AMERICAN JOURNAL OF SOCIAL AND MANAGEMENT SCIENCES ISSN Print: 2156-1540, ISSN Online: 2151-1559, doi:10.5251/ajsms.2010.1.2.102.123

© 2010, ScienceHuβ, http://www.scihub.org/AJSMS

Banking relationship and credit terms: empirical evidence from Portuguese small firms

Miguel Neves Matias1, Zélia Serrasqueiro2 and Carlos Arriaga Costa3 1Management and Economics Department of Polytechnic Institute of Leiria and Management and Sustainability Research Center Leiria, Campus 2, Morro do Lena - Alto do Vieiro, 2411-

901 Leiria Apartado 4163, Portugal e-mail: miguel.matias@estg.ipleiria.pt

2Management and Economics Department, Beira Interior University, Estrada do Sineiro, 6200-209, Covilhã Portugal, and CEFAGE Research Center, Évora University

e-mail: zelia@ubi.pt 3Economics Department of Minho University, Campus de Gualtar

4710 - 057 Braga, Portugal e-mail:caac@eeg.uminho.pt

ABSTRACT

Based on a sample of Micro and Small-sized Enterprises (MSE) financed by a large Portuguese bank, this study seeks to assess the importance of the banking relationship for MSE to obtain better credit terms, i.e. higher credit limit, lower risk premium and less probability that collateral must be provided. We conclude that long term banking relationship alone is not important for firms to obtain better bank credit terms. However, greater borrowing concentration allows MSE to obtain a lower risk premium and a higher limit of credit. The magnitude of these impacts are affected by some specificities related to firm size, economic-financial performance and the position of the lending bank in the local banking market.

Keywords: Banking relationship; Credit terms; Small firms.

INTRODUCTION

Small businesses, which play a key role in generating employment and creating economic wealth, face restrictions in accessing bank credit and in the terms of loan conditions. The significant difficulties in accessing external finance arise partly from small firms´ financial fragility and the legal opacity of information that they give to the lenders. The relationship between accounting information and the economic-financial performance, actually in evidence, is frequently in question, given the freedom of accounting "elaboration", rarely audited, not rated and with multiple schemes for presenting their financial statements, leading to the minimization of the tax burden (Le Guirriec, 1996). It is in this context that assessment of the banking relationship acquires vital importance in determining the credit risk of Small and Medium-sized Enterprises (SME). SME operate in a particular context, characterised by heavy interdependence; between the "personal sphere" of the business-owner and the "business sphere"; and the administrative and financial

organization are characterized by informality of relationships; and, by the vital and dominant importance of the business-owner (Levin and Travis, 1987; Belletante and Levratto; 1995). Indeed, the profile, know-how, and reputation of the SMEs´ business-owner emerge as the most important determinants in credit decisions (Levratto, 2001). However, most of those attributes are included in the so-called “soft information”, i.e. information of a qualitative nature that is not generally recorded uniformly in databases, and is not easily or faithfully transmitted between the various organs deciding bank credit (Stein, 2002). It should be noted that the changes imposed by the Basel II Capital Agreement have hastened the need for banks to acquire tools of a quantitative nature which allow, in the field of credit risk, establishment of capital requirements more adjusted to the risks of the assets present in bank (credit) portfolios. This process has required sophistication of the systems of managing individual credit risk, with the aim of

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

103

making the rating and scoring models more sensitive to knowledge about the client, and mainly, to his/her relationship with the bank. However, the capacity and intensity of the banking relationship cannot always be inserted in a rating or scoring model, because the relationship is a more and more complex, dynamic and multi-faceted concept, and so it is not always quantifiable or immune to the subjectivity of assessment by the bank manager who is the intermediary in the credit relationship. It therefore becomes difficult to reproduce, rigorously and uniformly, in the client database and introduce and contemplate, at each moment, in a standard model of credit risk assessment. Several studies (Riding et al., 1994; Blackwell and Winters, 1997; Johnson, 1998; Athavale and Edmister, 1999) show that repeated transactions over time between banks and SME contribute to reduce information asymmetry between the parties, diminishing the bank’s need to control SME activity and, consequently, it becomes possible for them to obtain an overall improvement in credit terms. Portugal has a bank-based financial system with an undeveloped stock market in comparison with stock markets of USA and other European countries. Portuguese SME represent about 99.8% of the Portuguese business sector, being fundamental for increased employment and the country’s economic growth. Most Portuguese SME do not have the necessary requirements to gain access to the stock market. Portuguese SME are extremely volatile in terms of operating income, have limited access to a variety of sources of finance and are strongly dependent on bank debt, making them vulnerable to economic fluctuations. Considering the importance of bank debt for Portuguese SME, especially for the smaller firms, the aim of this study is to identify and analyse the influence of the banking relationship on the loan terms of small firms in Portugal, using a sample of Portuguese Micro and Small Enterprises (MSE) taking out short-term credit. Besides assessment of the existing banking relationship, descriptive indicators of the quality of the firm, its management and its economic-financial performance are also analysed. Compared to the majority of studies investigating the influence of the banking relationship on credit terms

granted to SME, this study aims to contribute for the literature on the topic with a more rigorous definition of the variables used in measuring the banking relationship and a greater control of the influence of factors outside the investigation, highlighting the importance of the bank market structure and the distance from the bank’s credit department, allowing also in this way, a broader concept of the various implications of this phenomenon in financing SME. It intends to contribute to firms recognizing objectively observable determinants of the terms of finance obtained, highlighting those which are directly linked to the banking relationship, so as to minimize difficulties in the relationship and information asymmetry, with a view to achieving a more efficient performance at the lending bank. The current study is structured as follows, after this introduction; section 2 presents a bibliography review; section 3 presents the empirical analysis carried out; and finally in section 4 we draw the conclusions from this study.

The impact of the banking relationship on SME credit: theory and empirical evidence: Theoretical investigation has showed that the relationship between the borrower and the bank ensures a climate of confidentiality, improves contractual flexibility, reduces agency problems through increased control and allows the creation of reputation. However, the abstract nature of the theoretical benefits of the banking relationship has drawn investigators to observe its concrete effects, dealing more with the analysis of its impact in terms of access to credit, credit limits, risk premium, collateral requirements and the ease of renegotiating the contract terms, thereby facilitating implicit long term contracting. The majority of empirical investigations have concluded that the banking relationship increases access to and improves loan conditions and, consequently increases firm performance and value. The main conclusions are generally reached through evaluation and measuring of the two main indicators of the banking relationship: the length of the relationship and the extent of bank products acquired. However, it should be noted that the conclusions of the research carried out are far from consensual, since not only benefits are found in the relationship. Problems related to the banking relationship emerge,

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

104

when the value created is not shared equally between the bank and the financed firm. And so the banking relationship includes some disadvantages that affect both the bank and the borrowing firm, and which Boot (2000) sums up as the “soft-budget constraint problem” and the “hold-up problem”. The first problem concerns a weaker monetary restriction to which the borrowing firm is subject and which is explained by the bank’s greater ease in granting credit (and a higher amount than the risk actually presents) to the firm with which a certain relationship of confidence is already consolidated, based on the collection and regular analysis of soft information (Dewatripont and Maskin, 1995; Bolton and Scharfstein, 1996). While the second problem arises from the information monopoly that the bank develops over the length of the finance and could mean it practising higher pricings than those it would define in conditions of equal share of the benefits (Greenbaum et al. 1989; Sharpe 1990; Rajan, 1992; von Thadden, 1995), particularly in local markets where the bank enjoys significant market power.

Length of the banking relationship: The length of the banking relationship has been used as a proxy of the solidness of the relationship, and therefore, a determinant of the main loan terms obtained by SME. The benefits of the length on credit terms concerning the interest rate obtained on the credit find theoretical support in the studies by Diamond (1989; 1991). Later, Boot and Thakor (1994) show on basis of a theoretical model that, at first, firms may have to bear high interest rates, but as the credit relationship extends going through a series of successful credits, the interest rate decreases and there is also more probability of diminished collateral attached to the credit. In the first empirical studies that assess the importance of the length of the banking relationship in obtaining better loan conditions, particularly greater availability of credit and lower interest rates, the investigations by Berger and Udell (1995) and Petersen and Rajan (1994) stand out. Petersen and Rajan (1994)3 conclude that the length of the banking relationship does not have statistically significant influence on the interest rate on credit that the bank offers to the firm. However, it appears to influence positively the availability of bank credit to clients. Firms with longer banking relationships

record greater access to credit than those with a shorter relationship with the bank. In a later investigation that used the same set of data, Berger and Udell (1995) argue that Petersen and Rajan (1994) failed to determine a relationship between interest rates and length, because they did not focus on the most important form of commercial bank loan: the line of credit. Berger and Udell (1995) concluded that the interest rates charged on lines of credit diminish as the length of the relationship increases. The authors argue that the results are consistent with the fact that a banking relationship involves disclosure of “soft information” which improves contractual conditions for the firm, and so banks do not appear to apply monopoly pricings. Considering the studies by Petersen and Rajan (1994) and Berger and Udell (1995), Cole (1998)4 shows that the likelihood of access to credit increases according to the length of the relationship in the first year, but does not increase thereafter. Bodenhorn (2003) and Brick et al. (2007) also find evidence of benefits from the length of the credit relationship in terms of interest rates and the probability of attaching collateral to the credit. In general, the results obtained in studies carried out in American firms seem not to be easily transposed to the European case, particularly those related to the impact in terms of credit costs. Elsas and Krahnen (1998), based on data about credit granted by large German banks, do not identify significant relationships between the length of the banking relationship and interest rate or greater access to credit. Similarly, Elsas (2005) and Harhoff and Körting (1998) do not identify a significant relationship between the length of the banking relationship and the cost of, or access to credit by German SME. However, Harhoff and Körting (1998) obtain a significant and negative relationship between the length of the banking relationship and the requirement for collateral. Angelini et al. (1998) and Auria et al. (1999), conclude, in the case of Italy, that length of the banking relationship is positively associated with the risk premium of credit. Also Degryse and Van Cayseele (2000), for Belgium, and Vigneron (2007), for France, arrived at similar results. Regarding the impact of the length of the relationship in terms of the amount of credit obtained, little empirical investigation exists. We make special mention of the work by Fernando et al. (2002) who

