis. From a

policy standpoint, the findings propose antitrust measures that may either jeopardize

efficiency in industries where private monitoring and sufficient contracting

environment are present, or second countries with less stringent regulation and better

deposit insurance schemes.

Hagendorff and Keasey (2009) examined the ex post effect of different M&As

strategies on performance in US and Europe during 1996-2004 in a way to infer ex

ante merger strategies. Interestingly, US banks tend to capitalize on institutional

restructurings by increasing revenues from interest and non-interest income assets.

That, in practice, raises questions on the inefficacy of US governance to fortify any

kind of stakeholders against the inherent risk of value destruction. In Europe, the

integration of member states has rendered the lagged consolidation of credit

institutions to be an imperative factor in policy making. Contrary to the US findings,

post merger accounting data endorse the cost-cutting strategy of European M&As

along with the substantial cut on loan provisions.

Lensink et al. (2008) made a step further on the efficiency effect of foreign ownership

allowing for the institutional differences in quality between home and host countries

to explain the gap from the best practice (frontier). By pooling commercial banks of

105 countries worldwide over the period 1998-2003, the empirical results draw

remarks on the better efficiency of foreign banks, the greater the institutional quality

and the smaller the differences in the institutional governance between host and home

countries.

Lozano-Vivas and Pasiouras (2010) run an international empirical study on the extent

to which the so-called non-traditional bank activities have an impact on operational

performance. Sampling publicly traded commercial banks of 87 countries, they used a

multi-product translog cost specification and variables encompassing economic

conditions, industry features and the regulatory environment in order to control for

inefficiency differences all over the world. On average, the inclusion of off-balance

127

sheet items and non-interest income in the output vector points at facilitating

efficiency gains emanated from countries with less government- and foreign-owned

banks, greater stock market development, protection schemes for shareholders and

developed derivatives stock exchanges (more ascribed for cost efficiency).

Similarly, Maudos and Pastor (2001) analysed the profit and cost efficiency in the

European Union, Japan and USA for the years 1984-1995. The contribution of this

paper pertains to the estimation of the profit efficiency, its comparison with cost

efficiencies scores and hence its contribution to bank profitability since almost all

previous studies overlooked the potential bias of the effects of competition,

production qualities and regulatory environment on the cost side of efficiency. The

stochastic frontier analysis concluded that Japan constantly looses efficiency levels

from 1988 onwards while USA and Europe are able to reap the benefits from

competition intensities as demonstrated in the aggregate profitability data of the

OECD countries.

Meon and Weill (2005) used stochastic frontier analysis to examine the magnitude of

six indicators of governance on aggregate efficiency of banks in 62 countries. After

controlling for openness, latitude and „ethno-linguistic‟ features, they estimated six

frontiers plugging in the inefficiency structure one specific index at a time. The

empirical results show a positive correlation of efficiency and institutional quality

though government efficiency has greater explanatory power, followed secondarily,

by probity and rule of law. It is noteworthy that all the other coefficients of the

production function demonstrate a remarkable stability.

Another potential source of efficiency, namely regulation and supervision, is studied

by Pasiouras (2008b). He followed the DEA approach using a bank data of 95

countries around the world for the year 2003. In a nutshell, the efficiency results, as

decomposed into technical, pure and scale efficiency, are presented by region and by

the percentage of banks within each region that exhibit constant and variable returns

of scale. Asia Pacific and Western Europe stand out as more efficient ranging from

75.67% to 97.66% and from 68.8% to 94.61%, respectively, contrary to the ill-

performed counterparts Eastern Europe and Latin and Caribbean America, which

roughly take values in between the range of 53.71% and 90.16%. In a second stage of

analysis, he utilized tobit regressions for efficiency modeling - since it takes values

128

between the interval [0,1] - to see how and to what extent macroeconomic conditions,

bank-specific features, financial development, market structure, the legal system and

the access to banking services determine operational performance of banks.

Evidently, market discipline mechanisms remain robust across all the employed

specifications backed by the statistically significant factors of market concentration

and capitalisation, government and foreign ownership, property rights and the number

of branches.

In view of the shortcomings of the previous study with respect to the static analysis

(not time variant) of technical, pure and scale efficiency and not the broader concepts

of cost and profit efficiency, Pasiouras et al. (2009) uses stochastic frontier analysis

for commercial banks of 74 countries over the period 2000-2004. The remarks drawn

confer power on the supervision that enforce stringent disclosure requirements and

market discipline incentives on the grounds of controlling for various macroeconomic

conditions, market structure and financial development. In fact, it is the case of

negative (positive) effect of capital requirements on profit (cost) efficiency probably

due to the preference of less profitable assets (less risk management costs). Therefore,

in the context of policy making, the indirect interactions of competition, efficiency

and financial stability should be accounted given that deregulation may induce

competitive entities to imprudently engage in risk taking activities (Keeley, 1990).

Pastor et al. (1997) in response to the deregulation of the Spanish banking system

shed some light on the efficiency differences of the US and European banking sectors

for the year 1992. Applying DEA methodology along with the productivity index of

Malmquist (1953), all countries are analysed in terms of the catching-up effect within

the national borders and distance to the reference frontier; in other words, the latter

highlights how much efficient a bank of a country is with respect to the technology of

other countries. On those grounds, Spain, France, UK and USA exhibit high

efficiency internally but employing low levels of technology (productivity) as

compared with the exactly opposite cases of Germany, Austria, Italy and Belgium.

Pastor and Serrano (2006) fill the gap of specialisation effects omitted in the

internally compared efficiency scores of the previous study. The banking systems

under investigation are European during the period 1992-1998 and the analysis

postulates the decomposition of inefficiency in two distinct constituents: the

129

inefficiency difference both in the composed spectrum of specialisation in every

single banking system and in the per se specialisation of their own. Greece and Spain

lead the group of cost inefficient banking sectors, accompanied by France, Germany

and Denmark, which productively specialize more in retail banking. Given that the

specialisation effect discounts significantly the composite inefficiency scores, retail

banking increases its positive contribution to total efficiency since its relative –

though poor - performance dominates that of other activities (wholesale and universal

banking).

4.4.2. European cross-country comparisons

It is beyond doubt that the methodological and empirical investigation of efficiency in

banking sector constitutes a hot area for academic debate. US studies have been by far

numerous and dominant over any other regional research, though losing some pace

after the inundating European interest in late 1990s and onwards. Berger and

Humphrey (1997) reviewed 130 studies applying various frontier approaches in 21

countries. The average efficiency stands at around 77% across models and the relative

rankings do not seem to coincide with those of the computed efficiency scores of

firms. Ever since, the cross-country efficiency literature of EU developed economies

can be divided in specific strands: a) the research conducted to measure the technical

change, productivity growth, scale and scope economies in the European region

employing various models and output specifications, b) the considerable interest in

the environmental factors that influence the conjectural variation of bank efficiency,

namely, country-level features (location, (de)regulation), industry-wide (market

structure, share, M&As) and bank-specific (profitability, off-balance sheet items, risk-

taking, corporate governance) and c) integration of banking sectors measuring the

evolutionary convergence of efficiency scores.

The first studies of convergence are ascribed to Tomova (2005) who applied sigma

convergence, ANOVA and beta convergence tests on the grounds of regulation and

commercial bank objectives. The results show a significant convergence of profit

efficiency scores across all European countries and, especially, a faster pace in new

member states. However, there are persistent disparities of efficiency average levels

with respect to commercial objectives and statistically insignificant differences for

130

regulatory purposes in the old and CEE countries during 2001. In addition, panel data

analysis gives credit to such differences identifying greater efficiency to old EU

member states and some Eastern countries like Malta, Czech Republic, Croatia and

Poland; in contrast, Lithuania, Romania and Estonia demonstrate the poorest

performance. Last, in contrast with inflation, market share, GDP growth and equity

share of total assets turn out to be positive and significant determinants of efficiency.

According to Weill (2009), β- and σ-convergence methodology for panel data

supports bank integration in the European region in terms of both concepts of cost

efficiency convergence over the period 1994-2005. However, Koutsomanoli-Fillipaki

and Mamatzakis (2010) employing a quadratic loss function of a „forward-looking-

rational-expectations‟ model, assess the speed of adjustment in efficiency scores. The

findings endorse undoubted progress during 1998-2005 but, still, there is a long way

to go eliminating the obstacles that hinder cross-border competition in retail banking

and European integration.

Few studies have investigated scale and scope economies in the European context.

Schure and Wagenvoort (1999) followed the Recursive Thick Frontier Approach

(RTFA) for an augmented Cobb-Douglas model to conclude almost absent positive

economies of scale (up to 6-million size) and potential of 6% cost reduction in

savings banks. The average 4% reduction of inefficiency out of 16% potential

comprise the poor performance in Germany, Austria, Luxembourg and France and the

unexploited capacity in terms of scope economies. Accounting for bank types, X-

efficient mortgage banks incur lower costs than commercial and savings banks

altogether. Similarly, Turati (2003) drew the same results finding insignificant

differences in terms of mean efficiency across countries and unexploited scale and

scope economies. Elasticities of substitution and derived demand highlight the

constraining laws that make banks unable to substitute inputs in view of relative price

fluctuations.

As for the M&As effects on banks‟ operating performance, there is no supporting

evidence of the passing-through effect on X-efficiency according to Punt and Van

Rooij (1999). The matter of antitrust policy rests on unfavourable price setting that

currently seems „faint‟, and the potential social welfare gains. In contrast, Huizinga et

al. (2001) analysed 52 horizontal bank mergers during 1994-1998 and realized

positive correlations between mergers and cost efficiency and little, if any, impact on

131

profit efficiency. Their findings on motivation for consolidation corroborate most of

the empirical evidence about unexploited scale economies, inefficient management

and technological advancement. Along the same lines, domestic consolidation is

favoured by differences of banks in capitalisation, innovation strategies and

technology while dissimilarities in strategies concerning earnings, loans and deposits

undermine performance-enhancing goals (Altunbas and Marques, 2008); for cross-

border mergers, the remarks go the other way around. During the period 1996-2004,

Lensink and Maslennikova (2008) document a predominant domestic trend of

horizontal consolidation that provided abnormal returns for the acquirers roughly

around the deal of imminent mergers.

Other effects on bank performance comprise the contribution of off-balance sheet

items, IT investment, ownership, risk and regulation. Casu and Girardone (2005)

explicitly allege that the inclusion of OBS in bank output dismantles misspecification

problems and seems to increase productivity growth in terms of technological change

rather than efficiency variation. Becalli (2007) confronted a profitability paradox

when concluded that European banks having invested a lot in IT, during 1995-2000,

gleaned little gains in profits and efficiency. However, the heterogeneous impact of IT

types conveys performance-related advantage on IT services from external providers,

whereas hardware and software tend to diminish both profit efficiency and profits.

The probability, however, of better performance is greater for loan-oriented balance

sheets and limited sizes during 1993-2001, after the analysis Barros et al. (2007)

conducted with a mixed logit model. That is juxtaposed to the deregulatory endeavour

of policymakers to intensify cross-border activity and penetration to foreign bank

markets.

The omitted component of risk-taking preferences in the efficiency literature was

lately studied by Pastor (2002). He decomposed bad loan provisioning in its

exogenous and endogenous part so as to identify what extent of credit risk is

attributed to economic factors and bad management. The results show efficiency in

risk management to be enhancing in Spanish banks, stabilising in German and Italian

banks, and worsening in French banking sector. Pastor and Serrano (2005) further

contributed to the analysis with the estimation of risk-adjusted efficiency in EU – 11

to see whether risk behaviour was modest or imprudent. Average cost efficiency was

about 89.5% at its best in France and Luxembourg and essentially in commercial

132

banks. The worst performance was identified in Spain and Portugal at savings banks.

The adjusted cost efficiency gave no significant changes while discrepancies exist

between the already lower profit efficiency and risk-adjusted metric. Hence, it is quite

interesting that risk matters especially in the revenue side: France is the least profit

efficient and Portugal and Spain the most profit efficient contrary to the previous

results.

On the other hand, stable and great efficiency in risk management is reported in

Portugal, Luxemburg, Belgium, and Germany, in mortgage banks and those of

medium size. Altunbas et al. (2007) found that inefficiency in banks is positively

related with less risk-taking and more capital profoundly in line with the stringent

regulation. That is the case for all types of banks except for inefficient mortgage ones,

which tend to hold significantly lower capital. As far as ownership is concerned,

private and mutual banks are riskier and more profitable than public banks in 15

European countries for the period 1999-2004 (Iannotta et al., 2007). The mutual banks

engage in relationship banking enjoying better loan portfolio and lower operational

costs. Profitability, however, tends to shrink due to unexploited scale economies and

different asset mix; as they are getting closer to the services of large private banks,

policy makers should remove ownership hurdles of „one-member-one-vote‟ systems,

in order to facilitate market contestability through progressive restructurings.

Other studies have focused on the impact of regulatory initiatives on the European

market structure and banks‟ operational performance. From a theoretical perspective,

there are conflicting views as to which directives do work and whether a general

policy is applicable in a unified market. Empirical evidence delineates the key role of

the third pillar of Basel II, namely the market discipline. Policy-making, therefore,

should ensure the timely disclosure of information and create motives that intensify

private banking (Delis et al., 2011). Since 1999 the efficiency of acquiring banks is

greater than that of the acquired while the resultant competitive forces, due to the

combination of market entries and better local competitors, make incumbent banks

operate with lower costs and inefficiency (Evanoff and Ors, 2008). Indeed,

deregulation has entailed the elimination of entry barriers and the gradual

convergence of efficiency scores across the European banking systems. The

productivity gains, as precipitated by pertinent regulatory reforms, show Germany as

the most efficient banking region followed by Austria, Belgium and Denmark,

133

whereas Greece, Spain and Portugal are encompassed by the lowest rankings

(Kondeas et al., 2008).

4.4.3. Developed economies

An interesting way of reviewing the latest literature of cross-country differences in

efficiency of the developed European economies is to assess the empirical evidence

by the employed methodology from 90‟s and onwards. Berg et al. (1993) examined

the Nordic region in 1990, identifying significant efficiency discrepancies between

Norway and Finland, whose market structure contains the largest and the smallest

pool of banks, respectively. The Swedish banks demonstrate the highest scores of

Malmquist indices, constituting a productivity gap of 40-46% with Norwegian and

52-63% with Finnish banks. Accounting for constant and variable returns of scale,

Swedish banks are the most efficient performers than the least efficient Norwegian by

64%; a key remark is the technological advantage of small Norwegian banks since the

contribution of technology frontier contributes the most to the overall performance.

For 1993, Lozano-Vivas (2002) applied DEA in ten European countries by means of

incorporating the most influential environmental variables on the basic model and

comparing the corresponding output with that of the original model. In doing so,

cross-country efficiency comparisons are inferred by coupling score changes with bad

country-specific environmental conditions. Belgium, Portugal, Spain and Denmark

have had an average efficiency of 70% followed by Luxembourg, Germany, UK,

Netherlands, and the worst performers, namely France and Italy.

Casu and Molyneux (2003) examined the same period 1993-1997 utilising the DEA

methodology and bootstrapping technique in order to infer statistical estimates. That

powerful statistical approach was originated by Efron (1979) to correct the bias from

the inherent dependency of efficiency scores – first articulated by Xue and Harker

(1999) – by means of deriving confidence intervals from the resultant empirical

distribution of efficiency scores. Hence, after the estimation of DEA, they compare

the output of Tobit regression and that of bootstrap Tobit regressions: Productive

efficiency turns out to converge within the European market except for the case of

Italy, although ranging at low levels whatsoever. Eventually, empirical evidence

signifies the role of country-specific factors in explaining efficiency scores reflecting

134

thereby the national regulation and managerial stance vis-à-vis fierce competitive

pressures exacerbated by information technology and financial innovation. Extending

a bit the examined period for 1994-2002, Tomova (2005) estimated a common

frontier for 25 states of the European union plus the accession countries that have

been following the run-up procedure of convergence. The DEA approach revealed the

less efficient eastern part of Europe, as opposed to the developed western group,

whereas the hypothesis of convergence, or in other words the equality of relative

variances, could not have been rejected. The same pattern of decreasing dispersion of

average efficiency variability was identified in the accession countries, as well.

The SFA methodology and its amendments caught much more attention of the

researchers in the 90‟s till 2005. Thus, for the period 1989-1997, Altunbas et al.

(2001) applied Fourier-flexible form of the cost frontier model to produce X-

efficiency, technical change and scale economies. The results show that large-sized

banks exploit technological advantages and are potentially capable of accomplishing

efficiency through the reduction of managerial performance below the current high

levels 20-25%. Nonetheless, they tend to have no scale economies (average 5-7%)

over their smaller competitors. For the same period 1989-1996, Carbo et al. (2002)

followed the same methodology to remark that smaller-sized savings banks are

qualified to have relatively better managerial efficiency and strong motivation to

capitalize economies of scale through cross-border consolidation. Contrary to

Altunbas et al. (2001), scale economies are apparent in larger bank sizes and banks

persistently take refuge in them over time since competition has urged them to curtail

X-inefficiencies and take advantage of their size benefits.

Bikker (2002) employed multiproduct translog functional form to examine efficiency

in fifteen European countries over the period 1990-1996. Inefficiency scores are

creeping an average of 30% classifying Luxembourg, Switzerland as the most

efficient banking industries, Germany, Portugal, UK and Netherlands as intermediate

performers and Greece, France, Italy and Belgium as the least efficient regions. The

alternative measures that show, instead of measures of managerial (in)ability, cost

levels across countries in order to account for economic and institutional differences.

Such levels tend to highlight average cost differences of 20% between each group;

inefficiency and cost discrepancies are related but deviations are likely to exist.

Moreover, more inefficiency scores are identified in larger banks twice as much as in

135

smaller-sized, and in commercial banks of over double levels with respect to

cooperative and savings banks. However, the cost levels track a different pattern since

savings and investment banks operate with 10-20% higher cost levels than

cooperative and commercial banks. Last, deregulation and financial integration have

brought about inefficiency scores to cover up 45% reduction as opposed to the

respective 10% in cost levels.

In the same period, financial conglomerates exhibit higher operational and profit

efficiency in universal banking services whereas traditional banking confers no cost

disadvantage on other specialized institutions (Vennet, 2002). Moreover, Williams

and Gardener (2003) ended up with high rates of mean efficiency (around 93%) for

regional banks leaving the remainder to correlate with liquidity, capitalisation and

corporate structure.

It is noteworthy the endeavour of a few researchers to employ sophisticated or, at

least, new models that stand out of the common. For instance, De Guevara and

Maudos (2002) examined the inequalities of the European banking sectors by means

of decomposing the Theil index. By doing so, cost efficiency demonstrates great

differences across institutional clusters (savings, cooperative, commercial, other,

which the respective type of productive specialisation and country effects convey

explanatory power to. As for the profit efficiency, it is accredited to country

differences, rather than between specialisation clusters.

In addition, the metafrontier approach has come along in response to some studies

that set the research question „how is it possible to pin down a framework of making

comparison of efficiency across countries accounting for different production

technologies?‟. Such studies are Dietsch and Lozano-Vivas (2000), Lozano-Vivas et

al. (2001) and Bos and Kolari (2005), which early pointed out the mandate to correct

for potentially inherent estimation bias. Bos and Schmiedel (2007) are seeking to

identify whether there exists a unique cost frontier in the European commercial

banking for the period 1993-2004. Accounting for heterogeneity that may influence

efficiency modeling in many other ways (distribution of error term, shifts in frontiers

of bank groups), they establish a framework of comparing efficiency across countries

by assuming a common functional form, but the implied common production

technology is corrected by the location of country-specific frontiers and technology

136

gaps. The results converge to those of country-specific results without standing afar of

a single European frontier as low technology gap ratios predicate. From a policy

perspective, it is sufficiently nationalized to succumb to cross-border mergers once

the results advocate the low probability of banks‟ successful (equal-to-home)

performance abroad. On the other hand, profit efficiency scores are deterministically

explained by competitive power in the local market.

Another methodology followed in the efficiency literature is the accounting ratio

approach. Staikouras et al. (2008a) examined a bulk of explanatory factors/proxies of

size, ownership structure, risk and (non)traditional activities that determine

operational performance in European banking. Notwithstanding the unquestionable

reduction in bank inefficiency, its persistent high levels are positively related to the

loan portfolio quality, size, bank and economic development and negatively

associated with liquidity and loans ratio. Furthermore, Delis and Tsionas (2009)

applied an amended local maximum likelihood approach to simultaneously estimate

market power and efficiency at banking level. This method stems from the studies of

Uchida and Tsutsui (2005) and Brissimis et al. (2008) concerning the gratification and

application of Cournot-related theoretical framework. They examined US and

European markets for the years 1996-2005 confirming the competitive behaviour of

banks and the „quiet life‟ hypothesis. At higher levels of efficiency, the efficiency-

structure hypothesis is endorsed implying that banks exercize higher market power

than average.

Contrary to the vast majority of empirical studies, Maudos et al. (2002) focus on the

estimation of bank inefficiency lying both on the cost and revenue side. For the period

1993-1996, they applied REM, FEM, SFA and DFA to compare cost and profit

efficiencies of ten European countries. The results verify the lower levels of

efficiency in profits than those on the cost side. The variation of scores across

countries is structured by factors of size, productive specialisation, industrial and

economic conditions; higher efficiency is identified in banks of medium size and high

loans to assets ratio. The higher (lower) the market concentration is, the higher the

profit efficiency is (enhanced profitability by exploiting increased market

power/share) and the lower the cost efficiency (indication of „quiet life‟ hypothesis as

banks feel less bound to diminish costs). Profit efficiency is positively associated with

higher risk (proxied by the standard deviation of asset returns) and GDP growth along

137

with cost efficiency being adversely affected by high network density (structural

overheads).

In the same vein, Weill (2004a) ran SFA, DEA and DFA for five European banking

markets, namely France, Germany, Italy, Spain and Switzerland, over the period

1992-1998. Unfortunately, he came up with mixed results that lean towards the

substantially and persistently different estimates of all models, with some exceptions

that corroborate the rule. More research is needed to shed light on the differences of

the underlying efficiency measures as the latter is the outcome of specific

presuppositions of models. However, they have contributed to the tiny niche of the

vast literature that explicitly lays down the comparability of empirical evidence from

different methodologies.

4.4.4. Transition economies

The recent literature of efficiency in the European transition economies enumerates

the predominant application of stochastic frontier methodology with some few

exceptions (Grigorian and Manole, 2002; Stavarek. 2003; 2005). More specifically,

DEA methodology is employed by Grigorian and Manole (2005a) to assess the

performance of commercial banks during the period 1995-1998. As a second stage of

analysis, a censored tobit model was applied to explain cross-country differences in

terms of macroeconomic, regulatory and institutional factors. Evidently, foreign

ownership and consolidation are likely to contribute to the optimal efficiency of the

underlying banking region.

Between 1999-2002, Stavarek (2003) examined the Visegrad region by means of the

aforementioned methodology (DEA, tobit modeling). Given the integration that took

place in the Central Europe since the Communist period, the V4 countries are

reckoned for their heterogeneous structure and their lagging performance with

reference to their developed European counterparts. Slovakia seems to have

demonstrated the poorest performance 28% gap followed by Poland (12%) while

Czech Republic recorded a smooth and stable efficiency of up to 83% over the period,

with Hungary accomplishing to catch up by more than 10% efficiency gains. As for

the efficiency factors, country specific conditions best determine country

138

heterogeneity coupled with the significant impact of profitability, asset size and

foreign ownership. According to Stavarek (2005), with the addition of Portugal,

Greece, Romania and Bulgaria in the previous region for the years 2002-2003, the

first two countries seem to lead the CEE-6 group in exhibiting relatively higher

average efficiency levels. From a policy perspective, better input utilisation is

proposed to tackle the great scale inefficiency of large banks since higher real output

cannot be attained by further betaking themselves to mergers but, alternatively,

developing products and services of high quality.

Delis and Papanikolaou (2009) derived efficiency scores from the application of

input-oriented DEA over the period 1994-2005. Instead of using censored Tobit

model to regress environmental factors on (in)efficiency, the bootstrap procedure is

followed along the guidelines of Simar and Wilson (2007) to resolve the serial

correlation of efficiency scores of the previous step with output and inputs and

therefore with the standard errors of the second stage and the exogenous factors, as

well. Hence, better inference reveals greater explanatory power of bank size

(positive), concentration (negative) and investment environment (positive).

Weill (2002) used the Iterative Seemingly Unrelated Regression (ITSUR) to estimate

the efficiency scores of Poland and Czech Republic in 1994 and 1997. Czech

Republic is persistently reported as the most efficient banking system in the literature

of transition banking markets and Poland turns out to increase cost efficiency by

4.12% but, nonetheless, not sufficiently enough to make up for its lag in 1997 (1.2%).

Over the same period (1996), the degree of openness of the banking industry in terms

of foreign ownership positively determines efficiency scores not by its correlation

with asset size but, rather, by the impartment of know-how and diffusion of

technology. Besides, in the context of cross-country gap in the performance of

Western and Eastern European region, convergence has already taken place between

1996 and 2000; efficiency discrepancies are ill-defined by risk preferences or

environment and best fitted in the degree of managerial competence (Weill, 2004a).

Bonin et al. (2005a) conducted bank efficiency estimations for eleven economies for

the period 1996-2000. Their contribution consisted in the inclusion of time and year

dummies in the cost function and the examination of ownership impact on cost/profit

efficiency. The results give credit to the significant positive effect of asset size and a

139

combined impact of strategic foreign investment, majority foreign ownership and

international institutional participation on efficiency of over 18% on cost efficiency

and of 16% on profit efficiency. Without identifying any causality, foreign strategic

ownership incurs a more robust effect on both profit and cost specifications while

international investors tend to seek for higher returns (profit efficiency). Almost the

same results are confirmed by Dimova (2005) for Central and Eastern Europe with

respect to the performance superiority of foreign banks in terms of managerial

competence, loan portfolio quality and profitability. However, besides the amendment

of operational performance throughout those years, low levels of cost efficiency and

even much lower for profit efficiency are identified by Rossi et al. (2005). According

to Bonin et al. (2005b), greater inefficiency of large state-owned banks constitutes a

„drawing power‟ of privatisation strategies. The cream-skimming hypothesis is

persistent in foreign banking (Borovicka, 2007) whereas de novo and privatized banks

are used to collecting deposits, reducing loan provisioning and offering commission-

intensive services.

Kasman and Yildirim (2006) applied the Fourier functional form of SFA

methodology on commercial banks for the period 1995-2002 to end up with 20.7%

for average cost and 36.7% for profit inefficiency. Still, foreign ownership is charged

for advanced performance in the `90s after controlling for macroeconomic and

reforming conditions prevalent in that period. Along the lines, Yildirim and

Philippatos (2007b) applied SFA and DFA for 12 transition economies to find out

almost the same levels of inefficiencies posing significant positive (negative) effects

of competition on cost (profit) efficiency. Foreign ownership attributes more to cost

reduction rather than profit efficiency, which is often traced in domestic ownership

schemes of both private and public banking entities. On the contrary, Green et al,

(2004) rejected the hypothesis of more efficiency in foreign ownership schemes since

costs reductions have hardly been accomplished by scale and scope efficiency.

Fries et al. (2006) with a model of monopolistic competition verified higher margins

in privatized banks than in de novo and state-owned banks but such differences were

trimmed further down the road; any significant mark-ups were traced in privatized

banks of domestic owners. In addition, state-owned banks failed down to preserve

average mark-ups due to their inability of either keeping up with cost reductions or

attracting demand for their banking services. Holló and Nagy (2006) made a step

140

further drawing remarks in the comparison of 25 European countries over the period

1999-2003. As the cost efficiency gap between the old and new member states tends

to shrink, profit efficiency differences are detected if and only if there exist controls

for market conditions. That highlights the non-competitive pricing of oligopolistic

banking industries of new states and the possibility of whether efficiency gains are

going to move towards either an internally or externally operating process; in other

words, the former prerequisites managerial efficacy to squeeze out cost burden, and

the latter presupposes the passive avail out of the existent financial integration and

economic growth in developing countries.

Vo Thi (2009) and Vo Thi and Vencappa (2008) examine the role of foreign entry

either by mergers and acquisitions or by Greenfield investments in Poland, Czech

Republic and Hungary. Alongside the convergence period of the banking markets,

legal reforms have leant against drags on new market entries. Bank performance, as

broadly defined in terms of profitability, operational efficiency and net interest

margins, turns out to have been significantly affected by M&As and not so by

Greenfield projects at the very beginning of the foreign entry. Hence, cost efficiency

has been increased as well as profits and margins have diminished thereby

overshadowing the minor impact of Greenfield banks in domestic markets. After a

while (4 years or so), bank performance rebounds as M&As follow an inverse U-

shaped path although Greenfield mode of entry convey greater efficiency gains than

M&As and not statistical difference in those of M&As and domestic banks. Even

profit efficiency over the same period is positively correlated with structural reforms

that kicked in towards the liberalisation of the banking industry (Koutsomanoli-

Filippaki et al., 2009).

Staikouras et al. (2008b) and Mamatzakis et al. (2008) examine different emerging

banking markets of the Southeastern European Union for the years 1998-2003. They

applied SFA methodology identifying foreign ownership as a more efficient vehicle

of operational performance with significant differences across members of such

transition groups. As a normal outcome of financial integration, convergence of cost

efficiency scores is evident and contingent on the positive correlations with market

share and bank capitalisation. A more comprehensive study was carried out by

Poghosyan and Kumbhakar (2010), who examined 20 former communist countries on

141

the grounds of the heterogeneous nature of national economic conditions.

Technological regimes are classifying banking markets in clusters in line with the

latent class SFA methodology developed by Orea and Kumbhakar (2004); evidence

shows that inefficiency is attributed to the combination of otherwise distinct and

significant effects of technology differences, risk and market concentration.

4.4.5. Country – specific studies

Austria has drawn the attention of Hahn (2007) and Egger and Hahn (2010), who

endeavoured to quantify the endogenous effect of mergers and acquisitions on

banking performance. The former study samples banks over the period 1996-2002

and utilizes a two-stage analysis: DEA methodology and tobit regressions to account

for the underlying variation. Banks that engaged in domestic restructurings

experienced a higher productive efficiency than those, which did not. The latter study

complements the analysis with panel methodology to conclude that horizontal

mergers were evident over the same period producing greater cost performance gains

to smaller banks and, especially, before the merger takes place.

Podpiera and Podpiera (2005) estimated mean cost efficiencies of Czech Republic‟s

banks running three models: the stochastic frontier, the random and the fixed effects

models. They used their findings in the model of Cox proportional hazards to explain

to what extent cost efficiency affects the risk of bank failure. Statistically robust

results underline poor cost management one year before the subsequent failure almost

of all banks as a legitimate signal of warning. Pruteanu-Podpiera et al. (2007)

estimated competition with Lerner index and applied Granger-causality model to

verify the banking specificities and quiet life hypothesis. The results endorse the

former, that is the negative relationship of efficiency and competition, as fierce

competition dismantles relationship banking, making thereby monitoring costs

unbearable, and leave scale economies unexploited. Moreover, the evolution of

competition in such a transition economy over the period of 1994-2005 is pronounced

as falling concentration till 2000 and then turned off due to mergers and failures.

However, the European passport made possible the cross-border establishment of

142

branches and subsidiaries and hence, enhanced competition by the provision of

services and loan products.

For the Croatian banking sector, Kraft and Tirtiroglu (1998) and Kraft et al. (2006)

use stochastic frontier analysis employing the translog and the Fourier-flexible form,

respectively. In times of transition, the old and state-owned banks have performed

substantially in the industry although the regulatory framework, political and

macroeconomic conditions constitute major hindrance against competitiveness.

Privatisation to foreign investors could have brought about large efficiency gains had

liberalisation policies eliminated the free-riding problem. Nonetheless, Kraft et al.

(2006) in the period 1994-2000 attributed the profitability of the Croatian banking

system to bank management that treated costs and inherent risks prudently. In

contrast, the DEA methodology of Jemric and Vujcic (2002) found during the same

period the new banks to be more efficient than old ones and the smaller banks

globally efficient; large banks are, too, efficient once we consider variable returns of

scale. In the same vein, Primorac and Troskot (2005) applied DEA analysis to observe

relative technical efficiency between the range [34.46%, 50.92%], though in the

Malmquist productivity index the decomposed components - the change of efficiency

and technology – favours, on average, the former with 142% vis-à-vis the latter

(88.79%) over the examined period 2000-2003.

The French banking industry enumerates two studies. Dietsch (1993) demonstrated

the evidence of scale and scope economies, though of limited contribution.

Furthermore, universal banking tend to have the most competitive advantage amongst

the alternative productive specialisations. Brack and Jimborean (2010) juxtaposed

bank efficiency in French banks to that in Europe and the United States. Data

envelope analysis shows them to be relatively more efficient with the gap to be

narrowing from 1994 to 2000. Tobit regressions of efficiency scores on various

factors underline new capitalized banks, operating in countries with low GDP per

capita and employing austere Tier 1 capital ratios, to be the most cost-efficient.

Lang and Welzel (1996) analysed the efficiency and technical progress of the

cooperative banks of Germany using a translog multi-product specification for the

years 1989-1992. Evidence shows significant scale and scope economies most

favouring universal banking and moderate economies evident in all size classes. The

143

degree of cost efficiency is far away from the frontier. The classes of small-sized

banks tend to achieve greater productivity gains in terms of cost reductions due to

recorded technical progress. Fiorentino et al. (2006) compared the results of SFA and

DEA methodologies during the period 1993-2004. The most crucial issues involving

the underlying comparison is the heterogeneity of banking groups, time variation,

random errors and outliers. In particular, both approaches perform best for distinct

groups per year recording correlation in their ranking up to 44%. DEA is supported in

lieu of SFA if the researcher anticipates great measurement errors whereas the latter

fits in the cases of time series, present outliers and benchmarking whole banking

systems. Last, Lang and Welzer (1996) applied a panel-based SFA for the mergers

wave of cooperative banks during the period 1989-1997. The concluding remarks

show size effects in a merger if that endeavour closes part of the acquired bank‟s

branches, pre-merger motives and post-merge performance unfavouring efficiency

gains but rather a convergence in their operational performance. Bos et al. (2009)

accounted for heterogeneity in German cooperative and by plugging vectors of

variables in the deterministic kernel and in the exogenous part to affect the

distribution of the error term. Eventually, size, location and specific bank type explain

the significant systematic differences over the period 1993-2005.

The Hungarian experience was studied by Hasan and Marton (2003) throughout its

transition period towards a more liberalized and market-oriented system. During 90‟s,

banking system was reorganized through initiatives of privatisation and policies that

liberalized the status quo of corporate governance towards foreign ownership, or at

least high involvement of it. In line with the results of Claessens et al. (2001a),

foreign banks were more efficient and profitable that the domestic state-owned

structures of a transition economy like Hungary. In this way, new financial services

and business lines coupled with other comparative advantages of the region promoted

competitiveness of the banking sector.

The multiproduct bank of Ireland has been the subject of Glass and McKillop (1992),

in which the presence of scale and scope economies was investigated. The empirical

results show total diseconomies of scale in response to branch closures for cost-

cutting purposes occurring in that period; however, significant product-specific scale

efficiency traced to loans and inefficiency to investments. As for the scope

144

economies, there exist no cost complementarities, thus literally suggesting

diseconomies of productive activities.

Bos and Kool (2006) investigated the efficiency of the banking sector in Netherlands

for the years 1998-1999. Their method is based upon stochastic frontier analysis but

incorporating the exogeneity of input prices rather than endogeneity adopted widely

in the vast literature. Next, the inefficiency scores are regressed on indicators of

macroeconomic conditions, industry specific features and strategic choices.

Netherlands appears to be a region of larger bank inefficiencies with the new

methodology employed and, profoundly, more ascribed to managerial X-inefficiency.

Greek banks have been examined quite extensively since 2008. Apart from it,

Christopoulos et al. (2002) employed a translog cost function accounting for

heteroskedasticity in the error term; results show heteroskedastic inefficiency to be

appropriate for capturing the efficiency effects of bank‟s size. Large banks are less

efficient than small and medium-sized and in order to recoup some gains the

reduction of the asset portfolio is suggested. In addition, economic performance,

investments and loans are positively related to cost efficiency. Kamberoglou et al.

(2004) analysed the efficiency of Greek banks during the period 1993-1999, in which

the sector was exposed to financial deregulation in line with the economic

convergence towards EMU. Differences in inefficiency, that indirectly constitute the

unexploited potential, seem to be determined by ownership (public), size (large), risk

attitude (aversion), technological innovations and scale economies. The results are in

line of Noulas (1997), who found small banks to be more efficient and larger ones

being beneficially affected by technological progress. Giokas (2008a; 2008b) and

Noulas (2008) investigated the efficiency impact of Greek bank branches on

generating profits. On average, their operational performance has been poor hitting up

to 30%, rural regions do play a positive role on efficiency and, within branch

networks, larger and more profitable branches tend to be more efficient. Kosmidou

and Zopounidis (2008) highlighted the relatively strong tendency of performance-

ranked commercial banks to expand their customers‟ base and competitively operate

and gain profits; the results of cooperative banks were inconclusive as their profits

and market shares follow adverse pace. In addition, Pasiouras et al. (2008) regressed

the annual returns of bank share prices on the change of efficiency controlling for size

and risk; in a nutshell, the underlying relationship was significant for the period 2001-

145

2005 with no scale efficiency effects recorded. Chortareas et al. (2009) analysed

quantitatively the levels of bank efficiency utilising DEA and TFP methodologies.

Over the years 1998-2003, Greece demonstrated 4.3% increase in cost efficiency and

gains of profit efficiency by 93%; last, productivity has been enhanced by 15% with

off-balance sheet items showing no significant effect.

In the run-up of Polish banks towards the deregulation of its industry, Havrylchyk

(2006) found no evidence during 1997- 2001 in favour of cost efficiency gains and its

components, namely technical, allocative, pure technical and scale effects. Contrary

to these findings, Pawlowska (2003) examined the impact of mergers and acquisitions

on cost efficiency and productivity change and underlined the upward shift of all

efficiency and productivity measures, especially though amongst the large banks that

traditionally engage in M&As and perform higher efficiency already. The key

influencing factors of efficiency are recorded to lie upon bank‟s size and foreign

ownership. Along the same lines, Figueira et al. (2007) corroborate to the competitive

efficiency of polish banks with respect to the benchmark of UK banks‟ performance.

In the advent of economic changes, the banking industry seems to pass the „stress‟ test

of EU competition.

In the Portugal case, Canhoto and Dermine (2003) employed DEA methodology to

assess the efficiency of old and new banks between 1990 and 1995. The performance

of foreign and new banks (77%) dominated that of old ones (62%) recording a stable

tendency over time. Portela and Thanassoulis (2007) innovated upon DEA by

decomposing performance of bank branches into the transactional, operational and

profit components. The results show positive interconnections among service quality,

profit and operational efficiency; however, although there exists a positive

relationship between operational and transactional performance, the latter is,

surprisingly, not affected by service quality. On the other hand, SFA methodology

enumerates the studies of Mendes and Rebelo (1999), De Pinho (2001) and Boucinha

et al. (2009), Lima (2008) and Lima and Pinho (2008) along with Mendes and Rebelo

(2000), who applied them both. Pinho (2001) shows an increase in average efficiency

of Portuguese banks, identifying scale economies for smaller banks and cost

complementarities of loans and deposits. Rebelo and Mendes (2000) found strong

productivity progress both for small and large banks, positive efficiency shares of

bank consolidation (Lima, 2008) but comparatively greater productivity and

146

technological change for urban banks. In contradistinction, during the same period,

Mendes and Rebelo (1999) realized no profound evidence of stable increasing path of

efficiency and exact causality on size and cost efficiency since inefficient smaller

banks enjoy economies of scale and efficient larger institutions suffer by

diseconomies of scale. In the latest study of Boucinha et al. (2009) for the period

1992-2004, banks were found to be more efficient in the case of low credit risk, idle

liquidity and increased capitalisation; the latter is likely to explain the alleviation of

agency problems and the ability to attract the most efficient management. On average,

banks record total productivity increased by 31.4%, inefficiency levels of 9% and

higher share of technological innovation on cost minimisation (3.2% from 2.2%).

Romanian banks are analysed in terms of their performance during 1996-2002 as a

response to the tightening of regulatory framework by the National Bank of Romania.

Notwithstanding stochastic frontier analysis showing policy changes to operate

against technical inefficiency, such an effect is dominated by the short-run costs of

additional regulatory burden; indeed, technical inefficiency share on total costs has

been substantially diminished from 41% to 21%. After the examined period, the

restrictive framework paid off in the form of constructing a more sound mechanism of

absorbing shocks and limited losses of loan portfolios (from 35.4% to 1.1%).

In Sweden, Bergendahl and Lindblom (2008) measured the service efficiency of

savings banks and the extent to which the results are differentiated from a profit

oriented evaluation of a commercial bank. Employing a DEA methodology, more

small and medium-sized banks are service efficient, rather than profit efficient,

though the average efficiency scores tend to coincide over the period. Large savings

banks face fierce competition and exhibit lower profit efficiency as well as service

efficiency due to their operation in large networks of populated districts.

Rime and Stiroh (2003) examined the case of Switzerland, pooling a sample of banks

for the period 1996-1999. They applied stochastic frontier analysis with translog cost

and profit functions and distribution-free specification for the inefficiency component

of the error term. The evidence supports significant profit and cost efficiency as well

as economies of scale for the small and medium-sized banks. Greater dispersion in

performance is observed in large banks whereby the initiatives of consolidation and

restructurings explain the dominating trend over the decade. The proliferation of

147

universal banks with a broad mix of outputs – e.g. brokerage, OBS items, portfolio

management and trading - falls short of reaping the benefits of scale and scope

efficiencies.

The paper of Mertens and Urga (2001) examined the whole spectrum of efficiency

dimensions of the 79 commercial banks in Ukraine. They applied stochastic frontier

analysis and thick frontier approach for the year 1998, drawing remarks about the

small banks being more cost efficient and at the same time less profit efficient than

large banks; the latter is possibly explained by the presence of monopoly power in the

banking industry. Another finding is the remarkable dispersion of economies of scales

between differently-sized banks: large banks face diseconomies of scale while smaller

ones great scale efficiency. That, in part, could be explained by the hurdle of current

technology to diffuse in the production process and evolve in an efficient way. Market

concentration is, therefore, far from taking significant levels, though being the case

for the market of small banks.

During the period of 1982-1995, Drake (2001) and Webb (2003) conducted DEA

analysis for UK banks differentiating with each other by applying window analysis

and Malmquist productivity indices between periods, respectively. They both

corroborate on the findings that overall productivity has increased since „80‟s,

technically the upward frontier shifts dominating the negative catching-up effect.

Excess capacity has been a long-lasting problem and rationalisation in terms of pure

technical efficiency takes time to give out tangible results, and if any, they are

identified in banks clustering above 100bn asset sizes. Both consent to the thorny

problem in UK banking that scale inefficiency affected severely very large banks as

well as very small banks according to Drake (2001).

148

4.5. Conclusion

The methodology of competition modeling comprises the issues of market structure

and efficiency. The key question arises regarding the potential effects of

concentration and efficiency, as decomposed into economies of scale and scope

economies, technology, management practice, on bank behaviour. Conditional on the

line of reasoning, viz. SCP paradigm and efficiency hypothesis, several measures of

concentration and efficiency approaches are used to verify their explanatory power on

competitive conditions. However, the employment of particular methodology implies

particular shortcomings and advantages, according to which the empirical analysis

should be treated with cushion.

In particular, to the best interest of research questions, CRk and HHI are the most

widely used in the literature, without overlooking the consistency of all the others to

markets of different features. Efficiency stands out as a stand-alone theme of financial

literature, comprising traditional data envelope analysis, stochastic modeling and

semiparametric methodology. Whether it is the case of deterministically estimating

efficiency from deviations from the observed best practice, defining the functional

form of the inefficiency component of the random error or combining those two in

two-stages procedures, among others, technical and allocative efficiency may be the

source of survival in fierce competition or the ultimate end of prospective

consolidations amid conditions of integrated and deregulated European market.

149

Chapter 4 appendix

Table of reviewed papers

Author(s) Approach Period Countries

International comparisons

Allen and Rai

(1996)

SFA (distribution-

free) 1988-1992

AUSTR, AUS, CAN, SWISS, GER, DEN,

SP, IT, SW, JAP, BEL, FIN, FR, UK, US

Amel et al. (2004) Review – efficiency gains from M&As

Berger (2007) Review

Berger and

Humphrey (1997) Review – performance effects from M&As

Correa (2009)

Fitzpatrick and

McQuinn (2008) SFA (ML) 1996-2002 IR, UK, CAN, AUSTR

Gonzalez (2009)

EFF and Political

economy effects

on MS, CONC

1996-2002 69 countries (Europe, US, Latin and

Carribean countries, Asia)

Hahendorff and

Keasey (2009)

M&A effects on

performance 1996-2004 US, EU-15 plus SWISS

Kasman et al.

(2005) SFA (Fourier) 1996-2001 16 Latin and Carribean countries

Lensink et al.

(2008) SFA (translog) 1998-2003 105 countries worldwide

Lozano-Vivas and

Pasiouras (2010) SFA (translog) 1999-2006 87 countries

Maudos and

Pastor (2001) SFA (translog) 1984-1995

Europe, Japan, US

Meon and Weill

(2005) SFA 62 countries

Pasiouras (2008b) DEA 2003 95 countries

Pasiouras et al.

(2009) SFA 2000-2004 74 countries

Pastor and

Serrano (2006)

Specialisation

effect on

efficiency

1992-1998 GER, AUS, BEL, DEN, SP, FR, GR, IT,

LUX

Pastor et al.

(1997) DEA 1992 US, Europe

150

Developed economies

Berg et al. (1993)

DEA (malmquist

index)

1990 Finland, Norway, Sweden

Fried et al. (1993)

Berger and Mester

(1997)

Alternative SFA

(profit) 1990-1995 US

Mester (1997) SFA (translog) 1991-1992 US

Altunbas et al. (2001) SFA (Fourier) 1989-1997 EU - 15

Berger and DeYoung

(1997)

Granger causality

SFA (Fourier - OC)

Loans-efficiency

hypotheses

1985-1994 US

Berger and Humphrey

(1997) Review Before 1997 21 countries

Guevara and Maudos

(2002) Theil index 1993-1997 EU-14

Lozano-Vivas et al.

(2002) DEA 1993

Belgium, Denmark, France,

Germany, Italy, Luxembourg,

Netherlands, Portugal, Spain, UK

Maudos et al. (2002) DFA, FEM, REM 1993-1996 EU-10

Carbo et al. (2002) SFA (Fourier) 1989-1996 EU-12 (savings)

Bikker (2002) SFA - TCF 1990-1997 EU-15

Vennet (2002) SFA (translog) 1995-1996 EU-17

Casu and Molyneux

(2003) DEA 1993-1997

France, Germany, Italy, Spain,

UK

Williams and Gardener

(2003) SFA (Fourier) 1990-1998

Denmark, France, Germany, Italy,

Spain, UK

Weill (2004a) SFA (Fourier), DFA,

DEA 1992-1998

France, Germany, Italy, Spain,

Switzerland

Tomova (2005b) DEA 1994-2002 EU-25, accession countries

Bos and Schmiedel

(2007) Metafrontier approach 1993-2004 EU-15

Staikouras et al.

(2008a)

Regressions with

accounting ratios 1998-2005 EU-10, EU-15

151

Dellis and Tsionas

(2009) LML technique 1999-2006 EMU, US

Transition economies

Bonin et al. (2005a) SFA - ownership 1996-2000

Poland, Croatia, Hungary, Romania,

Bulgaria, Lithuania, Slovenia,

Slovakia, Latvia, Estonia, Czech

Republic

Bonin et al. (2005b) SFA 1994-2002 Bulgaria, Czech Republic, Croatia,

Hungary, Poland, Romania

Borovicka (2007) SFA 1995-2004 19 European transition countries

Dimova (2005) Ratio approach - SFA 1997-2002

Poland, Hungary, Czech Republic,

Estonia, Latvia, Lithuania, Bulgaria,

Romania, Slovenia, Slovakia,

Fethi et al. (2008)

Various

methodologies – see

references

1998-2003

Poland, Czech Republic, Estonia,

Latvia, Lithuania, Hungary, Slovenia,

Slovakia,

Fries and Taci (2005) SFA 1994-2001 15 East European countries

Fries et al. (2006) SFA (REM, FEM) 1995-2004 Post-communist countries in Eastern

region

Dellis and

Papanikolaou (2009) DEA, bootstrap 1994-2005 10 newly acceded countries

Green et al. (2004) SFA 1995-1999

Bulgaria, Croatia, Czech Republic,

Estonia, Hungary, Latvia, Lithuania,

Poland, Romania

Grigorian and

Manole (2002) DEA 1995-1998

Central, Southeastern Europe plus the

Commonwealth of Independent States

Hollo and Nagy

(2006) SFA (Fourier) 1999-2003 25 European countries

Kasman and Yildirim

(2006) SFA (Fourier) 1995-2002

Czech Republic, Estonia, Hungary,

Latvia, Lithuania, Poland, Slovak

Republic, Slovenia

Koutsomanoli-

Fillipaki et al. (2009) SFA 1999-2003

Hungary, Poland, Czech Republic,

Slovakia

Mamatzakis et al.

(2008) SFA 1998-2003

Cyprus, Czech Republic, Estonia,

Hungary, Latvia, Lithuania, Malta,

Poland, Poland, Slovak Republic,

Slovenia

Poghosyan and

Kumbhakar (2010) SFA 1995-2005 20 former socialist countries

152

Rossi et al. (2005) SFA 1995-2002 9 Central and Eastern European

countries

Staikouras et al.

(2008b) SFA 1998-2003

Bosnia-Herzegovina, Bulgaria, Croatia,

FYR of Macedonia, Romania, Serbia–

Montenegro

Stavarek (2003) DEA 1999-2002 Czech Republic, Hungary, Poland,

Slovakia

Stavarek (2005) DEA 2002-2003

Portugal, Greece, Czech Republic,

Hungary, Poland, Slovakia, Bulgaria,

Romania

Tong Wu (2006) SFA, SURE 2000-2004 7 accession and 10 non-accession

countries

Von Thi (2008) Accounting measures 1994-2004 Czech Republic, Hungary, Poland

Von Thi and

Vencappa (2008) SFA 1994-2004 Czech Republic, Hungary, Poland

Weill (2002) SFA 1994, 1997 Poland, Czech Republic

Weill (2004a) SFA (Fourier) – input

shares 1996, 2000

11 Western and 6 Eastern European

countries

Yildirim and

Philippatos (2007b) SFA, DFA 1993-2000

Czech Republic, Estonia, Croatia,

Hungary, Lithuania, Latvia, FYR

Macedonia, Poland, Romania, Russian

Federation, Slovenia, Slovakia


Egger and Hahn

(2010) Matching framework 1996-2002 Austria

Hahn (2007) DEA 1996-2002 Austria

Podpiera and

Podpiera (2006) SFA, REM, FEM 1994-2003 Chech Republic

Prutaneu-Podpiera et

al. (2007) SFA (translog) 1994-2005

Chech Republic

Podpiera and Weill

(2008) GMM panel 1994-2005

Jemric and Vujcic

(2002)

DEA

Croatia Kraft and Tirtiroglu

(1998) SFA (translog) 1994-1995

Kraft et al. (2006) SFA (Fourier) 1994-2000

153

Primorac and Troskot

(2005) Malmquist index 2000-2003

Dietsch (2003) SFA (translog) 1987

France Jimborean and Brack

(2010) DEA 1994-2006

Fiorentino et al.

(2006) SFA (translog), DEA 1993-2004

Germany Lang and Weizel

(1996) SFA (translog) 1989-1992

Lang and Weizel

(1999) SFA (translog) 1989-1997

Chortareas et al.

(2009)

DEA,

Malmquist index 1998-2003

Greece

Christopoulos et al.

(2009)

SFA (translog –

hetegeneity) 1993-1998

Giokas (2008a) DEA 2001

Giokas (2008b) DEA 2002

Kamberoglou et al.

(2004) FEM 1993-1999

Kosmidou and

Zopounidis (2008)

Promethee method

(CAMEL rating

system)

2003-2004

Noulas (1997) Malmquist index 1991-1992

Noulas et al. (2008) DEA 2000-2001

Pasiouras et al.

(2008) DEA 2001-2005

Rezitis (2008) Output distance

function 1993-2004

Hasan and Marton

(2003) SFA (translog) 1993-1998 Hungary

Glass and McKillop

(1992) SFA (hybrid translog) 1972-1990 Ireland

Bos and Kool (2006) SFA (translog) 1998-1999 Netherlands

Figueira et al. (2007) SFA (translog) 1999-2004

Poland Havrylchyk (2006) DEA 1997-2001

Pawlowska (2003) DEA 1997-2001

154

Boucinha et al.

(2009) SFA (translog) 1992-2004

Portugal

Canhoto and Dermine

(2003) DEA 1990-1995

Portela and

Thanassoulis (2007) DEA 2001-2002

Mendes and

Rebelo(1999) SFA (translog) 1990-1995

Rebelo and Mendes

(2000) Malmquist index 1990-1997

de Pinho (2001) SFA (translog) 1986-1992

Asaftei and

Kumbhakar (2008) SFA (translog) 1996-2002 Romania

Bergendahl and

Lindblom (2008) DEA 1987-1992 Sweden

Rime and Stiroh

(2003) SFA (translog), DFA 1996-1999 Switzerland

Beccalli (2004) SFA (translog) 1995-1998

UK

Drake (2001) DEA – Malmquist

index 1984-1995

Fitzpatrick and

McQuinn (2005) DEA – SFA (translog) 1996-2001

McKillop et al.

(2002) DEA 1996

Webb (2003) DEA window analysis 1982-1995

Mertens and Urga

(2001) SFA (translog) 1998 Ukraine

155

CHAPTER 5

The nexus between bank competition and financial stability

157

5.1. Introduction

Banks may be a channel-through conducive for instability and welfare loss11

in the

economy once considering liquidity, credit pause drying out payment services and

interbank lending, among others. Its definition may be conceptually clear depending

on the angle we evaluate incidents of financial distress. For example, a brief overview

should comprise financial stability to be a situation of information asymmetries that

jeopardize asset allocations (Mishkin, 1999), of systemic risks (domino effect)

inherent in the interconnections of financial institutions that witness distress of few

banks (Summer, 2003), of asset pricing that serves as a means of optimally planning

savings and investments (Haldane, 2004).

In addition, the last mergers wave has rendered the analysis of concentration on bank

risk-taking an imperative to absorb shocks of potential real output losses since

consolidation plans are subject to the approval of antitrust authorities. However, the

design of the so-called macroprudential regulation12

and supervision comes into play

to configure prudent conduct of banks as well as monetary and fiscal policies in order

to create indispensable synergies of all the relevant authorities.

5.2. Theory

The analysis13

of the underlying relationship stems from the seminal contribution of

Keeley (1990) following those of Marcus (1984) and Chan et al. (1986), who

proposed the „franchise (charter) value‟ paradigm, that is if the diminishing market

power of banks, in line with the emergence of great competitive pressures, squeezes

profit margins, banks in order to recoup increased returns take refuge in risky

projects. Bank failures are likely to occur when adverse selection and moral hazard

problems indicate that banks are getting more reluctant to monitor borrowers and

thereby falling short of exploiting the benefits of relationship banking (Boot and

Greenbaum, 1993). Thus, loan portfolios tend to comprise marginal applicants and

11

See Haldane et al. (2004) for the economic repercussions of financial stability and resolution

practices. 12

See Borio (2006) and Borio and Lowe (2002) for further discussion. 13

See Carletti and Hartman (2003), Beck (2008) and Vanhoose (2010) for a comprehensive theoretical

and empirical review.

158

potentially exacerbate the risk exposure (Allen and Gale, 2000). In contrast,

monopolistic banking markets promote the prudent conduct of banks in engaging in

less risky projects amid conditions of more profit-making opportunities and capital

cushions (Beck, 2008).

However, another severe source of instability is traced at the liability side; thanks to

the ongoing deregulation of financial markets, interest rates have slumped on the sly

ever since the endeavour of removing entry barriers and expanding restrictions

flourished. In such a situation, banks strive to curb low franchise value and

profitability engaging in riskier asset allocation given that in hard times of insolvency

and banks runs, deposit insurance schemes are stand by to intervene. Hence, it is

deemed to be essential for the authorities to impose restrictions on deposit

competition to discourage „gambling for resurrection‟ (Cole et al., 1995).

Matutes and Vives (2000), in an imperfect competition model, show that deposit

ceilings can produce some welfare gains in an uninsured competitive market when the

social cost of failure is prohibitive. With the employment of deposit insurance

schemes at a fixed rate deposit competition is escalating insofar as risk-taking and

profit margins can afford (Cordella and Yeyati, 2002). In the presence of risk-based

insurance on deposits, the prudent behaviour of banks can be restored improving

further social welfare. In the same vein, Hellmann et al. (2000) allege that the

imposition of capital requirements may have adverse effects on decreasing franchise

values inducing banks to embark on gambling behaviour. However, deposit-rate

controls are more Pareto-efficient instruments that promote prudency in investments

even off the equilibrium path. Even when there exist only large institutions in the

market, supervisory authorities should keep a vigilant eye on their prudent operation

while the limited role of deposit insurance schemes should not undermine their per se

existence along with government safety net in times of emergency (Mishkin, 1999).

In addition, they examine the welfare implications of deposit insurance equilibria:

notwithstanding the positive impact of insurance in preventing crises, mitigating

transport costs and extending the market, deposit insurance guarantee that all banks

are credible. Hence, in the absence of no expected diversification economies to

exploit, all banks are discounted at the same rate and the resulting higher competition

hits high failure probabilities. If decisive regulatory authorities allow for takeovers of

159

the failed banks, the incentive of behaving prudently dominates that of profiteering by

taking risks. Hence, there is a trade-off between the so-called „last bank effect‟ and

risk-taking on the grounds that competition is treated as endogenous: banks are

anticipating the failure of the others to reap short-term benefits from the resulting

concentration. (Perotti and Suarez, 2002). Should consolidation policies be coupled

with vibrant easy entry ones, the temporarily enhanced charter values of incumbent

banks will slump along the way of restoring long-run competition.

According to Allen and Gale (2000; 2004), it is Pareto optimal, though socially

undesirable, to have instability in cases of perfect competition and complete markets.

Contagion and welfare loss might well be an outcome triggered by the systemic risk

of a liquidity shock on a bank if all banks in the market rationally act as price takers

and are reluctant to liquidate its assets through the interbank market (Besanko and

Thakor, 1993). However, the alternative to rescue a troubled bank with liquidity

deficiencies by the coordination of few banks is strategically feasible (Saez and Shi,

2004).

On the top of that, Matutes and Vives (1996) consider the self-fulfilling expectations

of depositors to endogenously affect the quality – or the failure probability - of a

bank. In other words, the bank with high depositors‟ trust will enjoy higher margins

and greater market share as perceived to be diversified and hence safer in their eyes.

Such endogenous evaluation is not contingent on the perception that in concentrated

markets few banks are too big to fail or can freely exploit more opportunities;

multiple equilibria may come up even in the case of market concentration. The

assumption that concentration is conducive for lower competition is erroneous but, at

least, possible to bring about financial stability if the preservation of long-term

relationships can exploit the private and exclusive information about the way liquidity

needs are probabilistically distributed (Smith, 1984).

Even it is the case that anticompetitive conduct is not an inherent feature of large

banks, a relative market structure may induce stability effects in the presence of well-

diversified portfolios and economies of scale (Diamond, 1984; Williamson, 1986).

The size, therefore, does matter in concentrated markets and recent economic history

advocates fewer episodes of bank insolvency and runs occurred in Canada than in

U.S. (Allen and Gale, 2000). Capie (1995) show less competitive market structures in

160

UK to be stable during the period 1840 and 1940 and sustainably profitable in the past

few decades as compared to the variable performance of German banks. The

argument here is in favour of large banks and concentrated markets, upon which

supervisory authorities can bear the burden of monitoring and regulating.

Along the lines, Boyd et al. (2004) underscore inflation as another key determinant of

bank failure irrespectively to the underlying banking system. When the nominal rate

of interest (inflation) is below a certain threshold, a relatively higher probability of

bank failure is present in monopolies on the grounds that the incentive of loaning out

cash reserves dominates that of paying low rates on deposit accounts. Secondly, asset

loss is greater in competitive structures in times of a crisis, as monopolies tend to

make profits upon the liquidation of assets (e.g. deposits) except cash, for they are

able to provide inter-temporally much lower deposit insurance.

The other strand of literature contending the competition-stability nexus emanates

from Stiglitz and Weiss (1981), who showed that monopolistic market structures are

to be blamed for great charges on loans and thereby upcoming defaults. Safe

borrowers are repelled by high borrowing costs and information asymmetries render a

significant part of loans nonperforming and hence constituent of financial instability.

Boyd and De Nikolo (2005) employed the loan market channel in their analysis to

conclude that the positive relationship between risk and competition is fragile. As

monopolistic structures increase loan rates, borrowers surrender to riskier projects.

Thus, the probability of default rates is conditional on banks‟ pricing conduct in the

loan markets.

