Does Money Matter in Africa? New Empirics on Long- and Short-run Effects of Monetary Policy on...
40
Indian Growth and Development Review Does money matter in Africa?: New empirics on long- and short-run effects of monetary policy on output and prices Simplice A. Asongu Article information: To cite this document: Simplice A. Asongu , (2014),"Does money matter in Africa?", Indian Growth and Development Review, Vol. 7 Iss 2 pp. 142 - 180 Permanent link to this document: http://dx.doi.org/10.1108/IGDR-12-2012-0048 Downloaded on: 13 November 2014, At: 02:30 (PT) References: this document contains references to 69 other documents. To copy this document: [email protected] The fulltext of this document has been downloaded 21 times since 2014* Users who downloaded this article also downloaded: Philip Kostov, Thankom Arun, Samuel Annim, (2014),"Banking the unbanked: the Mzansi intervention in South Africa", Indian Growth and Development Review, Vol. 7 Iss 2 pp. 118-141 http://dx.doi.org/10.1108/ IGDR-11-2012-0046 Kenneth Backlund, Tomas Sjögren, Jesper Stage, (2014),"Optimal tax and expenditure policy in the presence of emigration: Are credit restrictions important?", Indian Growth and Development Review, Vol. 7 Iss 2 pp. 98-117 http://dx.doi.org/10.1108/IGDR-09-2012-0040 Michael Keen, (2014),"Targeting, cascading and indirect tax design", Indian Growth and Development Review, Vol. 7 Iss 2 pp. 181-201 http://dx.doi.org/10.1108/IGDR-02-2013-0009 Access to this document was granted through an Emerald subscription provided by Token:JournalAuthor:6F95467C-5349-48C0-AAF7-79BB16C495CA: For Authors If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service information about how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional customer resources and services. Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation. *Related content and download information correct at time of download. Downloaded by Mr Asongu Simplice At 02:30 13 November 2014 (PT)
Transcript of Does Money Matter in Africa? New Empirics on Long- and Short-run Effects of Monetary Policy on...
Indian Growth and Development ReviewDoes money matter in Africa?: New empirics on long- and short-run effects of monetarypolicy on output and pricesSimplice A. Asongu
Article information:To cite this document:Simplice A. Asongu , (2014),"Does money matter in Africa?", Indian Growth and Development Review, Vol.7 Iss 2 pp. 142 - 180Permanent link to this document:http://dx.doi.org/10.1108/IGDR-12-2012-0048
Downloaded on: 13 November 2014, At: 02:30 (PT)References: this document contains references to 69 other documents.To copy this document: [email protected] fulltext of this document has been downloaded 21 times since 2014*
Users who downloaded this article also downloaded:Philip Kostov, Thankom Arun, Samuel Annim, (2014),"Banking the unbanked: the Mzansi intervention inSouth Africa", Indian Growth and Development Review, Vol. 7 Iss 2 pp. 118-141 http://dx.doi.org/10.1108/IGDR-11-2012-0046Kenneth Backlund, Tomas Sjögren, Jesper Stage, (2014),"Optimal tax and expenditure policy in thepresence of emigration: Are credit restrictions important?", Indian Growth and Development Review, Vol. 7Iss 2 pp. 98-117 http://dx.doi.org/10.1108/IGDR-09-2012-0040Michael Keen, (2014),"Targeting, cascading and indirect tax design", Indian Growth and DevelopmentReview, Vol. 7 Iss 2 pp. 181-201 http://dx.doi.org/10.1108/IGDR-02-2013-0009
Access to this document was granted through an Emerald subscription provided byToken:JournalAuthor:6F95467C-5349-48C0-AAF7-79BB16C495CA:
For AuthorsIf you would like to write for this, or any other Emerald publication, then please use our Emerald forAuthors service information about how to choose which publication to write for and submission guidelinesare available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.comEmerald is a global publisher linking research and practice to the benefit of society. The companymanages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well asproviding an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committeeon Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archivepreservation.
*Related content and download information correct at time of download.
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Does money matter in Africa?New empirics on long- and short-run effects
of monetary policy on output and pricesSimplice A. Asongu
African Governance and Development Institute, Yaoundé, Cameroon
AbstractPurpose – The purpose of this paper is to examine the effects of monetary policy on economic activityusing a plethora of hitherto unemployed financial dynamics in inflation-chaotic African countries forthe period of 1987-2010.Although in developed economies, changes in monetary policy affect realeconomic activity in the short-run, but only prices in the long-run, the question of whether thesetendencies apply to developing countries remains open to debate.Design/methodology/approach – Vector autoregresion (VARs) within the frameworks of VectorError Correction Models and simple Granger causality models are used to estimate the long- andshort-run effects, respectively. A battery of robustness checks are also used to ensure consistency in thespecifications and results.Findings – The tested hypotheses are valid under monetary policy independence and dependence,except few exceptions. H1: Monetary policy variables affect prices in the long-run but not in theshort-run. For the first-half (long-run dimension) of the hypothesis, permanent changes in monetarypolicy variables (depth, efficiency, activity and size) affect permanent variations in prices in thelong-term. But in cases of disequilibriums, only financial dynamic fundamentals of depth and sizesignificantly adjust inflation to the cointegration relations. With respect to the second-half (short-runview) of the hypothesis, monetary policy does not overwhelmingly affect prices in the short-term. Hence,but for a thin exception, H1 is valid. H2: Monetary policy variables influence output in the short-termbut not in the long-term. With regard to the short-term dimension of the hypothesis, only financialdynamics of depth and size affect real gross domestic product output in the short-run. As concerns thelong-run dimension, the neutrality of monetary policy has been confirmed. Hence, the hypothesis is alsobroadly valid.Practical implications – A wide range of policy implications are discussed. Inter alia: the long-runneutrality of money and business cycles, credit expansions and inflationary tendencies, inflationtargeting and monetary policy independence implications. Country-/regional-specific implications, themanner in which the findings reconcile the ongoing debate, measures for fighting surplus liquidity andcaveats and future research directions are also discussed.Originality/value – By using a plethora of hitherto unemployed financial dynamics (that broadlyreflect monetary policy), we provide significant contributions to the empirics of money. The conclusionof the analysis is a valuable contribution to the scholarly and policy debate on how money matters as aninstrument of economic activity in developing countries.
Keywords Africa, Banking, Inflation, Monetary policy, VECM, Output effects
Paper type Research paper
1. IntroductionIn large industrial economies, changes in monetary policy affect real economic activityin the short-term but only prices in the long-run. In transitioning and developing
JEL classification – E51, E52, E58, E59, O55The author is highly indebted to the editor and referees for their useful comments.
The current issue and full text archive of this journal is available atwww.emeraldinsight.com/1753-8254.htm
IGDR7,2
142
Indian Growth and DevelopmentReviewVol. 7 No. 2, 2014pp. 142-180© Emerald Group Publishing Limited1753-8254DOI 10.1108/IGDR-12-2012-0048
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
countries, however, the question of whether monetary policy variables affect output inthe short-run is open to debate (Starr, 2005). Conversely, the evidence of real effects in adeveloped economy is supportive of the idea that monetary policy can be used to counteraggregate shocks. Economic theory traditionally suggests that money influencesbusiness cycles, not the long-term potential real output. This suggests monetary policyin neutral in the long-run. The neutrality of money has been substantially documentedin the literature (Olekalns, 1996; Serletis and Koustas, 1998; Bernanke and Mihov, 1998;Bullard, 1999; Bae et al., 2005; Nogueira, 2009). In spite of the theoretical and empiricalconsensus on this neutrality (Lucas, 1980; Gerlach and Svensson, 2003), the role ofmoney as an informational variable for decision-making has remained open to debate(Roffia and Zaghini, 2008; Nogueira, 2009; Bhaduri and Durai, 2012)[1].
The potential for using monetary policy in affecting real prices is also less clear. Incountries that have experienced high inflation or in which labor markets are chronicallyslack, prices and wages are unlikely to be particularly sticky, so that monetary policychanges could pass quickly through prices and have little real effect (Gagnon and Ihrig,2004). Moreover, the globalization of financial markets undercut the potential ofindependent policy by significantly eroding the ability of small-open economies todetermine interest rates independently of world markets (Dornbusch, 2001; Frankelet al., 2004).
In light of the above, three challenges are central in the literature. First, the extent towhich monetary policy affects output in the short-run, and prices in the long-run indeveloping countries remains an open debate. Hence, the need to contribute to thescholarly and policy debates on the manner in which money matters in economicactivity by providing an answer to the open question. Second, but for a few exceptions(Moosa, 1997; Bae and Ratti, 2000; Starr, 2005; Nogueira, 2009), the literature on thelong-run economic significance of money has focused on developed countries for themost part. Evidence provided by these studies may not be relevant for African countriesbecause their financial dynamics of monetary policy have different tendencies. Forexample, financial depth (liabilities) in the perspective of money supply is not asrelevant in African countries because a great chunk of the monetary base does nottransit through the banking sector (Asongu, 2014a). Third, the empirical investigationthat has focused on monetary aggregates has failed to take account of other proxies thatare exogenous to money supply. For instance, other financial dynamics of efficiency (atbanking and financial system levels), activity (from banking and financial systemperspectives) and size (credit of the banking sector in relation to that of the financialsystem) substantially affect the velocity of money and, hence, the effectiveness ofmonetary policy (both in the short- and long-run). Indeed, financial allocation efficiencyis a serious issue in African countries because of surplus liquidity issues (Saxegaard,2006) and, consequently, limited financial activity (credit).
The contribution of this paper to the literature is therefore fivefold. First, it assessesthe effects of monetary policy on output and prices in a continent (Africa) that has notreceived much scholarly attention. Second, there are certain specificities in monetarypolicy variables in developed economies that are not very relevant in Africa. Inter alia:on the one hand, financial depth in the perspective of money supply as applied in thedeveloped world cannot be transported to Africa because a great chunk of the monetarybase in the continent does not transit through the banking system; on the other hand, theeffectiveness of the credit channel of monetary policy is an issue in Africa owing to the
143
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
substantially documented surplus liquidity issues (Saxegaard, 2006; Fouda, 2009).Third, soaring food prices that have recently marked the geopolitical landscape ofAfrica have not been braced with adequate short-run monetary measures to stem therising price tide[2]. Fourth, embryonic African monetary unions like the proposed WestAfrican Monetary Zone (WAMZ) and East African Monetary Zone (EAMZ) need to beinformed on the relevance of money as an instrument of economic activity (growth andprice control). Fifth, the use of a plethora of hitherto unemployed monetary policyvariables complements existing literature by contributing to the empirics of monetarypolicy.
The rest of the paper is organized as follows. Section 2 highlights the debate anddiscusses monetary policy in Africa. The intuition motivating the empirics, the data andthe methodology are discussed in Section 3. Empirical analysis is covered in Section 4.Section 5 concludes.
2. The debate and African monetary policy2.1 The debateFor organizational purposes, we present the debate in two strands:
(1) the traditional discretionary monetary policy strand; and(2) the second strand of nontraditional policy regimes that limit the ability of
monetary authorities to use policy to offset output fluctuations.
In recent years, the rewards of shifting from traditional discretionary monetary policy toarrangements that favor commitments to price stability and international economicintegration (such as inflation targeting, monetary unions, dollarization, etc.) have beendiscussed substantially. An appealing prospect of discretionary policy is that themonetary authority can use policy instruments to offset adverse shocks to output bypursuing expansionary policy when output is below its potential and, contractionarypolicy when output is above its potential. For instance in the former situation, apolicy-controlled interest rate can be lowered in a bid to reduce commercial interest ratesand stimulate aggregate spending. Conversely, a monetary expansion that lowers thereal exchange rate may improve the competitiveness of a country’s products in domesticand world markets and thereby, boost demand for national output (Starr, 2005). As amatter of principle, a flexible countercyclical monetary policy can be practiced withinflation targeting (Ghironi and Rebucci, 2000; Mishkin, 2002; Cavoli and Rajan, 2008;Cristadoro and Veronese, 2011; Levine, 2012).
In the second strand, nontraditional policy regimes limit the ability of the monetaryauthorities to use policy to offset output fluctuations. The degree to which a givencountry can use monetary policy to affect output in the short-term is open to debate.Findings for the USA are consistent with the fact that, a decline in the key interest ratecontrolled by the Federal Reserve tends to boost output over the next two to three years.But the effect dissipates thereafter, so that the long-run effect is limited to prices (Starr,2005). Several studies have assessed whether the short-term effects of monetary policyon output in other countries are similar to those in the USA. Mixed results have beenfound in 17 industrialized countries (Hayo, 1999), and studies on two middle-incomecountries have found no evidence of Granger-causality from money to output,regardless of money used (Agénor et al., 2000). Hafer and Kutan (2002) find that interestrate generally plays a relatively more important role in explaining output in 20
IGDR7,2
144
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Organisation for Economic Co-operation and Development (OECD) countries, whileGanev et al. (2002) find no such evidence in Central and Eastern Europe. TheInternational Monetary Fund (IMF) places great emphasis on monetary policy in itsprograms for developing countries, especially in sub-Saharan Africa (SSA). It viewssuch a policy as crucial in managing inflation and stabilizing the real exchange rate.According to Weeks (2010), such an approach is absurdly inappropriate, as the vastmajority of governments in SSA lack the instruments to make monetary policyeffective[3].
2.2 Monetary policy in AfricaKhan (2011) has looked at the relationship between the growth of gross domesticproduct (GDP) and different monetary aggregates in 20 economies of SSA and foundempirical support for the hypothesis that credit-growth is more closely linked than inmoney-growth to the growth of real GDP. Mangani (2011) has investigated the effects ofmonetary policy on prices in Malawi and established the lack of unequivocal evidence insupport of the conventional channel of monetary policy transmission mechanism. Thefindings suggest that exchange rate was the most important variable in predictingprices. The study recommends that authorities should be more concerned with importedcost-push inflation than with demand-pull inflation[4]. In a slight contradiction,Ngalawa and Viegi (2011) have also investigated the process through which monetarypolicy affects economic activity in Malawi and found that, bank rate is a more effectivemeasure of monetary policy than reserve money.
