Master Thesis
The impact of bank capital on profitability in the EU:
An examination of multiple capital ratios
By Janine Lotte Pluimers
MSc. Finance & MSc. International Financial Management Faculty of Economics and Business
University of Groningen
Student nr: s2043645 Phone: +31 6 11402639
Email: lotte_pluimers@hotmail.com Date: 14 January 2016
1
The impact of bank capital on profitability in the EU:
An examination of multiple capital ratios
Janine Lotte PluimersABSTRACT: Using a panel of European banks, this research examines how capital affects
bank profitability and how this effect varies across non-crisis and crisis periods. Three types of capital ratios are examined; the equity-asset ratio, the tier 1 ratio, and the total capital ratio. The results indicate that bank profitability is positively influenced by the equity-asset ratio under all circumstances. Additionally, the total capital ratio increases profitability during crisis periods but deteriorates profitability during non-crisis periods. Therefore, Basel III capital requirements will positively influence bank performance during crisis periods, which enhances the stability the banking sector and the economy as a whole.
JEL classifications: G01, G28, G21.
Keywords: banking, profitability, bank capital, financial crisis.
1. Introduction
2 Not surprisingly, the influence of capital on bank performance has become of major interest. 1 Especially as literature offers conflicting predictions about the relation between capital and profitability. Various authors emphasize that banks should hold more capital, as more capital should provide banks with a safety net, improving social efficiency (see, e.g., Calomiris and Herring (2011); Hart and Zingales (2011)). However, Goddard et al. (2010) and Altunbas et al. (2007) find that increasing the level of capital has a negative impact on the performance of the bank. In addition, Haq and Heaney (2012) find capital deficits to be strongly negatively related to future profitability. Summarizing, capital might have positive as well as adverse effects on bank profitability. Hence this paper extends and continues on the observations from previous academic research.
The main research question of this paper addresses the empirical impact of bank capital on the financial performance of European Union banks. This research investigates this issue by performing a panel data analysis on 462 banks within the European Union for the period from 2006 to 2014. The research is in line with Berger and Bouwman (2013), however, a few deviations are made. In contrast to with Berger and Bouwman (2013), who mainly investigate the equity-asset ratio, this research examines various types of capital ratios. The equity-capital ratio does not distinguish any differences amongst assets, and does not take off-balance sheet activities into account. This research fills this gap by investigating two additional capital ratios, namely the tier 1 ratio and the total capital ratio, which also take risk-weighted assets and off-balance sheet activities into account. By examining all three ratios, a comparison can be made between the traditional capital ratio and the risk-based capital ratios, and their predictional power in explaining profitability.2
Furthermore, this research deviates from Berger and Bouwman (2013) by investigating bank profitability as a measure of bank performance, whereas their research is based on bank survival and market share. The relevance in shifting the attention to profitability, is that bank profitability is of fundamental importance to the stability of the banking industry, the impact of the banking industry on capital markets and, the economy as a whole (Dietrich and Wanzenried, 2011). This is particularly true in the aftermath of the recent financial crisis, for which separate analyses are included. Through this additional analysis, the effect of pre-crisis capital on subsequent bank performance is determined.
1 By capital we refer, here and elsewhere in the paper, to the book value of equity.
2 By risk-based capital ratios we refer, here and elsewhere in the paper, to the tier 1 capital ratio and the total
3 This research focuses on relatively large banks (assets in excess of $10 billion) and performs additional analyses for large and medium bank subsamples. According to Demirguc‐Kunt et
al. (2013), large banks are typically more sophisticated organizations with complex balance sheets, operating on a global scale. These banks are of major importance to the stability of the system as a whole, and hence particularly important to analyze.
The results of this paper contribute to the literature in several ways. Firstly, it will shed further insight on how capital relates to bank performance. Secondly, by examining all three capital ratios, a comparison can be made in explaining profitability between the equity-asset ratio, the tier 1 ratio and the total capital ratio. Thirdly, it studies how the new Basel III capital regulations will influence bank performance. This research will examine how pre-crisis capital influences bank performance during a crisis. This will give more insight in the effectiveness of capital regulations, since these capital regulations should prevent future crises. The results of this study should be of particular interest to a diverse set of stakeholders, including bank regulators, policy makers, practitioners, financial economists, as well as individual investors.
The remainder of this paper is structured as follows. Section two presents the theoretical background and the hypotheses development. Section three describes methodology, variables and the sample construction. Section four shows the empirical results on the influences of capital on bank profitability and performs additional robustness checks. Finally, in section five conclusions and final remarks are described.
