DRAFT God, Government and Outsiders: The Influence of Religious Beliefs on Depositor Behavior in an Emerging Market. Ayesha K. Khan January, 2010 Abstract This paper provides evidence that religious beliefs can have a significant impact on individual financial choices. Using proprietary panel data on the distribution of bank deposits across all commercial banks in Pakistan over a 33-month period, I find that Islamic banks enjoy substantially higher deposit growth rates than other banks and that this difference persists even after various other profit-maximizing determinants of bank demand are taken into account. I also find that while a recent financial crisis triggered a fall in deposit growth rates at all other types of banks, it had a positive impact on the religious banks despite the fact that these banks tend to have lower credit scores than other conventional banks. Together, these results reflect some of the complex factors influencing individual financial decisions and indicate that at least in the context of a religiously motivated population it makes economic sense to focus on the growth of institutional forms that reflect these preferences. 1 DRAFT Introduction The impact of religion on economic outcomes has been the subject of scholarly debate since Adam Smith’s analysis of religious institutions as competitive firms subject to market dynamics (1776). More recently, as religion has continued to remain deeply embedded throughout the world, understanding the influence of religiosity at the individual, the group and the national level has generated a new wave of theoretical and empirical work exploring the role of religion as an independent variable with a wide range of socioeconomic effects (see Iannaconne, 1998 for an overview). This paper focuses on effects of religious beliefs on financial markets and provides some insights into the complex set of monetary and non-monetary preferences that influence individual decisions to allocate savings to a specific bank under different economic conditions – including a severe financial crisis. In doing so, it also connects to existing literature on the impact of organizational differences between banks and how they respond to liquidity shocks (Berger, 2002, Libreti 2003, Goldberg et al. 2000). Much has already been written on the economic consequences of different bank structures. For example, Mian (2003) finds that domestic banks have an advantage in lending to soft information firms in comparison to foreign banks while Crystal et al. (2002) show that older foreign banks in Chile, Colombia and Argentina exhibited stronger and less volatile credit growth than domestic banks. However, this literature has focused exclusively on the bank lending channel with very little being said about the liability side of bank balance sheets and how depositors respond to different bank 2 DRAFT characteristics. Given that a bank’s deposit base is a critical source of liquidity in emerging markets and for smaller banks that are most vulnerable to financial crises, it is important to address this gap and understand the determinants of consumer demand for deposits in a way that goes deeper than simple profit maximization. This paper takes a step in that direction by examining how one specific bank characteristic of associated religiosity can affect its deposit base. As far as the connection between religion and economics is concerned, much of the evidence so far has emerged from cross-country studies of growth rates and religiosity (La Porta et. al 1997, Inglehart 1999, Landes 1998, Barro 2002) where it is often difficult to control for the existence of other institutional differences that may be correlated with dominant faith. Alternatively, at the micro level, various papers have examined how religious beliefs affect labor choices (Azzi and Ehrenberg 1975, Lehrer 1995, Berman 2000) or individual traits such as honesty, work ethic, trust and education that in turn influence economic performance (Chiswick 1983, Freeman 1986, Ellison 1991, Evans et al. 1995). However, it has been difficult to establish causal links, overcome endogeneity issues and disentangle the impact of religion from alternative explanations. This paper overcomes some of the specification problems associated with isolating the economic effects of religiosity by examining deposit growth rates at commercial banks that differ along a well-defined religious dimension. In order to do so I use a proprietary panel data set of demand deposits across all commercial banks in Pakistan where, like most other emerging markets, three of the dominant bank types – private banks, foreign 3 DRAFT banks and public sector banks – are identified on the basis of ownership. In addition, over the past decade, a fourth category of banks identified on the basis of religion and known as Islamic banks, has emerged and grown rapidly within the most financially sophisticated parts of the country. These religious banks conform to Islamic legal (shariah) restrictions on financial transactions while offering deposit accounts and credit services that look very similar to those being offered at mainstream, conventional banks. In fact, from the perspective of a depositor, a basic Islamic bank account is an almost perfect functional substitute for a conventional account of the same type – with the additional halo of associated religious compliance. This basic similarity makes it possible to compare the growth of simple deposit accounts across different bank types, including religious banks. In addition to the basic religion and ownership-based categorizations, I split the local banks in my sample into two categories based on the size of their asset base and divide the foreign banks into Western and Arab owned banks. This results in six distinct bank categories – Islamic banks, State owned banks, Western banks, Arab banks, small and large local banks. I find that over the 33-month period covered by the data, demand deposits at Islamic banks grew faster than they did at any other category of banks. More specifically, for every category of comparable basic deposits – from checking accounts to time deposits – growth rates were at least three times as fast at Islamic banks than they were at comparable conventional banks. These effects persist even after I control for other variables- such as credit ratings, bank size, bank age, branch network, deposit rates, leverage ratios – that may affect deposit growth. 4 DRAFT More interestingly, with all other controls factored in, when I focus the analysis on a recent two month period during which Pakistan went through an acute financial crisis that bought it to the brink of bankruptcy, I find while the crisis had a negative impact on deposits at conventional banks, it actually a positive effect on Islamic banks where the deposit base actually increased during the crisis. While the crisis effect is not significant at conventional levels due to the limitations of the panel dataset, it is particularly interesting to note its presence in the context of an emerging market like Pakistan where there is no FDIC type depositor insurance and bank failures can result in substantial losses for depositors. In fact, in combination with the earlier, robust finding of substantially higher deposit growth rates for religiously affiliated banks within the sample, it points towards a preference for holding religious assets that at least persists, and may even increase, during more difficult periods. These findings reflect some of the complex factors influencing individual decisions to hold bank deposits at a particular bank. They demonstrate that religious affiliations