Some studies have also focused on South Africa, with Gupta et al. (2010a) finding thathouse price inflation was negatively related to money policy shocks; Gupta et al. (2010b)showing that during the period of financial liberalization, interest rate shocks hadrelatively stronger effects on house price inflation irrespective of house sizes; and Ncubeand Ndou (2011) complementing Gupta et al. (2010a2010b)[5] with the suggestion that,the direct effects of high interest rates on consumption appear to be more important intransmitting monetary policy to the economy than through indirect effects. Hence, theinference that monetary policy tightening can marginally weaken inflationarypressures (arising from excessive consumption) operating via house wealth and thecredit channel. To demonstrate that monetary expansions and contractions may havedifferent effects in different regions of the same country, Fielding and Shields (2005)have estimated the size of asymmetries across the nine provinces of South Africa (overthe period 1997-2005) and found substantial differences in the response of prices tomonetary policy.
Consistent with the position of Weeks (2010) on the inherent ineffectiveness ofmonetary policy in African countries discussed above, the insights from the “Blindercredit-rationing model” are useful in solidifying the intuition for African empirics.According to Blinder (1987, p. 2), a rethinking new monetary policy dynamics is neededat times:
The reader should understand that this is merely an expositional device. I would not wish todeny that the interest elasticity and expectational error mechanisms have some validity. Butthe spirit of this paper is that those mechanisms do not seem important enough to explain thedeep recessions that are apparently caused by central bank policy.
145
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
The postulation of Blinder is even more relevant when existing monetary and exchangerate responses have not been effective in addressing the recent food inflation (VonBraun, 2008).
In light of the points presented in the introduction and the section above, thefollowing hypotheses will be tested in the empirical section:
H1. Monetary policy variables affect prices in the long-run but not in the short-run.
H2. Monetary policy variables influence output in the short-term but not in thelong-term.
3. Data and methodology3.1 Intuition for the empiricsAlthough there is vast empirical work on the effects of monetary policy on economicactivity based on aggregate measures of money supply, there is yet (to the best of ourknowledge) no usage of fundamental financial performance dynamics (that reflect thequantity of money supply) in the assessment of the long-run and short-term effects ofmonetary policy on output and prices. With this fact in mind, we are aware of the risksof “doing measurement without past empirical basis” and assert that reporting facts,even in the absence of past supporting studies (in the context of an outstandingtheoretical model), may be a useful scientific activity. Moreover, applied econometricshave other tasks than merely validating or refuting economic theories with existingexpositions and prior analytical frameworks (Asongu, 2014a, 2013e). Hence, the need tounderstand the economic/monetary intuition motivating the employment of a plethoraof financial performance measures in the assessment of the incidence of monetary policyvariables on economic activity.
From a broad standpoint, money supply can be understood in terms of the followingfour points:
(1) Financial intermediary depth could be defined both from an overall economicperspective and a financial system standpoint. This distinction, as will bedetailed in the data section, is worth emphasizing because unlike the developedworld, in developing countries, a great chunk of the monetary base does nottransit through the banking sector (Asongu, 2014b).
(2) Financial activity (observed from banking and financial system perspectives)reflects the ability of banks to grant credit to economic operators.
(3) Financial allocation efficiency (from banking and financial system standpoints)that reflects the fulfillment of the fundamental role of banks (in transformingmobilized deposits into credit for economic operators) could also intuitively beconceived as the ability of banks to increase the velocity of money.
(4) Financial size (deposit bank assets/total assets) mirrors the credit allocated bybanking institutions as a proportion of total assets in the financial system(deposit bank assets plus central bank assets).
Hence, financial intermediary performance dynamics are exogenous to money supplyand corrolarily monetary policy.
The choice of the monetary policy variables is broadly consistent with the empiricalunderpinnings of recent African monetary literature targeting inflation (Asongu, 2013a,
IGDR7,2
146
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
2013b) and real GDP output (Asongu, 2013c). These financial dynamic fundamentals entailall the dimensions identified by the Financial Development and Structure Database (FDSD)of the World Bank (WB). We are not the first to think out of the box when it comes to theempirics of monetary policy. Blinder (1987, p. 2) in assessing the effects of monetary policyon economic activity, completely banished interest rate elasticities:
In order to make credit rationing mechanism stand out in bold relief, most other channels ofmonetary policy (such as interest elasticities and expectational errors) are banished from the model.
3.2 DataWe investigate a panel of 10 African countries with data from African DevelopmentIndicators and the FDSD of the WB. The resulting balanced panel spans from 1987 to2010 due to constraints in data availability and the interest of obtaining results withupdated policy implications. In a bid for robustness, we are poised to control for theperiod of 2007-2009, during which the rise in food prices was very substantial. Hence, asub-panel of the period of 1987-2006 will be used to assess the consistency of findings.
We are limited to only 10 African cross-sections because some countries in the continentinherently do not exhibit a unit root in consumer price inflation. Owing to the problemstatements of the study, it is imperative to have non-stationary (chaotic) consumer priceinflation for consistent long-run modeling. Hence, in accordance with recent Africanlaw-finance literature (Asongu, 2011), African French Colonies (CFA) franc[6] countries ofthe Economic and Monetary Community of Central African States (CEMAC)[7] andEconomic and Monetary Community of West African States (UEMOA)[8] zones have notbeen included[9] in the sample. Beside these justifications for eliminating CFA franccountries, the seminal work of Mundell (1972) has shown that African countries with flexibleexchange rates regimes have more to experience in “money and inflation dynamics” thantheir counterparts with fixed exchange rate regimes[10].
Consistent with the literature, the dependent variables are measured in terms of annualpercentage change in the Consumer Price Index and real GDP output (Bordo and Jeanne,2002; Hendrix et al., 2009; Bae et al., 2005; Kishor and Ssozi, 2010; Moorthy and Kolhar, 2011).For organizational clarity, the independent variables are presented in terms of money(financial depth), credit (financial activity), efficiency and size. First, from a moneyperspective, we are consistent with the FDSD and recent African finance literature (Asongu,2013a, 2013b) in measuring financial depth both from overall-economic and financial systemperspectives with indicators of broad money supply (M2/GDP) and financial systemdeposits (Fdgdp), respectively. Whereas the former denotes the monetary base plus demand,saving and time deposits, the latter represents liquid liabilities of the financial system. It isinteresting to distinguish between these two aggregates of money supply because, as we aredealing exclusively with developing countries, a great chunk of the monetary base does nottransit via the banking sector. Second, credit is appreciated in terms of financialintermediary activity. Thus, the paper seeks to lay emphasis on the ability of banks to grantcredit to economic operators. We proxy for both banking-system-activity andfinancial-system-activity with “private domestic credit by deposit banks: Pcrb” and “privatecredit by deposit banks and other financial institutions: Pcrbof”, respectively. Third,financial size is measured in terms of deposit bank assets as a proportion of total assets(deposit bank assets plus central bank assets). Fourth, financial efficiency[11] measures theability of deposits (money) to be transformed into credit (financial activity). This fourth
147
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
measure appreciates the fundamental role of banks in transforming mobilized deposits intocredit for economic operators. We adopt indicators of banking-system-efficiency andfinancial-system-efficiency (respectively “bank credit on bank deposits: Bcbd” and “financialsystem credit on financial system deposits: Fcfd”).
Whereas definitions of the variables and their corresponding sources are presented inAppendix 2, summary statistics (with presentation of countries) and correlationanalysis are detailed in Appendix 1 and Appendix 3, respectively. From a preliminaryassessment of the summary statistics, we are confident that reasonable estimated nexuswould emerge. The correlation analysis provides empirical validity for the theoreticalcategorization of banking and financial system indicators into fundamentals of depth,efficiency, activity and size.
3.3 MethodologyThe estimation technique typically follows mainstream literature on testing the long-runneutrality of monetary policy (Nogueira, 2009) and the short-run effects of monetary policyvariables on output and prices (Starr, 2005). The approach involves unit root andcointegration tests that assess the stationary properties and long-term equilibriums,respectively. In these assessments, the Vector Error Correction Model (VECM) is applied forlong-run effects, while simple Granger causality is used for short-term effects. Whileapplication of the former model requires that variables exhibit unit roots in levels and havea long-run relationship (cointegration), the latter is applied on the condition that variables arestationary (do not exhibit unit roots). Impulse response functions are used to further assessthe tendencies of significant Granger causality results.
4. Empirical analysis4.1 Unit root testsWe investigate evidence of stationarity with first- and second-generation panel unit roottests. When the variables exhibit unit roots in levels, we proceed to test for stationarityin their first difference. Employment of the VECM requires that the variables have a unitroot (non stationary) in levels. Two main types of panel unit root tests have beendocumented: first-generation (that supposes cross-sectional independence) and thesecond-generation (based on cross-sectional dependence). Accordingly, it is convenientto perform several panel unit root tests to infer an overwhelming evidence to verify theorder of integration of a series, as none of these tests is immune from statisticalshortcomings in terms of size and power properties. With regard to the first-generationtests, the Levin et al. (2002) and Im et al.(2003) tests are applied. Whereas the former is ahomogenous-based panel unit root test (common unit roots as null hypothesis), the latteris a heterogeneous-oriented test (individual unit roots as null hypotheses). When theresults are different, Im et al.(2003) takes precedence over Levin et al. (2002) indecision-making because, consistent with Maddala and Wu (1999), the alternativehypothesis of Levin et al. (2002) is too powerful. While Im et al.(2003) controls forcross-sectional dependence to a certain degree, Pesaran (2007) has put forward a testwhich allows for the presence of more general cross-sectional dependence patterns. Thistest can be considered as a second-generation of panel unit root tests. In accordance withLiew (2004), goodness of fit (or optimal lag selection) is ensured by the Hannan-QuinnInformation Criterion (HQC) for the Levin et al. (2002) and the Akaike Information
IGDR7,2
148
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Criterion (AIC) for the Im et al.(2003) and Pesaran (2007) tests. Ultimately, Pesaran (2007)takes precedence over Levin et al. (2002) and Im et al.(2003) in decision-making.
Table I reports the panel unit root tests results. While Panel A presents the findingsfor the period of 1987-2010, Panel B is for the 1987-2006 span. Based on two informationcriteria, “constant and trend” and Pesaran (2007), only financial system efficiency inPanel A is stationary in levels. Hence, the findings indicate the possibility ofcointegration (long-run equilibrium) among the variables; because, in line with theEngel–Granger theorem, two variables that are not stationary may have a linearcombination in the long-run (Engle and Granger, 1987).
4.2 Cointegration testsConsistent with the cointegration theory, two (or more) variables that have a unit root inlevels may have a linear combination (equilibrium) in the long-run. In principle, if twovariables are cointegrated, it implies permanent movements of one variable affectingpermanent movements in the other variable. To assess the potential long-runrelationships, we test for cointegration using the Engle–Granger-based Pedroni test,which is a heterogeneous panel-based test. Whereas we have earlier used bothhomogenous and heterogeneous panel-based unit roots tests in Section 4.1, we disagreewith Camarero and Tamarit (2002) in applying a homogenous Engle–Granger-basedKao panel cointegration test because it has less deterministic components. Accordingly,application of Kao (1999), in comparison to Pedroni (1999), presents substantial issues indeterministic assumptions[12]. The same deterministic trend assumptions used in theIm et al.(2003) unit root tests are used in the Pedroni (1999) heterogeneous cointegrationtest. However, like in panel unit root tests, panel cointegration tests also suffer fromcross-sectional dependence. A second-generation cointegration test such as Westerlund(2007) can control for cross-sectional dependence via bootstrapping. The choice ofbivariate statistics has a twofold advantage (justification): on the one hand, it isconsistent with the problem statements and, on the other hand, it mitigatesmisspecification issues in causality estimations[13].
Tables II and III present the Pedroni and Westerlund cointegration results,respectively. Panel A (B) in both tables presents findings for the period 1987-2010(1987-2006). In Table II, based on the information criterion of “constant and trend”, thereis absence of cointegration between money supply (financial system activity) andinflation in Panel A (B). There is overwhelming (scanty) evidence of long-runequilibriums between monetary policy variables and inflation (output). These findingsare broadly consistent with the predictions of economic theory, which indicates thatmonetary policy has no incidence in real output in the long-run, but affects prices in thedistant future. Put in other words, the absence of a long-run relationship betweenmonetary policy variables and output shows the long-term neutrality of money. Itfollows that, permanent changes in financial intermediary dynamics (exogenous tomonetary policy) do not affect permanent changes in real GDP output in the long-run.Hence, the need to assess short-run effects by simple Granger causality (Section 4.4).Conversely, permanent movements in monetary policy variables influence prices in thelong-tem. Hence, the need to examine the short-term adjustments to the correspondingequilibriums with the VECM (Section 4.3).
The inferences above fail to consider the presence of common factors that affectmonetary policy across countries. Table III presents bivariate Westerlund cointegration
149
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table I.Panel unit root tests
Fina
ncia
ldep
th(M
oney
)Fi
nanc
iale
ffici
ency
Fina
nial
activ
ity(C
redi
t)Fi
nanc
ials
ize
Infla
tion
Out
put
M2
Fdgd
pB
cBd
FcFd
Pcrb
Pcrb
ofD
bacb
a(C
PI)
GD
P
Pane
lA:1
987-
2010
LLC
test
sof
hom
ogen
ous
pane
lLe
vel
c3.
231.
780.
24�
3.7
0**
*�
0.10
�1
.47
*2.
263.
242.
16ct
0.98
0.04
0.97
�5
.48
***
0.49
0.35
�1
.49
*1.