2. Literature review
In this section, an extensive literature research has been performed to analyze the current state of research with regards to capital and its interaction with bank profitability. Therefore, a review of the most important angles and relations will be presented.
2.1 Relationship between capital and profitability
4 (Modigliani and Miller, 1995). Furthermore, due to the tax deductibility of interest payments, a higher capital ratio will lower the after-tax income of banks (Berger, 1995).
A considerable amount of research states a negative relationship between capital and profitability, which supports the theory of Modigliani and Miller (1995). For example, Altunbas et al. (2007) find that increasing the level of capital can contributes negatively to the performance of the bank. They argue that inefficient European banks seem to hold more capital, compared to efficient European banks. In addition, Goddard et al. (2010) state that the negative relationship between capital and earnings reflects the standardized risk-return payoff. In addition, Berger and Bonaccorsi di Patti (2006) find capital to be negatively related to profit efficiency, whereas Haq and Heaney (2012) find capital deficits to be strongly negatively related to future profitability.
However, other studies provide a different view on the relationship between capital and earnings, whereby capital and profitability are positively related (see, e.g., Goddard et al. (2004) and Iannotta et al. (2007)). In addition, Demirguc-Kunt and Huizinga (2010) find a positive relation between capital and profitability, explained by the reduction of costs in funding, cost efficiency, managerial incentives, and asset monitoring. Berger and Bouwman (2013) state that capital improves the performance of medium and large banks, especially during banking crises. Beltratti and Paladino (2015) find that an increase in the equity-asset ratio increases the residual income of banks. Furthermore, literature argues that deviations from the Modigliani and Miller theorems are particularly relevant for the banking industry, by which they frequently derive from the expected bankruptcy costs theory (Buser et al. (1981); Berger (1995).
5 higher expected bankruptcy costs. Increasing the equity-asset ratio lowers the probability of bankruptcy, whereby the expected bankruptcy costs are lowered (Berger, 1995). To illustrate this, if a bank is below its optimal equity-asset ratio, expected bankruptcy costs might be relatively high. Therefore, an increase in its equity-asset ratio will result in an increase in expected profitability due to the lower interest expenses on uninsured debt (Goddard et al., 2004).
Berger (1995) argues that when banks are faced with a sudden increase in the probability of default, they are probably operating below their optimal equity-asset ratio. Therefore, a bank which is able to identify its optimal equity-asset ratio together with the ability to adjust to the altered optimal equity-asset ratio could reduce the probability of default. This is because being able to adapt to the new optimal ratio, will result in lower costs of uninsured debt and in higher profitability, ceteris paribus. Berger (1995) observes that the expected costs of bankruptcy hypothesis are most relevant for “risky banks”. These risky banks, with high leverage or high portfolio risk, are assumed to benefit the most from an increase in their equity, as these banks experience the greatest marginal benefit in reducing risk by increasing the equity-asset ratio.
An alternative theory in explaining the link between capital and profitability is explained by the signaling hypothesis. Thereby, bank managers can have a stake in the bank through ownership or options. Allowing bank managers to (partially) own the bank can be advantageous, because they could possibly have access to superior information. According to Berger (1995), it would be cheaper for a bank with “good” private information regarding future revenues to signal high quality by increasing the equity-asset ratio compared to a bank with “bad” information. Thereby a signaling equilibrium exists in which banks that have positive private information regarding future cash flows, increase their level of equity.
2.2 From profitability to capital
6 the long term, higher earnings can result in a smaller capital buffer, as more profitable banks know they will be able to allocate their internal funds to finance future investment opportunities. In addition, less profitable banks might also engage in higher portfolio risk and higher leverage (Milne and Whalley, 2001). This indicates a negative relation in the long run (Myers, 1984). On the contrary, Jensen (1986) argues that managers of more profitable banks retain excess profits because of personal incentives, i.e. for their own personal projects or ambitions, by which a negative relation can be observed between capital and earnings. However, the remainder of this paper concentrates on theories explaining the causality from capital to profitability, not vice versa. This relationship is most relevant for today’s policy debate concerning capital regulation and it challenges conventional theories (Berger, 1995).
2.3 Capital during financial crises
Berger and Bouwman (2013) observe that banking crises are commonly connected to a considerable amount of bank failures and near-failures. According to Osborne et al. (2009), banks aim to deleverage their balances during crises by acquiring higher capital ratios, which in turn decreases the required rate of return on risky debt and thus lowers the probability of default.
7 way with lesser-capitalized banks. The research of Demirguc‐Kunt et al. (2013) confirms this,
as their results show that the importance of a stronger capital position becomes more evident during crisis periods, especially for large banks.
