Analysis of Islamic banks’financing and economic growth:a panel cointegration approach

Yazdan Gudarzi Farahani and Masood DastanDepartment of Economics, University of Tehran, Tehran, Iran

Abstract

Purpose – This paper seeks to use empirical evidence to examine the role of Islamic banks’ financingon economic performance of selected countries (Malaysia, Indonesia, Bahrain, UAE, Saudi Arabia,Egypt, Kuwait, Qatar and Yemen).

Design/methodology/approach – Using quarterly data (2000:1-2010:4), this paper utilizes thepanel cointegration approach models framework.

Findings – The results generally signify that, in the long run, Islamic banks’ financing is positiveand significantly correlated with economic growth and capital accumulation in these countries. Theresults obtained from the Granger causality test reveal a positive and statistically significantrelationship between economic growth and Islamic banks’ financing in the short run and in the longrun. It also found that the long run relationship is stronger than the short run relationship.

Originality/value – This paper uses empirical evidence to show the effect of Islamic banks’financing on economic growth of selected Islamic countries. To the best of the authors’ knowledge,most of the studies in this field have applied the bound testing approach of cointegration, errorcorrection models (ECMs), Auto Regressive Distributed lag (ARDL) and Vector Autoregressive Model(VAR), and the coefficients obtained by these models cannot be deemed as a general finding applicablefor other countries. The superiority of this article is in applying the FMOLS model, which has stableand consistent coefficients and is also a dynamic model.

Keywords Islamic banking, Economic growth, Panel cointegration analysis, Financial analysis

Paper type Research paper

1. IntroductionToday, Islamic banks exist in all parts of the world, and are looked upon as a viablealternative system which has many things to offer. While it was initially developed tofulfill the needs of Muslims, Islamic banking has now gained universal acceptance.Islamic banking is recognized as one of the fastest growing areas in banking and finance.Since the opening of the first Islamic bank in Egypt in 1963, Islamic banking hasgrown rapidly all over the world. The number of Islamic financial institutionsworldwide has risen to over 300 today in more than 75 countries mainly concentrated inthe Middle East and Southeast Asia (with Bahrain and Malaysia are the biggest hubs).However, Islamic banks are also appearing in Europe and the USA. The Islamic bankingtotal assets worldwide are estimated to have exceeded $250 billion and are growing at anestimated pace of 15 percent a year (Sufian et al., 2008).

The theory of Islamic banking is based on the concept that interest is strictlyforbidden in Islam, and that Islamic teachings provide the required guidance on which tobase the working of banks. The basic principle that has guided all theoretical work onIslamic banking is that although interest is forbidden in Islam, trade and profit isencouraged.

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/1753-8394.htm

International Journal of Islamic andMiddle Eastern Finance andManagementVol. 6 No. 2, 2013pp. 156-172q Emerald Group Publishing Limited1753-8394DOI 10.1108/17538391311329842

IMEFM6,2

156

Islamic banking is a financial system whose key aim is to fulfill the teachings of theHoly Quran. Islamic law reflects the commands of God and this law regulates all theaspects of a Muslim’s life and hence Islamic finance is directly involved with spiritualvalues and social justice.

El-Ghattis (2011), stats that:

[. . .]the basic principle in Islamic law is that exploitative contracts or unfair contracts thatinvolve risk or speculation are impermissible. Under Islamic banking, all partners involved infinancial transactions share the risk and profit or loss of a venture and no one gets apredetermined return. This direct correlation between investment and profit is the keydifference between Islamic and conventional banking which its main objective is maximizingshareholders’ wealth (Dar and Presley, 2000).

El-Ghattis (2011), stats that Islamic banking has introduced itself as an emergingalternative to conventional banking system and has grown rapidly over the last twodecades both in Muslim and non-Muslim countries. Islamic banks have recorded highgrowth rates in both size and number and have operated in over 60 countries worldwideand bankers predict that Islamic banking would control over 50 percent of savings in theIslamic countries within the next decade (Ahmad, 2004; Abduh and Azmi Omar, 2012).

Recent articles and theoretical papers have called on economies to consider Islamiceconomic theories as an alternative solution to the current capitalist system. There arealso numerous good and well-organized papers respecting Islamic banking systemwhich attempt to clarify the effects of Islamic banking on economic growth incomparison to the effects of conventional banking on it. However, most of empiricalstudies conducted in this field were not able to explain the overall effect of Islamic bank’sfinancing on economy, due to the fact that they consider a single-country sample.

Islamic banking activities can be classified into two groups: in one group, theiractivities are without any competition with conventional banking based on interestrate, due to the domestic laws and regulations of some Islamic countries which do notallow any activities based on interest rate (ribah) for financial institutions, banks inthose countries. In the second group, there is a high competition between these twobanking systems which is because of this fact that they are operating in non-Muslimcountries or Muslim countries which do not forbid interest rate-based banking system.

Out of the extensive research carried out in this field, there are not sufficient worksconducted within the Islamic finance framework. The objective of this paper, therefore,is to narrow the gap between literature and practice by examining the short-run and thelong-run relationships between Islamic financial development and economic growth inthe selected Islamic countries, using full modified ordinary least square (FMOLS).However, most of studies in this field has applied the bound testing approach ofcointegration, error correction models (ECMs), auto regressive distributed lag (ARDL)and vector autoregressive model (VAR) which coefficients obtained by these modelscannot be deemed as a general finding applicable for other countries. The superiority ofour article is in applying FMOLS model that has stable and consistent coefficients and isalso a dynamic model. Also, with regard to previous studies, our paper includes morecountries which increase the reliability of results.

This study is guided by the following research questions:

RQ1. What is the relationship between Islamic financial development and economicgrowth in these selected countries?

Financing andeconomic growth

157

RQ2. Does Islamic financial development have significant effect on these countries’economic growth in the short-run and in the long-run?

This paper uses the nine selected Islamic countries’ information based on dataavailability and compatibility (in all of these countries both Islamic and conventionalbanking systems prevail) to test the validity of the theoretical findings. Accordingly,the following hypotheses are considered:

H1. There is a long-run relationship between Islamic financial development andeconomic growth.