can have a significant impact on bank success even if they provide no profit maximizing benefits to depositors. These findings are also relevant on a policy level – particularly in the context of emerging markets where the financial sector is often under developed, capital structures are relatively simple and the likelihood of an economic crisis is relatively high. As various papers have successfully documented (Kashyap and Stein 1994, 1995, 1999; Kishan 1999, Peek and Rosengren 1997, Kishan and Opiela , 2000; Khwaja and Mian, 2008), bank liquidity shocks are directly transmitted to the rest of the economy via the bank lending channel. This is especially true for smaller banks that are 5 DRAFT financed almost entirely through deposits and common equity and are often unable to access alternative sources of funding. A shock to the deposit base of these banks has a real effect on lending behavior and economic growth. As such, if endogenous consumer preferences for religiously denominated assets can have a stabilizing influence on bank deposits then it makes sense to support the growth of these religiously affiliated banks and allow them to absorb some of the effects of financial instability that would otherwise spill over to the rest of the economy. Very simply, the results in this paper indicate that non-monetary preferences matter and for a religiously motivated population it makes economic sense to focus on the growth of institutional forms that reflect these preferences. The rest of the paper is proceeds as follows. In Section 2, I provide an overview of the political and economic context in Pakistan during the period covered by the dataset including a review of the 2008 bankruptcy crisis and a brief description of the banking sector. In section 3, I discuss Islamic banking. In Section 4, I describe the construction of the dataset and specify all variables included in subsequent regression models. In Section 4, I report the results of various empirical tests and Section 5 concludes the paper. 2: Background Country Overview Pakistan, with a population of 175 million and a per capita GDP of $2,600 (2008, PPP) is the fifth largest emerging market by population size. It is also one of the most turbulent 6 DRAFT regions in the world, suffering from decades of political disputes and instability, low levels of social development, income inequality and persistent bouts of military rule that have undermined the growth of democratic political institutions. More recently, the level of violence has escalated even higher as militant Islamic groups operating along the Afghan border set off a wave suicide bombings that killed more than 3000 people in 20091. Over 97% of all Pakistanis follow Islam and like much of the Muslim world, religion has continued to exert a strong, possibly increasing, influence over the population. The level of religious self-identification is very high with over 91% of Pakistani’s considering themselves religious people and 82% regarding religion extremely important in their lives 2(World Values Survey, 2001). This high level of general religiosity is confirmed by a more recent survey conducted by Nielson (2009) that found that while only one out of seven respondents identified themselves as Pakistani, over three-quarters identified themselves primarily as Muslim. Furthermore, and perhaps not surprisingly given the constant political instability, rampant institutional failures and frequent bouts of military dictatorships, when the same survey asked what institutions were trusted the most in the country, it found that a majority of respondents rated Pakistan’s military and religious educational institutions the highest while the national government came last with less 1 eclan a l sh , Pakistan suffers record num b er deaths du e to m ilitan t violen ce ” Jan 11, 2010. .gu ardian.co.u k /w orld/2010/jan/11/pak istan- militant- violence- death - toll 2 a re h is o he v e ra g e s ll o u n t rie s mpled u rin g e rio d y he VS, 6% of resp ondents conside re d lig io n m e ly p o rt a n t h e ir es of http://w w w omp r h is h e re x t re nd 3% e n t ifie d 7 s lig io u s people. DRAFT than 10 percent of the respondents expressing trust in state run institutions (see Figure 2 in the Appendix for details). The Financial Crisis of 2008 For the most part, despite intermittent bursts of violence and internal instability, Pakistan’s economy grew rapidly under the government of General Pervez Musharraf, – the president and chief of army staff, who first came to power in a military coup in 1999 and introduced a broad based set of economic reforms in 2000. For the first four years of that decade, the Karachi Stock Exchange (KSE) was regarded as the best performing stock market index in the world and by 2005, the World Bank had named Pakistan the top reformer in its region and in the top 10 reformers globally. By late 2007, foreign investment had increased by 47.7% from the previous year to a peak of around $8 billion, foreign portfolio investment had increased by 222.5% to reach record high levels and despite the presence of double-digit inflation, the economy was expected to continue growing at a healthy rate of 7% for the near future. However, by 2008 the Musharraf regime was under strong pressure to relinquish power, triggering widespread political protests that had a severely negative impact on foreign and domestic investment. The supply of credit declined and local companies found it more difficult to access to money from international markets due to rising political risk insurance. As a consequence of increasingly chaotic political turmoil, real GDP growth rates slowed down, inflation hit a 30 year high of 24% in June, the trade deficit widened, industrial output fell by 5% and the KSE index that had peaked at a record high of 15,676 8 DRAFT points in April, dropped by more than a third in July 2008. The collapse was so severe that by mid August the stock exchange was forced to set a floor for prices to halt the plunge that had wiped out over $36.9 billion of market value since April (see Figure 1 in the Appendix for details). By October 2008, the KSE index had lost more than half its value, the national currency had depreciated to a historic low, foreign exchange reserves were depleted and the budget deficit was at a 10-year high. Pakistan's creditworthiness was ranked the second worst among all nations ranked by Standard and Poor's and markets were seized by such widespread rumors of imminent financial collapse that Pakistan’s newly elected President was forced to go public with an assurance that “Pakistan is not going bankrupt.”3 The erosion of consumer confidence in the economy extended to the banking system and triggered a run on foreign-currency accounts held at various banks in Karachi, Islamabad and other major cities. In October 2008, at least three commercial banks were the subject of intense default speculations. Newspapers and television stations reported these risks as fact, causing depositors to panic and start withdrawing savings and prompting public reassurances of financial stability and causing at least one bank CEO to hold a press conference categorically denying bankruptcy rumors.4 3 0 ctober 008, a rd a ri. http://w w w .daw n.com /w ps/w cm /connect/daw n- con ten t- lib ra ry/ da w n / n ew s/ p a k ista n / p a k ista n +is+n ot+goin g+b a n k ru p t+p residen t+za rda ri+a a h 4 he EOs a ssu ra n c e s e re llowed by a further public announcem ent by the CE O of the Abu Dhabi Group hat as a re h o l d e r he ank nd n v e st m e n t e h ic l e r he bu habi yal fa m il y . Tahir fu rt h er a ssu re d as o announcing