That is not the case for the model of mean shifting investment technologies by

Koskela and Stenbacka (2000), since higher competition diminishes the loan rates

without necessarily triggering default risk in equilibrium out of the increased

demanded volume of investments. Caminal and Matutes (2002) found that

monopolistic markets, bearing the costs of monitoring, tend to be more susceptible to

risky loans and thereby subsequent failures. Thus, failing credit rationing, bank

willingness to loan out places the bank course in jeopardy. On the other hand, De

Nikolo and Luchetta (2009) proved out of a general equilibrium framework that

efficiency, optimal portfolio quality and diversification rents best feature competitive

markets, though non-perfect competition constitutes second-best alternative.

161

Under the perspective of supervision and precautionary regulation, interesting

arguments have been articulated in the literature. The assertion that it is easier to

monitor few and large banks rather than much more and smaller banks, does not

cohere with operating complexity of large entities. Under conditions of expanding

restructurings, universal banks and conglomerates provide the whole spectrum (or

part of) of financial services - e.g. consulting (M&As), instrument and proprietary

trading (derivatives included), stock broking, investment management, insurance –

which have been previously offered by more specialized banks, such as commercial,

investment, and merchant banks. The subject is apparently multifarious for

supervisory authorities to inspect and timely intervene rendering financial instability a

presumptive reality.

However, the effect of deregulation and in particular the removal of branching

restrictions in US on banks‟ risk taking is likely contingent on the corresponding

market structure and organisational form (Goetz, 2010). His theoretical model meets

econometric verification that the liberalisation of intrastate restrictions affects risk

taking downwards for banks branching in other states but not following the trend of

other competitors to expand in real estate markets. That is even more pervasive in

markets where the information for borrowers is less verifiable.

In contradistinction to the aforementioned mandate of having concentrated markets at

the convenience of authorities, policies are likely to lean towards the provision of

large subsidies and bail-outs in an attempt to eschew from impawning the viability of

too-big-to-fail banks. Under the auspices of such tutelar policies, bank managers have

an incentive to take risks potentially exacerbating the contagion risk of large and

insolvent banks (Mishkin, 1999).

Without constraining ourselves to surrender to any side of contending hypotheses, it

could be the case that there exists a matching point. Martinez-Miera and Repullo

(2010) concur that there exists a U-shaped relationship between competition and bank

failure risk. In particular, monopolistic markets experience the risk-shifting effect,

that is more competition with low loan rates stabilizes banks as they run less risk of

default whereas the margin effect – lower revenues of total non-defaulting loans may

jeopardize banks in view of potential entries – occurs usually in competitive markets.

162

5.3. Empirical review

The pertinent literature of empirical analysis on whether it is the case that bank

competition drives to financial stability, enumerates several - though not adequate -

studies on European Union, USA, Asia and Latin America. Recent studies that focus

on regions outside European territory are Yeyati and Micco (2007) for Latin

American countries and Cerqueiro et al. (2008) for the US banking industry. In

addition, Boyd et al. (2009) and De Nikolo and Loukoianova (2007) conducted

broader investigation of 133 non-industrialized countries for the period 1993-2004.

However, the developed economies comprising several European states are analysed

by Beck et al. (2006a), Berger et al. (2009), Laeven and Levine (2009), Schaeck and

Cihak (2008) and Schaeck et al. (2009). Last, other papers deal with the EMU – e.g.

Agoraki et al. (2009)(CEE), Cipollini and Fiordelisi (2009), Uhde and Heimeshoff

(2009) - and/or specific European countries like Germany (Beck et al., 2009) and

Spain (Jimenez et al., 2010).

5.3.1. International studies

Other regions outside the EU have triggered less attention regarding contemporary

issues of financial stability partly due to the immature operation mechanism of

European financial markets under the umbrella of a single currency union. Hence,

Cerqueiro (2008) examined why concentration in US bank markets gives out peculiar

effects on loan supply. The reason comes into play providing interesting insights for

the research of financial stability: the credit quality is contaminated once banks raise

loan rates and are willing to grant loans to induced borrowers of relatively less credit

worthiness. In contrast, the Latin American territory witnessed high levels of

concentration and foreign penetration due to the operation of foreign branches and

subsidiaries (Yeyati and Micco, 2007). Τhe competitive conduct of banks lost ground

and the positive correlation of risk and competition highlights a positive relationship

between foreign banking and financial stability notwithstanding the recorded high-

risk indicators of foreign ownership.

Boyd et al. (2009) run several cross-section and panel regressions for US and

international samples. The common idea of the former is the set up of two models,

163

that is one running OLS with fixed effects and at the same time setting as dependent

variables Z-scores and loans to assets ratios, and the other utilising Cox

transformation of the loans to deposits ratio to capture the location-specific effects of

state clustering. The international sample is used for static and dynamic forms of

panel regressions; including HHI and other controls, the former type consists in two

specifications: one with all variables and the other omitting firm-specific variables.

The latter type applies the GMM procedure set out by Arellano and Bond treating all

dependent and independent variables as predetermined and, thus, instrumented by

their lags. The resultant trend of the empirical evidence lies in the negative

association of competition measures and probability of bank failures (Z-score). Other

robustness tests suggest higher competition implying either higher overall bank risk or

default probability of bank‟s borrowers (moral hazard effect) according to the way we

interpret the indication of loan losses ratio. In addition, notwithstanding the

ambiguous effect sign of loans-to-deposits ratio on competition, the propensity of

banks to lend is empirically favoured by competitive markets.

De Nikolo and Loukoianova (2007) articulated in a general equilibrium model the

significance of ownership in the competition-stability nexus since the underlying

effect seems much stronger for large and state-owned banks. They run several sets of

panel regression models examining the conditional means of intercepts and HHI beta

coefficients with respect to ownership type even after the inclusion of time-invariant

bank-specific controls. Last, the analysis is further enhanced by allowing for bank

heterogeneity: how much the risk profile (Z-scores) of banks of a specific ownership

type is sensitive to changes in the market structure (HHI) of the rest competitors on

the grounds that such a change is offset by adverse movements of the structure of the

whole market or that of banks of other ownership types. For the period 1993-2004, it

is indicative for regulation to promote the analysis of financial stability up to the point

of accounting of the specific relationship of ownership schemes and market structure,

along with the contemporary dimensions of risk-taking, consolidation and

competition, among others.

The IMF group of banking industries has been under investigation by Berger et al.

(2009). The GMM estimation of regressions comes along to resolve problems of

within-country correlation of the error term and heteroskedasticity after validating

164

market power instruments against potential endogeneity problems. The usual proxies

of financial distress are regressed on three versions of instrumented measures of

market power (Lerner index, HHI deposits/loans) and their quadratic terms providing

not a clear-cut answer. In fact, the opposing strands of the underlying relationship are

not necessarily mutually exclusive in the sense that banks of high market power does

exacerbate credit risk but their high franchise value may induce banks to protect

themselves mitigating the overall risk profile by higher equity capital, loan portfolios

of smaller size, among others.

Laeven and Levine (2009) empirically investigate how risk taking is interconnected

with management structures and national regulations in 48 countries for the period

1996-2001. They followed a pairwise examination by first estimating the extent to

which ownership (proxied by cash flow rights) affects solvency risk and, then,

quantifying the interaction of ownership and regulation on risk taking. The latter is

precipitated by the studies of Shleifer and Vishny, (1986) and Buser et al. (1981)

producing results that explicitly convey significant power on their interaction but

preserve individual effects whatsoever. Last, the interconnections among them could

be ascribed to bank valuations; that is, risk taking might be induced in the case of

having bank valuation be diminished as a response to national regulations (deposit

insurance, capital requirements, activity restrictions14

). Evidence highlights the

willingness of more powerful owners and equity holders to engage in riskier

investments than debt holders and non-shareholding managers.

Schaeck et al. (2009) examined 31 banking crises in a sample of 45 countries making

comparisons of duration and standard tobit modeling. Their explicit difference is that

the latter records a systemic financial crisis at the time of occurrence unconditionally

of the past, while the former defines the conditional probability of a crisis at a time,

until which no crisis has been experienced. Both models control for national legal

origins, key macroeconomic conditions under various specifications to conclude that

competition never appears to intensify systemic risk and produce equivocal effect

signs with concentration. In duration models, competition and concentration maintain

14

Barth et al. (2000) identified restrictions on securities activities to undermine bank viability in a large

sample of banking markets. However, the endogeneity of credit supply has been overlooked as an ex-

post anagnosis of the US crisis in 2007/8 (Shin, 2009).

165

the positive effect on the transitory period to crisis while both effects come along with

less inference after the inclusion of ownership, regulation and private monitoring.

Schaeck and Cihak (2008) used datasets of Europe and U.S. to assess first whether

competition Granger causes efficiency. The proxies employed here are the Lerner

Index and a translog specification of a cost as well as profit and alternative profit

functions. They also run the model of the so-called Boone indicator, which

presumably assumes that the relative performance of a bank is attributed to

differences in efficiency. Despite its shortcomings alleged in the literature, Z-scores

are regressed on it apart from control factors; however, instrumental variable

techniques are employed to resolve potential endogeneity bias of bank soundness,

Boone indicator and bank size. Some aspects here: First, the last two concepts are

correlated since vulnerable banks of little size may exacerbate their risk profile by

intensifying loan provision, which constitutes an indication of greater competition.

Second, this effect may acquire a negative sign if the Boone indicator captures the

entry of small firms that provide substitute banking services and thereby forcing

incumbent banks to curtail their costs more efficiently. The instruments employed are

the financial freedom index and fixed assets to total assets for the Boone indicator and

bank size, respectively. In so doing, efficiency tends to have an effect on competition

in both samples, by the construction of which efficiency is an active conduit of

competitive effects on bank‟s risk-taking conduct.

Beck et al. (2006a) examined the relationship of concentration and probability of

systemic instability in a large sample of 69 countries. They hypothesized mechanisms

– competition, monitoring and risk diversification - that are conducive to it in the

context of policy conducting and general academic debate. Controlling for several

factors that are treated as indicators of the above mechanisms, even different

measures of concentration turn out to express market features other than those

presumably supposed (competition) and used in the literature. The employed

methodology follows the logit probability model of Demirgüc-Kunt and Detragiache

(1998; 2002) and Gonzalez-Hermosillo et al. (1997). Empirical evidence gives credit

to concentration-stability advocating since less concentrated markets seem to

exacerbate the likelihood of systemic crisis to occur. The results are persistently

robust under different specifications that account for reverse causality, membership in

G10 group, non linearities and different samples.

166

5.3.2. European studies

In particular, Uhde and Heimeshoff (2009) examined the nexus between market

concentration and financial stability for the whole region of EU. Hence, upon the

theoretical arguments of Murphy et al. (1992) regarding the heterogeneous impact of

regulatory reforms in different countries, the employed methodology pertains to a

random-effect panel regression of Z-scores on concentration rates and other controls.

Sensitivity analysis that comprise other concentration measures, instrument variables

(among others) endorse the persistent negative effect of market concentration on

banks‟ soundness. In contrast, Cipollini and Fiordelisi (2009) identified a positive

effect in the EU-25 for the period 2003-2007. The employed methodology is inspired

by the GMM proposition of Bertscheck and Lechner (1998) for a panel probit model

that proxies financial distress as the lowest observations of the risk-adjusted

shareholder value.

Agoraki et al. (2009) studied Central and Eastern European banking industries

estimating the effect of competition and regulation on bank‟s risk-taking. The Lerner

index is used to proxy banks‟ market power by means of running a translog cost

function that includes deviations from the variable means (Uchida and Tsutsui, 2005;

Brissimis et al., 2008). Regulation is expressed quantitatively in terms of regulatory

indices along with the inclusion of other factors controlling for size, efficiency,

ownership and market discipline, among others. The results show that the stand-alone

effect of regulation on credit and solvency risk is contingent on the relative market

power of banks; contrary to capital requirements and restrictions on banks‟ activities,

official regulatory power turns out to diminish both risks.

Utilizing the same methodology and a unique database, Beck et al. (2009) examined

which ownership scheme contains higher stabilizing effects in Germany.

Corroborating the remarks of Cihak and Hesse (2007), cooperative and savings (of

any size) banks persistently exhibit lower insolvency and failure risk; stability levels

are intensified by lower volatility of recorded profits despite the negative, though

inadequate, forces of low profitability and high credit risk. Contrary to such findings,

the reckless and unsettled risk profile of large and private-owned banks jeopardizes

167

their imminent insolvency - withholding lower capital as they grow in size - while

lowers distress probabilities whatsoever.

At the country level, Beck et al. (2009) and Jimenez et al. (2010) studied Germany

and Spain, respectively. Beck et al. (2009) used a dataset for the period 1995-2007

contributing to the literature by commenting on the relationship of size and ownership

types with insolvency, credit risk and distress probability. Evidence showed stability

patterns for government-owned and cooperative banks and savings banks due to low

volatility in profits and credit risk and less distress probabilities for savings banks.

However, large and private-owned banks tend to withhold less capital along their

expansion thereby being less likely to fail but nonetheless closer to insolvency

conditions. On the other hand, Jimenez et al. (2010) tested the hypothesis that banks

with possessing market power lead to less risk taking along the lines of franchise

value paradigm. Their model verified it by identifying no causal effect of

concentration on risk-taking and significant persistence of market power (measured

by Lerner index).

5.4. Conclusion

The chapter reviews the theory and empirical applications on the trade-off between

market power and financial stability. Amid competitive conditions, we may

experience failure in coordination between investors and depositors although with

appropriate regulation we can accomplish stability alongside liberalisation

procedures. On the other hand, the dynamics towards market concentration through

mergers and acquisitions should be accompanied with policies that consider

intertemporal incentives of too-big- to-fail institutions and collaboration between

credible competition and regulatory authorities.

168

Chapter 5 Appendix


Theoretical foundations

Allen and Gale

(2004) Theoretical framework

Beck (2008) Good theoretical and empirical review

Besanko and

Thakor (1993)

Relationship banking, deposit insurance and bank portfolio choice

Theoretical framework

Borio (2003) Micro- and macroprudential dimensions of financial instability

Borio (2006) Macroprudential regulation and supervision

Borio et al.

(2002) Asset prices, credit, monetary stability to financial stability

Boyd and de

Nikolo (2005) General equilibrium modeling of risk-taking in concentrated bank markets

Boyd et al.

(2004) Relative probabilities of crises in monopolistic versus competitive markets

Boyd et al.

(2009) Theoretical framework of competition, risk and asset allocation

169

Carletti and

Hartman (2002) Good theoretical and empirical review

Crockett (1996) Good theory

Cordella and

Yeyati (2002) Spatial competition model (Salop (1979) under various scenarios (deposit insurance and information disclosure)

Martinez-Miera

and Repullo

(2010)

Risk-shifting effect (monopolistic) and marginal effect (competitive markets)

Matutes and

Vives (1996) Fragility induced by coordination problems of depositors – economies of scale, deposit insurance

Matutes and

Vives (2000) Welfare performance of market and appropriateness of regulation contingent on rivalry degree and deposit insurance regime

Mishkin (1999) Consolidation leads to inherent systemic risk but larger banks can be protected by vigilant supervision and government safety net

De Nikolo and

Lucchetta

(2009)

Financial intermediation, competition and risk – A GE exposition

Fischer and

Genard (1997) Financial liberalization and fragility

Goetz (2010) Bank risk taking as states remove branching barriers – channels of effect

170

Haldane et al.

(2004)

Financial stability and bank solvency

Economic effects of banking crises

Hellmann et al.

(2000)

Dynamic model of moral hazard

Capital requirements are not enough unless coupled with deposit-rate controls

Koskela and

Stenbacka

(2000)

Model of mean-shifting investment technologies

Market structure – risk taking – social welfare

Competition-stability without conditioning interest rates on investments and credit rationing

Perotti and

Suarez (2002) Dynamic relationship of charter value with future competition (takeover of failed banks)

Schaeck (2009) Good review of concentration, competition and stability - measures

Vanhoose

(2010) Good theoretical analysis of competition, stability and regulation

Review of empirical analysis

Author(s) Methodology Notes Period Countries

Agoraki et al.

(2009)

Non-performing loans to total loans, Z index regressed

on market power, 3 regulatory indices (Barth et al., 2001;

2006; 2008), interaction term, bank and macroeconomic

controls plus an AR(1) specification

Risk-taking (r), competition (Lerner-mc by translog a la

Uchida and Tsutsui, 2005; Brissimis et al., 2008),

(lagged) regulation

1998-2005 13 CEE countries

Barth et al.

(2000) Financial stability (competition, bank and securities

development): Regulation-banks ability to engage in

171

securities, insurance, real estate, restrictions on banking

mix and commerce (banks owning nonfinancial firms

and vice versa), countries with state-owned banks

Beck et al.

(2006a) DD (1998; 2002)

Growth, terms of trade change, real interest rate,

inflation, M2/reserves, depreciation, credit growth (t-2),

real GDPPC, moral hazard, concentration, entry

applications denied, activity restrictions, required

reserves, capital regulatory index, banking freedom,

economic freedom, KKZ_composite, dummies of

national legal origin, latitude, ethnic fractionalisation and

religion (3)

1980-1997 69 countries

Beck et al.

(2005)

Logit model (Cole and Gunther, 1995; Gonzalez-

Hermosillo et al., 1997; Demirguc-Kunt, 1989; DD,

1998; 2002)

Macroeconomic Controls: real GDPGR, terms of trade

change, inflation, real interest rate, international forces

on bank vulnerability: depreciation, M2/reserves), credit

growth (t-2), economic development: GDPPC, moral

hazard

Regulation: fraction of entry denied, activity restrictions,

required reserves, capital regulatory index, official

supervisory power

Ownership: state, foreign

Openness, competition, institutional: banking freedom,

economic freedom, KKZ_Composite


Beck et al.

(2009)

3 measures of financial stability

Comparison of 3 ownership groups

Relationship of the latter and bank size across different

ownership schemes

Financial stability proxied by Z-score, non-performing

loans, distress probabilities 1995-2007 Germany

172

Beck et al.

(2012)

FE

OLS (country, year dummies)

Cross-country heterogeneity in the competition-stability

relationship with respect to institutional and financial

development, regulation , supervision, herding and

market structure.

Z-score, conditional correlation between Lerner and Z-

score on information sharing, stock market turnover,

capital regulation, deposit insurance coverage, multiple

supervision, external governance index, activity

restrictions, heterogeneity in bank revenues, systemic

stability


Berger et al.

(2009)

GMM regression model of financial stability on market

structure, bank controls, business environment

NPLS, Z-score, E/TA on

Lerner derived by TCF, HHI-deposit, HHI-loan indices

Bank size, foreign ownership, ln(GDPPC), legal rights

index

Market power variables instrumented by activity

restrictions, banking freedom, percent of state-owned

banks

1999-2005 30 developed countries

(IMF)

Bonfirm and

Dai (2009) Fixed effects panel model

Interest rates on the number of bank relationships, time

dummies

Firm controls: turnover, tangible assets% debt, leverage,

credit risk, debt coverage, firm age, assets and assets2

3-month Euribor, number of banks granting credit,

GDPGR instead of time dummies

1996-2004 Portugal

Boyd et al.

(2009)

Stability proxies (Z score, NPL) on HHI, country- and

bank-specific controls

Z score, NPLs, - HHI

-GDPGR, inflation, GDPPC population

-total assets, non-interest operating costs to total income

June 2003

1993-2004

US

134 countries

173

Cerqueiro

(2008) Heckman (1979) – probit model

Probit model of Firm‟s (bank‟s) economic benefit of

applying for (granting) a bank loan on variables

Loan - rate equation

1993 US

Cipollini and

Fiordelisi

(2009)

Panel probit regression

GMM (Bertscheck and Lechner, 1998)

HHI, C5, id, size, GDPPC 2003-2007 EU - 25

Cole et al.

(1995)

OLS

FIML by Anemiya (1973)

Risk weighted assets, tangible net worth, net interest

margin, state charters, mortgages, loans, investments,

securities, goodwill, failure (time) dummies

1986-1989 US

De Nikolo and

Loukoianova

(2007)

Plus theory

Z scores on ownership, HHI, country and bank-specific

controls

Other various specifications

1993-2004 133 countries

Jimenez et al.

(2010)

Arellano and Bond (1991) procedure (first differences) -

GMM

Risk on AR(1), 2 structure factors, GDPG, AR(1), ROA,

size, loan ratio 1988-2003 Spain

Keeley (1990) Competition – asset risk and capital reductions (fragility) Risk on instrumented market value of bank assets and

financial and other controls 1952-1986 US

Laeven and

Levine (2009)

Ownership is instrumented by cash flows rights

Simultaneous estimation of risk and Tobin‟s q

Z scores, equity and earnings volatility on country

regulation controls (GDP volatility, risk: capital

regulations (requirements, stringency), activity

restrictions index, deposit insurance, shareholder

protection rights, national enforcement of laws, M&As),

bank - level controls: (managerial ownership, revenue


174

growth, market share, NYSE, size, NPLs, liquidity

ratios) and regulation measures, cash flow rights

Olivero et al.

(2010)

Dynamic Panzar-Rosse measure by GMM model of

Arellano and Bond (1991)

Bank lending channel

Loan growth rates on monetary policy indicator,

competition, their interaction term, crisis dummy,

GDPGR, bank size, capitalization and liquidity

1996-2006 10 Asian and Latin

American countries

Schaeck and

Cihak (2008)

Efficiency by TCF, Lerner,

Efficiency on AR(1), Lerner and AR(1), bank-specific

and other controls

Lerner on efficiency…(fixed effects panel)

Boone indicator (GMM)

Z score on Boone indicator/size (instrumented by bank‟s

market share, financial freedom index, fixed assets to

total assets, total assets, equity ratio and asset growth,

LLP to total assets, diversification index, [HHI, total

banks‟ assets, GDPPC, real interest rate]) plus country

and type dummies

1995-2005 Europe, US

Schaeck et al.

(2009)

H statistic

Duration analysis

Logit model

Inflation, GDP growth, real interest rate, depreciation,

terms of trade, credit growth, moral hazard index, legal

origins, continent dummies, H-stat, concentration

Activity restrictions, entry restrictions, foreign

ownership, government ownership, official supervisory

power, private monitoring index, capital regulatory

index.


Shin (2009) Origins of the us crisis - interesting Accounting framework- profile of banks dependent on

equity, leverage and funding

2001-2007

1998-2007

US

UK

175

Uhde and

Heimeshoff

(2009)

Z-score on concentration, macroeconomic, bank-specific,

regulatory and institutional factors

Good robustness tests – 2SLS

Z score, ROAA, capital ratio, s.d.ROAA, concentration

(4), HHI, basic economic attitude, EU-integration,

Duration, GDPPC, real GDPGR, inflation, real interest

rate (t-1), credit growth, net interest margin, loan loss

provisions, cost-income ratio, moral hazard index, entry

restrictions, activity restrictions, capital regulatory index,

governmental ownership, economic freedom, British-

French-German-Scandinavian-Soviet legal origin

1997-2005 EU-25

Yeyati and

Micco (2007)

Time invariant (OLS) - varying (WLS / panel)

H stat

H on concentration and foreign penetration

Z score and its components on foreign ownership, size,

country-year fixed effect

Z score on the share of foreign assets, H stat and macro

controls (real growth and exchange rate volatility),

[Including bank effects: proxies of bank size and foreign

ownership]

Foreign entry instrumented by average of the foreign

shares in the rest sample countries

3 alternative z-scores [Z, ln(Z), system Z] on foreign

assets to total assets, GDP, exchange rate volatility,

concentration (average CS3, average country lnTA, H

stat)

1993-2002 8 Latin American countries

177

CHAPTER 6

Income-specific estimates of competition in European banking

179

6.1. Introduction

A significant number of empirical applications of contemporary competition

modeling has proposed indicators of market structure that suffer from either limited

comparability across European countries or over time. The structural approach, that is

concentration ratios capturing the structural features of a market, is used in models or

interpreted in conjunction with other performance measures to explain the

competitive behavior of a specific industry without sufficing stand-alone to

extrapolate competitive conditions. Even changes in concentration can be deduced

regarding market entries and exits, a feature widely used in U.S for anti-trust

purposes.

Next step is to build up a link between structural changes and bank performance,

mainly on the grounds of the Relative Efficiency (RE) and the Structure-Conduct-

Performance paradigm (SCP). Non-structural measures developed in response to the

deficiencies of structural models to quantify bank competition are based on monopoly

power measures - the New Empirical Industrial Organization (NEIO) approach. It sets

out behavioral equations concerning the price/output specifications of Iwata (1974),

Bresnahan (1982), Lau (1982) and Panzar and Rosse (1987) upon the grounds of

Lerner (1934). Another case may be the degree of contestability in a bank market;

few banks implement competitive pricing as price takers in order to discourage „hit

and run‟ behavior of new entries and thereby second their monopoly power.

Due to the limited capacity of SCP and relative efficiency, Heffernan (2002) tests for

contestability and Cournot/Salop-Stiglitz behavior in UK banking industry by

proposing a general linear model of competitive pricing of the important retail

banking products: deposits, loans, credit cards and mortgages. There is also

considerable attention in the literature to Leuvensteijn et al. (2011) for the first

empirical application of the so-called Boone indicator that measures competitive

conditions insofar as they are expressed by efficiency dynamics. The most recent

models coming to shed some light on the cross-country comparability of alternative

competition measures and incorporate the switch of banking income to not-interest

bearing sources are those of Carbo et al. (2009) and Bolt and Humphrey (2010),

respectively. The econometric analysis applies an error-correction specification to

distill competition measures from country-specific effects, and stochastic frontier

180

methodology to provide rankings of banking sectors in terms of relative competitive

measures of market structure.

The paper follows the analysis of De Guevara et al. (2005) investigating the key

effects of market power as proxied by the Lerner index. The contribution is threefold:

a) the sample is focusing on nine developed countries within the European union

since the advent of Euro, b) the emphasis is more on the measurement of marginal

costs in order to abstain from potential bias triggered by traditional modeling in the

literature, c) it is the first time, at least to our knowledge, that effects on market power

are investigated at the income level.

6.2. Methodology

The underlying paper estimates the price mark-up over marginal cost combining the

estimation of average prices and marginal costs at the bank level. The average prices

are estimated over total assets (TA) along the lines of Shaffer (1993) and Berg and

Kim (1994), instead of other earning assets in an attempt to expand as much as

possible the observations of the sample since 2002. First, we have to estimate

marginal costs by means of running a translog cost function, similar to the version of

Ariss (2010b) that excludes the use of price of borrowed funds as input price on the

grounds that it presumably captures some degree of monopoly power of incumbent

banks in the deposit market. The employed model takes the following form15

:

(6.1)

where TC: total costs, Q: total assets, W1: price of labour (personnel expenses over

total assets), W2: price of physical capital (other operating expenses over fixed

15

The panel model utilizes observations of each bank i at time t. We omit subscripts for convenience

purposes.

181

assets), Z1: fixed assets deflated by total equity, Z2: Off-balance sheet activities (non-

interest operating income) deflated by total equity and T: time trend. We introduce

fixed effects to account for different bank specificities and run model (6.1) separately

for each banking market to reflect different technologies in the region. We also allow

for a time trend to interact with the deterministic kernel in order to capture time-

varying and non-neutral technological progress in the banking sector. Homogeneity of

degree one in input prices (Σγk=1) and symmetry conditions in all quadratic terms are

imposed in model (6.1).

When it comes to the estimation of the Lerner index, we extrapolate the marginal

costs by running the following model, which is schematically the partial derivative of

total costs with respect to total assets (see Berger et al. 2009; Ariss, 2010):

(6.2)

We are then able to construct the Lerner index (L) with respect to specific bank

activities before delving into the analysis of competition determinants. According to

the following structural model,

(6.3)

where AR denotes the average revenue estimated by total income over total assets and

MC the marginal cost derived through model (6.2). Their subscripts signify the use of

Lerner index as the only proxy of market power at the bank level over time. The

following model (6.4) encompasses the conditioning of market power to various

information sets that comprise some key effects that have been under scrutiny in the

literature and other variables depicting conditions in the banking industry,

institutional and macroeconomic environment. Therefore, we maintain the structure of

four specifications in models 6.4 and 6.5 in order to draw upon the changes of

significance in the key coefficients employed.

182

L f BANKS, INDUSTRY ,ECONOMY ,DUMMIES (6.4)

There are some econometric issues involved in this case. First, we run Hausman test

to see whether fixed or random effects are appropriate, an amended Wald test for

present groupwise heteroskedasticity, Wooldridge test for first degree of

autocorrelation and the significance of using time fixed effects (testparm). We come

up with concurrent heteroskedasticity, autocorrelation in the model urging us to opt

for standard errors clustered at the bank level.

As a next step, we go further down the analysis replacing the aforementioned Lerner

index with indexes that are based upon specific bank activities, namely income on

loans, other interest-bearing income, fees and commissions and other non-interest

income. We fell short of employing other sources of bank income since they lack

considerable amount of information. In the underlying case, we opt for the Seemingly

Unrelated Regressions (SUR) framework to account for error autocorrelation within

banks. In other words, running equation-by-equation OLS regressions would be

consistent but, nonetheless, inefficient since all equations are interrelated through the

correlation of the error term. It is also of interest to verify, through the Breusch-Pagan

test, the degree of error correlation across the equations of each bank and thereby the

imperative to employ the SUR framework.

The following model (6.5) comprises the contemporaneous pricing power of specific

bank products based upon various information sets. The latter refers to the key

features of banking entities, the market they operate within and other effects of

economic environment. We denote as li the income on loans, other interest income16

,

fees and commissions and other non-interest income17

:

16

Interest income stemming from the trading book, investment securities and other short-term funds. It

excludes insurance-related income (Bankscope definition).

17 Sustainable operating income that is related to the core business of a bank, totally demarcated from

trading, derivatives, other securities and insurance income (Bankscope definition).

183

l1

l2

l3

l4

f BANKS, INDUSTRY ,ECONOMY ,DUMMIES

1

2

3

34

(6.5)

For the SUR to be properly applied, the models of every income-specific Lerner index

should have exactly the same size but different information set. Otherwise, the

estimation falls into equation-by-equation OLS.

6.3. Competition determinants

Size is introduced in the form of log of total assets as a control variable in order to

allow for the heterogeneous European sample that is associated with either relative

market power or scale economies. We opt to plug in the model a quadratic term to

verify whether it is the case of non-linear relationship between size and competition.

We test the SCP paradigm with the significance of Herfindahl-Hirschman index

(concentration), which is the sum of the squared market shares of all banks operating

in a country. Taking into account Corvoisier and Gropp (2002), the effect of

concentration may be different if expressed in terms of total deposits and loans. On

the contrary to the use of aggregated information for the concentration proxy, we opt

for the estimation of HHI from the sample data in order to examine how the synthesis

of market shares may constitute an exogenous force towards market power. However,

data from non-consolidated accounts by no means exclude large banks that otherwise

would depict larger-than-national regions, since they comprise information of at least

a part of banks‟ operations disaggregated at national levels.

Market share reflects the ratio of a bank‟s total assets over those of a national banking

industry. We use it also in terms of total deposits and loans to specify the channel,

through which the efficiency hypothesis may hold. However, its statistical

significance should be interpreted in conjunction with that of efficiency and

concentration in order to give credit to the power of alternative competition theories.

We use the first order lag of loan impairment charges over average gross loans as a

proxy of credit risk. We abstain from the traditional non-performing loans over gross

loans since it lacks considerable amount of observations from 2002 and onwards.

184

Moreover, the former is a direct measure of loan losses as it is deducted at the end of

the year from profits and, hence, is taken into account when it comes to price bank

products the year after; the latter is more obscure when it comprises doubtful loans

that may or may not end up nonperforming.

We employ cost-to-income ratio as a direct measure of operational performance that

may be attributed to superior management or production technologies. The RE

hypothesis assumes that banks of higher efficiency engage in competitive pricing in

order to grasp greater market shares, leading to high market concentration. We also

utilize the degree of income diversification of bank portfolios as indicated by the

proxy of off-balance sheet activities over total assets. It is ambiguous whether the sign

of effect is going to be positive or negative since banks willing to engage in other than

traditional loan and deposit services may have a different strategic pricing contingent

on the bank type, region, economic cycle, among others.

Total equity as a percentage of total assets (capital) accounts for further size effects

on the Lerner index of other non-interest income. The intuition here is that banks of

greater size or high capital buffers are willing to expand non-traditional banking.

Liquidity proxied by the amount of liquid assets over customer deposits and short

term funding is an important driver in the models of interest-bearing Lerner indexes

so as to quantify the correlation of pricing conduct with the ability of banks to

facilitate a potential bank run.

Elasticity of aggregate demand. We opt for the bank claims on the private sector over

GDP so as to verify whether the dependence on bank financing may be associated

with benefits for the real economy or loan losses and bank instability.

Apart from the effect of real GDP growth (cyclicality) boiling down the procyclical or

countercyclical effect of market expansion on market power, we utilize the real GDP

per capita and inflation to see whether it is the case of population (GDPPC) or price

effect (Inflation) of the business cycle in Europe. In addition, we control for sector

regulation by the strength of legal rights index as it essentially measures the efficacy

of regulatory laws with respect to collateral and bankruptcy issues in order to

safeguard the rights of borrowers and lenders. It takes values between 0 and 10, with

higher scores indicating that access to credit is facilitated to expand further.

185

We use dummies for the divergent specialisation of European banks, viz. commercial

banks, savings banks, cooperative banks and „other‟ banks that incorporate any bank

holding and holding companies, clearing institutions and custody, finance companies,

group finance companies, investment and trust corporations, investment banks,

Islamic banks, other non-banking credit institutions, private banking and asset

management companies, real estate and mortgage banks, securities firms and

specialized governmental credit institutions. The last dummy „other‟ is excluded for

multicollinearity reasons. We apply the Dummy Variable Least Squares (DVLS)

version of the fixed effects modeling, using country dummies for Austria, Denmark,

France, Germany, Luxembourg, Netherlands, Spain, Sweden and UK; Netherlands

are excluded to eschew multicollinearity. We add time dummies to allow for time

fixed effects after verifying their significance each time the information set is altered

in the regression model.

Other effects already tested in the literature, namely liquidity risk, implicit interest

payments or degree of risk aversion turn out to be insignificant without affecting the

regression output whatsoever. We omit additional bank-specific and institutional

variables since they lack sufficient statistical information during the underlying

period.

6.4. Data

The sample encompasses features of financial statements of banks operating within

nine developed countries of the European Union as they enjoy the most available data

recently. Particularly, they amount to 19187 observations of 2950 banks in Austria,

Denmark, France, Germany, Italy, Luxemburg, Spain, Sweden and United Kingdom.

We retrieve data of unconsolidated accounts from the Bankscope database,

macroeconomic and other regulatory variables from the World Bank and Eurostat.

We drop banks out of the sample that are not capable of preserving a satisfactory time

dimension (3 years) during the period 2002-2010.

We get rid of the outliers in order to come up with appropriate values of the Lerner

index that lie in-between negative infinity and one. All the determinants of bank

186

competition as a second stage of our analysis narrow down the observations to 15219

and, even, to levels of 9503 observations. Last, the third stage substantiates the

persistence of market power effects on income-specific Lerner indexes analysing a

sample that fluctuates between 798 and 8798 observations.

Table 1: Number of banks

Country 2002 2003 2004 2005 2006 2007 2008 2009 2010 Total

Austria 27 125 143 147 155 159 146 133 106 1141

Denmark 11 44 47 47 50 50 62 61 56 428

France 25 62 65 75 74 75 66 62 53 557

Germany 326 1328 1346 1619 1625 1609 1565 1515 1344 12277

Italy 18 19 22 572 581 595 582 555 406 3350

Luxembourg 27 50 54 52 51 55 53 44 27 413

Spain 8 9 15 68 69 52 56 61 48 386

Sweden 25 75 73 77 71 69 58 56 56 560

UK 14 50 61 71 76 82 83 81 69 587

Total 481 1762 1826 2728 2752 2746 2671 2568 2165 19699

Source: Own estimations of data retrieved from Bankscope.

Table 1 displays the summary statistics of the key variables used in the stochastic

frontier model (1) averaged by country over the period 2002-2010. It is evident that

Germany dominates the sample enumerating more than the half of total observations

(62.3%) while the year 2002 displays the least available information for all countries

with a gradual escalation year by year. The sample includes fewer banks for Spain

and relatively more banks for Italy and Austria; all the rest observations are close to

the range of 413 to 587 banks. Table 2 summarizes the key statistics of the variables

employed in model 1. Total costs over total assets stand at 4.9% and ranges from 3.8

in Spain to 7.1% in France. In terms of total assets, banks in UK and Luxembourg

bear comparably higher costs in levels up to 4.7 and 8.2 million euro, respectively; the

rest sectors take values from 240.82 in Sweden to 2182.14 millions in France.

187

Table 2: Summary statistics of variables in model 6.1

Country TOTAL

COSTS

TOTAL

ASSETS

PERS

EXP

OTHER

OPER

EXP

FIXED

ASSETS OBS EQUITY

Austria 0.045 842.83 0.013 0.011 0.017 0.014 0.087

Denmark 0.053 625.65 0.019 0.016 0.012 0.018 0.140

France 0.071 2182.14 0.016 0.020 0.014 0.028 0.112

Germany 0.049 1443.27 0.015 0.011 0.015 0.011 0.068

Italy 0.044 1519.12 0.015 0.013 0.016 0.012 0.117

Luxembourg 0.052 4757.09 0.013 0.013 0.005 0.030 0.104

Spain 0.038 1345.14 0.011 0.009 0.017 0.009 0.117

Sweden 0.044 240.82 0.015 0.016 0.014 0.016 0.141

UK 0.062 8273.54 0.018 0.023 0.011 0.032 0.155

Average 0.049 1661.42 0.015 0.012 0.015 0.013 0.087

All figures are expressed as a percentage of total assets apart from total assets being in millions of Euro.

Source: Bankscope database.

Furthermore, personnel expenses over total assets stand at 1.5%, ranging from 1.1%

in Spain to 1.9 in Denmark. As for the other operating expenses over total assets, it

has a mean up to 1.2, with values from 0.9% in Spain to 2.3% in United Kingdom.

Fixed assets to total assets stand at 1.5%, ranging from 0.5% in Luxembourg to 1.7%

in Austria and Spain. Moreover, the average off-balance sheet activities deflated by

total assets is 1.3%, while it ranges from 0.9% in Spain to 3.2% in United Kingdom.

Last, capital ratio stands at 8.7% taking values within the range of 6.8% in Germany

and 15.5% in United Kingdom.

In table 3, we report the man values of the key variables of models 4 and 5 per

industry sector. Concentration creeps in moderate average levels below 15% except

the case of United Kingdom where it is about 27%. The degree of concentrated in

specific markets like deposits market ranges between 1% in Germany and 37.3 in UK,

whereas in loans markets the pattern is almost the same for Germany (1.1%) and UK

(42.4%). Market share stands at values in between the range of 0.8% in Austria and

2.3% in Luxembourg. Product-specific market shares coincide in almost all cases with

values between the extreme bounds 0.1% in Germany and 2.3% in Spain and

Luxembourg.

188

Cost efficiency as proxied by total cost over total income is identified in Luxembourg

(58.6%) whereas the worst performer is Germany (70.8%) and UK (80%). The quality

of loan books is proxied by net charge-offs over average gross loans as limited data

were available for the standard ratio of NPLs over gross loans. It gets values from

67.5% in Italy to 217.3% in Germany, which consists in the share of loans finally

purged from the books. The negative average value in France means that delinquent

debt classified as gross charge-offs turns out to be repaid; thus, the recoveries are

greater than charge-offs since French banks have been moderate in evaluating bad

loans.

Moreover, the diversification ratio of off-balance sheet activities as a percentage of

asset size stands at levels in between the range of 1.1% for the cooperative banks of

Germany and 3.2% in UK. The capital cushion against potential asset malfunction

(Equity/TA) that insulates a bank at least up to the degree of investment in it, has

figures from 6.8% in German to 15.5% in UK banks. Liquidity risk is proxied by

liquid assets to customer deposits and short-term funds, which essentially forms the

share of deposits and other funds that could be met in case of a sudden bank run. Its

figures take values between 16.6% in Sweden and 83.3% in Luxemburg.

As far as economic conditions are concerned, the average growth rate of GDP

annually performs worse by 0.2% in Italy and by 2.8% in Luxembourg. Italy also has

the least private credit granted to the private sector (deflated by GDP) at levels of

0.976 while Denmark performs much better with 1.878. In addition, in order to

account for the population effect, GDP over midyear population ranges between

20900 in Spain and 65344 in Luxembourg. Besides, the inflation rate (deflated by

GDP) had a much higher pace in Spain during the period 2002-2010 at about 2.83%

whereas Germany is considered to have accomplished the lowest rate at 1.6%. Last,

the strength of legal rights index indicates that bankruptcy and collateral laws are

reinforced to better protect lending in UK (10.0) and considerably less in Italy (3.0).

Table 3: Summary statistics of variables in models 6.4 and 6.5

Variable AUS DEN FRA GER IT LUX SP SWED UK

189

The estimation of CONC and market share along with their constitutents takes account of the available

information in Bankscope database, which enumerates up to 19761 observations of our sample. Source: World

Bank, Eurostat and own estimations.

6.5. Results

Table 4 depicts the mean values of the Lerner index and those indexes at the income

level per banking industry as well as the average marginal costs that used for the

estimation of the former. What really stands out in the column of marginal cost is the

high values of the banking sectors in United Kingdom and France at about 6.7%; the

remainder falls in-beween the range of 3.8% in Sweden to 5.4% in Luxembourg.

When it comes to compute the Lerner index, we expect sectors with high marginal

costs to enjoy relatively greater competitive conduct. In contrast to the almost perfect

competition in UK, France demonstrates 21%, which is higher than other values in

countries of higher marginal costs. However, Sweden and Denmark enjoy the highest

CONC 0.071 0.133 0.059 0.013 0.107 0.123 0.150 0.040 0.273

CONC (deposits) 0.073 0.075 0.062 0.010 0.113 0.127 0.155 0.040 0.373

CONC (loans) 0.071 0.142 0.053 0.011 0.136 0.126 0.127 0.048 0.424

Market share 0.008 0.021 0.016 0.001 0.003 0.022 0.023 0.016 0.015

Market share

(deposits) 0.008 0.021 0.016 0.001 0.003 0.022 0.023 0.016 0.016

Market share

(loans) 0.008 0.021 0.016 0.001 0.003 0.023 0.023 0.016 0.017

TC/TI 0.669 0.679 0.665 0.708 0.691 0.586 0.628 0.642 0.800

Loan impairment 0.893 1.204 -2.821 2.173 0.675 0.807 1.912 0.604 1.430

OBS/TA 0.014 0.018 0.028 0.011 0.012 0.030 0.009 0.016 0.032

EQ/TA 0.087 0.140 0.112 0.068 0.117 0.104 0.117 0.141 0.155

Liquidity 0.306 0.245 0.273 0.188 0.273 0.831 0.296 0.166 0.653

GDPGR 0.017 0.006 0.011 0.009 0.002 0.028 0.019 0.022 0.016

Private credit 1.147 1.878 0.996 1.116 0.970 1.486 1.633 1.167 1.754

GDPPC 30367 3798

9 27322 27789 24367 65344 20900 33044 30289

Ιnflation 1.811 1.944 1.867 1.600 2.222 2.689 2.833 1.811 2.200

Legal Strength 7.000 8.714 6.000 7.571 3.000 6.800 6.000 6.714 10.000

190

monopolistic rents at levels of circa 39 and 33.3%, respectively. The negative sign in

Spain indicates irrational behaviour of bank managers pricing products below

marginal costs.

Table 4: Marginal costs and Lerner indexes

Country MC L L1 L2 L3 L4

Austria 0.049 0.103 -4.107

(16.2%)

-0.488

(68.8%)

-16.312

(12.6%)

-90.441

(2.4%)

Denmark 0.048 0.332 -0.953

(60.4%)

-2.470

(21.3%)

-5.389

(14.4%)

-190.856

(3.9%)

France 0.068 0.210 -2.179

(47.4%)

-15.724

(27.8%)

-18.621

(18.1%)

-90.447

(6.7%)

Germany 0.052 0.118 -0.604

(66.4%)

-12.776

(18.2%)

-8.513

(12.3%)

-44.055

(3.1%)

Italy 0.049 0.126 -2.561

(70.3%)

-11.748

(15.9%)

-8.082

(10.9%)

-37.187

(2.9%)

Luxembourg 0.054 0.215 -3.081

(69.6%)

-77.182

(12.8%)

-5.010

(13.3%)

-72.152

(4.3%)

Spain 0.050 -0.010 -0.846

(2.1%)

-9.623

(61.5%)

-18.705

(35.4%)

-120.919

(1%)

Sweden 0.038 0.390 -1.061

(71.5%)

-21.066

(9.8%)

-3.650

(17.1%)

-89.223

(1.6%)

UK 0.066 0.003 -4.307

(7.7%)

-32.977

(41.7%)

-16.982

(47.1%)

-81.170

(3.5%)

Average 0.051 0.129 -1.062 -13.526 -9.255 -53.557

MC=Marginal Cost; L=Lerner index; L1=Lerner with respect to income on loans; L2=Lerner index

with respect to other interest income; L3=Lerner index with respect to Fees and Commissions;

L4=Lerner index with respect to other non-interest income. All estimates are expressed in average

terms per industry sector with the last row estimating the averages of the 9-country sample. The

percentage in the parentheses highlights the weights of income sources on banks‟ portfolio.

Supposing that irrational behaviour takes the form of competitive market in the eyes

of customers, the results are close to the competition efficiency scores estimated by

Bolt and Humphrey (2010). In addition, the negative values in the last four columns,

bank-specific indexes are supposed to constitute not a straightforward metric of

income-specific market power but, alternatively, the relative contribution of each Li to

the construction of the Lerner index. Ideally, the overall index (L) should be the sum

of the 4 sub-indexes, although some sources are excluded from the analysis as

unavailable and thus the lack of data does not permit us to delineate rankings of

relative market power at income level, but rather to investigate its possible effects

191

(table 4). We should note however, that the weighting of each source in the

construction of a banks‟ portfolio is such that explains the distance from the

respective level of marginal cost. However, they are indicative of the degree of

portfolio specialisation and the strategic orientation of profit-maximizing institutions.

As table 5 indicates, there is a negative relationship between size and market power

since, the logarithm of asset size enters negative at 1% level of significance. Non-

linearities are present through the quadratic term of asset size, which remains at the

same levels significant with negative sign; positive correlation does exist at least up to

a certain level of asset size, although turning insignificant once we apply fixed effects

methodology. If we treat the results with caution, banks are struggling to get bigger

and bigger implementing high profit margins, as far as their size makes them willing

to apply competitive conduct. The latter is reminiscent of market contestability, under

which incumbent banks demoralize potential market entries to jump in the market and

exploit any profit margin; however, such hit-and-run cases prerequisite a secondary

market to eschew from costs after closure. This issue is going to be elaborated further

when regressions are applied for banks of different productive specialization.

As for market structure, we find a positive stand-alone effect of concentration on

market power. It is nonetheless significant in the panel fixed-effects model (column

3), rendering the investigation of concentration on loans and deposits imperative

along the lines of De Guevara et al. (2005). The Cournot model is confirmed by the

significantly positive coefficient of deposit-related concentration, alongside the

concentration on loans that exhibits significance and persistent positive effect (except

the FE model), as well. Hence, we find evidence of the traditional SCP hypothesis

that collusion in deposits and loan markets drives banks to take advantage of cost

efficiency or monopolistic pricing to guarantee high profitability.

Table 5: Regression output (whole sample)

Competition

Determinants

MODEL SPECIFICATION

(1) (2) (3) (1) (2) (3)

LnQ -0.0148*** -0.0112*** -0.0367*** 0.0149* 0.0337*** 0.0577

(0.001) (0.002) (0.008) (0.008) (0.011) (0.038)

(LnQ)^2 - - - -0.0024*** -0.0037*** -0.0072**

(0.001) (0.001) (0.003)

Concentration 0.1399 0.1504 0.2784*** - - -

192

(0.101) (0.119) (0.101)

Concentration

(deposits) - - -

0.1852* 0.2347** 0.2653***

(0.098) (0.106) (0.081)

Concentration (loans) - - - 0.1630*** 0.1604*** 0.0530

(0.050) (0.054) (0.048)

Market share -0.5458 -0.7222 0.0857

- - - (0.344) (0.448) (0.770)

Market share (deposits) - - - -0.5367 -0.1282 1.0430

(0.413) (0.527) (0.662)

Market share (loans) - - - 0.1381 -0.2285 -0.6136

(0.399) (0.601) (0.547)

Cost/TI -0.5834*** -0.5738*** -0.5795*** -0.5821*** -0.5734*** -0.5722***

(0.034) (0.046) (0.066) (0.034) (0.046) (0.065)

Loan impairment 0.0103*** 0.0088* -0.0081 0.0107*** 0.0092* -0.0082

(0.004) (0.005) (0.008) (0.004) (0.005) (0.008)

OBS/TA 0.7486*** 0.7920*** 0.3526** 0.7433*** 0.7818*** 0.3682**

(0.181) (0.180) (0.142) (0.180) (0.178) (0.154)

GDPGR 0.0164 0.0757 0.1038*** 0.0489 0.1253 0.0874***

(0.149) (0.198) (0.017) (0.147) (0.193) (0.017)

Private credit -0.0582** -0.0330 -0.0106 -0.0979*** -0.1058*** -0.0433**

(0.028) (0.037) (0.019) (0.026) (0.034) (0.021)

Legal strength -0.0071** -0.0075* -0.0098*** -0.0047* -0.0026 -0.0087***

(0.003) (0.004) (0.001) (0.003) (0.004) (0.001)

Commercial 0.0522*** 0.0608***

-

0.0524*** 0.0603***

-

(0.013) (0.016) (0.013) (0.016)

Cooperative 0.0567*** 0.0614*** 0.0543*** 0.0578***

(0.012) (0.014) (0.011) (0.014)

Saving 0.0597*** 0.0580*** 0.0586*** 0.0580***

(0.012) (0.014) (0.012) (0.04)

Austria 0.1167*** 0.1357*** 0.0904*** 0.0837**

(0.036) (0.042) (0.034) (0.039)

Denmark 0.3852*** 0.3843*** 0.3860*** 0.3804***

(0.031) (0.032) (0.032) (0.033)

France 0.2479*** 0.2654*** 0.2173*** 0.2066***

(0.039) (0.048) (0.037) (0.044)

Germany 0.1679*** 0.1865*** 0.1474*** 0.1398***

(0.040) (0.049) (0.037) (0.043)

Italy 0.1253*** 0.1433*** 0.1111*** 0.1130***

(0.033) (0.039) (0.031) (0.035)

Luxembourg 0.2729*** 0.2679*** 0.2581*** 0.2472***

(0.033) (0.035) (0.034) (0.036)

Sweden 0.3849*** 0.4103***

-

0.3638*** 0.3634***

- (0.036) (0.043) (0.034) (0.039)

UK 0.1575*** 0.1516*** 0.0460 0.0197

(0.034) (0.037) (0.038) (0.044)

T2003 - -

-

- -

- T2004

-0.0017 - -

-0.0040

(0.001) (0.002)

T2005 - 0.0015 0.0036** -

193

(0.002) (0.001)

T2006 -0.0033 0.0008 -0.0027 -0.0068

-0.005 (0.006) (0.004) (0.007)

T2007 -0.0235*** -0.0217*** -0.0229*** -0.0280***

(0.003) (0.004) (0.003) (0.005)

T2008 -0.0425*** -0.0406*** -0.0393*** -0.0418***

(0.002) (0.003) (0.003) (0.003)

T2009 -0.0034 0.0048 0.0035 0.0096

(0.008) (0.012) (0.009) (0.010)

T2010 0.0138** 0.0142** 0.0167*** 0.0136*

(0.005) (0.006) (0.004) (0.007)

Intercept 0.5166*** 0.4372*** 0.8223*** 0.4701*** 0.3971*** 0.5511***

(0.051) (0.061) (0.055) (0.052) (0.062) (0.126)

R-squared 0.750 0.770 0.901 0.753 0.776 0.901

Obs 15219 15219 15219 15216 15216 15216

Column 1): standard OLS regression with standard errors clustered at the bank level using country and time

fixed effects to allow for unobserved heterogeneity; Column (2): Fixed effects model estimation with clustered

standard error at the bank level along with probability country weights. In so doing, we apply the inverse of the

number of banks operating within a national banking sector for cases where sample is overrepresented by some

countries (e.g. Germany); Column (3): Fixed effects model estimation with clustered standard error at the bank

level with time fixed effects. Standard errors are in parentheses while asterisks ***, **, * denote the

significance level being at 1%, 5% and 10%, respectively18

.

Contrarily opposed to the Relative market power hypothesis, market share performs

insignificant and negative coefficient whatsoever. However, if we juxtapose the

disaggregation of market share in the product categories of deposits and loans, it

remains of negligible contribution across all model specifications. Efficiency structure

hypothesis is verified by the negative and significant coefficient of cost-to-income

ratio at the 1% level. In fact, banks with better management or technology tend to

bear lower costs that are transmuted to higher market shares and overall great

concentration. In our specifications where the element of increased market share is

not verified empirically as it is assumed by the theory, banks with lower costs have

relatively higher margins along the lines of pure efficiency hypothesis.

Credit risk as proxied by loan impairment charges over average gross loans has a

significant effect on market power, which is in contradiction with theoretical

premises. It bears negligible effect in the FE model as indicated by its opposite

coefficient and statistical insignificance, though the omission of it causes almost a 3%

loss of explanatory power. That is attributable to the total absorption of bank

specificities in the panel FE model, since our heterogeneous sample may blur the

18

We drop the dummies of 2002 and Spain for multicollinearity reasons.

194

trend of banks with different specialisation. Furthermore, portfolio diversification

proclaimed by the use of off-balance sheet activities over total assets demonstrates a

stable positive pattern with a level of significance ranging between one and 5% level.

Thus, banks are willing to impose greater prices on products if their portfolios are

well diversified against potential market and credit risks.

The effect of GDP growth remains insignificant in all but one specification, while its

positive sign implies procyclical rather than countercyclical force. That means in

times of economic expansion, banks enjoy relatively higher margins exacerbating

thereby the economic conditions further down the road either seconding bubbles or

deepening the recession spiral. However, the significance is not robust and, thus, any

remark should be drawn with caution. The elasticity of aggregate demand proxied by

private credit over GDP, which is a bank-oriented form of firm financing, is

significant up to 1% level in four specifications. The negative effect means that as

more credit is demanded and granted to the private sector, banks tend to narrow down

profit margins. The elasticity of aggregate demand is assumed in this study to

empirically verify its considerable importance when coupled with concurrent

conditions.

After accounting of all fixed effects in European banking, the negative effect of

lending-facilitating laws is considerably significant on the Lerner index. Regulatory

and supervisory policies are expressed in the degree of law stringency acting

preventively against monopolistic practices and in favour of borrowers and lenders in

order to facilitate access to credit.

Country and specialisation effects in table 5 persistently appear to be significant at 1%

level in the first three columns, and once we allow for non-linearities and

disaggregation of concentration and market share they remain put; time effects are

present almost after 2007 and onwards. We next rerun the above regressions for four

bank categories, namely commercial, cooperative, savings and „other‟ banks, which

include all the rest failing to construct a sufficient size of panel series. The necessity

of conducting regressions per bank type is to explain how the pricing conduct of each

specialisation is formulated given the divergence of strategic priorities and corporate

expertise. We can also see how the model behaves in a disaggregated analysis and to

195

what extent the per se pricing conduct is determined by the following SUR framework

of income-specific market power.

Hence, table 6 displays the results of six regressions, out of which two models are

implemented to four bank categories. The first model employs bank size,

concentration and market share and all the other variables that proxy for market

conditions, whereas the second includes the quadratic term of total assets (in logs),

and disaggregates concentration and market share into the broad product categories of

deposits and loans. The fit of the models ranges between 79.6% (other banks) to

87.5% (savings banks) while commercial banks, which constitute by far the most

fragmented market in Europe, perform moderately up to 65.6%. Cooperative banks

dominate the half of the sample with 8798 observations and the rest is shared among

saving, commercial and „other‟ banks with 4009, 1612 and 800 observations,

respectively.

A notable result is that asset size and market power are negatively related in

cooperative, saving and „other‟ banks while commercial banks fail to demonstrate any

significance. In fact, cooperative banks show only linear association, while for

savings and „other‟ banks, the positive sign is significant up to a certain level of asset

size before it turns negative at much higher sizes. Collusion hypothesis holds in loans

markets for commercial banks and „other‟ banks whereas in the deposits markets, the

commercial category is significant at ten% level. That comes in contradistinction with

the significant effect of concentration in table 5 thereby implying that sample

heterogeneity questions the credibility of the whole sample.

196

Table 6: Regression output per bank type

Competition Determinants MODEL SPECIFICATION

Commercial Cooperative Savings Other

LnQ -0.0085 -0.0085 -0.0171*** -0.0163** -0.0089*** 0.0423** -0.0390*** 0.1419***

(0.006) (0.038) (0.001) (0.006) (0.001) (0.017) (0.009) (0.035)

(LnQ)^2 - -0.0001

- -0.0001

- -0.0036***

- -0.0124***

(0.002) (0.001) (0.001) (0.002)

Concentration 0.0799

- 0.0272

- -0.3895*

- 0.0937

- (0.176) (0.056) (0.215) (0.293)

Concentration (deposits) - 0.2490*

- -0.7215

- -0.0990

- 0.1804

(0.130) (0.520) (0.432) -0.217

Concentration (loans) - 0.1374**

- 0.7243

- -0.1395

- 0.2707*

(0.066) (0.504) (0.398) (0.153)

Market share -1.2176***

- -1.3911***

- 0.6267

- 1.0301*

- (0.417) (0.265) (0.404) (0.556)

Market share (deposits) - -1.2171

- -3.2271***

- 0.5640

- 0.8079

(0.738) (0.670) (0.831) (1.193)

Market share (loans) - 0.3098

- 1.8082***

- -0.3401

- 0.7306

(0.796) (0.566) (0.570) (0.689)

Cost/TI -0.5112*** -0.5075*** -0.6216*** -0.6218*** -0.6448*** -0.6356*** -0.6431*** -0.6362***

(0.072) (0.073) (0.012) (0.012) (0.020) (0.020) (0.030) (0.028)

Loan impairment 0.0197*** 0.0197*** 0.4730*** 0.4812*** 0.2974 0.2949 -0.0096* -0.0093

(0.004) (0.004) (0.128) (0.123) (0.223) (0.219) (0.006) (0.006)

OBS/TA 0.6101*** 0.6031*** 3.7762*** 3.7687*** 5.1893*** 4.8759*** 0.6541** 0.7350***

(0.191) (0.198) (0.517) (0.519) (1.071) (1.026) (0.269) (0.260)

GDPGR -1.2818*** -1.2030** 0.0814 0.1429 0.4602** 0.4771*** 0.5598 0.6401

(0.465) (0.467) (0.098) (0.113) (0.184) (0.180) (0.549) (0.546)

Private credit -0.2033*** -0.2502*** -0.0981*** -0.1060*** -0.0394 -0.0464 0.2093 -0.0132

(0.056) (0.053) (0.020) (0.022) (0.035) (0.035) (0.130) (0.154)

197

Legal strength -0.0053 -0.0013 -0.0124*** -0.0104*** -0.0007 -0.0025 -0.0149 -0.0052

(0.005) (0.005) (0.003) (0.003) (0.004) (0.004) (0.010) (0.011)

Austria 0.0081 0.0046 0.0783*** 0.0770*** 0.1280*** 0.1253** 0.3076* 0.1854

(0.078) (0.080) (0.021) (0.022) (0.045) (0.050) (0.180) (0.172)

Denmark 0.3735*** 0.4014*** 0.3408*** 0.3092*** 0.4030*** 0.4089*** -0.0069 0.0835

(0.072) (0.080) (0.041) (0.036) (0.033) (0.049) (0.137) (0.187)

France 0.1040 0.0974 0.2528*** 0.2460*** 0.2098*** 0.2201*** 0.3680** 0.2259

(0.076) (0.078) (0.029) (0.026) (0.052) (0.059) (0.186) (0.179)

Germany -0.0055 0.0009 0.1243*** 0.1209*** 0.1681*** 0.1606*** 0.3265* 0.2317

(0.078) (0.076) (0.024) (0.025) (0.053) (0.055) (0.196) (0.186)

Italy -0.0463 -0.0290 0.0644*** 0.0752*** 0.1654*** 0.1524*** 0.2472 0.1597

(0.076) (0.078) (0.020) (0.018) (0.043) (0.048) (0.175) (0.164)

Luxembourg 0.2462*** 0.2465***

- - 0.1953*** 0.2324*** 0.3059** 0.3022**

(0.067) (0.076) (0.070) (0.074) (0.145) (0.146)

Sweden 0.3899*** 0.3862***

- - 0.3644*** 0.3779*** 0.3951* 0.3128

(0.079) (0.078) (0.047) (0.051) (0.212) (0.205)

UK 0.1377** 0.0297

- - - - 0.1207 -0.0200

(0.066) (0.075) (0.147) (0.177)

T2003 -

T2004 0.0100

- -0.0022** -0.0020* -0.0132** -0.0103* -0.0040 -0.0128

(0.008) (0.001) (0.001) (0.006) (0.006) (0.012) (0.014)

T2005 - -0.0020

- - -0.0072 -0.0041

- - (0.009) (0.007) (0.007)

T2006 0.0223 0.0156 -0.0250*** -0.0260*** -0.0258*** -0.0235*** -0.0140 -0.0332

(0.012)* (0.013) (0.003) (0.003) (0.004) (0.004) (0.020) (0.024)

T2007 -0.0004 -0.0046 -0.0315*** -0.0327*** -0.0395*** -0.0376*** -0.0277 -0.0409*

(0.013) (0.014) (0.002) (0.003) (0.006) (0.005) (0.022) (0.024)

T2008 -0.0462*** -0.0508*** -0.0476*** -0.0454*** -0.0396** -0.0390*** -0.0505*** -0.0529***

198

(0.010) (0.011) (0.002) (0.003) (0.005) (0.005) (0.014) (0.015)

T2009 -0.0421 -0.0430 -0.0088* -0.0022 0.0222 0.0247 0.0112 0.0170

(0.025) (0.027) (0.005) (0.008) (0.016) (0.015) (0.028) (0.028)

T2010 0.0729*** 0.0654*** 0.0038 0.0040

- - -0.0003 -0.0049

(0.016) (0.016) (0.004) (0.004) (0.021) (0.019)

Intercept 0.7863*** 0.8063*** 0.7156*** 0.7085*** 0.4751*** 0.3196*** 0.3334 -0.0161

(0.123) (0.178) (0.035) (0.037) (0.071) (0.089) (0.274) (0.309)

R-squared 0.656 0.655 0.833 0.834 0.870 0.875 0.796 0.809

Obs 1612 1611 8798 8798 4009 4009 800 798

Note: OLS regressions for every single productive specialisation of banks (commercial, cooperative, savings, other) with clustered standard errors at the bank level

utilising country and time fixed effects to allow for unobserved heterogeneity. The second column per bank type is the expansion of the information set to

comprise the quadratic term of asset size as well as the concentration and market share with respect to deposits and loans markets. Standard errors are in

parentheses while asterisks ***, **, * denote the significance level being at 1%, 5% and 10%, respectively19

.