H2. Islamic financial development leads to economic growth.

H3. Economic growth leads to Islamic financial development.

In this paper, we use Abduh and Azmi Omar (2012) article “The relationship betweenIslamic financial development and economic growth for Asian countries” and Furqaniand Mulyany (2009) article “Islamic banking and economic growth: empirical evidencefrom Malaysia”.

In this paper the relationship between Islamic financing and economic growth istested by using panel data of nine selected Islamic countries over the period 2000-2010.This paper consists of four sections. Section 1, discusses the introduction, in which thebackground and rationale of the study is outlined. Section 2, covers the review ofliterature, the relationship between Islamic banking and economic growth. Section 3,covers the details of the data and research methodology employed in this study andreports the findings and discussions. The final section contains the conclusions.

2. Literature reviewIn recent years a number of both theoretical and empirical studies about the Islamicbanking have been published. Most of these studies have argued that Islamic bank’sproducts must submit to principles set by Quran and this prevents the emergence ofspeculative demand for money. Islamic banks also use profit and loss sharing (PLS)criterion. These advantages help the Islamic banks to cope with instability in the economy.

Abduh and Azmi Omar (2012) examines the short-run and the long-run relationshipsbetween Islamic banking development and economic growth in Indonesia. For thispurpose they use quarterly data (2003:1-2010:2), they utilizes the bound testing approachof cointegration and ECMs, developed within an ARDL framework. Their results showa significant relationship in short-run and long-run periods between Islamic financialdevelopment and economic growth. The relationship, however, is neither Schumpeter’ssupply-leading nor Robinson’s demand-following. It appears to be bidirectionalrelationship.

Furqani and Mulyany (2009) examine the dynamic interaction between Islamicbanking and economic growth in Malaysia by employing the cointegration test andvector ECM. Their results show that in the short-run, only fixed investment is thegranger cause of Islamic banking for period 1997:1-2005:4. While in the long-run, there isevidence of a bidirectional relationship between Islamic banking and fixed investmentand also there is evidence supporting demand following hypothesis of gross domesticproduct (GDP) and Islamic banking, where increases in GDP causes Islamic banking todevelop and not vice versa.

IMEFM6,2

158

Dahduli (2009) investigates the effectiveness of Islamic banking and financeconcepts and concludes that in addition to the traditional prohibitions, Islamic banksshould be aware of the immoral defrauding and selfish sales behavior that took place inWestern financial institutions, which only protects banks’ interests and the bonusespaid to management. In addition and through discussing developed countries bankingpractices in the nineteenth century, it has been concluded that the Islamic PLScontracts would have helped accelerate economic growth by its force of directing fundsinto profitable industries that increases banks’ cash flows and market confidence.

Mirakhor and Khan (1990) point out that reliance on profit sharing arrangementsmakes the Islamic system akin to an equity-based system, relatively straightforwardtheoretical models that have been developed for analyzing the working of the system. Inthese models, depositors are treated as shareholders (as in a mutual fund or investmenttrust, for example) and bank provides no guarantee on the rate of return or nominal valueof shares. Symmetrically, banks themselves become partners with the borrowers andaccordingly share in the returns obtained from the borrowed funds. They add that aninteresting result emerging from such models is that the Islamic system may be bettersuited than an interest-based system to adjust to shocks that can lead to banking crises.

Zangeneh (1995) discusses that in a loan-based economy instability can arise from twoseparate sources. One source of instability is refinancing and refinancing of businessinvestments. As long as markets are moving in the right direction loans are available onreasonable terms. But as soon as markets turn against expectations, especially againstsurplus spending units’ expectations, refinancing of the old debt and financing newprojects will prove to be difficult and such activities will have repercussions for theeconomy. This cannot happen in an equity-based economy. A second source of instabilityin a loan-based economy comes from the demand for real balances. Zangeneh concludesthat in equity-based economy, the speculative demand for money does not exist. As aresult, a major source of instability in the demand for real balances on the economy iseliminated since in Islamic banking profit sharing schemes are not applicable. Onepossible solution for purely personal loans is the use of zakat (wealth tax) to provideinterest free loans to those who need financial assistance (Hassani, 2010).

Mishkin (2006) posits that indirect finance, which involves the activities of financialintermediaries, is many times more important than direct finance, in which businessesraise funds directly from lenders in financial markets, towards economic growth. For theperiod of 1970-1996, for example, sources of external funds of non-financial businesses inJapan were 85 percent from bank loans and 15 percent from financial markets while inGermany were almost 80 percent from bank loans and the rest from financial markets(Mishkin, 2006).

Referring to the importance of financial development for a country, study on causalrelationship between the development of financial intermediaries’ activities andeconomic growth has been carried out extensively. Among the seminal works done inthis field is a study by McKinnon (1973), Shaw (1973), King and Levine (1993),Demetriades and Hussein (1996), Levine et al. (2000), Beck et al. (2000), Beck and Levine(2004) and recently Shen and Lee (2006).

King and Levine (1993) for instance, studied this issue using data from 80 countriesover the 1960-1989 periods. They constructed four indicators of the level of financialsector developments which is regressed with the real GDP per capita and its sources.First is “financial depth” which equals the ratio of liquid liabilities of the financial system

Financing andeconomic growth

159

to GDP. Second is the ratio of deposit money bank domestic assets to deposit moneybank deposit assets plus central bank domestic assets to measure the relativeimportance of specific financial institutions. The third and fourth financial developmentindicators are designed to measure domestic asset distribution.

In spite of the rather technical nature of their criteria, the data set contains countrieswith rich experiences in relation to both economic and financial development. All ofthese countries, however, displayed some evidence of reverse causation so that therelationship between financial development and growth appears to be bidirectional.Deidda and Fattouh (2002) and Rioja and Valev (2002) claim that there is no significantrelationship between financial depth and economic growth in countries with low incomeper capita.

