Islamic Banks, Deposit Insurance Reform, and Market Discipline: Evidence from a Natural Framework

Abstract

Although it has been intensively claimed that Islamic banks are subject to more market discipline, the empirical literature is surprisingly mute on this topic. To fill this gap and to verify the conjecture that Islamic bank depositors are indeed able to monitor and discipline their banks, we use Turkey as a test setting. The theory of market discipline predicts that when excessive risk taking occurs, depositors will ask higher returns on their deposits or withdraw their funds. We look at the effect of the deposit insurance reform in which the dual deposit insurance was revised and all banks were put under the same deposit insurance company in December 2005. This gives us a natural experiment in which the effect of the reform can be compared for the treatment group (i.e., Islamic banks) and control group (i.e., conventional banks). We find that the deposit insurance reform has increased the market discipline in the Turkish Islamic banking sector. This reform may have upset the sensitivities of the religiously inspired depositors, and perhaps more importantly it might have terminated the existing mutual supervision and support among Islamic banks.

Full text

D/2017/7012/03
FACULTEIT ECONOMIE
EN BEDRIJFSKUNDE
TWEEKERKENSTRAAT 2
B-9000 GENT
Tel. : 32 - (0)9 – 264.34.61
Fax. : 32 - (0)9 – 264.35.92
WORKING PAPER
Islamic Banks, Deposit Insurance Reform, and
Market Discipline: Evidence from a Natural
Framework
Ahmet F. Aysan – Mustafa Disli – Meryem Duygun – Huseyin Ozturk
January 2017
2017/929

Islamic Banks, Deposit Insurance Reform, and Market Discipline:
Evidence from a Natural Framework
Ahmet F. Aysan – Mustafa Disli – Meryem Duygun – Huseyin Ozturk
Abstract
Although it has been intensively claimed that Islamic banks are more subject to market discipline,
the empirical literature is surprisingly mute on this topic. To fill this gap and to verify the conjecture
that Islamic bank depositors are indeed able to monitor and discipline their banks, we use Turkey
as a test setting. The theory of market discipline predicts that when excessive risk taking occurs,
depositors will ask higher returns on their deposits or withdraw their funds. We look at the effect
of the deposit insurance reform in which the dual deposit insurance was revised and all banks were
put under the same deposit insurance company in December 2005. This gives us a natural
experiment in which the effect of the reform can be compared for the treatment group (i.e., Islamic
banks) and control group (i.e., conventional banks). We find that the deposit insurance reform has
increased market discipline in the Turkish Islamic banking sector. This reform may have upset the
sensitivities of the religiously inspired depositors, and perhaps more importantly it might have
terminated the existing mutual supervision and support among Islamic banks.
JEL codes: G23, G28, O52
Keywords: Depositor discipline, Islamic banks
Mustafa Disli (corresponding author)
Ghent University, Tweekerkenstraat 2, 9000 Gent, Belgium
mustafa.disli@ugent.be
Ahmet F. Aysan
Central Bank of the Republic of Turkey, Turkey
ahmet.aysan@tcmb.gov.tr
Meryem Duygun
Nottingham University Business School, United Kingdom
meryem.duygun@nottingham.ac.uk
Huseyin Ozturk
Central Bank of the Republic of Turkey, Turkey
huseyin.ozturk@tcmb.gov.tr

2
1. Introduction
The motivation behind market discipline is to increase the role of depositors to supplement
regulatory discipline. The growing complexity of banking activities and the move towards market-
based banking provide an explanation for why subsequent Basel reforms have increasingly stressed
the role of market discipline as a pillar for a safe and efficient financial system. Market discipline
refers to a market-based incentive scheme, whereby depositors or other creditors actively reward
or punish banks for their relative performance. The underlying theory of market discipline predicts
that depositors will ask higher returns on their deposits and/or withdraw their funds in response to
declining bank fundamentals. This mechanism of market discipline operates through both price and
quantity adjustments in bank liabilities, which, in turn, would force bank management to lessen its
risk taking (Flannery 1998, Park and Peristiani 1995, Martinez Peria and Schmukler 2001).
The existence of market discipline in conventional banking markets is well-documented
and reflected in the literature. Much of this evidence comes from countries with mature and well-
developed banking systems. Both price and quantity disciplining have been shown to play an
important role as a complement to supervisory efforts, particularly with respect to deposits that are
not fully insured. For example, using a sample of banks in thirty-two OECD countries, Nier and
Baumann (2003) find that riskier banks in general hold a bigger capital buffer, which confirms the
presence of market disciplining behavior in more mature institutional environments. Sironi (2003)
yields a similar conclusion by analyzing the risk sensitivity of European banks' subordinated notes
and debentures spreads. For a sample of both OECD and developing countries, Demirgüç-Kunt
and Huizinga (2004) find that depositors impose market discipline on banks, but to a lesser extent
when there is an explicit deposit insurance system.
The Islamic banking system distinguish itself from the conventional system because interest
(riba) is prohibited in Islam; i.e., banks are not permitted to charge predetermined interest rates on
loans or savings. In accordance with the Shariah, the Islamic legal rules, the Islamic banking model
is based on the profit-and-loss sharing mechanism (PLS), which is typically practiced through
Mudarabah (profit-sharing) and Musharaka (joint venture) contracts. Under the PLS arrangement,
bank assets and liabilities are balanced in such a way that borrowers share profits and losses with
banks, which in turn share these profits and losses with depositors. Given the emphasis on equity
financing, advocates of Islamic banking have argued that the deleveraged nature of Islamic banks

3
contribute to the stability of the financial system (Khan and Mirakhor 1989, Ebrahim and Safadi
1995, Iqbal 1997). Čihák and Hesse (2010), Hasan and Dridi (2011) and Beck et al. (2013),
amongst others, provide empirical evidence that Islamic banks better withstand negative shocks
than conventional banks. Instead of debt, the Islamic banking model introduces asset-backed
financing instruments, where the investor’s return is linked to the profit and loss of a pool of
heterogeneous assets (Askari 2012). In addition, Chapra (1992) and Mills and Presley (1999) argue
that the risk-sharing feature of the PLS paradigm allows Islamic banks to lend on longer-term basis
to projects with better risk-return profiles and, thus, promote economic growth.
More importantly for our study purposes, given the equity-like nature of savings and
investment deposits, it has been strongly claimed that Islamic banks are more subject to market
discipline (e.g., Errico and Farahbaksh 1998, El-Hawary et al. 2004, Beck et al. 2013). In other
words, as ‘quasi-shareholders’, Islamic bank depositors would have greater incentives to exercise
control over management to prevent excessive risk taking behavior. By assessing the presence of
market discipline, this article contributes to the small but growing Islamic finance literature in
several ways. To the best of our knowledge there has been no empirical study examining the
existence of market discipline in a dual banking system, where Islamic and conventional banks
operate side by side. Moreover, the existing literature does not provide us clear insights into how
the disciplining mechanism operates in the Islamic setting. We attempt to fill this gap by examining
depositors’ disciplining behavior in the Turkish banking market. While in a few countries, such as
Iran and Sudan, the entire banking industry operates according to Islamic rules, in many other
countries, such as Bangladesh, Egypt, Malaysia and Turkey, Islamic financial institutions are
running side-by-side with conventional banks. The Turkish case provides an interesting
opportunity to test the dynamics of depositors’ behavior. Turkey was the first country in the world
that had adopted a dual deposit insurance framework in 2001, in which the Islamic deposit
insurance scheme operated alongside its conventional counterpart. Unlike the conventional
scheme, which was administered by the government, the Islamic scheme was organized and
managed by Islamic banks. However, on December 2005, the dual deposit insurance framework
was revised and the Islamic scheme was absorbed by the conventional scheme. We exploit this
natural experiment to study the impact of the unified deposit insurance system on market discipline
using a difference-in-differences approach. By analyzing the Russian banking market, Karas et al.
(2013) adopted a similar empirical strategy in order to identify the differential effect of deposit

4
insurance on the behaviors of insured households and uninsured firms. We instead compare the
presence of market discipline among Islamic banks with a control group of conventional banks that
are not affected by the deposit insurance reform.
Our findings suggest that Islamic bank depositors indeed behave differently than their
conventional peers. In the pre-deposit insurance reform period, we observe that depositors of
conventional banks were sensitive to bank risk, whereas the vigilance of Islamic depositors was
not present. In the post-reform period, however, we find that Islamic bank depositors increased
their sensitivity to bank-specific risks. We interpret these findings as evidence that the reform may
have upset the sensitivities of religiously inspired depositors. Furthermore, the reform may have
terminated the mutual supervision and support among Islamic banks. Robustness checks, in the
form of incorporating the 2001 financial crisis in our analyses, confirm that the presence of the
Islamic deposit insurance scheme had a numbing effect on market discipline.
We organize this article as follows. In Section 2, we describe the disciplining process in the
Islamic banking sector. Section 3 presents a concise history of the dual-banking industry in Turkey.
In Section 4, we describe the sample of banks, and identify the impact of the deposit insurance
reform on market discipline. Section 5 presents the results of the difference-in-difference
estimation strategy. Section 6 extends our sample period, look at the influence of the 2001 financial
crisis on market discipline, and also functions as a robustness analysis. In Section 7, we summarize
our main findings and provide a discussion of the results found.
2. Market discipline in Islamic banks
Although the Islamic finance literature has highlighted the importance of market discipline, still
very little is known about the mechanism through which the disciplining occurs in a profit and loss
sharing framework (Aysan et al. 2015). Particular focus in this context is on the Mudaraba contract,
which involves a partnership between the bank as one of the several investors, with profits and
losses being shared in mutually agreed proportions. This mode of financing is manifested on the
banks’ liabilities side, with investment accounts or deposits that do not yield preset interest rates
but rather confer a proportion of profits. An Islamic deposit contract, in fact, contains neither debt
nor equity-based compensations – and this, in turn, may create additional agency conflicts. Agency