that vestment u rre n t e p o sit o rs his f grou p w ou ld b e a k is t a n , o n d it io n s . hat h e re a n k ru ptcy by in c re a s in g s e s p it e 9 he DRAFT This instability continued through November 2008 when it became clear that Pakistan was seriously close to a financial collapse that could paralyze its government and further complicate the ongoing war in Afghanistan. To prevent this from happening, the IMF announced on November 15th that it had reached an agreement with the government of Pakistan for a $7.6 billion bailout package. The bailout package was approved on November 24th and $3.1 billion were immediately released as the first tranche of funding support, pulling Pakistan back from the brink of imminent economic failure. The Banking Sector in Pakistan Like various emerging markets, the level of financial exclusion in Pakistan is very high. Only about 17% of the population (30 million) has a bank account while less than 4% (5.5 million) has access to credit. Given the scarcity of banking services, the real rate of return on deposits has remained negative and banking spreads are so high that Pakistan’s banking sector has been regarded as the second most profitable in the world after Colombia5. Historically, six government owned banks dominated the banking sector and accounted for 92% of market share with the remainder being distributed between foreign banks.6 During the 1990s, various banking sector reforms were introduced to liberalize the 5financial sector by privatizing four out of the five largest state owned banks and issuing een nt, Banking spread in a k is t a n ave s c il l a t e d etw 5.95 an d 9.58 p ercent du ring 1990 - 2008 and were around 8% in January 2009. or o m p a riso n purposes, the average interest rate spread in Can ada w as .3 e rc e he K .3 e rc e n t , p a in .4 e rcen t, the US 2.8 p ercent, A u stralia 3 p ercent and France 3.1 percen t. e n e ra l l y cceptable evels .5 f e rc e n t . 6 n a l iz a t io n ct f d he n t ire f a k is t a n o som e sp ecia lized credit in stitu tion s directing bank credit tow ards governm ent funding. 10 ank p re a d s re ank a t io 974 ad a t io n a l iz e a n k in g ector crea te these six , la rge b a n k s a n d w ith the ob jectives of specific sectors and ensuring he DRAFT licenses for the creation of new private sector banks (Bonaccorsi, 2005). By 2009 the banking sector was composed of 39 private domestic, public and foreign owned commercial banks, including 6 Islamic banks. 3: Islamic Banking Islamic banking is a prohibition-based industry based on shariah rules that prohibit transactions involving riba (a fixed rate of return), gharar (uncertainty or speculation) and particular sin sectors such as alcohol, pornography and gambling. The core prohibitions on interest are an outcome of the general belief that it is unjust to earn income without assuming risk (Chapra 1992). So unless a depositor assumes some of the risk inherent in investing her funds she is not entitled to a return on her money. In order to make sure that Islamic banks conform to strict Islamic principles of banking, a “Shariah Board” composed of various religious scholars is used to supervise the development and creation of all financing products and services offered by the bank. This board usually ranks above the Board of Directors within the organizational structure of an Islamic bank and is empowered to issue fatwas (legal pronouncements) on any matter proposed to it by different business units of the bank. Since any transaction involving a pre specified rate of interest is strictly prohibited in the Islamic banking model, deposit returns are based on the principles of profit and loss (PLS) sharing where the depositor enters into an equity type relationship with the bank. Instead of receiving interest on their accounts, depositors at an Islamic bank receive profit shares that tend to fluctuate over time. In theory this implies that the returns associated 11 DRAFT with holding an Islamic savings account could be very different from the fixed returns offered by a conventional account. However, in practice, the PLS returns usually follow the movements of ordinary interest rates and are very similar to the deposit rates being offered at conventional banks. This is partially because Islamic banks often place their deposits into bonds and other interest bearing instruments but also because Islamic banks are directly competing for deposits with conventional bank and as such maintain generally comparable stream of returns for comparable products (Kuran 1995)7. So while Islamic banks have the potential to provide a very different type of financial intermediation, at least at present, the products and services provided by them are shariah compliant, repackaged versions of similar products being offered by conventional banks. In fact, as pointed out by Kuran (1997), Islamic banks are often “an instrument of guilt reduction for depositors and borrowers who believe that, even if Islamic banking is not actually interest-free, it is at least morally superior to conventional banking.” Not surprisingly, Islamic banks tend to focus on this perceived moral superiority and signal their religious affiliations in various ways that include specifying Islamic dress codes for all bank employees, setting aside specific prayer areas within the bank and using advertising campaigns that make heavy use of particular images and colors that tap into religious beliefs. 4: Data Description and Definition of Variables 7 s em p lo y ees m ic anks ften n o fficia lly o ten tia l dep ositors retu rns no low er than interest rate… in cou ntries w here Islam ic b ank s com p ete 995) o in ts u t, u ra n f the w ith e ro m ise p revailing con ven tion a l b a n k s, essen tia lly 12 the osten sib ly in terest - fre e tu rn s of match the explicitly interest - based returns of the the form er latter.” DRAFT The dataset used in this paper is constructed from proprietary, bank level data on consumer deposits held within the commercial banking sector in Pakistan. This data consists of monthly information on various deposit accounts held at 32 private, public, foreign and Islamic banks over a period of 33 months (June 2006 – March 2009). The data was sourced from the State Bank of Pakistan (SBP), which is the central bank of Pakistan. The SBC data and other sources used to put together the dataset are described below: The SBP Data This SBP requires all scheduled banks to provide a monthly summary of balance sheet aggregates for regulatory and monitoring purposes. The data used in this paper was extracted from these proprietary monthly status reports submitted to the SBP and consists of information from the liability side of bank balance sheets. For each of the 39 banks operating in Pakistan in December 2008, I have monthly information on the level of checking, saving and time deposits held at the bank. As defined by the SBP, checking and saving accounts are standard demand deposits that earn zero and non-zero returns while time deposits are yield-bearing accounts that cannot be withdrawn for a specific 8duration ranging from a few months to ten years or more. The SBP information was aggregated at the bank/account type level and available for 33 periods ranging from June 2006 to March 2009. 