The significant relationship only appears in the high income countries. Some studieshave taken a more microeconomic approach and some used stock markets as the proxyfor financial development. For example, Fisman and Love (2003) revisited an earlierpaper by Rajan and Zingales (1998) by re-examining their assumptions, and therobustness of their results to alternative theories and interpretations. The result issupporting the hypothesis that financial development helps industries with goodgrowth opportunities. It also reinforces their hypothesis that the role of financialdevelopment is to reallocate resources to industries that have good growth opportunitiesand not to industries with “technological dependence” on external finance.

Another study by Beck and Levine (2004) investigates the impact of stock marketsand banks on economic growth using a panel data set for the period 1976-1998. Theresults strongly reject the notion that overall financial development is unimportant orharmful for economic growth. Therefore, they argue that stock markets and bankspositively influence economic growth.

With regard to the role of Islamic financial development in economic growth, Furqaniand Mulyany (2009) and Majid and Kassim (2010) are among the limited articles in thisarea. However, using not-so-different time span of quarterly data, their findings aredifferent in terms of the direction of the relationship. Furqani and Mulyany (2009), on theone hand, posit that the relationship between Islamic financial development andeconomic growth is following the view of “demand-following” which means thateconomic growth causes Islamic banking institutions to change and develop. On theother hand, findings of Majid and Kassim (2010) are in favor of the supply-leading view.

3. Data and methodologyPanel data provide a large number of point data, increasing the degrees of freedom andreducing the collinearity between regressors. Therefore, it allows for more powerfulstatistical tests and normal distribution of test statistics. It can also take heterogeneityof each cross-sectional unit into account, and give “more variability, less collinearityamong variables, more degrees of freedom, and more efficiency” (Baltagi, 2001).

In this paper, regressions are based on data concerning a group of nine Islamiccountries over the period 2000-2010. Data for real GDP growth (GDP), gross fixedcapital formation (GFCF), total Islamic bank’s financing (FIN) and trade activities thatinvolve export plus import (control variable) (TRADE) for nine Islamic countriesnamely Malaysia, Indonesia, Bahrain, UAE, Saudi Arabia, Egypt, Kuwait, Qatar andYemen come from the World Bank Statistics and International Financial Statisticspublished by the International Monetary Fund (IMF).

IMEFM6,2

160

Countries have been selected based on data availability. Also, in these countries bothIslamic and conventional banking systems exist. One of the main reasons for choosingthese countries is similarity in their financial markets which the use of homogeneousdata leads to more accurate coefficients and more consistent results.

GDP is a common statistic for representing the income level of a particular countrywithin a certain time range. Study about finance-growth nexus always use GDP as theprincipal variable reflecting economic growth. We use GFCF as a representation ofinvestment in order to measure net new investment during an accounting period. It is tobe noted that the financing variable applied in this model is a portion of total financing inthe economy provided by Islamic banks.

3.1 Model specificationIn this paper we pool cross-section and time series data to study relationships betweenIslamic financing, GDP and fixed capital formation. We get the following equation:

GDP ¼ f ðFIN ;GFCF;TRADEÞ ð1Þ

The empirical model form for this specification is given by:

GDPit ¼ b0 þ b1Finit þ b2GFCFit þ b3TRADEit þ 1it ð2Þ

where FIN, GFCF, TRADE and GDP are as defined earlier in equation (1). The b0 isa constant term andb1 tob3 are estimated parameters in the model and i is a cross-sectiondata for countries referred to, and t is a time series data and 1it is an error term.

3.2 Estimation procedureIn order to investigate the possibility of panel cointegration, first, it is necessary todetermine the existence of unit roots in the data series. For this study we have chosen theIm, Pesaran and Shin (IPS, 1997), which is based on the well-known Dickey-Fullerprocedure.

Im, Pesaran and Shin denoted IPS proposed a test for the presence of unit roots inpanels that combines information from the time series dimension with that from thecross-section dimension, such that fewer time observations are required for the test tohave power. Since researchers have found the IPS test to have superior test power foranalyzing long-run relationships in panel data, we will also employ this procedure inthis study. IPS begins by specifying a separate augmented Dickey-Fuller (ADF)regression for each cross-section with individual effects and no time trend:

Dyit ¼ ai þ riyi;t1 þXpij¼1

bijDyi;tj þ 1it ð3Þ

where i ¼ 1, . . . , N and t ¼ 1, . . . ,T.IPS use separate unit root tests for theN cross-section units. Their test is based on the

ADF statistics averaged across groups. After estimating the separate ADF regressions,the average of the t-statistics for p1 from the individual ADF regressions, tiTi

( pi):

�tNT ¼1

N

XNi¼1

tiT ð pibiÞ ð4Þ

Financing andeconomic growth

161

The t-bar is then standardized and it is shown that the standardized t-bar statisticconverges to the standard normal distribution as N and T ! 1. IPS (1997) showed thatt-bar test has better performance when N and T are small. They proposed across-sectional demeaned version of both test to be used in the case where the errors indifferent regressions contain a common time-specific component (Hussin et al., 2010).

3.3 Panel cointegration testsThe next step is to test for the existence of a long-run relationship among Islamicfinancing (FIN), GDP, GFCF and trade (TRADE). A common practice to test forcointegration is Johansen’s procedure. However, the power of the Johansen test inmultivariate systems with small sample sizes can be severely distorted. To this end, weneed to combine information from time series as well as cross-section data once again.In this context three panel cointegration tests are conducted.