5
problems do not only stem from the separation of ownership and control for shareholders (Berle
and Means, 1932; Jensen and Meckling, 1976; Fama and Jensen 1983), but are also sourced from
the separation of cash flow and control rights for depositors (Safieddine 2009). Although in such
an arrangement depositors turn from bank creditors to residual claimants on banks’ cash flows,
they are not granted the control rights that shareholders enjoy and their cash flow rights are
separated from the rights to control the investments.
Recent corporate finance theories posit that corporate governance is increasingly based on
exit strategies rather than on voice. These theories conjecture that, in the absence of intervention
power, dispersed blockholders can govern firms (Edmans 2009; Admati and Pfleiderer 2009; and
Edmans and Manso 2011). While such a dispersed structure appears to be a barrier for direct
intervention, it strengthens the role of a second governance mechanism: disciplining through
trading. This behavior involves an exit strategy upon negative information, thus leading to a stock
price that closely reflects the firm’s fundamental value. The drop in the stock price involves the
punishment of equity-linked managers, while ex-ante the mere threat of exit raises efficiency as it
induces managers to undertake value-enhancing activities.
The voice through exit mechanism is especially applicable to the Islamic model of banking.
According to this model, deposits are considered as shares (i.e., equity participation) and are thus
entitled to dividends if the investment operates at a profit (Bashir et al. 1993). Although demand
deposits are not tradable, they are highly liquid by their very nature. Deposits typically contain a
withdrawal option embedded with each account, which licenses the depositor to sell the deposit to
the bank at will. Calomiris and Kahn (1991) explain that, in a risk-sharing framework, liquid
deposits substantially help to align the bank’s portfolio choice with depositors’ preferences. If the
bank fundamentals turn out to be weak, depositors still could exercise power over bank
management due to their ability to withdraw their deposits when information indicates poor firm
prospects. Such an exit by depositors can be an effective governance mechanism, even when they
have no direct intervention power in the operation of the banks.
Although a profit-and-loss sharing framework may contribute to market discipline, the
influence of religious commitment on depositors’ sensitivity to bank financial conditions is not
unequivocal. On the one hand, the findings of prior research generally suggest that religious
individuals show more risk-averse characteristics than non-religious individuals. For example,
Miller and Hoffmann (1995) report a negative association at the individual level between religiosity

6
and attitudes towards risk. Similarly, Hilary and Hui (2009) show that corporate policies of US
firms located in more religious regions display lower degrees of risk exposure. Extending this
intuition to the market discipline mechanism, we expect that Islamic bank depositors will be more
vigilant and responsive to bank-specific risks. On the other hand, Abedifar et al. (2013) point that
Islamic depositors may have a strong sense of loyalty toward their banks, thus numbing the
sensitivity to bank riskiness. This loyalty argument is likely to be more relevant in dual-banking
systems where Islamic banks are minor players in the market. In other words, in such banking
systems, religious reasons may play a strong role in the decision to bypass conventional banks and
to deposit funds to Islamic banks. This conjecture has recently been evidenced by Baele et al.
(2014) with respect to the loan market in Pakistan. Their finding of much lower default rates on
Islamic loans than those on conventional loans suggests that the signing of an Islamic loan contract
activates religious moral norms.
In part because of the only recent institution of deposit insurance schemes for Islamic banks,
the literature is also silent regarding the impact of deposit coverage on market discipline. Since the
funds of the deposit insurance scheme may be invested in interest-bearing assets, concerns are
raised about its compliance with Shariah principles (Solé 2007). In addition, because of the
existence of different modes of Islamic deposit insurance, clear-cut interpretations about the impact
of deposit insurance on market discipline are elusive. There are broadly two forms of Islamic
deposit insurance, protection of Islamic deposits through the conventional insurance system or
through the establishment of a separate Islamic deposit insurance scheme. As for conventional
banks, we expect that deposit insurance will reduce the incentives for Islamic depositors to exert
market discipline
1
. However, both insurance forms can have different implications on the
disciplining intensity. In order to mitigate the lack of clarity in the insurability of profit-sharing
deposit accounts, a dual-deposit insurance framework may provide a clear signal to Islamic
depositors that the Islamic scheme is Shariah-compliant. Hence, the dual-scheme may create
1
The impact of deposit insurance on the disciplining behavior of conventional depositors is well-documented. The
existence of deposit insurance entails a trade-off between bank stability and moral hazard. Using cross-country data,
Demirgüç-Kunt and Detragiache (2002) provide evidence that explicit deposit insurance significantly increases the
incidences of financial crisis. Demirgüç-Kunt and Huizinga’s (2004) finding show that this increase in financial
fragility mainly stems from reduced depositors' incentives to monitor and discipline banks. In developing countries,
however, it is frequently observed that deposit insurance schemes are not fully credible. Martinez Peria and Schmukler
(2001) proved that depositors are also concerned about the solvency of the insurance fund by showing that small-
insured depositors still react to bank risk. Prean and Stix (2011) and Disli et al. (2013) show that the credibility of
generous deposit insurance schemes are especially affected in turbulent economic environments.

7
confidence about Shariah-conformity of the reimbursed funds in case of bankruptcy. Conversely,
the conventional insurance scheme, that covers both Islamic and conventional deposits, may face
difficulties in convincing depositors about the reliability of reimbursed funds when a bank failure
occurs. Depending on the intensity of religious devotion, even in the presence of blanket guarantee,
depositors still might discipline their Islamic bank under a conventional insurance scheme.
3. Turkish dual-banking system and deposit insurance reform
Parallel to the growth in the Islamic finance industry worldwide, Islamic banks in Turkey have also
been expanding and attracting new customers. Interest-free banking has long been present in
Turkey and was first made legal in 1983, as part of a plan to draw deposits from religious citizens
and from the Gulf states, under the government of Turgut Özal, a former prime minister and
president of the Turkish Republic. Islamic finance debuted in 1985 with the Bahrain-based
AlBaraka Turk and the Saudi-based Faisal Finans. The Kuwaiti-based Kuveyt Turk began its
operations in 1989. Afterwards, the special finance houses also began lending with domestic
capital, including Anadolu Finans in 1991, Ihlas Finans in 1995, and Asya Finans in 1996.
2
These
intermediaries introduced a different banking model into the system with banning interest on
deposits or loans. Instead, deposits were primarily invested into transactions based upon principles
of markups (Murabaha) and leasing services (Ijhara). To distinguish the Islamic financial
institutions from the conventional banks operating in Turkey, they were given the status of ‘Special
Finance Houses’. Despite being Islamic-compliant, such a name was given with the objective to
soften their Islamic image and to resonate with the ideological sensibilities of the ruling
administration. Initially, these institutions did not enjoy the same regulatory status as conventional
banks
3
, the Savings Deposit Insurance Fund (SDIF) did not cover their deposits, and they also could
not invest in government securities.
In the 1990s, otherwise known as the lost decade, the Turkish economy suffered severe
setbacks from large and volatile international capital flows. The first unpleasant experience was
2
We refer to Aysan et al. (2013) for an overview of the developments in the Turkish Islamic banking sector.
3
Decree No. 83/7506, issued by the Council of Ministers, allowed Islamic banks to operate in Turkey. This decree,
however, could just as easily be revoked by the same authority (Brown 2015).

8
due the 1991 Gulf War, which led to capital flow reversals and bank distress due to Turkey's
closeness to Iraq, and the first major banking collapse occurred in 1994
4
. Although the banking
sector recovered quickly from the 1994 crisis, the economy in Turkey continued to face significant
headwinds. In 1999, to reduce fiscal and monetary imbalances, the Turkish government launched
an exchange rate-based stabilization (ERBS) program with the strong support of the IMF. Massive
capital inflows contributed to the bubble in the stock market, the appreciation of the domestic
currency, the widening of current account deficits, which in turn led to an excessive expansion in
domestic credit. However, interest rates began to rise because of widening current account deficits
and delays in the privatization program. With rates moving up, levered banks were forced to offload
their treasury holdings at sale prices to maintain liquidity in the face of increasing financing costs.
5
Taken together, all these factors caused capital reversals, implying a sharp increase in the overnight
interest rates. In December 2000, an IMF rescue package was needed to ease the tensions in the
financial markets, and a resemblance of stability was returned. However, the confidence in the
disinflation program was completely lost, and the economy teetered on the brink of collapse. The
aftermath of the 2001 crisis witnessed a significant drop in the number of conventional banks from
more than 50 to just 33.
6
The financial crisis did not only affect conventional banks, but also the special finance
houses. Amid the mounting foreign exchange crisis, on 10 February 2001, Ihlas Finans’ license
was revoked by the Banking Regulation and Supervision Agency (BRSA) after charges that it had
siphoned off over $1 billion to its parent holding company (The Economist 2001). Further, since
Ihlas could not invest in government securities, it was exposed to a severe maturity mismatch. With
4
In order to contain budget deficits, the government introduced measures such as caps on Treasury bill rates, and
shifted towards deficit financing through monetization. The curbing of interest rates in the weekly tenders for Treasury
bills produced anxiety in financial markets and a loss of confidence in the government. The drop in interest rates caused
a decline in the profits from uncovered arbitrage opportunities and banks started to close their open positions by buying
foreign exchange in domestic markets. In order to contain the loss of foreign currency reserves, and to defend the
Turkish lira, the Central Bank was forced to heavily intervene in the interbank market and raised the overnight rate to
record levels. These developments undermined an already fragile system, and the banking sector faced panic
withdrawals of bank deposits. In order to restore confidence and prevent further capital outflows, the government had
to institute a blanket deposit guarantee.
5
The problem was compounded in October and early November with rumors of malpractices of some nationalized
banks. In October two more banks – Etibank and Bank Kapital – were brought under the management of the Savings
Deposit and Insurance Fund (SDIF). Moreover, not met privatization targets, due to ideological differences within the
ruling coalition, caused the IMF to postpone a scheduled fund transfer.
6
The fierce turmoil, between President Ahmet N. Sezer and Prime Minister Ecevit in February 2001, completely
agitated the financial markets. Because of the massive capital outflows, the overnight interest rates skyrocketed on
February 21 and the Central Bank had no choice but to abandon the peg. We refer to Akyüz and Boratav (2003) for a
detailed discussion about the development of the Turkish banking sector during the crisis.