84�
1.5
8*
Firs
tdiff
eren
cec
�8
.39
***
�6
.28
***
�4
.64
***
NA
�4
.42
***
�4
.53
***
�1
0.9
***
�2.7
7**
*�
8.4
6**
*ct
�7
.04
***
�4
.41
***
�3
.42
***
NA
�2
.96
***
�3
.52
***
�1
0.9
***
�2.7
3**
*�
8.2
6**
*
IPS
test
sfo
rhe
tero
gene
ous
pane
lLe
vel
c2.
771.
860.
05�
2.5
0**
*1.
240.
063.
94�
0.76
5.02
ct0.
830.
130.
12�
3.2
9**
*0.
200.
38�
0.96
�1.8
6**
�1.3
6*
Firs
tdiff
eren
cec
�7
.30
***
�7
.50
***
�7
.15
***
NA
�4
.92
***
�5
.03
***
�1
2.1
***
�13.0
***
�8.0
3**
*ct
�6
.79
***
�7
.01
***
�6
.19
***
NA
�3
.64
***
�4
.14
***
�1
1.9
***
�8.7
0**
*�
7.6
6**
*
Pesa
ran
CA
DF
Z(t
-bar
)tes
tsta
tistic
sLe
vel
c�
0.15
�0.
711.
64�
0.01
0.26
�1
.52
*�
2.1
8**
2.18
�1.7
9**
ct0.
370.
24�
0.15
�1
.90
**1.
100.
360.
13�
0.83
0.12
7
Firs
tdiff
eren
cec
�3
.40
***
�2
.71
***
�5
.64
***
�3
.62
***
�1
.65
**�
2.4
8**
*�
4.0
2**
*�
7.2
3**
*�
5.2
2**
*ct
�2
.10
**�
4.6
1**
*�
3.6
3**
*�
3.3
0**
*�
2.2
8**
�2
.64
***
�1
.93
**�
5.9
5**
*�
3.5
7**
*
Pane
lB:1
987-
2006
LLC
test
sof
hom
ogen
ous
pane
lLe
vel
c�
0.80
�0.
98�
1.6
0*
�4
.69
***
�5
.24
***
�6
.37
***
0.28
1.06
1.57
ct0.
620.
53�
1.4
3*
�1
.72
**0.
680.
94�
1.28
0.77
�0.
95(c
ontin
ued)IGDR
7,2
150
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table I.
Fina
ncia
ldep
th(M
oney
)Fi
nanc
iale
ffici
ency
Fina
nial
activ
ity(C
redi
t)Fi
nanc
ials
ize
Infla
tion
Out
put
M2
Fdgd
pB
cBd
FcFd
Pcrb
Pcrb
ofD
bacb
a(C
PI)
GD
P
Firs
tdiff
eren
cec
�7
.27
***
�5
.69
***
NA
NA
�2
.43
***
�2
.37
***
�6
.43
***
�12.0
***
�5.6
4**
*ct
�6
.21
***
�5
.35
***
NA
NA
�4
.81
***
�5
.03
***
�6
.21
***
�8.2
4**
*�
7.5
1**
*
IPS
test
sfo
rhe
tero
gene
ous
pane
lLe
vel
c�
2.0
7**
�0.
98�
1.9
7**
�3
.71
***
�3
.39
***
�4
.33
***
1.00
�2.1
8**
4.50
ct�
0.40
�0.
82�
1.4
4*
�2
.12
**0.
231.
12�
0.60
�2.5
0**
*�
0.02
Firs
tdiff
eren
cec
�5
.45
***
�5
.68
***
NA
NA
�3
.01
***
�3
.05
***
�7
.27
***
NA
�4.7
9**
*ct
�5
.03
***
�4
.80
***
NA
NA
�4
.81
***
�5
.31
***
�7
.54
***
NA
�6.1
4**
*
Pesa
ran
CA
DF
Z(t
-bar
)tes
tsta
tistic
sLe
vel
c0.
36�
0.06
�0.
72�
1.29
0.00
�0.
77�
2.8
5**
*13
.32
0.21
ct0.
310.
65�
1.06
�0.
632.
041.
90�
0.56
�0.
111.
47
Firs
tdiff
eren
cec
�2
.62
***
�1
.88
**�
4.5
6**
*�
2.1
1**
�2
.79
***
�2
.29
**�
3.2
7**
*�
5.9
5**
*�
3.3
3**
*ct
�1
.50
*�
3.2
7**
*�
3.2
0**
*�
2.2
2**
�1
.77
**�
1.7
0**
�1
.70
**�
3.5
7**
*�
2.6
5**
*
Not
es:
***,
**,
*den
ote
sign
ifica
nce
at1,
5an
d10
%,r
espe
ctiv
ely.
‘c’an
d‘ct
’:‘co
nsta
nt’a
nd‘co
nsta
ntan
dtr
end’
resp
ectiv
ely;
max
imum
lag
is6
and
optim
alla
gsar
ech
osen
via
HQ
Cfo
rLLC
test
and
the
AIC
forI
PSIm
etal
.(20
03)a
ndPe
sara
n(2
007)
test
s;LL
C:Le
vin
etal
.(20
02).
IPS:
Imet
al.(
2003
);M
2:m
oney
supp
ly;
Fdgd
p:liq
uid
liabi
litie
s;B
cBd:
bank
ing
syst
emef
ficie
ncy;
FcFd
:fina
ncia
lsys
tem
effic
ienc
y;Pc
rb:b
anki
ngsy
stem
activ
ity;P
crbo
f:fin
anci
alsy
stem
activ
ity;
Dba
cba:
depo
sitb
ank
asse
tson
tota
lass
ets;
CPI:
cons
umer
pric
einfl
atio
n;G
DP:
gros
sdom
estic
prod
uct;
CAD
F:cr
oss-
augm
ente
ddi
ckey
fulle
r;si
gnifi
canc
eofb
old
data
isre
ject
ion
ofth
eun
itro
otnu
llhy
poth
esis
151
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table II.Bivariate heterogeneouspedroni engle-grangerbased panel cointegrationtests
Fina
ncia
ldep
th(M
oney
)and
infla
tion
Fina
ncia
lallo
catio
nef
ficie
ncy
and
infla
tion
Fina
ncia
lact
ivity
(Cre
dit)
and
infla
tion
Fina
ncia
lsiz
ean
din
flatio
nM
oney
supp
lyLi
quid
liabi
lity
Ban
king
syst
emFi
nanc
ials
yste
mB
anki
ngsy
stem
Fina
ncia
lsys
tem
cct
cct
cct
Cct
cct
cct
cct
Pane
lA:1
987-
2010
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dIn
flatio
nPa
nelv
-sta
ts�
0.50
20.
095
�0.
396
�1.
263
�0.
410
�1.
821
NA
NA
�0.
421
�1.
659
�0.
227
�1.
396
�0.
009
�1.
931
Pane
lrho
-sta
ts1.
497
1.86
0�
0.75
3�
1.03
3�
2.2
78
**�
2.0
72
**N
AN
A�
1.19
8�
1.2
87
*�
1.3
08
*�
1.6
49
**�
2.6
34
***
�2
.14
7**
Pane
lPP-
stat
s1.
883
1.56
3�
1.5
40
*�
2.8
10
***
�3.5
83
***
�4.9
79
***
NA
NA
�1.9
72
**�
3.4
50
***
�2.0
06
**�
3.9
86
***
�4
.37
0**
*�
5.4
81
***
Pane
lAD
F-st
ats
1.98
51.
673
�1.6
92
**�
2.5
42
***
�3.5
92
***
�5.8
72
***
NA
NA
�2.2
29
**�
3.6
87
***
�1.8
98
**�
4.4
65
***
�5
.07
2**
*�
6.8
80
***
Gro
uprh
o-st
ats
2.50
42.
549
�0.
393
0.14
2�
1.3
66
*�
0.44
0N
AN
A�
0.77
2�
0.06
8�
1.26
0�
0.80
2�
0.83
9�
0.23
4G
roup
PP-s
tats
2.65
72.
176
�2.3
49
***
�4.2
88
***
�4.1
13
***
�4.4
75
***
NA
NA
�2.7
18
***
�3.5
80
***
�3.0
75
***
�4.6
17
***
�4
.95
8**
*�
5.6
16
***
Gro
upA
DF-
stat
s2.
735
1.46
6�
2.3
87
***
�3.9
27
***
�2.3
81
***
�3.7
11
***
NA
NA
�2.8
40
***
�3.7
59
***
�2.7
25
***
�4.7
30
***
�3
.76
2**
*�
4.3
41
***
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dR
ealG
DP
Out
put
Pane
lv-s
tats
1.2
85
*1.
364
�0.
954
2.0
30
**�
1.31
82.
157
NA
NA
�1.
056
1.9
83
**�
1.19
41.9
34
**0.
242
2.0
26
**Pa
nelr
ho-s
tats
�0.
530
0.85
50.
987
0.85
01.
595
0.42
3N
AN
A1.
549
1.06
61.
957
1.03
8�
0.87
60.
565
Pane
lPP-
stat
s�
1.27
50.
028
0.79
2�
0.44
51.
829
�0.
954
NA
NA
1.82
8�
0.09
72.
482
�0.
121
�1
.35
0*
�0.
563
Pane
lAD
F-st
ats
�2.6
37
***
�0.
118
0.57
3�
1.27
50.
444
�1.5
68
*N
AN
A0.
479
�0.
845
1.40
2�
0.82
7�
0.69
0�
0.25
9G
roup
rho-
stat
s0.
840
1.89
51.
908
1.72
12.
242
1.68
0N
AN
A2.
211
2.15
52.
723
2.14
40.
113
1.25
8G
roup
PP-s
tats
�0.
501
0.88
51.
453
0.18
82.
907
0.25
7N
AN
A2.
462
0.85
43.
412
0.84
7�
1.8
77
**�
0.11
5G
roup
AD
F-st
ats
�2.2
13
**0.
176
0.67
8�
0.94
30.
962
�2.2
07
**N
AN
A�
0.18
1�
2.0
11
**0.
998
�1.9
84
**�
0.02
2�
1.07
6
Pane
lB:1
987-
2006
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dIn
flatio
nPa
nelv
-sta
ts�
0.98
3�
1.30
2�
0.80
6�
0.98
1�
0.73
9�
1.68
3�
1.04
6�
1.60
8�
0.37
5�
0.76
9�
0.57
4�
0.87
8�
0.45
7�
2.04
2Pa
nelr
ho-s
tats
�0.
114
�1.3
41
*�
0.04
6�
1.25
5�
1.4
68
*�
1.2
95
*�
0.70
1�
1.27
9�
0.85
4�
1.9
84
**0.
419
1.75
2�
2.4
15
***
�1
.73
8**
Pane
lPP-
stat
s�
1.02
7�
3.9
16
***
�0.
736
�3.4
87
***
�2.9
13
***
�4.2
55
***
�2.0
46
**�
3.4
52
***
�1.6
04
*�
4.6
33
***
�0.
170
1.09
0�
4.1
97
***
�5
.14
0**
*Pa
nelA
DF-
stat
s�
1.09
0�
3.9
32
***
�0.
774
�3.2
25
***
�3.1
99
***
�5.2
53
***
�2.3
79
***
�3.5
30
***
�1.3
18
*�
3.5
17
***
�0.
367
�0.
563
�4
.73
0**
*�
6.9
70
***
Gro
uprh
o-st
ats
0.71
10.
093
0.69
9�
0.02
3�
0.40
60.
269
0.20
50.
168
�0.
208
�0.
610
1.37
62.
743
�0.
418
�0.
027
Gro
upPP
-sta
ts�
1.5
86
*�
4.1
51
***
�1.7
24
**�
3.5
73
***
�3.0
24
***
�3.5
10
***
�2.1
93
**�
3.3
36
***
�2.0
04
**�
4.2
54
***
0.21
02.
126
�4
.60
7**
*�
5.1
46
***
Gro
upA
DF-
stat
s�
1.40
3�
3.8
91
***
�1.3
82
*�
3.3
96
***
�2.9
70
***
�3.6
65
***
�2.3
12
**�
2.7
86
***
�1.7
08
**�
3.1
13
***
�1.4
09
*�
0.58
0�
3.7
11
***
�4
.30
6**
*
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dR
ealG
DP
Out
put
Pane
lv-s
tats
�0.
788
3.89
3�
0.54
43.8
59
***
�0.
497
2.9
18
***
�0.
844
2.0
85
**�
0.49
52.5
38
***
�0.
349
2.1
78
**0.
643
1.81
8Pa
nelr
ho-s
tats
2.29
60.
847
1.94
00.
768
1.82
40.
763
2.03
51.
091
1.64
41.
140
2.05
91.
007
�0.
522
0.41
1Pa
nelP
P-st
ats
3.19
40.
641
2.65
10.
481
2.48
5�
0.06
32.
800
0.41
22.
076
0.55
52.
943
0.41
8�
0.64
2�
0.72
3Pa
nelA
DF-
stat
s3.
348
�1.
056
2.57
5�
1.00
53.
100
�0.
923
3.22
7�
1.4
11
*1.
735
�0.
589
2.87
3�
0.62
4�
0.61
0�
1.5
75
*G
roup
rho-
stat
s2.
929
1.71
42.
711
1.46
32.
553
1.82
12.
979
2.19
12.
554
2.28
92.
931
2.19
80.
626
1.34
7G
roup
PP-s
tats
4.21
80.
807
3.76
90.
248
3.61
0�
0.22
84.
285
1.10
32.
988
1.59
74.
054
1.39
7�
0.41
8�
0.16
6G
roup
AD
F-st
ats
3.55
5�
0.82
02.
496
�1.
205
4.14
5�
1.7
36
**3.
815
�0.
877
2.17
50.