First, we use a test due to Levine and Lin (1993) in the context of panel unit roots, toestimate residuals from (supposedly) long-run relations. Levine and Lin (1993) considerthe model:

yit ¼ ri yi;t21 þ z 0itgþ uit ð5Þ

where zit are deterministic variables, uit is iid(0, s 2) and ri ¼ r. The test statistic is att-statistic on r given by:

tr ¼ðr2 1Þ

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPNi¼1

PTt¼1~y

2i;t21

qse

ð6Þ

where:

~yit ¼ yit 2XTs¼1

hðt; sÞyis; ~uit ¼ uit 2XTs¼1

hðt; sÞuis hðt; sÞ ¼ z 0t

XTt¼1

ztz0t

!zs;

s2e ¼ ðNTÞ21

XNi¼1

XTt¼1

~u2it;

And r is the OLS estimate of r. It can be shown that if there are only fixed effects in themodel, then: ffiffiffiffi

Np

Tðr2 1Þ þ 2ffiffiffiffiN

p! N 0;

91

9

� �And if there are fixed effects and a time trend:

ffiffiffiffiN

pðTðr2 1Þ þ 9:5Þ! N 0;

3425

136

� �

Second, we use the unit root tests developed for equation (4) by Harris and Tzavalis(1999). If there are only fixed effects in the model, then:

ffiffiffiffiN

pr2 1 þ

8

T þ 1

� �! N 0;

8ð18T 2 2 23T þ 18Þ

10ðT 2 1ÞðT þ 1Þ3

� �

IMEFM6,2

162

If there are fixed effects and a time trend, then:

ffiffiffiffiN

pr2 1 þ

19

5ðT þ 5Þ

� �! N 0;

19ð233T 2 2 849T þ 1528Þ

130ðT þ 5Þ3ðT 2 5Þ

� �

It must be noted that Levine and Lin (1993) tests may have substantial size distortion ifthere is cross-sectional dependence (O’Connell, 1998). Also, Harris and Tzavalis (1999)find that small T yields Levine and Lin tests which are substantially undersized andhave low power. A drawback of the Levine and Lin or Harris and Tzavalis tests is thatthey do not allow for heterogeneity in the autoregressive coefficient, r.

Finally, to overcome the problem of heterogeneity that arises in both tests we useFisher’s test to aggregate the p-values of individual Johansen maximum likelihoodcointegration test statistics (Maddala and Kim, 1998). If pi denotes the p-value of theJohansen statistic for the ith unit, then we have the result 22

PNi¼1log pi , x2

2N . Thetest is easy to compute and, more importantly, it does not assume homogeneity ofcoefficients in different countries (Christopoulos and Tsions, 2004).

The next step is to test for the existence of a long-run cointegration GDP andthe independent variables using panel cointegration tests suggested by Pedroni(1999, 2004). We will make use of seven panel cointegration by Pedroni (1999), since hedetermines the appropriateness of the tests to be applied to estimated residuals from acointegration regression after normalizing the panel statistics with correction terms(Hussin et al., 2010).

The procedures proposed by Pedroni make use of estimated residual from thehypothesized long-run regression of the following form:

yi;t ¼ ai þ dit þ b1ix1i;t þ b2ix2i;t þ · · · þ bMixMi;t þ 1i;t ð7Þ

For t ¼ 1, . . . , T; I ¼ 1, . . . , N; m ¼ 1, . . . , M.Where T is the number of observations over time, N number of cross-sectional units

in the panel, and M number of regressors. In this set up, ai is the member specificintercept or fixed effects parameter which varies across individual cross-sectionalunits. The same is true of the slope coefficients and member specific time effects, dit.

Pedroni (1999, 2004) proposes the heterogeneous panel and heterogeneous groupmean panel test statistics to test for panel cointegration. He defines two sets ofstatistics. The first set of three statistics Zv;N ;T , Z r;N ;T 1 and ZtN, T are based on poolingthe residuals along the within dimension of the panel. The statistics are as follows:

Z n;N ;T ¼ T 2N 3=2XNi¼1

XTt¼1

L2

11i e2i;t 1 ð8Þ

Z r;N ;T 1 ¼ TffiffiffiffiN

p XNi¼1

XTt¼1

L2

11i e2i;t 1

XNi¼1

XTt¼1

L2

11iðei;t 1 Dei;t l iÞ ð9Þ

ZtN ;T ¼ ~s 2N ;T

XNi¼1

XTt¼1

L2

11i e2i;t 1

1=2 XNi¼1

XTt¼1

L2

11iðei;t 1 Dei;t l iÞ ð10Þ

where ei;t 1 is the residual vector of the OLS estimation of equation (5) and where theother terms are properly defined in Pedroni.

Financing andeconomic growth

163

The second set of statistics is based on pooling the residuals along the betweendimension of the panel. It allows for a heterogeneous autocorrelation parameter acrossmembers. The statistics are as follows:

~ZrN ;T 1 ¼XNi¼1

XTt¼1

e2i;t 1

1XTt¼1

ðei;t 1 Dei;t liÞ ð11Þ

~ZtN ;T 1 ¼XNi¼1

XTt¼1

e2i;t 1

1=2XTt¼1

ðei;t 1 Dei;t liÞ ð12Þ

These statistics compute the group mean of the individual conventional time seriesstatistics. The asymptotic distribution of each of those five statistics can be expressedin the following form:

XN ;T mffiffiffiffiN

p

ffiffiffiffiV

p ! N ð0; 1Þ ð13Þ

where XNT is the corresponding form of the test statistics, while m and v are the meanand variance of each test, respectively. They are given in Table II in Pedroni (1999).Under the alternative hypothesis, panel v-statistics diverges to positive infinity.Therefore, it is a one sided test were large positive values reject the null of nocointegration. The remaining statistics diverge to negative infinity, which means thatlarge negative values reject the null (Al-Awad and Harb, 2005).

3.4 FMOLS estimationIn this section we adopt FMOLS procedure from Christopoulos and Tsions (2004). Inorder to obtain asymptotically efficient and consistent estimates in panel series,non-exogeneity and serial correlation problems are tackled by employing fully modifiedOLS (FMOLS) introduced by Pedroni (1996). Since the explanatory variables arecointegrated with a time trend, and thus a long-run equilibrium relationship existsamong these variables through the panel unit root test and panel cointegration test,we proceed to estimate the equation (2) by the method or FMOLS for heterogeneouscointegrated panels. This methodology allows consistent and efficient estimation ofcointegration vector and also addresses the problem of non-stationary regressors, aswell as the problem of simultaneity biases. It is well known that OLS estimation yieldsbiased results because the regressors are endogenously determined in the I(1) case.Following cointegrated system for panel data:

yit ¼ ai þ x0itbþ uit ð14Þ

xit ¼ xi;t21 þ eit ð15Þ

where jit ¼ [uit, e0it] is stationary with covariance matrix Vi. Following Phillips and

Hansen (1990) a semi-parametric correction can be made to the OLS estimator thateliminates the second order bias caused by the fact that the regressors are endogenous.Pedroni (2000) follows the same principle in the panel data context, and allows for theheterogeneity in the short-run dynamics and the fixed effects. Pedroni’s estimator is(Christopoulos and Tsions, 2004):

IMEFM6,2

164

bFM 2 b ¼XNt¼1

V22

22i

XTt¼1

ðxit 2 �xtÞ2

!21

·XNi¼1

V21

11i V21

22i

XTt¼1

ðxit 2 �xtÞu*it 2 Tgi

!