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

105

find some advantages of a long banking relationship in the amount of credit obtained by small North-American firms. As for the effect in terms of the probability of attaching collateral, Boot and Thakor (1994) demonstrate theoretically that firms with longer banking relationships are less likely to provide collateral to loans. Berger and Udell (1995), Cowling (1999) and Brick et al. (2007) also find an inverse relationship between the length of the banking relationship and the probability of attach collaterals to the loans, based on a sample of small North-American firms. Also Vigneron (2007), based on a sample of borrowers at a large French bank, finds evidence of an inverse relationship between the quality of the collateral attached to the loan and the length of the banking relationship. Hernández-Cánovas and Martínez-Solano (2006) conclude for Spain, dividing the sample collected in two sub-samples according to the median length of the relationship, that for relationships under 15 years, the bank would exercise its monopoly power by imposing higher risk premiums and by increasing the probability of attaching collateral to the granting credit. The extent of banking products acquired from the lending bank: The strength or intensity of a banking relationship is also defined in terms of the extent of financial products/services offered by the bank and used by the borrowing SME. There is little evidence to document the influence of the extent of products acquired on the loan terms offered to borrowing firms. The borrowing concentration with the lending bank or the number of credit products has been used more frequently. Together with this last indicator, there is also extensive investigation that analyses the theoretical and empirical impact of the number of banking relationships on credit terms obtained. Indeed, in order to minimize the difficulty of obtaining data that permit assessment of the extent of acquired products and the related information that can be extracted, sometimes only the financing aspect is used, through determination of the borrowing concentration with the bank. This method is used by several authors (Blackwell and Winters, 1997; Charlier, 1998; Harhoff and Körting, 1998; Athavale and Edmister, 1999; Auria et al., 1999) that conclude that a greater concentration of credit with the lender is associated with lower cost of credit, greater access to credit or even more granting credit.

Concerning particularly the impact on the interest rate, Bédué and Lévy (1997) find that the more a firm concentrates its loans in just one bank, the lower the cost of credit. Worthington (1999) arrives at a similar result, concluding that the interest rate associated with a loan granted to a new client was between 40 and 50 points higher than that offered to a client with earlier experience of credit. Auria et al. (1999), Blackwell and Winters (1997), Elsas (2005) find empirical evidence that shows that greater borrowing concentration with the lending bank allows also a reduction in interest rate. As for the impact in terms of the credit limit obtained, Charlier (1998) concludes most firms that concentrated their commitments with the main bank saw their credit limit almost doubled, compared to the other firms making up the analysed sample. Analysing the impact in terms of the probability of attaching collateral, Forestieri and Tirri (2002) and Ziane (2003) found empirical evidence of a greater concentration of credit with the main bank increasing the probability of attaching collateral. Structure of the bank market: Analysis of the bank market structure in the credit relationship with SME has multiple features, presents great specificity and interacts with other variables, that are not always understood or duly pondered, and is therefore complex and ambiguous. In spite of theoretical models in the area of industrial economics generally alerting to the inefficiencies of market power, reflected in less availability of credit and higher pricings, a more concentrated market can give value to the credit relationship, by allowing a reduction of information asymmetry and agency problems (Guzman, 2000). The investigation carried out is not consensual in the conclusions reached6. It has generated divergent results, presenting a relationship which is sometimes positive and sometimes negative between the degree of market power in the banking system and the value of the banking relationship, with diverse and rather unclear effects on credit terms for SME. The defenders of the virtues of banking concentration in the banking relationship, reflected in beneficial credit terms for SME, support their stance with the fact that banking relationships facilitate a continuous flow of information between the debtor-firm and the creditor-bank, which can in turn guarantee

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

106

continuous (uninterrupted) access to bank finance, and so it will be in a less competitive financial system that banks may give more importance to client relationships (Boot and Schmeits, 2005; Petersen and Rajan, 1995). In competitive banking markets, soft information tends to get scattered, its collection being therefore more costly for the lending bank integrated in that market, and consequently that cost will have to be reflected in higher credit pricings, particularly for small borrowers (Marquez, 2002). However, despite the various virtues referred to the influence of banking concentration on SME access to credit, for other investigators, credit terms do not seem to benefit from a market structure characterised by high banking concentration or reduced competitive intensity. Boot and Thakor (2000), based on a theoretical model, demonstrate that as the intensity of inter-bank competitiveness increases, banks concentrate more on the credit relationship but with less value for the borrowing firm. Processes of banking concentration reduce credit to SME (Berger et al., 1998) and are associated with significant spreads (Degryse and Ongena, 2005). Concentration seems to destroy existing relationships and marginalize small banks (more focused on credit to this type of firm). The increased competitiveness in the bank credit market forces the reduction of spreads, but can also motivate banks to compensate for the reduced financial income, through deepening the relationship, “tying” clients through the sale of other products that generate commission, via cross-selling The empirical investigation by Blackwell and Winters (2000) finds an inverse relationship, for the case of lines of credit destined to small firms, between competition in the local banking markets and average interest rates. Beck et al. (2004), from a database containing information about the obstacles in accessing credit and the financing patterns in 74 countries, also conclude that movements of banking concentration, whatever their intensity, increase firms’ difficulty in accessing credit, that impact tending to decrease as firm size increases. To summarize, the relationship, between the degree of concentration of banking markets and the value of the credit relationship for SME, does not seem to be monotonic. It is therefore no wonder that investigation carried out is not consensual in terms of its impact on the loan conditions obtained by SME. Therefore,

most investigators who tackle this subject begin by admitting that future research is still necessary to examine and explain these conclusions in greater detail (Boot, 2000; Boot and Schmeits, 2005; Degryse and Ongena, 2007; Udell, 2008).

Empirical analysis: Based on the review and systematization of the literature referred to above, we begin the empirical investigation, carried out from the position of the granter of credit and where we assess the effects of the banking relationship, the quality of the financed firm and some specific conditions on the loan terms obtained by SME. Research hypotheses: The banking relationship, evaluated by its length and by the degree of borrowing concentration with the lending bank, is positively associated with the loan conditions for SME assessed in respect of interest rate, credit limit and collateral requirements. Firms with a long relationship with a certain bank (associated to which there are also previous successful credit experiences) and with a greater borrowing concentration with the lending bank have better loan conditions7 than those obtained by firms with a recent relationship, or with a reduced borrowing concentration with that bank. As referred to, given that loan conditions are assessed concerning risk premium, credit limit and attachment of collateral, the following three hypotheses arise. A greater banking relationship with a SME allows for a reduction of information asymmetry (better knowledge of the borrower, based on the “soft information” acquired through his banking behaviour over time), with beneficial effects in less intense monitoring, allowing the lending bank to grant lower pricings. Considering this argument, we formulate the following hypothesis:

H1: the banking relationship is negatively associated with the risk premium (spread) of the loan, ceteris paribus.

The banking relationship also encourages more information clarity in the elements of accounting made available, and so the proportionate reduction of risk allows the lending bank to grant short-term operations with a higher credit limit. On basis of this argument, we formulate the second hypothesis:

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

107

H2: the banking relationship is positively associated with the credit limit (amount), ceteris paribus.

By allowing a reduction of information asymmetry, the banking relationship provides a minimization of the problems arising from adverse selection and moral hazard, which can also reduce the possibility of attaching collateral (e.g. real guarantees, rather than personal guarantees) to the credit obtained by the financed firm. On basis of this explanation, we formulate the following hypothesis: H3 the banking relationship is negatively associated with the quality of the guarantees attached to the loan, ceteris paribus. Research Sample: We obtained a suitable sample of 436 observations from a large Portuguese CI containing information related to the characteristics of the banking relationship of SME holding a short-term line of credit, in the form of a current account, with a limit not above 500.000 €., in a situation considered normal8. The data were collected in 2002 and refer to the centre region of Portugal. Relating to statistical treatment applied in this study, with special prominence for the use of multiple and logistic linear regression models, it was also necessary to subject the sample to a process of detection, analysis and later elimination of abnormal observations (outliers and points of leverage), referring to the dependent and explanatory variables, with the elimination of 10 observations, the number of observations in the sample settling at 426. The past financial-economic information gathered refers to the last period known on that date which was 2001. For reasons of banking secrecy, and also statistical secrecy, the information supplied omitted any data which could lead to identification of the borrowers. The information about the type of place and structure of the local bank market was collected from the Annual Statistical Account of the National Statistics Institute (INE) and the Biannual Bulletin of the Portuguese Association of Banks.