9
the outbreak of the liquidity crisis, Ihlas’ exposure led to its collapse when it experienced a
traditional bank run on its deposits. Since there was no insurance to dissuade Islamic depositors
from withdrawing their funds, it was feared that the panic would spread to depositors at other
Islamic financial institutions. The remaining Islamic banks, however, proved to be resilient to the
difficult operating environment. Another development in the same period was the purchase of
Faisal Finans by the Turkish holding company Ülker, changing its name to Family Finans.
Subsequently, in December 2005, Anadolu Finans and Family Finans were merged into Türkiye
Finans, leaving the number of Shariah-compliant banks operating in Turkey to fall to four.
7
The regulatory reform in Turkey actually started in June 1999 under Banks Act 4389. A
new and independent regulatory authority, the BRSA, was established to oversee banking system
stability. However, due to political disputes, the operationalization of the new agency was delayed
until September 2000, when it became a little too late to intervene appropriately.
8
The 1999 bank
law also brought Islamic banks under the same regulatory umbrella as conventional banks.
Although the reform sought to integrate the Islamic banks into the financial system, these banks
were not made part of the conventional deposit insurance scheme but were merely given the right
to create one of their own (Brown 2015). Further, the implementation of an Islamic deposit
insurance scheme did not materialize until the collapse of Ihlas. The Islamic deposit insurance
scheme provided insurance up to 50,000 TL for each deposit ownership in each bank, while the
conventional insurance offered at that time an unlimited coverage. Furthermore, unlike the
conventional system, which was managed by the government, the administration of the Islamic
deposit insurance system was delegated to the ‘Union of Private Finance Houses’. Membership to
the Union was compulsory for all licensed Islamic banks, and as the scheme was backed by Islamic
banks concerns about its Shariah-compliance were allayed. The operation of the system was funded
by premiums received from Islamic banks, and these were assessed based on the amount of both
7
In May 2015, Ziraat Participation Bank, was authorized by the country’s banking regulator (BRSA) to start its
operations as Turkey’s first state-owned Islamic bank, increasing the number of Islamic banks to five.
8
Although the institution of BRSA was not successful in preventing the crisis at the end of 2000, it is considered as
the first of subsequent reforms in the regulation of the banking sector. In May 2001, the BRSA launched the Banking
Sector Restructuring Program (BSRP) to recover fundamental fragilities in the banking sector, and for building a strong
base for the system by clearing it from weak banks (see BRSA 2010).The structural reforms in the banking industry
and the political stability after 2002 have facilitated a significant improvement in overall economic performance. In
the period between 2002 and 2010, Turkey’s public sector debt-to-GDP ratio firmly declined from 70% to 42%, a ratio
that is since 2004 consistently below the Maastricht criterion. The conducive economic environment in the period
2002-2007, with abundant global liquidity, also made Turkish treasury instruments very attractive, and massive capital
inflows steadily helped to reduce interest rates.

10
demand and profit/loss participation accounts (i.e., 0.25% on the ending of quarterly balances
9
).
This pioneering example was praised because of its conformity to Takaful principles, the Islamic
version of insurance, where members cooperate to pool resources in order to guarantee each other
against loss or damage.
However, on December 2005, upon enactment of the Banking Act No. 5411, the dual
deposit insurance system was revised and the management of the Islamic deposit insurance fund
was transferred to the Savings Deposit Insurance Fund (SDIF). Following amendments to the
banking law, ‘Special Finance Institutions’ were renamed as ‘Participation Banks’, which allowed
them to integrate fully into the financial system. Premiums paid by Islamic banks to the SDIF were
equalized with those of conventional banks, and the premium ratio had initially been calculated as
of 0.15% of the deposit amount covered under the deposit insurance scheme (current and profit/loss
participation accounts for Islamic banks and current and savings accounts for conventional
banks).
10
The Law, however, did not require to separate the premiums from conventional and
Islamic banks, nor did it require the investment of premiums into Shariah-compliant instruments
(Brown 2015). Since the Islamic fund is not separately managed anymore nor it is in accordance
with the Shariah rules, serious concerns have been raised in the properly resolution of failed Islamic
institutions.
Participation banking in Turkey has not traditionally made up a large portion of Turkey’s
finance sector due to the secular tradition of the republic. However, by becoming more and more
like banks in both image and reality, they progressively gained acceptance among depositors and
investors. Also, after the 2001 crisis, the ruling Justice and Development Party (AK Party) paved
the way to the ascent of Islamic finance. As illustrated in Table 1, slowly but surely, and even more
pronouncedly after the Banking Act of 2005, the sector has managed to increase its market share
both on the credits and deposits segments of the financial industry.
< INSERT TABLE 1 AROUND HERE>
9
For purposes of completeness, Islamic banks with higher risk profiles could end-up paying 0.26% on their end-of-
quarter deposit liabilities.
10
From 2009 onwards, the SDIF has implemented for both banking models a more risk weighted deposit insurance
premiums with a premium ratio varying between 0.11% and 0.19%. The maximum deposit insurance coverage for
both Islamic and conventional bank deposits was set at 50,000 TL, but was in February 2013 doubled to 100,000 TL.

11
4. Data analysis and empirical strategy
We collect an unbalanced panel of 48 commercial banks operating in Turkey from the various
issues of Banks in Turkey published by the Banks Association of Turkey. This publication includes
quarterly balance sheet and income statement information from 2002:4 to 2012:4. The fourth
quarter of 2002 corresponds to the effective start of the Erdogan-era. Of the 48 conventional banks,
23 banks are branches of foreign banks or are classified as foreign subsidiaries (more than 50% of
their shares are owned by non-residents), 22 banks are domestically owned commercial banks
(more than 50% of their shares are owned by Turkish residents), and 3 are classified as state-owned
deposit-taking banks (predominantly owned by the Turkish government). The Islamic bank data
contains an unbalanced panel of 6 banks and is obtained from the Central Bank of the Republic of
Turkey.
11
As with most similar studies, we analyze both the price and quantity reactions to bank risk
since this joint information allows us to better identify disciplining behavior. The use of this
combined information will help us to disentangle depositor discipline from demand shifts (e.g.,
Ioannidou and de Dreu 2006, Karas et al. 2013). A positive relation between bank risk and deposit
rates could reflect a demand effect rather than discipline, with risky banks pursuing a more
aggressive expansion strategy to meet new loan demand. But this would be discovered by looking
at the quantity regression, where the relation between bank risk and deposit quantity would also be
positive in case of a demand effect, and negative in case of true depositor discipline.
So far, most published studies have analyzed the differences in behavior to bank risk
between insured and uninsured deposits and have ascribed behavioral differences to the presence
of insurance. In their approach, however, it cannot be dismissed that other depositor group-specific
characteristics may possibly explain the observed differences as well. Other studies provide
empirical evidence on the incentive impact of deposit insurance by comparing the behavior of a
particular depositor group by looking at the before and after deposit insurance introduction. But
their approach cannot reject the possibility that the results are driven by other time specific
unobservable characteristics. The difference-in-differences approach allows us to study the effect
11
For both Islamic as well as conventional banks, these figures includes adjustments for mergers & acquisitions by
generating a new bank after tracing such an event.

12
of treatment, in this case the unification of the dual deposit insurance system, by comparing the
depositor sensitivity to risk of the treatment group (i.e., Islamic banks) pre- and post-treatment
relative to the depositor sensitivity to risk of the control group (i.e., conventional banks) pre- and
post-treatment. By comparing changes, we control for time-invariant characteristics that might
affect Islamic and conventional bank depositors differently and for time-varying factors that might
affect them in a similar fashion.
Our identification strategy is the most related to Karas et al. (2013). They employ a similar
difference-in-differences methodology to identify the differential impact of deposit insurance on
the behavior of insured households and uninsured firms and find evidence that insurance diminishes
the insured depositors’ sensitivity to bank risk.
We employ the following reduced-form difference-in-differences models:













































































(1)













































































(2)
Where ISL = 1 for Islamic banks and ISL = 0 for conventional banks. The variable REF is a
dichotomous variable for the post-treatment period, i.e., after the unification of the insurance
system in 2005Q4. From the perspective of depositors, we can fairly claim that the deposit
insurance reform constitutes an exogenous change and its impact only applies to Islamic banks but
not to conventional banks. The reaction variables are the traditional measures used in the depositor
discipline literature. The dependent variable in Eq. 1 is the first difference of the log of deposits of
bank type j (Islamic or conventional) for bank i during period t, and the dependent variable in Eq.
2 is the quarterly expenses for bank deposits divided by total deposits of type j for bank i during
period t. For both Islamic and conventional banks, according to the market discipline hypothesis,
the return on deposits should in principle be a reflection of the risk profile.
The  represents a matrix of bank fundamentals that might be of interest to depositors
concerned about bank safety. This vector contains capitalization, our primary indicator of bank-
level risk, bank liquidity, and non-performing loans. These variables are included with a one-
quarter lag to account for the delay in publication. The CAP variable is measured as the capital-to-

