MSc Business Economics, Finance Track
Master Thesis
Shifting Preferences among Bulgarian
Depositors
A Study on the Bulgarian Banking System with a Focus on
Depositor Behavior after the Failure of a Major Financial
Institution
Yordan S. Bogev
-July 2016-
Supervisor
2
Abstract
This paper examines the impact that the failure of Corporate Commercial Bank in 2014 has had on the behavior of Bulgarian depositors. The study incorporates quarterly supervisory data on institutions within the Bulgarian banking system for the period 2013-2015 and aims to find empirical relationships between changes in deposit levels and the origin of ownership of the banks. Deposit flows towards either domestic or foreign-owned banks should indicate depositors’ assessment about the riskiness of the two groups of institutions. The undertaken research reveals evidence in support of large foreign-owned institutions having increased their wholesale deposit bases by approximately 15% relative to domestic-owned competitors since the banking crisis of 2014. In regard to the retail funding of large-scale banks, a similar pattern is observed, although it is only supported by weak empirical evidence.
This study complements the existing literature on depositor behavior and can be related to depositors in most transition economies in Eastern Europe. The main contribution is based on the theory, that the ownership of a bank itself could be considered by depositors as a proxy for the quality of corporate governance and risk management. The outcome of this research can bring essential implications for bank managers to take into account when managing a financial institution within a transition country.
Acknowledgements
I am grateful to Dr. Tanju Yorulmazer, Associate Professor of Finance in the Faculty of Economics and Business at the University of Amsterdam, for his overall support during the research process. As my Thesis Supervisor, he encouraged me to proceed with my thesis proposal and demonstrated a genuine interest in the topic, which gave me the motivation to complete my research.
3
Contents
Abstract ... 2
1. Introduction & Background ... 4
A. The Bulgarian Banking System: Overview ... 7
B. The Bank Runs of June 2014: Corporate Commercial Bank and First
Investment Bank ... 10
2. Literature Review ... 13
3. Hypotheses & Methodology ... 19
A. Hypotheses ... 19
B. Model Testing ... 21
4. Data & Descriptive Statistics ... 23
5. Results ... 30
6. Robustness Checks ... 37
7. Concluding Remarks & Limitations ... 40
References ... 44
Appendices ... 46
Statement of Originality
This document is written by Student Yordan Bogev who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
4
1. Introduction & Background
The financial crisis of 2008 and the several bank runs that occurred as a result from it provoked researchers’ interest in depositor behavior during episodes of widespread panic. Since depositors are the major debt holders of banks, every time they become worried about a given bank’s ability to pay them back, this institution is facing the risk of experiencing a wave of withdrawals which may eventually result in a liquidity shock and even bankruptcy (Diamond and Dybvig, 1983).
Because of the specific economic regime, under which most Central and Eastern European citizens have been living for almost fifty years during the twentieth century, it is presumed that depositors in these countries are particularly sensitive to banking crises and economic shocks as a whole (Hasan et al., 2013; Ungan et al., 2007). In June 2014, Bulgaria witnessed an unprecedented banking crisis when the country’s fourth-biggest bank, Corporate Commercial Bank (CCB), experienced a bank run and, eventually, got its license revoked by Bulgaria’s National Bank in November 2014. The bank run spread to the country’s third-biggest bank, First Investment Bank (FIB), which received liquidity support from the government and was able to continue normal operations. Given the fact that these two banks represented the two biggest Bulgarian-owned financial institutions at the time, the following question arises: Was it a coincidence that the only two problematic banks happened to be Bulgarian-owned while foreign-owned banks managed to remain stable during the episode?
This paper examines whether Bulgarian depositors have developed a stronger preference for foreign-owned banks since the CCB episode. The underlying assumption is that the shock experienced by the two biggest domestic-owned institutions was a result of depositors’ beliefs that these banks had adopted worse corporate governance practices and weaker risk management relative to their foreign-owned counterparts. In order to investigate this, I aim to
5
provide empirical evidence that foreign-owned banks have attracted significantly more deposit funding relative to Bulgarian-owned competitors since the events of June 2014. The research output could suggest important implications about depositors’ assessment of the business and corporate governance practices of domestic-owned institutions. The study itself is based on the first-of-its-scale banking crisis in a European Union Member State in Eastern Europe. Unlike most of the recent bank runs across the continent, the events in Bulgaria were not a direct consequence of the financial crisis of 2008, but rather a local and isolated episode of panic. Given the specific psychological characteristics of depositors in former Socialist States, this study can yield important implications related to the prevention of similar future shocks not only in Bulgaria, but in other European transition economies as well.
For the purpose of the difference-in-difference analysis performed in this study, two time windows are constructed, each incorporating the six quarters prior and the six quarters after the run on CCB, namely the period starting in the first quarter of 2013 up to the last quarter of 2015. I also make a distinction between small banks (with less than BGN1 1 billion worth of
assets, as of the end of the last quarter of 2015) and large banks (assets worth more than BGN 1 billion) with the assumption that the economic impact of the 2014 crisis should be different for banks of different sizes and depositor bases. The study also differentiates between retail and wholesale funding, since the Bulgarian Deposit Insurance Fund only guarantees the full amount of deposits up to EUR 100,000. Therefore, I expect the impact on wholesale deposit levels to be more pronounced.
My results bring some important economic implications. Firstly, no significant shift of preferences among small banks’ depositors can be confirmed by the empirical findings.
6
This signifies that the narrow groups of depositors in these banks might have close connections to the bank management, which would cause an absence of adverse information. Moreover, depositors in these banks usually hold limited deposit amounts, which are likely to be under the protection of the deposit insurance fund.
For the banks of larger scale, I only find weak evidence for big foreign-owned banks having attracted significantly more retail funding relative to their Bulgarian-owned competitors. A plausible explanation could be the presence of the Bulgarian Deposit Insurance Fund, which was successfully put into practice for the first time in December 2014, when depositors of the bankrupted CCB were paid out the amount of their guaranteed deposits. It could be assumed that prior to this episode, most retail depositors in the country were not aware of the existence of the Fund and, since then, this awareness has been preventing new runs on domestic-owned institutions. In regard to wholesale funding, I detect statistically significant flows, indicating that large-scale foreign-owned banks have managed to attract approximately 15% more wholesale deposits relative to big domestic-owned counterparts. Since most of these deposits are not subject to deposit insurance by the state, wholesale depositors have more reason to be concerned about the quality of their bank and a “flight to quality” (Saunders and Wilson, 1996) towards foreign-owned banks is the most likely scenario, which was observed after the banking crisis of 2014.
The rest of the paper is structured as follows: Part A of Section 1 provides an overview of the Bulgarian Banking System, while Part B is focused on the bank runs of June 2014. Section 2 provides a review of the related literature on bank runs and depositor behavior. Part A of
Section 3 outlines the hypotheses to be tested, while Part B is focused on the empirical
approach for the research with a detailed description of the different regression variables.