006
3.48
8�
0.03
6�
1.12
9�
1.18
7
Not
es:
***
,**
,*
deno
tesi
gnifi
canc
eat
1,5
and
10%
resp
ectiv
ely;
‘c’a
nd‘c
t’:‘c
onst
ant’
and
‘con
stan
tand
tren
d’re
spec
tivel
y;Fi
n:Fi
nanc
ial.
PP:P
hilli
ps-P
eron
.AD
F:A
ugm
ente
dD
icke
yFu
ller;
na:n
otap
plic
able
beca
use
one
vari
able
inth
epa
iris
stat
iona
ryin
leve
ls;s
igni
fican
ceof
bold
data
isre
ject
ion
ofth
enu
llhy
poth
esis
ofno
coin
tegr
atio
n
IGDR7,2
152
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table III.Bivariate westerlund
panel cointegration tests
Fina
ncia
ldep
th(M
oney
)and
infla
tion
Fina
ncia
lallo
catio
nef
ficie
ncy
and
infla
tion
Fina
ncia
lact
ivity
(cre
dit)
and
infla
tion
Fina
ncia
lsiz
ean
din
flatio
nM
oney
supp
lyLi
quid
liabi
lity
Ban
king
syst
emFi
nanc
ials
yste
mB
anki
ngsy
stem
Fina
ncia
lsys
tem
Cct
cCt
cct
cct
cct
cct
cct
Pane
lA:1
987-
2010
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dIn
flatio
nG
t(Z-
valu
e)4.
079
2.84
84.
703
3.46
30.
011
�3.2
03
***
NA
NA
2.77
72.
422
2.03
51.
701
3.47
62.
509
Ga
(Z-v
alue
)2.
491
2.95
22.
860
3.18
91.
492
0.0
05
**N
AN
A3.
246
3.95
22.
811
3.46
92.
714
1.98
3Pt
(Z-v
alue
)2.
602
1.70
93.
194
2.58
6�
0.64
2�
3.8
28
***
NA
NA
�1.6
83
*�
0.64
3�
2.7
93
**�
0.94
13.
237
3.09
0Pa
(Z-v
alue
)1.
594
1.07
92.
031
1.42
1�
0.39
8�
1.7
60
**N
AN
A�
0.74
10.
927
�1.4
42
*0.
820
1.99
12.
197
BP
CDT
est(
Chi2 )
272.4
55
***
321.3
10
***
205.6
05
***
200.7
05
***
259.7
41
***
250.1
88
***
28
3.7
60
***
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dR
ealG
DP
Out
put
Gt(
Z-va
lue)
1.83
52.
983
2.58
23.
988
0.07
52.
412
NA
NA
1.57
96.
313
0.78
95.
377
�0.
064
2.43
6G
a(Z
-val
ue)
1.74
82.
861
1.63
13.
166
0.63
23.
148
NA
NA
0.78
04.
064
0.1
88
**3.
736
�0
.01
1*
2.73
9Pt
(Z-v
alue
)�
3.3
62
**0.
802
�3.5
95
**1.
061
0.47
43.
200
NA
NA
�3.5
65
**2.
554
�3.2
21
***
2.31
40.
614
3.37
7Pa
(Z-v
alue
)�
1.9
02
*1.
112
�3.1
51
**0.
864
0.46
92.
668
NA
NA
�4.2
79
**1.
056
�4.1
52
***
1.10
7�
1.25
82.
160
BP
CDT
est(
Chi2 )
203.3
72
***
210.3
10
***
205.1
28
***
181.0
99
***
160.2
62
***
174.2
66
***
16
4.5
90
***
Pane
lB:1
987-
2006
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dIn
flatio
nG
t(Z-
valu
e)1.
352
0.58
31.
339
0.35
8�
2.0
07
**�
1.5
55
*�
5.4
94
***
�3.8
59
***
�3.9
09
***
�2.0
32
**�
4.4
18
***
�2
.35
5*
2.27
60.
236
Ga
(Z-v
alue
)2.
053
2.77
22.
051
2.53
91.
081
2.28
70.4
87
*2.
076
0.95
93.
023
0.74
43.
131
2.65
93.
182
Pt(Z
-val
ue)
�0.
077
�0.
002
�0.
276
�0.
324
�3.9
30
**�
3.1
45
*�
6.3
17
***
�5.2
69
**�
4.5
60
**0.
083
�3.6
68
**�
0.00
32.
761
4.95
3Pa
(Z-v
alue
)�
0.67
10.
889
�0.
485
0.77
1�
0.67
11.
476
�1.5
30
**1.
106
�1.5
03
*1.
587
�1.4
94
**1.
535
1.55
02.
988
BP
CDT
est(
Chi2 )
220.3
03
***
251.1
17
***
170.6
52
***
174.7
23
***
184.0
55
***
206.7
27
***
19
2.4
62
***
Coi
nteg
ratio
nbe
twee
nM
onet
ary
Polic
yan
dR
ealG
DP
Out
put
Gt(
Z-va
lue)
3.09
81.
147
3.55
41.
222
�2.2
84
**0.
047
�4.7
96
***
�2.3
29
**�
1.4
10
*3.
186
�2.3
48
**2.
385
�1
.98
4**
�0.
266
Ga
(Z-v
alue
)3.
365
3.29
03.
263
3.29
30.2
48
*2.
484
0.5
15
*2.
164
1.61
34.
196
0.99
63.
869
0.9
78
*3.
009
Pt(Z
-val
ue)
�0.
728
1.33
8�
0.88
01.
962
�1.
549
�0.
467
�4.8
50
**�
2.17
9�
2.9
66
*0.
694
�1.
164
1.32
4�
1.9
10
*1.
898
Pa(Z
-val
ue)
0.56
92.
090
0.38
72.
254
�0.
282
1.53
0�
0.65
11.
194
0.03
12.
159
�0.
656
1.82
5�
0.61
61.
900
BP
CDT
est(
chi-s
quar
e)222.7
81
***
242.6
43
***
182.5
64
***
176.6
32
***
205.2
88
***
231.8
35
***
13
1.7
66
***
Not
es:
***
,**
,*
deno
tesi
gnifi
canc
eat
1,5
and
10%
,res
pect
ivel
y;‘c
’and
‘ct’:
‘con
stan
t’an
d‘c
onst
anta
ndtr
end’
,res
pect
ivel
y;Fi
n:Fi
nanc
ial;
the
lag
and
lead
leng
ths
are
sett
oon
e;ch
oosi
ngto
om
any
lags
and
lead
sca
nre
sult
ina
dete
rior
atio
nof
the
smal
l-sam
ple
prop
ertie
sof
the
test
;to
cont
rolf
orcr
oss-
sect
iona
ldep
ende
nce,
robu
stcr
itica
lval
ues
isob
tain
edth
roug
h30
0bo
otst
rap
repl
icat
ions
;BP
CDT
est:
Bre
usch
–Pag
anLM
test
ofin
depe
nden
ce;s
igni
fican
ceof
bold
data
isre
ject
ion
ofth
enu
llhy
poth
esis
ofno
coin
tegr
atio
n
153
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
tests that control for cross-sectional dependence[14]. Based on the information criterionof “constant and trend”, while the long-run neutrality of money is still broadlyconfirmed, evidence of cointegration between monetary variables and inflation isscanty.
Modeling with the VECM and simple Granger causality (unrestricted vectorautoregresion (VAR)) will be based on the hypotheses of cross-sectional dependence andindependence in cointegration for the following reasons. First, it eases comparison of thepresent study with prior 2007 Pedroni-based literature. Accordingly, studies before theemergence of the Westerlund (2007) test still constitute useful scientific activity. Second,both assumptions are necessary to fully examine the postulated hypotheses underinvestigation. Third, new comparative insights into monetary policy independence anddependence could emerge. For instance, if no significant differences in results emerge, itcould be established that monetary policy dependence does not really matter. Fourth, asan intuition for independence, African monetary union countries (with the likelihood ofcommon monetary policies) have not been considered in the dataset. Fifth, theassumption of independence matters only for the money–inflation nexus and not for themoney– output linkage because Westerlund (2007) is consistent with Pedroni (1999) onthe long-run neutrality of money. Hence, one of the underlying hypotheses motivatingthe study remains sound. Sixth, the label of “empirics” on the paper’s title means we canmake the assumption without being afraid of “econometrics polices”.
4.3 VECM for monetary policy and inflationLet us consider inflation and money with no lagged differences, such that:
Inflationi,t � �Moneyi,t (1)
The resulting VECMs are the following for equation (1):
�Inflationi,t � �(Inflationi,t�1 � �Moneyi,t�1) � �i,t (2)
�Moneyi,t � (Moneyi,t�1 � kInflationi,t�1) � ei,t (3)
In equations (2) and (3), the right-hand terms are the Error Correction Terms (ECTs). Atequilibrium, the value of the ECT is zero. When the ETC is non-zero, it means thatinflation and money have deviated from the long-run equilibrium; and the ECT helpseach variable to adjust and partially restore the equilibrium relationship. The speed ofthese adjustments are measured by � and for inflation and money, respectively.Therefore, equations (2) and (3) are replicated for all the “finance and inflation” pairs.The same deterministic trend assumptions used in the cointegration tests are used.
Table IV shows results for the VECM. While the findings in the upper-sections ofboth panels are consistent with equation (1), those in the lower-sections correspond toequations (2) and (3). Specifically, the first line in the lower-sections (D[Inflation])corresponds to equation (2), whereas the estimates shaping the diagonal are consistentwith equation (3). The signs of the cointegration relations in the upper-sections of thepanels are in accordance with the predictions of economic theory. This confirms theexisting consensus that in the long-run, money has a positive relationship with inflation.Among the long-run relationships of monetary policy variables, that of financial size isthe most significant. In other words, the ratio of bank assets in proportion of total assets
IGDR7,2
154
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table IV.Vector error correction
model (cointegration andshort-term adjustment
coefficients)
Pane
lA:1
987-
2010
Est
imat
esof
coin
tegr
atio
nre
latio
nshi
psFi
nanc
iald
epth
(Mon
ey)
Mon
eysu
pply
NA
––
––
––
Liqu
idlia
bilit
ies
–2.
854
––
––
–(0
.126
)
Fina
ncia
lallo
catio
nef
ficie
ncy
Ban
king
syst
em–
–4.
259
––
––
(0.2
45)°
Fina
ncia
lsys
tem
––
–N
A–
––
Fina
ncia
lact
ivity
(Cre
dit)
Ban
king
syst
em–
––
–13
.750
––
(0.5
00)
Fina
ncia
lsys
tem
––
––
–15
.716
–(0
.627
)
Fina
ncia
lsiz
eB
anki
ngsy
stem
––
––
––
29.5
42
**(2
.169)
Est
imat
esof
shor
t-ter
mad
just
men
tcoe
ffici
ents
D[In
flatio
n]N
A�
0.2
04
***
�0
.19
0**
*N
A�
0.2
00
***
�0
.20
2**
*�
0.2
53
***
(�5
.18
2)
(�5
.04
4)°
(�5
.13
6)
(�5
.139)
(�5.4
69)
Fina
ncia
ldep
th(m
oney
)D
[Mon
eysu
pply
]N
A–
––
––
–D
[Liq
uid
liabi
litie
s]–
�0
.00
01
**–
––
––
(�2
.55
2)
Fina
ncia
lallo
catio
nef
ficie
ncy
D[B
anki
ngsy
stem
]–
–�
0.00
03–
––
–(�
1.05
4)°
D[F
inan
cial
syst
em]
––
–N
A–
––
(con
tinue
d)
155
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table IV.
Fina
ncia
lact
ivity
(cre
dit)
D[B
anki
ngsy
stem
]–
––
–�
0.00
0088
7–
–(�
1.29
2)D
[Fin
anci
alsy
stem
]–
––
––
�0.
0000
960
–(�
1.34
2)
Fina
ncia
lsiz
eD
[Ban
king
syst
em]
––
––
––
�0.0
004
**(�
2.0
95)
Pane
lA:1
987-
2006
Est
imat
esof
coin
tegr
atio
nre
latio
nshi
psFi
nanc
iald
epth
(Mon
ey)
Mon
eysu
pply
9.36
5–
––
––
–(0
.404
)Li
quid
liabi
litie
s–
10.4
45–
––
––
(0.3
70)
Fina
ncia
lallo
catio
nE
ffici
ency
Ban
king
syst
em–
–8.
639
––
––
(0.3
98)°
Fina
ncia
lsys
tem
––
–7.
892
––
–(0
.398
)°Fi
nanc
iala
ctiv
ity(C
redi
t)B
anki
ngsy
stem
––
––
21.6
62–
–(0
.618
)Fi
nanc
ials
yste
m–
––
––
NA
–
Fina
ncia
lsiz
eB
anki
ngsy
stem
––
––
––
58.0
27
***
(4.7
42)
(con
tinue
d)IGDR7,2
156
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table IV.
Est
imat
esof
shor
tter
mad
just
men
tcoe
ffici
ents
D[In
flatio
n]�
0.2
03
***
�0
.20
7**
*�
0.1
89
***
�0
.19
0**
*�
0.2
02
***
NA
�0.2
86
***
(�4
.44
8)
(�4
.55
2)
(�4
.39
6)°
(�4
.40
3)°
(�4
.50
9)
(�4.8
40)
Fina
ncia
ldep
th(M
oney
)D
[Mon
eysu
pply
]�
0.0
00
2**
*–
––
––
–(�
3.0
59
)D
[Liq
uid
liabi
litie
s]–
�0
.00
01
**–
––
––
(�2
.45
5)
Fina
ncia
lallo
catio
nef
ficie
ncy
D[B
anki
ngsy
stem
]–
–�
0.00
02–
––
–(�
0.67
9)°
D[F
inan
cial
syst
em]
––
–�
0.00
01–
––
(�0.