Characterisation of firms making up the sample: In the universe of SME, the investigation centred its analysis specifically on MSE, as these firms are even more exposed to serious problems of information asymmetry, due to their reduced size and absence of rating (Hernández-Cánovas and Martínez-Solano, 2006), the even greater importance of the banking relationship being recognized in obtaining better loan conditions. Besides, the preference for MSE was to meet the aim of reaching a greater degree of homogeneity, and banish completely the possibility of observed firms turning to the capital market for finance. The option for the “short-term limit” form of credit is related to the fact of it not being used for any specific previously defined investment, assessed and later monitored by the bank, and requiring periodic revision of the credit terms, where (re)evaluation of the quality of the existing banking relationship becomes crucially important (Fernando et al. 2002). In particular, the selection of borrowers observes the following criteria: - Portuguese MSE, according to the ruling European definition at the time, i.e., firms with fewer than 50 employees, an annual business volume of no more than 7 million euros or an annual total balance sheet not exceeding 5 million euros and meeting the independence criterion defined in the Commission’s Recommendation 96/280/CE, of 3 April 1996; - firms unrelated to a group, understood as not owning or being owned9 by other firms in more than 25% of equity capital; - bank credit is used for its main economic activity, defined in terms of the NACE (Nomenclature of Classification of Economic Activities); - firms with organized accountability10; - firms with loans and deposits with the bank; - firms with headquarters situated in the same administrative area11 as the branch of the bank where the loan was granted. The composition the research sample by industry sectors is presented in appendix 1 table A1.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

108

Appendix Table A1: Composition of research sample by industry sectors

Sample Universe S ample Universe Sample Universe

A gric ulture 5,40% 2,49% 5,91% 1,46% 5,62% 1,34%

E xt ractive Indust ry 0,70% 0,31% 0,70% 0,55% 0,16% 0,90%

Manufacturing Industry 19,48% 14,22% 17,38% 23,62% 11,64% 21,46%

Construct ion and Rea l Estate 30,75% 28,23% 35,61% 18,76% 60,72% 46,40%

Construct ion 27,46% 12,28% 34,38% 10,40% 60,06% 20,31%

Commerce 28,64% 32,12% 30,66% 45,50% 15,90% 17,80%

A ccommodation 2,82% 9,41% 1,70% 2,72% 1,23% 2,08%

Transport 1,64% 6,17% 1,94% 4,99% 1,10% 5,03%

E duc ation , Health and Other Services 10,56% 7,04% 6,10% 2,40% 3,63% 4,98%

Industry sectorNumber of Firms Volume of Sales Bank Debt

Table 1: Definition of variables

VARIABLE DESIGNATION DEFINITION

I. Terms of Credit Operation

RPM Risk premium Spread (in %) that is added to the market index (Euribor rate), and expresses the risk premium of the loan attributed by the lending bank (in classes).

LIM Credit limit Natural logarithm of the limit (amount) of the credit operation (in thousands of euros) held with the lending bank

COLL Collateral Dummy variable with the value of "1" when the guarantee associated with the credit operation is a collateral, with a value equal to the loan and "0" in the remaining situations.

II. Banking Relationship

LBR Length of banking relationship Number of years of relationship with the lending bank (in classes)

DBC Borrowing Concentration Bank debt concentration (in %) with the lending bank (in classes)

III. Firm Quality

REP Reputation Number of years of activity (in classes)

SIZE Size Natural logarithm of Liquid Assets (in thousands of euros)

DEBT Debt = (Liabilities / Liquid Assets)*100

SAL Sales Natural logarithm of value of sales + services provided (in thousands of euros)

GLIQ General Liquidity = (Current Assets / Current Liabilities )*100

OPI Operating Income = Operating Revenue - Operating Costs (in thousands of euros)

DEVP

Deviation of EBITDA to Sales compared to

the median for the activity

= ESi / ESNACE

ES being ratio EBITDA to Sales (in %); i, the firm and NACE, the Code of Economic Activity (to 5 digits) of the activity firm i belongs to

MKTS Bank’s local market power

Market share (in %) in terms of the number of branches of the lending bank in the local market (unit of local government) where the borrowing firm has its headquarters, in relation to the total number of bank branches there

IND Industrial firm Dummy variable with the value of “1” if credit is used in an INDUSTRIAL firm and “0” otherwise.

CONST Construction firm

Dummy variable with the value of “1” if credit is used in a CONSTRUCTION or REAL ESTATE firm and “0” otherwise

RGB Regional branch Dummy variable with the value of "1" if the place (unit of local government) where the credit is granted is where the regional credit department is located, and “0” otherwise.

DEC Decision Number of decision organs of the credit operation

PART Partners Number of firm partners/shareholders, with participation of at least 5%

VII. Control Variables

IV. Bank’s Local Market Power

VI. Location

V. “Industry” Effect

Characterisation of the type of bank loan analysed: From the outset, our preliminary concern was that the information gathered on credit should refer to only one determined line of non-specialized finance available indistinctly to all the MSE chosen for the study. By choosing not to introduce various different forms of finance in the analysis, we avoided

the need to introduce other forms of risk assessment in the study, and therefore simplified the analysis. By selecting a non-specialized line of credit accessible indiscriminately to all MSE, we avoided the credit terms obtained being correlated with the special framework of a special type of credit more accessible to a certain sector or branch of activity or sub-group

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

109

of firms (e.g. lines of credit with special rates due to an agreement with an association in a certain sector), or with other discriminatory motives outside the investigation, besides the quality of the existing relationship between the bank and the financed firm and the economic-financial performance in evidence. For the previous reasons, in the diversity of credit products generally contracted by MSE, the one whose definition of credit terms depends most on the quality of the existing relationship and the risk profile of the applicant rather than the project, business, or activity, is the short-term (renewable) loan, in the form of a credit limit, destined to minimize cyclical financial needs arising from the firm’s operating activity, which is also the predominant form of financing in small firms.

It is a form of short-term credit destined to temporary working capital needs, arising from the firm’s operating activity. The bank monitors and revalidates its use periodically, which contributes to gaining more and better information about the firm, also allowing the result of periodic assessment of the risk of the operation to have possible repercussions in altered terms for the operation in the following period. In this attribution of a maximum value of short-term finance appropriate to the perceived risk of the client, besides the quality of the existing banking relationship11, there is evaluation of descriptive indicators of the quality of the firm, namely assessment of management and economic-financial performance.

In this way, we isolate the problem of the loans terms (amount, interest rate or collateral requirements) of the MSE included in the sample being influenced by external circumstances coming together at certain moments in time, caused by a change in the bank’s internal credit policy, or by the firm’s performance in the period before the first (or other) credit contract, which could happen in a medium or long-term credit, since loan terms are generally defined on the date the credit starts and do not change during the contract.

Definition of the research variables: As already mentioned, the principal aim of this research is to assess the importance of the banking relationship for small firms accessing bank finance, reflected in the loan conditions obtained. To do so, the first concern was to define the dependent variables that would characterise the credit terms obtained and introduce a first set of explanatory variables that would capture the quality of the existing banking relationship, and

subsequently other complementary determinant variables. Based on the previously presented investigation review, we now clarify objectively the main concepts and variables to be used in this study. The indicators included in the models tested were listed from the main studies published about this subject and already referred to in the theoretical and empirical review. The information extracted aimed to correspond to seven distinct components: considering the dependent variables they refer to generic contract terms of the type of loan selected; and, concerning the independent variables they include characteristics identifying the borrowing MSE banking relationship; firm’s economic-financial characteristics; identification of the local banking market structure; possible existence of “industry effects”; relevance of the distance from the regional credit department; control variables. Dependent variables: contract terms of the selected loan : The risk premium (RPM) is defined as the spread (in %) that is added to the market index (Euribor rate), and expresses the risk premium of the loan attributed by the lending bank (in classes). The risk premium is calculated in the same way as in several studies (Berger and Udell, 1995; Elsas and Krahnen, 1998; Forestieri and Tirri, 2002; Hernández-Cánovas and Martínez-Solano, 2006). Although it is common in the type of credit observed to set various types of commission (commission associated with the study, management, renewal and/or immobilization), which has the effect of increasing the cost of credit, the fact that the values for the research sample are almost invariably the same (and therefore unrelated to the quality of the operation or the proponent of credit), leads us to believe they are not defined as a function of the perceived risk of the operation or the client, and so they were not included in this analysis. Another variable related to the contract terms of the selected loan is the credit limit (LIM) defined as the natural logarithm (ln) of the limit (amount) of the short-term credit operation, borrowed by the observed firm from the lending bank (variable presented in this way in the study by Peltoniemi, 2007). Collateral (COLL) is a dummy variable with the value of "1" when the guarantee associated with the credit

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

110

operation is a collateral, with a value equal to the loan and "0" in the remaining situations (variable presented in this way in the study by Peltoniemi, 2007). The literature in the field also suggests that attachment of more and better guarantees reduces the cost of the credit. Indeed, the guarantee attached to the credit is associated with a certain consumption of capital by the bank. We generally find that loans with collateral attached will have a lower opportunity cost for the bank than a credit operation attached to personal guarantees (whatever the overall assessment of the operation’s risk), and so this variable can also be an explanatory factor of the final cost of the operation. We chose to introduce the variable “collateral” only as dependent variable, not including the “risk premium” variable in the set of independent variables. The collateral effect in definition of the risk premium is not considered relevant, in harmony with the conclusions obtained by Fredriksson (2007) and Degryse and Van Cayseele (2000). From observation of the bivariate correlations between the selected dependent variables12 we find low values (in absolute terms). The highest value detected (-0,581) refers to the correlation between the variables of “risk premium” and “credit limit”. This value is not worrying, considering the fact that in the process of analysing firm credit, the cost of credit is set independently of the definition of the credit limit granted/to be granted (Peltoniemi, 2007), and so the hypothesis of possible endogeneity between selected dependent variables is negligible. Next, we present the independent variables used in the models, in six groups: indicators of the banking relationship, firm quality, local market strength of the lending bank, “industry” effect, location and control variables.

Banking relationship: From what has been said, the banking relationship is considered to be an important criterion of risk analysis weighing decisively in the credit terms attributed to the borrower. Owing to the limitations in access to information about the sample obtained, the banking relationship is assessed by the length of the relationship under analysis and by the degree of borrowing concentration with the lending bank. The length of the banking relationship is a frequently used variable in various empirical studies (Petersen and Rajan, 1994; Berger and Udell, 1995; Blackwell

and Winters, 1997; Cowling, 1999; Cardone et al., 2005; Hernández-Cánovas and Martínez-Solano, 2006; Brick et al., 2007) as an indicator of the stability of the banking relationship. This study measures the length of the banking relationship (LBR) by the number of years of relationship with the lending bank, since becoming a customer, in classes. The degree of borrowing concentration variable (DBC) with the lending bank used in this study is a substitute of the number of lending banks used by the borrowing firm, which can also transmit an approximate picture of the stability of the banking relationship. This variable was used in several studies (Blackwell and Winters, 1997; Auria et al., 1999; Ewert and Schenk, 1998; Ziane, 2003).