Section 4 introduces the collected data and the descriptive statistics of Bulgaria’s banking
7
related robustness checks. Section 7 concludes by discussing the policy implications as well as the limitations faced during the research process.
A. The Bulgarian Banking System: Overview
The banking system in Bulgaria follows a traditional business model. Resulting from the domestic economic crisis of 1996-1997, and the implementation of the currency board, pegging the national currency against the Euro, the banking sector has been dominated by foreign-owned institutions. Bulgarian banks are rather conservative, with limited reliance on wholesale funding and an asset side that consists primarily of loans. The 2008 global financial crisis and the subsequent need for parent companies to strengthen their balance sheets, including through deleveraging abroad, decreased the appetite for expansion on the Bulgarian market. Bulgarian-owned banks saw this as a chance to gain market share through quick credit expansion, primarily addressed to domestic companies.
Table 1. Bulgarian Banking System: Breakdown by Institution (as of 31.12.2015)
Source: Bulgarian National Bank (in millions of BGN).
Banking Institution Amount (in millions of BGN)
Bank ID Name Total Assets % Retail Deposits % Wholesale Deposits %
660 UniCredit Bulbank 17368 20% 6720 15% 7855 26% 300 DSK Bank 11100 13% 7224 16% 2191 7% 150 First Investment Bank 8680 10% 5954 13% 1510 5% 200 United Bulgarian Bank 6559 7% 3625 8% 1669 6% 155 Raiffeisenbank Bulgaria 6450 7% 2653 6% 2821 9% 920 Postbank 5760 7% 3626 8% 1159 4% 400 SG Expressbank 5274 6% 2913 7% 1704 6% 790 Central Coop. Bank 4642 5% 3193 7% 992 3% 170 Piraeus Bank Bulgaria 2861 3% 1493 3% 770 3% 888 Cibank 2687 3% 868 2% 1500 5% Other Other 16143 18% 6138 14% 7767 26%
8
While overall credit growth after the 2008 economic crisis was very limited, the domestic-owned segment expanded very rapidly. The average annual growth rate of loans extended by the Bulgarian banking system between the first quarter of 2010 and the first quarter of 2014 stood at around 3.5%, according to the European Commission’s Country Report on Bulgaria (2015). The majority of the Greek-owned subsidiaries decreased their loan stock by 0.9% during this period, while the subsidiaries owned by other European Union banks increased it by 3.7%. In comparison, the average annual expansion of the most Bulgarian-owned banks reached 17.6%. Overall, Corporate Commercial Bank (CCB) achieved the highest growth rate, at 32.3% per year. Four domestic-owned banks grew by an annual rate higher than 10%, while no foreign-owned institution reached this growth rate.
Figure 1. Nominal GDP & Bulgarian Banking System Statistics
Source: Bulgarian National Bank (in millions of BGN).
81544 81971 83612 86373 75131 77557 85135 87524 59261 61839 66567 74457 56947 59202 66869 74346 2012 2013 2014 2015
9
It is important to note that this credit expansion took place in the context of low economic growth. Since 2009, nominal GDP has been growing by 3.3% per year on average and by the end of 2014, the banking sector’s total assets level has surpassed the country’s nominal GDP.
Figure 1 illustrates that at the end of December 2015, the banking sector accounted for 101%
of Bulgaria’s nominal GDP. The contrast between credit and economic growth rates posed questions about the allocative efficiency of financial intermediation institutions and the quality of the investments they made over this period.
The pricing of loans suggests higher risk levels for Bulgarian-owned banks relative to their foreign competitors. Domestic banks have been charging, on average, higher interest on the corporate loans they were granting. This profile raises concerns about the average credit quality of borrowers, the credit underwriting practices, and the underlying risk of the portfolios. During the rapid expansion before the 2009 recession, this behavior could have been justified by the very strong competition for lower-risk clients from the foreign-owned banks, which also accounted for the majority of lending during that period. However, this trend does not seem rational in the context of post-crisis years when competition for clients from the other segments of the banking system was much weaker.
Another clear pattern among Bulgarian-owned banks is the relatively low profitability compared to foreign-owned counterparts, which is empirically illustrated in Section 4. Despite the fact that the institutions within the system have managed to remain profitable throughout the crisis, their profitability was significantly reduced. This decline was higher for Bulgarian-owned banks, mainly due to a narrowing interest margin. Unlike foreign-owned banks, the profitability of their domestic competitors declined despite the relatively low impairment costs. Consequently, these banks may not be able to generate sufficient profits to absorb losses if asset quality deteriorates. As a result, even bringing provisioning costs up to the sector average could bring losses and erosion in capital buffers. Impairments for
loan-10
losses in Bulgarian-owned banks accounted for 2.2% of total assets at the end of the last quarter of 2013, while the system average stood at 5.9%.
Deposit insurance in Bulgaria is executed by the Bulgarian Deposit Insurance Fund, which was established under the Law on Bank Deposit Guarantee and began operations in April 1998. For the period between its enactment and December 2010, all deposits up to BGN 100,000 (EUR 51,000) were fully guaranteed regardless of their number. On December 31, 2010, the insurable limit was increased to BGN 196,000 (EUR 100,000) in accordance with European Union directives on deposit insurance. The Fund offers full coverage to all depositors, holding amounts up to EUR 100,000 in both domestic and foreign-owned banks operating within the country’s market. So far, the Bulgarian Deposit Insurance Fund has only been put into practice once. This episode took place in December 2014, when bankrupt CCB’s depositors received approximately EUR 1.89 billion worth of guaranteed deposits.
B. The Bank Runs of June 2014: Corporate Commercial Bank and First Investment Bank
Bulgaria’s banking system took a big hit in the summer of 2014. The banking sector and Bulgarian-owned banks, in particular, experienced a liquidity shock at the end of June 2014. Corporate Commercial Bank (CCB), the fourth-largest bank in the system and the second-largest domestic-owned bank with EUR 3.74 billion worth of assets, became a victim of a bank run and was placed under special surveillance by the Bulgarian National Bank. Legal and institutional hurdles did not allow the triggering of the deposit insurance scheme within the deadlines required by European Union law and depositors were able to access their funds no sooner than December 2014. The Central Bank revoked CCB’s license in early November after the bank was found to be in a position of deep negative equity, with EUR 1.79 billion worth of non-performing loans out of its EUR 2.76 billion loan portfolio.
11
Preliminary results from the administrators’ investigation on the available collateral presented evidence of malpractice in CCB’s operations. Based on the analysis of sixty-nine borrowers, or nearly half of those reviewed by the auditors, the following problems were identified: unpledged collateral, involving assets that are included in the lending contract but have not been legally pledged; special claims, which are not legally binding; and second-lien claims, including seventy-four claims by CCB and fourteen second-lien mortgages.