582)
°
Fina
ncia
lact
ivity
(Cre
dit)
D[B
anki
ngsy
stem
]–
––
–�
0.00
0076
8–
–(�
1.22
5)D
[Fin
anci
alsy
stem
]–
––
––
NA
–
Fina
ncia
lsiz
eD
[Ban
king
syst
em]
––
––
––
�0.0
007
***
(�3.5
22)
Not
es:
***,
**,
*den
ote
sign
ifica
nce
at1,
5an
d10
%,r
espe
ctiv
ely;
the
dete
rmin
istic
tren
das
sum
ptio
nsan
dla
gse
lect
ion
crite
ria
for
the
VE
CMar
eth
esa
me
asin
the
coin
tegr
atio
nte
sts;
():t
-sta
tistic
s;D
[]:F
irst
diff
eren
ce.°
:val
ues
also
appl
yfo
rth
em
onet
ary
polic
yde
pend
ence
hypo
thes
is.n
a:ab
senc
eof
coin
tegr
atio
nor
stat
iona
ryin
leve
ls;s
igni
fican
ceof
bold
data
isre
ject
ion
ofth
enu
llhy
poth
esis
ofno
coin
tegr
atio
nan
dre
ject
ion
ofth
enu
llhy
poth
esis
nosi
gnifi
cant
erro
rco
rrec
tion
term
s
157
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
(Deposit bank assets plus Central bank assets) has the most significant positiverelationship with inflation in the long-run.
The lower-sections of the panels of Table IV show feedback coefficients for thecointegrating vectors or the short-run adjustments of inflation and the monetary policyvariables. Some adjustments are significantly different from zero, implying that thesemonetary policy variables are not weakly exogenous with regard to the parameters ofthe cointegration relationships in the upper-sections. In case of any deviation from thelong-run equilibriums, these variables respond and adjust the system back to theequilibrium relationships. Only the monetary policy variables of financial depth andfinancial size are particularly significant in adjusting inflation to the equilibrium.Monetary policy fundamentals of credit and ability of banks to transform money intocredit are not significant in adjusting inflation to the equilibrium. Therefore, in event ofdisequilibriums in the long-run, short-term adjustments in the ability of banks totransform money into credit do not matter in significantly correcting inflation. Apossible and logical explanation for this outcome is the substantially documentedsurplus liquidity issues in African financial institutions (Saxegaard, 2006; Fouda, 2009).This is robustly confirmed by the insignificance of the credit-adjusting estimates offinancial activity. Hence, allocation inefficiency and correspondingly, limited financialactivity (credit) partially explain these insignificant contributions of credit andallocation efficiency in error correction. The ECTs have the expected signs and are in theright interval for a stable error correction mechanism (see the last point on robustnesschecks in Section 4.6 for discussion). It should be noted that the discussion above is bothrelevant for the hypotheses of independence and dependence in monetary policy. Thedegree sign (°) in Table IV has been used to emphasize values that represent both.
Consistent with the Engle–Granger theorem, we examine short-run effects of thenexus under investigation if the pairs are either not cointegrated or a variable isstationary in levels.
4.4 Granger causality for monetary policy and economic activityLet us consider the following basic bivariate finite-order VAR models:
Inflationi,t � �j�1
p
ijInflationi,t�j � �j�0
q
�ijMoneyi,t�j � �i � �i,t (4)
Outputi,t � �j�1
p
hijOutputi,t�j � �j�0
q
kijMoneyi,t�j � vi � ei,t (5)
Simple Granger causality is based on the assessment of how past values of a monetarypolicy variable could help past values of inflation in explaining the present value ofinflation (equation 4). In the same line of thought, it also implies investigating how pastvalues of monetary policy variables are significant in helping the past values of outputto explain the present value of output (equation 5). In mainstream literature, this modelis applied on variables that are stationary.
In light of the above, the resulting VAR models in first difference (for pairs that havefailed the cointegration test) are the following:
IGDR7,2
158
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
�Inflationi,t � �j�1
p
ij�Inflationi,t�j � �j�0
q
�ij�Moneyi,t�j � �i � �i,t (6)
�Outputi,t � �j�1
p
hij�Outputi,t�j � �j�0
q
kij�Moneyi,t�j � vi � ei,t (7)
The null hypothesis of equation (4) is the position that, Money does not, Granger causesInflation. Therefore, a rejection of the null hypothesis is captured by the significantF-statistics, which is the Wald statistics for the joint hypothesis that estimatedparameters of lagged values equal zero. Optimal lag selection for goodness of fit is inaccordance with the AIC (Liew, 2004).
Table V presents the Granger causality results. Within the framework of this study,we are applying the model only to pairs that have failed the cointegration test (na).Accordingly, the “not cointegrated” (nc) pairs may either be stationary in levels (sl):nc(sl) or in first difference (sfd): nc(sfd). While Panel A shows results based on thehypothesis of monetary policy independence, Panel B depicts findings grounded on thehypothesis of monetary policy dependence. From horizontal and vertical comparativestandpoints, two main findings can be established. First, for both samples and eitherhypothesis, monetary policy does not affect prices in the short-run. Second, consistentacross hypothetical specifications for the period of 1987-2006, financial depth andfinancial size affect real output in the short-run. The overwhelming absence ofsignificant causalities flowing from monetary policy to inflation in the short-term isconsistent with economic theory and in line with expectations. The simple fact is thatmonetary policy dynamics (of depth and size) Granger cause real GDP output is notenough to draw any economic inferences. Hence, impulse–response functions (IRFs) ofthe nexus will provide additional material on the scale and timing of output responses tomonetary policy innovations.
4.5 Impulse responsesUsing Choleski decomposition on VAR with ordering: output and a monetary policyvariable, we compute IRFs for output and financial fundamentals of depth and size.Appendix 4 shows Figures A1-A3 for the VAR ordering: output and monetary policy,while Appendix 5 shows Figures A4-A6 for the VAR ordering: monetary policy, output.In the graphical representation of the IRFs, the dotted lines are the two standarddeviation bands which are used to measure the significance (Agénor et al., 1997, p. 19).By virtue of the causality analysis, whereas we are more concerned with the first VARordering, the second VAR ordering has been presented simply for robustness purposes.
Based on the first VAR ordering, from intuition, we expect a positive shock(expansionary policy) in monetary variables to positively affect real GDP output in theshort-term and a negative shock (contractionary policy) to negatively affect real GDPoutput in the short-run. Accordingly, the innovations and responses in Figures A1-A3are consistent with our intuition and economic theory. First, from Figure A1 a onestandard deviation positive shock (innovation) in money supply sharply increasesoutput in the first period before the effect steadily decreases during the next three yearsand finally disappears in the fourth year. Second, the dynamic responses of Figure A2broadly confirm those of Figure A1. A one standard deviation innovation in liquid
159
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table V.Short-run simple Grangercausality analysis
Fina
ncia
ldep
th(M
oney
)Fi
nanc
iale
ffici
ency
Fina
ncia
lact
ivity
(Cre
dit)
Fina
ncia
lsiz
eM
2Fd
gdp
BcB
dFc
FdPc
rbPc
rbof
Dba
cba
Pane
lA:H
ypot
hesi
sof
mon
etar
ypo
licy
inde
pend
ence
Pane
lA1:
1987
-201
0M
onet
ary
polic
ydo
esno
tcau
seR
ealG
DP
Out
put
Leve
lsnc
(sfd
)nc
(sfd
)nc
(sfd
)0.
231
nc(s
fd)
nc(s
fd)
nc(s
fd)
D[M
2]D
[Fdg
dp]
D[B
cBd]
D[F
cFd]
D[P
crb]
D[P
crbo
f]D
[Dba
cba]
1std
iffer
ence
1.74
11.
812
0.01
98nc
(sl)
0.67
10.
511
1.46
3
Mon
etar
ypo
licy
does
notc
ause
Infla
tion
Leve
lsnc
(sfd
)N
AN
A1.
468
NA
NA
NA
D[M
2]D
[Fdg
dp]
D[B
cBd]
D[F
cFd]
D[P
crb]
D[P
crbo
f]D
[Dba
cba]
1std
iffer
ence
0.45
5N
AN
Anc
(sl)
NA
NA
NA
Pane
lA2:
1987
-200
6M
onet
ary
polic
ydo
esno
tcau
seR
ealG
DP
Out
put
Leve
lsnc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)D
[M2]
D[F
dgdp
]D
[BcB
d]D
[FcF
d]D
[Pcr
b]D
[Pcr
bof]
D[D
bacb
a]1s
tdiff
eren
ce2
.51
3*
2.6
95
*0.
028
0.00
51.
700
1.22
42.6
47
*
Mon
etar
ypo
licy
does
notc
ause
Infla
tion
Leve
lsN
AN
AN
AN
AN
Anc
(sfd
)N
AD
[M2]
D[F
dgdp
]D
[BcB
d]D
[FcF
d]D
[Pcr
b]D
[Pcr
bof]
D[D
bacb
a]1s
tdiff
eren
ceN
AN
AN
AN
AN
A0.
439
NA
Pane
lB:H
ypot
hesi
sof
mon
etar
ypo
licy
depe
nden
cePa
nelB
1:19
87-2
010
Mon
etar
ypo
licy
does
notc
ause
Rea
lGD
PO
utpu
tLe
vels
nc(s
fd)
nc(s
fd)
nc(s
fd)
0.23
1nc
(sfd
)nc
(sfd
)nc
(sfd
)D
[M2]
D[F
dgdp
]D
[BcB
d]D
[FcF
d]D
[Pcr
b]D
[Pcr
bof]
D[D
bacb
a]1s
tdiff
eren
ce1.
741
1.81
20.
019
nc(s
l)0.
671
0.51
11.
463
(con
tinue
d)IGDR7,2
160
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Table V.
Fina
ncia
ldep
th(M
oney
)Fi
nanc
iale
ffici
ency
Fina
ncia
lact
ivity
(Cre
dit)
Fina
ncia
lsiz
eM
2Fd
gdp
BcB
dFc
FdPc
rbPc
rbof
Dba
cba
Mon
etar
ypo
licy
does
notc
ause
Infla
tion
Leve
lsnc
(sfd
)nc
(sfd
)N
A1.
468
nc(s
fd)
nc(s
fd)
nc(s
fd)
D[M
2]D
[Fdg
dp]
D[B
cBd]
D[F
cFd]
D[P
crb]
D[P
crbo
f]D
[Dba
cba]
1std
iffer
ence
0.45
51.
153
NA
nc(s
l)0.
393
0.40
30.
118
Pane
lB2:
1987
-200
6M
onet
ary
polic
ydo
esno
tcau
seR
ealG
DP
Out
put
Leve
lsnc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)nc
(sfd
)D
[M2]
D[F
dgdp
]D
[BcB
d]D
[FcF
d]D
[Pcr
b]D
[Pcr
bof]
D[D
bacb
a]1s
tdiff
eren
ce2
.51
3*
2.6
95
*0.
028
0.00
51.
700
1.22
42.6
47
*
Mon
etar
ypo
licy
does
notc
ause
infla
tion
Leve
lsnc
(sfd
)nc
(sfd
)N
AN
Anc
(sfd
)nc
(sfd
)nc
(sfd
)D
[M2]
D[F
dgdp
]D
[BcB
d]D
[FcF
d]D
[Pcr
b]D
[Pcr
bof]
D[D
bacb
a]1st
Diff
eren
ce0.
387
0.42
0N
AN
A0.
424
0.43
90.
260
Not
es:
Nul
lHyp
othe
ses
ofPa
nelA
1:M
onet
ary
Polic
ydo
esno
tGra
nger
caus
ere
alG
DP
outp
utan
dM
onet
ary
Polic
ydo
esno
tGra
nger
caus
ein
flatio
n;M
2:m
oney
supp
ly;F
dgdp
:liq
uid
liabi
litie
s;B
cBd:
bank
cred
iton
bank
depo
sit(
Ban
king
Syst
emIn
term
edia
ryE
ffici
ency
);Fc
Fd:fi
nanc
ialc
redi
ton
finan
cial
depo
sits
(Fin
anci
alSy
stem
Inte
rmed
iary
);Pc
rb:p
riva
tedo
mes
ticcr
edit
from
depo
sit
bank
s(B
anki
ngSy
stem
Inte
rmed
iary
Act
ivity
);Pc
rbof
:pri
vate
dom
estic
cred
itfr
omde
posi
tban
ksan
dot
her
finan
cial
inst
itutio
ns(F
inan
cial
Syst
emIn
term
edia
ryA
ctiv
ity);
Dba
cba:
depo
sitb
ank
asse
ton
tota
lass
ets
(Ban
king
Syst
emSi
ze);
Fin:
finan
cial
.***
,**
,*d
enot
esi
gnifi
canc
eat
1,5
and
10%
,res
pect
ivel
y;na
:not
appl
icab
lebe
caus
eof
coin
tegr
atio
n;nc
(sl):
noco
inte
grat
ion
and
stat
iona
ryin
leve
ls;n
c(sf
d):n
oco
inte
grat
ion
and
stat
iona
ryin
first
diff
eren
ce;s
igni
fican
ceof
bold
data
isre
ject
ion
ofth
enu
llhy
poth
esis
ofno
Gra
nger
caus
ality
161
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
liabilities increases real GDP output sharply during the first period, the effect levels-upduring the second year before steadily decreasing during the next three years anddisappearing in the fifth year. Third, from Figure A3, a positive shock in financial sizesharply increases output over the next two years, after which the effect also sharplydrops for a year before falling steadily during the next two years and disappearing in thefifth year.
From the second VAR ordering in Appendix 5, the positive nexus among theinnovations of output in money supply (Figure A4), financial deposits (Figure A5) andfinancial size (Figure A6) are consistent with the predications of economic theory.Accordingly, a one standard deviation positive innovation (shock) in output increasesfinancial dynamic fundamentals of depth (size) during the first (first-two) year(s). Theeffects disappear in the third year, fourth year and fifth year for money supply (FigureA4), liquid liabilities (Figure A5) and financial size (Figure A6), respectively.