Firm quality: The first and main indicator for assessing the credit risk of SME continues, inevitably, to be their past financial-economic performance, judging by the quantitative techniques for measuring credit risk in SME, which include various variables characterising their quality. Added to which are indicators of reputation and firm size which we designate as characterising their informational opacity. Almost all empirical studies carried out that analyse specifically the influence of the banking relationship on SME finance incorporate in their models independent variables identifying the reputation and size of the financed firm and its economic-financial performance which represent the borrower’s risk and are associated positively with the credit terms obtained (Blackwell and Winters, 1997; Steijvers and Voordeckers, 2002). Concerning assessment of informational opacity, the characteristics of reputation and firm size are observed, selecting the variables REP and SIZE, respectively. The reputation variable (REP) was measured by the number of years the firm has been active. This variable is associated with the firm’s past, and it indicates its reputation and benefits credit terms. This is the most commonly used variable in empirical research into topics to do with SME capital structure or finance and also inevitably in assessing the banking relationship. Firm age, associated with its past, indicates its reputation and benefits credit terms, particularly in terms of cost and attachment of collateral to the credit. The firm size variable (SIZE) is defined by the natural logarithm of the firm’s Liquid Assets. This variable is generally used to represent

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

111

firm size13 in investigatory studies carried out in the field of firm finance. A firm’s performance, and consequently also its risk, can be assessed from the financial and economic point of view. From the financial point of view, concerning assessment of the need for short-term finance, an examination of firm stability is important. Short-term finance requires the debt service to be essentially interlinked with the solidity of the firm balance sheet, since the main concern is to ensure adequate conversion of assets (e.g. stock, clients) held, in the operating activities, and so it is important to introduce indicators of the degree of firm debt, its liquidity and the size of the activity undertaken. From the economic point of view, analysis of profitability indicators gets special attention, particularly profitability arising from the MSE operating activity (where the finance is used). The selection criterion for the economic-financial indicators was based initially on a review of the various investigations carried out in the field of credit scoring and rating models for SME. Aiming to determine the degree of intensity of recourse to debt to finance a firm, we used the level of debt (DEBT), measured by the ratio of Liabilities to Liquid Assets. Volume of sales (SAL) is the variable indicating the size of operating activity carried out in the year, and is used in these terms in various empirical studies on this subject. The General Liquidity variable (GLIQ) used in this study is determined by the relationship between current assets and current liabilities. This ratio shows to what extent current liabilities are covered by assets that are expected to be converted in liquid financial means, in a period supposedly corresponding to the expiry of short-term debts. This variable has negative effects on the probability of getting a loan according to Bebczuk (2004). As the more liquid assets are, the easier it is to decapitalize the firm when bankruptcy approaches. Besides, high levels of liquidity can lead the bank believe that the firm intends to transfer the risk to the bank to transfer its own risk to the lender in the event of default. This variable does not appear very often in empirical studies dealing with this topic. Operating income (OPR) is determined by the difference between the operating revenue and the operating expenses. This variable expresses the profitability of the firm’s activity and allows a more

accurate approximation to the operating capacity to create a liquid margin, rather than net income. The analytical importance of the economic-financial indicators lies in the possibility of making comparisons with other firms in the same sector14. However, not all the indicators allow extraction of good information by comparing them with reference values. The value of the indicator of sales profitability should vary little among firms belonging to the same sector, so it is therefore an indicator whose comparability seems useful. A value found that is under the sector average means insufficient sales volume, excessive purchases or high production costs, which shows the absence of competitive advantage and therefore an added credit risk. We used the variable relating to the deviation of gross profitability of sales compared to the median for the activity (DEVP), as it is an indicator that shows the relative deviation of the margin of the firm’s operating activity (which is independent of the amortization or provision policies followed by the firm), compared to the average for the activity sector. Introduction of this indicator aimed to find the firm’s competitive advantage, within its activity sector. To compare the indicator of gross sales profitability of each firm with the sector average, we turned to the Bank of Portugal, Balance-Sheet Database, using when possible only the sub-set of SME, rather than the total sample15, for each NACE selected to 5 digits. Inclusion of this variable aimed also to reduce some accounting irrelevance that SME give to Credit Institution.

Bank’s local market power: With the aim of controlling the impact of the lending bank’s market share on the credit terms offered to borrowing firms, in terms of risk premium and credit limit, ceteris paribus, it was necessary to introduce a variable indicating the bank’s market strength in the local banking market. With the aim of controlling the impact of the lending bank’s market quota, on the credit terms granted to borrowing firms, ceteris paribus, it was necessary to introduce a variable indicating the bank’s market power, in the local banking market. This variable corresponds to the market share (MKTS) in terms of the number of branches of the lending bank, in relation to the total number of bank branches in a given local market (local government area) where the borrowing firm has its headquarters. This variable assumes values between “0” and “1”. High values of this indicator

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

112

mean greater market power held by the lending bank and consequently less intensity of competition in that market.

Industry effect: The theory defines the “industry” effect as one of the explanatory factors of finance options, and does not focus on its direct influence on loan conditions. Even so, this hypothesis is generally raised in the main empirical studies dealing with this subject. Volatility of income generation (associated with the industry sector) due to the specific risk of the business and in turn the credit operation, can also influence the credit terms granted. In the empirical studies analysed, this hypothesis has sometimes been tested, obtaining statistical significance, especially in the construction and manufacturing sectors. Based on the studies by Harhoff and Körting (1998) and Hernández-Cánovas and Martínez-Solano (2006), we decided to include some dummy variables indicating the sector industry of the financed firms which could influence credit terms, ceteris paribus. The variables considered were the following: manufacturing firm (IND) which is a dummy variable with the value of "1" if the credit is used in a manufacturing firm and “0” otherwise; construction firm (CONST) which is a dummy variable with the value of "1" if the credit is used in a construction or real estate firm and “0” otherwise.

Bank Localization: Since besides the borrowing firm’s bank branch, the regional department of credit always interferes in the credit decision, it is relevant to introduce a variable to assess the importance of the distance from regional credit department of the lender, concerning loan terms, in this way allowing us to draw conclusions as to the influence of distance on the value of the banking relationship and the ease with which “soft information” circulates In this way, we introduce the regional branch (RGB) variable, a dummy variable with the value of "1" if the place (unit of local government) where the credit is granted is where the regional credit department is located, and “0” otherwise. A variable introduced by Cowling (1999) and Vigneron (2007). Although generally speaking, proximity to the lending branch may mean added value to the relationship (Degryse and Ongena 2005), location in the same

local authority area as the regional credit management or regional branch where the credit relationship exists, can reflect better loan terms, compared to location near local branches. According to the empirical study by Cowling and Westhead (1996), the relationship with the local branch, although providing better service, is penalized in terms of interest rates. According to the empirical study by Cowling (1999), when the credit decision is made at the local branch, rather than at a regional office, the incidence of collateralisation increases.

Control variables: Finally, we introduce two regression variables which, although not contributing to solve any theoretical or empirical problem pointed out in the literature of the field, influence loan terms for SME, and so we introduce them here in order to control results. Being normal, in the decision matrix of banking credit, that credit operations registering unadjusted (more favourable) conditions in relation to the previously determined risk, have their approval based on a high number of decision organs, we chose to insert a variable controlling that fact. Irrespective of the perceived credit risk or the typical banking decision matrix, a credit operation being decided by a greater number of hierarchical organs will have better credit terms. In this way, the decision variable (DEC) determined by the number of decision organs in the credit operation aims to control situations where, despite an overall credit risk that could indicate granting of certain terms, these will be improved if a greater number of decision organs becomes part of the operation. In addition, similarly to what is found in the majority of empirical studies analysing determinants of the banking relationship, we decided to introduce a variable relating to the number of partners in the firm (PART), so as to control the impact of the firm’s partnership structure on credit terms. Model estimation: This study uses multiple linear regression models (MRLM), with recourse to the method of ordinary least squares (OLS), to evaluate the impact of the selected independent variables on the risk premium and the credit limit, and a logit model, with recourse to the method of maximum likelihood estimation, to evaluate the impact of the dichotomic variable “Collateral”. These options, besides appearing the most suitable for this study, also make it easier to compare the

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

113

results obtained with those reached in other studies that follow the same methodology. Similarly to what was done in the majority of the empirical studies analysed, we introduced the described pre-selected independent variables in a single econometric model, for each dependent variable under analysis. Given the importance, the extent and the foreseeable high correlation of the size variable (through the value of net assets,

designated SIZE) with credit terms (particularly the credit limit) the sample was divided by the median of the SIZE variable in two sub-samples, with the aim of exercising some control over this variable and on which the previously defined econometric models were also tested.

Analysis and discussion of the results: The descriptive statistics of research variables appear in table 2.