CCB’s losses accounted for nearly two-thirds of the bank’s total assets. According to the final version of the officials’ report, the need for further impairments amounted to EUR 2.1 billion, most of which arise from loans, granted to newly-created holding companies. Exposures to these companies were not properly reported by the bank in its annual reports, so it is not clear why those statements were certified by the auditors as credible. Some of the loans in the bank’s portfolio have been renegotiated numerous times and the initially-reported collateral was largely unrealisable.
Figure 2. Timeline Summary: The Bank Runs on CCB and FIB (2014)
June 2014 Media spreading negative rumours about CCB; June 20, 2014 CCB depositors begin withdrawals, start of a bank run;
June 20, 2014 CCB closes down and is put under special surveillance by the Bulgarian National Bank; June 27, 2014 FIB reports BGN 800 million worth of daily withdrawals, a second bank run begins; June 29, 2014 European Commission approves EUR 1.65 billion support for Bulgaria's banking system; June 29, 2014 FIB receives EUR 0.61 billion as liquidity aid from the Government;
Nov 6, 2014 Bulgarian National Bank officially revokes the license of CCB;
Nov 25, 2014 EC approves the restructuring plan for FIB and extends liquidity support for another 18 months; Dec 4, 2014 Bulgarian Deposit Insurance Fund starts paying the guaranteed amounts to CCB's depositors.
12
The week after the CCB episode, the largest Bulgarian-owned bank and the third-largest bank in the system, First Investment Bank (FIB), also experienced a run and officially requested liquidity assistance from the authorities. A credit line of EUR 1.65 billion, approved under State aid rules by the European Commission, was made available to all Bulgarian banks. The authorities decided to put a EUR 0.61 billion state deposit into FIB for an initial duration of five months. On 25 November, the European Commission approved the State aid and the restructuring plan for the bank, extending the term of the originally-provided liquidity support by further eighteen months.
The bottom line from all facts and speculations surrounding the bankruptcy of CCB is that the true source of its difficulties was the fraudulent schemes practiced by the bank’s officials. Even though media attacks were the spontaneous trigger of the bank run in June 2014, the final report from the investigation by AlixPartners (2015) suggested that the huge losses and deep negative equity arise from granting uncollectible loans with no recorded collateral, mainly to companies with unknown ownership, but considered to be related to the last Chairman of CCB. By contrast, the bank run experienced by FIB could be classified as an “effect of contagion”. At the present moment, this bank is properly functioning and, at the end of May 2016, it managed to make the last BGN 350 million payment on the liquidity injection it had received from the government in late 2014.
13
2. Literature Review
When discussing banking institutions in general, it is important to take into consideration the literature focusing on the maturity transformation performed by banks, and also on the fragility of financial intermediaries expressed as the risk of a run and loss of funding sources. Bryant (1980) and Diamond and Dybvig (1983) are some of the first to come out with a framework that illustrates the role of financial intermediaries in performing maturity transformation by issuing short-term liabilities in the form of deposits and holding long-term assets in the form of loans. As a result, banks become exposed to fragility when depositors become insecure about a certain institution’s ability to pay them back at a later period and thus decide to withdraw their funds in an early period. This phenomenon resembles what is now seen as a “bank run” and sometimes leads to the failure of fundamentally solvent banks. The authors do not directly address the question of what events might cause depositors’ beliefs to shift and, hence, trigger a run. Although there is no doubt that the CCB case was indeed a bank run, the question of what has prompted the panic among its depositors is a fundamental one to answer.
According to Kindleberger (2000), bank runs can be set off by anything that provokes pessimism among depositors, including what might be called “mass hysteria”. Often, the trigger is an exogenous event, which is not necessarily related to the bank’s fundamental solvency. However, there is also evidence of a correlation between bank runs and the current condition of a given sector, or an economy as a whole. Gorton (1988) performs empirical analysis using U.S. data from the late nineteenth and early twentieth century in order to explore banking crises. He finds a close relationship between the occurrence of banking panics and the overall state of the economy. However, the run on CCB does not seem to be a natural consequence of the business cycle. The unique feature of this failure is that it occurred after the economy had already recovered from the post-2008-Crisis recession, and during a
14
time when the country’s banking system was deemed very stable. An addition, which seems to be applicable to the CCB case, is illustrated by Allen and Gale (1998). According to their view, in the presence of adverse information about the banks’ prospects, depositors anticipate the difficulties, which banks may face in making promised payments. The consequences from this anticipation may lead to massive withdrawals and even bank failure.
Occasionally, the news does not reflect the actual state of a banking institution. Depositors are sometimes provided with inaccurate information about a certain bank and this imperfect information can lead to a run on a solvent institution. According to Ennis (2003), the observed correlation between runs and economic fundamentals does not imply that healthy banks are immune to runs. Similar research is performed by He and Manela (2014), who focus on information acquisition and spontaneous withdrawal decisions when a spreading rumour leads to a run on a solvent bank. In their setting, uncertainty about the bank’s solvency and potential failure drives depositors to acquire additional noisy signals. This is exactly the case with First Investment Bank (FIB) in Bulgaria, which became victim of media speculations after the run on CCB. The result was a bank run on an otherwise sound institution.
When discussing the motivation behind a bank run, it is important to distinguish between two scenarios – fundamental-based and panic-based runs. Fundamental-based runs reflect depositors’ risk assessment of a given bank, which may lead to withdrawals of funds. From an institution perspective, Morris and Shin (2009) distinguish between and measure three separate types of risk: insolvency risk, the conditional probability of default resulting from deterioration in asset quality if there is no run by short-term creditors; total credit risk, which is the unconditional probability of default due to either a creditor run or asset insolvency; and illiquidity risk, which is the difference between the first two, specifically, the probability of failure due to a run when the banking institution would otherwise have been solvent.
15
The three aforementioned types of risk vary with different features of a bank’s balance sheet. In particular, they illustrate that illiquidity risk is decreasing in the “illiquidity ratio”, the ratio of realizable cash on the balance sheet to short-term liabilities; increasing in the “outside option ratio”, a measure of the opportunity cost of the funds used to roll over short-term liabilities; and, last but not least, increasing in the “fundamental risk ratio”, a measure of ex-post volatility of the portfolio of assets.
When one financial institution is experiencing liquidity distress, it is important to pay attention to the risk of contagion spreading among other banks in the system. Nier et al. (2007) explore this topic in detail and focus on the structure of a banking system as a determinant of possible “effects of contagion”. The authors show evidence that a system of better-capitalized banks is more resilient against contagious defaults. Besides capitalization, their research reveals that concentration in the banking system is another important factor, with less concentrated systems allowing for a contagion to stop before spreading to a large fraction of banks. These implications can provide insight about the events in Bulgaria in 2014, when the fundamentally solvent First Investment Bank (FIB) appears to have become a victim of contagion effects arising from the bank run on CCB. Although the Bulgarian banking sector is considered to be highly segmented, the fact that only one other bank has suffered “contagious” effects from the run on CCB could suggest that the other institutions within the system were well-capitalized.