4.6 Robustness checksTo ensure that our results and estimations are robust, we have performed and checkedthe following six points:
(1) For almost every monetary policy variable (money, efficiency or credit), twoindicators have been used. Thus, the findings have encapsulated measures ofmonetary policy dynamics both from banking and financial systemperspectives.
(2) By using bivariate analysis in cointegration tests and correspondingly in VECMestimations, we have focused on the problem statements and limited (mitigated)causality misspecification issues.
(3) Optimal lag selection for goodness of fit in model specifications has beenconsistent with the recommendations of Liew (2004)[15].
(4) Simple Granger causality has been carefully calibrated to be consistent with theEngle–Granger theorem.
(5) In the error correction analysis, the ability of financial allocation efficiency toinsignificantly contribute in restoring inflation to its long-run equilibrium isconsistent with the estimates of financial activity[16].
(6) The signs and intervals of the ECTs conform to theory. It is worthwhile layingemphasis on this seventh point.
As a matter of principle, the speed of adjustment should be between zero and “minusone” (0 and �1) for a stable error correction mechanism. Therefore, if the ECTs are notwithin this interval, then the model is misspecified (and needs adjustment), the data isinadequate (perhaps owing to issues with degrees of freedom)[17] or, ultimately, theerror correction mechanism is unstable.
4.7 Discussion and policy implications4.7.1 Retrospect to tested hypotheses.
H1. Monetary policy variables affect prices in the long-run but not in the short-run.
For the first-half (long-run dimension) of the hypothesis, permanent changes inmonetary policy variables (depth, efficiency, activity and size) affect permanent
IGDR7,2
162
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
variations in prices in the long-term. But in cases of disequilibriums, only financialdynamic fundamentals of depth and size significantly adjust inflation to thecointegration relations. With respect to the second-half (short-run view) of thehypothesis, monetary policy does not overwhelmingly affect prices in the short-term.Hence, but for a thin exception, H1 is valid.
H2. Monetary policy variables influence output in the short-term but not in thelong-term.
With regard to the short-term dimension of the hypothesis, only financial dynamics ofdepth and size affect real GDP output in the short-run. As concerns the long-rundimension, the neutrality of monetary policy has been confirmed. Hence, the hypothesisis also broadly valid.
Accordingly, had we used only the 1987-2006 sample period and restricted monetarypolicy variables to fundamental financial dynamics of depth and size, both hypotheseswould have been overwhelmingly valid.
4.7.2 Implications for the long-run neutrality of money and business cycles. Economictheory has traditionally suggested that monetary policy can influence the business cycle, butnot the long-run potential output. Despite theoretical and empirical consensus on moneyneutrality well documented in the literature, the role of money as an informational variablefor money policy decisions has remained opened to debate with empirical works providingmixed outcomes. By using hitherto unemployed monetary policy variables, results offer onlypartial support for the traditional economic theory. While evidence of the long-run neutrality ofmoney has been confirmed, the influence of monetary policy in short-run economic activity isapparent only with respect to monetary policy fundamentals of depth and size, with a greatereffect from the latter (see discussion on IRFs). It follows that contractionary and expansionarypolicies based on financial size would be more effective than those based on financial depth.Overall, this finding is not consistent with studies on two middle-income countries byAgenor et al. (2000), which establish no evidence of Granger-causality flowing from moneyto output, regardless of the measurement of money used.
The inability of monetary policy fundamentals of efficiency and activity to affectoutput in the short-run indicates that they cannot be used as policy instruments toinfluence business cycles (through expansionary and contractionary policies). Surplusliquidity issues (financial allocation inefficiency) and limited credit facilities (financialinactivity) could explain this side of the findings.
4.7.3 Implications for credit expansions and inflationary tendencies. There is a generalconsensus among analysts that significant money stock expansions that are not coupledwith sustained credit increases are less likely to have any inflationary tendencies. This is atleast true in the long-run because monetary policy variables theoretically have no incidenceon prices in the short-term. This paper has reframed the consensus into an importantquestion policymakers are most likely to ask today. In the long-run, do short-termadjustments in monetary policy variables matter in the restoration of the long-runinflation-money equilibrium? The findings have three mains implications on this account:
(1) there are significant long-run equilibriums between inflation and monetarypolicy variables;
(2) the error correction mechanism is stable, such that in the event of adisequilibrium, for all monetary policy variables under consideration, inflationis adjusted to the long-run relationship; and
163
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
(3) only adjustments in the monetary policy fundamentals of financial size andfinancial depth are significant in the restoration of inflation to its equilibrium.
This broadly implies that financial depths and size (if well calibrated) could be used forinflation targeting by the monetary authorities.
4.7.4 Implications for inflation targeting. Three main policy implications could bederived with respect to inflation targeting. First, we have seen that financial depth is asignificant tool in fighting inflation with the effect of money supply doubling that of liquidliabilities. This difference in magnitude is broadly consistent with the fundamentals offinancial development in African countries. It has been substantially documented that agreat chunk of the monetary base does not transit through the banking sector (Asongu,2014b). Hence, it is only normal to expect that adjustments in bank deposits (liquid liabilities)are less significant in magnitude in the restoration of inflation to its long-term equilibrium.Corollary to this explanation is the growing phenomenon of mobile banking and otherinformal financial activities that contribute to inflation but are not captured by formal(mainstream) financial (banking) activities (Asongu, 2013d). Second, we have also noticedthat the monetary policy variable of financial size (in terms of magnitude) more significantlyadjusts inflation than financial intermediary depth (from money supply and liquid liabilitiesperspectives). In plainer terms, decreasing financial intermediary assets (in relation to totalassets in the financial system) more substantially exerts deflationary pressures on consumerprices. It could thus be inferred that, tight monetary policy targeting the ability of banks togrant credit (in relation to central bank credits) is more effective[18] in fighting consumerprice inflation than that targeting the ability of banks to receive deposits (financial depth).Third, we have also seen that monetary policy variables of financial depth and financial sizeare more significant adjusters of inflation than financial efficiency and activity. The inherentsurplus liquidity issues in African banks still explain this insignificance in adjustments(Saxegaard, 2006).
4.7.5 Implications for the ongoing monetary policy debate. The monetary policyeffects (long-run and short-term) on economic activity (output and prices) are broadlyconsistent with traditional discretionary monetary policy arrangements that favorcommitment to price stability and international economic integration (such as inflationtargeting, monetary unions, etc). Hence, on the one hand, a flexible countercyclicalmonetary policy (with financial fundamentals of depth and size) can be practiced withinflation targeting (Ghironi and Rebucci, 2000; Mishkin, 2002) and, on the other hand,contractionary and expansionary monetary policies based on financial depth and sizecan be used to offset output in the short term (Starr, 2005).
Conversely, the absence of short-run effects from some monetary policyfundamentals on output is in line with the second strand of the debate which sustainsthat non-traditional policy regimes limit the ability of monetary authorities to use policyto offset output fluctuations. The inability of monetary policy authorities to use financialfundamentals of efficiency and activity to affect short-run real GDP is consistent withWeeks (2010) who views this IMF-oriented approach as absurdly inappropriate becausea vast majority of governments in SSA countries lack the instruments to make monetarypolicy effective[19]. From our findings, the monetary authorities can only use policyinstruments of depth and size to offset adverse shocks to output by pursuing either anexpansionary or a contractionary policy. In the same vein, we have also discovered that
IGDR7,2
164
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
financial fundamentals of efficiency and activity do not significantly adjust inflation tothe long-run equilibrium (Table IV).
4.7.6 Implications for monetary policy independence. Beside the underlyinghypotheses motivating the inquiry, the empirics have given birth to anotherhypothesis of monetary policy independence. While the intuition of and justificationfor the hypothesis have already been substantially covered in Section 4.2, it isrelevant to take stock of how the assumption of monetary policy independence hasinfluenced the results. Two points clearly stand out. First, the assumption has notaffected the relevance of H2 for two main reasons: on the one hand, the Westerlund(2007) and Pedroni (1999) tests are consistent on the long-run neutrality of money; onthe other hand, short-run Granger estimates are not affected (see Panel A2 and PanelB2 in Table V). Second, if the hypothesis of independence in monetary policy isrelaxed, two facts emerge: on one hand, only financial efficiency has a long-termequilibrium with inflation and, on the other hand, in case of disequilibrium,adjustments in the fundamental dynamics of allocation efficiency do notsignificantly adjust inflation to the cointegration relationship. Ultimately, the issueof surplus liquidity still emerges even without the hypothesis of monetary policyindependence.
4.7.7 Country-/regional-specific implications. We now devote space to discussingspecific implications for some of the sampled countries. The surplus liquidity issues may bedue to the weight of political instability in some of the sampled countries. Fielding andShortland (2005) have established a positive nexus between violent political incidents(arising from conflicts between radical Islamic groups and the Egyptian state) and excessliquidity. While categorizing “conflict-affected” countries present analytical and practicaldifficulties, essentially because few countries in the world are completely conflict-free, fewwould object the extension of the Fielding and Shortland interpretation to Nigeria andSudan. Also, Uganda is still technically at war with the Lord Resistance Army because itsleader Kony (who refused to sign the 2007 peace agreement) is still at large. Beyond pointingout countries that have experienced conflicts during a significant interval of the sampledperiod, it is also relevant to emphasize that in the same vein, the protracted Arab Springrevolutions (Egyptian and Tunisian) and sectarian crisis in Nigeria would only fuel excessliquidity issues.
Like most countries in the sample, since 2002, the Bank of Algeria has conducted activepolicy aimed at solving the problem of excess liquidity. The plethora of measuresimplemented to curb the issue has led to some satisfactory results (IMF, 2013, p. 22). Theinternal impediments to financial intermediation are being exacerbated by Lesotho’s smallsize and its proximity to South Africa. Accordingly, the excess liquidity managementmeasures may not sustain the domestic economy in the long run. While potential customersare taking businesses to the much better and developed South African market, at the sametime, instead of looking for business opportunities in Lesotho, the banks are investing excessliquidity in South African securities.
A recent short-run Schumpeterian trip to embryonic African monetary zones (Asongu,2013c) has established a positive causality flowing only from financial efficiency to real GDPoutput for the EAMZ and none for the WAMZ. Hence, these results indirectly confirm theneed to implement structural and institutional convergence reforms by candidate membersof the potential monetary unions (especially the WAMZ).
4.7.8 Fighting surplus liquidity. It is also relevant to devote space in discussing how to
165
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
fight surplus liquidity in African banking institutions for two main reasons. First, as wehave seen above, it is a common issue resulting from monetary policy independence anddependence scenarios. Second, the two hypotheses motivating the empirics would have beenoverwhelmingly validated only and only if we had used the 1987-2006 sample period andrestricted monetary policy variables to fundamental financial dynamics of depth and size.
Measures aim at tackling over-liquidity will be efficient if they are consistent with thereasons for holding liquidity: voluntary or involuntary. First, voluntary holding of excessliquidity could be mitigated by:
• easing difficulties encountered by banks in tracking their positions at the centralbank that may require them to hold reserves above the statutory limits;
• reinforcement of institutions that would favor interbank lending so as to easeborrowing between banks for contingency purposes; and
• improve infrastructure so that remote bank branches may not need to hold excessreserves due to transportation problems.
Second, involuntary holding of excess liquidity could also be avoided by:• decreasing the inability of banks to lend, especially in situations where interest
rates are regulated[20];• creating conditions to sustain the spread between bonds and reserves, so that
commercial banks can invest excess liquidity in the bond markets;• stifling the unwillingness of banks to expand lending by reducing asymmetric
information and lack of competition; and• developing regional stock exchange markets to broaden investment opportunities
for commercial banks.
4.7.9 Caveats and future directions. The first drawback to the analysis is the limitedsample in the dataset. We have limited our sample to only 10 countries because of substantialreasons already discussed in the data section. Hence, generalization of the results to the entireAfrican continent should be treated with caution. Second, while the intuition motivating theempirics is sound and the employment of all fundamental financial dynamics identified bythe FDSD of the WB, innovative interest rates and exchange rates have not been directlytaken into account. Third, given the length of the sample period, there could have beenstructural breaks or regime shifts which have not been taken into account due to technicaland logistical reasons. Fourth, we have only considered financial intermediary determinantsof output and inflation in the analysis. However, in the real world, economic activity in terms ofoutput and inflation are endogenous to a complex set of variables: exchange rates, wages, pricecontrols, etc. Therefore, the interaction of money, credit, efficiency and size elasticities of inflationand output with other determinants of economic activity could result in other dynamics ofconsumer price and output variations. Hence, replicating the analysis in a multivariate VARcontext would be interesting. Another interesting future research direction could be to assesswhether the findings apply to country-specific cases of the sample.
5. ConclusionWhile in developed economies, changes in monetary policy affect real economic activity inthe short-run, but only prices in the long-run, the question of whether these tendencies applyto developing countries remain open to debate. In this paper, we have examined the real
IGDR7,2
166
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
effects of monetary policy using a battery of estimation techniques in inflation-chaoticAfrican countries for the period of 1987-2010. By using a plethora of hitherto unemployedfinancial dynamics (that broadly reflect monetary policy), we have provided significantcontributions to the empirics of money. Two main hypotheses have been tested.
H1. Monetary policy variables affect prices in the long-run but not in the short-run.For the first-half (long-run dimension) of the hypothesis, permanent changes inmonetary policy variables (depth, efficiency, activity and size) affect permanentvariations in prices in the long-term. But in cases of disequilibriums, onlyfinancial dynamic fundamentals of depth and size significantly adjust inflationto the cointegration relations. With respect to the second-half (short-run view) ofthe hypothesis, monetary policy does not overwhelmingly affect prices in theshort-term. Hence, but for a thin exception, H1 is valid.