19

We drop the dummies of 2002 and Spain for multicollinearity reasons.

199

On the other hand, market power in terms of market share appears significant only in

the aggregate for commercial, cooperative and other banks; however, market share of

cooperative banks with respect to deposits and loans is highly correlated with the

Lerner index. Such results seem to replicate the falsities of the whole sample

explained above.

Efficiency hypothesis is again persistently existent in all bank categories as cost-to-

income ratio performs robustly at 1% level of significance. Moreover, credit risk is

positively related to Lerner index and significant for commercial and cooperative

banks thereby implying risk-taking behavior. The „other‟ category demonstrates a

negative relationship, though significant at ten% level in the first model. Moreover,

exposure to more off-balance sheet activities makes all banks insulate to potential

risks and, therefore, free to apply high profit margins.

The procyclical effect of GDP growth on competition is persistent for commercial and

saving banks while the elasticity of aggregate demand is significant for commercial

and cooperative banks at 5% level with a sign indicating a negative relationship

between credit granting to private sector and market power. Strength of legal rights is

significant only in the category of cooperative banks maintaining the same negative

sign as in table 5.

Up next in table 7, we apply seemingly unrelated regressions of income-specific

Lerner indexes on key features of market structure and power (Herfindahl index, asset

size, marker share) along with the quadratic term and disaggregated variables on bank

deposits and loans. We employ different information set for each index since

otherwise the methodology would fall into equation-by-equation OLS. The dependent

variables are Lerner on loans (L-loans), other interest income (L-otherint), fees (L-

fees) and other non-interest income (L-othernint). In particular, we include the

following new variables most akin to the previous factors employed:

L-loans: cost-to-income ratio, loan losses and liquidity

L-otherint: liquidity, GDP per capita and legal strength

L-fees: diversification, private credit and inflation

L-othernint: diversification, equity and private credit.

200

Table 7: SUR for commercial banks

Commercial L-loans L-otherint L-fees L-othernint L-loans L-otherint L-fees L-othernint

lnQ 0.034*** -0.716** -0.021 -3.629** 0.340*** -2.127 -2.515** -32.562***

2.84 -2.16 -0.09 -2.15 5.00 -1.27 -1.98 -3.60

(LnQ)^2 - -0.022*** 0.103 0.175** 2.044***

-4.73 0.84 2.20 3.36

Concentration -1.395 2.978 15.554** 98.523

- 1.57 0.19 2.20 0.94

Concentration (deposits) - -1.484 2.375 10.059 78.132

-1.38 0.10 0.86 0.71

Concentration (loans) - -0.691 7.454 4.973 -22.432

-1.42 0.60 0.57 -0.31

Market share -0.917 28.477 -19.847** 50.969

- -1.01 0.93 -1.99 0.54

Market share (deposits) - -4.316*** 78.435 -24.921 230.891

-3.14 1.62 -1.56 1.40

Market share (loans) - 5.635*** -65.045 -2.822 -308.762**

4.69 -1.56 -0.19 -2.21

Cost/TI -0.480***

- - - -0.416***

- - - -3.65 -3.36

Loan impairment -0.064***

- - - -0.052***

- - - -4.37 -3.88

OBS/TA - - 14.791 104.071

- - 13.811 93.689

0.53 0.59 0.51 0.60

EQ/TA - - - 14.561

- - - 10.694

0.75 0.55

201

Liquidity -0.157*** -0.118

- - -0.124** -0.510

- - -2.76 -0.13 -2.28 -0.52

GDPPC - 1.673

- - 2.032

- - 0.15 0.18

Private credit - - 1.017 9.263

- - 0.106 13.017

0.66 0.46 0.06 0.61

Inflation - - -46.226***

- - - -51.325***

- -2.83 -3.10

Legal strength - -1.036**

- - - -1.038**

- - -2.38 -2.38

Country effects YES YES

Time effects NO NO

R-squared 0.660 0.656 0.627 0.645 0.684 0.659 0.632 0.653

Obs 830 830 830 830 830 830 830 830

The table summarizes the results of commercial banks through two Seemingly Unrelated Regressions, the first one including total assets, market share and

concentration apart from other controls and country fixed effects and the other analysing the effect of total assets squared, concentration and market share with respect

to deposits and loans. We opt to exclude time fixed effects since otherwise standard errors could not be estimated. The estimation allows for correlation among errors

across the four models within each bank as well as heteroskedasticity by means of bootstrapping. The four models have as independent variables Lerner indexes

specific to four distinctive sources of bank income: Loans, other interest income, fees/commissions and other non-interest income. The information sets of the models

are equally sized but not equivalent whatsoever; otherwise we would fall into the standard case of equation-by-equation OLS. T-statistic is reported below each

coefficient while asterisks ***, **, * denote the significance level being at 1%, 5% and 10%, respectively.

202

First, the selection of these variables stems from the ability of the banking literature to

pin down a theory of the interconnection amongst them and to verify it empirically.

Thus, we pick up the factors (as well as those of the baseline model) that determine

the highest possible explanatory power for the whole sample before it comes to repeat

the regressions for every bank type. Otherwise, limited comparability across

heterogeneous subsets of the sample makes it hard for us to draw remarks on the

empirical evidence. We also use bootstrapping methodology to reduce the inference

bias induced by groupwise heteroskedasticity along with the within-panel correlation

of standard error across models. In some cases (commercial, other banks) where

standard errors could not be computed by bootstrapping due to limited observations,

we opt to exclude time effects from the analysis.

Table 8 exhibits the lowest fit of the 830 observations, as R-squared ranges between

62.7% and 68.4%. Along the lines of Petersen and Rajan (1995) and Berlin and

Mester (1999) that banks specialized in lending/deposit-taking enjoy higher margins

due to long-lasting relationships, we may justify the low explanatory power of the

commercial modeling. Furthermore, asset size seems to be highly significant for L-

otherint and L-othernint models with a negative bearing while it turns negative for L-

loans. In other words, as banks increase their size, prices on other interest and non-

interest products (loans) are plummeting (increasing). Non-linearities exist in all cases

but L-otherint; as banks are getting bigger in asset size they tend to offer lower other

interest and non-interest profit margins and higher loan rates. After a certain point,

higher bank size goes the other way around making banks follow the exactly opposite

trend.

Moreover, we find an insignificant correlation between concentrated structures in

loans (deposits) with all Lerner indexes (L-loans) failing to corroborate the studies of

Cetorelli and Gamberra (2001) and Angelini and Cetorelli (2003). The significant and

positive effect of aggregate concentration on the Lerner index of fees partly justifies

the ability of European member states to exploit concentrated markets in order to

boost their performance (Gondat-Larralde and Lepetit, 2001). Besides, the market

share is negatively correlated with monopolistic pricing in the L-fees case; however,

when it comes to their disaggregated constituents, banks tend to decrease (increase)

prices on loans as their market share in the deposits (loans) markets is getting higher.

203

Such evidence, coupled with relative efficiency gains (significant TC/TI), is

indicative of the „cream skimming‟ phenomenon, which assumes that the process of

integration forces foreigners to pick up banks of higher operational efficiency to grasp

higher market shares monopolising loan markets. Accounting for the negative

coefficient of market share in deposits we give more credit to the competitive conduct

of banks in the deposit market so as to attract new customers. It is also interesting that

the pattern of market share in the loan markets is opposite between traditional (loans)

and off-balance sheet activities (othernint) possibly on account of risk sharing or

hedging. These results are perhaps reminiscent of the hold-up problem of relationship

banking, namely the ability of banks with monopoly power and proprietary

information to charge higher loan rates on customers.

Loan losses have a significant and negative impact on L-loans, implying that banks

may tend to react by imposing lower loan prices and holding other risk-free

investments in order to rationalize their loan books. Commercial banks with high

share of liquid assets are willing to impose lower prices on loans, a pattern which is

not accepted ex ante. In fact, banks may consider liquid assets as a safety net for

potential losses and have no need for excessive risk-taking. In times of economic

booms when inflation pressures are considerable, commercial banks lessen their fees

and commissions in an attempt to boost the volume of transactions. In addition, legal

framework that protects both borrowers and lenders enhances competition in other

interest products.

In table 8, cooperative banks enumerate 6331 observations with a R-squared ranging

from 64.5% in L-otherint to 89.4% in L-fees; the rest are up to levels of 77.9% and

77.5% for L-loans and L-othernint, respectively. The natural logarithm of total assets

is negatively associated with all cases but L-fees at 1% level of significance while

there exists non-linear relationship in the models of L-loans and L-fees. It is positive

up to a certain level of assets when it turns negative at higher levels of scale

economies. However, we observe the opposite trend in L-othernint, according to

which the negative bearing switches to positive.

204

Table 8: SUR for cooperative banks

Cooperative L-loans L-otherint L-fees L-othernint L-loans L-otherint L-fees L-othernint

LnQ -0.019*** -0.353*** 0.399*** -3.747*** 0.053** -0.574 5.001*** -11.900***

-5.24 -4.30 10.87 -15.89 2.49 -1.12 16.13 -8.23

(LnQ)^2 - -0.006*** 0.018 -0.396*** 0.700***

-3.67 0.34 -16.69 4.63

Concentration 0.585 -3.215 0.355 16.908

- 1.62 -0.54 0.07 0.84

Concentration (deposits) - -13.175*** 80.076 -33.922 -822.439

-3.03 0.34 -0.20 -0.68

Concentration (loans) - 13.854*** -83.087 37.683 818.669

3.02 -0.35 0.23 0.66

Market share 0.947 -113.862*** -59.845*** 44.741

- 1.29 -3.87 -4.41 0.30

Market share (deposits) - -26.258*** 140.772 47.742 154.660

-4.94 1.50 1.17 0.62

Market share (loans) - 23.452*** -224.168*** -21.603*** -212.271

3.03 -4.11 -5.31 -1.37

Cost/TI -1.087***

- - - -1.095***

- - - -21.93 -41.96


- - - -4.612***

- - - -8.47 -5.05

OBS/TA - - 294.011*** 2166.794***

- - 271.227*** 2215.934***

9.26 13.26 9.16 32.33

EQ/TA - - - 115.077***

- - - 116.478***

5.08 59.67

Liquidity 0.007 -3.048*** - - 0.047 -3.263*** - -

205

0.24 -3.88 1.26 -4.62

GDPPC - 12.327

- - 16.067***

- - 1.11 8.65

Private credit - - -5.097*** 8.358

- - -6.281* 9.174

-3.17 0.79 -1.93 0.30

Inflation - - 66.989*

- - - 34.653

- 1.82 0.95

Legal strength - 1.000

- - - 1.000

- - 1.16 1.16


Time effects NO NO

R-squared 0.776 0.645 0.886 0.754 0.779 0.646 0.894 0.755

Obs 6331 6331 6331 6331 6331 6331 6331 6331

The table summarizes the results of cooperative banks through two Seemingly Unrelated Regressions, the first one including total assets, market share and

concentration apart from other controls and country/time fixed effects and the other analysing the effect of total assets squared, concentration and market share with

respect to deposits and loans. The estimation allows for correlation among errors across the four models within each bank as well as heteroskedasticity by means of

bootstrapping. The four models have as independent variables Lerner indexes specific to four distinctive sources of bank income: Loans, other interest income,

fees/commissions and other non-interest income. The information sets of the models are equally sized but not equivalent whatsoever; otherwise we would fall into the

standard case of equation-by-equation OLS. T-statistic is reported below each coefficient while asterisks ***, **, * denote the significance level being at 1%, 5% and

10%, respectively.

206

SCP paradigm is rejected on the grounds of insignificant effect of aggregate

concentration on the price of any income source. We nonetheless observe negative

association of concentration in the deposit market with L-loans. The positive sign in

the loans markets, as Corvoisier and Gropp (2002) argued also about, instantiates the

passing-through mechanism of causality in the specific case of cooperative banks.

In addition, banks with greater market share behave competitively when they impose

prices on other interest income, fees and commissions. Relative market power is

scantily existent in the L-loans model having a negative (positive) bearing in the

deposit (loan) market. It is also the case of Efficient structure hypothesis for L-loans

as the pertinent coefficient is negative at 1% level of significance with considerable

credit losses making also banks reluctant to succumb to high prices on loans.

According to it, amid competitive conditions in the market more efficient banks are

likely to survive by means of eliciting greater market share from less efficient

institutions (Demsetz, 1973).

The share of OBS to total assets has a considerably positive impact on L-fees and L-

otherint corroborating the tendency of well-diversified banks to impose high margins.

Bank capital demonstrates, as expected, a positive sign suggesting that a higher

degree of risk aversion (high capital ratio) is transmuted to higher margin on OBS

activities to make up for the inherent systematic risk20

. What is more, the degree of

liquidity of bank assets turns out to have a negative effect on L-otherint, as opposed to

the case of more liquid commercial banks that prefer to lower prices on loans.

Private credit is also an important factor for L-fees with a negative sign without losing

significance when there is disaggregation of concentration and market share or the

non-linear term of asset size. Moreover, we observe an indication that the price effect

of economic growth as well as proactive legal initiatives in favour of consumer

protection operates procyclically towards monopolistic practices in fee and other

interest income, respectively.

20

Alternatively, it may constitute a signal in the market of creditworthiness, but since the sample

comprises only developed economies with strong legal environment, there is need in having high

capital buffers in order to restore depositor confidence (Claeys and Vennet, 2008).

207

Table 9: SUR for savings banks

Savings L-loans L-otherint L-fees L-othernint L-loans L-otherint L-fees L-othernint

LnQ -0.005 0.185 -0.125*** 6.242*** -0.044*** 2.393* 2.216*** 6.305*

-1.28 1.12 -3.21 9.67 -3.01 1.66 6.03 1.70

(LnQ)^2 - 0.003*** -0.159* -0.162*** 0.012

3.34 -1.89 -7.36 0.04

Concentration -1.17 61.661** -11.216 -185.585

- -1.06 1.99 -1.32 -0.88

Concentration (deposits) - -3.643 51.943 -35.470 -1041.498

-1.03 0.24 -0.56 -1.34

Concentration (loans) - -0.268 138.899 31.849 689.327

-0.10 1.62 0.58 0.92

Market share 0.176 -108.646*** 5.136 -377.764

- 0.57 -3.87 0.52 -1.25

Market share (deposits) - -7.610 305.896 -21.090 -959.774

-1.15 0.73 -0.53 -0.73

Market share (loans) - 7.438 -385.343 6.085 500.170

0.87 -0.73 0.15 0.31

Cost/TI -1.032***

- - - -1.032***

- - - -28.73 -34.39


- - - -10.820***

- - - -21.10 -22.35

OBS/TA - - 415.473*** 4333.502***

- - 407.844*** 4474.753***

14.21 16.03 10.72 5.60

EQ/TA - - 74.371**

- - - 67.813*

2.10 1.68

Liquidity -0.321*** 11.551*** - - -0.304*** 10.803*** - -

208

-6.79 19.91 -4.56 3.62

GDPPC - 22.278

- - - 37.148

- - 0.83 1.63

Private credit - - -0.727 -103.363***

- - -2.108 -104.621***

-0.56 -7.62 -1.29 -8.14

Inflation - - 8.931

- - - 2.253

- 0.30 0.06

Legal strength - -0.990***

- - - -1.015***

- - -2.65 -3.03


Time effects YES YES

R-squared 0.778 0.605 0.953 0.777 0.780 0.610 0.954 0.779

Obs 3390 3390 3390 3390 3390 3390 3390 3390

The table summarizes the results of savings banks through two Seemingly Unrelated Regressions, the first one including total assets, market share and concentration

apart from other controls and country/time fixed effects and the other analysing the effect of total assets squared, concentration and market share with respect to

deposits and loans. The estimation allows for correlation among errors across the four models within each bank as well as heteroskedasticity by means of bootstrapping.

The four models have as independent variables Lerner indexes specific to four distinctive sources of bank income: Loans, other interest income, fees/commissions and

other non-interest income. The information sets of the models are equally sized but not equivalent whatsoever; otherwise we would fall into the standard case of

equation-by-equation OLS. T-statistic is reported below each coefficient while asterisks ***, **, * denote the significance level being at 1%, 5% and 10%, respectively.

209

We next analyse the regression output for savings banks (table 9) based upon 3390

observations in each model. The fit here is between 60.5% (L-otherint) and 95.4% (L-

fees) while the rest settles at about 78%. Asset size does have a positive effect on L-

othernint and a negative one on L-fees. Non-linearities are traced in the L-loans

following a negative trend as soon as it turns negative in higher scale economies. In

contrast, L-fees and L-otherint models report a positive-to-negative nexus, although

the latter comes with 10% significance level. Concentration in EU member states has

only a positive effect in the aggregate on L-otherint followed by the market share with

the same pattern, as well.

Cost efficiency along with credit risk is negatively interconnected with L-loans at 1%

level compatible with the practice of banks to exercize market power on the grounds

of reducing costs. Thus, there is no indication of „quit life‟ hypothesis, according to

which banks not vulnerable to intense competition, managers by no means seek to

maximize profits through an everlasting cost reduction (Berger and Hannan, 1998;

Delis and Tsionas, 2009). Rather pure efficiency hypothesis comes into play, which

sets out the ability of efficient banks to engage in monopolistic price without

intending to higher market share. In addition, diversified portfolios make banks feel

safer to charge higher margins on OBS activities offsetting at least in part the positive

effect of loan losses on loans competitive pricing.

Moreover, banks with high share of liquid assets engage in competitive pricing on

loans and monopolistic conduct on other interest-bearing activities; however, that is in

contradistinction to the pattern observed in cooperative banks. As cooperative banks

operate in a decentralized system of rather national as well as regional outreach,

liquid assets enable banks to enhance long-standing relationships through cheaper

loans and more expensive prices on trading and investment securities. That may also

be conducive from the opportunity cost that banks are bound to bear as a result of

their obligation to withhold liquid reserves. Thus, higher loan prices compensates for

potentially higher interest rates being available in the financial markets (Hawtrey and

Liand, 2008). It is intuitively relevant the argument of Lakonishok et al (1992),

according to which larger banks with low liquid assets share demonstrate a herding

behaviour in excessive risk-taking in other (interest) income sources by charging

higher profit margins.

210

Table 10: SUR for 'other' banks

Other L-loans L-otherint L-fees L-othernint L-loans L-otherint L-fees L-othernint

LnQ -0.051*** 0.651* -1.224*** -8.790*** 0.231*** -4.753** 3.736*** -0.196

3.24 1.88 -5.02 -6.12 2.79 -2.38 3.19 -0.02

(LnQ)^2 - -0.014** 0.389*** -0.324*** -0.626

-2.41 2.99 -4.29 -1.14

Concentration -0.635 17.604 16.706 78.663

- -0.58 0.92 0.72 0.36

Concentration (deposits) - -1.670** 38.086** -0.174 81.991

-2.27 1.97 -0.01 0.45

Concentration (loans) - -0.279 -9.189 10.526 92.924

-0.85 -0.78 0.80 1.15

Market share -6.835*** 89.601*** 59.033** -368.921**

- -4.02 2.83 2.57 -2.29

Market share (deposits) - -6.644** 36.130 157.260*** -204.002

-2.11 1.11 4.37 -0.82

Market share (loans) - 4.617 -38.753 -108.576** -93.461

1.26 -1.25 -2.55 -0.31

Cost/TI -0.358***

- - - -0.369***

- - - -3.85 -3.79

Loan impairment 0.022

- - - 0.082

- - - 0.03 0.11

OBS/TA - - 29.895 178.205**

- - 37.087* 191.014***

1.44 2.54 1.80 2.64

EQ/TA - - - -17.541 - - - -26.490

211

-0.87 -1.17

Liquidity -0.097 -1.411

- - -0.066 -1.825

- - -1.59 -1.10 -1.08 -1.38

GDPPC - 20.913

- - - 18.341

- - 1.32 1.11

Private credit - - 10.877** 3.890

- - 5.368 -40.142

2.55 0.11 1.05 -1.10

Inflation - - 0.777

- - - -6.282

- 0.02 -0.21

Legal strength - -1.666***

- - - -1.676***

- - -3.28 -3.06


Time effects NO NO

R-squared 0.746 0.661 0.781 0.663 0.747 0.664 0.314 0.667

Obs 396 396 396 396 396 396 396 396

The table summarizes the results of „other‟ banks through two Seemingly Unrelated Regressions, the first one including total assets, market share and concentration

apart from other controls and country fixed effects and the other analysing the effect of total assets squared, concentration and market share with respect to deposits and

loans. We opt to exclude time fixed effects since otherwise standard errors could not be estimated. The estimation allows for correlation among errors across the four

models within each bank as well as heteroskedasticity by means of bootstrapping. The four models have as independent variables Lerner indexes specific to four

distinctive sources of bank income: Loans, other interest income, fees/commissions and other non-interest income. The information sets of the models are equally sized

but not equivalent whatsoever; otherwise we would fall into the standard case of equation-by-equation OLS. T-statistic is reported below each coefficient while

asterisks ***, **, * denote the significance level being at 1%, 5% and 10%, respectively.

212

More private credit and legal stringency are persistently significant and negatively

related to L-othernint and L-otherint, respectively. As inflation and GDP per capita

are insignificant we conclude that there pricing conduct is not contingent on different

stages of economic development.

The remaining banking sector (table 10) constitutes a heterogeneous group of

different financial institutions. In table 10, the results are produced out of 396

observations concluding to moderate R-squared values ranging from 66.1% in L-

othernint to 74.7% in L-loans model; the 31.4% fit in L-fees model is the half of that

in the first specification of no disaggregation and non-linear term. That may be

attributed to the more terms that exacerbate the degree of multicollinearity.

As for asset size, it takes a positive coefficient at 1% significance level for L-otherint

and negative for L-loans, L-fees and L-otherint. Besides, non-linear relationship

between size and competition is evident in the first three models, in which savings

banks seem to charge higher (lower) prices on loans and fees (other interest-bearing

products) but further down the road they reap the benefits of scale economies to

reverse their strategy.

In the model of L-loan, some sort of contestability justifies the negative sign of

concentration in the deposits markets at 5% level of significance. On the other hand,

collusion in the deposit market motivates incumbent banks to monopolize the market

through higher margins on other interest income. Furthermore, banks tend to exploit

their relative market power by imposing higher prices on other interest-bearing

products and fees, and lower margins on loans and other non-interest activities. That

is further decomposed into a negative bearing of market share in deposits markets on

L-loans as well as positive (negative) effect of it in the deposits (loans) market.

Cost efficiency persistently contains a negative effect on L-loans corroborating the

results of Maudos and De Guevara (2007) diversification in bank portfolios and

demand elasticity positively affect L-othernint and L-fees, respectively. Legal

stringency maintains its preemptive repercussion against market monopoly.

213

6.6. Conclusion

The aim of the paper is to investigate potential effects of bank competition as the

latter is defined as the price mark-up over marginal cost. We end up with

monopolistic pricing to be persistently associated with low levels of asset size for

cooperative, savings and „other‟ banks. The relationship turns out to be non-linear for

savings and „other‟ banks following a positive pattern for low levels of total assets up

to a point where contestable conduct or economies of scale turn it negative.

Cooperative banks experience a negative pattern only in lower values of total assets

since the dominating cooperative banks of Germany constitute a homogenous set of

equally sized institutions.

The positive sign of concentration gives credit to the SCP paradigm once we see the

general picture of European banking sectors as well as the savings banks themselves.

When it comes to construct Herfindahl-Hirschman indices for deposits and loans

products, concentration is significant only for „other‟ banks in loans market, and for

commercial banks in both deposits and loans markets. Notwithstanding that relative

market power bears no significance in the whole sample, a positive indication is

traced in „other‟ banks and in greater levels of market share of cooperative banks.

However, a more competitive pricing seems a common practice for commercial banks

and for cooperative banks of relatively low market share.

Relative efficiency hypothesis is verified against „quit life hypothesis‟ by the negative

effect of total cost over total income variable at 1% level of significance along the

lines of Koetter et al. (2012), who concluded the same for US banks. We verify,

therefore, the strategic option of banks to exploit lower costs in favour of their

customers either in the form of lower loan rates or higher deposit rates (Vennet,

2002). Furthermore, credit risk motivates commercial and cooperative banks to apply

higher profit margins, a strategy which is contradistinction with that of „other‟ banks.

As for the economic conditions, we can only examine the effect of GDP growth rate

on banks of specific specialization. In fact, high rates of economic expansion operate

procyclically for savings banks and countercyclically for commercial banks. Higher

elasticity of aggregate demand also induces banks to offer lower prices, especially in

214

the case of commercial and cooperative banks. Legal stringency proves to have a

positive impact only on the competitive behavior of cooperative banks.

Next, we delve into the analysis of how such - or next akin to – effects influence the

pricing conduct of banks at the income level. In the same vein, commercial banks of

greater asset size tend to impose low prices on other interest and non-interest

products, a case that remains the same for cooperative banks, which opt for

competitive price on loans as well as higher fees. Economies of scale seem to produce

lower fees and higher prices on other non-interest products for savings banks whereas

„other‟ banks apply generally lower prices except for other interest activities. Non-

linearities have the same pattern for commercial, cooperative and „other‟ banks

yielding a negative sign on L-othernint as soon as it turns positive at higher levels of

total assets; in savings banks there is only a linear and positive association. Contrary

to commercial assets, all the other banks demonstrate first a positive effect on L-fees,

which then turns into negative whereas the same tendency is traced for L-loans in all

cases except savings banks. Profit margins on other interest products demonstrate a

diminishing (increasing) trend when „other‟ (savings) banks are getting bigger,

notwithstanding higher scale economies reverse the sign.

Collusion hypothesis does play a significant role for commercial banks in

implementing higher fees while savings banks follow exactly the same strategy for

other interest income. Greater collusion in the deposit market promotes competitive

loan rates by cooperative and „other‟ banks, with the former reaping the benefits of

collusion in the loans market through higher loan rates. „Other‟ banks utilize

monopolistic structure in the deposit market by enjoying relatively higher prices on


Higher market share is correlated with lower fees in commercial and cooperative

banks and lower other interest charges in cooperative and savings banks; the latter

also prefers the utilization of low other non-interest rates. „Other‟ banks apply lower

rates on loans and other non-interest products and higher prices on the remainder.

Higher relative market power in the deposit market drives all banks to compete each

other for loan rates, a behaviour which alters in commercial and cooperative banks

when they enjoy higher market share in the loans market. In addition, a common

215

feature for cooperative and „other‟ banks of greater market share in the loans market

consists in the use of lower fees but only „other‟ banks are able to impose higher fees

when they acquire market share in the deposits market. Moreover, charges on other

interest and non-interest activities contain their negative nexus market share in the

loans market.

Cost efficiency drives all banks to charge low loan rates, which are even affected by

loan losses in the portfolios of commercial, cooperative and savings banks. More

portfolio diversification makes cooperative and savings banks eager to charge fees

and other non-interest rates with an indication that same holds for „other‟ banks when

it comes to charge fees. Higher level of total equity that reflects risk aversion or an

explicit signal of bank creditworthiness enables cooperative and savings banks to

deploy a monopolistic strategy on other non-interest activities. Cooperative banks

with more liquid assets enjoy lower rates on other interest products followed by

commercial banks in offering cheaper loans. Savings banks do the same as

commercial banks but additionally compensate the underlying opportunity cost of

liquid assets through higher levels of non-interest rates.

Average country income has a positive impact on cooperative banks in charging

higher other interest prices, while demand elasticity motives cooperative and savings

banks to apply lower fees and other non-interest charges. On the other hand, „other‟

banks seem to take advantage of higher fees. In periods of high inflation rate,

cooperative banks operate in contradistinction to commercial banks enjoying

considerable fees in order to possibly counterbalance potential losses on fixed loan

rates. Last, stricter lending laws turns out to bring about higher other interest charges

on behalf of cooperative banks against the stream of diminishing returns for the rest

banking specialisations.

217

CHAPTER 7

Cost and profit efficiency in European banking: Comparison of

parametric methodologies and convergence dynamics

219

7.1. Introduction

Operational performance is a hot topic in the banking literature since it is related to

contemporary issues of policy making like macroprudential regulation, market

structure and systemic risk. There is a bulk of published papers investigating the way

we model cost functions and extrapolate inefficiency scores, the relationships between

performance and risk, competition, M&As, ownership, profitability and regulation

until convergence and cross-country comparisons.

Some topics though seem to be outdated as covering periods in the „90s and 80‟s,

skewed in favour of other markets (e.g. US), or even new areas of academic research.

For example, the comparison of different methodologies in estimating bank

performance is addressed in the study of Bauer et al. (1998) juxtaposing the results of

DEA, SFA, TFA and DFA for a sample of US banks over the period 1977-1988 to

conclude about the consistency within the set of parametric and nonparametric

approaches and inconsistency between them. The results are corroborated by

Cummins and Zi (1998), who studied the US life insurance industry during 1988-

1992 and alleged about the significant differences between the results of DEA/FDH

and DFA; however, nonparametric approaches (DEA, FDH) had considerable

divergence, as well. In the same vein, Huang and Wang (2002) covered the banking

sector of Taiwan from 1982 until 1997 coming up with robust evidence within

nonparametric approaches and parametric models and inconsistency between

SFA/DFA and DEA.

Other studies on European samples are those of Maudos et al. (2002), Casu et al.

(2004), Weill (2004a) and Delis et al. (2009). Maudos et al. (2002), which endorse

similar country rankings in Europe among DFA, FEM and REM during the period

1993-1996, a result which seems close to the conclusion of Casu et al. (2002) about

„not markedly different results‟ for DEA and TFA approaches in France, Germany,

Italy, Spain and UK. Moreover, considering an almost similar sample of countries and

time period, Weill (2004a) found significant correlation between SFA and DFA but

no relationship between them and DEA, whatsoever.

Next, another group of studies analyse cost and profit efficiency concurrently. Higher

performance in the cost side is evident in the majority of empirical studies, except

220

some few cases. The latter includes SFA methodology traced in Isik and Hasan

(2002), who concluded about the higher performance in the output mix of Turkish

banks during 1988-1996, Mertens and Urga (2001) for Ukraine in 1998, Vennet

(2002) over a European sample in 1995 and 1996 that traced high profit efficiency in

financial conglomerates and non-German universal and specialized banks, and Srairi

(2010) within the period 1997-2007 for Oman.

As for country specific studies, relatively higher cost efficiency is found in China

under SFA and DEA during 1993-2004 (Ariff and Can, 2008; Berger et al., 2009), in

India according to DEA from 1992 to 2004 (Das and Ghosh, 2009; Ray and Das,

2010), in Spain under DEA within the period 1985-1996 (Maudos and Pastor, 2003)

and Ukraine in 1998 for the special case of TFA methodology (Mertens and Urga,

2001). In addition, Bonin et al. (2005) and Yildirim and Philippatos (2007b)

corroborate cost-side effectiveness in samples of European developing countries

under SFA along the lines of Kasman and Yildirim (2006), who employ an enlarged

European group during the years 1995-2002. The remainder comprises international

samples, such as those of Gulf cooperation council countries (Srairi, 2010), EU, Japan

and US (Maudos and Pastor (2001), Africa, Asia and Middle East (Bader et al. 2008)

and 74 countries (Pasiouras et al., 2009), which all settle to high cost efficiency under

SFA and DEA approaches.

A quite smaller group of studies employ different definitions of bank outputs, namely

the intermediation and value-added approaches in particular. They are used either for

comparison purposes or for robustness checking purposes. Indeed, Glass et al. (1998)

implemented the cost function of Pulley and Braunstein (1992) for Japan within the

period 1977-1993 to come up with mixed results between economies of scale and

scope and better fit of VA approach. Sufian (1999) employed DEA for the Malaysian

banking industry and verified relatively higher average technical and scale efficiency

under VA approach, whereas Tortosa-Ausina (2002) highlighted similarly a higher

and more stable DEA efficiency scores in Spain.

The concept of economies of scale has been widely analysed for the US region since

the „80s dominating the literature by a clear mile. Hughes and Mester (1998)

identified explicit economies of scale through IA-based function augmented by OBS

activities as well as Evanoff et al. (1990) by means of a shadow-price model.

221

Under IA approach, Jagtiani et al. (1995) identify through the model of Christensen et

al. (1973) economies of scale, which turn out to be lower than 5% according to Berger

and Humphrey (1991). Scale economies do exist only in smaller sizes according to

Noulas et al. (1990), Gilligan and Smirlock (1984), Wheelock and Wilson (2001),

Cavallo and Rossi (2001), Gilligan et al. (1984), McAllister and McManus (1993),

who employed translog and quadratic specifications, kernel regressions, spline

regressions and polynomial smoothing, the model of Benston (1972) and Bell and

Murphy (1968), the intermediation, value-added and production approaches, among

others. Furthermore, diseconomies of scale are traced across branching and unit state

asset classes (Berger et al., 1987), money center and super-regional banks and

diversified US securities banks.

As for European studies, Altunbas and Molyneux (1996) observe scale economies in

banks of all size classes and different outputs over a sample of France, Germany, Italy

and Spain. Similar evidence comes out of dynamic modeling by Carvallo and Rossi

(2001) although some degree of exhaustion takes place at low output levels. Large

branch offices in Finland experience scale economies up until 2.5 million Finn marks

(Zardakoohi and Kolari, 1994) opposing to the case of Norway, which picks up

economies of scale in small banks and mixed results in any asset size class when total

assets are used as a single output variable (Humphrey and Vale, 2004). Scale

economies are evident across all asset classes in mutual funds of France in 1987 and

in the cooperative banks of Germany during the period 1989-1992. Last, Pasiouras

(2008a) draw remarks on the persistent scale economies of purely domestic banks and

relatively lower economies in international banks for the period 2000-2004.

Last, to our knowledge, beta and sigma convergence of bank efficiency enumerates

four papers. The US banking market experiences conditional convergence of DEA

scores among bank holding companies within the period 1996-2003 (Fung, 2006).

Casu and Girardone (2010) employ DEA and dynamic models over the developed

European group to endorse the lagging-behind rather than the catching-up tendency of

cost efficiency during the 1997-2003 period, whereas Weill (2009) across different

specifications found beta and sigma convergence in cost efficiency. At that time, the

developing European subgroup documents catching-up effect and dispersion

exclusively in the cost efficiency case.

222

Hence, the aim of this paper is threefold: First, to estimate the consistency of cost and

profit efficiency scores for the whole European region by means of various parametric

approaches and under different definitions of bank outputs. Second, we estimate scale

economies under the intermediation and value-added approach for each banking

market alongside different productive specialisations and asset classes. Third, we

verify whether the European challenge of market integration is accompanied by beta

and sigma convergence of cost and profit efficiency amongst financial institutions.

7.2. Methodological issues

This chapter comprises a set of methodological issues that need to be considered for

the sake of efficient data application. The first refers to the need of deciding upon the

best approach among parametric and non-parametric approaches; that is, econometric

modeling and linear programming. The observed inconsistency of the empirical

studies is attributed to the different structure of employed modeling as DEA and FDH

excludes the random error that accounts for shocks, measurement errors and data

problems.

In contrast, researchers have employed various structures of the inefficiency term in

SFA, apart from the other constituent of the composite error term that presumably

follows the standard normal distribution. SFA is also preferred against thick frontier

and distribution-free approaches since the former poses the strict assumption of

constancy on the error term and the latter runs the risk of ascribing cost differences to

the large values of the random error. In particular, panel data analysis constitutes a

better methodology for the estimation of technical inefficiency according to Baltagi

and Griffin (1988), Cornwell et al. (1990) and Kumbhakar (1993). The random effects

model is also suggested due to its fit in studies of large N bank groups since the fixed

effects model is arguably inappropriate for leading to reliable indicators of efficiency

(Gathon and Perelman, 1992; Simar, 1992) due to potential loss of freedom (Baltagi,

1995) and confounded efficiency scores across different asset sizes (Berger, 1993).

The way inputs and outputs of banks are defined determines the empirical analysis

quantitatively since the efficiency scores are different (Resti, 1997). Several

approaches have been proposed in the literature, namely the intermediation and

223

production approach as well as the asset, user cost and value-added approach. In

particular, the intermediation approach as first proposed by Sealey and Lindley (1977)

treat deposits as input; their cogitation tracks the operating process from the pool of

depository funds to their final transformation to loans and other assets through the

utilisation of capital and labour. The production approach by Ferrier and Lovell

(1990) treats deposits from a different angle. It measures loans and deposits in

quantities and considers them as the bank‟s production of services with the use of

labour and capital and the inclusion of interest expenses in total cost (as

intermediation approach sets out).

The other three are employed in relatively limited cases: the value-added approach

(Berger and Humphrey, 1992) defines output as any item, irrespectively of which side

of the balance sheet it gets into, being the product or a share of the capital-labour mix,

otherwise as input. According to Humphrey and Pulley (1997) it is assumed that

banks provide two types of financial services: intermediation and loan services and

payment, liquidity and safekeeping services. As for the user cost approach, deposits

and any other item are bank output if its net contribution to revenues is positive.

However, asset approach is sparsely used, though as a variant of the intermediation

approach much to the idea of accounting liabilities (deposits) as inputs and assets as

outputs. In this paper the intermediation and value-added approaches are adopted to

allow for cross-study comparisons in the European region.

The comparison of efficiency scores across borders and different sectors in another

issue that should be resolved. Anything that is out of managers‟ jurisdiction and area

of control might be proxied by country and industry specific factors that capture the

underlying variation. Otherwise, under the same cost frontier, the necessary

assumption that the efficiency differences are attributed only to managers‟ decisions

seems quite unrealistic and jeopardizes, for one thing, the empirical analysis.

However, we opt alternatively to compute as many frontiers as are the European

countries so as to exploit data availability, even in the Southeastern Union, and

thereby draw secure remarks on cross-country rankings.

Moreover, the appropriate specification of the cost function should be concluded upon

the data of the sample and amongst three alternatives: the translog cost function (TL),

the Box-Cox form (BC) and the Fourier-Flexible specification (FF). The TL form is

224

the most widely applied in empirical studies once it generalizes the important in

microeconomics Cobb-Douglas function and requires few restrictions on the cost

structure and its dual production function. In contrast, the underlying form stops short

at producing scale and scope economies at zero production output whereas White

(1980) suggests that the second order calculation of the Taylor expanded function is

not necessarily identified by TL estimates. The BC formulation alleviates the former

problem by accounting for zero values; however, non-homogeneity in input prices -

found by Shaffer (1994) - is surpassed by applying transformation in a specific

function each time that suits the output data. A pitfall might be the arbitrary

approximation of the modified version using various quadratic forms.

Last, the FF specification was first developed by Gallant (1981,1982), in response to

the critique of White (1980) on the restrictive property of second order Taylor

expansion function; it was further developed by Elbadawi et al. (1983) and Gallant

and Souza (1991) while first applied in banking later on by McAllister and McManus

(1993), Berger et al. (1994), Spong et al. (1995), Mitchell and Onvural (1996), Berger

and Mester (1997). It comprises in fact the TF form enriched with trigonometric

transformation of the variables as addends, any combination of which can

demonstrate a good fit in various multivariate functions (Tolstov, 1962). In so doing,

a better fit of the data is achieved through getting better inference on the coefficients

of the cost function (De Young and Hasan, 1998)21

, although it requires further

truncations of data (Hasan and Marton, 2003) and produces similar efficiency scores

to TL (Berger and Mester, 1997).

To contribute more on the ongoing cross-country literature that tests the consistency

of efficiency scores, the remaining error component is better depicted by the per se

utilisations of different methodologies and specifications. It is also indicative that

DEA is not part of the empirical analysis since pertinent software can by no means

tolerate missing values and therefore, a considerably shrank sample is not appealing

for the scope of our research. In addition, the literature has come to a consensus

regarding differences in results between parametric and nonparametric approaches

(Bauer et al., 1998; Cummins and Zi, 1998; Delis et al., 2009; Huang and Wang,

2002; Weill, 2004a). SFA too falls short at accomplishing curvature conditions when

21

Carbo et al. (2002) thoroughly explain the reasons why FF form is more appropriate.

225

applying maximum likelihood technique along the lines of Batesse and Coelli (1992,

1995).

To this end, I opt for the REM, DFA and TFA along with the use of the

intermediation and value-added approaches.

7.3. The model

The employed model examines the effect of banks‟ activities – whether traditional or

not – on efficiency in the European region. The framework of such analysis is based

upon the construction of a „paneuropean‟ efficiency frontier against which all banks

within national markets and across regions are compared. The DFA and TFA

methodologies to be followed hereinafter contribute to the literature of comparative

studies of profit and cost efficiency such as those of Maudos et al. (2002), Bauer et al.

(1998), among others.

The models of cost and alternative profit function are estimated to produce efficiency

scores over the examined period. The latter specification instead of the traditional

profit function allows for the profit function to be determined by input prices and

output quantities thus implying uncompetitive pricing or heterogeneous groups within

the European region (Kasman and Yildirim, 2006; Maudos et al., 2002; Pasiouras et

al., 2009).

The cost and profit models are a function of the output and input vector, the level of

inefficiency (u) and the standard error term (e). Schematically,

Y Y y,w,u,e (7.1)

where Y denotes total costs and total profits before taxes of bank i at time t22

. Taking

logs at both sides, we have:

lnY f (w,q) lne lnu (7.2)

22

Subscripts are omitted for convenience purposes.

226

where lnY is the logarithm of total costs (TC) and profits before taxes (PBT), q and w

are the vectors of bank outputs and inputs, respectively. The error term is decomposed

into two parts: the e factor being identically and independently distributed that

accounts for measurement errors and luck, and the u term denoting the inefficiency

term23

. Supposing the profit maximising and cost minimising conduct of banks,

inefficiency is bound to inflate costs and diminish profits. Thus, the plus and minus

sign is appropriately assumed in cost and profit functions, respectively.

The fully expanded model takes the following form:

lnY a i lnWi

i

ii

lnQi 1

2i, j

j

i

lnWi lnW j 1

2 i, j

j

i

lnQi lnQ j

i. jj

lnWi lnQ j

i

1 lnE 1

22 lnE2 3,i

i

lnE lnWi 4,i

i

lnE lnQi lne lnu

(7.3)

First, total costs (TC) and profits before taxes (PBT) are the dependent variables. To

avoid negative data when it comes to apply logarithmic time series, the minimum

absolute value of PBT over the whole sample plus „1‟ is added to the PBT so as to

give out in the extreme at least zero values [ln(PBT+|PBTmin|+1)]. The input vector

comprises the price of labour (P1) proxied by personnel expenses over total assets

since there is no reliability of the number of bank employees, price of physical capital

(P2) (operating costs of plant and equipment (net of personnel) over fixed assets).

As for the output, loans (Q1), other earning assets (Q2) are often used in the literature

whereas off-balance sheet items (Q3) in nominal terms are employed since otherwise

their exclusion would understate total output (Jagtiani and Khanthavit, 1996). The

level of equity (E) as fixed netput is quasi que non for its use according to Berger and

Mester‟s (1997) argumentation; risk preferences vis-à-vis informational asymmetries,

regulation and financial distress are scaled over the whole sample of European banks,

the omission of which would otherwise identify inefficiency in banks that behave

optimally in line with managers‟ stance for risk taking. We also employ the value-

added approach including in the input vector the price of labour (W1) and physical

23

It is modeled with positive sign in the cost function, as inefficiency can by no means diminish total

costs, whereas in the profit model the negative sign highlights reduction in banks profits.

227

capital (W2), while output consists of loans (Y1), deposits (Y2) and off-balance sheet

activities (Y3). It takes the form:

lnY ii

lnWi ii

lnYi 1

2i, j

j

lnWi

i

lnW j 1

2 i, j

j

lnYii

lnY j

i, j lnWi lnY j lne lnuj

i

(7.4)

We opt for the Random effects model (REM) that assumes a bank-specific constant as

(in)efficiency score, which is totally orthogonal to the model coefficients for the

whole period after the advent of Euro. It is estimated according to the constrained

linear regression framework with Huber/White/sandwich standard errors, which is

quite similar to the FGLS methodology. The reason is that the former offers coverage

to any kind of within-panel correlation of disturbances rather than intragroup

heteroskedasticity, whereas asymptotic properties mandate the latter to be used for

samples of few banks and many time periods. Homogeneity of degree one in input

prices ( ii

1 ) and symmetry conditions in all quadratic terms are imposed in

models 7.3 and 7.4.

DFA is then employed as an alternative version of REM, which differentiates itself as

permitting coefficients to vary over the time period to better capture differences in the

technology and economic environment. When it comes to produce efficiency scores,

the assumption of leaving the errors to cancel each other out in the course of time

allows us to make a simple estimation: the average of residuals for every single bank.

As a next step, we truncate the sample of average residuals at the points of 1%, 5%

and 10% to verify the consistency of country rankings in terms of cost and profit

efficiency. Last, we apply the Thick Frontier Approach (TFA), regressing only the

observations of the lowest average-cost quartile and the predicted residuals are

truncated to the most efficient 1% of the sample, against which all efficiency

estimates are allowed to vary over time. The difference in this approach lies in the per

se time variability of efficiency scores, since in REM and DFA, the mean of residuals

is used to define the distance from the frontier. We opt to estimate TFA also by

228

applying the cost function over the whole period in order to come up with average

efficiency scores similarly to the aforementioned methodologies.

After ending up with the parameter estimates of the translog cost function, we apply

the following expression:

EC exp[(lni lnimin )] (7.5) where lni lniti

T

.Or more formally,

Ec exp f y,w exp ln umin

exp f y,w exp ln u umin

u (7.6)

As for the profit efficiency, we estimate respectively the following expression:

E exp f y,w exp ln u 1

exp f y,w exp ln umax 1 (7.7)

7.4. Sample - Evidence

The unbalanced panel dataset comprises 11374 observations of at most 2228 banks

for the period 2003-2008 as retrieved from the database Bankscope Van Dijk. The

dataset covers the whole group of 27 countries in the European Union, viz. Austria,

Belgium, Bulgaria, Cyprus, Czech Republic, Denmark, Estonia, Finland, France,

Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta,

Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden and

United Kingdom. The year 2002 was excluded in the beginning since a huge amount

of missing values could undermine the panel setting of analysis. As table 11

highlights, the break down of the sample by year and country, the developed

economies dominate the sample with a considerable weight on Germany, Italy and

Austria. It is quite common in the literature to employ few banks for the Southeastern

Union due to the extreme data unavailability in Bankscope database.

229

Table 11: Number of banks

2003 2004 2005 2006 2007 2008 Total

Austria 113 132 136 151 156 132 820

Belgium 14 17 25 25 23 21 125

Bulgaria 5 66 8 8 9 10 106

Cyprus 8 9 8 11 12 10 58

Czech R. 8 13 13 15 14 11 74

Denmark 22 31 38 39 41 32 203

Estonia 4 5 5 6 4 3 27

Finland - 3 5 4 5 6 23

France 61 89 129 141 143 138 701

Germany 770 790 941 966 965 937 5369

Greece 2 13 15 16 14 13 73

Hungary 10 11 12 12 11 12 68

Ireland 3 6 15 13 16 12 65

Italy 4 11 394 408 409 414 1640

Latvia 10 12 12 13 13 12 72

Lithuania 8 8 9 9 9 7 50

Luxembourg 35 38 42 44 52 48 259

Malta 3 5 5 5 4 4 26

Netherlands 6 17 26 26 27 22 124

Poland 8 14 14 14 15 19 84

Portugal 2 7 19 21 19 16 84

Romania 8 10 12 11 12 11 64

Slovakia 6 6 7 7 7 7 40

Slovenia 6 7 10 9 9 9 50

Spain 2 34 88 91 81 89 385

Sweden 40 42 48 52 54 46 282

UK 38 57 98 101 104 104 502

EU-27 1196 1453 2134 2218 2228 2145 11374

Source: Bankscope database.

The following tables delineate the cost and profit efficiency scores as extrapolated

from the REM, DFA and TFA methodologies coupled with their truncated versions at

1%, 5% and 10% distribution level along with the specifications of intermediation and

value-added approach. Although the current financial crisis stems roughly from 2009,

the period under investigation commences from 2003 until 2008.

The first column (table 12) shows very low average efficiency scores following REM

methodology probably due to different parameter slopes prior to 2008, but also to the

extreme values of efficiency scores that may depict other random effects than the per

se operational performance that have not enough time to cancel each other out in the

course of time. Thus, the truncation point at 1%, 5% and 10% eliminates such effects

230

and efficiency levels rise considerably from 6% to almost 86%. DFA results in the

next four columns follow the same pattern although they are slightly bigger as long as

we come up with convergence at 5% and 10% truncation point.

Moreover, we apply TFA in the last two columns displaying the average cost

efficiency scores by country that, first, stem from average bank performance, and

secondly from time-varying efficient scores. At a closer look, TFA results are very

similar once we estimate them in averages by bank or allowing them to vary in the

time course. Furthermore, their levels also lie in between the scores of REM and DFA

when truncated at 1% and 5% distribution point. Corroborating the results of Maudos

et al. (2002), country rankings seem invariant across different methodologies as the

best performers (Denmark, Spain, Sweden) and poor performers (Hungary, Romania,

Bulgaria) remain remarkably stable.

TFA efficiency scores are estimated running regressions for the whole period as well

as by year to better reflect bank technology and economic conditions. Indeed, we

estimate relative efficiency scores against the most efficient frontier by taking the

average of residuals ending up with bank-specific constants of inefficiency over the

period (TFA column). The last column highlights efficiency scores after computing

the distance of the residuals from the 1% most efficient banks indentified per year and

truncated at the 1% and 99% points of the sample. It is evident that the level of

efficiency for each country differs considerably but, nonetheless, we compare all three

methodologies by picking up scores that come out of 5% truncation point, to which

they seem to converge in levels.

The coefficients of Spearman rank-order correlation (reported in the chapter

appendix) show how that REM and DFA demonstrate average rank correlation .99,

which is quite expected as the scores come out of the standard error of similarly

specified cost models. However, they both document .53 and .63 correlation with the

time-invariant TFA and time-varying case of TFA-Y, which seems to reflect

consistency in parametric estimation as Bauer et al. (1998). Hence, the resulting

differences of between DFA and TFA scores seem to vary from 9% to 22%, whereas

estimates of the same methodology differ substantially up to 85% in the case of

different truncations of DFA and between 25% and 83% for TFA. In addition, the

231

performance of the EU-15 group outpaces that of developing region by at least 6%

according to all methodologies.

Table 12: Cost efficiency (Intermediation approach)

REM REM-99 REM-95 REM-90 DFA DFA-99 DFA-95 DFA-90 TFA TFA-Y

Austria 0.0548 0.4698 0.7553 0.8466 0.1334 0.4863 0.7545 0.8454 0.6184 0.6010

Belgium 0.0608 0.4345 0.6672 0.7714 0.1488 0.4440 0.6731 0.7743 0.5573 0.5595

Denmark 0.0625 0.5145 0.8059 0.8902 0.1497 0.5242 0.8016 0.8874 0.6192 0.5998

Finland 0.0670 0.5385 0.7801 0.8549 0.1572 0.5380 0.7825 0.8556 0.6180 0.5567

France 0.0538 0.4495 0.7165 0.8101 0.1287 0.4641 0.7152 0.8074 0.5191 0.5143

Germany 0.0512 0.4488 0.7195 0.8076 0.1239 0.4620 0.7146 0.8014 0.5750 0.5535

Greece 0.0510 0.4489 0.7119 0.7920 0.1232 0.4613 0.7051 0.7854 0.5519 0.5208

Ireland 0.0718 0.5403 0.7650 0.8340 0.1636 0.5416 0.7481 0.8211 0.5907 0.5677

Italy 0.0539 0.4714 0.7542 0.8440 0.1310 0.4874 0.7519 0.8409 0.5890 0.5736

Luxembourg 0.0560 0.4147 0.6543 0.7508 0.1343 0.4332 0.6654 0.7594 0.4801 0.4734

Netherlands 0.0543 0.4559 0.7188 0.8024 0.1306 0.4662 0.7138 0.7965 0.5257 0.5097

Portugal 0.0423 0.3501 0.5911 0.6999 0.1028 0.3618 0.5887 0.6946 0.4324 0.4242

Spain 0.0563 0.4863 0.7723 0.8589 0.1377 0.5060 0.7751 0.8610 0.6381 0.6351

Sweden 0.0717 0.5589 0.8552 0.9269 0.1718 0.5726 0.8528 0.9252 0.6398 0.6239

UK 0.0629 0.3835 0.6224 0.7263 0.1378 0.3963 0.6266 0.7278 0.4746 0.4778

EU-15 0.0580 0.4644 0.7260 0.8144 0.1383 0.4763 0.7246 0.8122 0.5620 0.5461

Bulgaria 0.0394 0.3470 0.5682 0.6714 0.0975 0.3651 0.5776 0.6775 0.3999 0.4102

Cyprus 0.0442 0.3900 0.6340 0.7405 0.1119 0.4209 0.6639 0.7570 0.4642 0.4507

Czech R. 0.0546 0.4474 0.7379 0.8371 0.1370 0.4765 0.7401 0.8354 0.5566 0.5598

Estonia 0.0505 0.4450 0.7131 0.8196 0.1215 0.4550 0.7102 0.8224 0.4370 0.5060

Hungary 0.0358 0.2775 0.5165 0.6467 0.0865 0.2863 0.5180 0.6431 0.3569 0.3411

Latvia 0.0562 0.4949 0.7663 0.8337 0.1382 0.5176 0.7706 0.8373 0.5546 0.5460

Lithuania 0.0501 0.4410 0.7068 0.7845 0.1211 0.4533 0.7006 0.7779 0.5261 0.5119

Malta 0.0646 0.4849 0.6951 0.7856 0.1488 0.4866 0.6931 0.7846 0.5985 0.5567

Poland 0.0487 0.3970 0.6514 0.7480 0.1199 0.4159 0.6591 0.7496 0.4754 0.4960

Romania 0.0358 0.3182 0.5278 0.6494 0.0852 0.3214 0.5193 0.6350 0.3263 0.3348

Slovakia 0.0477 0.4200 0.6772 0.7575 0.1150 0.4307 0.6744 0.7551 0.5005 0.4862

Slovenia 0.0545 0.4802 0.7302 0.8054 0.1287 0.4817 0.7309 0.8060 0.5835 0.5201

EU-12 0.0485 0.4119 0.6604 0.7566 0.1176 0.4259 0.6631 0.7568 0.4816 0.4766

The table reports cost efficiency scores according to the intermediation approach on averages per banking industry across

different specifications. In particular, REM and DFA methodologies comprise versions of truncated distributions at 1%,

5% and 10% level of each tail. TFA produces average efficiency scores at the bank level over a common frontier whereas

TFA-Y replicates the procedure for every year in order to highlight the evolution of banking performance. EU-15 denotes

the average cost efficiency deflated by the number of banks operating within each country of the relatively more

developed European region, and EU-12 the developing region along with the inclusion of Bulgaria and Romania.

The results of table 13 come out of the VA approach giving estimates a bit lower than

those of table 2 persistently at about 10% across all specifications of REM and DFA;

the TFA results, however, may be up to 20% lower without altering the relative

232

rankings of banking markets. We also see the model fit relatively lower than that of

IA approach opposing the results of Glass et al. (1998). When it comes to compare

efficiency scores under VA approach, there is a convergent pace between DFA and

TFA estimates since the most efficient markets of Sweden, Czech Republic and

Latvia, as well as the least ones of Hungary, Romania and Portugal coincide in

relative rankings.

However, REM and DFA methodology fall short of depicting reliable results amid

conditions of economic growth and downturn; on those grounds, we opt to utilize the

truncated specifications at 5% of each side of the distribution for cross-country

comparisons. We also see the unobserved country-specific effects to impact REM and

DFA divergence from their own truncated scores at 88% the most. Thus, the

Spearman coefficient for the truncated versions of DFA and REM scores (at 5% of

the distribution) signifies identical rankings at .99 level, while that between them and

TFA and TFA-Y, correlation is still considerable (as in table 12) at levels .93 and .70,

respectively.

A notable difference between IA and VA approaches is that in the aggregate, the

comparison of the developed and developing world is not identical. In particular, the

value-added approach identifies rather a homogeneous sample with respect to cost

efficiency since EU-15 barely stands out by 1%. The result is reminiscent of the

stability in efficiency scores of Tortosa-Ausina (2002) albeit she argues – along with

Sufian (2009) - about higher levels under VA approach, which comes in

contradistinction with our results.

The intuition here is that in VA model we include deposits as bank output instead of

other earning assets, which highlights marked differences in the structure of banks‟

balance sheets. The fact that the EU-15 group performs better with the use of „other

earning assets‟, implies that extensive trading through derivatives and other

investment securities (property investments, government bills, insurance assets) has

yielded significant asset value on books given the existent cost structure. That,

however, is not related to potential losses on bank activities as the latter appear in

yearly income statements.

233

Table 13: Cost efficiency (Value-added approach)


Austria 0.0096 0.3686 0.6538 0.7581 0.0136 0.3890 0.6578 0.7589 0.3965 0.4267

Belgium 0.0109 0.3304 0.5840 0.7003 0.0158 0.3508 0.5923 0.7054 0.4211 0.4734

Denmark 0.0109 0.3874 0.6991 0.8070 0.0156 0.4088 0.7026 0.8080 0.4330 0.4648

Finland 0.0121 0.4403 0.7046 0.8186 0.0165 0.4445 0.7008 0.8129 0.4778 0.4731

France 0.0099 0.3764 0.6514 0.7478 0.0139 0.3969 0.6565 0.7508 0.4030 0.4315

Germany 0.0090 0.3576 0.6253 0.7197 0.0128 0.3765 0.6284 0.7198 0.3677 0.3928

Greece 0.0103 0.4161 0.7136 0.8005 0.0147 0.4380 0.7152 0.7989 0.4280 0.4480

Ireland 0.0142 0.4600 0.7090 0.7951 0.0185 0.4666 0.7029 0.7895 0.4840 0.4947

Italy 0.0098 0.3887 0.6768 0.7757 0.0140 0.4095 0.6806 0.7764 0.3920 0.4248

Luxembourg 0.0113 0.3615 0.6082 0.7161 0.0156 0.3774 0.6135 0.7193 0.3888 0.4185

Netherlands 0.0106 0.3836 0.6583 0.7574 0.0148 0.4009 0.6595 0.7545 0.4473 0.4591

Portugal 0.0079 0.2951 0.5395 0.6549 0.0113 0.3118 0.5437 0.6585 0.3143 0.3427

Spain 0.0119 0.4640 0.7971 0.8858 0.0168 0.4884 0.8016 0.8868 0.4825 0.5135

Sweden 0.0135 0.4799 0.8396 0.8997 0.0192 0.5044 0.8425 0.9010 0.5368 0.5766

UK 0.0241 0.2922 0.5498 0.6643 0.0255 0.3077 0.5530 0.6670 0.3867 0.4076

EU-15 0.0117 0.3868 0.6674 0.7668 0.0159 0.4047 0.6701 0.7672 0.4240 0.4499

Bulgaria 0.0093 0.3505 0.6263 0.7265 0.0131 0.3673 0.6260 0.7227 0.3898 0.4242

Cyprus 0.0090 0.3367 0.5794 0.6958 0.0128 0.3514 0.5831 0.6986 0.3560 0.3621

Czech R. 0.0113 0.4319 0.7246 0.8245 0.0161 0.4532 0.7281 0.8267 0.4740 0.5265

Estonia 0.0103 0.4153 0.7247 0.8279 0.0148 0.4393 0.7322 0.8340 0.3899 0.4681

Hungary 0.0070 0.2214 0.4726 0.6082 0.0098 0.2332 0.4740 0.6071 0.3157 0.3093

Latvia 0.0126 0.4749 0.7701 0.8483 0.0180 0.4948 0.7704 0.8459 0.4775 0.5177

Lithuania 0.0108 0.4327 0.7525 0.8369 0.0153 0.4545 0.7544 0.8362 0.4341 0.4620

Malta 0.0350 0.5116 0.7415 0.8257 0.0434 0.5192 0.7326 0.8132 0.5550 0.5058

Poland 0.0088 0.3235 0.5856 0.6909 0.0129 0.3512 0.6050 0.7046 0.3704 0.4106

Romania 0.0073 0.2808 0.5059 0.6179 0.0106 0.3010 0.5177 0.6214 0.3006 0.3505

Slovakia 0.0099 0.3968 0.6901 0.7809 0.0140 0.4162 0.6909 0.7785 0.4388 0.4622

Slovenia 0.0106 0.4286 0.7272 0.8143 0.0147 0.4388 0.7284 0.8125 0.4426 0.4392

EU-12 0.0118 0.3837 0.6584 0.7582 0.0163 0.4017 0.6619 0.7584 0.4120 0.4365

The table reports cost efficiency scores according to the value-added approach on averages per banking

industry across different specifications. In particular, REM and DFA methodologies comprise versions of truncated

distributions at 1%, 5% and 10% level of each tail. TFA produces average efficiency scores at the bank level over a

common frontier whereas TFA-Y replicates the procedure for every year in order to highlight the evolution of banking

performance. EU-15 denotes the average cost efficiency deflated by the number of banks operating within each country

of the relatively more developed European region, and EU-12 the developing region along with the inclusion of Bulgaria

and Romania.