13
assets ratio. The LIQ and NPL variables are calculated as liquid assets to total assets, and the ratio
non-performing loans to assets, respectively. Although we focus on the sensitivity of depositors to
bank capitalization, we estimate the most flexible specification by integrating each of these
fundamentals directly (coefficient  in Eq. 1/! in Eq. 2) as well as with three separate interaction
terms (coefficients ,  and  in Eq. 1/!, !, and ! in Eq.2) in each difference-in-differences
specification (Eq. 1 or Eq. 2).
Our primary measure of a bank’s risk level is its capital-to-assets ratio. Following the
introduction of the 1988 Basel Accord and the 1996 Market Risk Amendment, the importance of
bank capital buffers to financial safety has been emphasized and capital management has become
the main channel through which banks manage their risk of insolvency (Nier and Baumann 2006).
Further, more than any other measure, the capital ratio is extensively used as a proxy for bank risk
taking in market discipline studies, in both developed and emerging market economies (e.g.,
Hannan and Hanweck 1988, Park and Peristiani 1998, Martinez Peria and Schmukler 2001, Karas
et al. 2013, Disli et al. 2013, Berger and Turk-Ariss 2014). We also refer to Disli and Schoors
(2013) and Disli et al. (2013) who found that only bank capital proved to be unambiguously leading
to depositor discipline in the Turkish conventional banking market. From the perspective of
depositors, this simple but powerful indicator mitigates information asymmetries since a bank’s
decision to hold more capital subject its owners to a greater loss in case of failure. As holding
capital encourages banks to undertake less risk, depositors will reward these banks by supplying
more funds at lower deposit rates.
Evidence of depositor discipline requires " # and !$ # with increased bank
capitalization for conventional banks in the pre-treatment period. The slope parameters  and !
seize the difference in means between Islamic and conventional banks before the treatment takes
place. If the signs of  and ! are similar to the signs of  and !, respectively, Islamic depositors
are more vigilant to bank risk than their conventional counterparts. If on the other hand,  and !
have opposite signs to that of  and !, respectively, Islamic depositors are not as much sensitive
to bank risk as conventional depositors. The parameters  and ! capture for the conventional
banking market the change in depositor sensitivity in the unified deposit insurance period. The
coefficients  and ! quantify the additional shift in the sensitivity of Islamic depositors after the
treatment.

14
The % contains other bank specific controls potentially affecting the reaction variables.
The Bank size variable is calculated as the natural logarithm of total assets. As a measure for
institutional maturity, we define Bank age as the natural logarithm of quarter-years. The variable
Branch size is the average number of employees per branch and is used as a metric of bank service
quality. Since listed banks expose themselves more to public scrutiny, we also control for this and
use a dummy variable which equals to 1 when banks are listed in the Istanbul Stock Exchange
(Borsa Istanbul).
We estimate a model using bank-fixed effects, , for the quantity reaction, and !, for the
return reaction to control for unobserved characteristics across banks. Furthermore, in all
specifications, we include quarter dummy variables per bank type j (Islamic or conventional) to
account for nation-wide shocks that may have a different effect on the two types of depositors (i.e.,
& in the deposit growth equation and !& in the price equation).
Table 2 provides summary statistics on variables used in this analysis. The table compares
the treatment group of Islamic banks to the control group of conventional banks for the period
before as well as after the deposit insurance reform. Although the difference diminished in time,
we ascertain that Islamic banks were able to attract more deposits than their conventional
counterparts. Interestingly, they were able to do so even though they offered lower returns on
deposits. In both periods, Turkish Islamic banks exhibit poor fundamentals vis-à-vis their
conventional peers in terms of bank capitalization (CAP) and liquidity positions (LIQ). It seems,
however, that Islamic banks achieved better loan quality (NPL) than conventional banks.
Conventional banks have on average more personnel per branch than Islamic banks for both
periods. Finally, after the reform, two Islamic banks have been listed on the Istanbul Stock
Exchange, while there were no listed Islamic banks in the period before.
<INSERT TABLE 2 AROUND HERE>
5. Empirical results
Table 3 presents the results from the difference-in-differences estimations. Columns (1-4) exhibit
the estimation results of Eq. 1; columns (5-8) refer to the estimation results of Eq. 2. To test whether
our results are sensitive to alternative sample compositions, for each equation we produce four sets

15
of estimates. Columns (1) and (5) report the results for all banks with different ownership types. In
columns (2) and (6) we exclude state-owned banks since they provide depositors weaker incentives
for monitoring and disciplining (Caprio and Honohan 2004). Columns (3) and (7) restrict the
sample to foreign-owned banks as a control group, while in columns (4) and (8) compare the
behavior of Islamic depositors to that of their privately-owned peers.
Prior to the unified deposit insurance implementation, we in general observe that depositors
of conventional banks were sensitive to bank capitalization (CAP). An increase in the capital ratio
is associated with higher deposit growth ( " # and statistically significant for different sample
selections) and lower deposit rates (!$ # and statistically significant for different sample
selections except for column 8), which provides direct evidence of depositor discipline.
12
As to
whether Islamic depositors were more or less sensitive to bank capitalization than conventional
depositors prior to the deposit insurance reform (see the coefficient estimate of CAP x ISL), the
evidence suggests that quantity sensitivity to capitalization was annihilated (i.e.,  + ). The
parameter estimates of  and ! reveal the change in depositor sensitivity in the unified deposit
insurance scheme period for conventional banks (coefficient estimate of CAP x REF).
<INSERT TABLE 3 AROUND HERE>
In what follows, given our primary focus, we concentrate on the main coefficients of interest ( 
and !). In the quantity equation, through different samplings (except in columns 4 and 8), we
observe that the difference-and-differences coefficient ( ) is consistently positive and highly
significant. We also find that ! in general enters the deposit rate equation with negative
coefficients. These findings suggest that in general the sensitivity of Islamic bank depositors,
relative to conventional depositors, have increased substantially after the reform. Alternatively, we
might describe the coefficients  and ! as capturing the commonly felt impact of depositor
insurance reform on depositor vigilance, whereas the difference-in-difference coefficients seize the
differential impact on the vigilance of Islamic depositors. In sum, our results reveal that, in the pre-
reform period, conventional bank depositors were more sensitive to bank capitalization as predicted
by the market discipline theory: banks with lower capital ratios attracted fewer deposits while they
12
Conversely, depositors will punish banks with a lower capital ratio by decreasing the supply of funds, i.e., raising
the average yield on deposits and reducing the quantity of deposits.

16
paid higher risk premiums. On the other hand, Islamic banks were in the same period not disciplined
at all. The deposit insurance reform, however, consistently produced more sensitive Islamic
depositors to bank risk.
13
Confirming the importance of the capital ratio, other bank fundamentals (NPL and LIQ),
and their interactions with ISL and/or REF, do not allow us to draw clear conclusions concerning
their influence on market discipline.
14
Turning to our control variables, larger banks (Bank size) do
benefit from too-big-to fail effects as they on average succeed in attracting more deposits while
they do not pay higher returns on deposits. We do not find evidence that older banks (Bank age)
produce advantage in terms of deposits and interest rates compared to their younger counterparts.
Further, bank service quality (Branch size) does not seem to influence deposit flows nor the
expenses on deposits, while the deposit expenses of listed banks (Listed) were on average lower.
6. Banking crisis, dual deposit insurance and market discipline
6.1. Before and after the 2001Q1 crisis
Turkey's relatively large and historically troubled conventional banking sector had been covered
by deposit insurance since 1983; as we have noted, combined with problems of insider lending,
weak monitoring and control mechanism, provision of implicit and explicit government guarantees
had left the banking sector in a particularly vulnerable position. Although Islamic banks were not
much affected by the 1994 financial crisis, the 2001 crisis shook also this segment of the industry.
In the wake of the 2001 banking crisis, Ihlas Finans, the then largest finance house, faced a run on
its deposits. In February 2001, the BRSA revoked the operating license of Ihlas Finans on the
13
The estimation results for the sample composition of private and Islamic banks, columns 4 and 8 of Table 3, reveal
that while the signs of the estimated coefficients for the capital ratio and for its interactions are similar to those of
alternative sample compositions, most of them are not statistically significant.
14
Although conventional banks with high performing loans (NPL) face deposit withdrawals, it also appears to decrease
the interest rate that depositors demand from these banks. A similar finding was observed in the Columbian banking
sector (Barajas and Steiner 2000). In the post-reform period, our findings indicate that the conventional deposit growth
sensitivity to NPL has been reversed. On the other hand, albeit heavily reduced in the post-reform period, Islamic
depositors were sensitive to non-performing loans in the manner predicted by the market discipline hypothesis. The
effect of liquidity (LIQ) on the reaction of depositors, Islamic as well as conventional, is not consistent with the market
discipline hypothesis.