ð16Þ

u*it ¼ uit 2 V21

22i V21i; gi ¼ G21i þ V0

21i 2 V21

22i V21i G22i þ V0

22i

� �ð17Þ

where the covariance matrix can be decomposed as Vi ¼ V0i þ Gi þ Gi where V0

i is thecontemporaneous covariance matrix, andGi is a weighted sum of autocovariances. Also,

V0

i denotes an appropriate estimator of V0i .

In this study, we employed panel group FMOLS test from Pedroni (1996, 2000). Animportant advantage of the panel group estimators is that the form in which the data ispooled allows for greater flexibility in the presence of heterogeneity of the cointegratingvectors. Test statistics constructed from the panel group estimators are designed to testthe null hypothesis H0: bi ¼ b0 for all I against the alternative hypothesis HA: bi – b0,so that the values for bi are not constrained to be the same under the alternativehypothesis. Clearly, this is an important advantage for applications such as the presentone, because there is no reason to believe that if the cointegrating slopes are not equal toone, which they necessarily take on some other arbitrary common value. Anotheradvantage of the panel group estimators is that the point estimates have a more usefulinterpretation in the event that the true cointegrating vectors are heterogeneous.Specifically, point estimates for the panel group estimator can be interpreted as the meanvalue for the cointegrating vectors (Hussin et al., 2010).

3.5 Empirical resultTable I presents the results of the IPS panel unit root test at level indicating that allvariables are I(0) in the constant of the panel unit root regression. These results clearlyshow that the null hypothesis of a panel unit root in the level of the series cannot berejected at various lag lengths. We assume that there is no time trend. Therefore, wetest for stationarity allowing for a constant plus time trend. In the absence of a constantplus time trend, again we found that the null hypothesis of having panel unit root is

Level First order differenceVariable Constant Constant þ trend Constant Constant þ trend

FIN 20.872 21.832 23.538 23.648(0.891) (0.880) (0.000) (0.001)

GFCF 21.315 21.601 23.630 23.783(0.623) (0.904) (0.000) (0.002)

GDP 21.190 21.891 23.183 23.838(0.101) (0.485) (0.000) (0.002)

TRADE 21.201 21.911 22.053 22.102(0.352) (0.503) (0.032) (0.015)

Notes: Levels and first order differences denote the IPS t-test for a unit root in levels and firstdifferences, respectively; number of lags was selected using the AIC criterion; italic values denotesampling evidence in favour of unit roots; we use the Eviews software to estimate this valueSource: World Bank Statistics and IMF (2012)

Table I.Panel unit root test – IPS

Financing andeconomic growth

165

generally rejected in all series at level form and various lag lengths. We can concludethat most of the variables are non-stationary in with and without time trendspecifications at level by applying the IPS test which is also applied for heterogeneouspanel to test the series for the presence of a unit root. The results of the panel unit roottests confirm that the variables are non-stationary at level.

Table I also presents the results of the tests at first difference for IPS test in constantand constant plus time trend. We can see that for all series the null hypothesis of unit roottest is rejected at 95 percent critical value. Hence, based on IPS test, there is strongevidence that all the series are in fact integrated of orders one.

We can conclude that the results of panel unit root tests reported in Table I supportthe hypothesis of a unit root in all variables across countries, as well as the hypothesis ofzero order integration in first differences. At most of the 1 percent significance level, wefound that all tests statistics in both with and without trends significantly confirm thatall series strongly reject the unit root null. Given the results of IPS test, it is possible toapply panel cointegration method in order to test for the existence of the stable long-runrelation among the variables.

The next step is to test whether the variables are cointegrated using Pedroni’s (1999,2001, 2004). This is to investigate whether long-run steady state or cointegration existamong the variables and to confirm of what Coiteux and Olivier (2000) state that thepanel cointegration tests have much higher testing power than conventionalcointegration test. Since the variables are found to be integrated in the same orderI(1), we continue with the panel cointegration tests proposed by Pedroni (1999, 2001,2004). Cointegration are carried out for constant and constant plus time trend and thesummary of the results of cointegration analyses are presented in Table II.

In constant level, we found that three out of seven statistics reject null by hypothesisof no cointegration at the 5 percent level of significance for the ADF-statistic and groupr-statistic, while the group-adf is significant at 1 percent level. The results of the panelcointegration tests in the model with constant level show that independent variables dohold cointegration in the long-run for Islamic countries with respect to GDP. In the panelcointegration test for our model with constant plus trend level, the results indicate thatfour out of seven statistics reject the null hypothesis of no cointegration at the 1 and5 percent level of significance. It is shown that independent variables do holdcointegration in the long-run for Islamic countries with respect to GDP.

Test Constant trend Constant þ trend

Panel v-statistic 20.070 20.921Panel r-statistic 21.120 22.012Panel t-statistic: (non-parametric) 21.131 21.302Panel t-statistic (adf ): (parametric) 22.425 22.334Group r-statistic 22.021 22.830Group t-statistic: (non-parametric) 20.735 21.427Group t-statistic (adf ): (parametric) 23.329 23.724

Notes: All statistics are from Pedroni’s (1999) procedure where the adjusted values can be comparedto the N(0,1) distribution; the Pedroni (2004) statistics are one-sided tests with a critical value of 21.64(k , 21.64 implies rejection of the null), except the v-statistic that has a critical value of 1.64 (k . 1.64suggests rejection of the null); we use the Eviews software to estimate this valueSource: World Bank Statistics and IMF (2012)

Table II.The Pedroni panelcointegration test

IMEFM6,2

166

In Table III, we found that the estimate of the coefficient for FIN is positive andstatistically significant at the 1 percent level for all countries, except for Yemen that thisstatistic is significant at the 5 percent level. We conclude that there is a long-runrelationship between economic growth and Islamic banking for these counties. Thecoefficient on GFCF is positive and statistically significant at the 1 percent level for allcountries, except for Yemen and Kuwait that this statistic is significant at the 5 percentlevel. These results show that there is still a long-run cointegration between GFCF andeconomic growth (GDP). Being insignificant according to Table III, trade coefficient isomitted from the model.