In table 14, we see profit efficiency scores following the intermediation approach per

banking industry. The picture in this case differentiates from previous results as profit

performance accounts for efficiency gains not only from the cost side but also from

the output mix. Thus, negative values in profit efficiency scores are justified by

fluctuations in the profitability of differently specialized financial institutions rather

than the bearing of extreme outliers. It is clear that profit efficiency being much lower

than cost efficiency is evident and quite robust across empirical studies like Bonin et

234

al. (2005), Maudos and Pastor (2001; 2003), Kasman and Yildirim (2006), Yildirim

and Philippatos (2007b), among others. The developed group of the European Union

appears to be more vulnerable to profit performance since the developing markets

have more ground to cover up and, thus, more opportunities to exploit. Similar cases

of higher profit efficiency even above the level of cost efficiency are evident in the

developing markets of Turkey (Isik and Hasan, 2002), Ukraine (Mertens and Urga,

2001) and Oman (Srairi, 2010)24

. In fact, we come up with a considerable difference

up to 11.27% between the two European groups, especially in the time-varying case

of TFA.

When comparing REM and DFA results, we expect the same country rankings since

their only key difference lies in the latter allowing for variation of model parameters

over time vis-à-vis a common frontier. Hence, we identify Ireland, Malta and Cyprus

hitting the greatest negative scores (up to -23%) while Belgium, Hungary and Czech

Republic accomplish more than 30%. In the case of TFA, we verify that country

rankings remain stable as far as the least efficient markets are concerned, whereas the

best performers appear to be Slovakia, Bulgaria and Slovenia. This divergence lies in

the fact that TFA produces the distance from the frontier, which is constructed for

every single year while the other two methodologies construct the frontier and the

distance from it in average terms either the latter stems from regression over the

whole period or per year to allow for parameter variation. Moreover, negative profit

efficiency constitutes an issue hard to interpret from a rational perspective that all

banks should be profit-maximizers. In fact, it seems as if banks tend to completely

forego potential profits (Berger and Mester, 2000) possibly due to some random

effects or strategic failures.

24

What stands out in the developed world as more profit efficient is the financial conglomerates in the

EU during 1994-1995, the non-German universal and specialized banks of Germany (Vennet, 2002).

235

Table 14: Profit efficiency (Intermediation approach)


Austria 0.0185 0.1305 0.2194 0.2927 0.0182 0.1300 0.2207 0.2853 0.0712 -0.0109

Belgium 0.0362 0.3776 0.4795 0.5506 0.0390 0.3889 0.4904 0.5370 0.3116 0.0531

Denmark 0.0023 0.0754 0.0523 0.0545 -0.0032 0.0670 0.0417 0.0506 -0.0427 -0.1646

Finland 0.0014 0.1312 0.1943 0.2803 0.0144 0.1885 0.2957 0.3925 0.1495 -0.0835

France 0.0114 0.0927 0.1443 0.1650 0.0114 0.0897 0.1347 0.1684 0.0472 -0.0662

Germany -0.0073 -0.0287 -0.0627 -0.0944 -0.0071 -0.0303 -0.0646 -0.0888 -0.1943 -0.1399

Greece -0.0055 -0.0273 -0.0235 0.0060 -0.0078 -0.0393 -0.0365 -0.0095 -0.1214 -0.0762

Ireland -0.0847 -0.2310 -0.1729 -0.1177 -0.0753 -0.2220 -0.1207 -0.0449 -1.2949 -1.4855

Italy 0.0026 0.0248 0.0452 0.0659 0.0023 0.0233 0.0412 0.0624 -0.1028 -0.0946

Luxembourg 0.0341 0.1889 0.3513 0.4538 0.0323 0.1691 0.3329 0.4329 0.0527 -0.5090

Netherlands 0.0091 0.1229 0.2324 0.3045 0.0108 0.1312 0.2286 0.2871 0.0119 -0.2528

Portugal 0.0175 0.2087 0.3424 0.3669 0.0247 0.2357 0.3673 0.3952 0.1594 -0.0632

Spain 0.0361 0.2043 0.3510 0.4290 0.0388 0.2181 0.3541 0.4275 0.1411 0.0243

Sweden 0.0192 0.1821 0.3624 0.4783 0.0195 0.1835 0.3607 0.4779 -0.0045 -0.0433

UK 0.0077 0.2317 0.3230 0.3933 0.0081 0.2214 0.2981 0.3631 0.0780 -0.2824

EU-15 0.0066 0.1123 0.1892 0.2419 0.0084 0.1170 0.1963 0.2491 -0.0492 -0.2130

Bulgaria 0.0460 0.2299 0.3881 0.4730 0.0625 0.3160 0.5232 0.6100 0.2214 0.1571

Cyprus -0.0097 -0.0484 -0.0246 -0.0583 -0.0159 -0.0806 -0.0925 -0.0560 -0.4146 -0.4083

Czech R. 0.0659 0.3461 0.5262 0.5822 0.0504 0.3309 0.5158 0.5858 0.3307 -0.0174

Estonia -0.0015 -0.0073 0.0817 0.1461 -0.0049 -0.0246 0.0669 0.1536 -0.1025 -0.0996

Hungary 0.0070 0.3026 0.3750 0.3762 -0.0010 0.2730 0.3352 0.3451 0.2223 -0.1310

Latvia 0.0343 0.1714 0.3578 0.4679 0.0284 0.1437 0.3004 0.3982 0.0857 -0.0682

Lithuania 0.0445 0.2225 0.4031 0.4948 0.0439 0.2218 0.4237 0.5134 0.2719 0.1004

Malta -0.0922 -0.1875 -0.1280 -0.0948 -0.1097 -0.2772 -0.1687 -0.1181 -1.4742 -1.0797

Poland 0.0580 0.2573 0.3425 0.3926 0.0528 0.2446 0.3316 0.3764 0.2291 0.0939

Romania 0.0540 0.0947 0.0411 0.0354 0.0232 0.0644 -0.0211 -0.0134 0.0195 -0.0838

Slovakia 0.0402 0.2011 0.4249 0.6054 0.0436 0.2205 0.4610 0.5944 0.2821 0.1981

Slovenia 0.0417 0.2086 0.40994 0.5669 0.0436 0.2207 0.4492 0.5897 0.2369 0.1348

EU-12 0.0240 0.1492 0.2665 0.3323 0.0181 0.1378 0.2604 0.3316 -0.0076 -0.1003

The table reports profit efficiency scores according to the intermediation approach on averages per banking industry

across different specifications. In particular, REM and DFA methodologies comprise versions of truncated distributions at

1%, 5% and 10% level of each tail. TFA produces average efficiency scores at the bank level over a common frontier

whereas TFA-Y replicates the procedure for every year in order to highlight the evolution of banking performance. EU-

15 denote the average profit efficiency deflated by the number of banks operating within each country of the relatively

more developed European region, and EU-12 the Southeastern region along with the inclusion of Bulgaria and Romania.

In table 15, we observe the same pattern with the developed world coming short of

demonstrating efficiency on input-output mix at 22% lower than rest members of

European Union. Contrary to the analysis of cost efficiency, the VA approach better

fits the profit model along the lines of Glass et al. (1998). Drawing some comparisons

with respect to the relative rankings, we verify that Malta and Ireland keep their

considerable efficiency gap followed by Estonia, which is close to the performance of

Cyprus of the previous table. As for the best performers of REM and DFA, Hungary

still stands out while Slovenia and Portugal surpass Belgium and Czech Republic by

236

at most 3%. Furthermore, the trend of TFA leans towards the same pick-ups as the

other methodologies under both value-added and intermediation approaches. Thus, we

see Ireland, Malta and Luxemburg experiencing the lowest profit efficiency, or in

more abstract terms limited „loss‟ efficiency. The Spearman‟s rank correlation

between DFA and REM is again above .99, whereas both of them give similar

efficiency ranks to TFA and TFA-Y at correlation levels above .82 and .76,

respectively.

Therefore, to the extent that all employed methodologies and output-input definitions

sufficiently measure bank efficiency, negative scores may intuitively encompass

hidden (toxic) liabilities like collateralized debt obligations, among others. On the

other hand, the trinity of Bulgaria, Slovakia and Slovenia enjoy efficiency scores

between the range 24%-30% according to TFA. However, the results appear quite

ambivalent between TFA and REM/DFA with markedly divergence up to 25%.

In table 16, we see that relative rankings between IA and VA approaches are not the

same as in the cross-country analysis. In especial, the former concludes that

cooperative, savings and securities banks are the best performers in cost efficiency

(above 73%) while the latter picks up commercial, cooperative and private

banking/asset management companies at about 65%. As for profit efficiency, IA

approach identifies finance, real estate and specialized governmental credit

institutions leading the market with 28% to 34% score following Clearing institutions

and custody, finance companies and investment/trust corporations performing above

40% under VA approach.

In quantitative terms, the intermediation and value-added approach predictions are

very close in the cost/profit efficiency albeit the degree of divergence is

asymmetrically distributed endorsing thereby the relative ranking changes. In

contrast, with TFA and TFA-Y we depict average and yearly snapshots of rankings

subject to the inherent and potentially problematic assumptions of the underlying

methodology. In particular, sample heterogeneity makes efficiency predictions be

contingent on the bank technology of the lowest (highest) average cost (profit) profile

of each asset size quartile irrespectively of whether that thick frontier25

suffices as a

25

The thick frontier is constructed by truncating the best practice of the most efficient banks at the 1%

point of the efficiency distribution.

237

benchmark. Second, there exist cases where we observe negative profits and, thus, the

estimation of profit efficiency ends up negative according to expression (7.7).

Table 15: Profit efficiency (Value-added approach)


Austria 0.0250 0.1625 0.2818 0.3551 0.0128 0.1741 0.2857 0.3673 0.1206 0.0788

Belgium 0.0218 0.3605 0.4346 0.4975 0.0081 0.3471 0.4117 0.4681 0.3238 -0.0938

Denmark -0.0235 0.0766 -0.0317 -0.0230 -0.0098 0.0861 0.0113 0.0118 -0.0076 -0.2379

Finland -0.0162 0.0625 0.0474 0.1110 -0.0025 0.1042 0.1260 0.2145 -0.0284 -0.2708

France 0.0059 0.0670 0.1044 0.1599 0.0041 0.0773 0.1278 0.1790 0.0222 -0.0897

Germany -0.0052 -0.0043 -0.0283 -0.0333 -0.0030 -0.0101 -0.0310 -0.0393 -0.1105 -0.0825

Greece 0.0027 0.0117 0.0300 0.0787 0.0033 0.0314 0.0639 0.1132 -0.0657 -0.0801

Ireland -0.1512 -0.2259 -0.1816 0.0220 -0.0736 -0.2714 -0.1588 0.0129 -1.0358 -2.1541

Italy 0.0071 0.0510 0.0867 0.1211 0.0038 0.0520 0.0805 0.1094 -0.0295 -0.0537

Luxembourg 0.0046 0.0958 0.2345 0.3341 0.0010 0.0888 0.2273 0.3188 -0.2121 -0.6921

Netherlands -0.0290 -0.0057 -0.0065 0.0933 -0.0151 -0.0187 -0.0205 0.0665 -0.0944 -0.4038

Portugal 0.0548 0.3128 0.4618 0.5206 0.0289 0.3425 0.4757 0.5351 0.2943 0.0852

Spain 0.0479 0.1960 0.3665 0.4775 0.0301 0.2163 0.3646 0.4755 0.1168 0.0662

Sweden 0.0052 0.1695 0.2133 0.2756 0.0035 0.1783 0.2260 0.2929 0.1391 0.0067

UK 0.0057 0.3247 0.4204 0.5157 0.0012 0.3177 0.4119 0.5032 0.1898 -0.1443

EU-15 -0.0030 0.1103 0.1622 0.2337 -0.0005 0.1144 0.1735 0.2419 -0.0252 -0.2711

Bulgaria 0.0417 0.2580 0.4436 0.5401 0.0226 0.2845 0.4486 0.5491 0.2774 0.2489

Cyprus -0.0381 -0.0780 -0.1403 -0.0514 -0.0256 -0.1463 -0.1343 -0.0608 -0.3682 -0.3353

Czech R. 0.0543 0.2510 0.4566 0.5839 0.0248 0.2555 0.4465 0.5714 0.2407 0.0518

Estonia -0.0181 -0.0795 -0.1741 -0.0437 -0.0104 -0.0980 -0.1723 -0.0287 -0.2932 -0.0974

Hungary 0.0465 0.4776 0.5745 0.6461 0.0248 0.4879 0.5693 0.6212 0.4516 0.1518

Latvia 0.0149 0.0658 0.1528 0.2543 0.0069 0.0648 0.1415 0.2375 -0.0064 -0.0913

Lithuania 0.0399 0.1758 0.3700 0.4446 0.0222 0.2092 0.3898 0.4719 0.1768 0.1913

Malta -0.2225 -0.4843 -0.3607 -0.1305 -0.1062 -0.5384 -0.3372 -0.1490 -1.5587 -1.2112

Poland 0.0335 0.2514 0.3617 0.4042 0.0155 0.2488 0.3391 0.3799 0.2605 0.1677

Romania 0.0127 -0.0049 -0.1358 -0.0737 -0.0012 -0.0178 -0.1262 -0.0912 -0.0630 -0.2778

Slovakia 0.0423 0.1864 0.4145 0.5705 0.0246 0.2322 0.4612 0.6491 0.2756 0.2936

Slovenia 0.0600 0.2643 0.5132 0.6363 0.0325 0.3068 0.5215 0.6606 0.2829 0.3111

EU-12 0.0056 0.1070 0.2063 0.3151 0.0025 0.1074 0.2123 0.3176 -0.0270 -0.0497

The table reports profit efficiency scores according to the value-added approach on averages per banking industry across

different specifications. In particular, REM and DFA methodologies comprise versions of truncated distributions at 1%,

5% and 10% level of each tail. TFA produces average efficiency scores at the bank level over a common frontier whereas

TFA-Y replicates the procedure for every year in order to highlight the evolution of banking performance. EU-15 denote

the average profit efficiency deflated by the number of banks operating within each country of the relatively more

developed European region, and EU-12 the Southeastern region along with the inclusion of Bulgaria and Romania.

Last, DFA seems the most reliable measure of efficiency as it allows random effects

to cancel each other out as time passes leaving an average performance score at the

bank level; on the contrary, yearly frontiers of TFA-Y may pick up a portion of

238

random effects and potentially spoil relative performance. Hence, the per se fact that

country rankings demonstrate sufficient stability, is attributed to the symmetric

distribution of different bank types across European markets.

We next analyse the operational efficiency of the banking market with respect to size

classifying it into four distinct asset classes (table 6). In so doing, we define banking

groups to lie within each class among the 25%, 50% and 75% of the average size

distribution, namely 222.78, 710.56 and 3033.95 millions of Euro, respectively. Cost

efficiency seems more evident by at most 4% in the first two asset classes across all

specifications with a diminishing tendency in banks of higher asset profile. That may

be indicative of there being something more than scale economies and bank

specialisation; in fact, X-efficiency is deemed in contemporary efficiency analysis to

sufficiently explain cost dispersion even amongst banks of similar type or asset size.

Although it might be the case of allocative efficiency, that is the ability of a bank to

allocate its inputs on observable or potential growth opportunities, researchers

generally ignore it by assuming perfect allocation26

.

In particular, the last two classes turn out to be more efficient than smaller-sized

banks as far as we come up with the scores of the last rows. Intuitively, the

skyrocketing negative values in the higher-sized class are traced in some extreme

outliers that lie far off the yearly frontier. That issue becomes more discernible once

we compare the scores of the time-invariant TFA; that is, a more balanced trend

across asset classes implies no clear-cut winners. However, if we analyse the pattern

within each methodology, we identify a negative correlation between cost and profit

efficiency, implying that especially banks other than of commercial, cooperative or

savings type tend to make up for high average costs by utilising their high market

share (for monopolistically pricing) and different output composition.

26

Greene (1980), Schmidt (1984), Kumbhakar (1997), Schmidt and Lovell (1979), Maietta, (2002),

Kumbhakar and Tsionas (2005) and Brissimis et al. (2010) (among others) addressed the issue

theoretically and quantitatively in the literature.

239

Table 16: Cost/profit efficiency by productive specialisation and asset class

REM DFA TFA TFA-Y REM DFA TFA TFA-Y

Cost efficiency Profit efficiency

Bank Holding & Holding

Companies

0.6666 0.6650 0.4838 0.4884 0.1920 0.2051 0.0426 0.0136

0.5690 0.5767 0.3629 0.4116 0.1841 0.1739 0.0737 -0.0697

Clearing Institutions & Custody 0.6175 0.6249 0.4430 0.4648 0.3338 0.2910 -0.1709 -0.8980

0.5569 0.5572 0.4594 0.4764 0.4784 0.4435 0.0514 -1.2369

Commercial 0.7038 0.7043 0.5243 0.5136 0.1808 0.1787 0.0277 -0.1293

0.6472 0.6511 0.3919 0.4215 0.1588 0.1647 -0.0182 -0.1499

Cooperative 0.7520 0.7468 0.6135 0.5836 -0.0222 -0.0172 -0.1359 -0.0838

0.6750 0.6783 0.3955 0.4209 -0.0054 -0.0061 -0.0885 -0.0538

Finance Companies 0.6175 0.6179 0.4056 0.4305 0.4078 0.3945 0.0933 -0.3399

0.5762 0.5805 0.4343 0.4736 0.4159 0.4285 0.2783 -0.2851

Investment & Trust Corp. 0.5903 0.6303 0.3673 0.3803 0.2091 0.1711 -0.7210 -0.7221

0.4616 0.4866 0.3622 0.4028 0.7416 0.7255 0.5737 -0.4238

Investment 0.6707 0.6767 0.4354 0.4506 0.2622 0.2566 -0.0711 -0.2386

0.6026 0.6047 0.4227 0.4401 0.2222 0.2437 0.0019 -0.3103

Private Banking & Asset Mgt 0.7088 0.7116 0.4746 0.4725 0.1827 0.1631 0.0995 -0.0206

0.6657 0.6663 0.4286 0.4490 -0.0103 -0.0180 -0.0169 -0.1437

Real Estate & Mortgage 0.6353 0.6425 0.4823 0.4760 0.4021 0.3495 -0.0140 -0.3712

0.5655 0.5741 0.3555 0.3837 0.4547 0.4281 0.1755 -0.1099

Savings 0.7382 0.7349 0.6069 0.5792 0.0183 0.0140 -0.1313 -0.0860

0.6598 0.6628 0.3882 0.4079 0.0818 0.0800 -0.0432 0.0342

Securities 0.7380 0.7324 0.4581 0.4364 0.1462 0.1215 -0.3376 -1.1362

0.5858 0.5807 0.4102 0.4309 0.3845 0.3825 0.0254 -1.2499

Specialized Governmental

Credit

0.6704 0.6735 0.5072 0.5114 0.3548 0.2855 0.0714 -0.3205

0.4863 0.4925 0.2634 0.2936 0.2388 0.2526 0.0616 -0.5860

Asset class 1 (0-222.78)) 0.7286 0.7243 0.5547 0.5443 0.0798 0.0777 -0.0745 -0.0791

0.6584 0.6624 0.3955 0.4266 0.0596 0.0590 -0.0309 -0.1227

Asset class 2 (222.78-710.56) 0.7412 0.7376 0.5867 0.5652 0.0391 0.0422 -0.0934 -0.0803

0.6565 0.6590 0.3918 0.4149 0.0446 0.0491 -0.0338 -0.0219

Asset class 3 (710.56-3033.95) 0.7087 0.7083 0.5618 0.5475 0.1228 0.1151 -0.0624 -0.1005

0.6426 0.6466 0.3991 0.4190 0.1855 0.1825 0.0362 0.0160

Asset class 4 (3033.95-) 0.7051 0.7054 0.5586 0.5464 0.1269 0.1177 -0.0793 -0.2644

0.6313 0.6353 0.3843 0.4124 0.1427 0.1432 -0.0641 -0.2509

The table reports averaged cost and profit efficiency scores for every bank type and asset size class. For space considerations we include only those truncated versions of

REM and DFA (at 5% of each distribution tail) that they are closer with each other in levels; TFA and TFA-Y produce average and time-varying scores, respectively.

The IA and VA approach produce scores that appear in the first and second row for every combination of bank type and frontier methodology.

240

7.5. Scale economies

As a next step, the analysis moves on the examination of whether and to what extent,

banks are operating in efficient scale. Previous empirical studies of Altunbas et al.

(1996) and Cavallo and Rossi (2001) found limited, though many unexploited, scale

economies in banks of small size and no significance in cost complementarities.

Along the lines of Mester (1997) and Altunbas et al. (2001), we estimate scale

economies for each banking market per year by using the model parameters of the

employed methodologies and output-input definition approaches. In so doing, by

taking the partial derivative of total cost with respect to bank output we come up with

the cost elasticity that is evaluated at the mean bank output, input and equity capital

levels for every single asset class.

SEi,t 1 lnTC

lnQi i i, j lnQi

j

i

i

i

i, j lnWi

j

i

4,i lnEi

(7.8)

SEi,t 2 lnTC

lnQi i i, j lnQi

j

i

i

i

i, j lnWi

j

i

(7.9)

We have increasing returns of scale when SE get values less than unity SE 1 ,

constant returns of scale when SE equal unity SE 1 and decreasing returns of scale

if SE are more than unity SE 1 . Hence, we report in the following table scale

economies averaged by country from 2003 to 2008 under both intermediation and

value-added approach (table 17). At a first glance, we see that they are fully exploited,

if not over exploited so as to allege about diseconomies of scale. Nevertheless, there

exists a pattern of convergence or variation around the mean of unity; that is, in the

case of increasing returns of scale in Romania, Netherlands and France at the

beginning of the period, the tendency leans towards full employment of scale

dynamics as far as potential diseconomies strive for managerial correction. In

contrast, the majority of European banking industries starts off with decreasing scale

returns and, eventually, gets closer to constant scale economies.

241

Table 17: Scale economies by country

2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008

Intermediation approach Value-added approach

Austria 1.0899 1.0431 1.0807 1.0098 0.9681 0.9741 0.9558 0.9955 0.9951 0.9880 0.9653 0.9634

Belgium 0.9900 0.9972 0.9547 1.0583 1.0266 1.0149 1.0010 0.9864 0.9798 1.0045 1.0060 1.0334

Denmark 1.0249 1.0171 0.9967 1.0482 1.0187 1.0164 0.9605 0.9632 0.9649 0.9913 0.9897 1.0226

Finland 0.9920 0.9178 1.0744 1.0192 1.0231 0.9395 0.9579 1.0081 1.0300 1.0524

France 0.9620 0.9938 0.9615 1.0572 1.0195 1.0263 0.9946 0.9521 0.9617 0.9991 1.0118 1.0307

Germany 1.0484 1.0245 1.0489 0.9988 0.9782 0.9874 0.9734 0.9838 0.9871 0.9894 0.9714 0.9767

Greece 1.0866 1.0352 1.0262 1.0450 0.9914 0.9916 0.9598 0.9662 0.9803 0.9956 1.0023 0.9963

Ireland 1.2272 1.0515 1.0501 1.0785 1.0346 0.9900 0.9278 1.0236 1.0261 1.0054 0.9894 1.0387

Italy 1.0751 0.9873 1.0464 1.0480 1.0027 0.9763 0.9171 0.9457 0.9776 0.9862 0.9733 0.9684

Luxembourg 1.0088 0.9833 0.9893 1.0203 0.9998 1.0245 1.0115 0.9981 0.9902 1.0056 0.9864 1.0352

Netherlands 0.9709 1.0063 0.9701 1.0262 1.0096 1.0263 0.9671 0.9619 0.9751 1.0034 1.0157 1.0499

Portugal 1.0836 1.0300 1.0189 1.0490 1.0056 0.9945 0.9531 0.9634 0.9776 0.9962 0.9984 1.0015

Spain 1.0207 1.0875 1.1201 1.0391 0.9621 0.9340 0.9673 0.9949 1.0035 0.9855 0.9693 0.9308

Sweden 1.0581 1.0406 1.0107 1.0722 1.0321 1.0078 0.9544 0.9550 0.9578 0.9865 0.9854 0.9927

UK 0.9785 0.9924 0.9690 1.0536 1.0203 1.0169 1.0221 0.9685 0.9729 1.0002 1.0013 1.0295

EU-15 1.0446 1.0188 1.0107 1.0452 1.0059 1.0003 0.9690 0.9732 0.9805 0.9963 0.9931 1.0081

Bulgaria 1.0290 1.0244 1.0540 1.0138 0.9687 0.9832 0.9539 0.9713 0.9742 0.9871 0.9795 0.9580

Cyprus 1.0271 1.0249 1.0435 1.0151 0.9939 0.9670 0.9723 0.9809 0.9947 0.9917 0.9801 0.9676

Czech R. 1.0069 1.0355 1.0297 1.0505 0.9955 0.9770 0.9806 0.9739 0.9842 0.9926 0.9899 0.9839

Estonia 1.0042 1.0130 0.9941 1.1261 1.0312 1.0100 0.9487 0.9614 0.9460 0.9877 0.9879 0.9990

Hungary 0.9966 1.0150 0.9931 1.0548 1.0110 1.0402 0.9779 0.9568 0.9609 0.9935 1.0067 1.0466

Latvia 1.0521 1.0354 1.0368 1.0490 0.9872 0.9966 0.9575 0.9918 0.9731 0.9919 0.9804 0.9832

Lithuania 1.1097 1.0547 1.0833 1.0170 0.9557 0.9891 0.9198 0.9819 0.9886 0.9872 0.9800 0.9821

Malta 1.0524 1.0533 1.0831 1.0726 0.9887 0.9807 0.9512 1.0290 1.0164 0.9978 0.9667 0.9915

Poland 1.0037 0.9988 0.9778 1.0619 1.0122 1.0335 1.0033 0.9634 0.9764 1.0003 1.0082 1.0277

Romania 0.9804 0.9801 0.9886 1.0812 0.9944 1.0107 0.9640 0.9566 0.9557 0.9882 1.0059 0.9801

Slovakia 1.0425 1.0004 1.0056 1.0436 0.9966 1.0015 0.9624 0.9811 0.9783 0.9988 1.0018 1.0056

Slovenia 1.0294 1.0459 1.0666 1.0138 0.9637 0.9618 0.9599 0.9838 0.9881 0.9894 0.9845 0.9534

EU-12 1.0278 1.0235 1.0297 1.0500 0.9916 0.9959 0.9626 0.9777 0.9780 0.9922 0.9893 0.9899

The table reports (average) time-varying economies of scale for every European country under IA and VA approaches. The average values are deflated by the number of banks

operating within each banking market.

242

If we compare the results with those of value-added approach, there is a considerable

discrepancy in the measurement of scale economies. In fact, we hardly observe a

country that is above the threshold of unity with the exception of Belgium,

Luxembourg, UK and Poland. By and large, scale economies take values between the

range of 0.015 and 0.083 implying that 100% increase in bank outputs tend to inflate

total costs by 98.5% and 91.7%, respectively. Between the developed and developing

banking markets, differences in scale economies are on average 0.6% and evolve over

time by at most 1.2%. Italy, Finland, Estonia, Austria and Spain share the best scores

throughout the whole period with an escalation traced in 2006 and 2007 before it

settles in 2008. The results alo endorse the persistent scale economies found by

Pasiouras (2008a) in Greece (only under VA approach) and those of several countries

included in the studies of Altunbas and Molyneux (1996) and Carvallo and Rossi

(2001), though not robustly over both approaches.

From a micro-level perspective, we identify under the intermediation approach the

highest average of scale economies in investment banks and securities firms although

all bank types experience decreasing scale returns at some point in time (table 18). To

the contrary, the value-added approach turns out to produce increasing scale

economies in cooperative banks and finance companies followed by savings banks.

On the other side of the spectrum, under intermediation approach decreasing scale

economies (on average terms) are evident in specialized credit institutions, investment

and trust corporations. In contrast, securities banks and bank holding and holding

companies operate with scale diseconomies as far as the value-added approach

verifies the tendency. Notwithstanding the relative rankings of banking markets are

identical when efficiency scores are clustered by country, the underlying similarity

does not hold for different productive specialisations.

That is due to the way we define bank outputs and inputs in the applied cost function.

Intermediation approach, which has been widely used in empirical studies, defines

banks operation as transforming financial resources into loans and other investments.

In so doing, this approach overlooks bank production in terms of earning assets and

deposits (Aly et al., 1990; Berger and Humphrey, 1991). The latter issue as well as

any important output, which comes out of the contribution of labour and capital is

243

accounted for by value-added approach27

. Thus, both approaches see banks from a

different perspective, according to whether and to what extent banks are specialized to

alternative assets; in particular, no allowance of deposits make IA pick up

diseconomies of scale in banks whose outputs are not well specified in the cost

function. Along the lines of VAA, scale economies are higher in those financial

institutions that engage strategically in securities and other earning assets.

As the literature draws remarks regarding the existence of scale economies on

particular asset classes, there seems to be evident in the US territory (Hughes and

Mester, 1998; Jagtiani et al., 1995; Evanoff et al., 1990) and particularly in smaller-

sized banks (Berger and Humphrey, 1994; Gilligan and Smirlock, 1984; MacAllister

and MacManus, 1993; Goldberg et al., 1991). We corroborate similar results as Fries

and Taci (2005) did for Europe exclusively though under value-added approach in our

case. Besides, economies of scale are common in the European Union (under VA

approach) across all asset classes in samples of several developed economies

(Altunbas and Molyneux, 1996; Carvallo and Rossi, 2001), in France (Dermine and

Roller, 1992), Norway (Humphrey and Vale, 2004).

It seems that we verify about „moderate scale economies‟ in Germany as Lang and

Welzel did for the period 1989-1992, and „less than 5% scale economies‟ over the

whole period and across all productive specialisations and countries in line with the

findings of Berger and Humphrey (1991) for the US market in 1984. We should also

note that the quantile of the largest asset size depletes in scale economies after 2005,

since when IA approach shows some scale economies at below 1% level.

27

Other studies dealing with value-added approach are Olivei (1992), Resti (1993) and Berg et al.

(1993), among others.

244

Table 18: Scale economies by bank type and asset class

2003 2004 2005 2006 2007 2008 2003 2004 2005 2006 2007 2008

Intermediation approach Value-added approach

Bank Holding & Holding Companies 0.9734 0.9864 0.9674 1.0376 1.0066 1.0272 1.0001 0.9647 0.9823 1.0075 1.0221 1.0563

Clearing Institutions & Custody 1.0194 0.9780 0.8953 1.0891 1.0109 1.0218 1.0283 0.9665 0.9635 1.0155 1.0036 1.0438

Commercial 0.9943 1.0030 0.9931 1.0426 1.0075 1.0123 0.9888 0.9718 0.9747 0.9973 0.9959 1.0170

Cooperative 1.0772 1.0375 1.0701 1.0103 0.9768 0.9703 0.9617 0.9893 0.9874 0.9853 0.9631 0.9559

Finance Companies 0.9865 1.0034 0.9464 1.0836 1.0385 1.0425 0.9427 0.9208 0.9315 0.9877 1.0229 1.0159

Investment & Trust Corp. 0.9965 1.0422 1.0820 1.1499 1.0239 1.0430 0.9886 0.9888 0.9864 0.9950 0.9822 1.0459

Investment 0.8641 0.9481 0.9208 1.0700 1.0446 1.0585 1.0174 0.9498 0.9540 1.0049 1.0158 1.0599

Private Banking & Asset Mgt 0.9546 0.9392 0.8928 1.0458 1.0511 1.0771 1.0037 0.9507 0.9452 1.0018 1.0148 1.0722

Real Estate & Mortgage 1.0656 1.0457 1.0620 1.0265 1.0027 0.9785 0.9898 0.9742 0.9997 0.9877 0.9744 0.9881

Savings 1.0526 1.0388 1.0638 1.0119 0.9761 0.9757 0.9710 0.9868 0.9927 0.9896 0.9751 0.9727

Securities 0.9145 0.8897 0.8402 1.0948 1.0741 1.0940 1.0324 0.9644 0.9394 1.0188 1.0297 1.1003

Specialized Governmental Credit 1.0751 1.0272 0.9865 1.0708 1.0337 0.9950 1.0064 0.9926 0.9828 0.9978 0.9916 1.0221

Asset class 1 (0-222.78) 1.0798 1.0194 1.0438 1.0251 0.9995 0.9912 0.9631 0.9824 0.9746 0.9840 0.9615 0.9661

Asset class 2 (222.78-710.56) 1.0531 1.0222 1.0475 1.0148 0.9836 0.9855 0.9676 0.9836 0.9819 0.9878 0.9709 0.9721

Asset class 3 (710.56-3033.95) 1.0342 1.0260 1.0408 1.0247 0.9858 0.9887 0.9756 0.9782 0.9836 0.9910 0.9819 0.9844

Asset class 4 (3033.95-) 0.9967 1.0218 1.0154 1.0344 0.9957 0.9926 0.9948 0.9769 0.9909 1.0001 1.0020 1.0144

The table reports time-varying economies of scale for every bank type and asset size class under IA and VA approaches.

245

7.6. Convergence

The concepts of growth convergence in terms of β (beta) and σ (sigma) methods of

Barro and Sala-i-Martin (1992) and Sala-i-Martin (1996) are very well known in the

cross-country comparisons of GDP convergence in the literature. Expressed in terms of

bank efficiency, it takes the following form approximately,

ln effs,t 1 ln effs,t1 us,t (7.10)

where effi,t and effi,t-1 are respectively the means of bank efficiency scores of countries

s 1,...,N during the period 2003-2008. β-convergence is reflected by the significant

negative coefficient of β lying in the range 0,1 . We opt to employ a lagged

dependent variable to compare the results as to whether there are any dynamic

effects28

. The former regresses growth rates on its initial levels implying that banks

with low efficiency demonstrate higher rates of growth as convergence comes along to

make up for the lag.

Following panel data specification of Canova and Marcet (1995), we further extend it

by employing country fixed effects and a lagged dependent variable,

lneffs,t

effs,t1

ln effs,t1 ln

effs,t1

effs,t2

s

s

D us,t (7.11)

where Di the country dummies to capture a portion of fixed effects in the sample and

us,t the error term. Quah (1996) pointed out the conceptual limitation of the latter as it

implies that low efficient banks are growing faster and surpass in performance the

more efficient ones ending up, therefore, with no convergent tendency at all. In

addition, it can by no means identify how the efficiency trend is dispersed across

countries.

28

We use only OLS as two-step dynamic GMM methodology needs greater time dimension, especially

in our case where the respective test (estat abond) of zero autocorrelation in first-differenced errors

highlights zero p-values for higher than two-order autocorrelation.

246

That dispersion implies the evolution of variance as a function of its lagged values and

the β-convergence itself, as our sample comprises a large enough number of banks (N)

to imply that sample variance is close to the population variance. Hence, from equation

(7.10), we derive

(7.12)

and steady-state variance as

*2 u

2

1 1 2

. (7.13)

We substitute in equation (7.12) u

2from equation (7.13) to get,

(7.14)

As we can see, β-convergence is an important factor of σ-convergence, which settles in

equilibrium as β and u

2 are getting higher and lower values, respectively. We now

turn to empirically estimate σ-convergence with the application of pooled OLS in the

following model29

:

effs,t efft effs,t1 efft1

effs,t1 efft1

effs,t1 efft1 effs,t2 efft2

s

s

D us,t

(7.15)

29

Mamatzakis et al., 2008) regress the standard deviation of efficiency scores on a time trend to see

whether its (negative) coefficient is indicative of convergence. However, such methodology requires the

efficiency scores to be estimated under the same frontier through SFA. Alternatively, Weill (2009)

employs SFA under the Iterated Seemingly Unrelated Regression (ITSUR) framework.

247

where effs.t is cost (or profit) efficiency of country s 1,...,27 at t 1,...,6 , efft the

average efficiency of the EU-27 at time t, and us,t the standard error term. Convergence

does exist if β is statistically significant and negative converging from effs,t1 to

efft either estimating the lagged dependent variable (γ parameter) or not. All efficiency

scores are expressed in their natural logarithm and produced by TFA that allows for

time varying scores at the bank level each year. All the rest methodologies estimate

average scores of banks throughout the whole period and, thus, confine the analysis of

convergence30

.

Table 19 reports whether it is a case of β-convergence for cost efficiency in the

European Union under both intermediation and value-added approach. Equation (7.11)

endorses the tendency of convergence in cost efficiency scores under both approaches

(IA, VA) between the range 55.1% and 89.6%. The ADL model identifies significance

in the positive effect of lagged dependent variable as β-coefficient maintains its

converging pattern at about 35.9% to 78.7%. Moreover, we come up with converging

dispersion of cost efficiency scores ranging from 59.1% to 97.9%, (under OLS) as

soon as the lagged regressor absorbs dynamic effects and thereby shortens the

underlying bound to [56.3%, 86.4%]. The results are in line with Mamatzakis et al.

(2008) and Weill (2009), who examined Europe during a period that covers before and

after the time of EMU.

30

Other studies (e.g. Weill, 2008) run regressions on different sub-periods in the form of (e.g.) 2003-

2005, 2006-2008, 2009-2011, so as to construct a sufficient time dimension of averaged efficiency

scores; however, our small sample period rules out any such possibility.

248

Table 19: β-convergence and σ-convergence of cost efficiency

β-convergence σ-convergence

IA VA IA VA

α -0.299*** -0.778*** 0.034*** -0.004***

(0.003) (0.006) (0.002) (0.001)

β -0.551*** -0.896*** -0.591*** -0.979***

(0.007) (0.007) (0.007) (0.007)

Country dummies YES YES YES YES

R-squared 0.5336 0.7202 0.6142 0.7548

Obs 12667 12667 12667 12667

Autoregressive Distributed Lag model

α -0.415*** -0.312*** 0.060*** -0.013***

(0.005) (0.016) (0.002) (0.001)

β -0.787*** -0.359*** -0.864*** -0.563***

(0.010) (0.019) (0.013) (0.019)

γ - AR(1) 0.292*** 0.052*** 0.284*** 0.156***

(0.008) (0.004) (0.008) (0.004)


R-squared 0.6313 0.3636 0.6443 0.5007

Obs 10132 10132 10132 10132

We run pooled OLS with robust standard errors and AR(1) specification along with sampling

weights at the country level as some banking markets are asymmetrically represented; IA:

intermediation approach, VA: value-added approach; Standard deviation are reported in

parentheses denoting *, **, *** for significance at 10%, 5% and 1% level, respectively.

In table 20, we document the evidence from pooled OLS with robust standard errors

and an autoregressive distributed lag specification of order one to pick up, if any,

dynamic effects. Sampling weights are also employed at the country level to account

for the dominance of some markets (e.g. Germany) in the sample; otherwise we come

up with significance of unit roots and false sign changes (particularly in σ-

convergence).

In fact, IA approach shows a stable „catch-up effect‟ at 88.9%, which turns to 110.1%

under VA approach and even higher values in the ADL model. Similarly, efficiency

dispersion converge as well at 85.5% and 98.1% according to IA and VA approaches

notwithstanding the inclusion of lagged dependent variable decomposes σ-convergence

into positive AR(1) effect and negative β-parameter at unstable (>1) and economically

meaningless pace. That is due to the theoretical definition of β taking values between

the range [0,1] as it is conceptually close to income per capita convergence. Above-

unity values are possible if there are negative values31

and imply that previous values

31

Negative values of profit efficiency are addressed in the literature as a possibility, but once we intend

to investigate convergence dynamics we should express efficiency scores in natural logarithms. Hence,

if we just make the relative transformations the regression output will be misleading on the grounds that

249

of profit efficiency are negatively related to current levels, which potentially drive

panel series to oscillation between negative to positive values period by period.

However, this unstable tendency of ADL specification is mitigated by the positive

impetus of the lagged dependent variable.

Table 20: β-convergence and σ-convergence of profit efficiency

β-convergence σ-convergence

IA VA IA VA

α 1.013*** 1.709*** 0.956*** 1.477***

(0.050) (0.054) (0.038) (0.028)

β -0.889*** -1.101*** -0.855*** -0.981***

(0.044) (0.035) (0.035) (0.019)


R-squared 0.5185 0.6296 0.5210 0.5895

Obs 12667 12667 12667 12667

Autoregressive distributed Lag model

α 1.450*** 1.715*** 1.381*** 1.858***

(0.076) (0.042) (0.057) (0.027)

β -1.274*** -1.105*** -1.239*** -1.235***

(0.067) (0.027) (0.052) (0.018)

γ - AR(1) 0.329*** 0.166*** 0.332*** 0.333***

(0.052) (0.014) (0.047) (0.010)


R-squared 0.4658 0.7874 0.4933 0.7265

Obs 10132 10132 10132 10132

We run pooled OLS with robust standard errors and AR(1) specification along with sampling

weights at the country level as some banking markets are asymmetrically represented; IA:

intermediation approach, VA: value-added approach; Standard deviation are reported in

parentheses denoting *, **, *** for significance at 10%, 5% and 1% level, respectively.

The converging dispersion of profit efficiency does not verify the evidence of

Mamatzakis et al. (2008), according to which only cost efficiency demonstrates β and

σ-convergence. However, the next section gives some credit to the lagging-behind

pattern of EU-15 over the years 2004-2005 (particularly under IA approach) and 2007-

2008, exactly when efficiency dispersion settles down significantly.

negative values become missing as inconvertible. Therefore, we opt to rescale them according to the

standard practice of rescaling total profits as a dependent variable in a profit function:

ln eff effmin 1 .

250

7.7. Other stylized facts

Other measures on the symmetry of efficiency distribution that can provide further

information on the way convergence evolves since 2003 are skewness and kurtosis. We

also decompose average cost and profit efficiencies into the respective scores of its two

constituent sub-regions, namely the developed EU-15 and developing EU-12. In

addition, the notion of equality in the efficiency convergence by means of Theil and

Gini index facilitates whether convergence comes along with (in)equality of banking

markets in cost and profit efficiency.

Graph 2: Kernel Density of cost/profit efficiency under IA/VA approaches

Graph 2 summarizes in a snapshot how the kernel density of efficiency scores shifted

during the period. The distribution of cost had the level of its peak lowered

(diminishing kurtosis) by 23.3% and 32.6% according to IA and VA approaches,

respectively. Furthermore, the asymmetry of the probability function changed in shape

251

from right leaning Skewness2003 0.451 to left-leaning Skewness2008 1.172

tails; VA approach highlights the same tendency albeit towards normality (non-

negative skewness). On the contrary, profit efficiency gets a sharper peak across both

IA and VA approaches and the left tail seems quite long and thin enough to cover up a

small portion of financial institutions that forego 100% of potential profits.

Gini and Theil coefficients32

are alternative measures of statistical dispersion within

the range [0,1], with higher values effectively highlighting inequality amongst the

efficiency scores of their respective frequency distribution (see table 21). According to

VA approach, the evolution of equality seems to deteriorate up until 2006 whence it

gains ground as far as the onset of the financial crisis. In contrast, IA approach

identifies this „arched‟ path of equality, which diminishes in the advent of Euro, then

settles down during the period 2005-2006 and exacerbates back again thereafter. Thus,

although it turns out that ups and downs take place throughout the period, once we

compare the first and last year of the underlying period (even the year before),

inequality is evident over the whole European region.

Moreover, there is one question not addressed so far in the analysis: which banking

markets make the beta and sigma convergence be verified empirically on the realistic

presumption that there must exist centrifugal forces? And which are those? By and

large, the dispersion of cost and profit efficiency under both approaches attests

convergence during 2004-2005 and 2007-2008; that tendency is also indicative in some

banking industries even in 2003 and 2006, notwithstanding it is surpassed in the

aggregate.

Utilising the common pick-ups of both approaches (IA, VA), Austria, Denmark,

Portugal and Netherlands appear as out of the convergent pace experiencing a

reduction between 8.16% and 25.3%; in the other extreme, Italy, Germany, Romania

and Slovakia enhance their average cost efficiency by more than 16.9%. Belgium,

Germany, Sweden and UK stand out with considerable efficiency dispersion increased

at most by 66.5% throughout the whole period, during which we see plummeting in the

standard deviation of cost efficiency scores within the bound [-66.5%, -8.3%].

32

They are both applied in cost efficiency scores only, as profit efficiency lies on average below zero

thereby driving Gini and Theil indexes off their theoretical bound.

252

Belgium, Cyprus and Slovakia outpace the poor performance of Italy, Sweden, Malta

and Estonia in the β-convergence of profit efficiency. Sigma convergence is also

verified by the best performance of Greece, Italy and Slovakia, which counterbalances

that of Denmark, Spain, Cyprus and Malta. The reason why some of the

aforementioned countries do not perform beta and sigma convergence concurrently lies

in the equations (7.12)-(7.14) as Young et al (2008) argue about. In fact, the beta

coefficient depicts how fast a banking market regains its footing after a shock. When

such random events occur, their persistence is contingent upon the speed of

convergence to the steady-state variance. Hence, under β-convergence the initial levels

of efficiency dispersion govern the sign of adjustment towards a balanced growth path.

In other words, if the initial level of dispersion is high (low) relatively to the variance

of a random shock, then the convergence pattern takes place from above (below). That

is also consistent with the extent to which shock variance within the sample is widely

different from the population variance of shocks even if many N banks asymptotically

justify population inference.

There may be other reasons for the discrepancy between β and σ convergence. As our

sample comprises countries of different economic and institutional development, the

adoption of Euro identify banking sectors starting from different positions vis-à-vis

their balanced growth path. Alongside (beta) converging markets of similar efficiency

level, it might be the case that a market lagging considerably behind in the beginning is

bound to increase t

2much more than another that lies close to its balanced growth

path. Less likely, the dispersion of balanced growth paths ( *2 ) could get higher levels

in contrast with the ongoing institutional homogeneity of the European Union.

All the aforementioned reasons may effectively construe a divergent propensity

between efficiency catching-up and dispersion across certain markets or time periods.

However, in our analysis the tendency stands robustly convergent in the aggregate

although there exists some sort of centrifugal tendency.

253

Table 21: Distribution of cost/profit efficiency

2003 2004 2005 2006 2007 2008

Cost efficiency mean (EU) 0.517 0.508 0.499 0.531 0.530 0.495

(IA) mean (EU-15) 0.468 0.467 0.454 0.502 0.508 0.457

mean (EU-12) 0.451 0.431 0.442 0.502 0.521 0.470

s.d. (EU) 0.142 0.147 0.139 0.150 0.155 0.146

s.d. (EU-15) 0.126 0.141 0.122 0.135 0.144 0.127

s.d. (EU-12) 0.136 0.130 0.126 0.142 0.144 0.115

skewness 0.451 0.011 -0.402 -0.318 -0.905 -1.172

kurtosis 7.259 5.863 5.481 6.270 5.828 5.564

Gini 0.111 0.153 0.132 0.116 0.126 0.131

Theil 0.026 0.033 0.035 0.028 0.036 0.039

Cost efficiency mean (EU) 0.413 0.423 0.440 0.447 0.457 0.455

(VA) mean (EU-15) 0.410 0.444 0.453 0.462 0.447 0.448

mean (EU-12) 0.417 0.397 0.424 0.429 0.470 0.464

s.d. (EU) 0.160 0.154 0.151 0.159 0.161 0.161

s.d. (EU-15) 0.165 0.165 0.150 0.161 0.156 0.165

s.d. (EU-12) 0.154 0.141 0.153 0.155 0.168 0.156

skewness 1.666 1.775 1.809 1.171 0.937 1.020

kurtosis 10.762 9.263 8.961 7.245 7.775 7.245

Gini 0.141 0.159 0.159 0.162 0.151 0.149

Theil 0.040 0.049 0.048 0.049 0.046 0.043

Profit efficiency mean (EU) -0.095 -0.194 -0.176 -0.137 -0.124 -0.215

(IA) mean (EU-15) -0.180 -0.236 -0.181 -0.145 -0.210 -0.309

mean (EU-12) 0.004 -0.142 -0.169 -0.127 -0.017 -0.099

s.d. (EU) 0.584 0.623 0.559 0.632 0.622 0.687

s.d. (EU-15) 0.716 0.669 0.617 0.682 0.717 0.812

s.d. (EU-12) 0.431 0.565 0.486 0.571 0.504 0.532

skewness -4.653 -4.494 -4.425 -3.769 -3.928 -4.612

kurtosis 40.750 36.015 36.454 30.980 31.079 35.078

Profit efficiency mean (EU) -0.145 -0.378 -0.324 -0.206 -0.224 -0.312

(VA) mean (EU-15) 0.117 -0.087 -0.061 -0.100 -0.013 -0.087

mean (EU-12) 0.585 0.696 0.708 0.725 0.660 0.816

s.d. (EU) 0.681 0.774 0.800 0.748 0.754 0.943

s.d. (EU-15) 0.480 0.599 0.594 0.696 0.543 0.657

s.d. (EU-12) 0.560 0.651 0.649 0.645 0.624 0.732

skewness -3.866 -4.159 -3.860 -3.690 -3.859 -4.560

kurtosis 31.173 29.010 27.576 25.797 25.185 33.503

The table reports averaged values of the mean, standard deviation, skewness and kurtosis of cost/profit

distribution under IA/VA approach along with alternative measures of equality in cost efficiency scores, namely

the Gini and Theil index.

254

7.8. Conclusion

Our study aims to contribute to the literature that compares alternative parametric

methodologies of estimating profit and cost efficiency under widely employed

definitions of bank inputs/outputs, namely the intermediation and value-added

approach. Hence, we applied REM, DFA and TFA methodologies along with

differently truncated versions of them at 1%, 5% and 10% level of their distribution

tails in order to produce averaged efficiency scores at the bank level. Only the

specification of TFA estimated for each year enables us to utilize time-varying

efficiency scores of differently specialized financial institutions over the period 2003-

2008 for the enlarged EU-27 region.

We conclude that all methodologies turn out to produce different levels of efficiency

although they may be comparable if we juxtapose their truncated specifications.

Interestingly, the relative rankings of banking markets and asset classes with respect to

their averaged cost/profit efficiency are almost identical across all approaches; in

contrast, that is not the case for different productive specializations. According to the

latter, cost efficiency is prevalent in the developed European region and in

evolutionary dominating pace over that of EU-12; however, the developing sub-group

has experienced persistently higher profit efficiency reaping the benefits of their better

opportunities for operational expansion.

Next we estimate annual scale economies for every banking market and observe

considerable discrepancies between their average values. Apart from few exceptions33

in the analysis, both approaches start from different points, that is IA approach comes

from above to converge eventually to the unity level in 2008, while VA approach

identifies increasing scale economies at the beginning of the period and then settles to

full scale exploitation.

Last, we end up with significant „catching-up effect‟ and converging dispersion of cost

efficiency scores under both IA and VA approaches and dynamic specifications. Profit

efficiency demonstrates the same converging pattern at even higher levels albeit

beta/sigma convergence is clearly evident in so far ADL specification allows for

dynamic effects. Above-unity cases if assessed individually make little economic sense

33

Belgium, Luxembourg, United Kingdom and Poland.

255

as highlighting efficiency series fluctuate between negative and positive values year by

year. However, that pattern is counterbalanced by the positive bearing of AR(1)

variable.

Last, we also verify inequality amongst the cost efficiency scores within their

frequency distribution, while IA (VA) approach shows increasing (decreasing) path

until 2006 and exacerbates (ameliorates) thereafter.

256

Chapter 7 appendix


Literature on

Methodology comparison Period Countries Methodology Evidence

Bauer et al. (1998) 1977-1988 US DEA, SFA,

TFA, DFA

Consistency among parametric

approaches, and nonparametric

approaches

Inconsistency between

parametric and nonparametric

approaches.

Casu et al. (2004) 1994-2000

France,

Germany,

Italy, Spain,

UK

DEA (TFP),

TFA „Not markedly different results‟

Cummins and Zi (1998) 1988-1992

US life

insurance

industry

DEA, FDH,

DFA

Differences between DEA and

FDH, and between econometric

and mathematical programming

methods

Delis et al. (2009) 1993-2005 Greece SFA, DEA

Differences between parametric

(lower inefficiency) and non-

parametric approaches

Huang and Wang (2002) 1982-1997 Taiwan SFA, DFA,

DEA

Consistency among parametric

approaches, and nonparametric

approaches

Inconsistency between

parametric and nonparametric

approaches.

Maudos et al. (2002) 1993-1996 EU DFA, FEM,

REM Similar country rankings

Weill (2004a) 1992-1998

France,

Germany,

Italy, Spain,

Switzerland

SFA, DFA,

DEA

Correlation between SFA and

DFA

No relationship between

SFA/DFA and DEA

Literature on Cost-Profit

efficiency Period Country Methodology More efficient on

Ariff and Can (2008) 1995-2004 China DEA Cost

Bonin et al. (2005) 1996-2000 EU SFA Cost

257

developing

countries

Isik and Hasan (2002) 1988- 1996 Turkey SFA Profit

Maudos and Pastor (2003) 1985-1996 Spain DEA Cost

Maudos et al. (2002) 1993-1996 EU DFA, FEM,

REM Cost

Mertens and Urga (2001) 1998 Ukraine SFA, TFA Profit (SFA)

Cost (TFA)

Pasiouras et al. (2009) 2000-2004 74 countries SFA Cost

Vennet (2002) 1995-1996 EU SFA

Cost (+OBS)

Profit (Financial conglomerates,

non-German universal and

specialized banks

Maudos and Pastor (2001) 1984-1995 EU, Japan,

US SFA Cost

Bader et al. (2008) 1990-2005

21 countries

(Africa,

Asia, Middle

East)

DEA Cost

Berger and Mester (1999) 1991-1997 US

Berger et al (2009) 1994-2003 China SFA Cost

Das and Ghosh (2009) 1992-2004 India DEA Cost

Kasman and Yildirim (2006) 1995-2002 EU SFA Cost

Ray and Das (2010) 1997-2003 India DEA Cost

Srairi (2010) 1999-2007

Gulf

cooperation

council

countries

SFA Cost

Profit (Oman)

Yildirim and Philippatos

(2007b) 1993-2000

EU

(developing

countries)

SFA, DFA Cost

Literature on IA-VA

approaches Period Country Methodology Evidence

Glass et al. (1998) 1977-1993 Japan

Pulley and

Braunstein

(1992)

Better fit of VA approach

Mixed results between scale and

scope economies

258

Suflan (2009) 1995-1999 Malaysia DEA

Greater average technical and

scale efficiency under VA

approach

Tortosa-Ausina (2002) 1985-1997 Spain DEA Higher and more stable efficiency

scores under VA approach.

Literature on Scale

economies Period Country Methodology Evidence

Fries and Taci (2005) 1994-2001 15 East EU SFA Average-sized banks: CRS

Smaller banks: scale economies

Berger et al. (1987) 1983 US

SUR model

(Production

and

intermediation

approach)

Consistent diseconomies of scale

across branching and unit state

size classes.

Hughes and Mester (1998) 1989-1990 US IA approach

plus OBS Economies of scale

Jagtiani et al. (1995) 1988 US

Christensen et

al. (1973)

IA/VA

approach

Economies of scale under IA

approach

Noulas et al. (1990) 1986 US Production/IA

approach

Economies of scale for 1-3 billion

asset size class

Diseconomies for greater sizes

Berger and Humphrey

(1994) 1980-1992 US Review

Scale economies for smaller

banks only

Gilligan and Smirlock

(1984) 1973-1978 US

Benston

(1972); Bell

and Murphy

(1968)

Scale economies for smaller

banks only

Jagtiani and Klauthavit

(1996) 1984-1991 US

Employing

dummies for

technology

and risk-based

requirements

Diseconomies for money center

and super-regional banks

Hughes et al. (2001) 1994 US

Inclusion of

capital

structure and

risk-taking in

the production

unction

Scale economies may be elusive

since potentially obscured by risk-

taking

259

Berger and Humphrey

(1991) 1984 US

TF cost

function Less than 5% scale economies

Evanoff et al. (1990) 1972-1987 US Shadow-price

model Scale economies

Altunbas and Molyneux

(1996) 1988

France,

Germany,

Italy and

Spain

Intermediation

approach

„Scale economies over a wider

range of outputs, including the

largest banks‟

Wheelock and Wilson

(2001)

1985, 1989,

1994 US

Translog,

Fourier, kernel

regression and

polynomial

smoothing

Scale economies if banks expand

their assets to 300-500$.

Evident also in banks of $1 billion

asset size.

Zardakohi and Kolari (1994) 1988 Finland IA approach

Scale economies for large branch

offices – exhausted at 2.5 million

Finn marks.

Carvallo and Rossi (2001) 1992-1997

France,

Germany,

Italy,

Netherlands,

Spain, UK

GMM

(Arellano--

Bond)

Scale economies evident in small

banks, at any production scale and

organizational type albeit

exhausted at low levels of output.

Clark (1988) Review

Gilligan et al. (1984) 1978 US Translog Scale economies only in low

output levels

Humphrey and Vale (2004) 1987-1998 Norway

Translog,

fourier

flexible, spline

Scale economies in small banks

(translog)

Scale (dis)economies evident in

smaller and larger banks when

TA used as single output.

McAllister and McManus

(1993) 1984-1990 US

Translog,

fourier

flexible,

kernel, spline

VA approach

Scale economies for banks with

less that $500 million total assets.

Dermine and Roller (1992) 1987

France

(mutual

funds)

Translog.

quadratic

Scale economies for all asset

classes under translog (higher in

smaller ones though)

Scale economies for banks with

assets more than 100 and fewer

than 2900 FF million

Goldberg et al. (1991) 1983 US Translog Scale economies in smaller,

specialized firms and

260

(securities) diseconomies in more diversified

firms.