Table 2: Descriptive statistics

10 25 75 90

Dependent Variables

RPM 5,5 5,0 1,8 1,0 9,0 3,0 4,0 7,0 8,0 426

LIM 3,9 3,9 0,9 1,4 6,2 2,7 3,2 4,6 5,0 426

COLL 0,2 0,0 0,4 0,0 1,0 0,0 0,0 0,3 1,0 426Independents - Banking Relationship

LBR 2,9 3,0 0,8 1,0 4,0 2,0 2,0 3,0 4,0 426

DBC 2,9 3,0 1,1 1,0 4,0 1,0 2,0 4,0 4,0 426Independents - Firm quality

REP 4,5 4,0 1,4 1,0 7,0 3,0 4,0 5,0 6,0 426

SIZE 5,8 5,8 1,2 2,1 10,2 4,2 5,1 6,6 7,3 426

DEBT 80,1 82,0 29,5 2,6 275,0 46,3 65,1 93,6 103,5 426

SAL 5,4 5,7 1,6 0,0 8,9 3,9 4,8 6,4 7,0 426

GLIQ 1,1 0,9 1,9 0,0 31,7 0,3 0,6 1,1 1,6 426

OPI 27,2 12,0 75,3 -185,0 727,0 -14,3 1,0 30,0 79,6 426

DEVP 1,8 1,2 11,2 -101,6 172,4 0,0 0,6 2,1 3,7 426Independent - Bank’s Local Market Power

MKTS 0,2 0,1 0,1 0,0 0,5 0,0 0,1 0,2 0,3 426Independents - “Industry” effect

IND 0,2 0,0 0,4 0,0 1,0 0,0 0,0 0,0 1,0 426

CONST 0,3 0,0 0,4 0,0 1,0 0,0 0,0 1,0 1,0 426Independent - Location

RGB 0,3 0,0 0,5 0,0 1,0 0,0 0,0 1,0 1,0 426Independents - Control Variables

DEC 2,0 2,0 0,9 1,0 5,0 1,0 1,0 2,0 3,0 426

PART 2,3 2,0 0,9 1,0 6,0 1,0 2,0 3,0 3,3 426

Variables* Average Median Standard Deviation Minimum Maximum Percentiles N

(*) The complete designation of each variable can be found in Table 1. We find that the average (and the median) amount of each loan, was around 50.000 €. (value as a logarithm in thousands of euros being 3,9), with 25% of operations being no more than 25.000 €. (value as a logarithm in thousands of euros being 3,2). The average risk premium of the observed operations was in the spread interval (2% – 2,25%). Concerning guarantees attached to the operations, we see that

over 50% of the observed operations have no collateral attached. As for the characteristics of the existing banking relationship, we find that the average length of relationship is 7 years and the degree of borrowing concentration with the lending bank is 64%. In the bivariate correlation matrix, presented in table 3, we observe that the dependent variables present statistically significant correlations with the majority of explanatory variables, with only two coefficients registering over 50%.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

114

Table 3: Correlation matrix

RPM

LIM -0,581 **

COLL -0,369 ** 0,380 **

LBR 0,191 ** -0,028 -0,086

DBC -0,116 * -0,012 0,041 -0,095 *

REP 0,077 0,056 -0,012 0,622 ** -0,098 *

SIZE -0,368 ** 0,663 ** 0,186 ** 0,088 -0,241 ** 0,114 *

DEBT -0,003 0,004 0,105 * -0,112 * 0,017 -0,234 ** -0,033

SAL 0,004 0,067 -0,196 ** 0,212 ** -0,220 ** 0,227 ** 0,384 ** -0,144 **

GLIQ 0,039 -0,006 0,040 0,025 0,000 0,148 ** -0,125 ** -0,401 ** -0,010

OPI -0,234 ** 0,370 ** 0,083 0,042 -0,102 * 0,067 0,405 ** -0,048 0,231 ** 0,030

DEVP -0,077 0,078 0,007 -0,027 0,020 -0,047 0,048 -0,098 * -0,030 0,006 0,128 **

MKTS 0,209 ** -0,209 ** -0,045 0,135 ** 0,070 0,056 -0,166 ** -0,168 ** 0,024 0,042 -0,097 * 0,016

IND 0,040 -0,128 ** -0,032 -0,004 0,000 -0,020 0,014 -0,029 0,060 -0,102 * -0,058 -0,040 0,136 **

CONST -0,203 * 0,377 ** 0,193 ** 0,006 0,096 * -0,023 0,247 ** 0,109 * -0,108 * 0,070 0,205 ** -0,007 -0,024 -0,318 **

RGB -0,190 ** 0,147 ** 0,068 0,009 0,002 0,041 0,067 0,057 -0,056 -0,018 0,095 -0,005 -0,416 ** -0,162 ** 0,064

DEC -0,549 ** 0,446 ** 0,288 ** -0,207 ** -0,018 -0,146 ** 0,344 ** 0,272 ** -0,148 ** -0,105 * 0,001 -0,100 * -0,165 ** -0,039 0,222 ** 0,155 **

PART -0,178 ** 0,200 ** 0,010 0,017 0,030 -0,184 ** 0,167 ** 0,078 -0,038 -0,129 ** 0,122 * 0,109 * -0,133 ** -0,008 0,010 0,095 * 0,124 *

LBR RGB DECGLIQ OPI DEVP MKTSDBC IND CONSTREP SIZE DEBT SALRPM LIM COLL

The complete designation of each variable can be found in Table 1. (*) Correlations significant at 5% (**) Correlations significant at 1%

The “Risk premium” variable (RPM) is negatively correlated with the “Credit management” control variable (DEC) (r=-0,549, p<0,01), which agrees with the indications initially put forward, that to obtain lower risk premiums, operations would have to pass through a greater number of decision organs, ceteris paribus. The “Credit Limit” variable (LIM) is positively correlated with the “Size” variable (SIZE) (r=-0,663, p<0,01), indicating that the bigger the firm, the higher the credit limit, ceteris paribus. We also confirm a positive, and significant though not very pronounced correlation between the RPM variable and the SIZE variable. Concerning the correlation between dependent variables, as would be expected, the dependent variables of “Risk premium” (RPM) and “Credit limit” (LIM) also present high negative correlation (r=-0.581, p<0,01), which indicates that higher amounts of credit are associated with lower risk premiums, ceteris paribus.

Regarding the correlation between the dependent variables of “Risk premium” (RPM) and “Collateral” (COLL) we get a negative value but close to zero, which indicates that attaching higher quality guarantees tries to partly reduce problems arising from adverse selection, since when more or better guarantees are offered, lower interest rates are defined. In addition, the explanatory variables do not show a correlation completely in agreement with the relationships (signs) initially proposed in the hypotheses set out. Concerning assessment of the banking relationship on credit terms, the “length of the relationship” (LBR) presents contrary signs to those expected in relation to the RPM and LIM variables, indicating the disadvantage of a longer lasting relationship for credit terms. The “Concentration” variable (DBC) presents the expected signs in relation to the “Risk Premium” (RPM) and “Collateral” (COLL) variables. Besides, we find a significant and pronounced correlation between the “Credit management”

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

115

variable (RGB) and the “Bank’s local market strength” variable (MKTS), observing therefore that in local authority areas with a regional credit department (RGB=1), the market quota of the lending bank is low, and vice versa. As for the other explanatory variables, in general, the signs obtained by the significant correlation coefficients are as expected. The two econometric models, that have as dependent variables the “risk premium” (RPM) and the “credit limit” (LIM), allowing evaluation of hypotheses 1 and 2 respectively, and which were tested on the total sample (MRLM 1 and MRLM 4) and on two sub-samples: firms whose size is equal to or smaller than the median of the total sample (MRLM 2 and MRLM 5) and firms that are larger than the median of the total sample (MRLM 3 and MRLM 6), respectively, generated the results summarized in tables 4 and 5. The logit model, with the dichotomic dependent variable “Collateral” (COLL), allowing evaluation of

hypothesis 3 and which was also tested on the total sample (LOGIT 1) and on the two sub-samples: firms whose size is equal to or smaller than the total sample (LOGIT 2) and firms that are larger than the median of the total sample (LOGIT 3), respectively, generated the results summarized in table 6. Determinants of risk premium: Concerning hypothesis 1, formulated about the relationship between the banking relationship and the credit risk premium , we conclude, according to MRLM 1 (with dependent. variable RPM), that the length of the relationship (LBR) is positively associated with the risk premium of the loan, whatever the size of the firm, going against the formulated hypothesis and the main existing theory. This result was obtained by Angelini et al. (1998) and Auria et al. (1999) for Italy, by Degryse and Van Cayseele (2000) for Belgium, as well as by Vigneron (2007) for France. So, the results obtained in the current study are in line with the main European studies carried out.

Table 4: Determinants of the risk premium, depending on the MSE size (SIZE)

Dependent Variable

MRLM 1 MRLM 2 MRLM 3*** *** ***

( 13,193 ) ( 8,926 ) ( 6,8113 )

*** *( 2,625 ) ( 1,691 ) ( 1,273 )

*** *** ***( -4,838 ) ( -4,78 ) ( -4,686 )

( -1,442 ) ( -1,06 ) ( 0,0746 )*

( -1,82 ) ( 1,22 ) ( -1,295 )*** *** ***

( 4,4532 ) ( 3,115 ) ( 3,9679 )

( -0,848 ) ( -0,82 ) ( 0,8592 )*

( 1,3554 ) ( 1,771 ) ( 0,752 )*** ** ***

( -3,354 ) ( -2,201 ) ( -3,502 )**

( -2,384 ) ( -1,348 ) ( 0,6389 )

** ***( 2,447 ) ( 2,918 ) ( -0,368 )

*( 0,17 ) ( 1,833 ) ( -0,799 )

( -0,445 ) ( -1,41 ) ( 0,090 )

**( 1,200 ) ( -0,08 ) ( -2,468 )

*** *** ***( -13,05 ) ( -11,99 ) ( -11,16 )

** * *( -2,222 ) ( -1,69 ) ( -1,694 )

Nº obs.Adjusted R2

F statistic

“Industry” Effect

23,616

PART

0,61827,1070,498

RGB -0,185 -0,016

DEC -1,043 -1,206

0,380

CONST -0,072 -0,273IND 0,026

DEVP -0,009 -0,013

ICON 1,756 2,674Bank’s Local Market Power

GLIQ 0,060 0,115

OPI -0,004 -0,013

DEBT 0,011 0,009

SAL -0,043 -0,054

REP -0,083 -0,092

SIZE -0,122 0,107

DBC -0,266 -0,320

Firm Quality

LBR 0,255 0,240Banking Relationship

RPM(N=426)

RPM(LNAC ≤ 5,8 / N=210)

(Constant) 8,131 7,687

RPM(LNAC > 5,8 / N=216)

7,075

0,141

-0,300

0,005

-0,171

0,022

0,038

0,012

-0,003

0,009

-0,447

-0,169

0,018

-0,477

-0,895

0,59021,693

Location

Control Variables

-0,117

426 210 216

-0,167 -0,185

Independent Variables

The complete designation of each variable can be found in Table 1. The “t” statistic appears in brackets. (*) significant at 10%; (**) significant at 5% and (***) significant at 1%.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

116

This evidence is more robust for smaller firms (MRLM 2) and in places where the lending bank has greater market strength. We can therefore conclude lending banks make use of their information monopoly (hold up problem), due to the possibility of accumulated information, making it disadvantageous for the firm to prolong the banking relationship, as a way to obtain better credit terms. In the case of the larger firms in the sample (MRLM 3), the length of the credit relationship, despite presenting a positive relationship with the risk premium, is not significant, in harmony with the conclusions also obtained by Ziane (2003) for France or Cardone et al. (2005) for Spain.