The next step is to estimate the Granger causality model with a dynamic errorcorrection:

DGDPit ¼ a1j þXki¼1

b11ikDGDPit2k þXki¼1

b12ikDFinit2k þXki¼1

b13ikDGFCFit2k

þ b14ikDTRADEit2k þ d1i ECTit21 þ u1it

ð18Þ

DFinit ¼a2j þXki¼1

b21ik DGDPit2k þXki¼1

b22ik DFinit2k þXki¼1

b23ik DGFCFit2k

þ b24ik DTRADEit2k þ d2i ECTit21 þ u2it

ð19Þ

DGFCFit ¼a3jþXki¼1

b31ikDGDPit2kþXki¼1

b32ikDFinit2kþXki¼1

b33ikDGFCFit2k

þb34ikDTRADEit2kþd3i ECTit21 þu3it

ð20Þ

where D denotes first differencing and k is the lag length and is chosen optimally foreach country using a step-down procedure up to a maximum of two lags. Also, thetrade equation is omitted, because it is not a relevant variable in the model and themain aim of using it was just as a control variable.

Country FIN GFCF

Indonesia 4.12 (3.86) 3.21 (3.82)Malaysia 5.31 (2.54) 2.91 (2.17)Qatar 3.58 (2.03) 3.37 (2.45)Yemen 2.34 * (2.53) 1.28 * (2.14)UAE 5.25 (3.20) 3.30 (2.45)Saudi Arabia 5.72 (4.08) 3.01 (3.73)Bahrain 4.19 (3.37) 2.89 (4.20)Kuwait 3.89 (2.31) 1.98 * (3.10)Egypt 4.64 (4.35) 2.80 (2.87)Panel group 2.19 (2.17) 1.18 * (2.97)

Notes: Statistically significant at: *5 percent level; the null hypothesis for the t-ratio is H0 ¼ bi ¼ 0;figures in parentheses are t-statistics; we use the Eviews software to estimate this valueSource: World Bank Statistics and IMF (2012)

Table III.FMOLS regression

Financing andeconomic growth

167

The sources of causation can be identified by testing for the significance of thecoefficients of the dependent variables in equations (18)-(20). First, the short-run effectcan be considered transitory. For short-run causality, we can test H0: b12ik ¼ 0 for all iand k in equation (18) or H0: b21ik ¼ 0 for all i and k in equation (19). Finally in thecapital equation (20), short-run causality from GDP, Fin and TRADE to capital aretested, respectively, based on H0: b31ik ¼ 0;ik, H0: b32ik ¼ 0;ik and H0: b33ik ¼ 0.Next, the long-run causality can be tested by looking at the significance of the speed ofadjustment d, which is the coefficient of the error correction term, ECTit21. Thesignificance of k indicates the long-run relationship of the cointegrated process, and somovements along this path can be considered permanent. For long-run causality, wecan test H0: d1i ¼ 0 for all i in equation (18), H0: d2i ¼ 0 for all i in equation (19) orH0: d3i ¼ 0 for all i in equation (20). Finally, we can use the joint test to check for astrong causality test, where variables bear the burden of a short-run adjustment tore-establish a long-run equilibrium, following a shock to the system (Asafu-Adjaye,2000; Lee, 2005; Abbasinejad et al., 2012).

Because all variables enter the model in stationary form, a standardF-test can be usedto test the null hypothesis, which shows that all of the estimated country-specificparameters are significant. Table IV shows the result of a panel causality test betweenGDP, Islamic banks’ financing, GFCF and trade. An examination of the sum of thelagged coefficients on the respective variables indicates that Islamic banks’ financing(0.02) has a statistically more significant impact on GDP than real gross capitalformation (0.04). Moreover, the error correction term is statistically significant at the5 percent level denoting a relative high speed of adjustment to long-run equilibrium. Interms of equation (18), it appears that Islamic banks’ financing, real gross capitalformation and trade have a statistically significant impact on GDP. Real gross capitalformation has a positive and statistically significant impact on Islamic banks’ financing,indicative of their complementarily. However, the error correction term is statisticallysignificant which suggests that Islamic banks’ financing is responsive to adjustmentstowards long-run equilibrium. Based on results obtained from equation (19), it is notsurprising that GDP, real gross capital formation and trade have positive andstatistically significant impacts on Islamic banks’ financing. In regards to equation (20),GDP, Islamic banks’ financing and trade have positive and statistically significantimpacts on real gross capital formation in the short-run and the long-run.

Source of causation (independent variable)Short-run Long-run

Dependent variable DGDP DFin DGFCF 1 1/DGDP 1/DFin 1/DGFCF 1/DTRADE

DGDP – 0.98 1.98 3.14 – 2.84 2.17 1.21(0.02) (0.04) (0.00) (0.01) (0.02) (0.01)

DFin 2.15 – 2.04 2.76 2.23 – 2.59 0.88(0.03) (0.02) (0.01) (0.01) (0.00) (0.01)

DGFCF 3.26 2.26 – 3.15 2.32 2.67 – 1.02(0.00) (0.00) (0.03) (0.01) (0.00) (0.00)

Note: p-value in parenthesis; we use the Eviews (IHS Inc., Econometrics Software, USA) to estimatethis valueSource: World Bank Statistics and IMF (2012)

Table IV.Panel causality tests

IMEFM6,2

168

According to F-statistic in Table IV, the results reveal bidirectional Granger causalitybetween mentioned variables. However, the long-run bidirectional causality betweenthese variables is more significant than the short-run causality between them. As aresult, there is a bidirectional causality between Islamic bank’s financing and economicgrowth in which the impact of Islamic bank’s financing on economic growth is higherthan the reciprocal effect of economic growth on Islamic bank’s financing.