Upper and Worms (2002) adopt an additional view in their study on exposures and contagion. The authors make use of balance sheet data from the German banking sector and show that the failure of a single bank may lead to approximately 15% decrease in the total assets within the banking system. Moreover, the authors state that while contagion is usually narrowed to a limited number of relatively small banks, bank failures that affect a sizeable part of the banking system may still occur even in the presence of safety mechanisms like deposit
16
insurance. In the case of Bulgaria, the effects of contagion have spread to FIB despite the deposit insurance scheme, which appears to be in accordance with the authors’ aforementioned hypothesis.
Another perspective for discussing banking crises is to look at bank withdrawals as the mechanism employed by depositors to discipline financial institutions. This type of discussion has become even more extensive after the 2008 Crisis, since when depositors have become significantly more sensitive to the risks that banks take with their money. Goldsmith-Pinkham and Yorulmazer (2009) investigate the run experienced by Northern Rock in 2007 and examine in detail whether panic among investors was the only reason for the massive withdrawals, or their behavior was actually backed by a rational analysis of the balance sheet and funding characteristics of the bank. The authors provide empirical support on the latter: the bank run was indeed a rational response by depositors to the available information. This also turned out to be relevant for the CCB episode in Bulgaria. The investigations following the bank run and focusing on the bank’s activities revealed some unthought-of details.
Berger and Turk-Ariss (2010) explore the actual mechanisms which depositors exploit for disciplining risk-taking banks and also analyze how the impact of this discipline varies across small and large banking organizations. The authors’ findings indicate stronger effects for institutions below $50 billion in assets, which is consistent with the “too-big-to-fail” theory valid for larger organizations. It is also suggested that significant depository discipline did exist prior to the 2008 Financial Crisis, but the authors do not make any implications about post-Crisis discipline, which is observed within the time frame of the present research.
In his research of the U.S. banking system, covering the period from 1985 until 1989, Park (1995) provides empirical support on the presence of market discipline by large time depositors. The author reveals that riskier banks tend to offer higher interest rates on large time deposits, but at the same time attracted fewer deposits during the period of his study.
17
This implies that higher interest rates may serve as a proxy for depositors when assessing the riskiness of a certain bank. However, Park finds no empirical evidence that bank size actually affects depositors’ selection.
Martinez Peria and Schmukler (2001) evaluate the impact of banking crises on market discipline in their study, focused on experiences from Argentina, Chile and Mexico. They confirm the theory that depositors in these countries tend to punish banks for risky behavior, both by withdrawing their deposits and by requiring higher interest rates. In addition to the theory that bank size does not affect depositors’ preferences (Park, 1995), the authors present evidence that there is no significant difference across depositors; both large and small depositors discipline banks. They also suggest that, as a result of bank failures, depositors have become more aware of the risk of losing deposits. Thus, they start exercising a stricter market discipline.
Understanding the behavior of depositors within a specific type of economy and market is essential for the purpose of each related research. In order to get a more accurate understanding of depositor characteristics in Bulgaria, it is reasonable to examine existing literature on depositors with similar traits. Hasan et al. (2013) discuss examples of depositor behavior in Central European countries with transition economies. The authors conclude that non-financial depositors in those countries strongly react to sources of information other than financial statements. More specifically, they suggest that bank subsidiaries controlled by foreign parent companies rumoured to be in financial trouble, reported significantly lower deposit growth rates. An important similarity with the current study is the focus on depositors' behavior and their interpretation of public non-financial information. However, this thesis should contribute further by examining whether the bank institution’s origin of ownership itself (either domestic or foreign-owned) can influence the preferences of depositors.
18
Ungan et al. (2008) perform analysis of depositors’ assessment of bank riskiness in the Russian Federation – another Eastern European emerging market with similar features to Bulgaria. The authors measure the extent of market discipline in the Russian banking sector by exploring reactions of Russian depositors to excessive risk taking by large banks between January 2000 and January 2005. Their findings reveal that better-capitalized and more liquid banks significantly increase their deposits. In my setting, I want to test whether foreign-owned banks significantly increase deposit levels relative to domestic-foreign-owned banks after an episode of distress. Provided that such evidence is detected, it could be concluded that Bulgarian depositors classify foreign-owned banking institutions as generally “better” than domestic competitors.
Nowadays, deposit insurance is deemed as a useful mechanism for preventing runs and banking crises, but they still occur despite the presence of such safety policies. Iyer and Puri (2008) question this theory and investigate the influence of deposit insurance on depositor behavior during episodes of panic. In contradiction to conventional wisdom, the authors provide support for the theory that these policies are only partially effective, as it was the case with the run on CCB, where deposits up to EUR 100,000 were fully guaranteed by the state. Explicit deposit insurance is observed to increase the likelihood of banking crises, as reported by Demirgüç-Kunt and Detragiache (2000) in their study, which incorporates sixty-one countries during the period 1980-1997. The adverse impact of deposit insurance on bank stability becomes stronger with higher coverage offered to depositors where the scheme is run and funded by the government.
With the assumption that depositors do, ultimately, impose discipline on banking institutions, a consistent question would be: “What do investors do with their money after withdrawing them from a given bank?” Saunders and Wilson (1996) examine deposit flows in one hundred sixty-three failed and two hundred twenty-nine surviving banks over the Depression era of
19
1929-1933 in the United States. For the years 1929 and 1933, they found evidence of “flight to quality”, expressed by withdrawals from bankrupt banks leading to deposit increases in surviving banks. Their finding is, to some extent, linked to my research, which investigates whether such “flight to quality” occurred in the Bulgarian banking system, with the assumption that Bulgarian depositors consider foreign-owned banks to be institutions of “higher quality” and better corporate governance.
3. Hypotheses & Methodology
A. Hypotheses
The main objective of this research is to detect an empirical relationship between changes in deposit levels and the banks’ origin of ownership, while taking into account the extent to which this relationship was affected by the failure of CCB in June 2014. It is essential to investigate whether the shock from this episode was big enough to further shift depositors’ preferences towards foreign-owned banks, which, as a group, had been attracting more deposit funding prior to 2014 as well. I am going to differentiate between retail and wholesale funding, since the latter usually incorporates large corporate deposits, which exceed the amount of EUR 100,000.2 And as specified earlier, all deposits below this threshold are
fully-guaranteed by the state’s Deposit Insurance Fund, which was put into practice in December 2014 by paying out approximately EUR 1.9 billion to CCB’s retail depositors.