Lang and Welzel (1996) 1989-1992 Germany

(cooperative) Translog

Moderate scale economies in all

size classes

Pasiouras (2008a) 2000-2004 Greece DEA

Persistent scale economies in

purely domestic banks

Lower scale economies in

international Greek banks

Literature on Convergence Period Sample Methodology Result

Casu and Girardone (2010) 1997-2003 EU-15

DEA (cost

efficiency)

Dynamic

models

Lagging behind rather than

catching up

Mamatzakis et al. (2008) 1998-2003 EU-10

SFA

(cost/profit

efficiency)

Some beta and sigma

convergence in cost efficiency

(only)

Fung (2006) 1996-2003 US DEA Conditional convergence among

bank holding companies

Weill (2009) 1994-2005 10 EU

(developed)

SFA (Fourier)

Robustness

checks: FE,

DFA,

production

approach

Beta and sigma convergence in

cost efficiency

261

Spearman’s rank correlation among all efficiency scores

TFA-Y

(VA)

TFA-Y

(IA) TFA (VA) TFA (IA)

DFA-90

(VA)

DFA-95

(VA)

DFA-99

(VA) DFA (VA)

DFA-90

(IA)

DFA-95

(IA) Profit efficiency

REM (IA) 1 0.7839* 0.7774* 0.6600* 0.6861* 0.6862* 0.6865* 0.6865* 0.5630* 0.5638* TFA-Y (VA)

REM-99 (IA) 1.0000* 1 0.6459* 0.7116* 0.5458* 0.5473* 0.5475* 0.5475* 0.5647* 0.5657* TFA-Y (IA)

REM-95 (IA) 1.0000* 1.0000* 1 0.8443* 0.8343* 0.8357* 0.8357* 0.8357* 0.6936* 0.6949* TFA (VA)

REM-90 (IA) 0.9996* 0.9996* 0.9996* 1 0.7726* 0.7740* 0.7740* 0.7740* 0.8055* 0.8062* TFA (IA)

REM (VA) 0.6553* 0.6553* 0.6551* 0.6540* 1 0.9995* 0.9994* 0.9994* 0.8477* 0.8481* DFA-90 (VA)

REM-99 (VA) 0.6553* 0.6553* 0.6551* 0.6540* 1.0000* 1 0.9999* 0.9999* 0.8476* 0.8483* DFA-95 (VA)

REM-95 (VA) 0.6551* 0.6551* 0.6549* 0.6538* 0.9999* 0.9999* 1 1.0000* 0.8473* 0.8481* DFA-99 (VA)

REM-90 (VA) 0.6542* 0.6542* 0.6541* 0.6532* 0.9994* 0.9994* 0.9995* 1 0.8474* 0.8481* DFA (VA)

DFA (IA) 0.9907* 0.9907* 0.9907* 0.9903* 0.6537* 0.6537* 0.6534* 0.6523* 1 0.9996* DFA-90 (IA)

DFA-99 (IA) 0.9907* 0.9907* 0.9907* 0.9903* 0.6537* 0.6537* 0.6534* 0.6523* 1.0000* 1 DFA-95 (IA)

DFA-95 (IA) 0.9907* 0.9907* 0.9907* 0.9903* 0.6535* 0.6535* 0.6533* 0.6523* 1.0000* 1.0000* DFA-99 (IA)

DFA-90 (IA) 0.9903* 0.9903* 0.9903* 0.9906* 0.6528* 0.6529* 0.6527* 0.6518* 0.9996* 0.9996* DFA (IA)

DFA (VA) 0.6534* 0.6534* 0.6532* 0.6521* 0.9966* 0.9966* 0.9965* 0.9959* 0.6533* 0.6533* REM-90 (VA)

DFA-99 (VA) 0.6534* 0.6534* 0.6532* 0.6521* 0.9966* 0.9966* 0.9965* 0.9959* 0.6533* 0.6533* REM-95 (VA)

DFA-95 (VA) 0.6531* 0.6531* 0.6529* 0.6519* 0.9965* 0.9965* 0.9965* 0.9960* 0.6530* 0.6530* REM-99 (VA)

DFA-90 (VA) 0.6525* 0.6525* 0.6523* 0.6514* 0.9959* 0.9959* 0.9960* 0.9964* 0.6522* 0.6522* REM1 (VA)

TFA (IA) 0.5381* 0.5381* 0.5378* 0.5363* 0.5317* 0.5317* 0.5317* 0.5310* 0.5432* 0.5432* REM-90 (IA)

TFA (VA) 0.6406* 0.6406* 0.6404* 0.6394* 0.9404* 0.9404* 0.9402* 0.9394* 0.6389* 0.6389* REM-95 (IA)

TFA-Y (IA) 0.6361* 0.6361* 0.6361* 0.6352* 0.3269* 0.3269* 0.3266* 0.3256* 0.6392* 0.6392* REM-99 (IA)

TFAY (VA) 0.4896* 0.4896* 0.4894* 0.4882* 0.7046* 0.7046* 0.7043* 0.7026* 0.4939* 0.4939* REM (IA)

Cost efficiency REM (IA) REM-99

(IA)

REM-95

(IA)

REM-90

(IA) REM (VA)

REM-99

(VA)

REM-95

(VA)

REM-90

(VA) DFA (IA)

DFA-99

(IA)

262

DFA-99 (IA) DFA (IA)

REM-90

(VA)

REM-95

(VA)

REM-99

(VA) REM (VA)

REM-90

(IA)

REM-95

(IA)

REM-99

(IA) REM (IA) Profit efficiency

REM (IA) 0.5638* 0.5638* 0.6776* 0.6776* 0.6777* 0.6777* 0.5585* 0.5592* 0.5593* 0.5593* TFA-Y (VA)

REM-99 (IA) 0.5658* 0.5658* 0.5353* 0.5367* 0.5368* 0.5368* 0.5448* 0.5457* 0.5459* 0.5459* TFA-Y (IA)

REM-95 (IA) 0.6950* 0.6950* 0.8260* 0.8274* 0.8275* 0.8275* 0.6887* 0.6899* 0.6902* 0.6902* TFA (VA)

REM-90 (IA) 0.8063* 0.8063* 0.7666* 0.7679* 0.7679* 0.7679* 0.7962* 0.7969* 0.7970* 0.7970* TFA (IA)

REM (VA) 0.8483* 0.8483* 0.9906* 0.9903* 0.9902* 0.9902* 0.8424* 0.8426* 0.8427* 0.8427* DFA-90 (VA)

REM-99 (VA) 0.8485* 0.8485* 0.9903* 0.9909* 0.9908* 0.9908* 0.8424* 0.8428* 0.8430* 0.8430* DFA-95 (VA)

REM-95 (VA) 0.8483* 0.8483* 0.9903* 0.9908* 0.9909* 0.9909* 0.8421* 0.8426* 0.8429* 0.8429* DFA-99 (VA)

REM-90 (VA) 0.8483* 0.8483* 0.9903* 0.9908* 0.9909* 0.9909* 0.8421* 0.8426* 0.8429* 0.8429* DFA (VA)

DFA (IA) 0.9995* 0.9995* 0.8465* 0.8462* 0.8460* 0.8461* 0.9803* 0.9802* 0.9802* 0.9802* DFA-90 (IA)

DFA-99 (IA) 1.0000* 1.0000* 0.8469* 0.8469* 0.8468* 0.8468* 0.9802* 0.9807* 0.9807* 0.9807* DFA-95 (IA)

DFA-95 (IA) 1 1.0000* 0.8470* 0.8471* 0.8470* 0.8470* 0.9802* 0.9807* 0.9807* 0.9807* DFA-99 (IA)

DFA-90 (IA) 0.9996* 1 0.8470* 0.8471* 0.8470* 0.8470* 0.9802* 0.9807* 0.9807* 0.9807* DFA (IA)

DFA (VA) 0.6532* 0.6525* 1 0.9995* 0.9994* 0.9994* 0.8504* 0.8507* 0.8508* 0.8508* REM-90 (VA)

DFA-99 (VA) 0.6532* 0.6525* 1.0000* 1 0.9999* 0.9999* 0.8502* 0.8507* 0.8510* 0.8510* REM-95 (VA)

DFA-95 (VA) 0.6529* 0.6523* 0.9999* 0.9999* 1 1.0000* 0.8500* 0.8506* 0.8508* 0.8508* REM-99 (VA)

DFA-90 (VA) 0 0.6521* 0.6516* 0.9994* 0.9994* 0.9994* 1 0.8500* 0.8506* 0.8508* 0.8508* REM (VA)

TFA (IA) 0.5428* 0.5415* 0.5325* 0.5325* 0.5325* 0.5319* 1 0.9995* 0.9995* 0.9995* REM-90 (IA)

TFA (VA) 0.6388* 0.6382* 0.9378* 0.9378* 0.9376* 0.9366* 0.5660* 1 1.0000* 1.0000* REM-95 (IA)

TFA-Y (IA) 0.6390* 0.6385* 0.3304* 0.3304* 0.3302* 0.3294* 0.7637* 0.3385* 1 1.0000* REM-99 (IA)

TFA-Y (VA) 0.4938* 0.4929* 0.7120* 0.7120* 0.7117* 0.7097* 0.3565* 0.7509* 0.5264* 1 REM (IA)

Cost efficiency DFA-95 (IA) DFA-90

(IA) DFA (VA)

DFA-99

(VA)

DFA-95

(VA)

DFA-90

(VA) TFA (IA) TFA (VA)

TFA-Y

(IA)

TFA-Y

(VA)

The table reports Spearman‟s rank correlations between efficiency scores of all employed methodologies. For space considerations we include both profit and cost efficiency output in

one table, displayed in grey and white background, respectively. The asterisks denote significant correlation at 5% level.

263

CHAPTER 8

Bank competition and financial (in)stability in Europe:

A sensitivity analysis

265

8.1. Introduction

The deregulation process paves the way towards the intensification of competitive

conditions, amid which financial institutions struggle to survive posing a threat to the

potential incidence of financial crises. Systemic risk is large especially for incumbent

banks, whose high market share may imply negative externalities to national

economies in cases where cross-border activity is significant.

The ongoing restructuring towards aggregate concentration for income diversification

and risk management purposes has rendered financial regulation imperative building

upon the failing premises of Basel II. In the light of the present financial crisis, capital

regulation has been deficient falling short of taking account of systemic effects,

market discipline ineffective due to too-big-to-fail policies, risk evaluations deficient

once we consider the operation of credit ratings agencies while supervision exhausted

its jurisdiction to non-shadow banking system. In line with the ongoing debate on the

dynamics of competition and financial stability taking into consideration

contemporary institutional reforms we shed light on the dynamics of competition in

association with the evergreen topic of financial stability.

The contribution of the underlying paper is threefold: first, apart from Cipollini and

Fiordelisi (2009), Uhde and Heimeshoff (2009) and Agoraki et al. (2009) (for the

Southeastern region) there are no other empirical applications focusing exclusively on

Europe; comparative analysis of countries including Europe has been conducted by

Schaeck and Cihak (2008), Boyd et al. (2009), De Nikolo and Loukoianova (2007),

Berger et al. (2009) and Laeven and Levine (2009). Second, the U-shaped relationship

of competition and financial stability as theoretically proposed by Martinez-Miera and

Repullo (2010) is investigated empirically since no one but Berger et al. (2009) and

Beck et al. (2012) allowed for it. Third, it is interesting to see if the effects of

concentration of market power and concentration are persistent along with the

statistical significance of other sets of variables included in the deterministic kernel.

266

8.2. Methodology

The decision upon which key variables are theoretically and empirically grounded to

constitute the crux of deterministic group is not that apparent. The underlying

literature has proposed different econometric modeling – (e.g) GMM methodology,

panel, probit, logit models and duration analysis, etc. – and various determinants of

bank risk of failure. Such unwieldy bulk of variables amounts to almost 50

institutional, macroeconomic, financial, industrial and bank-specific variables that end

up with mixed results depending each time on the econometric specification and the

independent set of the regression model. On those grounds, this paper considers

interesting the prospect of analysing the statistical power of key effects (market

power, concentration) of bank risk conditioned to the variation of every single subset

of variables. In other words, utilising extreme bound analysis (EBA) as set out by

Leamer (1983; 1985) and Leamer and Leonard (1983), the partial correlations of

dependent and independent variables are examined to endorse whether such

relationships are fragile or robust at standard confidence levels.

Hence, the employed methodology is a sensitivity analysis of linear modeling

regarding multiple regressions of bank risks on explicit groups of variables. We

embark upon the group of interest that comprises the effects of market power and

concentration along with their quadratic terms to allow for non-linearities as

theoretically proposed by Martinez-Miera and Repullo (2010). Along the lines, both

centrifugal strands of literature can be verified without precluding the significant

power of the other. The so-called U-shaped relationship of competition and bank risk

of failure highlights that in monopolistic markets the risk-shifting effect is dominant –

greater competition lowers loans rates and makes banks safer – and the marginal

effect – greater competition lowers the revenues from non-performing loans providing

limited buffer for potential loan losses and thereby jeopardising bank viability –

prevails in more competitive markets.

The following steps of analysis include reasonable subsets of cautiously pooled

variables that one can identify as statistically significant in the literature.

Schematically,

267

Financial stability (Z) = f [Market structure variables (M) | institutional (I)

|macroeconomic and bank-specific factors (C)] + ε (8.1)

The market power set of variables comprises the Lerner index as applied to capture

the pricing conduct of banks. The mark-up of price seems appropriate in our case but

the procedure of extrapolating marginal costs should follow first the standard

stochastic modeling of equation (8.3) along with the estimation of the partial

derivative of total costs with respect to total assets (equation 8.4). Concentration is

delineated by Herfindahl-Hirschman ratio since it is deemed in the literature to depict

real market conditions in antitrust policies and represent the stepping-stone of SCP

paradigm. The selection of both measures that have drawn attention to the empirical

analysis is cautious of the fact that they have been overwhelmingly applied in the

literature interchangeably to explain market conditions34

; along the lines of the

analysis, we opt to plug them both in regressions to compare the accruing evidence.

All variables allowed to have variation at the bank level, namely competition and

several other controls that encompass the bank‟s business model, are lagged one

period to avoid reverse causality among them. In other words, the endogeneity of

market power may reflect the impact of insolvency on market structure and pricing

conduct of banks subsequently. The fact that country-level factors are not supposed to

interact with market power and stability comes at the expense of correlation of the

error term with some independent effects. However, the use of regulatory variables

that by nature have little variation over time leads to limited, if not negligible,

variation of market power at the bank level. In addition, fixed effects modeling is

more appropriate for a particular sample set, whereas largely heterogeneous N groups

(as in our case) would induce significant loss in the degrees of freedom (Baltagi,

1995). Thus, EBA methodology requires bank and time random effects as well as

clustered standard errors at the bank level in order to get a prima facie insight of the

potential drivers of country heterogeneity.

The bound analysis kicks off at the point where the baseline model regresses financial

stability on the variables of interest (M) after controlling other bank-specific factors

and adverse macroeconomic conditions (C) across the region to deduct as much a

34

See Schaeck (2009) for a thorough review on measures of concentration, market power and financial

stability in the literature.

268

possible the error effects and convey real depiction in levels and significance of the

underlying coefficients. Next, we add in various bank-specific factors that encompass

different aspects of banks‟ balance sheet, income structure, corporate governance and

general strategic planning investigating thereby their stand-alone and „en masse‟

effect. As a next step thereafter, institutional factors that encompass different legal

systems, regulatory schemes and market discipline mandates (I) are plugged in to pin

down the bounds of the coefficients of interest.

Thus, the extreme upper and low bounds of the coefficient values are constructed by

allowing for all possible combinations of (I) effects expanding the analysis up to 62

regressions for each model. The degree to which partial correlations between market

structure and financial stability are robust or fragile is defined by the persistence of

the sign and the statistical significance in-between the range of the bound. Otherwise,

we should treat the variables of interest with less confidence as far as their causal

effect on risk is concerned since the fragility in the underlying relationships in terms

of sign and statistical inference is contingent on the employed information set.

However, the analysis does not aim to come up with a single model that breaks down

the competition-(in)stability nexus. Rather, we assess the interconnections in between

on the grounds of what has been proposed in the empirical literature and always on

the mandate of ever-afresh institutional reform.

We also endeavour to resolve the adverse results on the relationship between

competition and concentration through the employment of the quadratic term of

Lerner index to verify whether it is the case of U-shaped relationship as argued by

Martinez-Miera and Repullo (2010). We plug it ad hoc only in the models that

construct the extreme bounds of competition and concentration so as to estimate the

infection point that is the value of competition above which the relationship of

soundness-competition alters; otherwise, the use of it in all model combinations

would blur the effective bounds. In case of no significance in the coefficient of the

Lerner index that constructs the extreme bounds, we opt to utilize a different

information set of the same size.

Moreover, the aim of gauging the impact of country-level variables on risk seems to

be insufficient, if not inappropriate due to the possibility that the underlying effect

may not be expressed in levels but also in slopes (model 2). Hence, we include

269

interaction terms between country-level factors and the Lerner index along with

lagged bank-specific controls. However, the only drawback with this approach is that

interaction terms may bring about multicollinearity problems that are partially

counterbalanced by more degrees of freedom due to the regression analysis conducted

on a whole sample. Such problems are depicted in inflated standard errors and,

consequently, in higher coefficients of (M) variables revealing thereby a potential

weak-data problem; the latter consists in the little variation of a specific independent

factor to determine cross-sectional differentials (Leamer, 1978; Levine and Renelt,

1992). The model is schematically the following:

Financial stability (Z) = f [Competition (L), Concentration*L | institutional (I)*L

|macroeconomic*L and bank-specific factors)] + ε (8.2)

In contrast with Levine and Renelt (1992), the variables of interest are only the Lerner

index and concentration, which are included in the analysis simultaneously since they

tend to reflect different features of bank competitiveness. Second, we utilize all

possible combinations of three I-variable sets in order to juxtapose the results with

those when different-sized sets are employed for robustness-checking reasons. In

addition, apart from the M-variables each model comprises 5 bank-specific variables,

three country-level variables and any combination of the I-variables, a procedure that

relaxes the „prescription‟ of Kormendi and Meguire (1985), Barro (1991) and Levine

and Renelt (1992).

8.3. The model

We should first estimate the price mark-up over marginal cost (Lerner index)

combining the estimation of average prices and marginal costs at the bank level. The

average prices are estimated over total assets (TA) along the lines of Shaffer (1993)

and Berg and Kim (1994), instead of other earning assets in an attempt to expand as

much as possible the observations of the sample since 2002. First, we have to estimate

marginal costs by means of running a translog cost function, similar to the version of

Ariss (2010b) that excludes the use of price of borrowed funds as input price on the

270

grounds that it presumably captures some degree of monopoly power of incumbent

banks in the deposit market. The employed model takes the following form:

(8.3)

where TC: total costs (total operating costs (interest expenses, personnel and other

costs), Q: total bank output or total assets, W1: price of labour (personnel expenses

over total assets), W2: price of physical capital (other operating expenses over fixed

assets), Z1: fixed assets deflated by total equity, Z2: Off-balance sheet activities

deflated by total equity and T: time trend35

. We employ fixed effects modeling (after

applying the hausman test) so as to account for different bank specificities and run

model (8.3) separately for each banking market to reflect different technologies in the

region. We also utilise a time trend to interact with the deterministic kernel in order to

capture time-varying and non-neutral technological progress in the banking sector.

Homogeneity of degree one in input prices (Σγk=1) and symmetry conditions in all

quadratic terms are imposed in model (8.3).

When it comes to the estimation of the Lerner index, we extrapolate the marginal

costs by running the following model, which is schematically the partial derivative of

total costs with respect to total assets (see Berger et al. 2009; Ariss, 2010):

MC TC

Q1 2 lnQ k lnW k lnZ

2

2

(8.4)

35

The model utilises observations of each bank i at time t. We omit the subscripts for convenience

purposes.

271

We are then able to construct the Lerner index (L) with respect to specific bank

activities before delving into the analysis of competition determinants. According to

the following structural model,

Li,t ARi,t MCi,t

ARi,t (8.5)

AR denotes the average revenue of banks estimated by total income over total assets

and MC the marginal cost derived through model (8.4). Their subscripts signify the

use of Lerner index as the only proxy of market power at the bank level over time.

We then estimate the natural logarithm of Z-score, which has been widely accepted in

the literature as the most reliable proxy of distance from a situation of insolvency. We

compute it with the following expression,

Zi,t ROAi,t E /TA

i,t

ROAi,t (8.6)

where ROA: returns on assets, EQ/TA: total equity to total assets, σ(ROA): standard

deviation of returns on assets, all expressed at for bank i at time t. It is interpreted as

the number of standard deviations by which ROA should fall under the mean so as to

extinguish the equity of a bank (Boyd and Runkle, 1993). We opt to estimate profit

volatility for the whole time period at the expense of subsequent time invariance of

the denominator; in some studies where time periods are much larger, a three-year

rolling window is implemented so as to potentially attribute the variation of the Z-

score not only to the variation of profitability and capital, but also to the volatility of

bank profitability. When it comes to set it as a dependent variable, we take the natural

logarithm of Z-score in order to normalize its extreme values due to high skewness. In

the literature, any possible negative values are alleviated by transforming them

through truncation at zero point [ln(1+Z-score)] or winsorising at 1% level and then

taking logs. Since the latter produces non-negative values, we opt for it.

272

8.4. Determinants of stability

8.4.1. Bank-specific variables

Asset size has been used in the literature by Bonfirm and Dai (2009), Liu et al. (2010)

and Beck et al. (2012) in an attempt to see whether financial stability comes from

managers‟ attitude to exploit scale economies or by the perception that too-big-to-fail

policies will constitute the facility of last resort in the form of governmental subsidies,

among others.

Capital ratio is the value of total equity deflated by a bank‟s total assets. We employ

it to account for differentials in risk preference behaviour of bank managers along the

lines of Stiroh (2004), Schaeck and Cihak (2008) and Berger et al. (2009).

Cost efficiency turns out to be the most widely employed accounting variable that

proxy for cost efficiency as contemporary efficiency modeling may produce bias due

to certain methodological and econometric assumptions. We prima facie expect a

negative effect on stability since inefficient banks tend to engage in risky behaviour to

make up for insufficient performance (Uhde and Heimeshoff, 2009; Liu et al. 2010).

However, a lagged variable may have stabilising repercussions when it is coupled

with divergent institutional reform.

Liquidity is controlled for by the ratio of liquid assets over customer deposits and

short-term funding. It measures what percentage of deposits and funding can be

served in case there is a sudden bank run. The higher this ratio is the less vulnerable a

bank is vis-à-vis a deposit run-off case. We see similar proxies in the literature, such

as liquid assets over liquid liabilities or over total assets (Laeven and Levine, 2009;

Olivero et al., 2010), with no substantial difference in practice.

Diversification indicates the ability of a bank to expand its operations to off-balance

sheet activities, namely to insurance, real estate and securities activities; thus, we

proxy it as the total non-interest operating income over total income (Liu et al., 2010).

We expect a negative association between diversification and risk but it also might be

the case that banks with high-income diversification are exposed to greater risks in

their attempt to accomplish economies of scope (Stiroh, 2004).

273

8.4.2. Macroeconomic variables

GDP growth rate has been employed by the studies of Bonfirm and Dai (2009),

Jimenez et al. (2010) and Olivero et al. (2010) in order to control for different stages

of economic development. As for its expected effect, higher customer demand after

the adoption of Euro may have resulted in better managerial efficiency in terms of a

relatively superior utilisation of production factors (Conrad et al., 2009; Liu et al.,

2010). Economic prosperity reduced the probability of a potential bank crisis, which

usually comes along with loan risk during economic recessions. On the other hand,

loan losses can occur during economic booms if high growth GDP rates promote

optimistic evaluations over borrowers‟ creditworthiness leading to less stringent

policies, and when competitive structures make managers more willing for risk-taking

activities (Jimenez et al., 2010).

Stock market turnover is defined as the total value of traded shares over the average

stock market capitalisation. We employ the degree of liquidity in stock markets in

order to take account of alternative funding means of firms, which may be related to

greater dissemination of credit information and, thus, to greater bank soundness (Beck

et al., 2012).

8.4.3. Regulatory environment

Capital regulatory index measures the degree of regulation on bank capital that

should be set aside as a buffer for potential market and credit risks. In particular, it is

about the initial capital stringency, that is which type - and to what extent - of

regulatory funds other than cash, governmental securities or borrowed funds, is

appropriate and verifiable by the official regulatory authorities. It is also about the

overall capital stringency, according to which the regulatory capital is estimated

accounting for risks and value losses. Thus, we quantify it by ascribing values of 0 or

1 to every single one of the nine questions included in the appendix, with the

observations ranging between 0 (no stringency) and 9 (high stringency). After the

advent of Euro, the necessity to build upon the inefficient or inadequate regulatory

directives as set out by the Accord of the Basel Committee (Basel I), Basel II and III

constitute the product of sedulous research. Empirical studies are split between the

274

invigorating effects of capital requirements on less loan losses and the detrimental

implications on risk-taking. Required reserves of capital may constitute sufficient

buffers vis-à-vis potential liquidity shocks notwithstanding the case of banks seeking

for monopolistic rents in order to make up for the utility loss of powerful bank owners

(Laeven and Levine, 2009).

Official supervisory power measures the degree of supervisory power exercized by

the official authorities and their „intervention‟ to the decisions of bank managers. It

takes values from 0 to 10 ascribing 0 and 1 to negative and positive responses,

respectively. From a theoretical perspective, strong supervision tends to demoralize

managers to engage in excessive risk-taking - especially in countries with low

accounting requirements (Fernandez and Gonzalez, 2005) - whereas it may be

associated with corruption in lending transactions, and obstruction of bank operations

(Barth et al., 2004).

Private monitoring index is indicating the degree of information released to the public

and officials relative to requirements of auditing authorities and credit rating agencies.

It takes values between 0 and 10 after taking into account the „no‟ and „yes‟ responses

of 10 questions, respectively. Hence, higher values highlight greater private insight

over the economic performance of banks. It has been overlooked in the literature and

only recently has been utilized by Schaeck et al. (2009), who argue about the

insignificant concentration due to the common practice of investors, regulatory

authorities and credit agencies to inspect large entities closely.

Activity restrictions is an interesting variable commensurate with the extent to which

bank activities like securities, insurance and real estate activities are under constraint.

In particular, it takes the responses of „prohibited‟, „restricted‟, „permitted‟ or

„unrestricted‟ and we quantify them by assigning the values of 4, 3, 2 and 1,

respectively. We, finally, get the average value of the overall index and draw remarks

over the degree of activity restriction. In the literature, there are two strands of

reasoning in favour or against their effect on bank soundness. In cases when such

restrictions forbid banks to engage in riskier fee-based projects (securities, real estate

and insurance), we observe financial stability (Uhde and Heimeshoff, 2009).

However, if banks are restricted to diversify their portfolio to non-interest bearing

275

products, the concomitant utility loss induces powerful bank owners to herding and

thus to riskier systemic conduct (Laeven and Levine, 2009).

Foreign ownership is calculated as the total assets of banks, which are owned by

foreigners with more than 50% stake, as a percentage to the total assets of the banking

system they operate within. It is addressed in many studies like those of Yeyati and

Micco (2007), Berger et al. (2009) and Laeven and Levine (2009), among others. The

issues related to penetration of foreign banks in a national market are the screening

costs of local customers that tend to attenuate through acquired experience, and the

guarantees of the parent bank that constitute a safety net in times of insolvency and

liquidity shocks (De Nikolo and Loukoianova, 2007; Claessens et al. 2001b).

There is also the option to pick up national banks of monopolistic markets (dodging

competition hypothesis), higher operational efficiency (cream skimming hypothesis),

or large market shares through branches and subsidiaries (quest for market power

hypothesis). There might be also the case that foreign ownership affects bank

soundness in slopes rather in levels, entering significantly in the competition-

(in)stability nexus; this possibility is included in the robustness-check section in order

to verify the persistency of the underlying relationship.

Fraction of entry denied consists in the fraction of denied entry applications of

domestic and foreign entities over their total volume. Its possible impact on bank

soundness has been profound in contestable markets, in which too-big-to-fail banks

operate competitively (Beck et al., 2006a); Beck, 2007). A positive sign is likely to

depict a monopolistic banking market where banks reap the benefits of diversifying

portfolios, stock market financing and proprietary information (Beck et al. 2012). In

contrast, financial crisis may occur in response to either unbearable operational

inefficiencies or to the limited relationship lending in a competitive market (Boot and

Thakor, 1993).

8.5. Data

The sample includes financial data of at most 2421 banks headquartering in the

enlarged European Union (27-EU). We retrieved data from consolidated accounts of

276

the Bankscope database, and when that is not possible, we use unconsolidated

accounts. The data amount to 9296 observations for the period 2002-2008 and pertain

to Austria, Bulgaria, Cyprus, Denmark, Estonia, France, Germany, Greece, Hungary,

Italy, Luxemburg, Portugal, Slovakia, Slovenia, Spain, Sweden and United Kingdom.

The decision to analyse a period up to the crack of the financial meltdown emanates

from the employment of supervision and regulatory variables36

, which are available

until June 2008 and we further assume that for the whole year of 2008 the data remain

the same.

To avoid losing observations-outliers in an already limited sample, we winsorize at

1% level of the distribution of the Lerner index and the Z-score to smooth out the

their impact and come up with robust standard errors. The sample also includes the

whole spectrum of productive specialisation in European banking, namely bank

holding and holding companies, clearing institutions and custody, commercial banks,

cooperative banks, finance companies (credit card, factoring and leasing), group

finance companies, investment and trust corporations, investment banks, islamic

banks, micro-financing institutions, other non-banking credit institutions, private

banking and asset management companies, real estate and mortgage banks, saving

banks, securities firms and specialized governmental credit institutions.

Table 22 contains summary statistics of the key bank-specific and regulation variables

employed throughout the main econometric part. Each column exhibits the mean

values per country as well as the average of the Euro zone (EU-15) and the rest

European Union (EU-17)37. In particular, EU-15 is far more stable throughout the

period 2002-2008 than European banking sectors; Germany and Spain stand out as

the best performers in contrast with Finland and Luxembourg, which are below the

average of EU-12. In the latter case, Lithuania, Czech Republic and Slovakia enjoy

greater bank soundness whereas Romania, Cyprus and Latvia lie below the lowest

score of EU-15 countries.

36

In particular, the variables activity restrictions, capital regulation, foreign ownership, official

supervision, fraction of entry denied and private monitoring index are quantified in order to produce

time series for every country through the use of questionnaires disseminated in many financial

institutions. The methodology (and data) followed to quantify more than one answers is explicitly

encompassed in the appendix as retrieved from Barth et al. (2004; 2005; 2008).

37

We include Bulgaria and Romania in the EU-10 as they constitute the enlarged Southeastern part of

European Union.

277

Table 22: Descritive statistics of key variables

Z-score Lerner TA TC/TI TNINTI/TI LIQ/DEP CONC GDPGR Inflation Stock MT

Activity

R.

Capital

Reg.

Foreign

Own.

Official

Sup.

Fraction

ED

Private

Mon.

Austria 55.071 0.305 5873.057 0.659 0.203 0.365 0.638 0.023 0.020 0.350 7.000 5.857 0.194 10.143 - 4.000

Belgium 29.233 0.074 44573.660 0.652 0.295 0.417 0.850 0.020 0.023 0.678 7.571 3.571 0.210 9.286 0.000 5.714

Denmark 34.256 0.208 9384.117 0.602 0.211 0.360 0.805 0.014 0.020 0.634 9.286 5.571 0.150 8.286 0.000 6.429

Finland 22.267 -0.030 34316.880 0.657 0.386 0.498 0.958 0.031 0.016 1.064 8.714 5.000 0.426 7.429 0.048 6.000

France 63.441 0.242 36018.890 0.659 0.306 0.426 0.580 0.015 0.021 0.812 8.143 6.571 0.155 6.286 0.000 6.429

Germany 168.847 0.168 5417.424 0.718 0.180 0.230 0.700 0.014 0.019 0.453 7.000 6.714 0.059 6.857 0.286 6.714

Greece 34.050 0.246 18160.340 0.657 0.215 0.317 0.842 0.035 0.035 0.587 8.571 4.571 0.108 9.143 0.076 6.429

Ireland 32.189 -0.171 40898.670 0.395 0.037 0.871 0.543 0.039 0.031 0.524 7.000 2.286 - 10.286 0.000 7.000

Italy 69.178 0.176 6903.262 0.669 0.199 0.349 0.422 0.008 0.025 0.425 11.429 4.286 0.096 6.286 0.130 6.714

Luxemburg 28.827 0.406 7723.828 0.551 0.283 0.834 0.304 0.040 0.031 1.638 8.143 7.000 0.946 9.429 0.000 6.000

Netherlands 30.412 0.019 88215.220 0.664 0.159 0.657 0.686 0.020 0.021 0.922 6.000 5.286 0.069 5.714 0.000 7.429

Portugal 41.970 0.181 12145.250 0.746 0.201 0.484 0.891 0.009 0.028 0.396 11.429 7.429 0.165 12.571 0.000 5.714

Spain 99.873 0.203 22986.740 0.569 0.171 0.219 0.774 0.030 0.034 0.880 6.714 9.000 0.097 9.000 0.026 7.429

Sweden 27.886 0.353 12957.060 0.597 0.210 0.236 0.960 0.027 0.018 1.007 9.429 3.000 - 5.143 0.010 5.714

UK 37.935 0.253 52002.790 0.735 0.351 0.726 0.587 0.023 0.020 1.247 4.286 5.714 0.519 7.429 - 7.000

EU-15 51.696 0.176 26505.146 0.635 0.227 0.466 0.703 0.023 0.024 0.774 8.048 5.457 0.246 8.219 0.044 6.314

278

Cyprus 14.255 -0.249 5458.085 0.627 0.090 0.468 0.873 0.036 0.028 0.579 11.000 6.429 0.301 9.429 0.333 6.714

Czech R. 40.790 0.238 6219.384 0.670 0.303 0.291 0.663 0.047 0.024 0.283 12.000 5.571 0.860 8.000 0.000 6.000

Estonia 19.374 0.281 985.089 0.675 0.372 0.782 0.976 0.062 0.048 0.314 7.143 5.143 0.992 11.857 0.000 6.000

Hungary 21.442 0.304 2134.387 0.823 0.180 0.648 0.679 0.032 0.054 0.260 11.000 7.143 0.940 12.571 0.000 6.429

Latvia 16.513 0.537 1318.662 0.638 0.296 0.485 0.551 0.073 0.071 0.107 7.714 6.286 0.563 9.429 0.167 6.714

Lithuania 43.599 0.205 1844.011 0.692 0.254 0.295 0.684 0.075 0.034 0.224 10.143 4.286 22.995 11.714 0.057 6.429

Malta 29.773 0.104 1891.901 0.704 0.109 0.597 0.764 0.025 0.025 0.536 10.286 6.714 0.632 12.571 0.000 6.714

Poland 25.862 0.372 3406.975 0.674 0.248 0.337 0.592 0.046 0.024 0.286 7.714 4.143 20.128 7.286 0.000 6.000

Slovakia 50.793 -0.040 4554.698 0.671 0.340 0.364 0.758 0.066 0.046 0.069 10.286 5.143 0.923 11.857 0.250 3.571

Slovenia 29.490 0.209 1428.511 0.709 0.201 0.250 0.711 0.045 0.045 0.310 10.286 5.857 0.197 11.286 - 6.714

Bulgaria 39.837 0.557 908.921 0.735 0.249 0.429 0.488 0.061 0.067 0.202 9.714 7.286 0.734 9.571 0.133 5.714

Romania 18.231 0.422 2253.744 0.797 0.258 0.506 0.662 0.064 0.112 0.170 11.286 5.714 0.473 8.286 0.273 5.000

EU-12 29.163 0.245 2700.364 0.701 0.242 0.454 0.700 0.053 0.048 0.278 9.881 5.810 4.145 10.321 0.110 6.000

Z-score: the unlogged version of Z-score before winsorizing it; Lerner: the Lerner index before winsorizing it, in order to draw remarks on its mean values across the European region; TA:

total assets; TC/TI: total cost over total income; TNINTI/TI: total non-interest income over total income; LIQ/DEPSTF: liquid assets over total deposits and short-term funding; CONC:

market concentration; GDPGR: the growth rate of GDP; Inflation: inflation rate; Stock MT: stock market turnover; Activity R.: activity restrictions; Capital Reg.: Capital regulation index;

Foreign Own.: the share of foreign-owned assets in a banking industry; Official Sup.: official supervisory power; Fraction ED: fraction of entry denied; Private Mon.: Private monitoring

index. EU-15: the average values of all variables deflated by the number of banks within a banking market; EU-12: the average values of all variables deflated by the number of banks

within a banking market including Bulgaria and Romania of the enlarged European Union. Source: World Bank, Eurostat and own estimations.

279

Furthermore, harsher monopolistic conditions occur in Luxemburg and, especially in

Latvia and Bulgaria. In contrast, Finland, Ireland, Cyprus and Slovakia have negative

values of the Lerner index, indicating an irrational behaviour on pricing their

products. The largest markets in terms of total assets are those of Netherlands, UK

and Belgium while Estonia and Bulgaria are quite underdeveloped in EU. Thus, EU-

15 region is ten times as much bigger than that of EU-12 along with their associated

costs. Total costs over total income take values above 60%, with Ireland

demonstrating twice as much lower than the rest. In addition, income diversification

as proxied by the share of non-interest income over total income is more profound in

the Southeastern Union, albeit more than 35% takes place in Finland, UK, Slovakia

and Hungary. Liquid assets as a percentage of deposits and short-term funding stand

at 46% for the whole European region notwithstanding Ireland, Luxembourg and

Estonia get more than 78% as opposed to Germany, Spain, Estonia and Czech

Republic.

As for country-level conditions, there is a considerable degree of concentration in

Finland, Sweden and Estonia approaching almost perfect monopoly; less colluding

markets are those of Luxemburg, Italy and Bulgaria creeping at levels of up to 50%.

Moreover, higher rates of economic expansion as depicted by GDPGR and inflation

rates are expected to lie in the underdeveloped EU-12 region, especially so in Latvia

and Lithuania that hit 7% economic growth. On the contrary, stock market activity

has been more intense in the most capitalized markets of UK, Luxembourg, Finland

and Sweden, followed by the EU-12 countries falling far behind.

Information on regulation and supervision are quite interesting to see whether

pertinent policies lean towards too-big-to-fail cases by protecting the domestic

markets or institutionally reforming information dissemination, supervisory control or

capital adequacy. What stands out is Romania and Germany that protect incumbent

banks by denying market entry to almost 28% of the submitted applications. Further

restrictions on activities are imposed to Italy and Portugal as well to the vast majority

of underdeveloped European territory. Capital adequacy requirements are more

common in Luxembourg, Portugal, Spain and, secondarily, to Hungary and Bulgaria.

In addition, the ability of official authorities to intervene in the banking process of

decision-making is more evident in Malta and Portugal whereas EU-15, and

280

especially the markets of Sweden and Netherlands, affords the opportunity for more

freedom in the hands of bank managers. We also observe higher information

requirements from auditing authorities and credit rating agencies in Netherlands and

Spain, with the case of Slovakia falling much less behind the standards of the

remainder. Last, the banking markets of Luxembourg, Estonia, Hungary and Slovakia

differentiate themselves from the rest European sample, as foreigners at more than

94% level own their assets. In contradistinction, Germany, Netherlands, Italy and

Spain lie in scores lower than 8% level.

Table 23 provides information on pairwise correlations of the country-specific

variables employed in the analysis. As we can see, correlation is significant at 1%

level between the variables of regulation, supervision and governance. In particular,

countries with high market concentration experience more liquidity in the stock

market and more restrictions on securities, insurance and real estate activities. Greater

growth rate of GDP and inflationary pressures are correlated with less control power

on bank managers‟ decisions and increased private insight on banks‟ economic

performance. Furthermore, the relationship of stock market turnover with both

activity restrictions and private monitoring index is negative at 28.5% level.

Table 23: Correlation between country-level variables

Concentration GDPGR Inflation

Stock

market

turnover

Activity

restrictions

Capital

regulation

Foreign

ownership

Official

supervision

Fraction

of denied

entries

Private

monitor

Concentration 1

GDPGR 0.031 1

Inflation -0.064 0.201 1

Stock market

turnover -0.421 -0.049 0.236 1

Activity

restrictions 0.32 0.064 0.024 -0.355 1

Capital

regulation -0.035 0.116 -0.058 -0.024 0.009 1

Foreign

ownership 0.175 0.104 0.09 -0.091 0.090 0.074 1

Official

supervision 0.031 -0.49 -0.322 -0.163 -0.088 -0.031 0.445 1

Fraction of

denied entries -0.077 0.016 0.014 0.035 0.242 -0.072 -0.510 -0.511 1

Private

monitor -0.099 0.359 -0.144 -0.285 0.066 -0.055 -0.112 -0.284 0.213 1

Source: Bankscope database, World Bank, own estimations.

281

Denied market entries are strongly related to activity restrictions as both determine

competitive conditions, and private monitoring index operating as a supplement

strategy to induce transparency in a protected market of incumbent banks. They are

also negatively correlated to less official supervision, since the latter may operate as a

substitute mechanism to correct imprudent practices not through the protection of the

national market but through the direct control of managerial decisions, and less

foreign ownership as restricted to integrate in other regions through the organic

growth of subsidiaries and branches. Last, official supervisory power is associated

with markets of more foreign-owned banks and less bank information required by

public authorities. Hence, in monopolistic markets we see more restrictions on market

entries and activities as well as more private monitoring through the dissemination of

bank information. In contrast, less protected markets experience the incoming of

foreign interest and the decisive intervention of supervisory authorities to the

decisions of bank managers.

8.6. Main results

Table 24 displays the results of the sensitivity analysis conducted to see whether and

to what extent factors of regulation and supervision have an independent impact on

bank soundness. The baseline model in the first column excludes everything but the

macroeconomic and bank-specific controls while the rest comprises the results we add

one key variable at a time in order to see the variables of market structure. The

coefficient of Lerner remains significant with a negative effect only when we include

activity restrictions and capital regulation. That highlights the devastating effect of

monopolistic conduct on bank stability in cases where managers are constrained to

diversify their portfolio risks and exploit their capital base. In contrast, market

concentration enters with positive bearing on bank soundness up to 1% significance

level except three cases. The first two are the same that bank competition enters

significantly therein, namely with the inclusion of restrictions on activities and capital

reserves. The third lies when we plug all institutional variables in the model bringing

about a negative resultant force for HHI; that might be the case of multicollinearity

problems when all institutional policies are assumed to co-exist in an economy.

282

There is an explicit significance of regulatory variables when they are assessed

individually. Once we plug them all in the model their „en masse‟ effect turns out to

be significant at 1% level. Significance of activity restrictions38

denotes that managers

are restricted to enter other non-traditional business lines for diversification purposes,

taking on more risks and exacerbating thereby their portfolio risk exposure. On the

contrary, capital regulation enters significant at 1% level fostering stability to the

banking sector. Hence, more capitalization makes banks immune to liquidity shocks

although the opportunity cost of such a „tax burden‟ in cases of powerful owners,

constitute a considerable motive for risk taking by means of monopolistic conduct

(Laeven and Levine, 2009).

Similarly, foreign ownership is linearly related to bank fragility meaning that the

openness to foreign institutions leads to excessive competition and diminishing profit

margins dominating their invigorating impact of adopting better practices that

enhance further operational performance. Bank soundness peters out as official

supervision becomes more stringent on the grounds that some degree of corruption in

lending activities undermines systemic efficiency due to limited firms access to bank

finance (Beck et al., 2006b; 2007). Μore rejections on potential market entries that

intensify the degree of contestability and leave incumbent banks free to take

advantage of profit opportunities, equity financing and other informational rents,

demonstrate no linear effect on stability. Last, private monitoring is highly significant

with a positive bearing on banking system stability, without losing the explanatory

power of market concentration. That underlines the significant role of too-big-to-fail

banks that have been subject to close monitoring and inspection due to their potential

systemic repercussions. However, once we employ the whole set of I-variables, we

come up with the same pattern notwithstanding denied market entries enter positive at

10% significance level and private monitoring changes sign possibly due to its

negative correlation with official supervisory power.

38

The stabilizing effect of activity restrictions is also found by Berger et al. (2009) and Beck et al.

(2011) unlike Beck et al. (2007), Laeven and Levine (2009) and Uhde and Heimeshoff (2009)

advocating to the contrary.

283

Table 24: Regression output of model 8.1

Variables Baseline Sensitivity analysis

Lerner -0.155 -0.210** -0.252*** -0.001 0.155 0.023 0.064 -0.077

(0.128) (0.098) (0.094) (0.120) (0.096) (0.138) (0.104) (0.151)

Concentration 0.537*** -0.019 0.069 0.689*** 0.614*** 0.696*** 0.516*** -0.442**

(0.082) (0.091) (0.081) (0.085) (0.087) (0.102) (0.082) (0.175)

Institutional variables (Regulation, supervision, governance)

Activity restrictions -0.089*** -0.091***

(0.007) (0.016)

Capital regulation 0.138*** 0.094***

(0.009) (0.016)

Foreign ownership -0.658*** -0.854***

(0.086) (0.221)

Official supervision -0.091*** -0.081***

(0.011) (0.018)

Fraction of denied entries -0.159 0.393*

(0.166) (0.238)

Private monitor 0.143*** -0.395***

(0.017) (0.054)

Country-specific variables

GDPGR 2.402*** 1.996*** 2.280*** 4.555*** 3.843*** 5.912*** 3.488*** 9.220***

(0.735) (0.726) (0.740) (0.862) (0.764) (1.315) (0.785) (1.886)

Inflation -4.831*** -4.458*** -6.338*** -2.126** -2.341*** -7.073*** -5.297*** 4.029**

(0.860) (0.823) (0.851) (0.974) (0.871) (1.607) (0.872) (1.851)

Stock market turnover -0.079*** -0.190*** -0.188*** 0.101*** -0.113*** -0.577*** -0.127*** -0.188*

(0.030) (0.032) (0.030) (0.029) (0.031) (0.073) (0.032) (0.097)

284

Bank-specific variables

Q 0.002 -0.001 -0.008 -0.013 0.004 0.015 -0.005 -0.007

(0.009) (0.009) (0.008) (0.009) (0.009) (0.011) (0.009) (0.011)

E/TA 0.343** 0.560*** 0.473*** 0.396** 0.361** 0.211 0.258 0.713***

(0.165) (0.161) (0.161) (0.168) (0.164) (0.238) (0.169) (0.238)

Cost to income -0.016 -0.048 -0.069 -0.001 -0.036 0.059 0.044 -0.041

(0.055) (0.044) (0.045) (0.063) (0.048) (0.076) (0.060) (0.057)

TNINTINC/TI -0.391*** -0.394*** -0.420*** -0.485*** -0.435*** -0.249** -0.426*** -0.442***

(0.080) (0.080) (0.079) (0.083) (0.080) (0.104) (0.080) (0.104)

Liquidity -0.039** -0.054*** -0.044*** -0.036** -0.036** -0.061** -0.030* -0.084***

(0.016) (0.016) (0.016) (0.017) (0.016) (0.023) (0.016) (0.028)

TIME DUMMIES YES YES YES YES YES YES YES YES

COUNTRY DUMMIES YES YES YES YES YES YES YES YES

OBSERVATIONS 9120 9120 9120 8683 9120 7541 9120 7066

BANKS 2408 2408 2408 2300 2408 1979 2408 1876

COUNTRIES 27 27 27 23 27 18 27 15

Random effects (RE) model with standard errors clustered at the bank level employing time dummies to capture time-varying random effects, as their significance is

evident after running standard F-tests. The first column reports the results of the baseline model, that is the regression of Z-score on banks specific and macroeconomic

variables totally devoid of any institutional controls. We then include one at a time controls for regulation, supervision and governance in order to verify their stand-

alone effect on bank stability as well as their „en masse‟ explanatory power in the last column. Z-score is logged for robustness in standard errors and all bank-specific

variables are lagged one period to avoid the possibility of reverse causality. Standard errors are in parentheses while asterisks ***, **, * denote the significance level

being at 1%, 5% and 10%, respectively.

285

Furthermore, times of high market demand as expressed by GDP growth make banks

utilize production factors in a more efficient way of diminishing average costs

(Conrad et al., 2009), while the combination of policies altogether are able to boost

further the impact of the business cycle towards more stabilising practices. In contrast

to the persistent positive effect of GDP growth across all specifications, the stand-

alone impact of inflation on bank stability is significantly negative. However, in the

last column we see that contemporary regulatory policies may attenuate the

destabilising effect of inflationary switching eventually its respective sign. More

liquidity in stock markets has persistent bearing on financial fragility across different

specifications, although there is indication at 1% significance level for the opposite.

That is, countries with high share of foreign-owned banks have more effective stock

markets that stabilize the market through alternative funding resources for the private

sector.

We also observe an insignificant association between asset size and bank soundness

persistently coupled with a limited effect of cost efficiency. Hence, we conclude that

financial fragility is not traced through economies of scale but also either through the

motivation of bank managers to take on more risk vis-à-vis potential government

subsidies and other „too-big-to-fail‟ policies. However, the inclusion of equity ratio

itself may eliminate the explanatory power of asset size to the extent that bank capital

is indicative of bank size. Thus, total equity over total assets maintains its positive

bearing at 1% significance level as it constitutes a buffer that insulates a bank from

low profitability or profit volatility commensurate with the degree of managers‟ risk

aversion. However, it loses power once we include denials on market entries and

private monitoring index as such policies focus directly on their protection and public

control and any change in capital stops short at explaining the variation of Z-

nominator. Moreover, portfolio diversification is explicitly significant in jeopardizing

bank stability on the grounds that banks on their way to accomplish economies of

scope engage in excessive risk-taking (Stiroh, 2004, Beck at al., 2009). The same

destabilizing pattern lies in the coefficient of liquid assets implying opportunity cost

that is potentially compensated by risky portfolio allocations.

We then construct the extreme bounds from every combination of regulatory

variables in an attempt to conclude for the persistence of bank competition and

concentration in sign and significance (table 25). Hence, the coefficient of Lerner

286

index takes significant values at 1% level ranging between -0.291 and 0.362 when we

include CAP, OFF, PRIV and ACT, FRACT in the model, respectively.

Concentration also switches the sign of its impact and proves to be fragile in its

relationship with bank risk-taking across all possible regressions. It takes values from

-0.625 when ACT, FRACT, PRIV come into play, to 1.053 with FOR and OFF

variables.

Table 25: Extreme bounds of model 8.1

Variables Bounds Coefficient Std. error t-value I-variables Significance

(1%)

Significance

(5%)

Lerner

low -0.291 0.095 -3.050 CAP, OFF,

PRIV Fragile (0) Fragile (0) base -0.155 0.108 1.440 -

high 0.362 0.138 2.620 ACT, FRACT

Lerner

low 0.103 0.125 0.830 CAP, OFF,

PRIV Fragile base 0.544 0.121 4.510 -

high 0.917 0.166 5.520 ACT, FRACT

Lerner^2

low -1.071 0.239 -4.490 CAP, OFF,

PRIV Robust

Infection points

[0.285,0.542]

base -1.909 0.244 -7.830 -

high -1.693 0.333 -5.090 ACT, FRACT

CONC

low -0.625 0.139 -4.500 ACT, FRACT,

PRIV Fragile (1) Fragile (1) base 0.537 0.092 2.640 -

high 1.053 0.089 11.840 FOR, OFF

Lerner

low 0.918 0.168 5.470 ACT, FRACT,

PRIV Robust

base 0.544 0.121 4.510 -

high 0.543 0.145 3.740 FOR, OFF

Lerner^2

low -1.792 0.338 -5.290 ACT, FRACT,

PRIV Robust

base -1.909 0.244 -7.730 -

high -1.398 0.277 -5.040 FOR, OFF

Following model 8.1, the table reports the extreme bounds of concentration and Lerner index with the respective standard

errors and t-values. The column „B-variables‟ indicates the specific information set that constructs the underlying bound,

and the last two underline the relationship between market structure and financial stability as fragile or robust at 1% and

5% significance level according to whether their sign and significance persistently remains stable over many

specifications. The rows in grey follow EBA utilizing only two and three-variable I-sets while in the two rows below

them, we include the quadratic term of Lerner index ad hoc for every extreme bound case in order to check the U-shaped

relationship as articulated by Martinez-Miera and Repullo (2010). Infection points refer to the level in Lerner distribution

where we see the switch in sign of the respective coefficient. The value „1‟ in the parentheses next to „fragile‟ indication

signifies the inclusion of one additional variable in order to make the parameter CONC significant. The zero value

indicates that in the baseline model the Lerner index comes insignificant.

287

In the next two arrows below the grey ones, we employ in the models that construct

the per se extreme bounds the quadratic term of market power and concentration to

verify whether it is a case of non-linear relationship between Lerner index and Z-

score. The linear effect of market power takes values between the range [0.103,

0.917] and the narrower one [0.918, 0.543] when we assess the extreme bounds of the

Lerner index and HHI, respectively. The former bounds indicate fragility in the

underlying relationship as opposed to the significance at 1% level in the latter. On the

other hand, the respective bounds for the quadratic term are [-1.071, -1.693] and [-

1.792, -1.398] while t-values show persistent explanatory power. We, thus, conclude

that the U-shaped relationship between competition and risk does exist and sign

change occurs at the infection points [0.285, 0.542].

In table 26, we replicate the procedure by allowing for the effect of country-level

factors not only to take place in levels but also in the slope of the competition-

stability nexus. The effect of Lerner index is quite unclear in the sensitivity analysis

of model 8.2. Market power drives to bank stability in concentrated markets where

we see effective systems of less restrictions on bank activities, more capital

requirements, less share of foreign-owned banks, limited intervention of supervisory

authorities on a bank‟s decisions and less private monitoring.

If we assess the stand-alone impact of capital requirements and private monitoring we

see their positive bearing on stability while market power is positively associated

with market instability. That may lie in individual policies failing to explain the

variation of the Z-score as long as they are seconded by supplementary initiatives. In

addition, the employment of interaction terms extracts some portion of explanatory

power of competition in the form of other conducive pass-through mechanisms but

their simultaneous use brings about a huge Lerner effect. The change in the sign of

private monitoring index depicts its negative correlation with official supervisory

power at almost 50% level.

288

Table 26: Regression output of model 8.2

Variables Baseline Sensitivity analysis

Lerner -0.010 1.937*** -1.539*** -0.083 0.669** 1.240*** -2.150*** 11.890***

(0.149) (0.272) (0.229) (0.169) (0.278) (0.297) (0.348) (1.521)

Concentration*L 0.604*** 0.469*** 0.463*** 0.867*** 0.607*** 0.741*** 0.592*** 0.500***

(0.083) (0.085) (0.081) (0.086) (0.084) (0.096) (0.083) (0.130)


Activity restrictions*L -0.171*** -0.204***

(0.024) (0.050)

Capital regulation*L 0.310*** 0.247***

(0.036) (0.065)

Foreign ownership*L -0.721*** -2.202***

(0.224) (0.673)

Official supervision*L -0.098*** -0.325***

(0.032) (0.060)

Fraction of denied

entries*L

-1.037 0.485

(0.691) (0.758)

Private monitor*L 0.343*** -1.375***

(0.051) (0.217)


GDPGR*L 2.311* 0.508 1.860 4.188*** 3.418** 15.217*** 3.845*** 23.044***

(1.246) (1.380) (1.304) (1.434) (1.318) (3.650) (1.412) (5.277)

Inflation*L -5.233*** -11.719*** -8.471*** -3.074* -3.811** -19.440*** -5.528*** 10.630

(1.575) (1.764) (1.659) (1.612) (1.586) (6.475) (1.650) (8.190)

Stock market turnover*L -0.122* -0.416*** -0.251*** 0.184*** -0.161** -1.727*** -0.185*** -0.512

289

(0.063) (0.069) (0.065) (0.062) (0.064) (0.214) (0.067) (0.339)


Q 0.000 -0.068*** -0.012 -0.014 -0.001 0.008 -0.003 -0.026**

(0.009) (0.008) (0.009) (0.009) (0.009) (0.011) (0.009) (0.011)

E/TA 0.316* -0.024 0.353** 0.382** 0.307* 0.196 0.251 0.512**

(0.168) (0.172) (0.163) (0.171) (0.168) (0.245) (0.175) (0.235)

Cost to income -0.013 -0.009 0.023 -0.033 -0.021 0.095 0.016 0.011

(0.055) (0.071) (0.061) (0.066) (0.062) (0.081) (0.063) (0.094)

TNINTINC/TI -0.367*** -0.909*** -0.393*** -0.432*** -0.376*** -0.275*** -0.393*** -0.432***

(0.081) (0.078) (0.080) (0.084) (0.081) (0.105) (0.081) 0.108

Liquidity -0.039** -0.082*** -0.045*** -0.039** -0.038** -0.057** -0.031* -0.073***

(0.016) (0.018) (0.017) (0.016) (0.016) (0.024) (0.016) (0.026)

TIME DUMMIES YES YES YES YES YES YES YES YES

OBSERVATIONS 9120 9120 9120 8683 9120 7415 9120 7066

BANKS 2408 2408 2408 2300 2408 1967 2408 1876

COUNTRIES 27 27 27 23 27 18 27 15

Random effects (RE) model with standard errors clustered at the bank level employing time dummies to capture time-varying random effects, as their significance is

evident after running standard F-tests. The first column reports the results of the baseline model, that is the regression of Z-score on banks specific and macroeconomic

variables totally devoid of any institutional controls; however, to attest the possibility of country-level factors affecting bank soundness in slopes through their

interaction with bank competition. We then include one at a time interacting controls of regulation, supervision and governance with the Lerner index, in order to verify

their stand-alone effect on bank stability as well as their „en masse‟ explanatory power in the last column. Z-score is logged for robustness in standard errors and all

bank-specific variables are lagged one period to avoid the possibility of reverse causality. Standard errors are in parentheses while asterisks ***, **, * denote the

significance level being at 1%, 5% and 10%, respectively.

290

We also verify considerable significance in the independent effects of interaction

terms between lagged competition and macroeconomic variables. We conclude that

GDP growth bolsters the stabilizing bearing of market power across all cases but

those of evaluating the individual effects of capital regulation and activity

restrictions. Furthermore, inflation pressures and stock market liquidity does harm to

the stability of banking industry. However, in the last column we see clearly the fact

that regulation policies tend to deprive stock market funding and price increases from

non-linear correlations with bank soundness.

Corroborating the results on table 24, income diversification and share of liquid

assets seem to undermine financial stability persistently at 1% level. Equity capital

maintains its positive impact although with some loss of its significance when

activity restrictions, denies entries and private monitoring index comes into play. On

the contrary, we end up with a significant of asset size when we include capital

regulation and all policies simultaneously indicating the possibility that managers

engage in risk-taking on expecting governmental too-big-to-fail policies for their own

„resurrection‟. We engage in constructing extreme bounds for Lerner and HHI index

through re-running the model for every possible combination of interaction terms. It

is interesting that bank competition bears no significance either at 5% level or 10%

level throughout 62 regressions.

In table 27, the extreme bounds of competition and concentration, emanated from two

to three-variable sets, range in-between the values of [-2.630, 7.073] and [0.209,

1.058], respectively. For the latter case, the model comprises ACT, FRACT (low

bound) and FOR, OFF and FRACT (high bound). In addition, the partial correlation

both at 5% and 1% significance level is fragile and it takes the replacement of one I-

variable to change sign or lose significance. Once we plug the squared term, the

linear effect across all extreme bounds of the Lerner index and its interaction with

HHI lies between the ranges [0.300. 0.557] and [0.929, 0.584]. The quadratic variable

gets values ranging from -1.262 to -1.909 with a persistent significance at 1% level

across all specifications.

291

Table 27: Extreme bounds of model 8.2

Variables Bounds Coefficient Std. error t-value B-variables Significance

(1%)

Significance

(5%)

Lerner

low -2.630 0.357 -7.370 CAP, PRIV

Fragile (0) Fragile (0) base -0.010 0.108 1.420 -

high 7.073 1.622 4.360 FOR, FRACT,

PRIV

Lerner

low 0.300 0.125 2.410 CAP, PRIV

Fragile base 0.542 0.121 4.490 -

high 0.557 0.215 2.590 FOR, FRACT,

PRIV

Lerner^2

low -1.262 0.242 -5.210 CAP, PRIV

Fragile base -1.909 0.244 -7.830 -

high -1.485 0.420 -3.540 FOR, FRACT,

PRIV

CONC*L

low 0.209 0.105 1.990 ACT, FRACT

Fragile (1) Robust base 0.604 0.092 2.680 -

high 1.058 0.110 9.580 FOR, OFF,

FRACT

Lerner

low 0.929 0.166 5.580 ACT, FRACT

Fragile base 0.542 0.121 4.490 -

high 0.584 0.207 2.820 FOR, OFF,

FRACT

Lerner^2

low -1.718 0.333 -5.160 ACT, FRACT

Robust base -1.909 0.244 -7.830 -

high -1.272 0.409 -3.110 FOR, OFF,

FRACT





specifications. The rows in grey follow EBA utilizing only two and three-variable I-sets while in the two rows below

them, we include the quadratic term of Lerner index ad hoc for every extreme bound case in order to check the U-shaped

relationship as articulated by Martinez-Miera and Repullo (2010). The value „1‟ in the parentheses next to „fragile‟

indication signifies the inclusion of one additional variable in order to make the parameter CONC*L significant. The zero

value indicates that in the baseline model the Lerner index comes insignificant.

292

8.7. Does the effect of market structure variables alter with the

interplay between regulation and ownership?

There exist some theories reckoning the correlation between ownership and risk

being variant in line with different national regulatory policies (John et al. 2008;

Laeven and Levine, 2009). We, therefore, embark on a sensitivity analysis including

interaction terms of regulatory variables with foreign ownership (F) to see whether

prudential policies are crucially determined by certain governance structures. We

additionally opt to keep the interactions between country-level factors and the Lerner

index extending the previous analysis one step further.

Financial stability (Z) = f [Competition (L), Concentration*L | institutional (I)*F


Τable 28 confirms no linear impact of market power on bank soundness and the

persistence of concentrated markets to less risk-taking behaviour. In the regressions

containing one I-variable each time, there is considerable stand-alone effect of

regulation and supervision on bank soundness that is materially contingent to the

degree of foreign ownership; however, exception to the trend constitutes the

interaction between foreign ownership and activity restrictions. In particular, foreign

ownership is positively associated with financial stability in markets where more a)

bank capital reserves are required, c) entry applications are granted and d)

requirements on bank credit information are utilized by rating agencies and auditing

authorities.