However, we conclude, according to MRLM 1, MRLM 2 and MRLM 3 that the more a firm concentrates its credit with the lending bank (DBC), the lower the risk premium of the loan, in harmony with the conclusions obtained by Bédué and Levy (1997) for France, Harhoff and Körting (1998) for Germany, and Auria et al. (1999) for Italy. Besides the banking relationship, we see that firm size (through the SIZE variable) presents a positive impact on the risk premium attributed (coefficient with negative sign), but only in the total sample (MRLM 1). As for firm age (REP), although presenting the expected sign, it is not a statistically significant variable in any of the models analysed.

Regarding the remaining variables concerning firm’s financial-economic performance, the importance of “debt” (DEBT) and “profitability of operating activity” (OPI) indicators stand out, their statistically significant coefficients presenting the expected signs in the 3 models, in harmony with the conclusions obtained in the main empirical studies carried out. Debt records a negative impact on the risk premium obtained and profitability records an inverse effect. General liquidity (GLIQ) is not statistically significant, in terms of its impact on the risk premium of operations.

Concerning “Bank’s local market strength” (MKTS) we obtain a significant negative impact on credit terms in MRLM1 and MRLM2, an increase in the bank’s market strength being associated with worse loan terms for MSE and vice versa. In the sub-sample made up of larger MSE (MRLM3) that effect is not noticeable. As for the “industry effect”, its impact is not significant. Branch location near regional branch (RGB) was shown to be important in improving loan terms, reflected in reduced interest

rate, judging from the sign associated with the coefficient of the independent variable referred to. This result is in agreement with the conclusions obtained by Cowling and Westhead (1996). However, only in the case of larger MSE (MRLM 3) was there statistical significance.

Regarding the “control variables”, the quite significant impact and with the expected sign of the coefficients associated with the “Decision” variable (DEC) stands out. The coefficients of the “Partners” variable (PART) record the expected signs, but without statistical significance.

Determinants of credit limit: Concerning hypothesis 2, concerning the relationship between the banking relationship and the credit limit attributed, we conclude, according to MRLM 4 (dep. var. LNMONT), that the coefficient associated with the length of the banking relationship presents a negative sign, contrary to what was expected in the hypothesis formulated. However, the estimated coefficient is not statistically significant.

Once again, and also in accordance with MRLM 4, and with greater emphasis in smaller firms (MRLM 5), there is the marked importance of another indicator of the banking relationship, the borrowing concentration with the lending bank, as a way to obtain higher credit limits, in harmony with existing theory and with the empirical conclusions obtained by Charlier (1998), confirming therefore the hypothesis formulated.

The variables assessing firm quality, both the variables characterising information opacity and those characterising economic-financial performance, generally present the expected signs. Beginning with the variables characterising information opacity (“reputation” - REP and “size” - SIZE), we conclude that firm reputation (age), despite only being a significant variable when considering the total sample (MRLM4), it presents the expected signs in the 3 models, i.e. the greater the reputation, the higher the credit limit obtained. Firm size is seen to be significant and also with a positive impact on the credit limit in the 3 models.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

117

Table 5: Determinants of the credit limit, depending on the MSE size (SIZE) Dependent Variable

MLRM 4 MLRM 5 MLRM 6*** *** **

( 3,574 ) ( 6,458 ) ( 2,583 )

( -0,397 ) ( -0,287 ) ( -0,838 )*** ***

( 3,593 ) ( 4,068 ) ( 0,894 )

**( 2,498 ) ( 1,162 ) ( 0,697 )

*** *** ***( 11,569 ) ( 3,100 ) ( 6,306 )

*** ***( -0,519 ) ( 0,112 ) ( 0,254 )

***( -1,074 ) ( 1,76 ) ( -3,933 )

*** *** **( 3,1627 ) ( -0,244 ) ( 2,325 )

*** *** ***( 6,3899 ) ( 1,28 ) ( 4,939 )

( 0,8938 ) ( 2,422 ) ( -0,735 )

** **( -2,491 ) ( -1,477 ) ( -0,124 )

*** ***( -2,996 ) ( -2,195 ) ( 0,382 )

*** *** ***( 3,1152 ) ( 2,917 ) ( 2,789 )

*** *** ***( 7,3328 ) ( 3,573 ) ( 5,467 )

* *( 1,7776 ) ( 2,267 ) ( 1,572 )

Nº obs.Adjusted R2

F statistic0,604

24,458

0,032

0,279

0,224

0,051

426 210 216

0,053 0,071

0,001

-0,083

0,023

0,001

-0,003

-0,043

LIM(SIZE > 5,8 / N=216)

1,274

-0,037

0,026

0,023

0,393

Banking Relationship

LIM(N=426)

LIM(SIZE ≤ 5,8 / N=210)

(Constant) 0,797 1,694

DBC 0,075 0,086

Firm quality

LBR -0,015 -0,013

REP 0,064 0,034

SIZE 0,354 0,088

DEBT -0,001 0,000

SAL -0,025 0,066

Bank’s Local Market Power

GLIQ 0,020 -0,002

OPR 0,002 0,003

-0,134

CONST 0,228 0,242

DEVP 0,003 0,009

MKTS -0,491 -0,308

DEC 0,269 0,173

IND -0,180

17,490

PART

0,524

“Industry” effect

Control Variables

52,8300,63

Independent Variables

The complete designation of each variable can be found in Table 1. The “t” statistic appears in brackets. (*) significant at 10%; (**) significant at 5% and (***) significant at 1%.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

118

Regarding the variables characterising economic-financial performance, we find that the indicators of firms’ financial stability, reflected in the level of debt (DEBT) and profitability of operating activity (OPI) are the ones presenting statistical significance and the expected signs. In the models tried on the total sample (MRLM4), debt registers a negative impact on the credit limit obtained and profitability registers a positive impact. General liquidity (GLIQ), in turn, presents coefficients that are significant from a statistical point of view, although its impact is reduced and presents opposite signs in the sub-samples analysed (MRLM5 and MRLM6). In the case of smaller firms (MRLM5), the impact of liquidity is negative. This result corroborates the arguments of Bebczuk (2004). However, in the case of larger firms (MRLM6) the impact of general liquidity on the credit limit is positive. This result suggests that banks attribute less risk to larger firms with higher levels of general liquidity, to which they will associate a higher level of cash-flow.

For the “bank’s local market strength” variable (MKTS), we obtain a significant negative impact on the amount of credit in models MRLM4 and MRLM5, an increase in the bank’s market strength being associated with worse credit terms for firms, in terms of credit limit. These results are close to the ones obtained by Repetto et al. (2002) and are in harmony with the conclusions obtained by the theoretical models that warn of the inefficiencies of market strength for the consumer, reflected here in less availability of credit.

As for larger firms (MRLM6), although the relationship between the bank’s local market strength and the amount of credit is negative, it is not significant. This result agrees with the main conclusions of the study by Beck et al. (2004).

Regarding the “industry effect”, we find the impacts of the activity sectors analysed are not indifferent. Firms

in the manufacturing sector are exposed to greater restrictions in terms of credit limit than firms in the construction sector, in agreement with Fernando et al. (2002). The greater size of credit destined to construction firms can be related to the fact of this type of credit being directed to the integral development of investments in fixed assets, whereas for manufacturing firms it may be destined exclusively to finance working capital needs.

Finally, concerning the “control variables”, we highlight the positive and quite significant impact in all models and with the expected sign of the “Decision” variable (DEC). This result suggests that when a greater number of decision organs are involved in the credit operation, the credit limit was raised. The “Partners” variable (PART) registered a positive impact on the credit limit and presents statistical significance in MRLM4 and MRLM5. However, the number of partners seems to lose relevance when firms are larger (MRLM6). This result is not surprising, given that larger firms are managed by professional managers, and it is they who negotiate credit with banks, firm ownership structure (including the figure of the owner-manager) tending to lose importance.