17
grounds that it failed to fulfill its liabilities. However, unlike the conventional deposits, the deposits
of Islamic banks did not enjoy insurance coverage with the rationale that profit-and-loss accounts
involved no guarantee of return (Starr and Yilmaz 2007).
15
In a way, the actual creation of the
Islamic deposit insurance scheme right after the collapse of Ihlas provides us a second natural
experiment: both banking models were affected by the crisis with the difference being that
conventional banks enjoyed deposit insurance, whereas Islamic banks only did so right after the
collapse Ihlas.
We first estimate the sensitivity of deposit flows to bank financial conditions, allowing for
different sensitivities across the two banking models and two distinct periods: before and after the
2001Q1 crisis. Since expenses on Islamic deposit accounts were not available before 2002, we
cannot estimate the equivalent price equation. Specifically, we estimate the following panel
specification:













































































(3)
As in Eq. 1, the dependent variable is the first difference of the log of deposits of bank type j for
bank i during period t. The right-hand side variables are the same as in Section 4.2., with the
exception that the deposit insurance reform dummy variable (REF) is now replaced by the financial
crisis (CRIS) dummy variable. The financial crisis dummy variable takes the value of 1 for the
period after the collapse of the banking sector (2001Q1 and thereafter), and 0 otherwise. The
coefficient  capture the sensitivity of deposit growth to bank capitalization for conventional
banks before the financial crisis. The slope parameter  represents the difference in means
between Islamic and conventional banks before the crisis. The coefficient  seize the effect of a
macroeconomic shock – that is, the banking crisis – on the vigilance of conventional (wholly) and
Islamic depositors (partly, in conjunction with ). The coefficient  (the difference-in-
differences coefficient) captures the differential effect of Islamic deposit insurance on the vigilance
of Islamic depositors.
15
The liquidation process of Ihlas would drag on and, as of today, has still not been settled in full. Profit and loss
sharing participation accounts were fourth in line for repayment after debts owed to the government, personnel
payments and current account debts. On December 31, 2013, 20,780 depositors were still waiting to be serviced.
Further details about the payout schedule can be found at http://www.ifk.com.tr/ (last visited on May 25, 2015).

18
Since the outbreak of the financial crisis occurred concurrently with the introduction of the
Islamic deposit insurance scheme, we are not able to disentangle their potential opposing effects
on market discipline. However, it might be possible that in the immediate aftermath of the crisis,
the wake-up call effects might transcend the numbing effect of the newly instituted deposit
insurance scheme. Therefore, we estimate the model for two sample periods: 1998Q1-2002Q3 (the
period before AK Party came into power) and 1998Q1-2005Q3 (the period before the unification
of deposit insurance schemes).
Table 4 reports the estimation results of Eq. (3). Columns (1)-(4) present the results for the
1998Q1-2005Q3 sample period, whereas Columns (5)-(8) display the results for the shorter sample
period (1998Q1-2002Q3). We estimate the most flexible specification with capital (CAP), loan
quality (NPL), and liquidity (LIQ) and full set of interactions. In order to facilitate the interpretation
of the results, however, we only report the coefficient estimates of CAP and its interactions with
ISL and/or CRIS since the other fundamental variables do not allow us to draw meaningful
conclusions concerning their influence on market discipline.
16
As in Table 3, each column in both
panels reflect the estimation results for alternative sample compositions. Our results in general
indicate that, prior to the outbreak of the financial crisis, conventional ( " #) as well as Islamic
(  " #) depositors were sensitive to bank capitalization. Even though the conventional
depositors were backed by a deposit insurance fund, they continued to impose market discipline on
their banks. This finding is in line with the results of Martinez Peria and Schmukler (2001) who
reported that even insured depositors may respond to bank risk if they are concerned about the
insurance fund solvency, or that depositors still feared costs related to the recovery of deposits in
case of failure (i.e. costs due to late payments and the foregone interest earnings). Likewise, for a
sample of banks in CEEC countries, the recent study of Hasan et al. (2013) found that global
financial crisis did not alter the sensitivity of capital flows to bank risk.
17
Before the crisis, although somewhat weakened when considering the shorter sample
period, Islamic depositors behaved in accordance to the market discipline hypothesis. More
importantly, in the post-crisis period, with the introduction of an Islamic deposit insurance scheme,
16
Full results are available upon request. Disli et al. (2013) and Karas et al. (2013) employed a similar projection of
their results in their analyses of market discipline in the Turkish and Russian banking sector, respectively.
17
Fueda and Konishi (2007) show that depositor sensitivity in Japan was most significant in the period 1997-2001
despite the presence of a blanket guarantee. Likewise Forssbæck (2011), who analyzed several hundred banks
worldwide over the period 1995-2005, could not find evidence of increased depositor sensitivity during financial crises.

19
and independent of the examined sample period, the capital ratio lost its indicative power to
discipline Islamic banks.
18
Our findings indicate that the wake-up call effect of Ihlas’ failure on
Islamic depositors was substantially muted with the creation of an Islamic deposit insurance
scheme.
19
Furthermore, at the onset of the failure of Ihlas, Starr and Yilmaz (2007) show that the
deposit levels of a competing Islamic bank quickly recovered after an initial decline. Rather than
raising doubts about the Islamic banking conduct, it appears that Islamic depositors perceived the
Ihlas case as an isolated event.
20
This finding confirms our previous observation that market
discipline was not present in the Islamic banking segment before the unification of deposit
insurance schemes, i.e., in the period where the Islamic deposit insurance scheme was instituted.
<INSERT TABLE 4 AROUND HERE>
6.2. Full sample period
In this section, by making use of the full sample period, we integrate the two previous analyses into
one. Specifically, we estimate the sensitivity of deposit growth to bank risk factors by looking at
the differences in sensitivities across the two depositor groups and three distinct periods: before the
2001Q1 crisis, the 2001Q1 (outbreak of the crisis) and 2005Q3 (before deposit insurance reform)
period and the post-reform period. For this purpose, we estimate the following regression model:



































'











(

























'


)










(















(4)
with the dependent variable being the first difference of the log of deposits of bank type j for bank
i during period t. The right-hand side variables include the same bank fundamentals (X) and bank
controls (C) as in the previous (sub)sections. While the period before the 2001Q1 crisis serves as a
reference period, we define P2 as equal to 1 for the period 2001Q1-2005Q3 and 0 otherwise, P3 is
18
A notable exception is the sample composition of foreign and Islamic banks (columns 3 and 7 of Table 4): although
the sign of the coefficient of CAP x ISL x CRIS is similar to those of other sample compositions, it is not statistically
significant.
19
We refer to Cubillas et al. (2012) and Berger and Turk-Ariss (2014) for evidence about the weakening impact of
government intervention on market discipline.
20
Likewise no bank panic occurred when banking regulators took over management of Bank Asya in February 2015.

20
equal to 1 for the period 2005Q4 to 2012Q4 and 0 otherwise. The differences in deposit flow
sensitivities to bank risk across the two depositor groups and the three periods are summarized as
follows:
Conventional Banks Islamic Banks
Period 1 (1998Q1-2000Q4)







Period 2 (2001Q1-2005Q3)





*





+
+
*





+
Period 3 (2005Q4-2012Q4)





*





+
+
*




)
+
The estimation results are reported in Table 5. As in the previous tables, the different columns
represent different sample compositions.
21
Prior to the crisis, in Period 1, we observe a positive
relationship between deposit flows and bank capital for both modes of banking, and in fact no
difference in market disciplining between them (the coefficient  of CAP is positive, whereas the
coefficient  of the interaction term CAP x ISL is non-significant). In the post-crisis period, in
Period 2, conventional depositors kept disciplining their banks (  , i.e., the sum of the
coefficients of CAP and CAP x P2, respectively, is positive for all different sample compositions).
Results for Islamic depositors, however, indicate that the capital ratio had not an effect on the
direction of deposit flows in a way that market discipline operates. In the post-deposit insurance
reform period, in Period 3, the sensitivity of conventional deposits to bank capitalization is positive.
More importantly, it seems that the unification of the deposit insurance schemes has restored
market discipline of Islamic depositors (sum of the coefficients *  + and *  )+ is positive
and statistically significant for all sample compositions), confirming our previous findings. In the
full sample estimation (1998Q1-2012Q4), we, in fact, show that sensitivity of deposit flows to bank
capitalization between Islamic and conventional depositors is only different in the period when
Islamic banks had to operate under an Islamic deposit insurance scheme. This finding implies that
the control group of conventional banks constitutes a valid counterfactual, which is in support for
the difference-in-differences estimator used in this study.
<INSERT TABLE 5 AROUND HERE>
21
In order to facilitate the interpretation of the results, we only report the coefficient estimates of CAP and its
interactions with ISL and/or time frame dummies (Period 2 and Period 3) since the other fundamental variables do not
allow us to draw clear conclusions concerning their influence on market discipline.