8 im e s e p o sit s e rm e p o sit s. re e p o sit s, he l so onds t u rn nown r n ed h e se accounts longer the depends term the on holding tim e higher the yield. and in 13 m ost cases the DRAFT The SBP dataset provided deposit level information for all commercial banks – including four specialized banks providing development focused financial services for rural, agricultural needs or term finance for small and medium enterprises. Since the focus of this paper is on understanding depositor behavior in response to particular bank characteristics, I excluded these four specialized banks from the dataset because they did not cater to the general population. I also excluded three foreign banks that were primarily catering to home country business interests via a single branch and did not have a relevant commercial banking presence within the local population. After these exclusions, I was left with a panel of 32 banks and 1056 bank-month observations. Out of the 32 banks in the SBP sample, two banks did not have a complete set of 33 monthly observations since they did not start operating until mid 2007. This means that the panel data is not balanced since observations for earlier periods are missing for two banks. However, this minor gap has had no effect on our empirical results and they do not change in any way if these banks are excluded from the sample. During the time period covered by the panel data, there was one change in ownership and two bank acquisitions. Since the change in ownership did not change the bank type (it remained a local, private bank but with different shareholders), it did not affect our general results. As far as the acquisitions were concerned, in order to keep the data consistent, I extended the balance sheet aggregation of merged banks to the beginning of the dataset in June 2006. So, for example, while NIB Bank acquired PICIC Bank in December 2007, I manually aggregated the deposit account information for NIB and 14 DRAFT PICIC from June 2006 till the December 2007 to make sure that any changes in deposit levels seen at NIB signified depositor response to the bank – and not a sudden deposit growth via acquisition. While this ex-ante acquisition adjustment does not account for all the effects of balance sheet consolidation between two banks, it only affected less than 1% of the banks for part of the time period used in this study and the exclusion of these banks does not affect our results. Bank Types Despite the banking sector liberalization of the 1990s, older, previously state owned banks have continued to dominate the banking sector through their extensive branch presence, their significantly larger asset base, deeper networks and stronger brand reputations. The dominant position enjoyed by these older banks is particularly striking in smaller cities and rural areas where they are often the only source of formal banking services along with, in some cases, the remaining public sector banks (PBs). In contrast, the newer private banks are much smaller, have significantly less market power and are clustered within urban areas. Given the highly skewed size distribution of private banks, it makes sense to use cutoffs to assign banks to different categories. Accordingly, I divide the private banks into two categories based on total assets. The small local banks (SLBs) have total assets that are below the 50% percentile for all banks while the large, local banks (LLBs) have assets that are above that cutoff. Within the category of foreign banks operating in Pakistan, it is possible to distinguish between Western foreign banks like Citibank, HSBC or Deutsche bank and Arab owned 15 DRAFT foreign banks financed by proximate by Gulf petrodollars. While Western foreign banks (hereby referred to as “Foreign”) have primarily focused on a higher-income market segment and clustered in urban centers, Arab banks (referred to as “Arab”) have catered to a more middle market segment and tend to have larger branch networks. Finally, despite previous attempts to “Islamize” the entire Pakistani banking sector and bring it in line with Islamic legal codes, Islamic banks (IBs) really only emerged as a distinct category after 2002. Since that time, these banks have grown at a rapid pace to account for almost 5% of the banking sector in 2009. Like most other commercial banks in Pakistan, Islamic banks are concentrated in major cities where they provide banking services that are very similar to and directly compete with conventional locally owned, state owned and foreign owned banks. Credit Ratings and Leverage Ratios Measures of bank quality were drawn from credit ratings on each of the banks in the sample. These ratings are issued by PACRA and JCR-VIS, two independent credit rating 9agencies that cover the entire SBP sample. Rating scales used are similar to those used by Standard & Poor’s/Fitch and ranged from AAA (Prime/almost zero risk) to D (in bankruptcy/default) with additional +/-intermediate ratings. For the purposes of this paper, I transform the rating scale to a numerical variable ranging from a 10 for AAA to a 91 for BBB-, the minimum investment grade assigned (details in Appendix 1). The banks ACR A he gency) CA a tin g a k ist a n re d it in t im it e d g en cy), he a t in g e n t u re etween In tern a tion a l Ct e tern a tion a l en y, n a n ce ora tion (IF n d e h ore Stock Exchange. W hile JCR- VIS is a joint ventu re etwe he pan re d it a t in g genc td. R ), Vital Inform ation Services (Pvt.) Lim ited (VIS) – a k ist a n ’s nly ata ank nd a n c ia l se a rc h organization, K arachi Stock E xchange and Islam ab ad Stock E xchange. 16 DRAFT in the dataset range from AAA banks corresponding to the maximum numerical rating of 10 to BBB+ banks earning a 3 on the numerical scale. The ratings focus on long-term bank stability and for the most part remain fairly stable during the 33-month period under investigation. The ratings themselves were based on financial and business risk measures. Specifically, these included indicators based on the CAMEL(S) criteria (capital adequacy, asset quality, management quality, earnings, liquidity and sensitivity to market risk) along with additional checks on market position, corporate governance, earning stability and ownership structure. I also talked to local banking experts to verify that these credit scores were a credible reflection of bank reputations. Since these scores take into account qualitative factors such as bank strategy, auditor quality and expected news affecting future performance as well as institutional factors such as ownership structures and support mechanisms make them a useful indicator of bank quality that goes beyond accounting measures. In fact, since I are specifically focusing on individual perceptions of bank characteristics as reflected in the willingness to hold deposits at a specific bank, the credit ratings serve as a consistent and valid indicator of underlying bank quality. Additional measures of bank quality were estimated from leverage ratios derived from quarterly SBP reports on bank balance sheets. These simple ratios measure capital adequacy by expressing Tier 1 bank capital as a percentage of total assets and are used as a prudential tool to limit excessive leverage in a banking system. The leverage ratio indicates the amount of equity or capital support that can protect the bank from 17 DRAFT unexpected events. The smaller this support gets, the greater the risk the bank may become insolvent. Accordingly, banking agencies publish threshold levels for the leverage ratio. In the US, the minimum is at 3% for otherwise strong banks and 4% for all other banks. While in other countries a minimum of 5% is considered adequate10. Deposit Rates The SBP does not collect monthly data on the interest rates offered on different types of bank accounts. However, I have a separate six monthly SBP dataset that includes this information. Given that returns on bank accounts do not fluctuate significantly, I