Determinants of collateral requirements: Concerning hypothesis 3 formulated to test the relationship between the banking relationship and the guarantees attached to credit, although the variables indicative of the banking relationship introduced in the logit models do not present statistical significance, finding in general a greater impact of some variables indicating firm quality, similarly to what was found in the study by Elsas and Krahnen (1999), it is nevertheless possible to draw some conclusions, bearing in mind the signs presented by the coefficients.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

119

Table 6: Determinants of collateral, depending on the MSE size (SIZE)

Dependent Variable

LOGIT 1 LOGIT 2 LOGIT 3*** *

( 9,762 ) ( 0,549 ) ( 2,681 )

*( 1,073 ) ( 2,336 ) ( 0,064 )

( 0,995 ) ( 0,012 ) ( 0,794 )Firm Quality

( 1,504 ) ( 1,281 ) ( 0,059 )*** **

( 9,205 ) ( 2,251 ) ( 5,338 )** *

( 2,26 ) ( 3,946 ) ( 2,646 )*** *** ***

( 13,214 ) ( 7,773 ) ( 7,186 )

( 1,725 ) ( 0,004 ) ( 0,039 )

( 0,406 ) ( 0,034 ) ( 0,064 )

( 0,142 ) ( 0,013 ) ( 0,206 )“Industry” Effect

( 0,348 ) ( 0,106 ) ( 0,598 )**

( 0,961 ) ( 0,685 ) ( 3,988 )

( 0,016 ) ( 0,028 ) ( 0,276 )

*** ** ***( 6,932 ) ( 1,125 ) ( 12,497 )

*( 0,995 ) ( 1,838 ) ( 3,349 )

Nº obs.

Nagelkerke R2

-2LogL Statistic

PART

DEC

DEVP

IND

GLIQ

OPI

DEBT

SAL

REP

SIZE

LBR

DBC

(Constant)

COLL(N=426)

-3,699

-0,206

0,115

Banking Relationship-0,525 -0,068

-0,021 0,142

0,153

0,442

0,008

-0,361

0,122

0,001

0,004

0,192

COLL(LNAC ≤ 5,8 / N=210)

COLL(LNAC > 5,8 / N=216)

-1,575 -3,661

0,227 0,042

0,427 0,638

0,014 -0,018

-0,576 -0,330

0,019 0,097

0,002 0,001

-0,001 -0,012

0,154 0,381

0,304 -0,475 0,880

Location

CONST

-0,0760 0,2070

Control Variables-0,335 0,774

0,0350

0,435

RGB

0,337 -0,379

426 210 216

-0,152

0,170 0,345

410,765 178,326 208,836

0,217

Independent Variables

The complete designation of each variable can be found in Table 1. The “Wald” statistic appears in brackets. (*) significant at 10%; (**) significant at 5% and (***) significant at 1%.

We conclude that the banking relationship, assessed by its length, is important for reduced probability of attaching collateral to the credit, this effect being more significant in the case of smaller firms (LOGIT 2), in harmony with what was demonstrated

theoretically by Boot and Thakor (1994). It is also in agreement with the conclusions extracted from the empirical work of Berger and Udell (1995) for the case in North America and Vigneron (2007) for France, who found evidence of an inverse

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

120

relationship between the quality of the collateral attached to the credit operation and the length of the banking relationship.

However, borrowing concentration with the lending bank (DBC) seems to increase the probability of the borrower attaching collateral to the credit, judging only by the sign obtained in LOGIT 1 and to a certain extent in harmony with the conclusions obtained by Forestieri and Tirri (2002) for Italy and by Ziane (2003) for France. But that variable is not significant and presented different signs, but also with significance, in the attempt to control the “size” effect (LOGIT 2 and 3).

Regarding the impact of firm quality on the guarantees attached, firm size (SIZE), the activity and level of debt (DEBT) affect the guarantees of the loan in the expected direction, allowing the solution of problems arising from moral hazard. This agrees with Brau (2002) and Elsas and Krahnen (1999) and confirms hypothesis 3 of this study.

The impacts of the different industry sectors are not indifferent (particularly industry and construction). In the case of larger MSE (LOGIT 3), firms in the construction sector would be more likely to see collateral attached to their credit than firms in the manufacturing sector, similarly to the conclusions of Elsas and Krahnen (1998), Harhof and Körting (1998), and Hernández-Cánovas and Martínez-Solano (2006).

As for the impact of location on collateral requirements, besides not obtaining statistically significant values, the signs were also contradictory, depending on whether we considered the sub-sample composed of smaller firms (LOGIT 2) or larger ones (LOGIT 3), in the case of smaller firms obtaining a positive sign of the coefficient, revealing proximity of a regional department to be beneficial in terms of attaching guarantees of poorer quality, in harmony with Cowling (1999).

CONCLUSIONS

The risk of bank credit for SME generally measured through models of credit rating and scoring, gives special importance to past economic-financial performance, which due to the high informational opacity that these firms present, ends up forming the main determinant of the high credit risk perceived by banks. Refocusing on the value of the banking relationship should not be considered a backward step in the techniques of evaluating credit risk, but rather the need to go back to basics.

In this study, based on a sample of MSE financed by a large national bank, the value of the banking relationship for loan terms is assessed by the length of the banking relationship and by the degree of borrowing concentration with the lending bank.

Generally speaking, a longer-lasting banking relationship seems not to bring benefits to MSE, in terms of improved conditions of credit, which ties in with the main European studies carried out. However, a greater concentration of credit with the lending bank generally presents a positive impact on loan conditions, in the form of lower interest rates and higher credit limits.

Indeed, the length of the banking relationship only presents benefits concerning the reduction of collateral requirements and only in the case of larger firms, and is not reflected in lower risk premiums or higher credit limits. This indicator is seen to be quite weak in the impact it registers on loan terms. This conclusion is in harmony with the principal studies carried out in European countries, namely Spain, France, Belgium and Germany, and unlike the conclusions obtained in more representative studies carried out mainly in USA, based on samples drawn from the NSSBF.

On its own, the length of the banking relationship does not reveal the degree of solidity of the relationship, as the length of the relationship in isolation does not transmit any relevant information to the bank, and therefore some empirical studies carried out (Blackwell and Winters, 1997; Ewert and Schenk, 1998 and Uzzi, 1999 among others) have added a complementary variable indicating the number of bank products acquired, especially “capturing” products that allow at least two types of benefits for the lending bank: improved customer profitability (which, in an overall analysis, can allow to compensate for possible reductions in the risk premium on the credit) and accumulation of soft information from the way the various products are used, with direct benefits for the credit terms obtained by SME.

Therefore, from analysis of the process of evaluating credit, we see that the length of the banking relationship, if devoid of any record allowing assessment of past behaviour of movements and the quality of existing accounts, particularly credit ones, does not in the end make a positive contribution to determining perceived risk, and consequently credit terms agreed.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

121

The degree of concentration of bank credit with the bank presents greater analytical value in evaluating credit, and consequently in defining the credit terms obtained by SME. Indeed, a greater concentration of credit with the bank shows a firm’s loyalty to its bank and/or, alternatively, also the costs of changing, especially for smaller MSE, with greater economic-financial risk and/or in places where the bank enjoys significant market strength.

The degree of concentration of credit, together with some length of the banking relationship, has crucial importance in reducing the effects arising from adverse selection, as it lets the credit analyst obtain very important information that reveals experience of credit, which if positive, will favour the terms of new credit granted to the firm, ceteris paribus.

A firm’s informational opacity and economic-financial performance, considered characteristics of its intrinsic quality, are also determinants of loan terms obtained. Besides the banking relationship, reduced informational opacity together with good economic-financial performance observed in the period preceding the granting of credit, influence positively the conditions of bank finance obtained. In general, we obtained the expected signs in the coefficients associated with the main variables revealing firm quality (indicators of informational opacity and economic-financial performance), in harmony with the conclusions obtained in the main empirical studies carried out.

In addition, we found evidence that in local markets, where the bank has a dominant position, there is a gain in terms of information, which means reduced risk in credit operations. However, these gains may not be reflected in reduced cost or fewer demands in terms of the quality of guarantee attached to granting credit, as the bank benefits from its information monopoly and MSE may have difficulty in changing banks.

In general, it is firms that concentrate an important part of their bank credit with the lending bank that obtain the best credit terms (particularly lower risk premiums and higher credit limits), the size of this benefit however being affected by firm size, by their status in economic-financial terms, by the structure of the local banking market and by their proximity to where banking decisions are made.

From the investigation carried out, we can conclude that the cost of changing banks which is progressively generated by the bank for its credit client (through deepening of the relationship and

affected by significant market strength in the local banking market) lets the bank establish less competitive rates above the average market pricing (up to a given limit) especially for small MSE, the differential of which would form the cost of changing to another Credit Institution.

To sum up, SME that try to benefit from the advantages of a loyal (concerning concentration of their credit) and lasting relationship with a bank should bear in mind that the information monopoly acquired by the lending bank over the course of the relationship, together with some opacity by the borrowing firm and the bank’s greater strength in the local market, can mean disadvantages for the firm. Therefore, it should try mobility in terms of changing banks, something which is only viable in places where the local banking market is not very concentrated.

Briefly, the banking relationship will be beneficial for all parties in the bank credit market (for banks as supply side and for SME as demand side), and consequently for the economy in general, as long as there is no abuse of informational advantage by any party in the relationship (particularly the party that generally has more negotiating power – the lending bank) and concerning a possible reduction of incentives to monitor the development of granting.

REFERENCES Angelini, P., Salvo, R. and Ferri, G. (1998). Availability and

cost of credit for small businesses: customer relationships and credit cooperatives. Journal of Banking and Finance. 22: 925-954.

Athavale, M. and Edmister, R. (1999). Borrowing relationships, monitoring and the influence of loan rates. Journal of Finance Research. 22(3): 341-352.

Auria, C. Foglia, A. and Reedtz, P. (1999). Bank interest, rates and credit relationships in Italy. Journal of Banking and Finance. 23: 1067-1093.

Bebczuk, R. (2004). What determines the access to credit by SMEs in Argentina?. Documento de Trabajo No.48, Facultad de Ciencias Económicas, Universidad de la Plata. Spain.

Beck, T., Demirgüç-Kunt, A. and Maksimovic, V. (2004). Bank competition and access to finance: international evidence. Journal of Money, Credit and Banking. 36(3): 627-648.

Bédué, A. and Lévy, N. (1997). Relation banque-entreprise et cout du crédit. Revue d'Économie Financière.39(1) :179-200.