4. ConclusionThis paper is an empirical study on the relationship between Islamic banking andeconomic growth. For that reason we use the panel cointegration approach. The unit roottest (IPS) is used to confirm the stationarity of all variables before the cointegration testcan be performed. After confirming that all variables are non-stationary at level, thepanel cointegration approach is applied. Using Pedroni’s, the long-run cointegration testis performed to investigate the existence of the long-run cointegration among thevariables. Results obtained indicate the presence of the long-run and the short-runrelationship between economic growth and Islamic financing for nine selected Islamiccountries. In the short-run and in the long-run, Islamic banks’ financing is found to beGranger cause of economic growth, and vice versa. The results of a bidirectionalrelationship in the short-run and in the long-run show that Islamic banks’ financingleads economic growth. Our results support that current as well as past changes inIslamic banks’ financing have significant impact on the changes in income in thesecountries. It is clear for these countries in general that in short-run Islamic banks’financing is an important ingredient for economic development. The results generallyshow that in the long-run, Islamic bank’s financing is positive and significantlycorrelated with economic growth and capital accumulation of these countries. In thisregard, Islamic banking has effectively played its main role as financial intermediariesthat facilitate the transmission of saving from surplus households to deficit households.Thus, we can argue that the current policies of these countries’ banks to develop acomprehensive Islamic financial system are strongly linked to economic growth.Furthermore, results show the reliability and contribution of Islamic banking to the realeconomic sectors of these countries specifically economic growth and capitalaccumulation. These results reveal that improvement of the Islamic financial systemin these countries may benefit economic development and it is important in the long-runfor economic welfare.

This interesting finding provides a number of implications. First, these countries’government should continue to promote Islamic banking as it has shown to benefit theeconomy. This can be done by setting a target ratio of Islamic banking assets to totalbanking assets to be achieved by a certain year as was done in these countries. As acorollary to this, the governments need to encourage and promote the establishment ofmore Islamic commercial banks, Islamic windows, and Islamic rural banks whilst at thesame time encourage existing Islamic banks to establish more branches. In addition,allowing foreign Islamic banks to operate in these countries can also help to foster moreinnovation in the domestic Islamic banking industry. Second, as the number of Islamicbanks and Islamic financial institutions increases, there is also a need to have sufficientskill manpower to manage these institutions. Consequently, there is a need to relook atthe current regulations and guidelines in order to bring it at par with the development ofIslamic banking worldwide.

Financing andeconomic growth

169

References

Abbasinejad, H., Gudarzi Farahani, Y. and Asghari Ghara, E. (2012), “The relationship betweenenergy consumption, energy prices and economic growth: case study (OPEC countries)”,OPEC Energy Review, Vol. 36, pp. 272-286.

Abduh, M. and Azmi Omar, M. (2012), “Islamic banking and economic growth: the Indonesianexperience”, International Journal of Islamic and Middle Eastern Finance andManagement, Vol. 5, pp. 35-47.

Ahmad, A. (2004), “Economic development in Islamic perspective revisited”, Journal of KAU:Islamic Economies, Vol. 17 No. 1, pp. 53-83.

Al-Awad, M. and Harb, N. (2005), “Financial development and economic growth in theMiddle-East”, Applied Financial Economics, Vol. 15, pp. 1041-1051.

Asafu-Adjaye, J. (2000), “The relationship between energy consumption, energy prices andeconomic growth: time series evidence from Asian developing countries”, EnergyEconomics, Vol. 22, pp. 615-625.

Baltagi, B.H. (2001), Econometric Analysis of Panel Data, 2nd ed., Wiley, New York, NY.

Beck, T. and Levine, R. (2004), “Stock markets, banks, and growth: panel evidence”, Journal ofBanking & Finance, Vol. 28, pp. 423-442.

Beck, T., Levine, R. and Loayza, N. (2000), “Finance and the sources of growth”, Journal ofFinancial Economics, Vol. 58, pp. 261-300.

Christopoulos, K.D. and Tsions, G.E. (2004), “Financial development and economic growth:evidence from panel unit root and cointegration tests”, Journal of Development Economics,Vol. 73, pp. 55-74.

Coiteux, M. and Olivier, S. (2000), “The saving retention coefficient in the long run and in theshort run: evidence from panel data”, Journal of International Money and Finance, Vol. 19,pp. 535-548.

Dahduli, M.S. (2009), “Islamic banking and economic development”, Working Paper Series,available at: http://ssrn.com/abstract¼1616624 or http://dx.doi.org/10.2139/ssrn.1616624

Dar, H. and Presley, J. (2000), “The lack of profit loss sharing in Islamic banking: managementand control imbalances”, Economic Research Paper No. 00/24.

Deidda, L. and Fattouh, B. (2002), “Non-linearity between finance and growth”, EconomicsLetters, Vol. 74, pp. 339-345.

Demetriades, P.O. and Hussein, K.A. (1996), “Does financial development cause economicgrowth? Time series evidence from 16 countries”, Journal of Development Economics,Vol. 51, pp. 387-411.

El-Ghattis, N. (2011), Islamic Banking’s Role in Economic Development: Future Outlook, availableat: www.cba.edu.kw/wtou/download/conf4/nedal.pdf

Fisman, R. and Love, I. (2003), “Financial dependence and growth revisited”, NBER WorkingPaper No. 9582, National Bureau of Economic Research, Cambridge, MA.

Furqani, H. and Mulyany, R. (2009), “Islamic banking and economic growth: empirical evidencefrom Malaysia”, Journal of Economic Cooperation and Development, Vol. 30, pp. 59-74.

Harris, R.D.F. and Tzavalis, E. (1999), “Inference for unit roots in dynamic panels where the timedimension is fixed”, Journal of Econometrics, Vol. 91, pp. 201-226.

Hassani, M. (2010), “Islamic banking and monetary policy: experience of Iran (1982-2006)”,International Review of Business Research Papers, Vol. 6 No. 4, pp. 430-443.