use the six monthly interest rate figures as an approximation for the monthly interest rates being offered by the banks. While I have deposit rate information for all interest bearing accounts, the substantial inter bank variation in the structure of time deposits makes it difficult to do rate based comparisons across this category. Also, time deposits are not very good indicators of short-term consumer preferences due to the constraints that they impose on withdrawals before term. In contrast, savings accounts (along with non interest bearing checking accounts) are very flexible indicators of consumer preferences. They form the bulk of all consumer deposits and their structure does not differ substantially across different banks. As such the paper focuses on simple, demand deposits and controls for the deposit rates offered on these accounts. The dataset also contains information on bank age and branch network drawn from bank annual reports. It is likely that in an unstable environment like Pakistan, bank age will 10 ee http://crisistalk .w orldb ank .org/files/B ank ing% 20and% 20the% 20Leverage% 20R atio.pdf r a m ore detailed discussion of banking and the leverage ratio by the W orld B ank. 18 DRAFT have a positive impact on the demand for a particular bank’s services, particularly in the absence of any depositor insurance. Similarly, a broad network of branches is likely to increase the demand for holding deposits at a particular bank because it allows for more flexibility in withdrawing deposits. Table 1 summarizes the different bank characteristics by category and shows that IBs are very comparable to SLBs in terms of almost all bank descriptive variables in our dataset including size, age, credit scores, deposit rate, leverage ratios and coverage. Table 1: Large Local Foreign Arab Public Islamic Small Local # Banks 9 4 3 4 6 6 Years open 35.78 42.25 15.33 29.50 4.17 4.67 Branches 549.51 62.56 124.51 393.18 31.32 29.70 Assets 273840.30 116629.20 163269.90 234577.70 22576.99 24733.39 Equity 24678.76 12951.70 10886.96 30610.04 3682.81 4694.38 Credit score 8 8.75 6.33 6 4.5 4.5 Deposit rate 3.41 1.73 4.58 2.95 4.31 4.53 Leverage ratio 8.95 9.42 7.41 12.58 29.17 23.48 Coverage Major and Major cities Major cities Major and Major Major cities minor cities. only and some minor cities cities only only Rural areas smaller cities 5: Empirical Results I: Cross sectional Analysis: Growth rates and Deposit Levels I start by examining the difference in deposit levels and growth rates between different types of banks. To do this I collapse the panel data by taking the averages of each variable across different bank types. Table 2 summarizes the differences between different types of banks in terms of deposit levels and growth rates. 19 DRAFT Table 2: Summary Statistics - Deposit Levels and Growth Rates Large Local Foreign Arab Public Islamic Small Local Checking 26344.87 14781.69 12017.23 9287.7 1779.37 642.01 Saving 54391.3 16876.8 15728.81 27575.04 3077.13 1707.92 Time 21732.04 17290.57 12936.68 4816.14 4189.21 2126.81 Checking Growth 2.16 2.29 1.87 1.34 13.22 5.51 Saving Growth 1.48 1.62 0.50 0.01 14.63 4.76 Short Term Growth 1.34 1.90 0.94 0.12 12.16 4.15 Time Growth 4.70 2.14 2.29 6.52 18.07 7.26 Total Deposit Growth 1.65 1.90 1.10 0.83 13.51 4.82 As expected, the larger domestic banks have a dominant market position and account for about 42% of total deposits. Western foreign banks account for around 20% of deposits, Arab banks for 16% and government owned PB’s for 17%. The SLBs and the Islamic banks hold the smallest market share. More than half of the total deposits in the larger LLBs and the PBs is held in saving accounts, while the IBs and the SLBs hold a majority of their deposits in long term, fixed time accounts (46% and 47% respectively). Together, the checking and saving accounts, i.e. the demand deposits, form the bulk of all deposits at every bank in the sample. As far as deposit growth is concerned, the IBs have a significantly higher growth rate than any other bank category. While total deposits at most other banks grew at a moderate rate of 1- 2% over the 33-month period covered by the dataset, they grew at a substantially higher rate of 13.5% at Islamic banks – a rate that was at least three times as high as the rate of growth at comparable, conventional SLBs. 20 DRAFT Figure 1 shows the distribution of average growth rates across different deposit categories for each bank in the sample and reflect the same, striking difference in growth rates between banks. Distribution by Bank and Bank TypeDeposit Growth Rates Deposit Growth Rates Figure 1: Growth Rates by Bank and Bank Type 00551010151520202525303035354040Growth, %Growth, %Large LocalLarge LocalForeignForeignArabArabPublicPublicIslamicIslamicSmall LocalSmall LocalBank TypeBank TypeTotal DepositsTotal DepositsShort Term DepositsShort Term DepositsTime DepositsTime DepositsDistribution by Bank and Bank TypeDeposit Growth Growth, % g n e iF or A i cPub l r Bank Type mi cl aI s L ocl Sma la l a b 0 5 10 15 20 25 30 35 40 RatesDistribution by Bank and Bank Type L ar ge a lL oc Total Deposits Short Term Deposits Time Deposits Panel Data Analysis: Growth of short-term deposits. II: The existence of a significant cross sectional difference between the deposit growth rates at different types of banks suggests that organizational structure may be a relevant factors in determining consumer demand for financial assets. Even when quantifiable factors such as credit ratings, bank returns and size are taken into account, it is possible that individuals place a certain value on the particular, non-monetary characteristics that describe a particular bank. So, for example, it is possible that consumers may have more (or less) implicit trust in public sector banks depending on their general levels of trust in the government. They may place a high value on religious beliefs that make Islamic 21 DRAFT banks appear more attractive than conventional banks. Or they may be more likely to trust foreign, multinational banks because they believe that these banks were more trustworthy than local banks. In order to test the impact of bank organizational characteristics on deposit growth rates, I construct bank type dummies that, along with a set of bank specific explanatory variables, can be used to estimate deposit growth rates. In effect, the bank type dummies function as fixed effect predictors in a panel regression where standard errors are clustered at the individual bank level. As such, the model controls for time constant, unobservables that may be correlated with the regressors and estimates the effect of bank types as the coefficient on its associated dummy variable. More formally, I estimate the following general model: =η + ' Xitß + Git i eit (1) Where Git is the monthly rate of deposit growth, ηi, (i = 1, ..., n ) is a set of time invariant bank type dummies, Xit is a matrix of bank specific control variables that includes deposit rates, credit scores, branches, average size, years open and eit is the idiosyncratic ~ IID ( 0, error, eit se) Table 3 shows the results of estimating (1) without bank specific control variables for different types of deposit accounts. Demand deposits are the sum of checking and saving deposits. I focus on this category as the dependent variable because these short-term 22 DRAFT deposits that form the bulk of all deposits in the sample are relatively standardized products that are easy to compare across banks. Also, unlike time deposits, there are no restrictions on withdrawing or closing these short-term accounts so the growth in these deposits is not subjected to