21
6. Discussion and concluding remarks
Our results reveal that, with the introduction of an Islamic deposit insurance scheme, Islamic bank
depositors did behave differently than their conventional counterparts and that these differences
are substantial. More specifically, in the period when Islamic deposit insurance was introduced,
and compared with their conventional peers, we consistently observe that the sensitivity of Islamic
depositors to bank risk was annihilated. More specifically, we do not observe risk aversion of
Islamic depositors as they do not exhibit behavior that is consistent with both quantity and price-
based disciplining. At first glance, our results might seem puzzling why Islamic depositor reactions
are different from their conventional peers, and subsequently switch decisions with the unification
of the deposit insurance. However, this is less so if we take into account the very nature of the
deposit insurance reform. This finding is likely due to the specific design of the Islamic deposit
insurance scheme that was created immediately after the crisis. This scheme was administered by
the ‘Union of Private Finance Houses’, consisting of only a handful of Islamic banks. The Union
was empowered to act as the resolution authority for non-viable banks, and was charged to promote
sound risk management practices over its members. The Islamic deposit insurance fund was interest
free and Sharia compliant, and once the license of an Islamic bank was revoked, the liquidation had
to be executed by the Union, and the insured deposits had to be settled by the Islamic insurance
fund. Furthermore, the Union was mandated to detect early warning signals so that it could
intervene timely in the resolution of troubled banks. By remaining silent, Islamic depositors may
have delegated their monitoring responsibility to this Union. It can be argued that the replacement
of market supervision by mutual supervision among member banks may be simply a consequence
of rational behavior. Particularly, the stakes of Islamic banks were large in case of a failure since
the scheme was backed by them, and the limited number of banks was conducive to robust
oversight.
Conversely, in the post-reform period, we observe that Islamic depositors have started to
discipline their banks. In other words, if the bank fundamentals turn out to be weak, depositors
indeed want to exercise power over bank management by withdrawing their funds. One explanation
for this behavioral shift is that depositors may have started to hesitate about the Shariah-compliance
of the deposit insurance under SDIF protection. This is a viable argument since the purpose of a

22
separate Islamic deposit insurance scheme, which avoided the use of premiums paid by
conventional banks, was to signal a radical break from the conventional banking system in order
to soothe the sensitivities of the religiously inspired depositors. Hence, in Turkey, the Islamic
deposit insurance allowed Islamic banks to give more security to their depositors, while keeping
their operations ostensibly interest-free (El-Gamal and Inanoglu, 2000).
The Islamic deposit insurance was instituted as a response to the failure of Ihlas Finans in
2001. By expanding the temporal scope of our analysis, i.e., including the time period before the
2001 financial crisis, we confirm our findings that depositors’ sensitivity to the Islamic banks’
capitalization was substantially muted with the introduction of an Islamic deposit insurance
scheme.
We also can state that the risk aversion of depositors is contingent on the Shariah-
compliance of Islamic banking. If depositors become suspicious about the operating environment
of banks, they become more vigilant and pay greater attention to bank risk. This intuition suggests
that the degree of risk aversion depends on the religious commitment of Islamic banks. In
environments where the commitment is high, loyalty may win over risk aversion (and vice versa).
Since the new deposit insurance lacks religious quality, Islamic depositors show a greater tendency
to discipline their banks (risk aversion). From a regulatory point of view, perhaps unintentionally,
the reform has increased market discipline among Islamic depositors, which has reinforced the first
line of defense for a safe and sound banking system.
References
Abedifar, P., Molyneux, P., & Tarazi, A. (2013). Risk in Islamic banking. Review of Finance,
17(6), 2035-2096.
Admati, A.R., & Pfleiderer, P. (2009). The “Wall Street Walk” and shareholder activism: Exit as a
form of voice. Review of Financial Studies, 22(7), 2645-2685.
Akyüz, Y., & Boratav, K. (2003). The making of the Turkish financial crisis.World
Development, 31(9), 1549-1566.
Askari, H. (2012). Islamic finance, risk-sharing, and international financial stability. Yale Journal
of International Affairs, 7(1), 1-8.

23
Aysan, A. F., Dolgun, M. H., & Turhan, M. I. (2013). Assessment of the participation banks and
their role in financial inclusion in Turkey. Emerging Markets Finance and Trade, 49(sup5), 99-
111.
Aysan, A. F., Disli, M., Ozturk, H., & Turhan, I. M. (2015). Are Islamic Banks Subject to Depositor
Discipline?. The Singapore Economic Review, 60(01), 1550007.
Banking Regulation and Supervision Agency BRSA (2010). From Crisis to Financial Stability
(Turkey Experience). Working Paper (Revised third edition).
Barajas A, Steiner, R (2000) Depositor Behaviour and Market Discipline in Colombia. Working
Paper WP/00/214, International Monetary Fund.
Bashir, A., Darrat, A. F., & Suliman, M. O. (1993). Equity capital, profit sharing contracts, and
investment: theory and evidence. Journal of Business Finance & Accounting, 20(5), 639-651.
Baele, L., Farooq, M., & Ongena, S. (2014). Of religion and redemption: Evidence from default on
Islamic loans. Journal of Banking & Finance, 44, 141-159.
Beck, T., Demirgüç-Kunt, A., & Merrouche, O. (2013). Islamic vs. conventional banking: Business
model, efficiency and stability. Journal of Banking & Finance, 37(2), 433-447.
Berle, A.A., & Means, G.G.C. (1932). The modern corporation and private property. Transaction
Books.
Berger, A. N., & Turk-Ariss, R. (2014). Do depositors discipline banks and did government actions
during the recent crisis reduce this discipline? An international perspective. Journal of Financial
Services Research, 48(2), 103-126.
Brown Jr, J.R. (2014). Secularism, Sharia, and the Turkish Financial Markets. Brooklyn Journal
of International Law, 40(2), 407-463.
Calomiris, C. W., & Kahn, C. M. (1991). The role of demandable debt in structuring optimal
banking arrangements. The American Economic Review, 81(3), 497-513.
Caprio, G., & Honohan, P. (2004). Can the unsophisticated market provide discipline? Policy
Research Working Paper No. 3364. World Bank.
Chapra, M.U. (1992). Towards a Just Monetary System. Leicester: The Islamic Foundation.
Čihák, M., & Hesse, H. (2010). Islamic banks and financial stability: an empirical analysis. Journal
of Financial Services Research, 38(2-3), 95-113.
Cubillas, E., Fonseca, A. R., & González, F. (2012). Banking crises and market discipline:
International evidence. Journal of Banking & Finance, 36(8), 2285-2298.

24
Demirgüç-Kunt, A., & Detragiache, E. (2002). Does deposit insurance increase banking system
stability? An empirical investigation. Journal of Monetary Economics, 49(7), 1373-1406.
Demirgüç-Kunt, A., & Huizinga, H. (2004). Market discipline and deposit insurance. Journal of
Monetary Economics, 51(2), 375-399.
Disli, M., & Schoors, K. (2013). Bank rebranding and depositor loyalty. Ghent University Working
Paper No. 2013/867.
Disli, M., Schoors, K., & Meir, J. (2013). Political connections and depositor discipline. Journal of
Financial Stability, 9(4), 804-819.
Ebrahim, M.S., & Safadi, A. (1995). Behavioral norms in the Islamic doctrine of economics: a
comment. Journal of Economic Behavior & Organization, 27(1), 151-157.
Edmans, A. (2009). Blockholder trading, market efficiency, and managerial myopia. The Journal
of Finance, 64(6), 2481-2513.
Edmans, A., & Manso, G. (2011). Governance through trading and intervention: A theory of
multiple blockholders. Review of Financial Studies, 24(7), 2395-2428.
El-Gamal, M., & Inanoglu, H. (2000). Islamic banking in Turkey: boon or bane for the financial
sector. In Proceedings of the Fifth Harvard University Forum on Islamic Finance (pp. 7-20).
Center for Middle Eastern Studies, Harvard University.
El-Hawary, D., Grais, W., & Iqbal, Z. (2004). Regulating Islamic financial institutions: the nature
of the regulated. World Bank Policy Research Working Paper No. 3227. World Bank.
Errico, L., & Farahbaksh, M. (1998). Islamic banking: issues in prudential regulations and
supervision. IMF Working Paper No. WP/98/30. International Monetary Fund.
Fama, E.F., & Jensen, M.C. (1983). Separation of ownership and control. Journal of Law and
Economics, 26(2), 301-325.
Flannery, M. J. (1998). Using market information in prudential bank supervision: A review of the
US empirical evidence. Journal of Money, Credit and Banking, 30(3), 273-305.
Forssbaeck, J. (2011). Ownership structure, market discipline, and banks’ risk-taking incentives
under deposit insurance. Journal of Banking & Finance,35(10), 2666-2678.
Fueda, I., & Konishi, M. (2007). Depositors’ response to deposit insurance reforms: Evidence from
Japan, 1990–2005. Journal of Financial Services Research, 31(2-3), 101-122.

25
Hasan, M., & Dridi, J. (2011). The effects of the global crisis on Islamic and conventional banks:
A comparative study. Journal of International Commerce, Economics and Policy, 2(2), 163-
200.
Hasan, I., Jackowicz, K., Kowalewski, O., & Kozłowski, Ł. (2013). Market discipline during crisis:
Evidence from bank depositors in transition countries. Journal of Banking & Finance, 37(12),
5436-5451.
Hannan, T.H., & Hanweck, G.A. (1988). Bank insolvency risk and the market for large certificates
of deposit. Journal of Money, Credit, and Banking, 20(2): 203-211.
Hilary, G., & Hui, K. W. (2009). Does religion matter in corporate decision making in America?.
Journal of Financial Economics, 93(3), 455-473.
Ioannidou, V., & De Dreu, J. (2006). The impact of explicit deposit insurance on market discipline.
De Nederlandsche Bank.
Iqbal, Z. (1997). Islamic financial systems. Finance and Development, 34, 42-45.
Jensen, M.C., & Meckling, W.H. (1976). Theory of the firm: Managerial behavior, agency costs
and ownership structure. Journal of Financial Economics, 3(4), 305-360.
Karas, A., Pyle, W., & Schoors, K. (2013). Deposit insurance, banking crises, and market
discipline: Evidence from a natural experiment on deposit flows and rates. Journal of Money,
Credit and Banking, 45(1), 179-200.
Khan, M. S., & Mirakhor, A. (1989). Islamic banking: Experiences in the Islamic Republic of Iran
and in Pakistan. IMF Working Paper No. WP/89/12. International Monetary Fund.
Martinez Peria, M.S., & Schmukler, S. L. (2001). Do depositors punish banks for bad behavior?
Market discipline, deposit insurance, and banking crises. The Journal of Finance, 56(3), 1029-
1051.
Miller, A.S., & Hoffmann, J.P. (1995). Risk and religion: An explanation of gender differences in
religiosity. Journal for the Scientific Study of Religion, 34(1), 63-75.
Mills, P.S., & Presley, J.R. (1999). Islamic finance: Theory and practice. Macmillan, London.
Nier, E., & Baumann, U. (2006). Market discipline, disclosure and moral hazard in banking. Journal
of Financial Intermediation, 15(3), 332-361.
Park, S., & Peristiani, S. (1998). Market discipline by thrift depositors. Journal of Money, Credit
and Banking, 30(3), 347-364.