2
As of December 2015, 77% of Retail funding constitutes of deposits below the insurable EUR 100,000 threshold, while 85% of total Wholesale deposits are above the EUR 100,000 limit. Source: Statistics on Deposits and Loans by Amount Category and Economic Activity, 2015 (http://bnb.bg/statistics).
20
For the purpose of separating the two types of funding, I am going to create two dependent variables: Retail Deposit Growth and Wholesale Deposit Growth, each measuring the percentage change in deposit levels since the previously reported quarter.
Another distinction in this study is between small and large-scale banks. Since the economic significance of banks with less than BGN 1 billion worth of assets is considered to be relatively smaller, I assume that their low liquidity levels and small depositor bases will yield results of lower significance when examining the impact of the CCB episode on depositors’ preferences. On the other hand, all banks with more than BGN 1 billion worth of assets will be classified as large-scale institutions. These banks attract wider bases of depositors, who are supposed to be more concerned with the institution’s overall health and solvency since the amounts they hold as deposits are expected to be higher. Having made the two major distinctions, it is now possible to develop the four hypotheses to be empirically tested:
Hypothesis 1: Large foreign-owned banks have attracted significantly more wholesale funding relative to large Bulgarian-owned banks since the run on CCB in June 2014.
Hypothesis 2: Large foreign-owned banks have attracted significantly more retail funding relative to large Bulgarian-owned banks since the run on CCB in June 2014.
Hypothesis 3: The joint impact of the run on CCB and ownership has been more pronounced for wholesale than for retail funding.
Hypothesis 4: The joint impact of the run on CCB and ownership has been more pronounced for large than for small banks (for either Retail or Wholesale set of regressions).
21
B. Model Testing
In order to test the aforementioned hypotheses and examine whether the ownership origin of a given bank has affected Bulgarian depositors’ preferences after the CCB episode, two separate difference-in-difference regression analyses will be run: one employing Retail
Deposit Growth as dependent variable, and one using Wholesale Deposit Growth. The
regressions make use of panel data for the period starting the first quarter of 2013 up to the fourth quarter of 2015. Firm and year-fixed effects will be incorporated in the empirical model for the purpose of controlling for any unobserved differences among banks and which can potentially affect the growth in deposit levels. Clustered standard errors are being used in order to allow for autocorrelation in the error term within entities.
Retail Deposit Growthit = α + β1foreignit + β2CCBit + β3foreignit*CCBit + controls + εit
Wholesale Deposit Growthit = α + β1foreignit + β2CCBit + β3foreignit*CCBit + controls + εit
In this setting, both Retail Deposit Growth and Wholesale Deposit Growth measure the percentage change in deposit levels for each bank between the observed and the previous quarter. Foreignit is a dummy, indicating whether the bank is domestic or foreign-owned (0
if domestic; 1 if foreign-owned), CCBit is a dummy revealing whether the observation is
recorded before or after the bank run on CCB in June 2014 (0 if prior to June 30, 2014; 1 if after), and foreignit*CCBit is the interaction term between the two dummies. Controls
represent the set of control variables, included to help control for any factors, unrelated to ownership, which could be inflicting changes in retail and wholesale deposit flows. The set of control variables includes log (Total Assets) and Market Share, both controlling for size and
22
market power of a given bank; Interest Costs/Total Liabilities to control for the interest payments to depositors; Total Equity/Total Assets to control for the bank’s leverage level;
Loans/Total Assets for measuring the fraction of risky assets each bank holds on its Balance
Sheet; and Operating Return on Assets to control for a bank’s profitability.
In order to be able to test for all four hypotheses, it is crucial to examine separately small and large banks within the dataset. For this purpose, the data on the two types of banks will be split, with all banks with assets worth BGN 1 billion or more (as of the fourth quarter of 2015) being classified as large. Banks with assets worth less than BGN 1 billion are placed into the small category. For each of the two groups, a total of four difference-in-difference panel data regression are run, using either Retail Deposit Growth or Wholesale Deposit
Growth as a dependent variable. First, the dependent variable will be regressed on the two
dummy variables foreignit and CCBit, and their interaction foreignit*CCBit only. Afterwards,
the aforementioned control variables will be incorporated as well. For both sets of regressions – with or without the control variables employed, firm and time-fixed effects, as well as clustered standard errors, will be utilized.
The coefficient of interest throughout all eight regressions is the coefficient on the interaction term foreignit*CCBit. The sign on this coefficient is expected to be positive for both Retail
and Wholesale Deposit Growth, indicating that depositors have developed an even stronger preference for foreign-owned banks as a result of the CCB episode. Therefore, a positive sign on this coefficient will confirm a significant flow of retail and wholesale funding towards institutions of foreign ownership. The economic logic indicates that the effect, and thus the significance of the coefficient of interest, should be better-pronounced for wholesale rather than for retail funding. Since, according to Bulgarian National Bank’s data, 85% of wholesale deposits do not fall under the deposit insurance protection offered to retail depositors, large corporate depositors should be more concerned with the health of their bank. As a result, they
23
will have a stronger incentive to transfer funds to a “safer” institution when they become anxious about corporate governance issues and excessive risk-taking by their current bank.
4. Data & Descriptive Statistics
This study makes use of several components from the banks’ financial statements, which are collected from the Supervisory Disclosure section of Bulgarian National Bank’s website.3 In
order to allow for comparison between the supervisory practices across Europe, the Community Law and the Bulgarian Law on Credit Institutions require achievement of an adequate level of supervisory transparency. The Supervisory Disclosure section provides quick and easy access to information concerning the adopted national laws and regulations and also summarizes statistical data on key aspects of the implementation of Directive 2013/36/EU and Regulation (EU) No 575/2013. This database incorporates the Balance Sheets, Income Statements, Loans and Deposits statistics for each banking institution operating within the Bulgarian banking system – both domestic banks and foreign banks’ branches. Data is reported on a quarterly basis and is currently available for the period starting in the first quarter of 2004 up to the first quarter of 2016.
The financial statement components of interest, which serve the purpose of dependent variables in this research, are the Retail and Wholesale Deposits reported by each bank for every quarter. These numbers can be found in a separate “Attracted Funds” report for the period ending in the first quarter of 2015. Since then, the Macro-Prudential Form 1 (MPF 1)
3
24
has been employed and deposits are being reported in a joint report alongside securities and loans and advances. Retail deposits are displayed as Households deposits within the report, while Wholesale deposits represent the sum of General governments, Credit institutions,
Other financial corporations and Non-financial corporations’ deposits. From this report are
also collected two of the inputs, which serve as control variables – Interest Expenses and
Total Loans and Advances. Data on the rest of the control variables is gathered from the
published Balance Sheets and Income Statements of the banks. Total Assets, Total
Shareholders’ Equity and Total Liabilities are retrieved from the Balance Sheets, while Net Income and Earnings Before Interest and Taxes are collected from the Income Statement.