structural constraints. The excluded comparison group comprises of the larger, local banks. Here, and for all subsequent regression models, the standard errors are clustered at the bank level. Table 3: Regression Estimates for Deposit Growth (No Covariates) (1) (2) (3) (4) Dependent Variable (Growth rates, %) Demand Deposits Checking Deposits Saving Deposits Total Deposits Foreign 0.557 0.135 0.139 0.256 (1.104) (1.634) (0.820) (1.003) Arab -0.406 -0.291 -0.978 -0.549 (0.562) (0.745) (0.626) (0.462) Public -1.222** -0.821 -1.474** -0.823* (0.488) (0.655) (0.607) (0.412) Islamic 10.820*** 11.067** 13.152*** 11.864*** (3.661) (4.463) (4.149) (4.265) Small Local 2.804* 3.356** 3.282* 3.174*** (1.378) (1.491) (1.662) (1.028) _cons 1.344*** 2.157*** 1.479*** 1.648*** (0.437) (0.648) (0.473) (0.333) N 995 963 995 995 r2 0.061 0.038 0.061 0.069 Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 As seen in the cross section aggregates earlier, the results of the Table 3 shows that for all types of deposits, IBs have been growing significantly faster than all other types of banks. This difference is greatest in comparison to public sector banks that appear to be growing at an even slower rate than the larger, conventional banks in the sample. All other types of banks, with the exception of the SLBs also do not appear to be growing at a 23 DRAFT significantly higher rate than the 1-2% growth rates at the LLBs. And while the SLBs grow almost faster than the older banks, in almost every case the IBs have grown almost three times as fast as the SLBs. For the purposes of clarity and relevance, given that the IBs are most comparable to the SLBs, the rest of the empirical analysis focuses on comparing these two bank categories to other banks while leaving a more general comparison between all six bank types for the robustness checks. Table 4 shows the results of estimating estimates equation (1) along with other covariates. Bank returns are estimated by drate, which is the deposit rate being offered 11by different banks on their savings accounts. Bank quality is estimated by cscore, which is the credit score assigned to the bank and ranges from a minimum of 3 (BBB+) to a maximum of 10 (AAA). Bank size is estimated by branches, the number of branches and lnavgtotal, which is the natural log of average total deposits at the bank over the 33month period covered by this dataset. Logyrsopen is the natural log of the number of years that a bank has been operating. 11 a sit sh o rt in c e return on te he deposits, ffered 24 h e c k in g the only ep o sits n accounts do relevant rate of s e v in g s not offer return for ep o cco u n ts. DRAFT Table 4: Regression Estimates for Islamic vs. Conventional Banks (1) (2) (3) Dependent Variable (Growth rates, %) Demand Deposits Checking Deposits Saving Deposits Islamic 3.423** 1.990 5.511** (1.577) (1.860) (2.302) Small Local -1.087 -2.039 -0.793 (1.798) (2.514) (2.170) cscore 0.339 0.207 0.502 (0.267) (0.343) (0.441) leverage 0.292*** 0.354*** 0.294*** (0.080) (0.119) (0.100) lnavgtotal 0.073 -0.187 -0.358 (0.626) (0.822) (1.193) logbranches -0.379 -0.158 -0.110 (0.486) (0.636) (0.631) logyrsopen -0.759 -0.740 -1.123 (0.628) (0.728) (0.876) drate 0.099 -0.015 (0.162) (0.206) _cons -0.924 2.207 2.515 (5.219) (6.607) (9.436) N 987 955 987 r2 0.094 0.068 0.084 Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 Again, as seen earlier, the results show that even after various other factors that may influence deposit growth, the Islamic banks grow substantially faster than other types of banks. In fact, Regression (1)-(3) show that while comparable conventional banks grow slower than average, demand deposits at IBs grow 3.4% faster than they do at other banks and that most of this growth comes from savings deposits, where aside from the leverage ratio, none of the monetary indicators including deposit rates associated with saving accounts appear to be a significant predictor of deposit growth (3). 25 DRAFT Figure 2 shows the distribution of point estimates for estimating equation (1) using fixed effects for individual banks, instead of bank category dummy variables. Again, it is clear from the figure that the point estimates of growth rates for Islamic banks – both in the case of short term demand deposits and for savings deposits – are higher than those for .04 .06 .08 .1 .12 .04 .06 .08 .1 .12 comparable conventional banks. Figure 2: Distribution of Point Estimates for Growth Rates by Individual Bank, Islamic vs. Conventional. All Covariates Included. Islamic Conventional This persistent growth rate difference between IBs and SLBs is particularly interesting because after controlling for age, credit scores, leverage ratios and size, the main difference between SLBs and IBs – both of which started operating around the same time, are located in the same areas, cater to the same market segment and have similar credit scores - is the religious affiliation of the IBs. 26 -10 -5 0 5 10Short Term deposit growth -5 0 5 10 15Saving deposit growth DRAFT Panel Data Analysis: Depositor Response to Economic Shocks As described earlier in the paper, Pakistan went through a severe financial crisis during the October- November, 2008 period. Accordingly, I define a variable crisis that takes on the value 1 if the time period is October or November 2008 and 0 otherwise. Also, in order to control for possibly cyclical end of year effects, I define yearend as a control variable that is equal to 1 for October and November in each of the three years included in the dataset. Formally, the model described in Equation (1) can be rewritten to estimate the effect of the crisis on deposit growth as follows: Gi =η + crisis * + ' Xitß + (2) t i ηi eit Table 5 shows the results of this specification. The results show that instead of the negative effect on deposit growth rates experienced by conventional banks during the crisis, the Islamic banks experienced an almost equivalent but positive effect on the growth of demand deposits during the same period. As seen in Regression (3), this positive effect came from a significant increase in checking account deposits during that period while conventional banks suffered a sharp reduction in the same deposit category. 27 DRAFT Table 5: Regression estimates of the Impact of an Economic Crisis on Short Term Deposit Growth Rates at Different Types of Banks Demand Deposits Savings Deposits (1) (2) (3) Dependent Variable (Growth rates, %) Checking Deposits Islamic 3.258* 5.720** 1.294 (1.700) (2.467) (1.787) Small Local -0.953 -0.783 -1.769 (1.792) (2.086) (2.577) Crisis -2.899 -4.659* -2.474 (1.796) (2.318) (3.060) Islamic*Crisis 2.460 -1.727 8.583 (5.977) (5.459) (11.231) Small Local*Crisis -2.147 -0.198 -5.296 (4.327) (7.705) (4.080) yearend 1.577 2.752 1.456 (1.594) (2.039) (2.633) drate 0.183 0.124 0.149 (0.160) (0.204) (0.209) cscore 0.373 0.593 0.210 (0.266) (0.453) (0.332) leverage 0.294*** 0.288*** 0.368*** (0.079) (0.099) (0.110) lnavgtotal 0.032 -0.538 -0.068 (0.631) (1.214) (0.713) logbranches -0.370 -0.043 -0.256 (0.497) (0.634) (0.648) logyrsopen -0.702 -1.046 -0.597 (0.620) (0.858) (0.766) _cons -1.368 2.554 0.223 (5.413) (9.602) (6.186) N 987 987 955 r2 0.096 0.087 0.071 Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 While the results of this analysis are striking in their dramatic mirror effect, it is important to interpret them with caution. The panel data used to derive these results is aggregated at the monthly level and the crisis variable includes only two observations for each bank. This, in addition to the fact that the model requires a large number of fixed 28 DRAFT effects and interactions