26
Prean, N., & Stix, H. (2011). The effect of raising deposit insurance coverage in times of financial
crisis–Evidence from Croatian microdata. Economic Systems, 35(4), 496-511.
Safieddine, A. (2009). Islamic financial institutions and corporate governance: New insights for
agency theory. Corporate Governance: An International Review, 17(2), 142-158.
Sironi, A. (2003). Testing for market discipline in the European banking industry: Evidence from
subordinated debt issues. Journal of Money, Credit, and Banking, 35(3), 443-472.
Solé, J. (2007). Introducing Islamic banks into conventional banking systems. IMF Working Paper
No. WP/07/175. International Monetary Fund.
Starr, M. A., & Yilmaz, R. (2007). Bank runs in emerging-market economies: evidence from
Turkey's special finance houses. Southern Economic Journal, 73(4), 1112-1132.
The Economist (2001). Full of Interest. February 15.

27
Table 1: Turkish participation banks’ main figures
December 2005
(thousand US$)
Growth
compared to
2001 (%)
December 2012
(thousand US$)
Growth
compared to
2005 (%)
Compared to
banking sector
2001 (%)
Compared to
banking sector
2005 (%)
Compared to
banking sector
2012 (%)
Loans
4,824,584
551
27,002,700
460
3.14 4.50 6.70
Funds collected
6,237,260
371
26,946,445
332
1.79 3.44 6.62
Total Assets
7,412,005
353
39,547,705
434
1.49 2.60 5.62
Total Shareholders' Equity 708,817 404 4,149,868 485 1.23 2.07 4.70
Source: Banks Association of Turkey, Participation Banks Association of Turkey. In the last three columns, total
figures from participation banks are compared with those of the banking sector (state-owned, foreign-owned and
private banks
)

Table 2 Summary statistics. The table reports summary statistics of bank-specific variables with each observation
representing a measure for a single bank in a specific quarter. The DEPG is calculated as the first difference of the log
of deposits. The RDEP represents the implicit returns on deposits, calculated as the quarterly deposit expenses divided
by total bank deposits. The bank fundamentals are represented by CAP, NPL and LIQ. The CAP is the book value of
equity to total assets. The LIQ is equal to liquid assets (cash and central bank reserves) to total assets. The NPL is ratio
of loans under follow-up to total credits. The control-vector contains Bank size, Bank age, Branch size and Listed. The
Bank size variable is computed as the natural logarithm of total assets. The Bank age variable is the natural logarithm
of quarter-years the bank exists. The Branch size variable is the average number of employees per branch . The Listed
variable equals to 1 when banks are listed in the Istanbul Stock Exchange (Borsa Istanbul).
Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.
DEPG 0.026 0.385 0.085 0.075 0.052 0.392 0.064 0.050
RDEP 0.030 0.073 0.021 0.005 0.023 0.145 0.017 0.006
CAP 0.188 0.146 0.114 0.029 0.193 0.180 0.119 0.021
NPL 0.127 0.222 0.108 0.063 0.061 0.155 0.041 0.015
LIQ 0.235 0.217 0.145 0.061 0.202 0.211 0.209 0.090
Bank size 14.119 2.135 13.936 0.399 15.283 2.234 15.747 0.615
Bank age 4.662 0.786 4.007 0.366 4.890 0.645 4.354 0.301
Branch size 29.595 21.875 18.338 3.504 27.945 20.152 20.562 2.912
Listed 0.299 0.458 0.000 0.000 0.399 0.490 0.422 0.496
Conventional Banks Islamic Banks Conventional Banks Islamic Banks
2005Q4-2012Q42002Q4-2005Q3

Table 3 Tests for market discipline, sample period 2002Q4-2012Q4. This table reports estimates in the time dimension
covering the period 2002Q4-2012Q4. The dependent variable in Columns (1-4) is DEPG, which is calculated as the
first difference of the log of deposits. The dependent variable in Columns (5-8) is RDEP, which is calculated as the
quarterly deposit expenses divided by total bank deposits. The bank fundamentals are represented by CAP, NPL, and
LIQ. The CAP is the book value of equity to total assets. The LIQ is equal to liquid assets (cash and central bank
reserves) to total assets. The NPL is ratio of loans under follow-up to total credits. ISL is equal to 1 for Islamic banks,
and 0 otherwise. REF is equal to 1 for the post-reform period (2005Q4-2012Q4), and 0 otherwise. The control-vector
contains Bank size, Bank age, Branch size and Listed. The Bank size variable is computed as the natural logarithm of
total assets. The Bank age variable is the natural logarithm of quarter-years the bank exists. The Branch size variable
is the average number of employees per branch . The Listed variable equals to 1 when banks are listed in the Istanbul
Stock Exchange (Borsa Istanbul). Bank-type (Islamic or conventional) quarter-year dummy variables are included in
all specifications but their coefficient estimates are not reported. The regression method is fixed effect estimator with
heteroskedasticity and within-panel serial correlation robust standard errors. Standard errors in parentheses.
(1) (2) (3) (4) (5) (6) (7) (8)
Capital ratio
CAP
1.09620 *** 1.11065*** 1.3875 5*** 1.05162*** -0.0207 4*** -0.02013*** -0.01806* -0.00867
(0.2979) (0 .2998) (0.40 09) (0.3345 ) (0.0066 ) (0.0067) (0 .0092) (0.011 4)
CAP x ISL
-1.04493* -1.07747* -1.57317 ** -0.94200 0.03 779* 0.03857* 0.03222 0.02544
(0.6187) (0 .6306) (0.74 48) (0.7078 ) (0.0191 ) (0.0197) (0 .0224) (0.022 1)
CAP x REF
-0.48808** -0.485 16** -0.7444 0** -0.27920 0.026 85*** 0.02506 *** 0.01934 * 0.013 65
(0.1916) (0 .1922) (0.27 35) (0.3804 ) (0.0069 ) (0.0073) (0 .0101) (0.017 3)
CAP x ISL x REF
1.190 91** 1.201 85** 1.54 665** 0 .8588 6 -0.0512 0** -0 .0515 7** -0.04586* -0. 03897
(0.5135 ) (0.5192 ) (0.625 9) (0.69 35) (0.0198 ) (0.019 1) (0.02 31) (0.0 253)
Other fundamentals
NPL
-0.00031*** -0.000 32*** 0 .05776 -0.00 026*** -0.0000 1*** -0.00001*** -0.00060 -0.00001***
(0.0001) (0 .0001) (0.09 58) (0.0001 ) (0.0000 ) (0.0000) (0 .0019) (0.000 0)
NPL x ISL
-0.88466*** -0.904 06*** -1.42186*** -0.85213** 0.0254 3** 0.02 417** 0.01459 0.01777
(0.3189) (0 .2914) (0.46 68) (0.3570 ) (0.0102 ) (0.0089) (0 .0125) (0.012 1)
NPL x REF
0.06922 ** 0.0702 3** 0 .04011 0.31170 -0.00292*** -0.00281 *** -0.00195 -0.0084 8
(0.0311) (0 .0326) (0.11 68) (0.3937 ) (0.0009 ) (0.0009) (0 .0023) (0.011 5)
NPL x ISL x REF
0.58082 0 .59089 0 .54585 0.08 485 0.04860 0 .04440 0 .03537 0.05 188
(0.5578) (0 .5862) (0.80 88) (0.7671 ) (0.0311 ) (0.0297) (0 .0281) (0.032 0)
LIQ
-0.23532* -0.23019 -0.1517 7 -0.28566 ** 0.00404 0.00424 0.00500 -0.00015
(0.1399) (0 .1403) (0.20 53) (0.1133 ) (0.0061 ) (0.0061) (0 .0042) (0.009 0)
LIQ x ISL
0.45955 ** 0.4752 3** 0.44 865* 0 .45234 0.0139 3 0.00828 0.0 0075 0.011 63
(0.2096) (0 .1983) (0.26 25) (0.3325 ) (0.0099 ) (0.0110) (0 .0121) (0.014 7)
LIQ x REF
0.02480 0 .02660 -0.10679 0.09 746 0.00235 0 .00171 0 .00082 -0.013 94***
(0.0873) (0 .0882) (0.15 30) (0.0879 ) (0.0059 ) (0.0060) (0 .0034) (0.004 4)
LIQ x ISL x REF
-0.31350 -0.3 3167 -0.295 45 -0.3337 4 -0.0 0645 -0.002 14 0.003 13 0 .01277
(0.2819) (0 .2824) (0.32 50) (0.3739 ) (0.0096 ) (0.0103) (0 .0109) (0.011 6)
Bank controls
Bank size
0.19989 ** 0.2016 7** 0.237 84** 0.14645 -0.000 29 -0.00040 -0.0 0096 0.00 135
(0.0785) (0 .0800) (0.10 13) (0.0958 ) (0.0011 ) (0.0011) (0 .0007) (0.002 3)
Bank age
-0.02556 -0 .00266 0 .15511 -0.04021 0.009 43*** 0.00567 0.00624 0.00444
(0.0801) (0 .0931) (0.25 42) (0.0590 ) (0.0034 ) (0.0035) (0 .0059) (0.003 3)
Branch size
-0.00215 -0 .00223 -0.00 263* 0.0039 7 0.00001 0.00001 0 .00003 -0.0001 0
(0.0014) (0 .0014) (0.00 13) (0.0036 ) (0.0000 ) (0.0000) (0 .0000) (0.000 1)
Listed
-0.02510 -0 .02802 0 .01273 -0.01088 -0.00644*** -0.00500*** -0.00242** -0.00398 ***
(0.0378) (0 .0297) (0.03 40) (0.0269 ) (0.0015 ) (0.0017) (0 .0010) (0.001 3)
State-owned banks Yes ----- ----- ----- Yes ----- ----- -----
Private-owned banks Yes Yes ----- Yes Yes Yes ----- Yes
Foreign-owned banks Yes Yes Yes ----- Yes Yes Yes -----
Islamic banks Yes Yes Yes Yes Yes Yes Yes Yes
Bank-fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Time-fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Observations 1,366 1,246 72 5 683 1,324 1,21 2 7 07 667
R-squared 0.1053 0.1 110 0.158 3 0.3432 0.5285 0.4 723 0.335 1 0.7937
Deposit growth (Eq. 1) Deposit rate (Eq. 2)
Statistical significance is indicated by *** p<0.01, ** p<0.05, * p<0.1