The period, for which data was collected, is between the first quarter of 2013 and the fourth quarter of 2015. The decision to limit the time frame within these twelve quarters is motivated by the regression method adopted for the purpose of the research – the difference-in-difference analysis. Since the bank run on Corporate Commercial Bank (CCB) occurred at the end of June 2014, or the end of the second quarter of 2014, the research is focused on the “before-after” effects during the six quarters prior to and the six quarters after this event. The main idea is to have comparable time windows before and after the “treatment” has taken place.
For the aim of the study, it is important to determine the criteria for classifying banks as either domestic or foreign-owned. To achieve that, it is necessary to obtain information about the shareholder structure of each institution. This data is collected from the Commercial
Register of Bulgaria’s Ministry of Justice 4 database, where each institution discloses
information on shareholder structure and how ownership is distributed among shareholders.
4
25
For a bank to be classified as domestic-owned (foreign-owned), the controlling interest should be held by a domestic (foreign) individual or legal entity. Table 2 illustrates a classification of the banking institutions according to their controlling shareholder as of the fourth quarter of 2015. The dataset consists of a total of twenty-eight banking institutions, with foreign-owned banks controlling more than three-quarters of all the assets in the banking system. The biggest stake is held by foreign-controlled banks, which represent legal entities, registered in the Republic of Bulgaria and whose controlling shareholder is either a foreign legal entity or an individual foreign citizen. Most of these banks were initially functioning as state-controlled Bulgarian banks, but after the privatization wave at the end of the twentieth century, they have been acquired by other foreign institutions. Foreign Branches, on the other hand, are subsidiaries, which are subject to dual banking regulation both in Bulgaria and their country of origin. Since these branches' loan limits are dependent on the parent bank's capital, they can provide relatively fewer loans compared to banks, which are solely under the banking regulation of Bulgaria.
Table 2. Distribution of Bank Ownership (as of 31.12.2015)
Banking Institution Number Share of Total Assets
State-controlled 2 3.2%
Bulgarian-controlled 8 21.1%
Foreign-controlled 12 71.8%
Foreign Branches 6 3.9%
Total 28 100.0%
Source: Bulgarian National Bank & Commercial Register of Bulgaria’s Ministry of Justice.
It is important to note that observations during periods, for which neither retail nor wholesale deposits are reported, were excluded from the dataset. This refers to the case of the relatively small Commercial Bank Victoria (EUR 86 million in Total Assets, as of 31.12.2015), which
26
was placed under special surveillance during the second and third quarters of 2014 and thus no data has been reported for this bank during that period. However, this institution has continued with normal operations during the fourth quarter of the year and data on it has been reported ever since.
As visible from Table 3, the banking sector in Bulgaria is strongly segmented and a small fraction of banks holds the majority of deposits, making it hard for smaller banks to take some of their market share. Retail deposits appear to be highly-concentrated and there is a big mismatch between the mean value of BGN 1.37 billion and the median of BGN 0.54 billion.
The average wholesale deposit level is slightly below BGN 1 billion, but its distribution among bank seems to be more homogeneous since the discrepancy with the BGN 0.67 billion is not as big as the one for retail deposits. The mismatch between the mean and the median is also significant for the banks’ Total Assets, which again confirms the segmentation in the industry.
Table 3. Descriptive Statistics: Overall
Variable Mean Median St. Deviation Min Max N
Retail Deposits 1367.96 535.52 1799.38 0.00 7224.47 334 Wholesale Deposits 950.42 668.41 1152.37 7.87 7855.38 334 Total Assets 2916.44 1441.37 3391.60 7.81 17368.32 334 Total Equity 384.15 165.57 537.59 -228.72 2507.79 334 Total Loans 2307.60 1031.86 2912.26 7.10 15224.20 334 Interest Costs 29.48 13.61 42.09 0.02 284.08 334 EBIT 19.52 3.67 53.03 -134.46 339.48 334 Net Income 17.47 3.27 47.94 -134.50 305.25 334
27
As suggested by the overview of the country’s banking system, loans represent the biggest part of the banks’ assets, with the average fraction being around 72%. The profitability also varies a lot and the gap between the highest annual loss and highest annual profit reported during the observed period accounts for nearly BGN 440 million.
By taking a separate look at the summarized statistics of Bulgarian and foreign-owned banks for the period 2013-2015 in Table 4 and Table 5, several conclusions could be derived about market power and profitability for both groups within the country’s banking system. First, foreign-owned banks attract more deposits, both from retail and wholesale customers. The difference is particularly remarkable for wholesale deposits, where both the mean and the median for foreign-owned institutions is nearly twice the size of their Bulgarian-owned counterparts. In regard to maximum rates of wholesale funding, the amount reported by a foreign-owned institution is more than three times larger, accounting for BGN 7.9 billion compared to BGN 2.3 billion recorded for a Bulgarian-owned bank.
Having advantage in average volume of assets, equity, and amount of issued loans, foreign-owned banks make, on average, a smaller amount of interest payments to their depositors – BGN 28 million compared to BGN 31 million average interest expenses by Bulgarian-owned banks. However, judging by both measures included, foreign-owned banks report significantly higher profitability. They record an average annual operating profit of BGN 28 million compared to less than BGN 4 million for Bulgarian-owned banks. This, of course, could be easily explained by the discrepancy in average total assets size, as well as the bigger depositor base that foreign-owned banks enjoy. According to mean estimates of two of the employed profitability ratios, Operating Return on Assets and Return on Assets, foreign-owned institutions report three times higher results. The difference is even more pronounced in terms of the Return on Equity ratio, where domestic-owned institutions report an average return of 1.51%, compared to approximately 10% for foreign-owned competitors.