means that most of coefficients are not significant at conventional levels. As such, they should be considered suggestive of particular directional trends that need to be interpreted in the context of the stronger evidence presented earlier and should be confirmed through further analysis using a wider panel or a narrower time dimension. Overall, however, these results they confirm and strengthen the finding that a bank’s religious affiliation can influence individual preferences for holding deposits at that bank in a way that goes beyond easily quantifiable measures of size, reputation and returns. The fact that these preferences hold during a crisis is particularly relevant in the context of Pakistan where, like many other emerging markets, there is no FDIC type depositor insurance and the decision to allocate savings to a particular institution is even riskier. I will discuss the implications of these findings and possible alternative interpretations in the final section of the paper after a brief discussion of robustness checks that show that my conclusions remain valid even when the analysis is extended to include a wider range of bank types. Robustness Checks Table 6 shows the results of estimating equation (1) with all related covariates for all six bank types. The excluded comparison group comprises of the larger, local banks and as before, standard errors are clustered at the bank level. 29 DRAFT Table 6: Regression Estimates for Islamic vs. all Other Bank Types Dependent Variable (1) (2) (3) (Growth rates, %) Demand Deposits Checking Deposits Saving Deposits Foreign 0.326 0.041 -0.140 (1.670) (2.359) (1.901) Arab -0.442 -0.315 -1.033 (0.503) (0.726) (0.693) Public -2.929*** -3.500*** -3.734** (0.818) (0.939) (1.427) Islamic 2.390* 0.801 4.004** (1.377) (1.864) (1.911) Small Local -2.355 -3.446 -2.592 (1.745) (2.522) (2.217) cscore 0.320 0.272 0.498 (0.215) (0.278) (0.406) leverage 0.279*** 0.339*** 0.277** (0.081) (0.122) (0.103) lnavgtotal -0.843 -1.336 -1.451 (0.865) (1.030) (1.607) logbranches 0.214 0.471 0.499 (0.855) (1.095) (1.142) logyrsopen -0.628 -0.464 -0.936 (0.746) (0.969) (1.018) drate 0.026 -0.103 (0.160) (0.202) _cons 6.481 10.739 11.850 (6.513) (7.833) (12.169) N 987 955 987 r2 0.096 0.070 0.086 Standard errors in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 Again, as before, the results indicate that growth rates for all demand deposits are higher for the IBs. Furthermore, it is clear that government owned PBs have the worst growth rates in the sector. This is possibly because two of the largest PBs are located in areas with relatively low income levels and limited potential for deposit growth. It may also be due to the lower standards of service that associated with public sector banks in most 30 DRAFT emerging markets or the immediate socio-political context of this analysis during which time, Pakistan experienced a particularly chaotic and violent political transition from an unpopular military dictatorship to an unstable and eventually equally unpopular democratic system. Since this political turbulence caused a severe economic crisis and led to further popular dissatisfaction with the government, it is possible that the negative public perceptions of the government may have spilled over into a negative view of public sector banks. On the other hand, foreign banks – both Arab and Western- seem to be growing in much the same way as the locally embedded, older conventional banks. Even though western foreign banks seem to be doing better than Arab banks. This is possibly because the Western banks in Pakistan have been operating locally for a long period of time and as such may have been assimilated into the local financial structure. Finally, as seen previously, smaller, newer, conventional banks seem to be doing as poorly as the public sector banks in terms of deposit growth. Unlike Islamic banks that also started operating recently and enjoy a very positive institutional effect, SLBs have lower deposit growth rates than almost every other type of bank. Finally, Table 7 shows regression estimates for equation (2), testing the impact of the financial crisis on deposit growth. 31 DRAFT Table 7: Regression Estimates for response to Crisis. Islamic vs. all Other Bank Types (1) Demand Deposits Foreign 0.791 (1.668) Arab -0.317 (0.547) Public -2.786*** (0.828) Islamic 2.362 (1.546) Small Local -2.103 (1.788) Crisis -1.482 (2.403) Foreign*Crisis -3.600 (2.361) Arab*Crisis -2.514 (2.160) Public*Crisis -1.575 (2.282) Islamic*Crisis 0.743 (6.250) Small Local*Crisis -3.530 (4.653) yearend 1.649 (1.568) drate 0.114 (0.158) cscore 0.333 (0.208) leverage 0.280*** (0.080) lnavgtotal -0.881 (0.888) logbranches 0.274 (0.865) logyrsopen -0.611 (0.746) _cons 5.872 (6.870) N 987 r2 0.098 Standard errors in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01 32 DRAFT Once again, the results show that the IBs are able to overcome the negative effects of the crisis while all other bank types suffer significant losses. This negative effect is quite large in the case of foreign banks where it appears as though the generally positive reputations associated with foreign banks do not prevent a liquidity shock12. This strong, negative reaction may possibly be because account holders at foreign banks usually come from a higher income group than those at local banks. As such, they may be better informed on the severity of the crisis, more capable of shifting their deposits to alternative destinations relatively quickly and more likely to do so. 5. Discussion Our results show that at least within the context described by our dataset, religious banks are associated with higher growth rates and enjoy a more stable deposit base during periods of economic instability than comparable conventional banks. While the interpretation of these results as evidence that individual religious beliefs can influence financial choices even if they have no direct monetary payoffs seems to fit the data, various alternative explanations are also possible. Alternative Interpretations It is possible that Islamic banking customers may consider these banks safer because they 12believe that in the event of a financial crisis and in the absence of depositor insurance for f o u rse , a n c ia l iv e n risis hat ru p t e d he u rin g ocal ia l risis banks, it Citibank or e rio d hen h ad already com prom ised the is possible that that brand an HSB C w ould have m ore 33 he S reputation of equity associated relevance at anc w estern w ith a other tim es. DRAFT any bank, these banks would be more likely to be bailed out in a religious country like Pakistan than other conventional banks. However, there are no special safety nets in place for IBs and regulatory authorities have categorically denied that these banks could receive special treatment under duress. In fact, the most striking cases of Islamic bank failure – in Egypt during the 1980s and in Turkey in 2001 – have occurred in Muslim majority countries where the depositors did not receive any special help in recovering their losses. Furthermore, at least in Pakistan, if any type of bank were more likely to receive special treatment and be safer than others, it would be one of the older conventional banks that are still deeply embedded within the government and have a strong network of supportive political alliances in place. Alternatively, Islamic banking customers may believe that these banks can earn higher returns via the PLS model than the rates being offered at conventional banks. But since we are looking at simple demand deposits, where returns are either non-existent or trivial, it is difficult to imagine