However, when it comes to the analysis of the whole set simultaneously, the evidence

ends up equivocal. First, the coefficient of concentration follows the same trend

irrespectively of the degree of co linearity amongst interaction terms. Moreover, we

find again insignificant in-level bearing of market power, whereas foreign-owned

financial institutions guarantee the stability of the financial system. That verifies the

fact that banks operating within a national industry feel safer when competition is

obstructed by pertinent policies and thus keep up a low risk profile. An indication of

10% significance withstands in the inclusion of I-variables altogether.

293

The whole set of I-variables enters significant, albeit differently when juxtaposed to

the output of previous columns. In fact, the coefficient of capital regulation turns

from -0.012 to 0.255 and the positive bearing of denied entries turns insignificant.

Both capital stringency and capital requirements, when coupled with other regulatory

policies, tend to make bank managers abstain from risk-taking behaviour so as

otherwise make up for potential utility loss; especially in cases of „sufficiently large‟

foreign owners (Laeven and Levine (2009). In the same vein, information of

information to the private sector demonstrate a negative sign at 10% significance

level as foreign-owned incumbent banks have ample space to exploit profit

opportunities amid conditions of ongoing institutional reforming.

As for the rest controls, GDP growth signifies its stabilising effect of market power

on the banking system across all specifications. On the other hand, inflation comes in

line with risk-taking conduct except the last column, in which the inclusion of the

whole interacting set of I-variables blurs the analysis. Moreover, the coefficient of

stock market activity is quite unclear as its contribution to the stabilising impetus of

market power is only evident when we allow for non-linearities between entry

restrictions and foreign ownership.

From the set of bank-specific variables, income diversification and asset liquidity

enters robustly negative corroborating the results of previous sections. Equity capital

also enjoys a positive impact, which persistently maintains its sign across all

specifications, as it constitutes one of the stabilising forces in banking through equity

capital and appears in the nominator of the Z-score.

294

Table 28: Sensitivity analysis of model 8.7

Random effects (RE) model with standard errors clustered at the bank level employing time dummies to capture

time-varying random effects, as their significance is evident after running standard F-tests. The first column reports

the results of the baseline model, that is the regression of Z-score on banks specific and macroeconomic variables

totally devoid of any institutional controls; however, to attest the possibility of country-level factors affecting bank

soundness in slopes through their interaction with bank competition. We then include one at a time interacting

controls of regulation, supervision and governance with foreign ownership, in order to verify whether, and to what

degree, the (significant) effect of market structure on bank stability does change. Z-score is logged for robustness in

standard errors and all bank-specific variables are lagged one period to avoid the possibility of reverse causality.

Standard errors are in parentheses while asterisks ***, **, * denote the significance level being at 1%, 5% and 10%,

respectively.

Variables Sensitivity analysis

Lerner -0.149 -0.153 -0.150 0.024 -0.152 0.087

(0.118) (0.117) (0.117) (0.155) (0.117) (0.152)

Concentration*L 0.858*** 0.854*** 0.844*** 0.939*** 0.850*** 0.494***

(0.089) (0.084) (0.084) (0.111) (0.084) (0.116)


Foreign ownership -0.086 0.070** -0.098*** -0.024*** 0.182*** 3.805***

(0.102) (0.032) (0.018) (0.004) (0.025) (1.308)

Activity restrictions*F 0.011 -0.247***

(0.014) (0.046)

Capital regulation*F -0.012** 0.255***

(0.005) (0.051)

Official supervision*F -0.010*** 0.041

(0.002) (0.033)

Fraction of denied entries*F 0.172*** -1.518

(0.017) (1.038)

Private monitor*F -0.034*** -0.594***

(0.004) (0.204)


GDPGR*L 2.362*** 2.406*** 2.393*** 7.320*** 2.380*** 12.110***

(0.811) (0.788) (0.787) (1.619) (0.785) (1.682)

Inflation*L -5.306*** -5.078*** -5.170*** -8.394*** -4.891*** -0.215

(0.982) (0.939) (0.940) (1.474) (0.933) (1.854)

Stock market turnover*L 0.015 0.015 0.016 -0.301*** 0.018 -0.489***

(0.029) (0.028) (0.028) (0.081) (0.028) (0.088)


Q -0.013 -0.013 -0.013 -0.008 -0.013 -0.004

(0.009) (0.009) (0.009) (0.011) (0.009) (0.011)

E/TA 0.437** 0.442** 0.444** 0.444* 0.448*** 0.586**

(0.172) (0.172) (0.172) (0.245) (0.171) (0.243)

Cost to income -0.035 -0.039 -0.037 0.019 -0.039 0.026

(0.059) (0.059) (0.059) (0.078) (0.059) (0.075)

TNINTINC/TI -0.449*** -0.450*** -0.448*** -0.361*** -0.451*** -0.385***

(0.083) (0.083) (0.083) (0.107) (0.083) (0.105)

Liquidity -0.042** -0.043*** -0.043** -0.075*** -0.043*** -0.087***

(0.016) (0.016) (0.016) (0.027) (0.016) (0.028)

TIME/SPEC DUMMIES YES YES YES YES YES YES

OBSERVATIONS 8683 8683 8683 7066 8683 7066

BANKS 2300 2300 2300 1876 2300 1876

Wald χ2 1135.02 1155.09 1195.41 2382.75 1205.44 5950.43

295

8.8. Are there other governance indicators that enter non-linearly?

The fact that supervisory authorities practice has come over an uphill path since

„80s coming from a quiet life into their obligation to fulfil the expectations of

„peckish‟ financial system. That probably stems from the process of financial

liberalisation and advancement in risk management techniques that incentivized for

more competitive pressure, leverage and the implicit option of bank managers at the

same time to surrender in safety net schemes (Crockett, 2003). Although we defined

different regulation and supervision policies, we need to see how market agents and

official authorities influence not only each other but also the competition-stability

nexus. That is because the way their operating role in the market is supposed to

govern their shared responsibility of systemic resilience and efficiency (Das et al.,

2003).

In table 29, we conduct sensitivity analysis on six dimensions of governance39

for

the enlarged (27-EU) group of European Union. We replicate the previous

methodology, namely the effect of different governance features on the interaction

between bank competition and financial stability, maintaining lagged values on all

bank-specific variables to take account of reverse causality problems. We

acknowledge the human capacity to produce estimation bias when quantifying

qualitative data on supervision and regulation and, therefore, employ 6 alternative

indicators that measure the institutional quality that any regulatory policy relies to

and thrives.

First, we include 1) voice and accountability - capturing the general perception of

the ability of a country‟s citizens to pick their government exercising their

constitutional rights of free expression, association, etc – 2) political stability -

measuring the probability of that national government of losing control or

overturning by violent or terrorist parties – 3) government effectiveness estimates

the quality of public and civil service, the degree of its independence from political

pressures, the quality of policy formulation and conducting as well as the

commitment for accomplishment – 4) rule of law engulfing the whole spectrum of

law enforcement by police and judicial system so as to prevent crime, violence and

breach of personal rights. It is also indicative of how much agents trust and comply

39

The data are based upon 30 indicators that have been constructed by the interaction of private

firms, research institutes, non-governmental organisations, think tanks and international

organisations. See their definition in the chapter appendix.

296

with rules – 5) control of corruption indicates the degree of corruption as being

expressed by the exploitation of public power to the best interests of elites and other

private parties – 6) credit depth of information sharing, that is the quality and

accessibility of credit information disseminated by public or private registries in

order to facilitate lending transactions. Hence we run the following model:

Financial stability (Z) = f [Competition (L), Concentration*L | Governance*L


Financial stability (Z) = f [Competition (L), Concentration*L | Governance*I


The Lerner index shows persistent significance with negative bearing on financial

stability unless we account for the exercising power of constitutional rights (table

29). On the contrary, market concentration maintains its significance and the

relative coefficient between the same bounds. We plug in the model one governance

indicator at a time as they are theoretically interconnected with respect to the level

of institutional development over a particular region and thus, high degree of

correlation would incur multicollinearity bias. All six governance indicators show

the expected positive non-linear effect on market power-stability nexus, indicating

that the stabilising effect of market power is stronger in countries where the

exercize of a) constitutional rights, b) political stability, c) efficient civil and public

service, d) law enforcement, e) control of corruption and f) accessibility of

qualitative information on credit, are more developed. It is, therefore, crucial when

monopolistic practices are present that the need for smooth and efficient functioning

of the state is by all means fulfilled in tandem with higher GDP growth rates and

limited burden of inflation and non-bank finance.

The positive correlation between market power and bank soundness strengthens in

times of high market demand, low stock market funding and inflationary pressures.

However, the latter is only significant when political stability and credit information

come in individually. In addition, high Z-scores tend to go in line with high levels

of equity capital as well as limited income diversification and liquidity in asset

structure.

297



Lerner -0.162 -0.178* -0.186* -0.238*** -0.215** -0.181*

(0.102) (0.103) (0.102) (0.102) (0.101) (0.099)

CONC*L 0.193** 0.501*** 0.065 -0.045 0.001 0.325***

(0.087) (0.082) (0.098) (0.103) (0.102) (0.084)

Other Governance indicators

Accountability*L 0.735***

(0.080)

Political stability*L 0.126***

(0.039)

Government effectiveness*L 0.275***

(0.032)

Rule of law*L 0.309***

(0.032)

Control of corruption*L 0.245***

(0.027)

Information depth*L 0.146***

(0.015)


GDPGR*L 3.524*** 1.773** 1.980*** 2.449*** 3.018*** 3.458**

(0.733) (0.735) (0.710) (0.716) (0.726) (0.724)

Inflation*L -1.183 -4.200*** -3.054*** -2.633*** -2.170** -3.459***

(0.959) (0.885) (0.881) (0.882) (0.892) (0.855)

Stock market turnover*L -0.170*** -0.072** -0.170*** -0.182*** -0.203*** -0.068**

(0.031) (0.029) (0.032) (0.033) (0.034) (0.028)


Q 0.003 0.005 0.003 0.004 0.005 0.005

(0.009) (0.009) (0.009) (0.009) (0.009) (0.009)

E/TA 0.438*** 0.390** 0.465*** 0.515*** 0.501*** 0.452***

(0.162) (0.165) (0.160) (0.159) (0.160) (0.171)

Cost to income -0.007 -0.018 -0.022 -0.036 -0.030 -0.029

(0.056) (0.055) (0.052) (0.050) (0.050) (0.054)

TNINTINC/TI -0.389*** -0.385*** -0.431*** -0.391*** -0.393*** -0.356***

(0.079) (0.080) (0.079) (0.078) (0.078) (0.083)

Liquidity -0.047*** -0.043** -0.046*** -0.052*** -0.051*** -0.039**

(0.016) (0.016) (0.016) (0.016) (0.016) (0.016)


OBSERVATIONS 9120 9120 9120 9120 9120 8811

BANKS 2408 2408 2408 2408 2408 2385

Wald χ2 1101.2 1075.84 1109.55 1110.46 1103.25 1044.47

COUNTRIES 1101.2 1075.84 1109.55 1110.46 1103.25 1044.47

Random effects (RE) model with standard errors clustered at the bank level employing time dummies to capture

time-varying random effects, as their significance is evident after running standard F-tests. The first column reports

the results of the baseline model, that is the regression of Z-score on banks specific and macroeconomic variables

totally devoid of any institutional controls; however, to attest the possibility of country-level factors affecting bank

soundness in slopes through their interaction with bank competition, we include one at a time interacting controls of

governance with the Lerner index, in order to verify their stand-alone effect on bank stability as well as their „en

masse‟ explanatory power in the last column. Z-score is logged for robustness in standard errors and all bank-

specific variables are lagged one period to avoid the possibility of reverse causality. Standard errors are in

parentheses while asterisks ***, **, * denote the significance level being at 1%, 5% and 10%, respectively.

298

In table 30, we report the regression output of the sensitivity analysis of the model

8.10. For space considerations, 30 distinct regressions are depicted for every single

regulation/supervision variable (I-variables, 1-5 columns) and combination of each

governance indicator itself alongside its interaction with every I-variable. The last

two columns report the extreme bounds of the Lerner index and the non-linear

relationship between the latter and market concentration. In order to keep track of the

origin of each bound, we report a number in parentheses, which corresponds to the

regression model that comprises the particular regulation/supervision variable. We

also highlight in grey the models that produce insignificant market power.

The evidence shows that activity restrictions, official intervention in the decision

making of bank managers are positively associated with bank risk-taking if and only

if institutional development lies in high levels. Put differently, in countries where the

public sector governance is well founded, restrictions on bank activities and

intervention of official authorities deprive banks to venture in other markets and

diversify their risk exposure through alternative asset allocations. In contrast,

regulation on capital reserves and private sector monitoring tend to stabilize the

financial system in markets where 1) agents are able to elect and monitor their

government, 2) sound policies are followed and 3) institutions that control economic

and social interactions are fully respected (Kaufmann et al., 2009). Moreover, the

degree of foreign ownership in a national banking market fosters financial stability

when corruption is persistent and requirements on credit information sufficiently

substandard.

It is also interesting that banks‟ pricing conduct has no linear effect on financial

stability, when the model allows for interactions between political stability and

activity restrictions, information depth, foreign ownership and official supervisory

power, while all non-linear effects of governance with private monitoring dry out the

impact of market power persistently. We, therefore, conclude that market power is

linearly related to bank risk-taking in cases of banks operating in institutionally

developed markets, considerable capital regulation and foreign-owned bank assets.

299



ACT CAP FOR OFF PRIV Extreme bounds

(1) (2) (3) (4) (5) Lerner CONC

Accountability 0.874*** -0.081 0.739*** 1.078*** -0.021

-0.250 (2) 0.246 (4) (0.081) (0.094) (0.090) (0.094) (0.106)

Accountability*I -0.061*** 0.107*** -0.007 -0.060*** 0.130***

-0.160 (4) 0.449 (3) (0.006) (0.007) (0.011) (0.008) (0.013)

Political stability 0.959*** -1.078*** 0.207*** 0.739*** -0.929***

-0.302 (2) 0.162 (1) (0.086) (0.076) (0.041) (0.106) (0.111)

Political stability*I -0.116*** 0.192*** -0.002 -0.078*** 0.176***

-0.170 (4) 0.781 (3) (0.010) (0.011) (0.015) (0.012) (0.017)

Government

effectiveness

0.525*** -0.353*** 0.259*** 0.720*** -0.500*** -0.253 (3) -0.201 (5)

(0.050) (0.051) (0.035) (0.057) (0.068)

Government

effectiveness*I

-0.041*** 0.095*** -0.009 -0.064*** 0.130*** -0.192 (1) 0.328 (3)

(0.006) (0.006) (0.020) (0.007) (0.011)

Rule of law 0.557*** -0.383*** 0.274*** 0.768*** -0.446***

-0.313 (3) -0.419 (5) (0.050) (0.053) (0.035) (0.057) (0.066)

Rule of law*I -0.044*** 0.104*** -0.022 -0.065*** 0.133***

-0.214 (1) 0.236 (3) (0.006) (0.007) (0.024) (0.007) (0.010)

Control of corruption 0.496*** -0.375*** 0.216*** 0.660*** -0.472***

-0.261 (3) -0.343 (5) (0.049) (0.046) (0.031) (0.054) (0.061)

Control of corruption*I -0.040*** 0.100*** -0.049* -0.058*** 0.123***

-0.194 (1) 0.288 (3) (0.006) (0.006) (0.029) (0.007) (0.010)

Information depth 0.181*** -0.102*** 0.165*** 0.275*** -0.045*

-0.273 (2) -0.297 (2) (0.015) (0.018) (0.016) (0.020) (0.024)

Information depth*I -0.014*** 0.033*** -0.184*** -0.021*** 0.028***

-0.220 (1) 0.413 (4) (0.002) (0.002) (0.018) (0.002) (0.003)

COUNTRY-LEVEL

CONTROLS YES YES YES YES YES

BANK-SPECIFIC

CONTROLS YES YES YES YES YES

TIME/SPEC

DUMMIES YES YES YES YES YES

The table reports the evidence of 30 regressions in a condensed output along with the extreme bounds of the coefficients of

Lerner index and CONC*L. With respect to each regulation/supervision proxy (I-variable, columns 1-5), every regression

model comprises the linear effect of one-at-a-time governance indicator and its interaction with the respective I-variable.

The cells in grey indicate the cases where market power fails to demonstrate significance, while the extreme bounds are

constructed by the significant only coefficients of market power and CONC*L. Hence, we pick up the lowest and highest

bound among the regressions that appear horizontally in the table, and indicate in parentheses the number, which

corresponds to the regulation/supervision factor that appears in the particular model of the bound. We also employ country-

level variables interacting with the Lerner index, lagged bank-specific variables and the standard time and specialisation

dummies. Standard errors are in parentheses while asterisks ***, **, * denote the significance level at 1%, 5% and 10%,

respectively.

300

8.9. Are the results robust to alternative measures of risk?

Last, we test alternative measures of market stability and to what degree the fully-

fledged model (8.10) can predict their variation. We, therefore, employ all the

components that build up the Z-score along with the logged nominator, the Z-score

with the denominator being averaged over the whole period and credit risk as proxies

by loan loss charges over average gross loans.

Risk = f [Market structure variables (M) | institutional (I) |macroeconomic and bank-

specific factors (C)] + ε (8.10)

Market power determines at 1% significance level the variation of bank profitability,

the logged nominator of Z-score and loan losses (table 31). It loses significance in

models of average equity capital and overall risk (Z-score) as dependent variables,

whereas market concentration turns out significant in predicting stand-alone

constituents of Z-score except for the case of bank profitability. We also see that the

volatility of ROA itself cannot be determined by market structure variables thereby

showing the lowest fit of the model.

Activity restrictions enter positively up at 1% significance level across all models and

seem to contribute to bank soundness through not only profitability and equity capital

but also through fewer loan losses. On the other hand, they seem to exacerbate the

fragility of the banking system by not allowing banks to diversify their risk and take

advantage of OBS activities. Thus, they are bound to experience considerable income

variation extend although the theoretical premise of granting more loans either to

marginal applicants or to customers-defaulters does not seem to hold. Furthermore,

more capital requirements undermine the ability of banks to gain profits by

experiencing considerable loan losses. Likewise, the share of foreign owned assets in

a banking system does harm to stability to the extent that is associated with lower

profitability that stems from losses on loans.

Official supervision demonstrates a negative impact on bank soundness, through

intensifying credit risk and secondarily through equity capital. The latter is expected

since stricter authorities tend to intervene in bank decisions through suspending

dividends, superseding the rights of shareholders thereby such burden on bank

301

management may provoke corruption in lending activities and other centrifugal

forces of risk-taking. Restrictions on market entries lead to bank soundness by

lowering profits volatility as banks are released from potential competitive pressures

and therefore managers are engaged in less risk-taking activities when there exist

enough profit opportunities to exploit within a monopolistic environment. Moreover,

the fact that financial institutions are allowed to be subject to the scrutiny of private

monitoring has a negative bearing on financial stability, as it is the product of the

opposite forces of greater Z-score nominator and even higher credit risk.

We also observe GDP growth to predict significantly bank stability and profitability

along with less credit risk and variation in profits. In contrast, inflationary pressures

drive to diminishing asset returns and high volatility in banks‟ profits. Furthermore,

stock market activity tends to have no impetus on bank soundness although it seems

to predict asset returns mitigate losses occurring in lending activities.

As for the bank-level controls, financial fragility is attributed to asset size, which

tends to diminish every single component of Z-score at 1% significance level. Equity

capital seems to stabilize banks particularly through the Z-score nominator that

outpace its adverse repercussions on profit volatility. Cost inefficiency is associated

with low profitability and greater losses without the per se impact being sufficient to

determine bank stability. Last, income diversification and asset liquidity are

significant determinants of ROA, equity capital and sore but once we combine them

to construct the Z-score as dependent variable its effect fades out.

302

Table 31: Sensitivity analysis with alternative measures of stability

Random effects (RE) model with standard errors clustered at the bank level employing time dummies to capture time-

varying random effects, as their significance is evident after running standard F-tests. The first column reports the results

of the baseline model, that is the regression of alternative measures of risk on banks specific and macroeconomic

variables totally devoid of any institutional controls. We then include one at a time controls of regulation, supervision and

governance, in order to verify their stand-alone effect on bank stability as well as their „en masse‟ explanatory power in

the last column. All bank-specific variables are lagged one period to avoid the possibility of reverse causality. Standard

errors are in parentheses while asterisks ***, **, * denote the significance level being at 1%, 5% and 10%, respectively.

All dependent variables are winsorized at 1% of their distribution to mitigate the effect of outliers.

Variables ROA E/TA σROA Lon(ROA+

E/TA) Loan losses Z

Lerner 0.014*** 0.015 -0.005 0.409*** 0.015*** -0.901

(0.004) (0.019) (0.005) (0.121) (0.005) (5.456)

Concentration 0.002 0.022* 0.006** 0.205* 0.006* 17.531***

(0.009) (0.012) (0.003) (0.107) (0.003) (6.189)


Activity restrictions 0.001** 0.007*** 0.0005* 0.078*** -0.001* -12.344***

(0.000) (0.002) (0.000) (0.011) (0.000) (2.075)

Capital regulation -0.001 -0.0001 -0.0002 -0.030*** 0.001** 4.450

(0.000) (0.003) (0.000) (0.009) (0.000) (3.088)

Foreign ownership -0.0002** 0.0003 0.0001 0.002 0.001** -35.368

(0.000) (0.001) (0.000) (0.006) (0.000) (22.468)

Official supervision 0.0001 -0.001 0.0001 0.003 -0.001* -9.045***

(0.000) (0.002) (0.000) (0.011) (0.000) (3.045)

Fraction of denied

entries

0.002 0.001 -0.009* 0.049 -0.001** 49.659***

(0.004) (0.018) (0.005) (0.140) (0.003) (16.865)

Private monitor 0.001 0.004 0.001 0.064* 0.003*** -19.250***

(0.001) (0.008) (0.001) (0.038) (0.001) (7.127)


GDPGR 0.062** -0.116 -0.052** 1.489** -0.073*** 144.442***

(0.024) (0.120) (0.022) (0.695) (0.025) (39.303)

Inflation -0.093** -0.080 0.083** -1.430 0.033 45.916

(0.123) (0.109) (0.037) (1.486) (0.032) (44.175)

Stock market turnover 0.007*** 0.001 -0.001 0.351*** -0.014*** 4.281

(0.002) (0.008) (0.001) (0.044) (0.002) (3.203)


Q -0.001*** -0.017*** -0.0004** -0.071*** -0.000 -3.240

(0.000) (0.002) (0.000) (0.010) (0.000) (2.133)

E/TA 0.012 0.013* 3.307*** 0.003 62.171

(0.009) (0.008) (0.723) (0.007) (40.159)

Cost to income -0.004** 0.010 -0.003 0.069 0.007** 0.834

(0.002) (0.011) (0.003) (0.068) (0.003) (2.637)

TNINTINC/TI 0.010*** 0.050*** 0.008*** 0.258*** 0.001 0.201

(0.003) (0.018) (0.003) (0.091) (0.004) (7.427)

Liquidity 0.004** 0.031*** 0.003*** -0.049 0.002 0.769

(0.002) (0.007) (0.003) (0.043) (0.001) (2.506)


OBSERVATIONS 7186 7188 7125 7175 6944 7186

BANKS 1888 1888 1877 1888 1823 1888

Wald χ2 775.77 622.70 269.29 3715.29 340.62 436.69

303

8.10. How market structure variables behave per specialization

group?

We opt to run model 1 for every bank group productive specialisation in order to see

how the results of the whole – albeit extremely heterogeneous - sample is determined.

To this end, if we employ the institutional variables set out in the previous analysis we

are bound to end up with incomparable regression output since data are not available

in all countries, in which banks are headquartering. Hence, we utilize the governance

indicator of World bank that proxies for general institutional quality instead of

encompassing distinct regulatory and supervisory policies. In an attempt to ease the

sensitivity analysis, we run model 8.1 for four large groups of bank specialisation; that

is, commercial, cooperative, savings and „other‟ banks. The latter further comprises all

the rest financial system40

that cannot construct a different group of appropriate size.

In table 32, we observe a negative bearing of market power on financial stability when

assessing the results of the full sample, which is further decomposed into the

respective destabilising effect in cooperative and savings banks as well as the opposite

tendency in „other‟ banks. Concentration maintains no explanatory power when the

analysis is conducted for specific bank types, whereas the full sample concludes on the

stabilising significance of HHI at 5% level. Moreover, regulatory quality enters

negative (positive) in commercial (cooperative, savings) banks resulting in an

increasing impact on bank soundness.

Other effects of the business cycle in terms of GDP growth boil down to a procyclical

tendency for the whole sample, with an indication at 10% significance level for the

cooperative banking group. Along the same lines, stock market turnover demonstrates

a negative effect only in the aggregate European banking sector. In contrast, the price

effect of economic development tends to diminish stability across all bank types but

the case of cooperative banks.

40

See the data section (8.5), which stipulates the structure of the sample.

304

Table 32: Model 8.1 output per bank type

Variables Commercial Cooperative Savings Other Full sample

Lerner 0.116 -1.032** -0.819* 0.242* -0.172*

(0.120) (0.506) (0.444) (0.142) (0.103)

Concentration -0.133 0.447 -0.053 0.206 0.191**

(0.119) (0.376) (0.392) (0.169) (0.091)


Regulatory quality -0.137* 0.750*** 0.537** -0.077 0.389***

(0.070) (0.201) (0.210) (0.097) (0.049)


GDPGR -0.791 6.243* 1.301 -1.721 2.790***

(0.767) (3.502) (2.563) (1.750) (0.734)

Inflation -6.366*** 2.417 -6.774** -4.666** -3.622***

(1.043) (3.982) (3.356) (2.157) (0.876)

Stock market turnover 0.026 0.036 -0.461 -0.000 -0.164***

(0.039) (0.137) (0.177) (0.064) (0.033)


Q -0.053*** 0.017 0.070** -0.025* 0.005

(0.011) (0.019) (0.030) (0.013) (0.009)

E/TA 0.798*** 2.842*** -5.584*** 0.654*** 0.469***

(0.258) (1.039) (0.987) (0.177) (0.162)

Cost to income -0.095 0.037 0.401** -0.038 -0.011

(0.076) (0.364) (0.169) (0.060) (0.054)

TNINTINC/TI -0.674*** 1.374*** -0.235 -0.846*** -0.389***

(0.126) (0.328) (0.519) (0.098) (0.079)

Liquidity -0.038 -0.321** -0.507* -0.035** -0.050***

(0.029) (0.126) (0.303) (0.015) (0.016)

TIME DUMMIES YES YES YES YES YES

SPECIALISATION

DUMMIES YES YES YES YES YES

OBSERVATIONS 1866 3887 2089 1278 9120

BANKS 509 1016 466 417 2408

COUNTRIES 27 13 16 23 27

The table reports evidence of model 8.1 replicating regressions for each productive specialisation group.

Because some of the I-variables are unavailable in countries with certain bak types, we alternatively

employ the indicator of regulatory quality as constructed by the World Bank, which gives an overall

picture of the institutional framework. Standard errors are in parentheses while asterisks ***, **, * denote

the significance level being at 1%, 5% and 10%, respectively.

Last, asset size follows a stabilising pattern in savings banks as opposed to the

opposite force in commercial and „other‟ banks; however, the underlying effect stands

305

negligible in the last column. In addition, the robust relationship between lagged

equity capital and Z-score across all previous results is addressed in the whole sample

but also in commercial, cooperative and „other‟ banks. In contrast, savings banks

appear to emit negative equity signals on their solvency levels. Additionally, cost

inefficiency has negative albeit insignificant impact on Z-score in line of the argument

that banks take on more risks so as to enhance their operating performance. On the

contrary, savings banks enjoy an opposite tendency probably due to their strategic

orientation to internalize potential costs.

Income diversification through OBS activities tends to plummet Z-scores in all bank

categories except for the savings and cooperative banks, in which risk diversification

takes place and insulates them from potential shocks. Therewithal, the partial

correlation between asset liquidity and stability comes negative over the majority of

financial institutions due to incurring opportunity costs. Exception to this rule

constitutes the case of commercial banks, the operation of which is more vulnerable to

a range of market risks.

8.11. Other robustness checks

In table 33, we re-estimate the model that gives out the extreme bounds of competition

and concentration taking account of all the possible combinations from four to five-

variable information sets. The low bound of the Lerner index is getting smaller to -

0.319 at 1% significance level with the extra addition of PRIV. However, the upper

bound stays put even after utilizing more than three-variable sets. The Herfindahl-

Hirschman index spreads the bound margin [-0.909, 1.053] by altering only the lower

bound when employing ACT, CAP, FOR, FRACT and PRIV. Thus, we conclude

about the fragility of market structure that needs only one replacement of I-variable set

in order to see an opposite sign or significance loss in the underlying partial

correlations.

When it comes to account for non-linearities, the U-shaped relationship is not traced in

the lower bound of market power whereas in that of HHI there exists no linear

association. In the other two cases (base and high bound), competition tends to

destabilize the banking market at the point where the Lerner index takes values 0.285

306

and 0.542, respectively. Apart from that, the base and high values of market structure

variables are similar to those in table 25, and the lower bound expands further in both

cases. On those grounds, we end up with fragile relationship between Z-score and

market power when it ranges from -0.179 to 0.917; however, the non-linear impact

becomes robust across the extreme bounds of market concentration [-1.005, -1.398] in

contradiction with insignificant bearing in the lower bound of market power.

Table 33: Further EBA of model 8.1

Variables Bounds Coefficient Std.

error t-value I-variables

Significance

(1%)

Significance

(5%)

Lerner

low -0.319 0.129 -2.48 CAP, OFF, FRACT,

PRIV Fragile(0) Fragile(0) base -0.155 0.108 1.44 -

high 0.362 0.138 2.62 ACT, FRACT

Lerner

low -0.179 0.165 -1.08 CAP, OFF, FRACT,

PRIV Fragile base 0.544 0.121 4.51 -

high 0.917 0.166 5.52 ACT, FRACT

Lerner^2

low -0.412 0.331 -1.24 CAP, OFF, FRACT,

PRIV Fragile

Infection points

[0.285, 0.542)

base -1.909 0.244 -7.83 -

high -1.693 0.333 -5.09 ACT, FRACT

CONC

low -0.909 0.160 -5.69 ACT, CAP, FOR,

FRACT, PRIV Fragile(1) Fragile(1)

base 0.537 0.092 2.64 -

high 1.053 0.089 11.840 FOR, OFF

Lerner

low 0.188 0.208 0.90 ACT, CAP, FOR,

FRACT, PRIV Fragile

base 0.544 0.121 4.51 -

high 0.543 0.145 3.74 FOR, OFF

Lerner^2

low -1.005 0.404 -2.49 ACT, CAP, FOR,

FRACT, PRIV Robust base -1.909 0.244 -7.83 -

high -1.398 0.145 3.74 FOR, OFF





specifications. The rows in grey follow EBA utilizing four and five-variable I-sets for robustness-checking reasons while

in the two rows below them, we include the quadratic term of Lerner index ad hoc for every extreme bound case in order

to check the U-shaped relationship as articulated by Martinez-Miera and Repullo (2010). Infection points refer to the

level in Lerner distribution where we see the switch in sign of the respective coefficient. The value „1‟ in the parentheses

next to „fragile‟ indication signifies the inclusion of one additional variable in order to make the parameter CONC

significant. The zero value indicates that in the baseline model the Lerner index comes insignificant.

307

We next conduct EBA from a pool of interaction terms that are affected by the

inclusion of four to five B-variables (table 34). We observe a different pattern in this

case as we end up with the same lower bound for Lerner index and the same upper

bound for HHI. In particular, the former ranges between [-2.630, 12.173] whereas

concentration withholds significance between the bounds [0.148, 1.058], with the use

of ACT, FOR, OFF, FRACT, PRIV (upper bound - Lerner) and CAP, FRACT (low

bound - HHI).

In contrast with table 27, in which all partial relationships are fragile and it takes only

the replacement of one variable to observe a switch in sign or loss in significance, at

least for the case of assessing two to three I-variables as conditioning information sets

we come up with a robust non-linear relationship between concentration and bank

soundness. In other words, market concentration affects the slope of market power on

risk-taking as long as we utilize different combinations of institutional variables.

When it comes to expand the size of I-variables set, the CONC*L interaction seizes to

demonstrate such robustness in a single case; that is, when the regression model

comprises ACT, CAP, FRACT and PRIV.

In this case, the argument of Martinez-Miera and Repullo (2010) about the risk-

shifting and margin effects that makes the relationship between competition and bank

failure risk switch sign still holds water across the whole range of bounds. In

particular, the purely linear bearing of Lerner index lies in-between 0.300 and 0.633,

which are very close to the results of table 6. On the contrary, the squared Lerner

index enters negative ranging from -1.084 and -1.272 for the CONC*L bounds and

even expanding in absolute levels between -1.262 and -1.400.

308

Table 34: Further EBA of model 8.2

Variables Bounds Coefficient Std. error t-value B-variables Significance

(1%)

Significance

(5%)

Lerner

low -2.630 0.357 -7.37 CAP, PRIV

Fragile(0) Fragile(0) base -0.010 0.108 1.42 -

high 12.173 1.650 7.38 ACT, FOR, OFF,

FRACT, PRIV

Lerner

low 0.300 0.125 2.41 CAP, PRIV

Fragile base 0.542 0.121 4.49 -

high 0.633 0.209 3.04 ACT, FOR, OFF,

FRACT, PRIV

Lerner^2

low -1.262 0.242 -5.21 CAP, PRIV

Fragile base -1.909 0.244 -7.83 -

high -1.400 0.416 -3.37 ACT, FOR, OFF,

FRACT, PRIV

CONC*L

low 0.148 0.106 1.40 ACT, CAP, FRACT

Fragile(1) Fragile(1) base 0.604 0.092 2.68 -

high 1.058 0.110 9.58 FOR, OFF, FRACT

Lerner

low 0.380 0.167 2.28 ACT, CAP, FRACT

Fragile base 0.542 0.121 4.49 -

high 0.584 0.207 2.82 FOR, OFF, FRACT

Lerner^2

low -1.084 0.328 -3.31 ACT, CAP, FRACT

Robust base -1.909 0.244 -7.83 -

high -1.272 0.409 -3.11 FOR, OFF, FRACT





specifications. The rows in grey follow EBA utilizing four and five-variable I-sets for robustness-checking reasons while

in the two rows below them, we include the quadratic term of Lerner index ad hoc for every extreme bound case in order

to check the U-shaped relationship as articulated by Martinez-Miera and Repullo (2010). The value „1‟ in the parentheses

next to „fragile‟ indication signifies the inclusion of one additional variable in order to make the parameter CONC*L

significant. The zero value indicates that in the baseline model the Lerner index comes insignificant.

8.12. Conclusion

This study addresses whether the relationship between market structure and financial

stability is significant under different specifications for the European Union since the

advent of the single currency. Thus, it extends the bulk of empirical analysis to draw

remarks on the validity of the underlying relationship subject to different information

sets out of a pool of various bank-specific, macroeconomic and institutional factors

that have been already addressed in the literature.

309

EBA methodology systematically investigates the partial correlations of market

concentration and bank competition with bank soundness employing a) linear effects

of regulatory and supervisory variables, b) non-linear effects I-variables with

competition, c) interactions of I-variables with bank market power, d) foreign

ownership, e) non-linearities between governance indicators with competition, f)

different dependent variables that encompass different aspects of bank risk and g)

sensitivity analysis per bank specialization.

We found a fragile relationship between the Lerner index and Z-score both at 5% and

1% significance level, when utilising the effect of I-variables in levels (model 8.1).

When interactions come into play, the Lerner index follows the same pattern taking

account of different-sized information sets. However, once we run model 8.1 and 8.2

with the inclusion of quadratic term, the underlying fragility stems particularly from

the pure linear impact of market power, as the squared term is persistently significant

across all extreme bounds. We also trace this U-shaped relationship to take place in

cooperative and savings banks and the whole sample of the European financial system

in general. Market power seems to empower bank solvency up to the level of 28.5%,

where monopolistic behaviour have devastating repercussions.

Concentrated markets are highly correlated with bank soundness across any

specification and robustness check. Such impact is traced in the tendency of

concentration to predict significantly equity capital at 10% level, respectively.

However, fragility emanates from bank managers who engage in risk-taking in lending

transactions, and model 1, which investigates its impact on Z-score in levels. Once we

analyze whether HHI affects the relationship between market power and Z-score in

slopes, we end up with robust bearing in the standard EBA (table 27) whereas in table

34 it becomes fragile in only one case out of 20 regressions of different I-variable

information sets.

Besides, we come up with collateral issues that have been appealing in many studies in

terms of the policy implications they put forward. In general, the majority of

institutional variables is capable of affecting bank stability individually. When we

assess their significance, more financial stability is traced in markets where we

observe more capital regulation and requirements of information dissemination

whereas restrictions on non-traditional activities as well as supervisory intervention

310

and foreign ownership tend to destabilize the financial system. Besides, when they are

interacting with market power, we identify the same tendency amid conditions of

monopolistic competition.

Going one step beyond the non-linear effects of country-level variables, we assess

how market variables affect Z-score when I-variables interact with foreign ownership.

In short, banking markets, within which foreign-owned banks are subject to 1) more

capital regulation, 2) official supervisory intervention, 3) freer market entries and 4)

much information requirements for the private sector, experience relatively higher

degree of risk-taking. Market power has no impact in bank risk-taking while market

concentration determines the market power-stability nexus at considerable level.

The same pattern is evident when utilising governance indicator to be interacting with

bank-level competition. The Lerner index maintains its significance in all cases except

that of plugging voice and accountability in the model as opposed to the non-linear

explanatory power of HHI taking place when market power underperforms. The price

mark-up over marginal cost has a considerable bearing on bank soundness the more

the state fulfils individual rights and implements policies free of potential political

interventions. In particular, bank soundness comes in tandem with low activity

restriction and official intervention as well as higher capital requirements and private

monitoring within a context of well-developed and respectful institutions.

Once we employ different aspects of bank risk, some interesting results pop up. First,

ROA and credit risk are sufficiently predicted by market power while market

concentration is associated with more capital buffers and losses on lending activities.

Amid conditions of ongoing regulatory policies, cooperative and savings banks are

more vulnerable to the bearing of market power as opposed to the operational strength

of „other‟ banks.

311

Chapter 8 Appendix

Information on the construction of Regulatory and Supervisory Variables

Variable(s) Definition Source and Quantification World Bank Guide Questions

(a) Securities

Activities

The extent to which banks

may engage in

underwriting, brokering

and dealing in securities,

and all aspects of the

mutual fund industry.

OCC and WBG 4.1 (higher values, more

restrictive)

4.1 What is the level of regulatory

restrictiveness for bank participation in

securities activities (the ability of banks

to engage in the business of securities

underwriting, brokering, dealing, and all

aspects of the mutual fund industry)?

Unrestricted = 1 = full range of activities can be

conducted directly in the bank; Permitted = 2 = full

range of activities can be conducted, but some or

all must be conducted in subsidiaries; Restricted =

3 = less than full range of activities can be

conducted in the bank or subsidiaries; and

Prohibited = 4 = the activity cannot be conducted in

either the bank or subsidiaries.

(b) Insurance

Activities


may engage in insurance

underwriting and selling.


restrictive)



insurance activities (the ability of banks

to engage in insurance underwriting and

selling)?









(c) Real Estate

Activities


may engage in real estate

investment, development

and management.


restrictive)



real estate activities (the ability of banks

to engage in real estate investment,

development, and management)?









312

(d) Bank

Owning

Nonfinancial

Firms


may own and control

nonfinancial firms.


restrictive)


restrictiveness for bank ownership of

nonfinancial firms? Unrestricted = 1 = a bank may own 100 percent of

the equity in any nonfinancial firm; Permitted = 2 =

a bank may own 100 percent of the equity of a

nonfinancial firm, but ownership is limited based

on a bank's equity capital; Restricted = 3 = a bank

can only acquire less than 100 percent of the equity

in a nonfinancial firm; and Prohibited = 4 = a bank

may not acquire any equity investment in a

nonfinancial firm.

Activity

restrictions

(overall index)

The sum of (a), (b), (c) and

(d).

(a)+(b)+(c)+(d)

Higher values indicate more actovity restrictions.

Fraction of

Entry

Applications

Denied

The degree to which

applications to enter

banking are denied.

WBG (1.9.1 + 1.10.1) / (1.9 + 1.10) (pure number)

1.9 In the past five years, how many

applications for commercial banking

licenses have been received from

domestic entities?

1.9.1 How many of those applications

have been denied?

1.10 In the past five years, how many

applications for commercial banking

licenses have been received from

foreign entities?

1.10.1 How many of those applications

have been denied?

(a) Overall

Capital

Stringency

Whether the capital

requirement reflects certain

risk elements and deducts

certain market value losses

from capital before

minimum capital adequacy

is determined.

WBG 3.1.1 + 3.3 + 3.9.1 + 3.9.2 + 3.9.3 + (1 if 3.6

< 0.75)

3.1.1 Is the minimum capital-asset ratio

requirement risk weighted in line with

the Basel guidelines? Yes / No

Yes = 1; No = 0 3.3 Does the minimum ratio vary as a

function of market risk? Yes / No

Higher values indicating greater stringency.

3.9.1 Are market value of loan losses

not realized in accounting books

deducted? Yes / No

3.9.2 Are unrealized losses in securities

portfolios deducted? Yes / No 3.9.3 Are

unrealized foreign exchange losses

deducted? Yes / No

313

3.6 What fraction of revaluation gains is

allowed as part of capital?

(b) Initial

Capital

Stringency

Whether certain funds may

be used to initially

capitalize a bank and

whether they are officially

verified.

WBG 1.5: Yes = 1, No = 0: WBG 1.6&1.7: Yes=0,

No=1.

1.5 Are the sources of funds to be used

as capital verified by the

regulatory/supervisory authorities? Yes /

No

Higher values indicating greater stringency.

1.6 Can the initial disbursement or

subsequent injections of capital be done

with assets other than cash or

government securities? Yes / No

1.7 Can initial disbursement of capital

be done with borrowed funds? Yes / No

Capital

Regulatory

Index

The sum of (a) and (b).

(a) + (b)

Higher values indicate greater stringency.

Official

Supervisory

Power

Whether the supervisory

authorities have the

authority to take specific

actions to prevent and

correct problems.

WBG 5.5 + 5.6 + 5.7 + 6.1 + 10.4 + 111.2 + 11.3.1

+ 11.3.2 +11.3.3 + 11.6 + 11.7 + 11.9.1 + 11.9.2 +

11.9.3

5.5 Does the supervisory agency have

the right to meet with external auditors

to discuss their report without the

approval of the bank? Yes / No

Yes = 1; No = 0

5.6 Are auditors required by law to

communicate directly to the supervisory

agency any presumed involvement of

bank directors or senior managers in

elicit activities, fraud, or insider abuse?

Yes / No

Sum of these assigned values, with higher values

indicating greater power.

5.7 Can supervisors take legal action

against external auditors for negligence?

Yes / No

6.1 Can the supervisory authority force a

bank to change its internal

organizational structure? Yes / No

10.4 Are off-balance sheet items

disclosed to supervisors? Yes / No

11.2 Can the supervisory agency order

the bank's directors or management to

constitute provisions to cover actual or

potential losses? Yes / No

11.3 Can the supervisory agency

suspend the directors' decision to

distribute: 11.3.1 Dividends? Yes / No

11.3.2 Bonuses? Yes / No

11.3.3 Management fees? Yes / No

11.6 Can the supervisory agency legally

314

declare-such that this declaration

supersedes the rights of bank

shareholders-that a bank is insolvent?

Yes / No

11.7 Does the Banking Law give

authority to the supervisory agency to

intervene that is, suspend some or all

ownership rights-a problem bank? Yes /

No 11.9 Regarding bank restructuring

and reorganization, can the supervisory

agency or any other government agency

do the following: ? Yes / No

11.9.1 Supersede shareholder rights?

Yes / No

11.9.2 Remove and replace

management? Yes / No

11.9.3 Remove and replace directors?

Yes / No

(a) Certified

Audit Required

Whether there is a

compulsory external audit

by a licensed or certified

auditor.

WBG 5.1 * 5.3 (Yes = 1; No = 0)

5.1 Is an external audit a compulsory

obligation for banks? Yes / No 5.3 Are

auditors licensed or certified? Yes / No

(b) Percent of

10 Biggest

Banks Rated by

International

Rating

Agencies

The percentage of the top

ten banks that are rated by

international credit rating

agencies.

WBG 10.7.1 (percent)

10.7.1 What percent of the top ten banks

are rated by international credit rating

agencies (e.g., Moody's, Standard and

Poor)?

(c) No Explicit

Deposit

Insurance

Scheme

Whether there is an explicit

deposit insurance scheme

and, if not, whether

depositors were fully

compensated the last time a

bank failed.

WBG 1 if 8.1=0 and 8.4=0; 0 otherwise 8.1 Is there an explicit deposit insurance

protection system? Yes / No

Yes =1; No =0

8.4 Were depositors wholly

compensated (to the extent of legal

protection) the last time a bank failed?

Yes / No

Higher values indicate more private supervision

(d) Bank

Accounting

Whether the income

statement includes accrued

or unpaid interest or

principal on nonperforming

loans and whether banks

are required to produce

consolidated financial

statements.

WBG (10.1.1 - 1)*(-1) + 10.3 + 10.6

10.1.1 Does accrued, though unpaid

interest/principal enter the income

statement while the loan is still non-

performing?

Yes=1; No=0 10.3 Are financial institutions required

to produce consolidated accounts

covering all bank and any non-bank

financial subsidiaries? 10.6 Are bank

directors legally liable if information

disclosed is erroneous or misleading?

Sum of assigned values, with higher values

indicating more informative bank accounts.

315

Private

Monitoring

Index

Whether (a) occurs, (b)

equals 100%, (c) occurs,

(d) occurs, off-balance

sheet items are disclosed to

the public, banks must

disclose risk management

procedures to the public,

and subordinated debt is

allowable (required) as a

part of regulatory capital.

WBG: (a) + [1 if (b) equals 100% ; 0 otherwise] +

(c) + (d) + 10.4.1 + 10.5 + 3.5

10.4.1 Are off-balance sheet items

disclosed to the public? Yes / No

Yes = 1; No = 0

10.5 Must banks disclose their risk

management procedures to the public?

Yes / No

Higher values indicating more private supervision. 3.5 Is subordinated debt allowable

(required) as part of capital? Yes / No

Foreign-Owned

Banks

The extent to which the

banking system's assets are

foreign owned.

WBG 3.8 (percent)

3.8 What fraction of the banking

system's assets is in banks that are 50%

or more foreign owned?

Definition of Governance indicators

Variables Description

Voice and accountability

It captures the general perception of the ability of a country‟s citizens

to pick their government exercising their constitutional rights of free

expression, association, etc.

Political stability It measures the probability of that national government of losing

control or overturning by violent or terrorist parties.

Government effectiveness

It estimates the quality of public and civil service, the degree of its

independence from political pressures, the quality of policy

formulation and conducting as well as the commitment for

accomplishment

Rule of law

It engulfs the whole spectrum of law enforcement by police and

judicial system so as to prevent crime, violence and breach of

personal rights. It is also indicative of how much agents trust and

comply with rules.

Control of corruption

It indicates the degree of corruption as being expressed by the

exploitation of public power to the best interests of elites and other

private parties.

Credit depth of information

sharing

It measures the quality and accessibility of credit information

disseminated by public or private registries in order to facilitate

lending transactions.

317

CHAPTER 9

Synopsis

319

9.1. Major findings of the research

In chapter 6, we investigate potential effects of bank competition as the latter is

defined as the price mark-up over marginal cost. We end up with monopolistic pricing

to be persistently associated with low levels of asset size for cooperative, savings and

„other‟ banks. The relationship turns out to be non-linear for savings and „other‟

banks following a positive pattern for low levels of total assets up to a point where

contestable conduct or economies of scale turn it negative. Cooperative banks

experience a negative pattern only in lower values of total assets since the dominating

cooperative banks of Germany constitute a homogenous set of equally sized

institutions.

The positive sign of concentration gives credit to the SCP paradigm once we see the

general picture of European banking sectors as well as the savings banks themselves.

When it comes to construct Herfindahl-Hirschman indices for deposits and loans

products, concentration is significant only for „other‟ banks in loans market, and for

commercial banks in both deposits and loans markets. Notwithstanding that relative

market power bears no significance in the whole sample, a positive indication is

traced in „other‟ banks and in greater levels of market share of cooperative banks.

However, a more competitive pricing seems a common practice for commercial banks

and for cooperative banks of relatively low market share.

Relative efficiency hypothesis is verified against „quit life hypothesis‟ by the negative

effect of total cost over total income variable at 1% level of significance along the

lines of Koetter et al. (2012), who concluded the same for US banks. We verify,

therefore, the strategic option of banks to exploit lower costs in favour of their

customers either in the form of lower loan rates or higher deposit rates (Vennet,

2002). Furthermore, credit risk motivates commercial and cooperative banks to apply

higher profit margins, a strategy which is contradistinction with that of „other‟ banks.

As for the economic conditions, we can only examine the effect of GDP growth rate

on banks of specific specialisation. In fact, high rates of economic expansion operate

procyclically for savings banks and countercyclically for commercial banks. Higher

elasticity of aggregate demand also induces banks to offer lower prices, especially in

320

the case of commercial and cooperative banks. Legal stringency proves to have a

positive impact only on the competitive behaviour of cooperative banks.

Next, we delve into the analysis of how such - or next akin to – effects influence the

pricing conduct of banks at the income level. In the same vein, commercial banks of

greater asset size tend to impose low prices on other interest and non-interest

products, a case that remains the same for cooperative banks, which opt for

competitive price on loans as well as higher fees. Economies of scale seem to produce

lower fees and higher prices on other non-interest products for savings banks whereas

„other‟ banks apply generally lower prices except for other interest activities. Non-

linearities have the same pattern for commercial, cooperative and „other‟ banks

yielding a negative sign on L-othernint as soon as it turns positive at higher levels of

total assets; in savings banks there is only a linear and positive association. Contrary

to commercial assets, all the other banks demonstrate first a positive effect on L-fees,

which then turns into negative whereas the same tendency is traced for L-loans in all

cases except savings banks. Profit margins on other interest products demonstrate a

diminishing (increasing) trend when „other‟ (savings) banks are getting larger,

notwithstanding higher scale economies reverse the sign.

Collusion hypothesis does play a significant role for commercial banks in

implementing higher fees while savings banks follow exactly the same strategy for

other interest income. Greater collusion in the deposit market promotes competitive

loan rates by cooperative and „other‟ banks, with the former reaping the benefits of

collusion in the loans market through higher loan rates. „Other‟ banks utilize

monopolistic structure in the deposit market by enjoying relatively higher prices on


Higher market share is correlated with lower fees in commercial and cooperative

banks and lower other interest charges in cooperative and savings banks; the latter

also prefers the utilisation of low other non-interest rates. „Other‟ banks apply lower

rates on loans and other non-interest products and higher prices on the remainder.

Higher relative market power in the deposit market drives all banks to compete each

other for loan rates, a behaviour which alters in commercial and cooperative banks

when they enjoy higher market share in the loans market. In addition, a common

feature for cooperative and „other‟ banks of greater market share in the loans market

321

consists in the use of lower fees but only „other‟ banks are able to impose higher fees

when they acquire market share in the deposits market. Moreover, charges on other

interest and non-interest activities contain their negative nexus market share in the

loans market.

Cost efficiency drives all banks to charge low loan rates, which are even affected by

loan losses in the portfolios of commercial, cooperative and savings banks. More

portfolio diversification makes cooperative and savings banks eager to charge fees

and other non-interest rates with an indication that same holds for „other‟ banks when

it comes to charge fees. Higher level of total equity that reflects risk aversion or an

explicit signal of bank creditworthiness enables cooperative and savings banks to

deploy a monopolistic strategy on other non-interest activities. Cooperative banks

with more liquid assets enjoy lower rates on other interest products followed by

commercial banks in offering cheaper loans. Savings banks do the same as

commercial banks but additionally compensate the underlying opportunity cost of

liquid assets through higher levels of non-interest rates.

Average country income has a positive impact on cooperative banks in charging

higher other interest prices, while demand elasticity motivate cooperative and saving

banks to apply lower fees and other non-interest charges. On the other hand, „other‟

banks seem to take advantage of higher fees. In periods of high inflation rate,

cooperative banks operate in contradistinction to commercial banks entertaining

considerable fees in order to possibly counterbalance potential losses on fixed loan

rates. Last, stricter lending laws turns out to bring about higher other interest charges

on behalf of cooperative banks against the stream of diminishing returns for the rest

banking specialisations.

Chapter 5 aims to contribute to the literature that compares alternative parametric

methodologies of estimating profit and cost efficiency under widely employed

definitions of bank inputs/outputs, namely the intermediation and value-added

approach. Hence, we applied REM, DFA and TFA methodologies along with

differently truncated versions of them at 1%, 5% and 10% level of their distribution

tails in order to produce averaged efficiency scores at the bank level. Only the

specification of TFA estimated for each year enables us to utilize time-varying

322

efficiency scores of differently specialized financial institutions over the period 2003-

2008 for the enlarged EU-27 region.

We conclude that all methodologies turn out to produce different levels of efficiency

although they may be comparable if we juxtapose their truncated specifications.

Interestingly, the relative rankings of banking markets and asset classes with respect

to their averaged cost/profit efficiency are almost identical across all approaches; in

contrast, that is not the case for different productive specializations. According to the

latter, cost efficiency is prevalent in the developed European region and in

evolutionary dominating pace over that of EU-12; however, the developing sub-group

has experienced persistently higher profit efficiency reaping the benefits of their

better opportunities for operational expansion.

Next, we estimate yearly scale economies for every banking market and observe

considerable discrepancies between their average values. Apart from few exceptions41

in the analysis, both approaches start from different points, that is IA approach comes

from above to converge eventually to the unity level in 2008, while VA approach

identifies increasing scale economies at the beginning of the period and then settles to

full scale exploitation. Last, we end up with significant „catching-up effect‟ and

converging dispersion of cost efficiency scores under both IA and VA approaches and

dynamic specifications. Profit efficiency demonstrates the same converging pattern at

even higher levels albeit beta/sigma convergence is clearly evident in so far ADL

specification allows for dynamic effects. Above-unity cases if assessed individually

make little economic sense as highlighting efficiency series fluctuate between

negative and positive values year by year. However, that pattern is counterbalanced

by the positive bearing of AR(1) variable.

In chapter 8, we provide evidence on whether the statistical relationship between

market structure and financial stability turns out significant under different

specifications for the European Union since the advent of the single currency. Thus, it

extends the bulk of empirical analysis to draw remarks on the validity of the

underlying relationship subject to different information sets out of a pool of various

bank-specific, macroeconomic and institutional factors that have been already

addressed in the literature.

41

Belgium, Luxembourg, United Kingdom and Poland.

323

EBA methodology systematically investigates the partial correlations of market

concentration and bank competition with bank soundness employing a) linear effects

of regulatory and supervisory variables, b) non-linear effects I-variables with

competition, c) interactions of I-variables with bank market power, d) foreign

ownership, e) non-linearities between governance indicators with competition, f)

different dependent variables that encompass different aspects of bank risk and g)

sensitivity analysis per bank specialization.

We found a fragile relationship between the Lerner index and Z-score both at 5% and

1% significance level, when utilising the effect of I-variables in levels (model 1).

When interactions come into play, the Lerner index follows the same pattern taking

account of different-sized information sets. However, once we run model 1 and 2 with

the inclusion of quadratic term, the underlying fragility stems particularly from the

pure linear impact of market power, as the squared term is persistently significant

across all extreme bounds. We also trace this U-shaped relationship to take place in

cooperative and savings banks and the whole sample of the European financial system

in general. The significance of squared Lerner index keeps pace with that of the

Lerner index itself, as we run ad hoc regressions when competition turns significant.

Reconciling of seemingly contending strands of the relevant literature, competition

and bank soundness follow a U-shaped path with a reverse in the pertinent coefficient

at the points 0.285 and 0.542 of the distribution.

Concentrated markets are highly correlated with bank soundness across any

specification and robustness check. Such impact is traced in the tendency of

concentration to predict significantly equity capital at 10% level, respectively.

However, fragility emanates from bank managers who engage in risk-taking in

lending transactions, and model 1, which investigates its impact on Z-score in levels.

Once we analyze whether HHI affects the relationship between market power and Z-

score in slopes, we end up with robust bearing in the standard EBA (table 6) whereas

in table 12 it becomes fragile in only one case out of 20 regressions of different I-

variable information sets.

Besides, we come up with collateral issues that have been appealing in many studies

in terms of the policy implications they put forward. In general, stringent activity

324

restrictions coupled with harsh exercize of supervisory power on managerial decisions

and foreign ownership are conducive to instability in the financial system.

Going one step beyond with I-variables interacting with foreign ownership we verify

that in banking markets of considerably foreign-owned bank assets, activity more

capital regulation, official supervisory intervention, freer market entries and sufficient

information requirements for the private sector, experience relatively higher degree of

risk-taking. We therefore end up with contradictory results, which endorse the

mandate for special treatment of foreign institutions since they require fewer

restrictions on capital reserves, loose intervention of supervisory authorities and

limited information transparency for their operational robustness.

The same pattern is evident when utilising governance indicator to be interacting with

bank-level competition. The Lerner index maintains its significance along with the

explanatory power of HHI when market power comes at 10% significance level. The

price mark-up over marginal cost has a considerable bearing on bank soundness the

more the state fulfils individual rights and implements policies free of potential

political interventions. In particular, bank soundness comes in tandem with low

activity restriction and official intervention as well as higher capital requirements and

private monitoring within a context of well-developed and respectful institutions.

Once we employ different aspects of bank risk, some interesting results pop up. First,

ROA and credit risk are sufficiently predicted by market power while market

concentration is associated with more capital buffers and losses on lending activities.

Amid conditions of ongoing regulatory policies, cooperative and savings banks are

more vulnerable to the bearing of market power as opposed to the operational strength

of „other‟ banks.

325

9.2. Policy implications of the research

The advent of Euro and the concomitant formulation of a single market brought about

considerable challenges and opportunities for the whole banking industry. In chapter

6, our analysis of the market power in 9 developed markets suggests imperfect

competition but not at levels that signify collusive tendency. However, the market of

Denmark and Sweden seem to experience more than 33% of price mark-up over

marginal cost whereas Spain lies in the other extreme with negative Lerner index. The

latter is justified by possible predatory pricing or irrational pricing conduct since cost-

raising strategy is not illustrated by high marginal costs.

We then attempt to explain competition in terms of which are the significant drivers

conducive to it. In particular, monopolistic pricing appears a common practice in

small-sized savings and „other‟ financial institutions as far as their increased size

makes them act competitively. In contrast, cooperative banks tend to impose low

prices in an attempt for high market shares albeit at greater levels the same positive

pattern holds. We take account of non-linearities, if any, between asset size and

market power and provide evidence that different bank types tend to behave

asymmetrically along the pace of increasing assets. Indeed, monopolistic pricing on

other non-interest bearing products comes along in savings banks and „other‟ banks of

higher size while banks turn out to impose high prices on loans (except savings

banks), fees (except commercial banks); other interest income activities are highly

charged by savings banks of limited asset class. Our analysis, therefore, gives insight

to the contribution of size on pricing behaviour and its role in upcoming bank failures.

We endorse also the necessity for antitrust policies that stand up against collusive

practices of concentrated markets, as savings (commercial) banks implement higher

other interest income (fees). With concentrated loans markets, market power is also

evident in „other‟ and commercial banks while the same holds for the latter in

deposits markets. Even so, cooperative („other‟) banks enjoy higher profit margins on

loans (other interest products) if the loans (deposits) market is considerably

concentrated. Likewise, banks with high market shares as well as those striving to get

the maximum out of it tend to behave competitively opposing to the operation of

„other‟ banks, which appear to enjoy higher fees due to their high market share in the

deposits market. Hence, we cannot verify (except for „other‟ banks) the relative

326

market power hypothesis and suggest policy makers should take account of certain

bank specialisations that enjoy high margins in specific product categories amid

collusive practices across either loans or deposits markets.

In addition, the efficient operation turns out significant in European banking as low

costs induce banks to charge lower interest rates on loans or higher deposit rates.

However, that gives no credit to the inherent impetus of banks for higher market

shares in order to exploit their power, with the exception of „other‟ banks and in high

levels of market share of cooperative banks. In contrast, commercial, cooperative with

higher market shares impose low fees, while cooperative and savings banks utilize

low other interest charges. Cost economies, therefore, constitute an indispensable

catalyst of convergence towards competitive prices while incentives in favour of

income diversification induce banks (cooperative, commercial, „other) in charging

higher fees and others (cooperative, commercial) for high other non-interest rates.

More capitalized banks (cooperative, savings) also tend to surrender in high prices of

other non-interest bearing activities, an issue which is directly linked to capital

requirements given that they constitute a considerable part of a bank‟s portfolio.

Furthermore, we signify the extension of Basel III in incorporating liquidity

requirements towards stable institutions that provide liquidity to market participants

effectively and manage payment transactions across different regions. Along the lines,

the analysis shows that more liquid cooperative (commercial) banks offer other

interest products (cheap loans) at low prices although savings banks tend also to

counterbalance the opportunity cost of liquid assets through higher non-interest rates.