IMEFM6,2

170

Hussin, A., Nor’Aznin, A.B. and Sallahuddin, H. (2010), “Analysis of FDI inflows to China fromselected Asean countries: a panel cointegration approach”, paper presented atInternational Economic Conference on Trade and Industry (IECTI).

Im, K.S., Pesaran, M.H. and Shin, Y. (1997), “Testing for unit roots in heterogeneous panels”,working paper, University of Cambridge, Cambridge.

King, R.G. and Levine, R. (1993), “Finance and growth: Schumpeter might be right”,The Quarterly Journal of Economics, Vol. 108 No. 3, pp. 717-737.

Lee, C.-C. (2005), “Energy consumption and GDP in developing countries: a cointegrated panelanalysis”, Energy Economics, Vol. 27 No. 3, pp. 415-427.

Levine, A. and Lin, C.F. (1993), “Unit root tests in panel data: asymptotic and finite sampleproperties”, working paper, Department of Economics, University of California atSan Diego, La Jolla, CA.

Levine, R., Loayza, N. and Beck, T. (2000), “Financial intermediation and growth: causality andcauses”, Journal of Monetary Economics, Vol. 46, pp. 31-77.

McKinnon, R.I. (1973), Money and Capital in Economic Development, Brooking Institution,Washington, DC.

Maddala, G.S. and Kim, I.-M. (1998), Unit Roots, Cointegration, and Structural Change,Cambridge University Press, Cambridge.

Majid, S.A. and Kassim, S. (2010), “Islamic finance and economic growth: the Malaysianexperience”, paper presented at Kuala Lumpur Islamic Finance Forum, Kuala Lumpur,2-5 August.

Mirakhor, A. and Khan, S.M. (1990), “Islamic banking: experiences in the Islamic Republic of Iranand Pakistan”, Economic Development and Cultural Change, Vol. 38, pp. 353-375.

Mishkin, F.S. (2006), The Economics of Money, Banking and Financial Markets, PearsonInternational Edition, Singapore.

O’Connell, P.G.J. (1998), “The overvaluation of purchasing power parity”, Journal of InternationalEconomics, Vol. 44, pp. 1-19.

Pedroni, P. (1996), “Fully modified OLS for heterogenous cointegrated panels and the case ofpurchasing power parity”, working paper, North American Econometric Society SummerMeeting, pp. 96-120.

Pedroni, P. (1999), “Critical values for cointegration tests in heterogeneous panels with multipleregressors”, Oxford Buellton Economic Statistics, Vol. 61, Special Issue, pp. 653-678.

Pedroni, P. (2000), “Fully-modified OLS for heterogeneous cointegration panel”, NonstationaryPanels, Panel Cointegration andDynamic Panels, Advances in Economic, Vol. 15, pp. 93-130.

Pedroni, P. (2001), “Purchasing power parity tests in cointegrated panels”, The Review EconomicStatistics, Vol. 83 No. 4, pp. 727-731.

Pedroni, P. (2004), “Panel cointegration: asymptotic and finite samples properties of pooled timeseries tests with an application to the PPP hypothesis”, Economic Theory, Vol. 20,pp. 597-625.

Phillips, P.C.B. and Hansen, B.E. (1990), “Statistical inference in individual variables regressionwith I(1) process”, Review of Economic Studies, Vol. 57, pp. 99-125.

Rajan, R. and Zingales, L. (1998), “Financial dependence and growth”, American EconomicReview, Vol. 88 No. 3, pp. 559-586.

Rioja, F. and Valev, N. (2002), “Finance and the sources of growth at various stages of economicdevelopment”, Working Paper No. 02-17, Andrew Young School of Policy Study, GeorgiaState University, Atlanta, GA.

Financing andeconomic growth

171

Shaw, E. (1973), Financial Deepening in Economic Development, Oxford University Press, Oxford.

Shen, C.-H. and Lee, C.C. (2006), “Same financial development yet different economic growthwhy?”, Journal of Money, Credit, and Banking, Vol. 38 No. 7, pp. 1907-1944.

Sufian, F., Mohamad, A.M.N. and Muhamed-Zulkhibri, A.M. (2008), “The efficiency of Islamicbanks: empirical evidence from the MENA and Asian countries Islamic banking sectors”,The Middle East Business & Economic Review, Vol. 20 No. 1, pp. 1-19.

Zangeneh, H. (1995), “A macroeconomic model of an interest-free system”, The PakistanDevelopment Review, Pakistan Institute of Development Economics, Vol. 34 No. 1, pp. 55-68,available at: www.financeinislam.com/article/1_35/8/287

Further reading

De Gregorio, J. and Guidotti, P.E. (1995), “Financial development and economic growth”,World Development, Vol. 23 No. 3, pp. 433-448.

Mulya, B. (2010), “Indonesian economy: recent development and challenges”, paper presented atthe Oversea-Chinese Banking Corporation Global Treasury Economic and BusinessForum, Singapore, 9 July.

Narayan, P.K. (2004), “Reformulating critical values for the bounds F-statistics approach tocointegration: an application to the tourism demand model for Fiji”, discussion papers,Department of Economics, Monash University, Melbourne.

Pesaran, H.M. and Shin, Y. (1995), “Autoregressive distributed lag modelling approach tocointegration analysis”, DAE Working Paper Series No. 9514, Department of Economics,University of Cambridge, Cambridge.

Pesaran, H.M., Shin, Y. and Smith, R. (1996), “Testing the existence of a long-run relationship”,DAE Working Paper Series No. 9622, Department of Applied Economics, University ofCambridge, Cambridge.

Pesaran, H.M., Shin, Y. and Smith, R. (2001), “Bounds testing approaches to the analysis of levelrelationship”, Journal of Applied Econometrics, Vol. 16, pp. 289-326.

Corresponding authorYazdan Gudarzi Farahani can be contacted at: yazdan.gudarzi@ut.ac.ir

IMEFM6,2

172

To purchase reprints of this article please e-mail: reprints@emeraldinsight.comOr visit our web site for further details: www.emeraldinsight.com/reprints