2
Table 4 Crisis and the sensitivity of deposits to bank capitalization. This table reports estimates in the time dimensions
covering 1998Q1-2005Q3 (Columns 1-4) and 1998Q1-2002Q3 (Columns 5-8). The dependent variable is DEPG,
which is calculated as the first difference of the log of deposits. The bank fundamentals are represented by CAP, NPL,
and LIQ. The CAP is the book value of equity to total assets. The LIQ is equal to liquid assets (cash and central bank
reserves) to total assets. The NPL is ratio of loans under follow-up to total credits. Although we estimate the most
flexible specification by integrating each of these fundamentals directly (coefficient 

in Eq. 3) as well as with three
separate interaction terms (coefficients 

, 

and 

in Eq. 3), we only report CAP and its interactions to facilitate the
interpretation of results. ISL is equal to 1 for Islamic banks, and 0 otherwise. CRIS is equal to 1 for the post-crisis
period (2001Q1-2005Q3 or 2001Q1-2002Q3), and 0 otherwise. The control-vector contains Bank size, Bank age,
Branch size and Listed. The Bank size variable is computed as the natural logarithm of total assets. The Bank age
variable is the natural logarithm of quarter-years the bank exists. The Branch size variable is the average number of
employees per branch . The Listed variable equals to 1 when banks are listed in the Istanbul Stock Exchange (Borsa
Istanbul). Bank-type (Islamic or conventional) quarter-year dummy variables are included in all specifications but their
coefficient estimates are not reported. The regression method is fixed effect estimator with heteroskedasticity and
within-panel serial correlation robust standard errors. Standard errors in parentheses.
VARIABLES (1) (2) (3) (4) (5) (6) (7) (8)
CAP
1.46496 *** 1.49293*** 2.2540 5*** 1.31093*** 1.91663*** 1.9878 0*** 2.88335*** 1.48 295***
(0.2334) (0 .2349) (0.39 05) (0.2019 ) (0.326 0) (0.3380 ) (0.5975) (0.2 351)
CAP x ISL
0.05780 -0 .00628 0.4956 2 -0.12617 -0.93121* -1 .02474** -1 .09741 -0.69215
(0.7139) (0 .7209) (1.31 63) (0.4235 ) (0.486 5) (0.5101 ) (1.0103) (0.5 598)
CAP x CRIS
-0.27735 -0.3 2162 -0.92004* -0.24174 -0.05801 -0.1 4441 -0.535 65 0.01 051
(0.2667) (0 .2852) (0.47 43) (0.2515 ) (0.457 2) (0.4933 ) (1.1628) (0.2 560)
CAP x ISL x CRIS
-1.43450** -1.35 925* -1.88661 -1.0949 4** -2.31 839*** -2.23255** -2.55551 -1.92 963**
(0.7103) (0 .7253) (1.36 50) (0.4552 ) (0.823 3) (0.8563 ) (1.6262) (0.8 244)
Bank controls
Bank size
0.28431 *** 0.28727*** 0.3559 3*** 0.19694*** 0.35084*** 0.3622 6*** 0.47058*** 0.20 090***
(0.0488) (0 .0494) (0.07 65) (0.0563 ) (0.080 2) (0.0824 ) (0.1377) (0.0 639)
Bank age
-0.24439*** -0.260 32*** 0 .02957 -0 .24501*** -0.1414 1 -0.15992 0.2434 2 -0.2252 0***
(0.0921) (0 .0949) (0.13 43) (0.0615 ) (0.142 0) (0.1514 ) (0.1614) (0.0 809)
Branch size
-0.00130 -0.0 0119 -0.004 09 0.00 438 0.00033 0.0004 1 -0.0 0162 0.00 772**
(0.0025) (0 .0025) (0.00 31) (0.0032 ) (0.002 9) (0.0029 ) (0.0030) (0.0 036)
Listed
-0.02471 -0.0 2499 0.08604 0.05 908 0.06294 0.10548***
(0.0458) (0 .0424) (0 .0617) (0.0381) (0.0403 ) (0.0372 )
State-owned banks Yes ----- ----- ----- Yes ----- ----- -----
Private-owned banks Yes Yes ----- Yes Yes Yes ----- Yes
Foreign-owned banks Yes Yes Yes ----- Yes Yes Yes -----
Islamic banks Yes Yes Yes Yes Yes Yes Yes Yes
Bank-fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Time-fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Observations 1,404 1,3 00 578 860 956 888 38 3 585
R-squared 0.131 5 0.1360 0 .1871 0.31 68 0.1275 0.134 1 0.1915 0 .2687
Deposit growth (sample period 199 8Q1-2005Q3) Deposit growth (sample period 199 8Q1-2002Q3)
Statistical significance is indicated by *** p<0.01, ** p<0.05, * p<0.1

3
Table 5 Crisis, deposit insurance reform, and the sensitivity of deposits to bank capitalization. This table reports
estimates in the time dimension covering 1998Q1-2012Q4. The dependent variable is DEPG, which is calculated as
the first difference of the log of deposits. The bank fundamentals are represented by CAP, NPL, and LIQ. The CAP is
the book value of equity to total assets. The LIQ is equal to liquid assets (cash and central bank reserves) to total assets.
The NPL is ratio of loans under follow-up to total credits. Although we estimate the most flexible specification by
integrating each of these fundamentals directly (coefficient 

in Eq. 3) as well as with three separate interaction terms
(coefficients 

, 

and 

in Eq. 3), we only report CAP and its interactions to facilitate the interpretation of results.
ISL is equal to 1 for Islamic banks, and 0 otherwise. While the period before the 2001Q1 crisis serve as a reference
period, we define PERIOD2 as equal to 1 for the period 2001Q1-2005Q3 and 0 otherwise, and PERIOD3 as equal to
1 for the period 2005Q4 to 2012Q4 and 0 otherwise. The control-vector contains Bank size, Bank age, Branch size and
Listed. The Bank size variable is computed as the natural logarithm of total assets. The Bank age variable is the natural
logarithm of quarter-years the bank exists. The Branch size variable is the average number of employees per branch .
The Listed variable equals to 1 when banks are listed in the Istanbul Stock Exchange (Borsa Istanbul). Bank-type
(Islamic or conventional) quarter-year dummy variables are included in all specifications but their coefficient estimates
are not reported. The regression method is fixed effect estimator with heteroskedasticity and within-panel serial
correlation robust standard errors. Standard errors in parentheses.
VARIABLES (1) (2) (3) (4)
CAP 1.27116*** 1.33225*** 2.09319*** 1.14524***
(0.2197) (0.2275) (0.4206) (0.1800)
CAP x PERIOD2 -0.30974 -0.36737 -0.83291* -0.48461**
(0.2345) (0.2517) (0.4714) (0.2315)
CAP x PERIOD3 -0.61155*** -0.64632*** -1.25641*** -0.45183
(0.1845) (0.1935) (0.2983) (0.4619)
CAP x ISL 0.81389 0.75626 0.36319 0.19130
(0.6856) (0.6860) (0.9266) (0.4674)
CAP x ISL x PERIOD2 -2.07830*** -2.02474*** -2.06266* -1.14886**
(0.6804) (0.7011) (1.0793) (0.4330)
CAP x ISL x PERIOD3 -1.14456 -1.11038 -1.46762 -0.55153
(0.9447) (0.9718) (1.3620) (0.6623)
Bank controls
Bank size 0.20401*** 0.20952*** 0.29293*** 0.10146***
(0.0363) (0.0383) (0.0693) (0.0274)
Bank age -0.17262*** -0.18201*** -0.19773** -0.15202***
(0.0491) (0.0512) (0.0948) (0.0361)
Branch size -0.00225* -0.00225* -0.00302** 0.00305
(0.0013) (0.0013) (0.0014) (0.0020)
Listed -0.05325 -0.05393* 0.01587 0.01376
(0.0373) (0.0319) (0.0322) (0.0256)
State-owned banks Yes ----- ----- -----
Private-owned banks Yes Yes ----- Yes
Foreign-owned banks Yes Yes Yes -----
Islamic banks Yes Yes Yes Yes
Bank-fixed effects Yes Yes Yes Yes
Time-fixed effects Yes Yes Yes Yes
Observations 2,361 2,170 1,125 1,292
R-squared 0.1157 0.1188 0.1599 0.4270
Deposit growth (sample period 1998Q1-2012Q4)
Statistical significance is indicated by *** p<0.01, ** p<0.05, * p<0.1