The ongoing process of regulation with Basel III that sets out additional requirements

on a countercyclical basis, our analysis verifies such need only for savings given that

commercial banks react contrarily to the business cycle. The latter, however, may

imply poor monitoring and screening practices as well as (limited) securitisation of

low „quality‟. In addition, in economies of high GDP per capita and inflation

cooperative banks tend to charge higher other interest prices to make up for possible

losses on fixed loan rates even under strict lending loans. Hence, there is room for

further institutional reforming in such cases especially in the group of „other‟ banks,

which seems to exploit higher elasticity of aggregate demand and impose higher fees.

327

Although there has been a considerable progress towards the contestability of the

European banking industry, the concomitant pressure on profitability is bound to

bring about restructuring and consolidation procedures. Hence, our analysis of cost

and profit efficiency across various methodologies and approaches aims at providing

evidence if convergence dynamics is taking place. However, we acknowledge the

heterogeneity of financial institutions and the utopian scope of perfect competitive

banking market due to the per se idiosyncratic features of banks.

In chapter 7, we report similar rankings of banking industries and asset classes across

different specifications with respect to cost and profit efficiency although of different

level, whatsoever. We need to know the underlying differences of European countries

so as for policy makers to boost the integration procedure especially for the

developing subgroup. Although rankings with respect to various productive

specialisations diverge due to bank specificities, the way that banking markets are

structured gives more insight at a disaggregated level. The gap in cost efficiency

between EU-15 and EU-12 seems to be closing while profit efficiency of the latter is

in dominating pace. After the adoption of Euro, fierce competition drove banks of the

developed world to less developed regions as deregulation gave permission to new

market entries in the form of branches and subsidiaries or through mergers and

acquisitions. In so doing, these entries have set new standards in risk management and

operational performance by means of economising on input costs as well as on input

and output mix.

The analysis on scale economies produces mixed evidence if we employ different

definitions of bank output. In general, by the onset of the financial crisis banks settled

to constant returns of scale converging from different starting points. If we define cost

function in terms of producing securities, European countries seem to experience

diseconomies of scale due to irrational profligacy of trading losses whereas if we

consider deposits as part of the production output, economies of scale have been

considerable up to the point of a „deadlock‟ in 2008. On the whole, the catching-up

effect in cost and profit efficiency turns out significant according to various

specifications as well as the their dispersion across the whole European market.

Coupled with intensified competitive conditions, financial institutions have employed

the best practices to enhance their operating performance and it poses challenges for

328

further progress on regulation, supervision and corporate governance towards market

integration and level-playing field in the financial system.

The aforementioned analysis on competition and efficiency turns out to be conducive

to economic development and financial stability. After conducting sensitivity analysis

of market structure variables on bank soundness (Z-score), we utilized various

information sets of regulation and supervision variables up to the most recent year

available (2008). As there exist diverging strands of the literature on the competition-

stability nexus, the results report robust evidence on the fragility of the inderlying

relationship as being contingent on specific aspects of banks‟ institutional framework.

We also reconcile them by investigating the significance of the quadratic term of the

Lerner index, concluding about the inverse U-shaped relationship of market power

and stability. At the lowest level of its coefficient, market power has only negative

bearing on stability, while all the other possibilities illustrate that the change in sign

occurs within the bound [0.285, 0.542], which corroborates the respective values of

0.28 and 0.33 of Beck et al. (2012).

Evidence on concentration verifies its persistent significance in determining bank

solvency although excessive loan provision undermines the underlying relationship.

Hence, we endorse the importance of too-big-to-fail policies in the financial system in

conjunction with pertinent regulatory standards on capital requirements, liquidity

issues and stress testing. The set of institutional variables used in the analysis enters

highly significant if assessed individually or collectively. In particular, low activity

restrictions, foreign ownership and supervisory power are conducive catalysts of

financial instability, whereas higher capital requirements and private monitoring

operating to the contrary. Therefore, efforts to more effective capital regulation as set

out in Basel III should continue unabated considering some interesting issues. First,

there is little room for income diversification so the authorities should reconsider the

destabilising impact of banks‟ engagement in risky projects.

In addition, foreign ownership brings about negative repercussions on bank soundness

possibly due to the potential cost burden that the parent bank has to bear vis-à-vis

insolvency and liquidity shocks (Claessens et al., 2001b). The latter is evident if

exploitation of market power in foreign markets takes the form of acquiring national

banks in monopolistic markets, taking advantage of their high cost efficiency or

329

establishing branches and subsidiaries in quest of large market shares and there exists

concurrently considerable inadequacy of „self-protection‟. Thus, policies need to

incorporate – though they have done so to some extent - additional requirements on

capital and liquidity within the foreign-owned banking group. Moreover, supervisory

authority should take into consideration that high degree of intervention to managerial

decisions ends up destabilising the market if associated with corruption in lending and

obstruction of bank operations (Barth et al., 2004).

On the other hand, in markets of high degree of foreign ownership the analysis

suggests few capital requirements, less intensified supervision, limited market entries

and information dissemination to the private sector. The results show that the previous

recommendations have an adverse effect if institutional controls demonstrate non-

linear effects on bank soundness through foreign-owned structures.

9.3. Limitations of research

Before moving into some possible recommendations for further research in European

banking, we should mention several methodological problems and limitations that are

quite common amongst similar studies of empirical research.

Considering the first contribution on bank competition and its possible determinants

even at the level of income sources, a considerable drawback is the lack of

information with respect to the price of borrowed funds as a standard bank input.

However, this drawback constitutes the explanatory strength of our model since the

inclusion of the underlying input may capture some degree of market power of the

incumbent banks in deposits markets, and thus may produce bias on the level of

marginal cost and, concomitantly, that of competition (Ariss, 2010). Thus, we can by

no means juxtapose the results of our model with that specification comprising the

additional input, albeit the necessity of doing so is dominated by the potential bias of

the results and the significant fit of our model specification.

As for the estimation of the Lerner index, with respect to specific income sources, we

utilize only those with sufficient information. Our analysis would be fully fledged had

we had a balanced sample of all the relevant constituents of banks‟ portfolio structure.

330

In addition, all empirical studies that employ bank-specific factors presumably

suppose that bank outputs and inputs are homogenous or at least the degree of

heterogeneity is not enough to render the analysis erroneous. Surmising, for example,

the inputs of loans and other investments, the Bankscope database does not account of

different categories of loans and investments but alternatively aggregates them into

one general group.

The second contribution on the efficiency analysis of the European banking industry

is contingent on some methodological assumptions. For example, the DFA approach

assumes that over the sample period the error term cancels each other out and what is

left off is the average efficiency level for each financial institution. If the period is too

short, we cannot disentangle the overall bank inefficiency out of the composed error

term, and in case of being too long market conditions may pose a threat on the

stability of bank inefficiency. Notwithstanding we follow the recommendation of

Mester (2003) for 6-year period as the best time interval for counterbalancing the

underlying centrifugal forces, the introduction of EMU may challenge the robustness

of the results. Furthermore, the TFA approach may also highlight the fact that

efficiency analysis in not a matter of a unique approach. Indeed, it is theoretically

founded in the arbitrary assumption that the thick frontier, constructed by the

technology of the most cost and profit efficient 1% within each asset class of the

distribution, is well defined and represented. Even the REM is potentially suspect if

we consider that the inherent bank specificities42

, or in other words the correlation

between the error term and regressor(s) may obscure the evidence; however, a

heterogeneous sample of different institutional types may be more appropriate in that

it considers the stochastic nature of bank efficiency (Simar, 1992).

As different methodologies of efficiency analysis should be treated with caution, we

constrain the comparison in various parametric approaches but not in non-parametric

one. This shortcoming, whoever, stems from the fact that previous studies have

concluded on considerable differences between them, and thus a more rigorous

analysis was imperative especially if coupled with different definitions of bank

output. We should mention that not all approaches are applicable in our sample; for

42

In the FEM case, the most efficient bank is that having the lowest fixed effect. Thus, it has a

deterministic character contrary to the inefficiency extrapolated by REM, which assumes it as part of

the error term.

331

example, SFA could not achieve concavity conditions on MLE iterative process while

distance function methodology demonstrated limited fit. Moreover, a highly

unbalanced sample makes DEA analysis potentially biased due to many missing

values in the EU-27 region.

Last, the third contribution on the competition-stability nexus includes the estimation

of the Lerner index through a cost function that omits price of funds but, as already

mentioned, illustrates better the effect of bank market power on bank soundness. In

doing so, we employ regulation and supervision variables updated up to 2008, prior to

the onset of the financial crisis. That stresses considerable limitation to the analysis

although we figure out the way we have been forming the conditions conducing to the

crisis. Moreover, we conduct sensitivity analysis along the lines of Extreme Bound

Analysis (EBA) given a set of macroeconomic, institutional and bank-specific

variables that we consider as parsimonious and sufficiently representative.

9.4. Suggestions for further research

There exist plenty of topics for academic research despite the ceaselessly progressive

nature of banking literature. From a theoretical perspective, research should lean

towards the detailed mapping-out of contemporary reforms in international banking

that have emerged since 2008 in order to be assessed in relation to the degree of

market concentration and bank market power. We need to develop new methods of

modeling financial risks in an attempt to forecast with perfect hindsight the causes of

financial bubbles and take lessons from other recent crises. Thus, it is an imperative

factor to investigate the nexus between politics and banking in restructuring the

European financial system with respect to institutional reforms that may resolve

„culpable‟ interconnections between the banking markets, fiscal policies and

investment strategies in EU.

We also need to shed some light on the patterns of profitability, corporate governance

and operational performance in banks of cross-border activity. As financial

institutions struggle to survive in a single market, their strategic options lie in the

structure of their portfolios, which may enclose potential threats for systemic risk

332

within a context of relaxed supervision and owners‟ control. Hence, further research

may also focus on models for forecasting risk on banks‟ future earnings through the

employment of (dis)aggregated accounting data. The per se issue of bank size and its

optimal level may constitute the principal scope before engaging in risk management

policies since it engulfs all the endogenous drivers of bank failures.

Other useful extension closely pertaining to the financial system and banking, could

be how fiscal policies (e.g. taxation) interact with competition and macroprudential

regulation, or the general equilibrium and empirical framework of public debt

management coupled with bank competitiveness. Additionally, the passing-through

mechanism of monetary policy in (cross-border) banking is of utmost importance

since their impact on stabilising inflation and output may occur through interest rates,

asset prices, exchange rates and bank credit (see Mishkin, 1995; Bernanke and

Gertler, 1995; Hubbard, 1995). On the top of that, amid conditions of sovereign debt

crisis, credit ratings at the bank and country level may predict financial distress

through its impact on bond yields and market structure of banks.

Financial institutions in the European union have been operating in a fast-changing

environment, which is subject to the ongoing trend of technological advancement and

deregulation. Thus, we need to analyse the key determinants of bank integration and

consolidation so as to perceive how banks are adapting their operations through their

competitive conduct at any point of the business cycle. Last, all issues that this thesis

touches upon constitute fruitful areas of academic research while the debate on

challenging institutional reforms continues unabated.

333

References

Abbasoglu O.F., Aysan A.F. and Gunes, A. (2007). Concentration, competition,

efficiency, and profitability of the Turkish banking sector in the Post-Crisis Period,

Banks and Bank Systems, vol. 2, no. 3, pp. 106-115.

Acharya, V.V. (2009). A theory of systemic risk and design of prudential bank

regulation, Journal of financial stability, vol. 5, pp. 224-255.

Adams, R.M., Berger, A.N. and Sickles, R.C. (1999). Semiparametric approaches to

stochastic panel frontiers with applications in the banking industry, Journal of

Business and Economic Statistics, vol. 17, pp. 349–358.

Adelman, M.A. (1951). The Measurement of Industrial Concentration, The Review of

Economics and Statistics, vol. 33, no. 4, pp. 269-296.

Adelman, M.A. (1969). Comment on the "H" concentration measure as a numbers-

equivalent, The Review of economics and statistics, vol. 51, no. 1, pp. 99-101.

Agoraki, M-E, Delis, M.D. and Pasiouras, F. (2009). Regulations, competition and

bank risk-taking in transition countries, Journal of Financial Stability,

doi:10.1016/j.jfs.2009.08.002.

Aigner, D.J., Lovell, C.A.K. and Schmidt, P. (1977). Formulation and estimation of

stochastic frontier production models, Journal of Econometrics, vol. 6, pp. 21–37.

Akhavein, J.D., Berger, A.N. and Humphrey, D.B. (1997). The effects of

megamergers on efficiency and prices: Evidence from a bank profit function, Review

of industrial Organization, vol. 12, no. 1, pp. 95-139.

Akin, G.G., Aysan, A.F., Borici, D. and Yildiran, L. (2012). Regulate one service,

tame the entire market: Credit cards in Turkey, Journal of Banking and Finance, vol.

37, no. 4, pp. 1195-1204.

Al Fayoumi, N. and Abuzayed, B. (2010). Competitive Conditions in MENA Banking

Markets.

Al-Jarrah, I.M. (2010). Competition in the Jordanian‟s Banking Sector. Dirasat:

334

Administrative Sciences, vol. 37, no. 2.

Al-Muharrami, S. (2008). Testing the contestability in Kuwait banking industry,

Studies in Economics and Finance, vol. 25, no. 4, pp. 253-266.

Al-Muharrami, S. (2009a). Analysis of competitiveness in Qatar banking industry,

International Journal of Business Innovation and Research, vol. 3, no. 2, pp. 168-181.

Al-Muharrami, S. (2009b). The competition and market structure in the Saudi Arabia

banking, Journal of Economic Studies, vol. 36, no. 5, pp. 446-460.

Allen, F. (1990). The market for intermediation and the origin of financial

intermediation, Journal of Financial Intermediation, vol. 1, pp. 3-30.

Allen, F. and Gale, D. (2000). Comparing financial systems. Cambridge, MA: MIT

Press.

Allen, F. and Gale, D. (2004). Competition and financial stability, Journal of Money,

Credit and Banking, vol. 36, no. 3, pp. 453-480.

Allen, L. and Rai, A. (1996). Operational efficiency in banking: An international

comparison, Journal of Banking and Finance, vol. 20, no. 4, pp. 655-672.

Altunbas, Y. and Marques, D. (2008). Mergers and acquisitions and bank

performance in Europe: The role of strategic similarities, Journal of Economics and

Business, vol. 60, no. 3, pp. 204-222.

Altunbas, Y. and Molyneux, P. (1996). Economies of scale and scope in European

banking, Applied Financial Economics, vol. 6, no. 4, pp. 367-375.

Altunbas, Y., Carbo, S., Gardener, E.P. and Molyneux, P. (2007). Examining the

relationships between capital, risk and efficiency in European banking, European

Financial Management, vol. 13, no. 1, pp. 49-70.

Altunbaş, Y., Gardener, E. P., Molyneux, P. and Moore, B. (2001). Efficiency in

European banking, European Economic Review, vol. 45, no. 10, pp. 1931-1955.

Aly, H.Y., Grabowski, R., Pasurka, C. and Rangan, N. (1990). Technical, scale, and

allocative efficiencies in US banking: an empirical investigation, The Review of

335

Economics and Statistics, pp. 211-218.

Andersen, P. and Petersen, N.C. (1993). A procedure for ranking efficient units in

DEA, Management Science, vol. 39, pp. 1261–1264.

Andrieş, A.M. and Capraru , B. (2012). Competition and efficiency in EU27 banking

systems, Baltic Journal of Economics, vol. 12, no. 1, pp. 41-60.

Angelini, P. and Cetorelli, N. (2003). The effects of regulatory reform on competition

in the banking industry, Journal of Money, Credit and Banking, vol. 35, pp. 663-684.

Anzoategui, D., Martinez Peria, M. and Rocha, R. (2010). Bank competition in the

Middle East and northern Africa region, World Bank Policy Research Working Paper

Series, vol. 5363.

Anzoategui, D., Martinez Peria, M.S. and Melecky, M. (2010). Banking Sector

Competition in Russia (October 1, 2010). World Bank Policy Research Working

Paper Series, no. 5449.

Aparicio, J., Ruiz, J. and Sirvent, I. (2007). Closest targets and minimum distance to

the Pareto-efficient frontier in DEA, Journal of Productivity Analysis, vol. 28, pp.

209-218.

Aragon, Y., Daouia, A. and Thomas-Agnan, C. (2005). Nonparametric frontier

estimation: a conditional quantile-based approach, Econometric Theory, vol. 21, pp.

358-389.

Ariff, M. and Can, L. (2008). Cost and profit efficiency of Chinese banks: A non-

parametric analysis. China Economic Review, vol. 19, no. 2, pp. 260-273.

Ariss, R.T. (2009). Competitive behaviour in Middle East and North Africa banking

systems. The Quarterly Review of Economics and Finance, vol. 49, no. 2, pp. 693-

710.

Ariss, R.T. (2010a). Competitive conditions in Islamic and conventional banking: A

global perspective, Review of Financial Economics, vol. 19, no. 3, pp. 101-108.

Ariss, R.T. (2010b). On the implications of market power in banking: Evidence from

336

developing countries, Journal of Banking and Finance, vol. 34, no. 4, pp. 765-775.

Atkinson, S.E. and Dorfman J.H. (2005). Bayesian measurement of productivity and

efficiency in the presence of undesirable outputs: crediting electric utilities for

reducing air pollution, Journal of Econometrics, vol. 126, pp. 445-468.

Bader, M.K.I., Mohamad, S.H.A.M.S.H.E.R., Ariff, M. and Hassan, T. (2008). Cost,

revenue and profit efficiency of Islamic versus conventional banks: international

evidence using data envelopment analysis, Islamic Economic Studies, vol. 15, no. 2,

pp. 23-76.

Baltagi, B. (1995). Econometric Analysis of Panel Data, John Wiley and Sons, New

York.

Baltagi, B.H. and Griffin, J.M. (1988). A general index of technical change, The

Journal of Political Economy, vol. 96, no. 1, pp. 20-41.

Baltagi, H.B. (2001). Econometric analysis of panel data, New York: John Wiley and

Sons.

Banker, R.D. and Maindiratta, A. (1992). Maximum likelihood estimation of

monotone and concave production frontiers, Journal of Productivity Analysis, vol. 3,

pp. 401–415.

Banker, R.D. and Morey, E.C. (1986). The use of categorical variables in data

envelopment analysis, Management Science, vol. 32, no. 12, pp. 1613–1627.

Banker, R.D. and Natarasan, R. (2004). Statistical tests based on DEA efficiency

scores. In: Cooper, W.W., Seiford, L.M., Zhu, J. (Eds.), Handbook on Data

Envelopment Analysis, Kluwer Academic Publishers, Norwell, MA (Chapter 11).

Banker, R.D., Charnes, A. and Cooper, W.W., (1984). Some models for estimating

technical and scale inefficiencies in data envelopment analysis, Management Science,

vol. 30, pp. 1078–1092.

Barajas, A. (1996). Interest Rates, Market Power, and Financial Taxation: An

Application to Colombian Banks 1974–1988. (unpublished; Washington: IMF ,

November).

337

Barro, J.R. and Sala-i-Martin, X. (1992). Convergence, The Journal of Political

Economy, vol. 100, no. 2, pp. 223-251.

Barro, R.J. (1991). Economic growth in a cross section of countries, The Quarterly

Journal of Economics, vol. 106, no. 2, pp. 407-443.

Barro, R.J. and Sala-i-Martin, X. (1992). Convergence, Journal of Political Economy,

vol. 100, no. 2, pp. 223-251.

Barros, C.P., Ferreira, C. and Williams, J. (2007). Analysing the determinants of

performance of best and worst European banks: A mixed logit approach, Journal of

Banking and Finance, vol. 31, no. 7, pp. 2189-2203.

Barros, P.P., Berglof, E., Fulghierri, J., Gual, J., Mayer, C. and Vives, X. (2005).

Integration of European Banks: The Way Forward. Centre for Economic Policy

Research, London.

Barth, J. R., Caprio, G. and Levine, R. (2005), Rethinking bank regulation: Till angels

govern, Cambridge University Press.

Barth, J.R., Caprio Jr. and Levine, R. (2000). Banking systems around the globe: do

regulation and ownership affect performance and stability?, World Bank Policy

Research Working paper, Washington DC.

Barth, J.R., Caprio, G. and Levine, R. (2004). Bank regulation and supervision: what

works best?, Journal of Financial intermediation, vol. 13, no. 2, pp. 205-248.

Barth, J.R., Caprio, G. and Levine, R. (2008). Bank regulations are changing: For

better or worse?, Comparative Economic Studies, vol. 50, no. 4, pp. 537-563.

Battese, G.E. and Coelli T.J. (1995). A Model for Technical Inefficiency Effects in a

Stochastic Frontier Production Function for Panel Data, Empirical Economics, vol.

20, pp. 325–332.

Battese, G.E. and Coelli, T.J. (1992). Frontier production functions, technical

efficiency and panel data: with application to paddy farmers in India, Journal of

productivity analysis, vol. 3, no. 1, pp. 153-169.

338

Bauer, P.W., Berger, A.N. and Humphrey, D.B. 1993. “Efficiency and productivity

growth in US banking”. In The Measurement of Productive Efficiency. Techniques

and applications, Edited by: Fried, H. O., Lovell, C. A. and Schmidt, S. S. 386–413.

Oxford: Oxford University Press.

Bauer, P.W., Berger, A.N., Ferrier, G.D. and Humphrey, D.B. (1998). Consistency

conditions for regulatory analysis of financial institutions: a comparison of frontier

efficiency methods, Journal of Economics and Business, vol. 50, no. 2, pp. 85-114.

Baumol, W. (1982). Contestable markets: An uprising in the theory of industry

structure, American Economic Review, vol. 72, pp. 1-15.

Baumol, W., Panzar, J. and Willig, R. (1982). Contestable markets and the theory of

industry structure, Harcourt Brace Jovanovich, San Diego.

Beccalli, E. (2007). Does IT investment improve bank performance? Evidence from

Europe, Journal of Banking and Finance, vol. 31, no. 7, pp. 2205-2230.

Beck, T. (2007). Bank Concentration and Fragility. Impact and Mechanics, In The

Risks of Financial Institutions (pp. 193-234), University of Chicago Press.

Beck, T. (2008). Bank Competition and Financial Stability: Friends or Foes?, World

Bank Policy Research, Working Paper 4656, June.

Beck, T., De Jonghe, O. and Schepens, G. (2012). Bank competition and stability:

cross-country heterogeneity, Journal of Financial Intermediation.

Beck, T., Demirguc-Kunt, A. and Levine, R. (2006a). Bank concentration,

competition, and crises: first results, Journal of Banking and Finance, vol. 30, pp.

1581–1603.

Beck, T., Demirgüç-Kunt, A. and Levine, R. (2006b). Bank supervision and

corruption in lending, Journal of Monetary Economics, vol. 53, no. 8, pp. 2131-2163.

Beck, T., Demirguc-Kunt, A. and Maksimovic, V. (2003). Bank competition,

financing obstacles and access to credit, World Bank policy research working paper

no. 3041.

339

Beck, T., Hesse, H., Kick, T. and von Westernhagen, N. (2009). Bank ownership and

stability: evidence from Germany, Bundesbank Working Paper Series.

Benston, G.J. (1972). Economies of scale of financial institutions, Journal of Money,


Berg, S.A., Førsund, F.R., Hjalmarsson, L. and Suominen, M. (1993). Banking

efficiency in the Nordic countries, Journal of Banking and Finance, vol. 17, no. 2, pp.

371-388.

Berg, S.A. and Kim, M. (1994). Oligopolistic interdependence and the structure of

production in banking: an empirical evaluation, Journal of Money, Credit and

Banking, vol. 26, no. 2, pp. 309-322.

Berg, S.A. and Kim, M. (1998). Banks as multioutput oligopolies: An empirical

evaluation of the retail and corporate banking markets, Journal of Money, Credit and

Banking, vol. 30, no. 2, pp. 135-153.

Bergendahl, G. and Lindblom, T. (2008). Evaluating the performance of Swedish

savings banks according to service efficiency, European Journal of Operational

Research, vol. 185, no. 3, pp. 1663-1673.

Berger, A.N., Hasan, I. and Zhou, M. (2009). Bank ownership and efficiency in

China: What will happen in the world‟s largest nation?, Journal of Banking and

Finance, vol. 33, no.1, pp. 113-130.

Berger, A., Leusner, J. and Mingo, J. (1994). The efficiency of bank branches. Center

for Financial Institutions Working Papers.

Berger, A.N. and Humphrey, D.B. (1997). Efficiency of financial institutions:

International survey and directions for future research, European Journal of

operational research, vol. 98, no. 2, pp. 175-212.

Berger, A.N., Hanweck, G.A. and Humphrey, D.B. (1987). Competitive viability in

banking: Scale, scope, and product mix economies, Journal of Monetary Economics,

vol. 20, no. 3, pp. 501-520.

Berger, A.N., Klapper, L.F. and Ariss, R.T. (2009). Bank competition and financial

340

stability, Journal of Financial Services Research, vol. 35, no. 2, pp. 99-118.

Berger, N.A. (1993). “Distribution-free” estimates of efficiency in the US banking

industry and tests of the standard distributional assumptions, Journal of Productivity

Analysis, vol. 4, no. 3, pp. 261-292.

Berger, N.A. (1995). The profit-structure relationship in banking – tests of market-

power and efficient-structure hypotheses. Journal of Money, Credit, and Banking, vol.

27, pp. 404-415.

Berger, N.A. (1999). The big picture about relationship-based finance, Proceedings,

Federal Reserve Bank of Chicago, issue March, pp. 390-400.

Berger, N.A. and Hannan, H.T. (1992). The price-concentration relationship in

banking: A reply, The Review of Economics and Statistics, vol. 74, no. 2, pp. 376-379.

Berger, N.A. and Hannan, H.T. (1998). The Efficiency Cost of Market Power in the

Banking Industry: A Test of the "Quiet Life" and Related, The Review of Economics

and Statistics, vol. 80, no. 3, pp. 454-465.

Berger, N.A. and Hannan, Η.Τ.. (1989). The price-concentration relationship in

banking, Review of Economics and Statistics, vol. 71, no. 2, pp. 291-299.

Berger, N.A. and Humphrey, D.B. (1991). The dominance of inefficiencies over scale

and product mix economies in banking, Journal of Monetary Economics, vol. 28, pp.

117–148.

Berger, N.A. and Humphrey, D.B. (1992). Measurement and Efficiency Issues in

Commercial Banking. In Output Measurement in the Service Sectors, Z. Griliches

(Eds.). National Bureau of Economic Research Studies in Income and Wealth, vol.

56, Chicago, University of Chicago Press.

Berger, N.A. and Mester, L.J. (1997). Inside the black box: What explains differences

in the efficiencies of financial institutions? Journal of Banking and Finance, vol. 21,

no. 7, pp. 895–947.

Berger, N.A., Klapper, F.L. and Ariss, R.T. (2009). Bank competition and financial

stability, Journal of Financial Service research, vol. 35, pp. 99-118.

341

Berlin, M. and Mester, J. L. (1992). Debt covenants and renegotiation, Journal of

Financial Intermediation, vol. 2, no. 2, pp. 95-133.

Berlin, M. and Mester, L.J. (1999). Deposits and relationship lending, Review of

Financial Studies, vol. 12, no. 3, pp. 579-607.

Bertschek, I. and Lechner, M. (1998). Convenient Estimators for the Panel Probit

Model, Journal of Econometrics, vol. 87, pp. 329‐371.

Besanko, D. and Thakor, A. (1993). „Relationship banking, deposit insurance and

bank portfolio choice, in C. Mayer and X. Vives, eds., Capital markets and financial

intermediation, Cambridge University Press.

Bessent, A., Bessent, W., Elam, J. and Clark, T. (1988). Efficiency frontier

determination by constrained facet analysis, Journal of Operational Research Society,

vol. 36, pp. 785–796.

Biekpe, N. (2011), The Competitiveness of Commercial Banks in Ghana. African

Development Review, 23: 75–87, pp.1467-8268.

Bikker A.J. and Haaf, K. (2002). Competition, concentration and their relationship:

An empirical analysis of the banking industry, Journal of Banking and Finance, vol.

26, no. 11, pp. 2191-2214.

Bikker, A.J. (2003). Testing for imperfect competition on EU deposit and loan

markets with Bresnahan's market power model, Kredit und Kapital, vol. 36, no. 2, pp.

167-212.

Bikker, A.J. and Groeneveld, J.M. (2000). Competition and concentration in the EU

banking industry, Kredit Und Kapital, vol. 33, pp. 62–98.

Bikker, A.J., Shaffer, S. and Spierdijk, L. (2012). Assessing Competition with the

Panzar-Rosse Model: The Role of Scale, Costs, and Equilibrium, vol. 94, no. 4, pp.

1025-1044.

Bikker, J.A. (2002). Efficiency and cost differences across countries in a unified

European banking market, De Nederlandsche Bank.

342

Bolt, W. and Humphrey, D. (2010). Bank competition efficiency in Europe: A frontier

approach, Journal of Banking and Finance, vol. 34, no. 8, pp. 1808-1817.

Bonfim, D., Dai, Q. and Franco, F. (2009). The number of bank relationships,

borrowing costs and bank competition, Borrowing Costs and Bank Competition

(February 5, 2009).

Bonfim, D., Dai, Q. and Franco, F. (2009). The number of bank relationships,

borrowing costs and bank competition, Borrowing Costs and Bank Competition

(February 5, 2009).

Bonin, J.P., Hasan, I. and Wachtel, P. (2005). Bank performance, efficiency and

ownership in transition countries, Journal of Banking and Finance, vol. 29, no. 1, pp.

31-53.

Bonin, J.P., Hasan, I. and Wachtel, P. (2005b). Privatization matters: Bank efficiency

in transition countries, Journal of Banking and Finance, vol. 29, no. 8, pp. 2155-2178.

Boone J., Griffith, R. and Harrison, R. (2004). Measuring Competition, presented at

the Encore Meeting 2004 Measuring competition.

Boone, J. (2000). Measuring Product Market Competition, CEPR Working Paper

2636.

Boone, J. (2001). Intensity of competition and the incentive to innovate, International

Journal of Industrial Organization, vol. 19, no. 5, pp. 705-726.

Boone, J. (2004). A new way to measure competition, CEPR Discussion Paper Series

No. 4330.

Boot, A.W. and Greenbaum, S. (1993). Bank regulation, reputation, and rents: Theory

and policy implications. In: Mayer, C., and Vives, X. (eds), Capital markets and

financial intermediation. Cambridge, UK: Cambridge University Press, 292-318.

Boot, A.W.A and Thakor, A.V. (2000). Can relationship banking survive

competition?, Journal of Finance , vol. 55, pp. 679–713.

Borio, C. (2006). „The Macroprudential Approach to Regulation and Supervision:

343

Where do we Stand?‟, Chapter 7 in Erfaringer og utfordringer Kredittilsynet 1986–

2006 (the special volume celebrating the 20th anniversary of Kredittilsynet).

Kredittilsynet : Norway.

Borio, C. and Lowe, P. (2002). Asset prices, financial and monetary stability:

exploring the nexus, BIS Working Paper No. 114.

Borio, Cl. (2003). Towards a macroprudential framework for financial supervision

and regulation?, BIS Working papers, no. 128.

Borovicka, J. (2007). Banking Efficiency and Foreign Ownership in Transition: Is

There an Evidence of a “Cream-Skimming” Effect, CERGE-EI Working Paper.

Bos J. (2002). European banking: market power and efficiency, University Press,

Maastricht.

Bos, J.W. and Kolari, J.W. (2005). Large Bank Efficiency in Europe and the United

States: Are There Economic Motivations for Geographic Expansion in Financial

Services?, The Journal of Business, vol. 78, no. 4, pp. 1555-1592.

Bos, J.W. and Kool, C.J. (2006). Bank efficiency: The role of bank strategy and local

market conditions, Journal of Banking and Finance, vol. 30, no. 7, pp. 1953-1974.

Bos, J.W., Koetter, M., Kolari, J.W. and Kool, C.J. (2009). Effects of heterogeneity

on bank efficiency scores, European Journal of Operational Research, vol. 195, no.

1, pp. 251-261.

Boucinha, M. and Ribeiro, N. (2009). An assessment of competition in the Portuguese

banking system in the 1991-2004 period. Financial Stability Report| 2007, 165.

Boucinha, M., Ribeiro, N. and Weyman-Jones, T. (2009). An assessment of

Portuguese banks‟ costs and efficiency, Banco de Portugal, Working Paper, 22.

Boutillier, M., Gaudin, J. and Grandperrin, S. (2005). La situation concurrentielle des

principaux secteurs bancaires européens entre 1993 et 2000: quels enseignements

pour la future structure des marchés financiers issue de l'UEM?, Revue d'économie

financière, vol. 81, no. 4, pp. 15-42.

344

Boyd, H.J., De Nicolò, G. and Jalal M.A. (2009). Bank competition, risk and asset

allocations, IMF Working Paper, no. 09/143.

Boyd, J.H. and De Nicolo, G. (2005). The theory of bank risk taking and competition

revisited, Journal of Finance, vol. 60, pp. 1329–1343.

Boyd, J.H. and Runkle, D. E. (1993). Size and performance of banking firms: Testing

the predictions of theory, Journal of Monetary Economics, vol. 31, no. 1, pp. 47-67.

Boyd, J.H., De Nicoló, G. and Smith, B.D. (2004). Crises in competitive versus

monopolistic banking systems, Journal of Money, Credit and Banking, vol. 36, pp.

487–506.

Boyd, J.H., De Nikolo, G. and Al Jalal, A. (2006). Bank risk taking and competition

revisited: New theory and evidence, IMF Working paper, no. 06/297.

Boyd, J.H., De Nikolo, G. and Smith, B. D. (2004). Crises in competitive versus

monopolistic banking systems, Journal of Money, Credit and Banking, vol. 36, no. 3,

pp. 487-506.

Boyd. J. and De Nicoló, G. (2005). The theory of bank risk taking revisited, Journal

of Finance , vol. 60, pp. 1329–1343.

Bramer, P., Gischer, H., Richter, T. and Weiß, M. (2012). Diverse Degrees of

Competition within the EMU and their Implications for Monetary Policy (no. 10).

Otto- von-Guericke University Magdeburg, Faculty of Economics and Management.

Bresnahan, T.F. (1982). The oligopoly solution concept is identified, Economic

Letters, vol. 10, pp. 87- 92.

Bresnahan, T.F., (1989). Empirical studies of industries with market power. In:

Schmalensee, R., Willig, R.D.(Eds.), In: Handbook of Industrial Organization, vol. 2.

Elsevier Science Publishers, pp. 1011–1058.

Brissimis, S.N. and Delis, M.D. (2011). Bank-level estimates of market power,

European Journal of Operational Research, vol. 212, no.3, pp. 508-517.

Brissimis, S.N., Delis, M.D. and Papanikolaou, N.I. (2008). Exploring the nexus

345

between banking sector reform and performance: evidence from newly acceded EU

countries, Journal of Banking and Finance, vol. 32, pp. 2674–2683.

Cameron, A.C. and Trivedi, P.K. (2010). Microeconometrics using stata (Vol. 5).

College Station, TX: Stata Press, p. 166.

Caminal, R. and Matutes, C. (2002). Market power and banking failures,

International Journal of Industrial Organization, vol. 20, pp. 1341-61.

Campa M.J. and Hernando, I. (2006). M&As performance in the European financial

industry, Journal of Banking & Finance, vol. 30, no. 12, pp. 3367-3392.

Canhoto, A. (2004). Portuguese banking: A structural model of competition in the

deposits market, Review of Financial Economics, vol. 13, no. 1, pp. 41-63.

Canhoto, A. and Dermine, J. (2003). A note on banking efficiency in Portugal, New

vs. Old banks, Journal of banking and finance, vol. 27, no. 11, pp. 2087-2098.

Canova, F. and Marcet, A. (1995). The poor stay poor: Non-convergence across

countries and regions.

Capie, F. (1995). Prudent and stable but inefficient?: Commercial Banks in Britain,

1890-1940. In: Bordo, M.D., Sylla, R. (Eds.), Anglo-American Financial Systems:

Institutions and Markets in the Twentieth Century, Irwin, New York.

Carbo, S., Humphrey, D. and Rodriguez , F. (2003). Deregulation, bank competition

and regional growth, Regional Studies, vol. 37, pp. 227-237.

Carbo, S., Humphrey, D., Maudos, J. and Molyneux, P. (2009). Cross-country

comparisons of competition and pricing power in European banking, Journal of

International Money and Finance, vol. 28, no. 1, pp. 115-134.

Carletti, E. (2006). Competition and regulation in banking. In: Boot, A., Thakor, A.

(Eds.), Handbook of Corporate Finance, Financial Intermediation and Banking,

North Holland, London.

Carletti, E. and Hartmann, P. (2003). Competition and financial stability: what‟s

special about banking? In: Mizen P (ed) Monetary history, exchange rates and

346

financial markets: essays in honour of Charles Goodhart, vol. 2. Edward Elgar,

Cheltenham, UK.

Casu, B. and Girardone, C. (2005). An analysis of the relevance of off-balance sheet

items in explaining productivity change in European banking, Applied Financial

Economics, vol. 15, no. 15, pp. 1053-1061.

Casu, B. and Girardone, C. (2006). Bank competition, concentration and efficiency in

the single European market, The Manchester School, vol. 74, no. 4, pp. 441-468.

Casu, B. and Girardone, C. (2009). Competition issues in European banking, Journal

of Financial Regulation and Compliance, vol. 17, no. 2, pp.119 – 133.

Casu, B. and Girardone, C. (2010). Integration and efficiency convergence in EU

banking markets. Omega, vol. 38, no. 5, pp. 260-267.

Casu, B. and Molyneux, P. (2003). A comparative study of efficiency in European

banking, Applied Economics, vol. 35, no. 17, pp. 1865-1876.

Casu, B., Girardone, C. and Molyneux, P. (2004). Productivity change in European

banking: A comparison of parametric and non-parametric approaches, Journal of


Cavallo, L. and Rossi, S. P. (2001). Scale and scope economies in the European

banking systems, Journal of multinational financial management, vol. 11, no. 4, pp.

515-531.

Cazals, C., Florens, J.P. and Simar, L. (2002). Nonparametric Frontier Estimation: A

Robust Approach. Journal of Econometrics, vol. 106, no. 1, pp. 1–25.

Cerqueiro, G. (2008). Bank concentration, credit quality and loan rates, CentER

mimeo.

Cetorelli, N. (1999). Competitive analysis in banking: Appraisal of the

methodologies, Economic perspectives, Federal Reserve Bank of Chicago, pp. 2-15.

Cetorelli, N. (2004). Real effects of bank competition, Federal Reserve Bank of

Chicago working paper no. 00-09.

347

Cetorelli, N. and Gamberra, M. (2001). Banking market structure, financial

dependence and growth: International evidence from industry data, Journal of

Finance, vol. 56, pp. 617–648.

Chan, Y.S., Greenbaum, S. and Thakor, A. (1986). Information reusability,

competition and bank asset quality, Journal of Banking and Finance, vol. 10, pp. 243-

53.

Chang, S.C., Chang, J.C.D. and Huang, T.H. (2012). Assessing market power in the

US commercial banking industry under deregulation, Economic Modeling, vol. 29,

no. 5, pp. 1558-1565.

Charnes, A. and Cooper, W.W. (1962). Programming with linear fractional

functionals, Naval Research Logistics Quarterly, vol. 9, pp. 67–88.

Charnes, A., Clarke, C., Cooper, W.W. and Golany, B. (1985a). A development study

of DEA in measuring the effect of maintenance units in the U.S. Air Force, Annals of

Operations Research, vol. 2, pp. 95–112.

Charnes, A., Cooper, W.W. and Rhodes, E.L. (1978). Measuring the efficiency of

decision making units, European Journal of Operational Research, vol. 2, pp. 429–

444.

Charnes, A., Cooper, W.W., Golany, B., Seiford, L.M. and Stutz, J. (1985b).

Foundations of data envelopment analysis and Pareto–Koopmans empirical

production functions, Journal of Econometrics, vol. 30, pp. 91– 107.

Charnes, A., Cooper, W.W., Huang, Z.M. and Sun, D.B. (1990). Polyhedral cone-

ratio DEA models with an illustrative application to large commercial banks, Journal

of Econometrics, vol. 46, pp. 73–91.

Chen, Y., Seiford, L.M. and Zhu, J. (2002). Imprecise data envelopment analysis,

Management Science.

Chen, Y.Y. and Yang, W.L. (2012). Financial Market Development and Competition

Degree in China's Banking Industry - An Empirical Study Based on Panzar-Rosse

Model, Journal of Dalian University of Technology (Social Sciences), 1, 003.

348

Chenard, K., Das, U.S. and Quintyn, M. (2004). Does Regulatory Governance Matter

for Financial System Stability? An Empirical Analysis (EPub). International

Monetary Fund.

Chortareas, G.E., Garza-Garcia, J.G. and Girardone, C. (2012). Competition,

Efficiency and Interest Rate Margins in Latin American Banking. International

Review of Financial Analysis, vol. 24, pp. 93-103.

Chortareas, G.E., Girardone, C. and Ventouri, A. (2009). Efficiency and productivity

of Greek banks in the EMU era, Applied Financial Economics, vol. 19, no. 16, pp.

1317-1328.

Christensen, L.R., Jorgenson, D.W. and Lau, L. J. (1973). Transcendental logarithmic

production frontiers, The review of economics and statistics, vol. 55, no. 1, pp. 28-45.

Christopoulos, D.K., Lolos, S.E. and Tsionas, E.G. (2002). Efficiency of the Greek

banking system in view of the EMU: a heteroscedastic stochastic frontier approach,

Journal of Policy Modeling, vol. 24, no. 9, pp. 813-829.

Cihak, M. and Hesse, H. (2007). Cooperative Banks and Financial Stability. IMF

Working Paper No. 2, Washington D.C.

Cipollini, A. and Fiordelisi, F. (2009). The Impact of Bank Concentration on

Financial Distress: The Case of the European Banking System (November 30, 2009).

EMFI Working Paper No. 2. Available at SSRN: http://ssrn.com/abstract=1578718

Claessens, S. (2009). Competition in the financial sector: overview of competition

policies, The World Bank Research Observer, vol. 24, no. 1, pp. 83-118.

Claessens, S. and Laeven, L. (2004). What drives bank competition? Some

international evidence, Journal of Money, Credit and Banking, vol. 36, no.2, pp. 563-

583.

Claessens, S., Demirgüç-Kunt, A. and Huizinga, H. (2001a). How does foreign entry

affect domestic banking markets?, Journal of Banking and Finance, vol. 25, no. 5, pp.

891-911.

http://ssrn.com/abstract=1578718

349

Claessens, S., Klingebiel, D. and Laeven, L. (2001b). Financial restructuring in

banking and corporate sector crises: What policies to pursue? (No. 8386), National

Bureau of Economic Research.

Claeys, S. and Vennet, R.V. (2008). Determinants of bank interest margins in Central

and Eastern Europe: A comparison with the West, Economic Systems, vol. 32, no. 2,

pp. 197-216.

Coccorese, P. (2004). Banking competition and macroeconomic conditions: A

disaggregate analysis, Journal of Financial Markets, Institutions and Money, vol. 14,

pp. 203-219.

Coccorese, P. (2005). Competition in markets with dominant firms: A note on the

evidence from the Italian banking industry, Journal of Banking and Finance, vol. 29,

no. 5, pp. 1083-1093.

Coccorese, P. (2009). Market power in local banking monopolies, Journal of Banking

and Finance, vol. 33, no. 7, pp. 1196-1210.

Coccorese, P. (2012). Information sharing, market competition and antitrust

intervention: a lesson from the Italian insurance sector, Applied Economics, vol. 44,

no. 3, pp. 351-359.

Coccorese, P. and Pellecchia, A. (2010). Testing the „quiet life‟ hypothesis in the

Italian banking industry, Economic Notes, vol. 39, no. 3, pp. 173-202.

Coelli, T. and Perelman, S. (2000). Technical Efficiency of European Railways: A

Distance Function Approach, Applied Economics, vol. 32, no. 15, pp. 1967–76.

Cole, R.A., McKenzie, J. A. and White, L.J. (1995). Deregulation gone awry: Moral

hazard in the savings and loan industry, The Causes and Costs of Depository

Institution Failures, pp. 29-73.

Conrad, A., Neuberger, D. and Trigo Gamarra, L. (2009). The impact of regional and

demographic factors on the efficiency of German savings banks (No. 111), Thünen-

series of applied economic theory.

Cook W.D. and Seiford L.M. (2009). Data envelopment analysis (DEA) - thirty years

350

on, European Journal of Operational Research, vol.192, pp. 1–17.

Cook, W.D. and Green, R.H. (2005). Evaluating power plant efficiency: A

hierarchical model, Computers and Operations Research, vol. 32, pp. 813–823.

Cook, W.D. and Zhu, J. (2007). Classifying inputs and outputs in data envelopment

analysis, European Journal of Operational Research, vol. 180, no. 2, pp. 692–699.

Cook, W.D. and Zhu, J. (2008). CAR-DEA: Context dependent assurance regions in

DEA, Operations Research, vol. 56, no. 1, pp. 69-78.

Cook, W.D., Kress, M. and Seiford, L.M. (1993). On the use of ordinal data in data

envelopment analysis, Journal of the Operational Research Society, vol. 44, pp. 133–

140.

Cook, W.D., Kress, M. and Seiford, L.M. (1996). Data envelopment analysis in the

presence of both quantitative and qualitative factors, Journal of the Operational

Research Society, vol. 47, pp. 945–953.

Cook, W.D., Liang, L., Zha, Y. and Zhu, J. (2008). A modified super-efficiency DEA

model for infeasibility, Journal of the Operational Research Society, vol. 60, pp. 276-

281.

Cooper, R.J. and McLaren, K.R. (1996). A system of demand equations satisfying

effectively global regularity conditions, Review of Economics and Statistics, vol. 78,

pp. 359–364.

Cooper, W.W., Deng, H., Huang, Z.M. and Li, S.X. (2002). Change constrained

programming approaches to technical efficiencies and inefficiencies in stochastic data

envelopment analysis, Journal of the Operational Research Society, vol. 53, pp.

1347– 1356.

Cooper, W.W., Huang, Z. and Li, S. (1996). Satisficing DEA models under chance

constraints, Annals of Operations Research, vol. 66, pp. 279–295.

Cooper, W.W., Huang, Z. and Li, S. (2004). Chance constraint DEA. In: Cooper,

W.W., Seiford, L.M., Zhu, J. (Eds.), Handbook on Data Envelopment Analysis,

Kluwer Academic Publishers, Norwell, MA.

351

Cooper, W.W., Park, K.S. and Yu, G. (1999). IDEA and AR-IDEA: Models for

dealing with imprecise data in DEA, Management Science, vol. 45, pp. 597– 607.

Cooper, W.W., Seiford, L.M. and Tone, K. (2006). Introduction to Data Envelopment

Analysis and its Uses, Springer Science.

Cordella, T. and Yeyati, E.L. (2002). Financial opening, deposit insurance, and risk in

a model of banking competition, European Economic Review. vol. 46, pp. 471–485.

Cornwell, C., Schmidt, P. and Sickles, R.C. (1990). Production frontiers with cross-

sectional and time-series variation in efficiency levels, Journal of econometrics, vol.

46, no. 1, pp. 185-200.

Corvoisier, S. and Gropp, G. (2002). Bank concentration and retail interest rates,

Journal of Banking and Finance, vol. 26, no. 11, pp. 2155-2189.

Cowling, K. (1976). On the theoretical specification of industrial structure-

performance relationships, European Economic Review, vol. 8, no. 1, pp. 1-14.

Cowling, K. and Waterson, M. (1976). Price-cost margins and market structure,

Economica, vol. 43, pp. 267-274.

Cummins, J.D. and Zi, H. (1998). Comparison of frontier efficiency methods: An

application to the US life insurance industry, Journal of Productivity Analysis, vol.

10, no. 2, pp. 131-152.

Cybo-Ottone, A. and Murgia, M. (2000). Mergers and shareholder wealth in European

banking, Journal of Banking and Finance, vol. 24, no. 6, pp. 831-859.

Daley, J. and Matthews, K. (2012). Competitive conditions in the Jamaican banking

market 1998–2009, International Review of Financial Analysis, vol. 25, pp. 131-135.

Daouia, A., and Simar, L. (2007), Nonparametric efficiency analysis: A multivariate

conditional quantile approach, Journal of Econometrics, vol. 140, pp. 375–400.

Das, A. and Ghosh, S. (2009). Financial deregulation and profit efficiency: a

nonparametric analysis of Indian banks, Journal of Economics and Business, vol. 61,

no. 6, pp. 509-528.

352

Davies, St. (1979). Choosing between concentration indices: The Iso-concentration

curve, Economica, New series, vol. 46, no. 181, pp. 67-75.

De Bandt, O. and Davis, P. (2000). Competition, contestability and market structure

in European banking sectors on the eve of EMU, Journal of Banking and Finance,

vol. 24, pp. 1045-1066.

De Guevara, J.F. and Maudos, J. (2002). Inequalities in the efficiency of the banking

sectors of the European Union, Applied Economics Letters, vol. 9, no. 8, pp. 541-544.

De Guevara, J.F., Maudos, J. and Perez, F. (2005). Market power in European

banking sectors, Journal of Financial Services Research, vol. 27, no. 2, pp. 109-137.

De Guevara, J.F., Maudos, J. and Perez, F. (2007). Integration and competition in the

European financial markets, Journal of International Money and Finance, vol. 26, no.

1, pp. 26-45.

De Nicolo G. and Loukoianova, E. (2007). Bank ownership, market structure, and

risk, IMF Working Paper, Washington, D.C.

De Nicolo, G. and Lucchetta, M. (2009). Financial intermediation, competition, and

risk: A general equilibrium exposition, IMF Working Papers 09/105.

De Pinho, P.S. (2001). Using accounting data to measure efficiency in banking: an

application to Portugal, Applied Financial Economics, vol. 11, no. 5, pp. 527-538.

De Rozas, L.G. (2007). Testing for competition in the Spanish banking industry: The

Panzar-Rosse approach revisited (No. 0726), Banco de Espana.

Degryse, H. and Ongena, S. (2008). Competition and regulation in the banking sector:

a review of the empirical evidence on the sources of bank rents, Handbook of

financial intermediation and banking, 2008, pp. 483-554.

Degryse, H., Ongena, S. (2001). Bank relationships and firm profitability. Financial

Management, vol. 30, pp. 9–34.

Degryse, H., Ongena, S. (2005). Distance, lending relationships and competition.

Journal of Finance, vol. 60, pp. 231–266.

353

Delis, M.D. (2010). Competitive conditions in the Central and Eastern European

banking systems, Omega, vol. 38, no. 5, pp. 268-274.

Delis, M.D. and Papanikolaou, N.I. (2009). Determinants of bank efficiency: evidence

from a semi-parametric methodology, Managerial Finance, vol. 35, no. 3, pp. 260-

275.

Delis, M.D., Koutsomanoli-Fillipaki, A., Staikouras, C.K. and Katerina, G. (2009).

Evaluating cost and profit efficiency: a comparison of parametric and nonparametric

methodologies, Applied Financial Economics, vol. 19, no. 3, pp. 191-202.

Delis, M.D., Staikouras, K.C. and Varlagas, P.T. (2008). On the measurement of

market power in the banking industry, Journal of Business Finance and Accounting,

vol. 35, no. 7-8, pp. 1023-1047.

Dell‟ Ariccia, G. (2000). Learning by lending, competition, and screening incentives

in the banking industry. Center for Financial Institutions working paper no. 00-10,

Wharton School Center for Financial Institutions, University of Pennsylvania.

Deltuvaitė, V., Vaškelaitis, V. and Pranckeviciute , A. (2007). The impact of

concentration on competition and efficiency in the Lithuanian banking sector,

Engineering Economics, vol. 4, no. 54, pp. 7-19.

Demirgüç-Kunt, A. and Detragiache, E. (1998). The determinants of banking crises in

developing and developed countries, IMF Staff Papers, vol. 45, no.1, pp. 81–109,

Washington, D. C.: International Monetary Fund.

Demirgüç-Kunt, A. and Detragiache, E. (2002). Does deposit insurance increase

banking system stability? An empirical investigation, Journal of Monetary

Economics, vol. 49, pp. 1373–1406.

Demirgüç-Kunt, A. and Huizinga, H. (1999). Determinants of commercial bank

interest margins and profitability: Some international evidence. The World Bank

Economic Review, vol. 13, no. 2, pp. 379-408.

Demirguc-Kunt, A. and Martinez Peria, M. (2010). A framework for analyzing

competition in the banking sector: an application to the case of Jordan, World Bank

354

Policy Research Working Paper Series, no. 5499.

Demsetz, H. (1973). Industry structure, market rivalry and public policy, Journal of

Law and Economics, vol. 16, pp. 1-10.

Deng, W.-S. and Huang T.-H. (2008). A Semiparametric Approach to the Estimation

of the Stochastic Frontier Model with Time-Variant Technical Efficiency, Academia

economic papers, vol. 36, no. 2, pp. 167–193.

Deprins, L., Simar, L. and Tulkens, H. (1984). Measuring labour efficiency in post

offices. In: Marchand, M., Pestieau, P., Tulkens, H. (Eds.), The Performance of Public

Enterprises: Concepts and Measurement, North Holland, Amsterdam, pp. 243–267.

Dewatripont, M. and Maskin, E. (1995). Credit and Efficiency in centralized and

decentralized economies, The Review of Economic studies, vol. 62, no. 4, pp. 541-

555.

Diamond, D.W. (1984). Financial intermediation and delegated monitoring, Review of

Economic Studies, vol. 51, pp. 393-414.

Dickson, V.A. (1981). Conjectural variation elasticities and concentration, Economics

Letters, vol. 7, pp. 281-285.

Dietsch, M. (1993). Economies of scale and scope in French commercial banking

industry, Journal of productivity Analysis, vol. 4, no. 1, pp. 35-50.

Dietsch, M. and Lozano-Vivas, A. (2000). How the environment determines banking

efficiency: A comparison between French and Spanish industries, Journal of Banking

and Finance, 24, no. 6, pp. 985-1004.

Dimova, D.D. (2005). The Benefits of Privatizing Banks to Strategic Foreign

Investors: A Survey of Central and Eastern Europe, Undergraduate Economic

Review, vol. 2, no. 1.

Doku, J.N., Abor, J., Adjasi, C.K. and Andoh, C. (2012). Assessing Competitive

Behaviour in Emerging Banking Market: African Evidence. Research in Accounting

in Emerging Economies, vol. 12, pp. 25-51.

355

Doyle, J. and Green, R. (1994). Efficiency and cross efficiency in DEA: Derivations,

meanings and the uses, Journal of the Operational Research Society, vol. 45, no. 5,

pp. 567–578.

Drake, L. (2001). Efficiency and productivity change in UK banking. Applied

Financial Economics, vol. 11, no. 5, pp. 557-571.

Drakos, K. and Konstantinou, P. (2005). Competition and contestability in transition

banking: An empirical analysis, South-Eastern Europe Journal of Economics, vol. 2,

pp. 183-209.

Drummond, P., Maechler, A.M. and Marcelino, S. (2007). Italy: Assessing

competition and efficiency in the banking system, International Monetary Fund

Working Paper 07/26.

Efron, B. (1979). Bootstrap methods: another look at the jackknife, The annals of

Statistics, vol. 7, no. 1, pp. 1-26.

Egger, P. and Hahn, F.R. (2010). Endogenous bank mergers and their impact on

banking performance: Some evidence from Austria, International Journal of

Industrial Organization, vol. 28, no. 2, pp. 155-166.

Elbadawi, I., Gallant, A.R. and Souza, G. (1983). An elasticity can be estimated

consistently without a priori knowledge of functional form, Econometrica: Journal of

the Econometric Society, vol. 51, no. 6, pp. 1731-1751.

Elyasiani, E., Goldberg, L. (2004). Relationship lending: A survey of the literature,

Journal of Economics and Business, vol. 56, pp. 315–330.

Evanoff, D.D. and Ors, E. (2008). The competitive dynamics of geographic

deregulation in banking: Implications for productive efficiency, Journal of Money,


Evanoff, D.D., Israilevich, P.R. and Merris, R.C. (1990). Relative price efficiency,

technical change, and scale economies for large commercial banks, Journal of

Regulatory Economics, vol. 2, no. 3, pp. 281-298.

Fadzlan S. and Muzafar S.H. (2013). Financial Sector Consolidation and Competition

356

in Malaysia: An Application of the Panzar-Rosse Method, Journal of Economic

Studies, vol. 40, no. 3.

Fan, Y., Li, Q. and Weersink, A. (1996). Semiparametric estimation of stochastic

production frontier models, Journal of Business and Economic Statistics, vol. 14, pp.

460–468.

Fare R., Martins-Filho C. and Vardanyan M. (2010). On functional form

representation of multi-output production technologies, Journal of Productivity

analysis, vol. 33, no. 2, pp. 81-96.

Fare, R. and Grosskopf, S. (1996). Intertemporal Production Frontiers: With Dynamic

DEA, Kluwer Academic, Boston, MA.

Fare, R.S. and Lovell, C.A.K. (1978). Measuring the technical efficiency of

production, Journal of Economic Theory, vol. 19, pp. 150–162.

Fare, R.S., Grosskopf, S. and Lovell, C.A.K. (1994). Production Frontiers, Cambridge

University Press.

Farinha, L.A., Santos, J.A.C. (2002). Switching from single to multiple bank lending

relationships: Determinants and implications. Journal of Financial Intermediation,

vol. 11, pp. 124–151.

Fernandez, A.I. and Gonzalez, F. (2005). How accounting and auditing systems can

counteract risk-shifting of safety-nets in banking: Some international evidence,

Journal of Financial Stability, vol. 1, no. 4, pp. 466-500.

Fernandez, C., Koop, G. and Steel, M.F.J. (2000). A Bayesian Analysis of Multiple-

Output Production Frontiers, Journal of Econometrics, vol. 98, pp. 47–79.

Fernandez, C., Koop, G. and Steel, M.F.J. (2002). Multiple-Output Production with

Undesirable Outputs: An Application to Nitrogen Surplus in Agriculture, Journal of

the American Statistical Association, Applications and Case Studies, vol. 97, pp. 432–

442.

Fernandez, C., Osiewalski, J. and Steel, M.F.J. (1997). On the Use of Panel Data in

Stochastic Frontier Models with Improper Priors, Journal of Econometrics, vol. 79,

357

pp. 169–193.

Ferrier, G.D. and Hirschberg, J.G. (1997). Bootstrapping Confidence Intervals for

Linear Programming Efficiency Scores: With an Illustration Using Italian Bank Data,

Journal of Productivity Analysis, vol. 8, pp. 19–33.

Ferrier, G.D. and Lovell, C.A. (1990). Measuring cost efficiency in banking:

econometric and linear programming evidence, Journal of Econometrics, vol. 46,

no.1, pp. 229-245.

Figueira, C. and Nellis, J. (2009). Bank merger and acquisitions activity in the EU:

much ado about nothing?, The Service Industries Journal, vol. 29, no. 7, pp. 875-886.

Figueira, C., Nellis, J.G. and Parker, D. (2007). Challenges facing the polish banking

industry: A comparative study with UK banks, Managing Global Transitions, vol. 5,

no. 1, pp. 25-44.

Fiorentino, E., Karmann, A. and Koetter, M. (2006). The cost efficiency of German

banks: a comparison of SFA and DEA, Available at SSRN 947340.

Fitzpatrick, T. and McQuinn, K. (2008). Measuring bank profit efficiency, Applied


Frei, F. and Harker, P. (1999). Projections onto efficient frontiers: Theoretical and

computational extensions to DEA, Journal of Productivity Analysis, vol. 11, pp. 275–

300.

Fried, H. O., Lovell, C. A. K. and Schmidt, S. (eds.). (1993). The Measurement of

Productive Efficiency: Techniques and Applications, New York: Oxford University

Press.

Fries, S., Neven, D. and Seabright, P. (2002). Bank performance in transition

economies.

Fung, M.K. (2006). Scale economies, X-efficiency, and convergence of productivity

among bank holding companies, Journal of banking and finance, vol. 30, no. 10, pp.

2857-2874.

358

Gaertner, M. and Sanya, S. (2012). Assessing Bank Competition within the East

African Community, IMF working paper series, WP/12/32.

Gajurel, D.P. and Pradhan, R.S. (2012). Concentration and competition in Nepalese

banking, Journal of Business, Economics, vol. 1, no. 1, pp. 5-16.

Gallant, A.R. (1982). Unbiased determination of production technologies, Journal of

Econometrics, vol. 20, no. 2, pp. 285-323.

Gallant, A.R. (1981). On the bias in flexible functional forms and an essentially

unbiased form: the Fourier flexible form, Journal of Econometrics, vol. 15, no. 2, pp.

211-245.

Gallant, A.R. and Souza, G. (1991). On the asymptotic normality of Fourier flexible

form estimates, Journal of Econometrics, vol. 50, no. 3, pp. 329-353.