H2. Monetary policy variables influence output in the short-term but not in thelong-term. With regard to the short-term dimension of the hypothesis, onlyfinancial dynamics of depth and size affect real GDP output in the short-run. Asconcerns the long-run dimension, the neutrality of monetary policy has beenconfirmed. Hence, the hypothesis is also broadly valid.
Accordingly, had we used only the 1987-2006 sample period and restricted monetarypolicy variables to fundamental financial dynamics of depth and size, both hypotheseswould have been overwhelmingly valid. Moreover, but for slight exceptions, the testedhypotheses are valid under monetary policy independence and dependence. Theconclusion of the analysis is a valuable contribution to the scholarly and policy debateon how money matters as an instrument of economic growth. A wide range of policyimplications have been discussed.
Notes1. As a matter of fact, empirical literature provides mixed results and the outcomes are
contingent on selected countries and historical periods under investigation (Stock andWatson, 1999; Dwyer and Hafer, 1999; Trecroci and Vega-Croissier, 2000; Leeper and Roush,2002; Bae et al., 2005).
2. According to the Director General of the International Food Policy Research Institute,monetary and exchange rate responses were not effective in addressing food inflation (VonBraun, 2008).
3. Weeks asserts that SSA lacks two main channels for implementing monetary policy: trying toinfluence the creation of private credit through so-called open market operations or; seeking toinfluence the borrowing rates for private sector by adjusting the interest rate at whichcommercial banks can borrow from the central bank.
4. According to Mangani (2011), in the short-run, pursuing a prudent exchange rate policy thatrecognizes the country’s precarious foreign reserve position could be critical in deepeningdomestic price stability. Beyond the short-run, policy stability could be sustained via theimplementation of policies directed toward the construction of a strong foreign exchangereserve base (as well as developing a sustainable approach to the country’s reliance ondevelopment assistance).
5. Gupta et al. (2010a, 2010b) do not quantify the indirect effects of interest rate changes workingthrough changes in house prices on consumer spending. Ncube and Ndou (2011) fill this gap
167
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
by estimating and quantifying the role of house wealth in South Africa using disaggregatedhouse prices.
6. The CFA franc is the name of two currencies used in sub-Saharan Africa (by some former Frenchcolonies) which are guaranteed by the French treasury. The two currencies though theoreticallyseparate are effectively interchangeable and include: the West African CFA franc (used in theUEMOA zone) and the Central African CFA franc (used in the CEMAC zone).
7. Economic and Monetary Community of Central African States.
8. Economic and Monetary Community of West African States.
9. The need for inflation to reflect a unit root to accommodate the problem statement (and theexclusion of CFA franc countries) also draws from an inflation uncertainty theory in recentAfrican finance literature. “The dominance of English common–law countries in prospects forfinancial development in the legal– origins debate has been debunked by recent findings.Using exchange rate regimes and economic/monetary integration oriented hypotheses, thispaper proposes an ‘inflation uncertainty theory’ in providing theoretical justification andempirical validity as to why French civil–law countries have higher levels of financialallocation efficiency. Inflation uncertainty, typical of floating exchange rate regimes accountsfor the allocation inefficiency of financial intermediary institutions in English common–lawcountries. As a policy implication, results support the benefits of fixed exchange rate regimesin financial intermediary allocation efficiency” Asongu (2011, p. 1). Moreover, before limitingthe dataset, we have found from preliminary analysis that, African CFA franc countries havea relatively very stable inflation rate.
10. “The French and English traditions in monetary theory and history have been different […]The French tradition has stressed the passive nature of monetary policy and the importanceof exchange stability with convertibility; stability has been achieved at the expense ofinstitutional development and monetary experience. The British countries by opting formonetary independence have sacrificed stability, but gained monetary experience and betterdeveloped monetary institutions.” (Mundell, 1972, pp. 42-43).
11. By financial efficiency here, we neither refer to the profitability related concept (notion) nor tothe production efficiency of decision-making units in the financial sector (through DataEnvelopment Analysis: DEA).
12. Whereas Pedroni (1999) is applied in the presence of both “constant” and “constant and trend”,Kao (1999) is based only on the former (constant).
13. For example, multivariate cointegration and corresponding error correction model mayinvolve variables that are stationary in levels (Gries et al., 2009).
14. The Breusch–Pagan (BP) LM test of cross-sectional independence is overwhelming rejected.The BP LM test is chosen because T � N (24 � 10 for the 1987-2010 period and 20�10 for the1987-2006 period).
15. “The major findings in the current simulation study are previewed as follows. First, thesecriteria managed to pick up the correct lag length at least half of the time in small sample.Second, this performance increases substantially as sample size grows. Third, with relativelylarge sample (120 or more observations), HQC is found to outdo the rest in correctlyidentifying the true lag length. In contrast, AIC and FPE should be a better choice for smallersample. Fourth, AIC and FPE are found to produce the least probability of under estimationamong all criteria under study. Finally, the problem of over estimation, however, is negligible
IGDR7,2
168
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
in all cases. The findings in this simulation study, besides providing formal groundworksupportive of the popular choice of AIC in previous empirical researches, may as well serve asuseful guiding principles for future economic researches in the determination ofautoregressive lag length” (Liew, 2004, p. 2).
16. As we have observed, if allocation inefficiency is a substantial cause for the insignificantcontribution of the monetary policy variable of financial efficiency in adjusting inflation to itslong-run equilibrium, then correspondingly, financial activity (credit) which indirectlyreflects bank efficiency (ability of banks to transform deposits into credit) should also beinsignificant in adjusting inflation to its long-run equilibrium relationship. This hypothesis isconsistent with our findings.
17. “The error correction term tells us the speed with which our model returns to equilibriumfollowing an exogenous shock. It should be negatively signed, indicating a move backtowards equilibrium, a positive sign indicates movement away from equilibrium. Thecoefficient should lie between 0 and 1, 0 suggesting no adjustment one time period later, 1indicates full adjustment. The error correction term can be either the difference between thedependent and explanatory variable (lagged once) or the error term (lagged once), they are ineffect the same thing” (Babazadeh and Farrokhnejad, 2012, p. 73).
18. Effectiveness here refers to the rate at which inflation is adjusted to its long-run equilibrium.
19. The International Monetary Fund (IMF) places great emphasis on monetary policy in itsprograms for developing countries in sub-Saharan Africa (SSA). It views such policy ascrucial in managing inflation and stabilizing the real exchange rate.
20. For instance, this is the case of the CEMAC region where the central bank sets a floor forlending rates and a ceiling for deposit rates above and below which interest rates arenegotiated freely.
ReferencesAgénor, P.R., McDermott, C.J. and Ucer, E.M. (1997), “Fiscal imbalances, capital inflows, and the
real exchange rate: the case of Turkey”, IMF Working Paper 97/1, Washington DC.
Agénor, P.R., McDermott, C.J. and Prasad, E.S. (2000), “Macroeconomic fluctuations in developingcountries: some stylized facts”, World Bank Economic Review, Vol. 14 No. 2, pp. 251-286.
Asongu, S.A. (2011), “Why do French civil-law countries have higher levels of financialefficiency?”, Journal of Advanced Research in Law and Economics, Vol. 2 No. 2, pp. 94-108.
Asongu, S.A. (2014a), “Are proposed African Monetary Unions optimal currency areas? Real,monetary and fiscal policy convergence analysis”, African Journal of Economic andManagement Studies, Vol. 5 No. 1, pp. 9-29.
Asongu, S.A. (2013e), “African stock market performance dynamics: a multidimensionalconvergence assessment”, Journal of African Business, Vol. 14 No. 3, pp. 186-201.
Asongu, S.A. (2014b), “African financial development dynamics: big time convergence”, AfricanJournal of Economic and Management Studies, Vol. 5 No. 2, pp. 160-194.
Asongu, S.A. (2013a), “Fighting consumer price inflation in Africa: What do dynamics in money,credit, efficiency and size tell us?”, Journal of Financial Economic Policy, Vol. 5 No. 1,pp. 39-60.
Asongu, S.A. (2013b), “Correcting inflation with financial dynamic fundamentals: whichadjustments matter in Africa”, Journal of African Business.
169
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Asongu, S.A. (2013c), “A short-run Schumpeterian trip to embryonic African monetary zones”,Economic Bulletin, Vol. 33 No. 1, pp. 859-873.
Asongu, S.A. (2013d), “How has mobile phone penetration stimulated financial development inAfrica?”, Journal of African Business, Vol. 14 No. 1, pp. 7-18.
Babazadeh, M. and Farrokhnejad, F. (2012), “Effects of short-run and long-run changes in foreignexchange rates on banks’ Profit”, International Journal of Business and Management, Vol. 7No. 17, pp. 70-77.
Bae, S. and Ratti, R. (2000), “Long-run neutrality, high inflation, and bank insolvencies inArgentina and Brazil”, Journal of Monetary Economics, Vol. 46, pp. 581-604.
Bae, S., Jensen, M. and Murdock, S. (2005), “Long-run neutrality in a fractionally integratedmodel”, Journal of Macroeconomics, Vol. 27 No. 2, pp. 257-274.
Bernanke, B. and Mihov, I. (1998), “The liquidity effect and long-run neutrality”,Carnegie-Rochester Conference Series on Public Policy, Vol. 49, pp. 149-194.
Bhaduri, S. and Durai, S.R.S. (2012), “A note on excess money growth and inflation dynamics:evidence from threshold regression”, MPRA Paper No. 38036.
Blinder, A.S. (1987), “Credit rationing and effective supply failures”, The Economic Journal,Vol. 97 No. 386, pp. 327-352.
Bordo, M.D. and Jeanne, O. (2002), “Monetary policy and asset prices: does ‘benign neglect’ makesense?”, International Finance, Vol. 5 No. 2, pp. 139-164.
Bullard, J. (1999), “Testing long-run neutrality propositions: lessons from the recent research”,Federal Reserve Bank of St. Louis Review, Vol. 81 No. 6, pp. 57-78.
Camarero, M. and Tamarit, C. (2002), “A panel cointegration approach to the estimation of thepeseta real exchange rate”, Journal of Macroeconomics, Vol. 24 No. 3, pp. 371-393.
Cavoli, T. and Rajan, R.S. (2008), “Open economy inflation targeting arrangements and monetarypolicy rules. Application to India”, Indian Growth and Development Review, Vol. 1 No. 2,pp. 237-251.
Cristadoro, R. and Veronese, G. (2011), “Monetary policy in India: is something amiss?”, IndianGrowth and Development Review, Vol. 4 No. 2, pp. 166-192.
Dornbusch, R. (2001), “Fewer monies, better monies”, American Economic Review, Vol. 91 No. 2,pp. 238-242.
Dwyer, G.P. and Hafer, R.W. (1999), “Are money growth and inflation still related”, FederalReserve Bank of Atlanta Economic Review, pp. 32-43.
Engle, R.F. and Granger, W.J. (1987), “Cointegration and error correction: representation,estimation and testing”, Econometrica, Vol. 55 No. 2, pp. 251-276.
Fielding, D. and Shields, K. (2005), “Asymmetries in the effects of monetary policy: the case ofSouth Africa”, Economic Discussion Papers No. 0509, University of Otago, Otago.
Fielding, D. and Shortland, A.K. (2005), “Political violence and excess liquidity in Egypt”, TheJournal of Development Studies, Vol. 41 No. 4, pp. 542-557.
Fouda, O.J.P. (2009), “The excess liquidity of banks in Franc zone: how to explain the paradox inthe CEMAC”, Revue Africaine de l’Integration, Vol. 3 No. 2, pp. 1-56.
Frankel, J., Schmukler, S. and Serven, L. (2004), “Global transmission of interest rates: Monetaryindependence and currency regime”, Journal of International Money and Finance, Vol. 23No. 5, pp. 701-733.
Gagnon, J. and Ihrig, J. (2004), “Monetary policy and exchange rate pass through”, InternationalDiscussion Paper, Federal Board, Washington, DC.
IGDR7,2
170
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Ganev, G.Y., Molnar, K., Rybinski, K. and Wozniak, P. (2002), “Transmission mechanism ofmonetary policy in Central and Eastern Europe”, Report No. 52, Center for Social andEconomic Research (CASE), Warsaw.
Gerlach, S. and Svensson, L. (2003), “Money and Inflation in the Euro area: a case for monetaryindicators?”, Journal of Monetary Economics, Vol. 50 No. 8, pp. 1649-1672.
Ghironi, F. and Rebucci, A. (2000), “Monetary rules for emerging market economies”, WorkingPaper No. 476, Boston College Economics Department, Boston, MA.
Gries, T., Kraft, M. and Meierrieks, D. (2009), “Linkages between financial deepening, tradeopenness, and economic development: causality evidence from Sub-Saharan Africa”, WorldDevelopment, Vol. 37 No. 12, pp. 1849-1860.
Gupta, R., Jurgilas, M. and Kabundi, A. (2010a), “The effect of monetary policy on real house pricegrowth in South Africa: a factor augmented vector auto regression (FAVAR) approach”,Economic Modelling, Vol. 27 No. 1, pp. 315-323.
Gupta, R., Miller, S.M. and Van Wyk, D. (2010b), “Financial market liberalization, monetary policyand housing price dynamics”, Working Paper 2010-2006, Department of Economics,University of Connecticut, CT.
Hafer, R.W. and Kutan, A. (2002), “Detrending and the money– output link: internationalevidence”, Southern Economic Journal, Vol. 69 No. 1, pp. 159-174.
Hayo, B. (1999), “Money– output Granger causality revisited: an empirical analysis of EUcountries”, Applied Economics, Vol. 31 No. 11, pp. 1489-1501.
Hendrix, C., Haggard, S. and Magaloni, B. (2009), “Grievance and opportunity: food prices,political regime and protest”, paper prepared for presentation at the International StudiesAssociation Convention, New York, NY.