that profitability would be the critical determinant of bank choice. Finally, it is possible that Islamic banking customers simply believe that the shariah compliant banking model is superior to conventional banking alternatives. This explanation seems plausible in light of the 2009 global financial crisis that exposed some of the risks inherent in a highly leveraged, Western banking system. Since Islamic banks prohibit excessive debt, constrain risk taking and investments in financial derivatives along with a host of other restrictions, they may be considered safer alternatives to 34 DRAFT conventional banks and their growth rates could simply reflect better quality, not religiosity. While this may be true, it is also true that growth in IBs preceded the Western financial crisis and shariah compliant baking products were growing at double-digit rates even before weaknesses in conventional banks were exposed. Also, Islamic banking customers are not particularly skewed towards the more financially sophisticated investors in the world. Rather, they tend to exclusively focus on competing and gaining market share in Muslim majority regions13. Conclusion The results in this paper show that religion can affect economic outcomes and that institutional success may be based on a variety of locally specific monetary and nonmonetary factors. While LaPorta et al. (1997) conclude that levels of trust are lower in religious countries because religion deters the formation of horizontal networks of cooperation, this paper shows that religiosity is a more complex variable that can have a different impact in different areas and that may even increase the level of trust in particular institutions. As fa r as ou r a na l ysis ssib l e t ha t rel ig iou s t ru st p remiu m in ist a n w here g en er al h dividu a l s piritual ources . ks ha ve b een grow ing at signi nt l y ess f is concerned, it s banks enjoy t u rb u l en t a rea s l ik e Pa k nstitutional vol at il it y can tds eeking tability owever, 13Isl a mic ban a ra p id pace ot her urb ul ent a l a y sia a m ic anks x t e n siv e l y b y non - Muslims as How ever, this is prim arily has been actively perusing n re they are because the the goal he a se x c e p t io n h e re se d s by M uslim s. M alaysian governm ent of becom ing a oted regional Islam ic banking hub that can serve the financial needs of petrodollar rich, M uslim m ajority Gulf o u n t rie s. s ch, as c t iv e l y ro m sl a m ic banks and offers IB custom ers very favorable term s in order to encourage grow th in this sector. 35 DRAFT M u sl im cou n t ries l ik e B a hra in , t he UAE and Indonesia so it seems l ik el y t ha t t here is more t o t heir p op u l a rit y a con t ex t of econ omic and p ol it ica l t u rmoil . esults st rongl y indica t e t ha t eligiou s p referen ces ma t t er d ca n ha ve a significa nt econ omic imp a ct . At least in the context of a religiously motivated population it makes economic sense to focus on the growth of institutional forms that positively reflect these preferences. 36 DRAFT APPENDIX 1: Rating Scale Equivalence Ratings Numerical Scale AAA 10 AA+ 9 AA 8 AA- 7 A+ 6 A 5 A- 4 BBB+ 3 BBB 2 BBB- 1 Figure 2: Trust in Institutions. The Niel son R esea rch su rvey was con du ct ed from Ju l y - S ep t emb er 2009 a n d el icit ed resp on ses from 1212 you n g ma l e a n d fema l e Pa k ist a n i’ s (18 - 29 yea r ol ds) from A, B and C in come g rou p s a cross ru ra l a n d u rb a n a rea s Responses to Survey Question: How much confidence do you have in Pakistan’s institutions? Source: Nielson. Pakistan: The Next Generation, November 2009. 37 DRAFT Graph 3: The KSE 100 Index. Jan 2006 – Jan 2009. Daily Price movements REFERENCES Alesina, A. and R. Perotti , Taxation and redistribution in an open economy. European Economic Review 39 (1995), pp. 961–979. Barro (1991). Economic growth in a cross section of countries. Quarterly Journal of Economics 106 2 (1991), pp. 407–443. Barro, R.J., McCleary, R.M., 2002. Religion and political economy in an international panel. Manuscript, Harvard University. Bonaccorsi di Patti, Emilia Daniel C. Hardy, (2005) Financial sector liberalization, bank privatization, and efficiency: Evidence from Pakistan, Journal of Banking & Finance, Volume 29, Issues 8-9, Pages 2381-2406 Chiswick B.R. (1983) The earnings and human capital of American Jews. Journal of Human Resources 18 3 (1983), pp. 313–336. Easton and Walker (1997). S.T. Easton and M.A. Walker , Income, growth, and economic freedom. American Economic Review 87 2 (1997), pp. 328–332. Eisenstadt S.N. (1968). The Protestant Ethic and Modernization: a Comparative View, Basic Books, New York (1968). Ellison (1991). C.G. Ellison , Religious involvement and subjective well-being. Journal of Health and Social Behavior 32 1 (1991), pp. 80–99. Freeman (1986). R.B. Freeman , Who escapes? The relation of churchgoing and other background factors to the socioeconomic performance of black male youths from inner- 38 DRAFT city tracts. In: B.F. Richard and J.H. Harry, Editors, The Black Youth Employment Crisis, University. Chicago Press, Chicago, London (1986), pp. 353–376. Grier, R. (1997). The Effect of religion on economic development: a cross-national study of 63 former colonies. Kyklos 50 1 (1997), pp. 47–62. Guiso, Luigi. Paola Sapienza, Luigi Zingales. 2003. People's opium? Religion and economic attitudes, Journal of Monetary Economics, Volume 50, Issue 1, Pages 225-282 Iannaccone, L.R. (1998). Introduction to the economics of religion. Journal of Economic Literature 36 3 (1998), pp. 1465–149 Inglehart (1999), Trust, well-being and democracy. In: M. Warren, Editor, Democracy and Trust, Cambridge University Press, New York, Cambridge (1999), pp. 88–12 J.S. Levin and H.Y. Vanderpool , Is frequent religious attendance really conducive to better health? Toward and epidemiology of religion. Social Science Medical Journal 24 7 (1987), pp. 589–600 Kuran, Timur (1997), The Discontents of Islamic Economic Morality, AER Vol. 86, Issue. 2; pg. 438. La Porta et al (1999). R. La Porta, F. Lopez-de-Silanes, A. Shleifer and R.W. Vishny , The quality of government. Journal of Law Economics and Organization 15 1 (1999) Lessnoff (1994). J.S. Levin , Religion and health: Is there an association, is it valid, and is it causal?. Social Science Medical Journal 38 11 (1994), pp. 1475–1482 Levin and Vanderpool (1987). P. Mauro , Corruption and growth. Quarterly Journal of Economics 110 3 (1995), pp. 681–712. McCleary, Rachel M., and Robert J. Barro. 2006. "Religion and Economy." Journal of Economic Perspectives 20, no. 2: 49-72. Health Business Fulltext Elite, Peek, Joe and Eric S Rosengren. 1997. The international transmission of financial shocks: The case of Japan. The American Economic Review 87, no. 4, (September 1): 495-505. Porta et al (1997). R. La Porta, F. Lopez-de-Silanes, A. Shleifer and R.W. Vishny , Trust in large organizations. American Economic Review 87 2 (1997), pp. 333–338 Putnam (1993). K. Samuelsson Religion and Economic Action: the Protestant Ethic, the Rise of Capitalism, and the Abuses of Scholarship, University of Toronto Press, Toronto (1993). Steuart (1998). Stulz, R., Williamson, R., 2001. Culture, openness and finance. NBER working paper 8222. Stulz and Williamson (2001). R.H. Tawney Religion and the Rise of Capitalism, Harper and Row, New York (1926). T.D. Evans et al., Religion and crime reexamined: The impact of religion, secular controls, and social ecology on adult criminality. Criminology 33 2 (1995), pp. 195–224. Uppal (1986). M. Weber 1956, The Protestant Ethic and The Spirit of Capitalism, Unwin, London (1905). 39