Gathon, H.J. and Perelman, S. (1992). Measuring technical efficiency in European

railways: a panel data approach, Journal of Productivity Analysis, vol. 3, no. 1, pp.

135-151.

Gelfand, M.D. and Spiller, P.T. (1987). Entry barriers and multiproduct oligopolies:

Do they forebear or spoil?, International Journal of Industrial Organization, vol. 5,

no. 1, pp. 101-113.

Gilbert, R. (1984). Bank market structure and competition – A survey, Journal of

Money, Credit and Banking, vol. 16, pp. 617-645.

Gilligan, T., Smirlock, M. and Marshall, W. (1984). Scale and scope economies in the

multi-product banking firm, Journal of Monetary Economics, vol. 13, no. 3, pp. 393-

405.

Gilligan, T.W. and Smirlock, M.L. (1984). An empirical study of joint production and

scale economies in commercial banking, Journal of Banking and Finance, vol. 8, no.

1, pp. 67-77.

Giokas, D.I. (2008a). Assessing the efficiency in operations of a large Greek bank

branch network adopting different economic behaviors, Economic Modeling, vol. 25,

no. 3, pp. 559-574.

359

Giokas, D.I. (2008b). Cost efficiency impact of bank branch characteristics and

location: An illustrative application to Greek bank branches, Managerial Finance,

vol. 34, no. 3, pp. 172-185.

Gischer, H. and Stiele, M. (2009), Competition Tests with a Non-Structural Model:

the Panzar–Rosse Method Applied to Germany's Savings Banks. German Economic

Review, vol. 10, pp. 50–70.

Giustiniani, A. and Ross, K. (2008). Bank competition and efficiency in the FYR

Macedonia, South-Eastern Europe Journal of Economics, vol. 2, pp. 145-167.

Glass, J.C. and McKillop, D.G. (1992). An empirical analysis of scale and scope

economies and technical change in an Irish multiproduct banking firm, Journal of


Glass, J.C., McKillop, D.G. and Morikawa, Y. (1998). Intermediation and value-

added models for estimating cost economies in large Japanese banks 1977-93, Applied

financial economics, vol. 8, no. 6, pp. 653-661.

Goddard, J. and John Wilson, O.S. (2009). Competition in banking: A disequilibrium

approach, Journal of Banking and Finance, vol. 33, no. 12, pp. 2282-2292.

Goddard, J., Molyneux, P. and Wilson, J.O.S. (2001). European Banking: Efficiency,

Technology and Growth, John Wiley, Chichester, UK.

Goetz, M. (2010). Bank Organization, Market Structure and Risk Taking: Theory and

Evidence from U.S. Commercial Banks, Working Paper, Brown University.

Gondat-Larralde, C. and Lepetit, L. (2001). 19. The impact of market structure and

efficiency on bank profitability: an empirical analysis of banking industries in Central

and Eastern Europe. Financial and Monetary Integration in the New Europe:

Convergence Between the Eu and Central and Eastern Europe, 416.

Gonzalez-Hermosillo, B., Pazarbasioglu, C. and Billings, R. (1997). Banking system

fragility: Likelihood versus timing of failure - An application to the Mexican financial

crisis, IMF Staff Papers.

Gonzalez, F. (2009). Determinants of Bank‐Market Structure: Efficiency and Political

360

Economy Variables, Journal of Money, Credit and Banking, vol. 41, no. 4, pp. 735-

754.

Green, C.J., Murinde, V. and Nikolov, I. (2004). The efficiency of foreign and

domestic banks in Central and Eastern Europe: Evidence on economies of scale and

scope, Journal of emerging market finance, vol. 3, no. 2, pp. 175-205.

Green, R., Doyle, J. and Cook, W.D. (1996). Efficiency bounds in data envelopment

analysis, European Journal of Operational Research, vol. 89, pp. 482–490.

Green, R.H., Cook, W.D. and Doyle, J. (1997). A note on the additive data

envelopment analysis model, Journal of the Operational Research Society, vol. 48,

no. 4, pp. 446–448.

Greenberg, J. B. and Simbanegavi, W. (2009, November). Testing for Competition in

the South African Banking Sector. In Faculty of Commerce University of Cape Town,

(November).

Grigorian, D.A. and Manole, V. (2002). Determinants of commercial bank

performance in transition: an application of data envelopment analysis, World Bank

Policy Research Working Paper, (2850).

Gropp. R. and Kashyap A. (2009). A new metric for banking integration in Europe,

Centre for European Economic research, Discussion Paper No. 08-102.

Grosskopf, S. (1996). Statistical Inference and Nonparametric Efficiency: A Selective

Survey, Journal of Productivity Analysis, vol. 7, pp. 161–176.

Gruben, W.C. and Koo, J. (1997). Contestability and Capital Flows in Argentina‟s

Banking System (unpublished; Dallas, Texas: Federal Reserve Bank of Dallas).

Gual, J. (1999). Deregulation, integration and market structure in European banking.

European Investment Bank Papers, vol. 4, pp. 35–48.

Habte, Y. (2012). Competitive Conditions in the Swedish Banking System.

Hahn, F.R. (2008). Testing for profitability and contestability in banking: evidence

from Austria. International Review of Applied Economics, vol. 22, no. 5, pp. 639-653.

361

Hagendorff, J. and Keasey, K. (2009). Post‐merger strategy and performance:

evidence from the US and European banking industries, Accounting and Finance, vol.

49, no. 4, pp. 725-751.

Hahn, F.R. (2007). Domestic mergers in the Austrian banking sector: a performance

analysis, Applied Financial Economics, vol. 17, no. 3, pp. 185-196.

Hailu, A. and Veeman, T. (2000). Environmentally Sensitive Productivity Analysis of

the Canadian Pulp and Paper Industry, 1959–1994: An Input Distance Function,

Journal of Environmental Economics and Management, vol. 40, no. 3, pp. 251–274.

Haldane, A., Hoggarth, G., Saporta, V. and Sinclair, P. (2004). “Financial Stability

and Bank Solvency,” Chapter 5 from the book “Systemic Financial Crises. Resolving

Large Bank Insolvencies,” editors: Evanoff, D., and Kaufman, G., World Scientific

Publishing (2005), London.

Haldane, A.G. (2004), Defining monetary and financial stability, Bank of England

mimeo.

Hall M. and Tideman N. (1967). Measures of concentration, American Statistical

Association Journal, vol. 62, no. 317, pp. 162-168.

Hamza, R.A. Validation Panzar-Rosse Model in determining the structural

characteristics of Tunisian banking industry, Journal of Economics and International

Finance, vol. 3, no. 5, pp. 259-268.

Hannah, L. and Kay, J.A. (1977). Concentration in modern industry, MacMillan

Press, London.

Hannan, T.H. and Liang, J.N. (1993). Inferring market power from time-series data:

The case of the banking firm, International Journal of Industrial Organization, vol.

11, no. 2, pp. 205-218.

Hart, P.E. (1975). Moment distribution in Economics: An exposition, the Journal of

the Royal Statistical Society, series A, vo. 138, pp. 423-434.

Hasan, I. and Marton, K. (2003). Development and efficiency of the banking sector in

a transitional economy: Hungarian experience, Journal of Banking and Finance, vol.

362

27, no. 12, pp. 2249-2271.

Hause, J.C. (1977). The measurement of concentrated industrial structure and the size

distribution of firms, Annals of Economic and Social Measurement, vol. 6, pp. 73-

107.

Havrylchyk, O. (2006). Efficiency of the Polish banking industry: Foreign versus

domestic banks, Journal of Banking and Finance, vol. 30, no. 7, pp. 1975-1996.

Hawtrey, K. and Liang, H. (2008). Bank interest margins in OECD countries, The

North American Journal of Economics and Finance, 1vol. 9, no. 3, pp. 249-260.

Heffernan, S.A. (2002). How do UK financial institutions really price their banking

products?, Journal of Banking and Finance, vol. 26, no. 10, pp. 1997–2016.

Heffernan, S.A. (2005). Modern Banking, Wiley, London.

Hellman, T.F., Murdock K. and Stiglitz J. E. (2000), Liberalisation, moral hazard in

banking and prudential regulation: Are capital requirements enough?, American

Economic Review, vol. 90, no. 1, pp. 147-165.

Hempell, H. (2002). Testing for competition among German banks.

Hoggarth, G.A., Milne, A. and Wood, G.E. (1998). Alternative routes to banking

stability: A comparison of UK and German banking systems, Financial Stability

Review, vol. 5, pp. 55-68.

Holló, D. and Nagy, M. (2006). Bank efficiency in the enlarged European Union. BIS

Papers, no. 28, pp. 217-235.

Hondroyiannis, G., Lolos, S. and Papapetrou, E. (1999). Assessing competitive

conditions in the Greek banking system, Journal of International Financial Markets,

Institutions and Money, vol. 9, pp. 377-391.

Horvath, J. (1970). Suggestion for a comprehensive measure of concentration,

Southern Economic Journal, vol. 36, no. 4, pp. 446-452.

Houston, J.F. and James, C.M. (2001). Do relationships have limits? Banking

relationships, financial constraints and investment, Journal of Business, vol. 74, pp.

363

347–374.

Huang, H.C. (2004). Estimation of technical inefficiencies with heterogeneous

technologies, Journal of Productivity Analysis, vol. 21, no. 3, pp. 277-296.

Huang, T.H. and Wang, M.H. (2002). Comparison of economic efficiency estimation

methods: Parametric and non–parametric techniques, The Manchester School, vol. 70,

no. 5, pp. 682-709.

Hughes, J. P. and Mester, L.J. (1998). Bank capitalization and cost: Evidence of scale

economies in risk management and signaling, Review of Economics and Statistics,

vol. 80, no. 2, pp. 314-325.

Huizinga, H.P., Nelissen, J.H. and Vennet, R.V. (2001). Efficiency Effects of Bank

Mergers and Acquisitions (No. TI 01-088/3), Tinbergen Institute.

Humphrey D.B. (1992). Flow versus stock indicators in bank output: Effects on

Productivity and scale economy measurement, Journal of financial services research,

vol. 6, pp. 115-135.

Humphrey, D.B. and Pulley, L.B. (1997). Banks' responses to deregulation: Profits,

technology, and efficiency, Journal of Money, Credit, and Banking, vol. 29, no.1, pp.

73-93.

Humphrey, D.B. and Vale, B. (2004). Scale economies, bank mergers, and electronic

payments: A spline function approach, Journal of Banking and Finance, vol. 28, no.

7, pp. 1671-1696.

Hussain, M. and Hassan, M. (2012). Competition, Risk Taking and Efficiency in the

US Commercial Banks Prior to 2008 Financial Crisis. Available at SSRN 2003066.

Hussain, M.E. and Mustapha, N. (2010). Determinants of Bank Competition (March

1, 2010). Available at SSRN: http://ssrn.com/abstract=2004746 or http://dx.doi.org/

10.2139/ssrn.2004746

Iannotta, G., Nocera, G. and Sironi, A. (2007). Ownership structure, risk and

performance in the European banking industry, Journal of Banking and Finance, vol.

31, no. 7, pp. 2127-2149.

364

Isik, I. and Hassan, M.K. (2002). Technical, scale and allocative efficiencies of

Turkish banking industry, Journal of Banking and Finance, vol. 26, no. 4, pp. 719-

766.

Iwata, G. (1974). Measurement of conjectural variations in oligopoly, Econometrica,

vol. 42, pp. 947-966.

Jagtiani, J., Nathan, A. and Sick, G. (1995). Scale economies and cost

complementarities in commercial banks: On-and off-balance-sheet activities, Journal

of Banking and Finance, vol. 19, no. 7, pp. 1175-1189.

Jemric, I. and Vujcic, B. (2002). Efficiency of banks in Croatia: A DEA approach.

Comparative Economic Studies, vol. 44, no. 2-3, pp. 169-193.

Jeong, S.-O. and Simar, L. (2006). Linearly interpolated FDH efficiency score for

nonconvex frontiers, Journal of Multivariate Analysis, vol. 97, no. 10, pp. 2141-2161.

Jimborean, R. and Brack, E. (2010). The cost-efficiency of French banks.

Jiménez, G., Lopez, J. and Saurina Salas, J. (2010). How does competition impact

bank risk-taking?.

John, K., Litov, L. and Yeung, B. (2008). Corporate Governance and Risk‐Taking,

The Journal of Finance, vol. 63, no. 4, pp. 1679-1728.

Kamberoglou, N.C., Liapis, E., Simigiannis, G.T. and Tzamourani, P. (2004). Cost

efficiency in Greek banking, Bank of Greece WP.

Karasulu, M. (2007). Competition in the Chilean Banking Sector: A Cross-Country

Comparison, Economia, vol. 7, no. 2, pp. 1-32.

Kasman, A. (2010). Consolidation and Competition in the Banking Industries of the

EU Member and Candidate Countries, Emerging Markets Finance and Trade, vol. 46,

no. 6, pp. 121-139.

Kasman, A. and Turgutlu, E. (2008). Competitive Conditions in the Turkish Non-Life

Insurance Industry, Review of Middle East Economics and Finance, vol. 4, no. 1, pp.

81-96.

365

Kasman, A. and Yildirim, C. (2006). Cost and profit efficiencies in transition

banking: the case of new EU members, Applied Economics, vol. 38, no. 9, pp. 1079-

1090.

Kaufmann, D., Kraay, A. and Mastruzzi, M. (2009). Governance matters VIII:

aggregate and individual governance indicators, 1996-2008, World bank policy

research working paper, (4978).

Keeley, M. (1990). Deposit insurance, risk and market power in banking. American

Economic Review, vol. 80, pp. 1183–1200.

Khiabani, N. and Hamidisahneh, M. (2012). The effects of entry regulation on bank

competition: The case of the Iranian banking industry, Journal of Applied Economics,

vol. 15, no. 1, pp. 119-137.

Kim S.H., Park C.G. and Park K.S. (1999). An application of data envelopment

analysis in telephone offices evaluation with partial data, Computers and Operations

Research, vol. 26, pp. 59–72.

Kim, J.Y. (2002). Limited information likelihood and Bayesian analysis, Journal of

Econometrics, vol. 107, pp. 175–193.

Kim, Y. and Schmidt, P. (2000). A Review and Empirical Comparison of Bayesian

and Classical Approaches to Inference on Efficiency Levels in Stochastic Frontier

Models with Panel Data, Journal of Productivity Analysis, vol. 14, pp. 91–118.

Kneip, A. and Simar, L. (1996). A General Framework for Frontier Estimation with

Panel Data, Journal of Productivity Analysis, vol. 7, pp. 161–176.

Kneip, A., Park, B.U. and Simar, L. (1998). A Note on the Convergence of

Nonparametric DEA Estimators for Production Efficiency Scores, Econometric

Theory, vol. 14, pp. 783–793.

Koetter, M., Kolari, J.W. and Spierdijk, L. (2012). Enjoying the quiet life under

deregulation? Evidence from adjusted Lerner indices for US banks, Review of

Economics and Statistics, vol. 94, no. 2, pp. 462-480.

Kondeas, A.G., Caudill, S.B., Gropper, D.M. and Raymond, J.E. (2008). Deregulation

366

and productivity changes in banking: evidence from European unification, Applied

Financial Economics Letters, vol. 4, no. 3, pp. 193-197.

Koop, G. (2001). Comparing the Performance of Baseball Players: A Multiple Output

Approach, Working paper in http://www.gla.ac.uk/Acad/PolEcon/Koop/.

Koop, G., Osiewalski, J. and Steel, M.F.J. (1994). Bayesian Efficiency Analysis with

a Flexible Form: The AIM Cost Function. Journal of Business and Economic

Statistics, vol. 12, pp. 339–346.

Kormendi, R.C. and Meguire, P.G. (1985). Macroeconomic determinants of growth:

cross-country evidence, Journal of Monetary economics, vol. 16, no. 2, pp. 141-163.

Koskela, E. and Stenbacka, R. (2000). Is there a tradeoff between bank competition

and financial fragility? Journal of Banking and Finance, vol. 24, no.12, pp. 1853–

1873.

Kosmidou, K. and Zopounidis, C. (2008). Measurement of bank performance in

Greece, South Eastern Europe Journal of Economics, vol. 1, pp. 79-95.

Koutsomanoli-Filippaki, A. and Mamatzakis, E.C. (2010). Estimating the speed of

adjustment of European banking efficiency under a quadratic loss function. Economic

Modeling, vol. 27, no. 1, pp. 1-11.

Koutsomanoli-Filippaki, A., Mamatzakis, E. and Staikouras, C. (2009). Structural

reforms and banking efficiency in the new EU States, Journal of Policy Modeling,

vol. 31, no. 1, pp. 17-21.

Koutsomanoli-Fillipaki, N. and Staikouras, C. (2004). Competition in the new

European banking landscape, European Financial Management, vol. 12, no. 3, pp.

443-482.

Kraft, E. and Tirtiroğlu, D. (1998). Bank efficiency in Croatia: A stochastic-frontier

analysis, Journal of Comparative Economics, vol. 26, no. 2, pp. 282-300.

Kraft, E., Hofler, R. and Payne, J. (2006). Privatization, foreign bank entry and bank

efficiency in Croatia: a Fourier-flexible function stochastic cost frontier analysis,

Applied Economics, vol. 38, no. 17, pp. 2075-2088.

367

Kumbhakar, S.C. (1993). Production risk, technical efficiency, and panel data.

Economics Letters, vol. 41, no. 1, pp. 11-16.

Kumbhakar, S.C. and Tsionas, E.G. (2005). The Joint Measurement of Technical and

Allocative Inefficiencies, Journal of the American Statistical Association, vol. 100,

no. 471, pp. 736-747.

Kumbhakar, S.C. and Tsionas, E.G. (2008). Scale and efficiency measurement using a

semiparametric stochastic frontier model: evidence from the US commercial banks.

Empirical Economics, vol. 34, no. 3, pp. 585-602.

Kumbhakar, S.C., Ghosh, S. and McGuckin, J.T. (1991). A generalized production

frontier approach for estimating determinants of inefficiency in US dairy farms,

Journal of Business and Economic Statistics, vol. 9, pp. 279–286.

Kumbhakar, S.C., Park, A., Simar, S. and Tsionas, E.G. (2007). Nonparametric

stochastic frontiers: A local maximum likelihood approach, Journal of Econometrics,

vol. 237, pp. 1–27.

Kwoka, J. (1985). The Herfindahl index in theory and practice, Anti-trust bulletin,

vol. 30, no. 4, pp. 915-947.

Laeven, L. and Levine, R. (2007). Is there a diversification discount in financial

conglomerates?, Journal of Financial Economics, vol. 85, pp. 331-367.

Laeven, L. and Levine, R. (2009). Bank governance, regulation and risk taking,

Journal of Financial Economics, vol. 93, pp. 259–275.

Lakonishok, J., Shleifer, A. and Vishny, R.W. (1992). The impact of institutional

trading on stock prices, Journal of financial economics, vol. 32, no. 1, pp. 23-43.

Land, K.C., Lovell, C.A.K. and Thore, S. (1992). Productive efficiency under

capitalism and state socialism: The chance constrained programming approach, Public

Finance in a World of Transition, vol. 47, pp. 109–121.

Land, K.C., Lovell, C.A.K. and Thore, S. (1994). Production efficiency under

capitalism and state socialism: An empirical inquiry using chance constrained data

envelopment analysis, Technological Forecasting and Social Change, vol. 46, pp.

368

139– 152.

Lang, G. and Welzel, P. (1996). Efficiency and technical progress in banking

Empirical results for a panel of German cooperative banks, Journal of Banking and

Finance, vol. 20, no. 6, pp. 1003-1023.

Lau, L. (1982). On identifying the degree of competitiveness from industry price and

output data, Economics Letters, vol. 10, pp. 93-99.

Leamer, H. (1983). Let's Take the Con Out of Econometrics, American Economic

Review, vol. 73, pp. 31-43.

Leamer, H. (1985). Sensitivity Analyses Would Help, American Economic Review,

vol. 75, pp. 308-13.

Leamer, H. and Leonard, H. (1983). Reporting the Fragility of Regression Estimates,

Review of Economics and Statistics, vol. 65, pp. 306-17.

Lensink, R. and Maslennikova, I. (2008). Value performance of European bank

acquisitions, Applied financial economics, vol. 18, no. 3, pp. 185-198.

Lensink, R., Meesters, A. and Naaborg, I. (2008). Bank efficiency and foreign

ownership: Do good institutions matter?, Journal of Banking and Finance, vol. 32,

no. 5, pp. 834-844.

Lerner, A. (1934). The Concept of Monopoly and the Measurement of Monopoly

Power, Review of Economic Studies, vol. 1, pp. 157-175.

Levine, R. and Renelt, D. (1992). A sensitivity analysis of cross-country growth

regressions, The American Economic Review, vol. 82, no. 4, pp. 942-963.

Levy-Yeyati, E. and Micco, A. (2007). Concentration and foreign penetration in Latin

American banking sectors: impact on competition and risk, Journal of Banking and

Finance, vol. 31, no. 6, pp. 1633–1647.

Liang, L.F., Wu, J., Cook, W.D. and Zhu, J. (2008). The DEA cross efficiency model

and its Nash equilibrium, Operations Research, forthcoming.

Lima, F. (2008). Product differentiation and the measurement of cost efficiency in

369

banking: the case of Portuguese banks, New Developments in Financial Modeling,

vol. 1, no. 1, pp. 34-76.

Lima, F. and Soares de Pinho, P. (2008). Financial disintermediation and the

measurement of efficiency in banking: the case of Portuguese banks, International

Journal of Banking, Accounting and Finance, vol. 1, no. 2, pp. 133-148.

Liu, H., Molyneux, P. and Wilson, J.O. (2010). Competition and stability in European

banking: A regional analysis, The Manchester School, vol. 81, no. 2, pp. 176-201.

Lloeyd-Williams D.M., Molyneux P. and Thornton J. (1994). Market structure and

performance in Spanish banking, Journal of Banking and Finance, vol. 18, pp. 433-

443.

Lovell, C.A.K., Rouse, A.P.B. (2003). Equivalent standard DEA models to provide

super-efficiency scores, Journal of the Operational Research Society, vol. 54, no.1,

pp. 101–108.

Lozano-Vivas, A. and Pasiouras, F. (2010). The impact of non-traditional activities on

the estimation of bank efficiency: international evidence, Journal of Banking and

Finance, vol. 34, no. 7, pp. 1436-1449.

Lozano-Vivas, A., Pastor, J.T. and Hasan, I. (2001). European Bank Performance

Beyond Country Borders: What Really Matters?, European Finance Review, vol. 5,

no. 1-2, pp. 141-165.

Lozano-Vivas, A., Pastor, J.T. and Pastor, J.M. (2002). An efficiency comparison of

European banking systems operating under different environmental conditions,

Journal of Productivity Analysis, vol. 18, no. 1, pp. 59-77.

Lucinda, C.R. (2010). Competition in the Brazilian loan market: an empirical

analysis. Estudos Economicos (Sao Paulo), vol. 40, no. 4, pp. 831-858.

Luckett, G.D. (1980). Money and Banking, 2nd

ed. New York: McGraw-Hill.

Macit, F. (2012). Recent evidence on concentration and competition in Turkish

banking sector, International Research Journal of Finance and Economics, (96).

370

Malmquist, S. (1953). Index numbers and indifference surfaces, Trabajos de

Estadistica y de Investigacion Operativa, vol. 4, no. 2, pp. 209-242.

Mamatzakis, E., Staikouras, C. and Koutsomanoli-Fillipaki, N. (2005). Competition

and concentration in the banking sector of the South Eastern European region,

Emerging Markets Review, vol. 6, no. 2, pp. 192-209.

Mamatzakis, E., Staikouras, C. and Koutsomanoli-Filippaki, A. (2008). Bank

efficiency in the new European Union member states: Is there convergence?,

International Review of Financial Analysis, vol. 17, no. 5, pp. 1156-1172.

Marcus, A.J. (1984). Deregulation and bank financial policy, Journal of Banking and

Finance, vol. 8, pp. 557-565.

Marfels, C. (1971). Absolute and relative measures of concentration reconsidered,

Kyklos, vol. 24, pp. 753–66.

Marius, A.A. and Bogdan, C. (2011). Competition in Romanian Banking Sector.

Annals of Faculty of Economics, vol. 1, no. 1, pp. 455-460.

Martin S. (1993). Advanced industrial Economics, Blackwell, New York and Oxford.

Martinez-Miera, D. and Repullo, R. (2010). Does competition reduce the risk of bank

failure?, Review of Financial Studies, vol. 23, no. 10, pp. 3638-3664.

Masood, O. and Aktan, B. (2010). The State of Competition of the Turkish Banking

Industry: An Application of the Panzar-Rosse Model, Journal of Business Economics

and Management, no. 1/2010, pp. 131-145.

Masood, O. and Sergi, B.S. (2011). China‟s banking system, market structure, and

competitive conditions, Frontiers of Economics in China, vol. 6, no. 1, pp. 22-35.

Matthews, K., Murinde, V. and Zhao, T. (2007). Competitive conditions among the

major British banks, Journal of Banking and Finance, vol. 31, no. 7, pp. 2025-2042.

Matutes C. and Vives X. (1996). Competition for deposits, fragility and insurance,

Journal of Financial Intermediation, vol. 5, no.2, pp. 184-216.

Matutes, C. and Vives, X. (2000). Imperfect competition, risk taking, and regulation

371

in banking, European Economic Review, vol. 44, pp. 1-34.

Maudos, J. and De Guevara, J.F. (2007). The cost of market power in banking: Social

welfare loss vs. cost inefficiency, Journal of Banking and Finance, vol. 31, no. 7, pp.

2103-2125.

Maudos, J. and Pastor, J.M. (2001). Cost and profit efficiency in banking: an

international comparison of Europe, Japan and the USA, Applied Economics Letters,

vol. 8, no. 6, pp. 383-387.

Maudos, J. and Pastor, J.M. (2003). Cost and profit efficiency in the Spanish banking

sector (1985–1996): A non-parametric approach, Applied Financial Economics, vol.

13, no. 1, pp. 1-12.

Maudos, J. and Perez , F. ( 2003). Competencia vs . poder de mercado en la banca

espanola, Moneda y Credito, vol. 217, pp. 139-166.

Maudos, J., Pastor, J. M., Perez, F. and Quesada, J. (2002). Cost and profit efficiency

in European banks, Journal of International Financial Markets, Institutions and

Money, vol. 12, no. 1, pp. 33-58.

McAllister, P.H. and McManus, D. (1993). Resolving the scale efficiency puzzle in

banking, Journal of Banking and Finance, vol. 17, no. 2, pp. 389-405.

Meeusen, W. and van den Broeck, J. (1977). Efficiency Estimation from Cobb-

Douglas production functions with composed error. International Economic Review,

vol. 18, pp. 435–444.

Mendes, V. and Rebelo, J. (1999). Productive efficiency, technological change and

productivity in Portuguese banking, Applied Financial Economics, vol. 9, no. 5, pp.

513-521.

Mensi, S. (2010). Measurement of competitiveness degree in Tunisian deposit banks:

An application of the Panzar and Rosse model, Panoeconomicus, vol. 57, no. 2, pp.

189-207.

Méon, P.G. and Weill, L. (2005). Does better governance foster efficiency? An

aggregate frontier analysis, Economics of Governance, vol. 6, no. 1, pp. 75-90.

372

Mercan, M. (2012). Competitive Conditions in the Banking Industry of Georgian

Economy: PR H Model, Procedia-Social and Behavioral Sciences, vol. 62, pp. 1260-

1264.

Mertens, A. and Urga, G. (2001). Efficiency, scale and scope economies in the

Ukrainian banking sector in 1998, Emerging Markets Review, vol. 2, no. 3, pp. 292-

308.

Mester, L.J. (2003). Applying Efficiency Measurement Techniques to Central Banks,

FRB of Philadelphia Working Paper No. 03-13. Available at SSRN: http://ssrn.com/

abstract=572442 or doi:10.2139/ssrn.572442

Mester, L.J. (2012). A multiproduct cost study of savings and loans, The Journal of

Finance, vol. 42, no. 2, pp. 423-445.

Mishkin F. S. (1999). Financial consolidation: Dangers and opportunities, Journal of

Banking and Finance, vol. 23, pp. 675-691.

Mitchell, K. and Onvural, N. M. (1996). Economies of scale and scope at large

commercial banks: Evidence from the Fourier flexible functional form, Journal of

Money, Credit and Banking, vol. 28, no. 2, pp. 178-199.

Mlambo, K. and Ncube, M. (2011). Competition and efficiency in the banking sector

in South Africa. African Development Review, vol. 23, no.1, pp. 4-15.

Molyneux P. and Forbes W. (1995). Market structure and performance in European

banking, Applied Economics, vol. 27, pp. 155-159.

Molyneux P., Altunbas Y. and Gardener E. (1996). Efficiency in European banking,

John Wiley and Sons, Ltd.

Molyneux, P., Lloyd-Williams, D. and Thornton, J. (1994). Competitive conditions in

European banking, Journal of Banking and Finance, vol. 18, pp. 445-459.

Moussawi, C.E. and Saad, W. (2012). Measuring the Degree of Competition in the

Lebanese Banking System, International Research Journal of Finance and

Economics, no. 94.

373

Muniz, M., Paradi, J., Ruggiero, J. and Yang, Z.J. (2006). Evaluating alternative DEA

models used to control for non-discretionary inputs, Computers and Operations

Research, vol. 33, pp. 1173–1183.

Murphy, K., Shleifer, A. and Vishny, R. (1992). The transition to a market economy:

Pitfalls of partial reform, Quarterly Journal of Economics, vol. 107, pp. 889-906.

Mwega, F. (2011). The competitiveness and efficiency of the financial services sector

in Africa: A case study of Kenya, African Development Review, vol. 23, no. 1, pp. 44-

59.

Nagarajan, S. and Sealey, C.W. (1995). Forbearance, deposit insurance pricing, and

incentive compatible bank regulation, Journal of Banking and Finance, vol. 19, no. 6,

pp. 1109-1130.

Nathan, A. and Neave, H.E. (1989). Competition and Contestability in Canada‟s

Financial System: Empirical Results, Canadian Journal of Economics, vol. 22, 576–

594.

Nelder, M. (2003). Competition necessarily tends to produce excess: The experience

of free banking in Switzerland, German Economic review, vol. 4, no. 3, pp. 389-408.

Neralic, L. (1997). Sensitivity in data envelopment analysis for arbitrary perturbations

of data, Glasnik Matematicki, vol. 32, pp. 315–335.

Neralic, L. (2004). Preservation of efficiency and inefficiency classification in data

envelopment analysis, Mathematical Communications, vol. 9, pp. 51– 62.

Neven, D. and Roller , L.H. (1999). An aggregate structural model of competition in

the European banking industry, International Journal of Industrial Organization, vol.

17, no. 7, pp. 1059-1074.

Nguyen, M., Skully, M. and Perera, S. (2012). Bank market power and revenue

diversification: Evidence from selected ASEAN countries, Journal of Asian

Economics. vol. 3, no. 6, pp. 688-700.

Nguyen, M., Skully, T.M. and Perera, S. (2012). Bank Market Power and Liquidity:

Evidence from 113 Developed and Developing Countries, 25th Australasian Finance

374

and Banking Conference 2012.

Noulas, A.G. (1997). Productivity growth in the Hellenic banking industry: state

versus private banks, Applied Financial Economics, vol. 7, no. 3, pp. 223-228.

Noulas, A.G., Ray, S.C. and Miller, S.M. (1990). Returns to scale and input

substitution for large US banks, Journal of Money, Credit and Banking, vol. 22, no. 1,

pp. 94-108.

Olesen, O.B. and Petersen, N.C. (1995). Chance Constrained Efficiency Evaluation,

Management Science, vol. 41, pp. 442–457.

Olivero, M.P., Li, Y. and Jeon, B. N. (2011). Competition in banking and the lending

channel: Evidence from bank-level data in Asia and Latin America. Journal of


Orea, L. and Kumbhakar, S.C. (2004). Efficiency measurement using a latent class

stochastic frontier model, Empirical Economics, vol. 29, 1, pp. 169-183.

Osiewalski, J. and Steel, M.F.J. (1998). Numerical Tools for the Bayesian Analysis of

Stochastic Frontier Models, Journal of Productivity Analysis, vol. 10, pp. 103–117.

Ou, Chao-min and Tan, Yue-Jin. (2011). Multi-perspective Studies on Competition

Degree of Chinese Banking Industry: Empirical Analysis Based on Panzar-Rosse

Method, Systems engineering, 02/2011.

Oxenstierna, G. (1998). An Asymmetric Oligopoly Model and a Method for its

Empirical Application, Journal of Economics, vol.67, no.1, pp 39-61.

Oxenstierna, G. (2000). Testing for Market Power in the Swedish Banking Oligopoly,

working paper 2/2000, Sodertorns hogskola.

Panzar, J.C., Rosse, J.N. (1982). Structure, conduct and comparative statistics, Bell

Laboratories Economics discussion paper, Bell Laboratories.

Panzar, J.C., Rosse, J.N. (1987). Testing for „monopoly‟ equilibrium, The Journal of

Industrial Economics, vol. 35, pp. 443-456.

Park, B.U. and Simar, L. (1994). Efficient semiparametric estimation in a stochastic

375

frontier model, Journal of the American Statistical Association, vol. 89, pp. 92–936.

Park, B.U., Sickles, R.C. and Simar, L. (2007) Semiparametric efficient estimation of

dynamic panel data models, Journal of Econometrics, vol. 136, pp. 281-301.

Park, B.U., Simar, L. and Sickles, R.C. (1998). Stochastic panel frontiers: a

semiparametric approach, Journal of Econometrics, vol. 84, pp. 121–152.

Park, B.U., Simar, L. and Weiner, Ch. (2000). The FDH Estimator for Productivity

Efficiency Scores: Asymptotic Properties, Econometric Theory, vol. 16, pp. 855–877.

Park, K.H. (2009). Has bank consolidation in Korea lessened competition?, The

Quarterly Review of Economics and Finance, vol. 49, no. 2, Pages 651-667.

Pasiouras, F. (2008a). Estimating the technical and scale efficiency of Greek

commercial banks: the impact of credit risk, off-balance sheet activities, and

international operations, Research in International Business and Finance, vol. 22, no.

3, pp. 301-318.

Pasiouras, F. (2008b). International evidence on the impact of regulations and

supervision on banks‟ technical efficiency: an application of two-stage data

envelopment analysis, Review of Quantitative Finance and Accounting, vol. 30, no. 2,

pp. 187-223.

Pasiouras, F., Liadaki, A. and Zopounidis, C. (2008). Bank efficiency and share

performance: evidence from Greece, Applied Financial Economics, vol. 18, no. 14,

pp. 1121-1130.

Pasiouras, F., Tanna, S. and Zopounidis, C. (2009). The impact of banking regulations

on banks' cost and profit efficiency: Cross-country evidence, International Review of

Financial Analysis, vol. 18, no. 5, pp. 294-302.

Pastor, J., Perez, F. and Quesada, J. (1997). Efficiency analysis in banking firms: An

international comparison, European Journal of Operational Research, vol. 98, no.2,

pp. 395-407.

Pastor, J.M. (2002). Credit risk and efficiency in the European banking system: A

three-stage analysis, Applied Financial Economics, vol. 12, no. 12, pp. 895-911.

376

Pastor, J.M. and Serrano, L. (2005). Efficiency, endogenous and exogenous credit risk

in the banking systems of the Euro area, Applied Financial Economics, vol. 15, no. 9,

pp. 631-649.

Pastor, J.M. and Serrano, L. (2006). The Effect of Specialisation on Banks'

Efficiency: An International Comparison, International Review of Applied Economics,

vol. 20, no. 1, pp. 125-149.

Pastor, J.T., Ruiz, J.L. and Sirvent, I. (1999). An enhanced DEA Russell graph

efficiency measure, European Journal of Operational Research, vol. 115, pp. 596–

607.

Pawlowska, M. (2003). The Impact of M&A on Technical Efficiency, Scale

Efficiency and Productivity Change in the Polish Banking Sector: a Non-Parametric

Approach (No. 29), National Bank of Poland, Economic Institute.

Pawlowska, M. (2012). Competition, concentration and foreign capital in the Polish

banking sector (prior and during the financial crisis). National Bank of Poland

Working Paper, (130).

Peltzman, S. (1977). The gains and losses from industrial concentration, Journal of

Law and Economics, vol. 20, pp. 229-263.

Perotti, C.E. and Suarez, J. (2002). Last bank standing: What do I gain if you fail?,

European Economic Review, vol. 46, pp. 1599-1622.

Petersen, M.A. and Rajan, R.G. (1995). The effect of credit market competition on

lending relationships, Quarterly Journal of Economics, vol. 110, no. 2, pp. 407-443.

Podpiera, A. and Podpiera, J. (2005). Deteriorating Cost Efficiency in Commercial

Banks Signals an Increasing Risk of Failure, Czech National Bank.

Poghosyan, T. and Kumbhakar, S.C. (2010). Heterogeneity of technological regimes

and banking efficiency in former socialist economies, Journal of Productivity

Analysis, vol. 33, no. 1, pp. 19-31.

Portela, M. and Thanassoulis, E. (2007). Comparative efficiency analysis of

Portuguese bank branches, European Journal of Operational Research, vol. 177, no.

377

2, pp. 1275-1288.

Portela, M., Castro, P. and Thanassoulis, E. (2003). Finding closest targets in non-

oriented DEA models: The case of convex and non-convex technologies, Journal of

Productivity Analysis, vol. 19, pp. 251–269.

Poshakwale, S.S. and Qian, B. (2011). Competitiveness and Efficiency of the Banking

Sector and Economic Growth in Egypt, African Development Review, vol. 23, no. 1,

pp. 99-120.

Prasad, A. and Ghosh, S. (2007). Competition in Indian Banking An Empirical

Evaluation, South Asia Economic Journal, vol. 8, no. 2, pp. 265-284.

Primorac, M. and Troskot, Z. (2005). Measuring the Efficacy and Productiveness of

Croatian Banks with the Malmquist Index of Change in Total Factor Productivity,

Financial Theory and Practice, vol. 29, no. 4, pp. 309-325.

Pruteanu-Podpiera, A., Weill, L. and Schobert, F. (2007). Market Power and

Efficiency in the Czech Banking Sector. Working Papers, 6.

Pulley, L.B. and Braunstein, Y.M. (1992). A composite cost function for multiproduct

firms with an application to economies of scope in banking, The Review of Economics

and Statistics, vol. 74, no. 2, pp. 221-230.

Punt, L.W. and Van Rooij, M.C.J. (1999). The profit-structure relationship, efficiency

and mergers in the European banking industry: an empirical assessment, Research

Memorandum WO&E, 604.

Quah, D.T. (1996). Empirics for economic growth and convergence, European

economic review, vol. 40, no. 6, pp. 1353-1375.

Rajan, R.G. (1992). Insiders and outsiders: the choice between informed and arm‟ s-

length debt, Journal of Finance, vol. 47, pp. 1367–1400.

Ray, S.C. and Das, A. (2010). Distribution of cost and profit efficiency: Evidence

from Indian banking, European Journal of Operational Research, vol. 20, no. 1, pp.

297-307.

378

Rebelo, J. and Mendes, V. (2000). Malmquist indices of productivity change in

Portuguese banking: the deregulation period, International Advances in Economic

Research, vol. 6, no. 3, pp. 531-543.

Reid, G.C. (1987). Theories of industrial organisation, Blackwell, New York and

Oxford.

Resti, A. (1997). Evaluating the cost-efficiency of the Italian banking system: what

can be learned from the joint application of parametric and non-parametric

techniques, Journal of Banking and Finance, vol. 21, no. 2, pp. 221-250.

Rezitis, A.N. (2010). Evaluating the state of competition of the Greek banking

industry, Journal of international financial markets, institutions and money, vol. 20,

no. 1, pp. 68-90.

Rhoades, A.S. (1994). A summary of merger performance studies in banking, 1980-

93, and an assessment of the operating performance and event study methodologies,

Staff Studies 167, Board of Governors of the Federal Reserve System (U.S.).

Rhoades, A.S. (1995). Market share inequality, the HHI, and other measures of the

firm-composition of the market, Review of Industrial Organisation, vol. 10, no. 6, pp.

657-674.

Rime, B. (1999). Mesure de degre de concurrence dans le systeme bancaire Suisse a

l‟aide du modele de Panzar et Rosse , Revue Suisse d’Economie Politique et de

Statistique, vol. 135, no. 1, pp. 21-40.

Rime, B. and Stiroh, K.J. (2003). The performance of universal banks: Evidence from

Switzerland, Journal of banking and finance, vol. 27, no. 11, pp. 2121-2150.

Rockoff, H. (1975). The Free Banking Era: A Reconsideration, New York: Arno.

Roll, Y., Cook, W.D. and Golany, B. (1991). Controlling factor weights in data

envelopment analysis, IIE Transactions, vol. 23, pp. 2–9.

Rolnick, A.J. and Weber, W.E. (1983). New evidence on the free banking era, The

American Economic Review, vol. 73, no. 5, pp. 1080-1091.

379

Rolnick, A.J. and Weber, W.E. (1984). The causes of free bank failures: A detailed

examination, Journal of Monetary Economics, vol. 14, no. 3, pp. 267-291.

Rosenbluth, G. (1955). Measures of concentration, in business concentration and

Price policy, National Bureau Committee for Economic Research, Princeton, pp. 57-

99.

Rossi, S.P., Schwaiger, M. and Winkler, G. (2005). Managerial behavior and

cost/profit efficiency in the banking sectors of Central and Eastern European

countries, Oesterr. Nationalbank.

Ruggiero, J. (1996). On the measurement of technical efficiency in the public sector,

European Journal of Operational Research, vol. 90, pp. 553– 565.

Saez, L. and Shi, X. (2004). Liquidity pools, risk sharing and financial contagion,

Journal of Financial Services Research, vol. 25, pp. 5-23.

Sala-i-Martin, X.X. (1996). The classical approach to convergence analysis. The

Economic Journal, ωολ. 106, νο. 437, pp. 1019-1036.

Salamanca, D. (2005). Competencia en los mercados de credito y depositos en

Colombia: aplicacion de un modelo de oligopolio fijador de precios. Universidad de

los Andes.

Salop, S. and Stiglitz J. (1977). Bargains and rip-offs: A model of monopolistically

competitive price dispersion, Review of economic studies, vol. 4, pp. 493-510.

Schaeck, K. (2006). Bank Competition and Bank Soundness: New evidence.

Schaeck, K. (2009). Bank market structure, competition and stability: Issues and

concepts. In: M. Fratianni, A. Z., Alessandrini, P. (Eds.), The Changing Geography of

Banking. Springer, pp. 133–153.

Schaeck, K. and Cihak, M. (2008). How does competition affect efficiency and

soundness in banking? New empirical evidence. European Central Bank Working

Paper 932.

Schaeck, K., Cihak, M. and Wolfe, S. (2009). Are competitive banking systems more

380

stable?, Journal of Money, Credit and Banking, vol. 41, no. 4, pp. 711-734.

Scherer, F.M. and Ross D. (1990). Industrial Market structure and Economic

performance, Houghton Mifflin, Boston.

Schmidt, P. and Sickles, R.C. (1984). Production Frontiers and Panel Data, Journal of

Business and Economic Statistics, vol. 2, pp. 299–326.

Sealey, C.W. and Lindley, J.T. (1977). Inputs, outputs, and a theory of production and

cost at depository financial institutions, The Journal of Finance, vol. 32, no. 4, pp.

1251-1266.

Seiford, L.M. and Zhu, J. (1997). An investigation of returns to scale in data

envelopment analysis, OMEGA, vol. 27, pp. 1–11.

Seiford, L.M. and Zhu, J. (1999). Profitability and marketability of the top 55 US

commercial banks, Management Science, vol. 45, no. 9, pp. 1270–1288.

Sengupta, J.T. (1990). Transformation in Stochastic DEA Models. Journal of

Econometrics, vol. 46, no. 1–2, pp. 109–124.

Sepulveda, J.P. (2012). On the relationship between concentration and competition:

evidence from the Chilean private pension system, Applied Economics Letters, vol.

19, no. 14, pp. 1385-1389.

Sexton, T.R., Silkman, R.H., Hogan, A.J. (1986). Data envelopment analysis: Critique

and extensions. In: Silkman, R.H. (Ed.), Measuring Efficiency: An Assessment of

Data Envelopment Analysis, vol. 32. Jossey-Bass, San Francisco, pp. 73–105.

Shaffer, S. (1989) Competition in the U.S. Banking Industry, Economics Letters, vol.

29, no. 4, pp. 321–23.

Shaffer, S. (1993). A Test of Competition in Canadian Banking, Journal of Money,


Shaffer, S. (1994). A revenue-restricted cost study of 100 large banks, Applied


Shaffer, S. (2001). Banking conduct before the European single banking license: a

381

cross-country comparison, The North American Journal of Economics and Finance,

vol. 12, no. 1, pp. 79-104.

Shaffer, S. and DiSalvo, J. (1994). Conduct in a banking duopoly, Journal of Banking

and Finance, vol. 18, pp. 1063-1082.

Shin, D.J. and Kim, B.H. (2012). Bank consolidation and competitiveness: empirical

evidence from the Korean banking industry, Journal of Asian Economics, vol. 24, pp.

41-50.

Shin, H. (2009). Securitization and Financial Stability, Economic Journal, vol. 119,

pp. 309-332.

Sickles, R.C. (2005). Panel estimators and the identification of firm-specific

efficiency levels in parametric, semi-parametric and non-parametric settings, Journal

of Econometrics, vol. 126, pp. 305-334.

Sickles, R.C., Good, D.H. and Getachew, L. (2002). Specification of distance

functions using semi- and nonparametric methods with an application to the dynamic

performance of Eastern and Western European air carriers, Journal of Productivity

Analysis, vol. 17, pp. 133–155.

Simar, L. (1992). Estimating efficiencies from frontier models with panel data: a

comparison of parametric, non-parametric and semi-parametric methods with

bootstrapping, Journal of productivity analysis, vol. 3, no. 1, pp. 171-203.

Simar, L. and Wilson, P.W. (1998). Sensitivity Analysis of Efficiency Scores: How to

Bootstrap in Nonparametric Frontier Models, Management Science, vol. 44, no. 11,

pp. 49–61.

Simar, L. and Wilson, P.W. (1999). Estimating and Bootstrapping Malmquist Indeces,

European Journal of Operations Research, vol. 115, pp. 459–471.

Simar, L. and Wilson, P.W. (2000a). Statistical Inference in Nonparametric Frontier

Models: The State of the Art, Journal of Productivity Analysis, vol. 13, pp. 49–78.

Simar, L. and Wilson, P.W. (2000b). A General Methodology for Bootstrapping in

Nonparametric Frontier Models, Journal of Applied Statistics, vol. 27, no. 6, pp. 779-

382

802.

Simbanegavi, W., Greenberg, J. and Gwatidzo, T. (2012). Testing for competition in

the South African banking sector.

Simpasa, A.M. (2011). Competitive conditions in the Tanzanian commercial banking

industry, African Development Review, vol. 23, no.1, pp. 88-98.

Simper, R. and Hall, M. (2012). Efficiency and Competition in Korean Banking,

Nottingham University Business School Research Paper, (2012-07).

Sjoberg, P. (2004), Market Power and Performance in Swedish Banking, Department

of Economics (Goteborg University).

Skully, M. and Perera, S. (2012). Bank Market Power and Liquidity: Evidence from

113 Developed and Developing Countries. Available at SSRN 2136743.

Smirlock, M. (1985). Evidence on the (non) relationship between concentration and

profitability in banking, Journal of Money, Credit, and Banking, vol. 17, no. 1, pp.

69-83.

Smith, B.D. (1984). Private information, deposit interest rates, and the „stability‟ of

the banking system, Journal of Monetary Economics, vol. 14, pp. 293-317.

Soedarmono, W. (2010). Bank competition, institution and economic development:

Evidence from Asia during 1999-2007, Economics Bulletin, vol. 30, no. 3, pp. 2119-

2133.

Soedarmono, W., Machrouh, F. and Tarazi, A. (2011). Bank market power, economic

growth and financial stability: Evidence from Asian banks, Journal of Asian


Spiller, P.T. and Favaro, E. (1984). The effects of entry regulation on oligopolistic

interaction: The Uruguayan banking sector, The Rand Journal of Economics, vol. 15,

no. 2, pp. 244-254.

Spong, K., Sullivan, R.J. and DeYoung, R. (1995). What makes a bank efficient? A

look at financial characteristics and bank management and ownership structure,

383

Federal Reserve Bank of Kansas City, Financial Industry Perspectives, December, 1-

19.

Srairi, S.A. (2010). Cost and profit efficiency of conventional and Islamic banks in

GCC countries, Journal of Productivity Analysis, vol. 34, 1, pp. 45-62.

Staikouras, C.K. and Koutsomanoli‐Fillipaki, A. (2006). Competition and

concentration in the new European banking landscape, European Financial

Management, vol. 12, no. 3, pp. 443-482.

Staikouras, C., Mamatzakis, E. and Koutsomanoli-Filippaki, A. (2008a). An empirical

investigation of operating performance in the new European banking landscape,

Global Finance Journal, vol. 19, no.1, pp. 32-45.

Staikouras, C., Mamatzakis, E. and Koutsomanoli-Filippaki, A. (2008b). Cost

efficiency of the banking industry in the South Eastern European region, Journal of

International Financial Markets, Institutions and Money, vol. 18, no. 5, pp. 483-497.

Stavarek, D. (2003). Banking efficiency in Visegrad countries before joining the

European Union. Available at SSRN 671664.

Stavarek, D. (2005). Efficiency of banks in regions at different stage of European

integration process. Available at SSRN 672184.

Stavarek, D. and Repkova, I. (2011). Estimation of the competitive conditions in the

Czech banking sector, pp. 299-305.

Steen, F. and Salvanes K. (1999). Testing for Market Power using a Dynamic

Oligopoly Model, International Journal of Industrial Organization, vol. 17, pp. 147–

77.

Stiglitz, E.J. and Weiss, A. (1981). Credit rationing in markets with imperfect

information, American Economic Review, vol. 71, no. 3, pp. 393-410.

Stiroh, K.J., (2004). Do community banks benefit from diversification? Journal of

Financial Services Research, vol. 25, pp. 135-160.

Sueyoshi, T. (1990). A special algorithm for the additive model in DEA, Journal of

384

the Operational Research Society, vol. 41, no. 3, pp. 249–257.

Sufian, F. (2009). Determinants of bank efficiency during unstable macroeconomic

environment: Empirical evidence from Malaysia, Research in International Business

and Finance, vol. 23, no. 1, pp. 54-77.

Sufian, F. (2011). The nexus between financial sector consolidation, competition and

efficiency: empirical evidence from the Malaysian banking sector, IMA Journal of

Management Mathematics, vol. 22, no. 4, pp. 419-444.

Sufian, F. and Habibullah, M.S. (2012). Financial Sector Consolidation and

Competition in Malaysia: An Application of the Panzar-Rosse Method, Journal of

Economic Studies, vol. 40, no. 3.

Summer, M. (2003). Banking regulation and systemic risk, Open Economies review,

vol. 14, pp. 43-70.

Sun, Yu. (2011). Recent Developments in European Bank Competition, IMF Working

Papers, vol. 11, no. 146, pp. 1-25, 2011.

Suominen, M. (1994). Measuring Competition in Banking: A Two-Product Model,

Scandinavian Journal of Economics, vol. 96, no. 1, pp. 95–110.

Sutton, J. (1991). Sunk Costs and Market Structure. MIT Press, London.

Swank, J. (1995). Oligopoly in loan and deposit markets: an econometric application

to the Netherlands, De Economist, vol. 143, no. 3, pp. 353-366.

Syrjanen, M.J. (2004). Non-discretionary and discretionary factors and scale in data

envelopment analysis, European Journal of Operational Research, vol. 158, pp. 20–

33.

Tabak, B.M., Gomes, G. and Junior, M. (2012). The Impact of Market Power at Bank

Level in Risk-Taking: The Brazilian Case (No. 283).

Thanassoulis, E. and Allen, R. (1998). Simulating weights restrictions in data

envelopment analysis by means of unobserved DMUs, Management Science, vol. 44,

no. 4, pp. 586–594.

385

Thompson, R.G., Dharmapala, S. and Thrall, R.M. (1995). Linked-cone DEA profit

ratios and technical inefficiencies with applications to Illinois coal mines,

International Journal of Production Economics, vol. 39, pp. 99– 115.

Thompson, R.G., Langemeir, L.N., Lee, C., Lee, E. and Thrall, R.M. (1990). The role

of multiplier bounds in efficiency analysis with application to Kansas farming,

Journal of Econometrics, vol. 46, pp. 93–108.

Thompson, R.G., Singleton, F.D., Thrall, R.M. and Smith, B.A. (1986). Comparative

site evaluations for locating a high-energy physics lab in Texas, Interfaces, vol. 16,

no. 6, pp. 35-49.

Thore, S. (1987). Chance-constrained activity analysis, European Journal of

Operational Research, vol. 30, pp. 267–269.

Tomova, M. (2005). X-efficiency of European banking-inequality and convergence,

Free University of Brussels.

Tone, K. (2001). A slacks-based measure of efficiency in data envelopment analysis,

European Journal of Operational Research, vol. 130, pp. 498–509.

Toolsema, L. (2002). Competition in the Dutch consumer credit market, Journal of

Banking and Finance, vol. 26, pp. 2215-2229.

Tortosa-Ausina, E. (2002). Bank cost efficiency and output specification, Journal of

Productivity Analysis, vol. 18, no. 3, pp. 199-222.

Tran, K.C. and Tsionas, E.G. (2009). Estimation of nonparametric inefficiency effects

stochastic frontier models with an application to British manufacturing, Econ Model,

vol. 26, pp. 904–909.

Trivieri, F. (2007). Does cross-ownership affect competition?: Evidence from the

Italian banking industry, Journal of International Financial Markets, Institutions and

Money, vol. 17, no. 1, pp. 79-101.

Tsionas, E.G. (2002). Stochastic Frontier Models with Random Coefficients, Journal

of Applied Econometrics, vol. 17, pp. 127–147.

386

Tulkens, H. (1993). On FDH efficiency analysis: Some methodological issues and

applications to retail banking, courts and urban transit, Journal of Productivity

Analysis, vol. 4, pp. 183–210.

Turati, G. (2003). Cost Efficiency and Profitability in European Commercial Banking:

Implications for Antitrust Analysis. Universitá di Torino, downloaded on Sept, 29.

Uchida, H. and Tsutsui, Y. (2005). Has competition in the Japanese banking sector

improved?, Journal of banking and Finance, vol. 29, no. 2, pp. 419-439.

Uhde, A. and Heimeshoff, U. (2009). Consolidation in banking and financial stability

in Europe: empirical evidence, Journal of Banking and Finance, vol. 33 pp. 1299–

1311.

Van den Broeck, J., Koop, G., Osiewalski, J. and Steel, M. (1994). Stochastic Frontier

Models: a Bayesian Perspective, Journal of Econometrics, vol. 61, pp. 273–303.

Van Leuvensteijn, M., Bikker, J.A., van Rixtel, A.A. and Sørensen, C.K. (2011). A

new approach to measuring competition in the loan markets of the euro area, Applied


Vaškelaitis, V. and Deltuvaitė, V. (2007). Competition, concentration, efficiency, and

their relationship in the Lithuanian banking sector, Taikomoji ekonomika: sisteminiai

tyrimai, (1/1), pp. 11-30.

Vennet, R.V. (1996). The effects of mergers and acquisitions on the efficiency and

profitability of EC credit institutions, Journal of Banking and Finance, vol. 20, pp.

1158–1531.

Vennet, R.V. (2002). Cost and profit efficiency of financial conglomerates and

universal banks in Europe, Journal of Money, Credit, and Banking, vol. 34, no. 1, pp.

254-282.

Vesala, J. (1995). Testing for competition in banking: Behavioral evidence from

Finland, Bank of Finland Studies, E1.

Vives, X. (2001). Competition in the changing world of banking, Oxford Review of

Economic Policy, vol. 17, pp. 535–545.

387

Vo Thi, N. (2009). Banking Market Liberalization and Bank Performance: the Role of

Entry Modes.

Vo Thi, N. and Vencappa, D. (2008). Does the Entry Mode of Foreign Banks Matter

for Bank Efficiency? Evidence from Czech Republic, Hungary, and Poland.

Von Thadden, E.-L. (1995). Long-term contracts, short-term investment and

monitoring, Review of Economic Studies, vol. 62, no. 4, pp. 557-575.

Wagenvoort, R.J. and Schure, P.H. (2006). A Recursive Thick Frontier Approach to

Estimating Production Efficiency, Oxford Bulletin of Economics and Statistics, vol.

68, no. 2, pp. 183-201.

Walkner, C. and Raes, J.P. (2005). Integration and Consolidation in EU Banking - An

Unfinished Business, European Commission – DG Economics and Financial Affairs,

Economic Papers 226, April.

Webb, R. (2003). Levels of efficiency in UK retail banks: a DEA window analysis,

International Journal of the Economics of Business, vol. 10, no. 3, pp. 305-322.

Weill, L. (2002). Does restructuring improve banking efficiency in a transition

economy?, Applied Economics Letters, vol. 9, no. 5, pp. 279-281.

Weill, L. (2004a). Measuring cost efficiency in European banking: A comparison of

frontier techniques. Journal of Productivity Analysis, vol. 21, no. 2, pp. 133-152.

Weill, L. (2004b). On the relationship between competition and efficiency in the EU

banking sector, Kredit und Capital, vol. 37, pp. 329–352.

Weill, L. (2009). Convergence in banking efficiency across European countries,

Journal of International Financial Markets, Institutions and Money, vol. 19, no. 5, pp.

818-833.

Weill, L. (2011). Do Islamic Banks Have Greater Market Power&quest, Comparative

Economic Studies, vol. 53, no. 2, pp. 291-306.

Weinstein, M. (1964). The sum of values from a normal and a truncated normal

distribution, Technometrics, vol. 6, pp. 104-105, 469-470.

388

Wheelock, D.C. and Wilson, P.W. (2001). New evidence on returns to scale and

product mix among US commercial banks, Journal of Monetary Economics, vol. 47,

no. 3, pp. 653-674.

White, A.P. (1982). A note on market structure measures and the characteristics of the

markets that they „Measure‟, Southern Economic Journal, vol. 49, no. 2, pp. 542-549.

White, H. (1980). Using least squares to approximate unknown regression functions,

International Economic Review, vol. 21, no. 1, pp. 149-170.

Williams, J. and Gardener, E. (2003). The efficiency of European regional banking,

Regional Studies, vol. 37, no. 4, pp. 321-330.

Williamson, S.D. (1986). Costly monitoring, loan contracts and equilibrium credit

rationing, Quarterly Journal of Economics, vol. 102, pp. 135-46.

Wilson, P.W. (1995). Detecting influential observations in data envelopment analysis,

Journal of productivity analysis, vol. 6, no. 1, pp. 27-45.

Winton, A. (1995). Delegated monitoring and bank structure in a finite economy,

Journal of Financial Intermediation, vol. 4, pp. 158-187.

Xue, M. and Harker, P.T. (1999). Overcoming the inherent dependency of DEA

efficiency scores: a bootstrap approach, Unpublished Working Paper, Wharton

Financial Institutions Center, University of Pennsylvania.

Yeyati, E.L. and Micco, A. (2007). Concentration and foreign penetration in Latin

American banking sectors: Impact on competition and risk, Journal of Banking and

Finance, vol. 31, no. 6, pp. 1633-1647.

Yildirim, H.S. and Philippatos, G.C. (2007a). Competition and contestability in

Central and Eastern European banking markets, Managerial Finance, vol. 33, no. 3,

pp. 195-209.

Yildirim, H.S. and Philippatos, G.C. (2007b). Efficiency of banks: recent evidence

from the transition economies of Europe, 1993–2000, European Journal of Finance,

vol. 13, no. 2, pp. 123-143.

389

Zardkoohi, A. and Kolari, J. (1994). Branch office economies of scale and scope:

Evidence from savings banks in Finland, Journal of Banking and Finance, vol. 18, no.

3, pp. 421-432.

Zellner, A. and Tobias, J. (2001). Further results on Bayesian method of moments

analysis of the multiple regression model, International Economic Review, vol. 42,

pp. 121-140.

Zhang, T. and Garvey, E. (2008). A comparative analysis of multi-output frontier

models, Journal of Zhejiang University – Science A, vol. 9, no. 10, pp. 1426-1436.

Zhu, J. (2003). Quantitative Models for Performance Evaluation and Benchmarking:

Data Envelopment Analysis with Spreadsheets, Kluwer Academic Publishers.

Zhu, J. (2004). Imprecise DEA via standard linear DEA models with a revisit to a

Korean mobile telecommunication company, Operations Research, vol. 52, no.2, pp.

323–329.

Zhu, J. and Shen, H.Z. (1995). A discussion of testing DMUs‟ returns to scale,

European Journal of Operational Research, vol. 81, pp. 590–596.

390

Thesis Appendix

Due to space considerations, we omit the regression output since it possibly covers

more than the half of the actual thesis size.

We maintain our commitment to provide evidence upon request.

Competition in the European banking sector

Documents

Transcript of Competition in the European banking sector