Im, K.S., Pesaran, M.H. and Shin, Y. (2003), “Testing for unit roots in heterogeneous panels”,Journal of Econometrics, Vol. 115, pp. 53-74.
IMF (2013), “Algeria: selected issues paper”, IMF Country Report No. 13/48.Kao, C. (1999), “Spurious regression and residual-based tests for cointegration in panel data”,
Journal of Econometrics, Vol. 90, pp. 1-44.Khan, M.S. (2011), “The design and effects of monetary policy in Sub-Saharan Africa”, Journal of
African Economies, Vol. 20 No. S2, p. 35.Kishor, N. and Ssozi, J. (2010), “Inflation and currency unions: the case of the East African
Community”, Indian Growth and Development Review, Vol. 3 No. 1, pp. 36-52.Leeper, E.M. and Roush, J.E. (2002), “Putting ‘M’ back into monetary policy”, Journal of Money
Credit and Banking, Vol. 35 No. 6, pp. 1217-1256.Levin, A., Lin, C.F. and Chu, C.S. (2002), “Unit root tests in panel data: asymptotic and
finite-sample properties”, Journal of Econometrics, Vol. 108 No. 1, pp. 1-24.Levine, P. (2012), “Monetary policy in an uncertain world: probability models and the design of
robust monetary rules”, Indian Growth and Development Review, Vol. 5 No. 1, pp. 70-88.Liew, V.K. (2004), “Which lag selection criteria should we employ?”, Economics Bulletin, Vol. 3
No. 33, pp. 1-9.Lucas, R.E. (1980), “Two illustrations of the quantity theory of money”, American Economic
Review, Vol. 70, pp. 1005-1014.Maddala, G.S. and Wu, S. (1999), “A comparative study of unit root tests with panel data and a new
simple test”, Oxford Bulletin of Economics and Statistics, Vol. 61, pp. 631-652.Mangani, R. (2011), The Effects of Monetary Policy on Prices in Malawi, Department of Economics,
University of Malawi.
171
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Mishkin, F.S. (2002), “The role of output stabilization in the conduct of monetary policy”,International Finance, Vol. 5 No. 2, pp. 213-227.
Moorthy, V. and Kolhar, S. (2011), “Rising food inflation and India’s monetary policy”, IndianGrowth and Development Review, Vol. 4 No. 1, pp. 73-94.
Moosa, I. (1997), “Testing the long-run neutrality of money in a developing economy: the case ofIndia”, Journal of Development Economics, Vol. 53 No. 1, pp. 139-155.
Mundell, R. (1972), “African trade, politics and money”, in Tremblay, R. (Ed), Africa and MonetaryIntegration, Les Editions HRW, Montreal, pp. 11-67.
Ncube, M. and Ndou, E. (2011), “Monetary policy transmission, house prices and consumerspending in South Africa: an SVAR approach”, Working Paper Series No. 133, AfricanDevelopment Bank, Tunis, Tunisia.
Ngalawa, H. and Viegi, N. (2011), “Dynamic effects of monetary policy shocks in Malawi”, SouthAfrican Journal of Economics, Vol. 79 No. 3, pp. 224-250.
Nogueira, R.P. (2009), “Is monetary policy really neutral in the long-run? Evidence for someemerging and developed economies”, Economics Bulletin, Vol. 29 No. 3, pp. 2432-2437.
Olekalns, N. (1996), “Some further evidence on the long-run neutrality of money”, EconomicsLetters, Vol. 50, pp. 393-398.
Pedroni, P. (1999), “Critical values for cointegration tests in heterogeneous panels with multipleregressors”, Oxford Bulletin of Economics and Statistics, Vol. 61 No. S1, pp. 653-670.
Pesaran, H.M. (2007), “A simple panel unit root test in the presence of cross-section dependence”,Journal of Applied Econometrics, Vol. 22 No. 2, pp. 265-312.
Roffia, B. and Zaghini, A. (2008), “Excess money growth and inflation dynamics”, Working PaperNo. 657, Bank of Italy Temi di Discussione.
Saxegaard, M. (2006), “Excess liquidity and effectiveness of monetary policy: evidence fromsub-Saharan Africa”, IMF Working Paper 06/115, Washington DC.
Serletis, A. and Koustas, Z. (1998), “International evidence on the neutrality of money”, Journal ofMoney, Credit and Banking, Vol. 30 No. 1, pp. 1-25.
Starr, M. (2005), “Does money matter in the CIS? Effects of monetary policy on output and prices”,Journal of Comparative Economics, Vol. 33 No. 3, pp. 441-461.
Stock, J.H. and Watson, M.W. (1999), “Forecasting inflation”, Journal of Monetary Economics,Vol. 44 No. 2, pp. 293-335.
Trecroci, C. and Vega-Croissier, J.L. (2000), “The information content of M3 for future inflation”,ECB Working Paper No. 33, Frankfurt.
Von Braun, J. (2008), “Rising food prices: dimension, causes, impact and responses”, Key NoteAddress at World Food Programme, 9 April, available at: http://documents.wfp.org/stellent/groups/public/documents/resources/wfp175955.pdf (accessed 21 1 2012).
Weeks, J. (2010), “Why monetary policy is irrelevant in Africa South of the Sahara”, DevelopmentViewpoint No. 53, School of Oriental and African Studies, Center for Developmentand Policy Research, available at: www.soas.ac.uk/cdpr/publications/dv/file59766.pdf(accessed 29 October 2012).
Westerlund, J. (2007), “Testing for error correction in panel data”, Oxford Bulletin of Economicsand Statistics, Vol. 69 No. 6, pp. 709-748.
IGDR7,2
172
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Appendix 1
Appendix 2
Table AI.Summary statistics and
presentation of countries
Variables Mean SD Min. Max. Obser.
Panel A: Summary statisticsMonetary policy dynamicsFinancial depth Money supply 0.384 0.256 0.001 1.141 240
Liquid liabilities 0.310 0.218 0.001 0.948 240Financial efficiency Banking system efficiency 0.642 0.346 0.139 2.103 240
Financial system efficiency 0.644 0.343 0.139 1.669 240Financial activity Banking system activity 0.203 0.194 0.001 0.825 240
Financial system activity 0.213 0.207 0.001 0.796 240Financial size Banking system size 0.672 0.299 0.017 1.609 240Output variable Real gross domestic product 10.193 0.628 8.561 11.340 240Price variable Consumer price index 19.714 31.862 �9.616 200.03 240
Panel B: Presentation of countries (10)Algeria, Egypt, Lesotho, Morocco, Nigeria, Sudan, Tunisia, Uganda, Zambia, Tanzania
Notes: SD: Standard Deviation. Min: Minimum. Max: Maximum. Obser: Observations. Fin: Financial.
Table AII.Variable definitions
Variables Signs Variable definitions Sources
Inflation Infl. Consumer Price Index (Annual %) World Bank (WDI)Real output Output Logarithm of Real GDP World Bank (WDI)Economic financialdepth (Money Supply)
M2 Monetary Base plus demand, saving andtime deposits (% of GDP)
World Bank (FDSD)
Financial system depth(Liquid liabilities)
Fdgdp Financial system deposits (% of GDP) World Bank (FDSD)
Banking systemallocation efficiency
BcBd Bank credit on Bank deposits World Bank (FDSD)
Financial systemallocation efficiency
FcFd Financial system credit on Financialsystem deposits
World Bank (FDSD)
Banking systemactivity
Pcrb Private credit by deposit banks (% ofGDP)
World Bank (FDSD)
Financial systemactivity
Pcrbof Private credit by deposit banks and otherfinancial institutions (% of GDP)
World Bank (FDSD)
Banking system size Dbacba Deposit bank assets/Total assets (Depositbank assets plus Central bank assets)
World Bank (FDSD)
Notes: Infl: Inflation; M2: money supply; Fdgdp: Liquid liabilities; BcBd: Bank credit on Bank deposits;FcFd: Financial system credit on financial system deposits; Pcrb: Private domestic credit by deposit banks;Pcrbof: Private domestic credit by deposit banks and other financial institutions; WDI: World developmentindicators; FDSD: financial development and structure database; GDP: Gross domestic product
173
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Appendix 3
Table AIII.Correlation analysis
Financialdepth
Financialefficiency
Financialactivity
Financial.size Inflation GDP Variables
M2 Fdgdp BcBd FcFd Pcrb Pcrbof Dbacba Infl. Output
1.000 0.993 0.146 0.183 0.796 0.778 0.513 �0.332 0.501 M21.000 0.157 0.189 0.801 0.782 0.538 �0.354 0.464 Fdgdp
1.000 0.949 0.630 0.646 0.392 �0.173 0.284 BcBd1.000 0.668 0.708 0.371 �0.195 0.274 FcFd
1.000 0.987 0.536 �0.321 0.422 Pcrb1.000 0.546 �0.339 0.412 Pcrbof
1.000 �0.323 0.385 Dbacba1.000 �0.219 Inflation
1.000 Output
Notes: M2: Money Supply; Fdgdp: Liquid liabilities; BcBd: Bank credit on Bank deposit (Banking SystemIntermediary Efficiency); FcFd: Financial credit on financial deposits (Financial System IntermediaryEfficiency); Pcrb: Private domestic credit by deposit banks (Banking System Intermediary Activity); Pcrbof:Private credit domestic credit from deposit banks and other financial institutions (Financial SystemIntermediary Activity); Dbacba: Deposit bank asset on total assets (Banking system size); Infl: Inflation. Fin:Financial
IGDR7,2
174
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Appendix 4
IRFs with VAR ordering: economic activity; monetary policy
−0.0
2
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(L
OG
RE
AL
GD
P)
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(M
2)
−0
.02
−0
.01
0.0
0
0.0
1
0.0
2
0.0
3
0.0
4
−0.0
2
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.02
−0
.01
0.0
0
0.0
1
0.0
2
0.0
3
0.0
4
12
34
56
78
91
0
Re
sp
on
se
of D
(M2
) to
D(L
OG
RE
AL
GD
P)
12
34
56
78
91
0
Re
sp
on
se
of D
(M2
) to
D(M
2)
Re
sp
on
se
to
Ch
ole
sky O
ne
S.D
. In
no
va
tio
ns ±
2 S
.E.
Figure A1.Dynamic responses of
output and money supply
175
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
−0.0
2
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(L
OG
RE
AL
GD
P)
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(F
DG
DP
)
−0
.02
−0
.01
0.0
0
0.0
1
0.0
2
0.0
3
−0.0
2
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.02
−0
.01
0.0
0
0.0
1
0.0
2
0.0
3
12
34
56
78
91
0
Re
sp
on
se
of D
(FD
GD
P) to
D(L
OG
RE
AL
GD
P)
12
34
56
78
91
0
Re
sp
on
se
of D
(FD
GD
P) to
D(F
DG
DP
)
Re
sp
on
se
to
Ch
ole
sky O
ne
S.D
. In
no
va
tio
ns ±
2 S
.E.
Figure A2.Dynamic responses ofoutput and liquidliabilities
IGDR7,2
176
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(L
OG
RE
AL
GD
P)
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(D
BA
CB
A)
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(DB
AC
BA
) to
D(L
OG
RE
AL
GD
P)
12
34
56
78
91
0
Re
sp
on
se
of D
(DB
AC
BA
) to
D(D
BA
CB
A)
Re
sp
on
se
to
Ch
ole
sky O
ne
S.D
. In
no
va
tio
ns ±
2 S
.E.
Figure A3.Dynamic responses of
output and financial size
177
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
Appendix 5
IRFs with VAR ordering: monetary policy; economic activity
−0.0
1
0.0
0
0.0
1
0.0
2
0.0
3
0.0
4
12
34
56
78
91
0
Re
sp
on
se
of D
(M2
) to
D(M
2)
12
34
56
78
91
0
Re
sp
on
se
of D
(M2
) to
D(L
OG
RE
AL
GD
P)
−0
.04
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0.0
1
0.0
0
0.0
1
0.0
2
0.0
3
0.0
4
−0
.04
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(M
2)
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(L
OG
RE
AL
GD
P)
Re
sp
on
se
to
Ch
ole
sky O
ne
S.D
. In
no
va
tio
ns ±
2 S
.E.
Figure A4.Dynamic responses ofoutput and money supply
IGDR7,2
178
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
−0
.01
0.0
0
0.0
1
0.0
2
0.0
3
12
34
56
78
91
0
Re
sp
on
se
of D
(FD
GD
P) to
D(F
DG
DP
)
12
34
56
78
91
0
Re
sp
on
se
of D
(FD
GD
P) to
D(L
OG
RE
AL
GD
P)
−0
.04
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.01
0.0
0
0.0
1
0.0
2
0.0
3
−0
.04
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(F
DG
DP
)
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(L
OG
RE
AL
GD
P)
Re
sp
on
se
to
Ch
ole
sky O
ne
S.D
. In
no
va
tio
ns ±
2 S
.E.
Figure A5.Dynamic responses of
output and liquidliabilities
179
Does moneymatter in Africa?
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(DB
AC
BA
) to
D(D
BA
CB
A)
12
34
56
78
91
0
Re
sp
on
se
of D
(DB
AC
BA
) to
D(L
OG
RE
AL
GD
P)
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
−0
.02
0.0
0
0.0
2
0.0
4
0.0
6
0.0
8
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(D
BA
CB
A)
12
34
56
78
91
0
Re
sp
on
se
of D
(LO
GR
EA
LG
DP
) to
D(L
OG
RE
AL
GD
P)
Re
sp
on
se
to
Ch
ole
sky O
ne
S.D
. In
no
va
tio
ns ±
2 S
.E.
Figure A6.Dynamic responses ofOutput and financial size
IGDR7,2
180
Dow
nloa
ded
by M
r A
song
u Si
mpl
ice
At 0
2:30
13
Nov
embe
r 20
14 (
PT)