28
Table 4. Descriptive Statistics: Bulgarian-Owned Banks
The sample excludes Corporate Commercial Bank. Source: Bulgarian National Bank (in millions of BGN). Bulgarian-Owned Banks
Variable Mean Median St. Deviation Min Max N
Retail Deposits 1079.43 420.21 1633.49 1.06 6066.38 118 Wholesale Deposits 592.03 471.81 482.44 36.33 2314.77 118 Total Assets 2024.59 1158.99 2345.56 112.26 8927.28 118 Total Equity 238.00 151.12 241.87 10.94 745.53 118 Total Loans 1364.08 678.00 1780.28 43.67 7089.48 118 Interest Costs 31.68 13.07 51.08 0.38 284.08 118 EBIT 3.82 1.24 10.18 -16.54 42.61 118 Net Income 3.29 1.17 9.37 -16.55 38.35 118 OROA 0.22% 0.09% 0.38% 0.00% 2.64% 118 ROA 0.20% 0.08% 0.35% 0.00% 2.38% 118 ROE 1.51% 0.76% 1.87% 0.00% 8.69% 118
29
Table 5. Descriptive Statistics: Foreign-Owned Banks
The sample excludes Corporate Commercial Bank. Source: Bulgarian National Bank (in millions of BGN). Foreign-Owned Banks
Variable Mean Median St. Deviation Min Max N
Retail Deposits 1525.58 664.46 1868.71 0.00 7224.47 216 Wholesale Deposits 1146.20 823.33 1349.50 7.87 7855.38 216 Total Assets 3403.65 2079.73 3761.28 7.81 17368.32 216 Total Equity 463.99 184.63 630.60 -228.72 2507.79 216 Total Loans 2823.04 1399.00 3264.32 7.10 15224.20 216 Interest Costs 28.28 13.86 36.32 0.02 171.76 216 EBIT 28.10 6.81 63.95 -134.46 339.48 216 Net Income 25.22 6.15 57.80 -134.50 305.25 216 OROA 0.64% 0.35% 0.83% 0.00% 5.89% 216 ROA 0.59% 0.33% 0.77% 0.00% 5.41% 216 ROE 9.98% 2.88% 21.73% 0.00% 111.91% 216
30
5. Results
This study aims to investigate whether the failure of CCB in 2014 had an influence on depositors’ preference for banks of foreign ownership. Two important distinctions are made in my analysis: between the different ex-post effects of the banking crisis on retail and wholesale funding, and between the impact on large (> BGN 1 million worth of assets) and small (≤ BGN 1 billion) banks. The analysis is split into two separate categories, each employing either Retail Deposit Growth or Wholesale Deposit Growth as a dependent variable. Each category incorporates four regressions – two per large and small banks. I present the various regression results related to Retail Deposit Growth in Table 6, and
Wholesale Deposit Growth in Table 7.
In terms of significance, all the coefficients of interest throughout the four regressions in
Table 6 are similar. In line with the initial expectations, I do not find a significant shift of
preference among depositors in smaller banks, which would have been illustrated by significant and positive coefficients on the interaction term foreign*CCB. Even after inserting the additional control variables in the model, no causal effect of the interaction term on Retail
Deposit Growth could be detected. Surprisingly, to some extent, I do not find strong evidence
of stronger retail depositor preference towards large foreign-owned banks as well.
As for the signs on the coefficients of interest, I do not observe the same pattern across large and small banks. Columns (1) and (2) in Table 6 report the output from the regression analysis related to small banks and, for both regressions, the sign on the interaction term
foreign*CCB is negative. This would be a reliable indication that small foreign-owned banks
have actually experienced an outflow of retail funds relative to small domestic banks, provided that the coefficients were statistically significant. However, this is not the case and no general implications could be derived from this regression output. Even though the
31
regression in column (2) is incorporating variables to control for bank-specific characteristics, their coefficients cannot be interpreted as economically meaningful since fixed effects have been included in the regression model as well.
Columns (3) and (4) in Table 6 report the empirical results from the two regressions related to large banking institutions. In contrast to small banks, the coefficients on the interaction term
foreign*CCB are positive. This confirms the expectations that large foreign-owned banks
have become more attractive to depositors after the failure of CCB and managed to increase their retail level of funding relative to their Bulgarian-owned competitors. Once again, firm and time-fixed effects are being utilized in order minimize the bias related to the independent variable of interest. However, the coefficients on the interaction term foreign*CCB remain below the 10% significance level, even after plugging in the control variables in column (4).
The bottom line from this part of the analysis is that no definite implications could be made about the impact of the bank run on CCB on the preference of Bulgarian retail depositors towards either domestic or foreign-owned banks. Due to the absence of statistically significant coefficients of interest, not enough evidence is provided in support of Hypothesis
2 and Hypothesis 4. Even though the logic suggested that Bulgarian-owned banks might be
seen by small depositors as riskier and poorly managed, I do not find strong evidence supporting the hypothesis that the crisis of 2014 has caused further damage for large domestic-owned retail deposit-takers. Moreover, no convincing difference is observed between small and large-scale banks since the results for both groups of institutions do not indicate significant deposit flows. A few plausible explanations will be discussed later in this section.
32
Table 6. Retail Deposit Flows after the Failure of CCB: Estimates for Banks of Different Size
This table looks at the determinants of the growth rate of Retail Deposits for both small and large banking institutions. In columns 1-2, results are reported for small banks with less than BGN 1 million worth of assets (as of Q4 2015). In columns 3-4, results are related to large banks, with at least BGN 1 billion worth of assets. The dependent variable is used as a proxy for customer deposit flows. Corporate Commercial Bank has been excluded from the dataset, as well as the observations for Commercial Bank Victoria for Q3 and Q4 of 2014, for which no data has been reported. Foreign is a dummy variable that indicates whether the banking institution is domestic or foreign-owned and takes the value 1 for the foreign-owned banks. CCB is a dummy variable, which takes the value 1 if the observation has been recorded after the CCB episode (end of Q2 2014), and 0 if before. Foreign*CCB is the interaction term between the two dummies. The Foreign dummy is automatically omitted by the statistical software in order to avoid multicollinearity. For definitions of all control variables, see Section 3. All regressions use quarterly panel data from 31.01.2013 until 31.12.2015. Clustered standard errors are reported in parentheses. *, **, and *** indicate significance at 10%, 5%, and 1%, respectively.
Dependent Variable: Retail Deposit Growth (%)
(1) (2) (3) (4) Foreign omitted omitted omitted omitted
CCB 1.0734 1.6321 -0.0299 0.0203 (0.91) (1.80) (0.06) (0.04) Foreign*CCB -1.7853 -1.8379 0.0556 0.1025
(1.71) (1.77) (0.07) (0.09) log (Total Assets) 5.3652 0.2742
(4.41) (0.22) Market Share -359.3626 -1.1498
(650.65) (1.85) Total Equity/Total Assets 11.4086 0.3092
(12.70) (0.99) Operating Return on Assets -17.2993 -10.7470
(44.99) (9.09) Total Loans/Total Assets 19.4931 0.4255
(16.33) (0.43) Interest Costs/Total Liabilities -135.3090 -8.6108
(115.63) (5.99) Bank Size Small Small Large Large Firm-Fixed Effects Yes Yes Yes Yes Time-Fixed Effects Yes Yes Yes Yes Clustered Standard Errors Yes Yes Yes Yes
N 97 97 216 216
33
The regressions in Table 7 should be addressed with the assumption that wholesale depositors are more sensitive to episodes of banking crises, since 85% of wholesale deposits are exceeding the insurable threshold of EUR 100,000 (as of December 2015). All four regressions included in the table measure the causal effect that the failure of CCB, combined with the origin of ownership, has had on wholesale deposit levels for both small and large banks within the Bulgarian banking sector. The dependent variable is Wholesale Deposit
Growth, measured as the percentage change between two neighbouring quarters. The results
referring to small banks are matching the initial expectations again: no significant changes in wholesale funding levels have been experienced by smaller foreign-owned banks relative to their domestic-owned competitors since June 2014. On the other hand, the analysis of large banking institutions yields statistically significant and positive coefficients on the interaction term foreign*CCB, thus providing empirical support for the hypothesis that big foreign-owned banks have attracted significantly more wholesale deposits relative to domestic-foreign-owned counterparts.