Belletante, B. and Levratto, N. (1995). Finance et PME: quels champs pour quels enjeux?’, Revue Internationale P.M.E.. 8 (3-4): 5-42.

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

122

Berger, A. and Udell, G. (1995). Lines of credit and relationship lending in small firm finance. Journal of Business. 68: 351-382.

Berger, A., Saunders, A., Scalise, J. and Udell, G. (1998) ‘The Effects of Bank Mergers and Acquisitions on Small Business Lending’, Journal of Financial Economics. 50: 187-229.

Blackwell, D. and Winters, D. (1997). Banking relationships and the effect of monitoring on loan pricing. Journal of Financial Research. 20(2): 275-289.

Blackwell, D. and Winters, D. (2000). Local Lending Markets: What a Small Business Owner/Manager Needs to Know. Quarterly Journal of Business and Economics, 39 ( 2): 62-79.

Bodenhorn, H. (2003). Short-term loans and long-term relationships: relationship lending in early America. Journal of Money, Credit and Banking. 35 (4): 485-505.

Bolton, P. and Scharfstein, D. (1996). Optimal debt structure and the number of creditors’, Journal of Political Economy. 104: 1-25.

Boot, A. and Thakor, A. (1994). Moral hazard and secured lending in an infinitely repeated credit market game. International Economic Review. 35(4): 899-920.

Boot, A. (2000). Relationship banking: what do we know. Journal of Financial Intermediation. 9: 7–25.

Boot, A. and Schmeits, A. (2005). The competitive challenge in banking. Amsterdam Center for Law & Economics. Working Paper 2005-08.

Boot, A. and Thakor, A. (2000). Can relationships banking survive competition?. Journal of Finance. 55 (2), pp. 679-713.

Brau, J. (2002). Do banks price owner-manager agency costs? An examination of small business borrowing. Journal of Small Business Management. 40 (4): 273-286.

Brick, I., Kane, E. and Palia, D. (2007). Evidence of jointness in the terms of relationship lending’, Journal of Financial Intermediation. 16 (3): 452-476.

Cardone, C., Casasola, M. and Samartín, M. (2005). Do banking relationships improve credit conditions for Spanish SMEs?. Working Paper 05-28, Universidad Carlos III de Madrid.

Charlier, P. (1998). L'Information comptable dans la relation de crédit bancaire. Revue Française de Géstion (121) : 108-116.

Cole, R. (1998). The importance of relationships to the availability of credit. Journal of Banking and Finance. 22, pp. 959-977.

Cowling, M. (1999). The incidence of loan collateralization in small business lending contracts: evidence from the UK. Applied Economics Letters. 6: 291– 293.

Cowling, M. and Westhead, P. (1996). Bank lending decisions and small firms: does size matter?’,

International Journal of Entrepreneurial Behaviour. 2(1): 52 - 68.

Dewatripont, M. and Maskin, E. (1995). Credit efficiency in centralized and decentralized economies’, Review of Economic Studies. 62(4): 541-555.

Degryse, H. and Ongena, S. (2001). Bank relationships and firm profitability. Financial Management. 30(1): 9-34.

Degryse, H. and Ongena, S. (2005). Distance, lending relationships and competition. Journal of Finance. 60 (1): 231-266.

Degryse, H. and Ongena, S. (2007). The impact of competition on bank orientation. Journal of Financial Intermediation. 16 (3): 399-424.

Degryse, H. and Van Cayseele, P. (2000). Relationship lending within a bank-based system: evidence from European small business data. Journal of Financial Intermediation. 9: 90-109.

Diamond, D. (1989). Reputation acquisition in debt markets. Journal of Political Economy. 97: 828-862.

Diamond, D. (1991). Monitoring and reputation: the choice between bank loans and privately placed debt. Journal of Political Economy. 99:689-721.

Elsas, R. (2005). Empirical determinants of relationship lending. Journal of Financial Intermediation. 14: 32–57.

Elsas, R. and Krahnen, J. (1998). Is relationship lending special? Evidence from credit-files data in Germany. Journal of Banking and Finance. 22: 1283-1316.

Elsas, R. and Krahnen, J. (1999). Collateral, default risk and relationship lending: an empirical study on financial contracting. Working Paper Nº 1999/13, Centre for Financial Studies, University of Frankfurt.

Ewert, R. and Schenk, G. (1998). Determinants of banking lending performance. CFS Working Paper 98/06, Centre for Financial Studies, University of Frankfurt.

Fernando, C., Chakraborty, A. and Mallic, R. (2002). The importance of being known: relationship banking and credit limits. WOPEC - Finance 0209007.

Forestieri, G. and Tirri, V. (2002). Verso un sistema bancario e finanziario Europeo? -Rapporto Banca-Impresa: Struttura del Mercato e Politiche di Prezzo. Ente per gli studi monetari, bancari e finanziari Luigi Einaudi. Quaderni di Ricerche Nº 31.

Fredriksson, A. (2007). The effect of relationship intensity and value on loan pricing. 20th Australasian Finance & Banking Conference 2007 Paper.

Freixas, X. (2005). Deconstructing relationship banking Investigaciones Económicas. 29(1): 3-31.

Granovetter, M. (1985). Economic action and social structure: the problem of embeddedness’, American Journal of Sociology. 91: 481-501.

Greenbaum, S., Kanatas, G. and Venezia, I. (1989). Equilibrium loan pricing under the bank-client

Am. J. Soc. Mgmt. Sci., 2010, 1(2): 102-123

123

relationship. Journal of Banking and Finance. 13: 221-235.

Groessl, I. and Levratto, N. (2004). Problems of evaluating small firms: quality as a reason for unfavourable loan conditions. Working Paper No. 0406014, series finance, Economics Working Paper Archive at WUSTL.

Guzman, M. (2000). Bank structure, capital accumulation and growth: a simple macroeconomic model. Economic Theory. 16: 421-455.

Harhoff, D. and Körting, T. (1998). Lending relationships in Germany - empirical evidence from survey data. Journal of Banking and Finance. 22: 1317-1353.

Hernández-Cánovas, G. and Martínez-Solano, P. (2006). Banking relationships: effects on debt terms for small Spanish firms. Journal of Small Business Management. 44(3): 315-333.

Johnson, S. (1998). The effect of bank debt on optimal capital structure’, Financial Management. 27(1): 47-56.

Kremp, E. and Sevestre, P. (2000). L’appartenance a un groupe facilite le financement des entreprises. Économie et Statistique .36( 6) : 79-96.

Le Guirriec, S. (1996) : La Représentativité des Comptes des PME-PMI. Banque. 573 : 28-30.

Levin, R. and Travis, V. (1987). Small company finance: what the books don't say. Harvard Business Review. 65 (6): 30-32.

Levratto, N. (2001). L’Évaluation des entreprises a fin de faciliter l’accès au crédit: quelle intermédiation informationnelle?. Rapport Final, DECAS Ministry of Economic Affairs, France.

Marquez, R. (2002). Competition, adverse selection, and information dispersion in the banking industry. Review of Financial Studies. 15 (3): 901-926.

Memmel, C., Schmieder, C. and Stein, I. (2007). Relationship lending – empirical evidence for Germany. Deutsche Bundesbank Discussion Paper Series: Banking and Financial Studies 14/2007.

Ongena, S., Tümer-Alkan, G. and von Westernhagen, N. (2007). Creditor concentration: an empirical investigation. Discussion Paper Series 2: Banking and Financial Studies 15/2007.

Peltoniemi, J. (2007). The benefits of relationship banking: evidence from small business financing in Finland. Journal of Financial Services Research. 31(2): 153-171.

Petersen, M. and Rajan, R. (1994). The benefits of lending relationships: evidence from small business data. Journal of Finance. 49(1): 3-37.

Petersen, M. and Rajan, R. (1995). The effect of credit market competition on lending relationships. Quarterly Journal of Economics. 110: 407-443.

Pezzetti, R. (2004). Bank-firm customer relations in the new competitive environment: prospects for relationship banking. Quaderno di Ricerca. 7, Faculty of Economics, University of Pavia.

Rajan, R. (1992). Insiders and outsiders: The choice between relationship and arm´s-length debt. Journal of Finance. 47: 1367-1400.

Repetto, A., Rodríguez, S. and Valdés, R. (2002). Bank lending and relationship banking: evidence from Chilean firms. Documento de Trabajo 146, Centro de Economía Aplicada, Universidad de Chile.

Riding, A., Haines, G. and Thomas, R. (1994). The Canadian small business-bank interface: a recursive mode. Entrepreneurship Theory & Practice. 1: 5-24.

Sharpe, S. A. (1990). Asymmetric information, bank lending, and implicit contracts: a stylized model of customer relationships. Journal of Finance. 45(4) 1069-1087.

Steijvers, T. and Voordeckers, W. (2002). The incidence of personal collateral in small business lending. Conference Proceedings, Research in Entrepreneurship and Small Business, 16th Workshop. 1: 573-592.

Stein, J. (2002). Information production and capital allocation: decentralized versus hierarchical firms. Journal of Finance. 57(5): 1891−1921.

Udell, G. (2008). What's in a relationship? The case of commercial lending. Business Horizons. 51: 93-103.

Uzzi, B. (1999). Embeddedness in the making of financial capital: how social relations and networks benefit firms seeking financing. American Sociological Review. 64 : 481–505.

Vigneron, L. (2007). Relations de clientèle bancaire et conditions de financements des PME: une étude empirique sur le portefeuille d’une grande banque Française. In: Sahut, J. (ed), Finance de Marché et Finance d'Entreprise: Complémentarités et Nouvelles Approches, Hermès Sciences.

von Thadden, E. L. (1995). Long term contracts, short-term investment monitoring. Review of Economic Studies. 62: 557-575.

Ziane, S. (2003). Number of banks and credit relationships: empirical results from French small business data. European Review of Economics and Finance. 2(3): 33-48.