Columns (1) and (2) in Table 7 report the output from the two regressions, which explore whether small foreign-owned banks have significantly increased their levels of wholesale funding relative to Bulgarian-owned banks. In contrast to the results for retail deposits, now it can be noticed that the sign on the interaction term foreign*CCB’s coefficient is different in each of the two columns. In column (1), where the dependent variable is only regressed on the two dummies foreign and CCB and their interaction term, I observe a negative, though insignificant coefficient on the independent variable of interest. This would bring the implication that small foreign banks have actually become less attractive to wholesale depositors, provided, of course, that the coefficient was significant. In column (2), with the introduction of bank-specific characteristics as control variables, the sign on the foreign*CCB coefficient turns out to be positive. However, its level of significance is again below the 10%
34
limit. Both regressions incorporate fixed effects and clustered standard errors, allowing for autocorrelation within the entities. Given the lack of significance of the coefficients of interest, no reliable evidence for significant shifts in small-bank depositors’ preference towards either foreign or domestic-owned institutions can be derived.
The output in columns (3) and (4) from Table 7 matches the expectations that a shift in depositor preferences would most probably be detected when investigating wholesale deposit levels in large-scale banks, thus confirming Hypothesis 1. In column (3), I regress the
Wholesale Deposit Growth on the two dummies and their interaction term, including time and
firm-fixed effects. The coefficient of interest is positive and significant at the 5% level, indicating that since the failure of CCB, large foreign-owned banks have experienced an approximate 11% increase of wholesale deposit levels relative to large Bulgarian-owned competitors. After inserting the control variables in the model in column (4), the coefficient on the foreign*CCB interaction remains positive and significant at the 5% level. Now, with increased explanatory power, the coefficient of interest has an even higher estimate, implying a relative growth of more than 15% in wholesale deposits for large foreign-owned banking institutions. Moreover, the results also provide confirmation of Hypothesis 3 stating that the joint impact of ownership and the run on CCB on deposit growth should be stronger for wholesale relative to retail funding. Within this set of regressions employing Wholesale
Deposit Growth as a dependent variable, Hypothesis 4 receives justification as well. In
contrast to the weak results referring to small banks, the coefficient measuring the impact on large banks is significant on the 5% level, thus confirming the expectations that bigger institutions have been more affected by the events of 2014.
35
Table 7. Wholesale Deposit Flows after the Failure of CCB: Estimates for Banks of Different Size
This table looks at the determinants of the growth rate of Wholesale Deposits for both small and large banking institutions. In columns 1-2, results are reported for small banks with less than BGN 1 million worth of assets (as of Q4 2015). In columns 3-4, results are related to large banks, with at least BGN 1 billion worth of assets. The dependent variable is used as a proxy for customer deposit flows. Corporate Commercial Bank has been excluded from the dataset, as well as the observations for Commercial Bank Victoria for Q3 and Q4 of 2014, for which no data has been reported. Foreign is a dummy variable that indicates whether the banking institution is domestic or foreign-owned and takes the value 1 for the foreign-owned banks. CCB is a dummy variable, which takes the value 1 if the observation has been recorded after the CCB episode (end of Q2 2014), and 0 if before. Foreign*CCB is the interaction term between the two dummies. The Foreign dummy is automatically omitted by the statistical software to avoid multicollinearity. For definitions of all control variables, see Section 3. All regressions use quarterly panel data from 31.01.2013 until 31.12.2015. Clustered standard errors are reported in parentheses. *, **, and *** indicate significance at 10%, 5%, and 1%, respectively.
Dependent Variable: Wholesale Deposit Growth (%)
(1) (2) (3) (4) Foreign omitted omitted omitted omitted
CCB 0.1640 0.1185 -0.0630 -0.1781 (0.27) (0.20) (0.05) (0.17) Foreign*CCB -0.1132 0.0557 0.1094** 0.1543**
(0.20) (0.15) (0.05) (0.07) log (Total Assets) 0.2711 0.3754
(0.50) (0.27) Market Share -121.5863 -4.1914
(103.93) (4.53) Total Equity/Total Assets 0.0741 -2.7139**
(0.95) (1.08) Operating Return on Assets -7.4762 2.2624
(5.08) (7.23) Total Loans/Total Assets -0.3742 -0.2300
(0.43) (0.37) Interest Costs/Total Liabilities -9.8334** 3.8041
(4.71) (9.50) Bank Size Small Small Large Large Firm-Fixed Effects Yes Yes Yes Yes Time-Fixed Effects Yes Yes Yes Yes Clustered Standard Errors Yes Yes Yes Yes N 117 117 216 216 R-squared: within 0.1154 0.2142 0.1495 0.2346
36
Despite the fact that not all four hypotheses were confirmed, the empirical results from this research appear to be economically meaningful and essentially logical. Due to the strong segmentation of the Bulgarian banking industry, it could not be expected to easily detect significant deposit flows in terms of either retail or wholesale funding for smaller banks (≤ BGN 1 billion). It can be assumed that these institutions have very narrow bases of depositors, holding relatively smaller deposit amounts. Moreover, some of these customers may even be insiders or in close connection to the bank management. Therefore, it is expected to observe a lower level of adverse information among these depositors, which is one of the main triggers of bank runs, according to Allen and Gale (1998).
Having a wider depositor base with higher deposited amounts, large banks (> BGN 1 billion) might be more fragile when depositors have reasons to become insecure about a given bank’s perspectives. This was the case in June 2014, when the two biggest Bulgarian-owned financial institutions witnessed severe difficulties. CCB experienced a bank run, the effects from which spread on FIB as well. As a result, the first bank went into bankruptcy, while the latter had to receive liquidity injection from the state in order to continue operations. With respect to the period after the outburst of this crisis, however, I do not detect evidence for sudden shifts in retail depositors’ behavior, which would result in foreign-owned banks attracting significantly more retail funding relative to Bulgarian-owned competitors. A plausible explanation could be the Deposit Insurance Policy of the state, which was put into practice for the first time after CCB had closed down and its retail depositors had to be fully compensated for deposit amounts up to EUR 100,000. Even though most of the retail depositors in Bulgaria did not seem to know about the existence of the Insurance Fund prior to the events of 2014 (which could be one reason behind the two banks runs), it would be a plausible guess that this new awareness has prevented large retail deposit outflows from Bulgarian towards foreign-owned banks.