2.4 Basel III and European banking
During the recent financial crisis, numerous banks struggled for their survival. According to BCBS (2010), one of the main causes of the recent financial crisis was that the banking sector had built up excessive on- and off-balance sheet leverage, combined with a substantial decrease of the level and quality of a bank’s capital base. To prevent such future crises, the BCBS introduced fundamental reforms to the international regulatory framework, known as Basel III. Thereby, new capital, leverage, and liquidity standards are implemented in order to enhance regulation, supervision, and risk management. New capital standards require banks to hold more capital and increase the quality of their capital base.
Härle et al. (2010) studied the impact, possible responses, and the challenges of implementation of Basel III within European banks. They claim that the impact of the new requirements is substantial, as they estimate a capital deficit of €1.1 trillion and a short-term liquidity deficit of €1.3 trillion within Europe.3
Also, short-term and long-term funding deficits are expected to have a substantial impact, as they estimate a long-term funding deficit of €2.2 trillion. Responding to these shortfalls will have a major influence on bank profitability. Holding all things equal and assuming full implementation in 2019, Basel III would reduce ROE with approximately 4.0% for the average bank compared to the pre-crisis levels. In line with Härle et al. (2010), the Institute for International Finance (2010) predicted an annual decrease in bank ROE of 3.5% for European banks over the years 2011–2020, as a result of higher capital requirements. Banks are already trying to mitigate the impact of the capital requirements by building their capital and funding stocks, and taking risk off their books. Hereby, raising capital requirements will have a negative impact on ROE, indicating a negative influence of capital on profitability.
2.5 Bank capital buffers
Existing literature indicates that banks hold far more capital than required by regulators (see, e.g., Jokipii and Milne (2008); Demirguc‐Kunt et al. (2013). Thereby, implying that banks hold a portion of their capital as a capital buffer. Several reasons have been put forward in
3
8 explaining why banks hold capital buffers. For example, shareholders may have incentives to add to bank capital, reducing bank risk and the cost of deposits, in case of not totally insured bank liabilities (Fonseca and González, 2010). Also, banks might hold excess capital in order to signal soundness and satisfy the expectations of rating agencies (Jokipii and Milne, 2008), which is in line with the signaling hypothesis. Furthermore, the excess capital might protect banks against potential violations of the regulatory minimum (Milne and Whalley, 2001). Another possible reason for excess capital is to take advantage of potential growth opportunities. Empirical studies on capital buffers mostly focus on analyzing the cyclical behavior of capital buffers.4 However, it is not clear how capital buffers relate to bank profitability.
2.6 Hypotheses development
New capital standards require banks to hold more capital and increase the quality of their capital base. However, the academic literature offers conflicting predictions about how the increase in capital relates to bank profitability. This emphasizes the importance of this study, as this paper examines the effect of capital on profitability within European banks.
Following Modigliani and Miller (1995), bank profitability is a negative function of the capital ratio. Whereas, for example, Demirguc-Kunt and Huizinga (2010) find a positive relation between capital and profitability, explained by the reduction in the cost of funding, cost efficiency, managerial incentives, and asset monitoring. All in all, the existing empirical studies show that there is evidence of a significant positive as well as a significant negative relationship between capital and bank profitability. Therefore, the sign of the coefficient examining the relation between capital and profitability cannot be predicted.
Berger and Bouwman (2013) and Demirguc‐Kunt et al. (2013) find that better-capitalized
banks are more resilient to an unexpected negative shock, and their financial performance not as much deteriorated compared to poorly capitalized banks. This suggests a positive relation between pre-crisis capital and bank profitability. Regarding non-crisis periods, Berger and Bouwman (2013) observe capital to be negatively related to profitability for large banks. In line with this, the following hypotheses will be examined:
H1: Capital will negatively affect bank profitability during non-crisis periods. H2: Capital will positively affect bank profitability during financial crisis periods.
4
9 3. Methodology, variables and data
This section first describes the empirical approach. Then, the key variables and control variables used in the model are discussed. Further, the sample selection is explained and descriptive statistics are provided.
3.1. Methodology
The general empirical approach is based on the studies of Berger and Bouwman (1995),
Demirguc‐Kunt et al. (2013), Goddard et al. (2010), and Lee and Hsieh (2013), whereby the
effect of lagged capital on the profitability of banks is examined. In contrary to Berger et al. (2013) and Demirguc‐Kunt et al. (2013), who use quarterly data, yearly data is used. Most of
the explanatory variables are available on a yearly basis rather than a quarterly basis. Furthermore, yearly data is more consistent and reliable, as this data is published in annual reports. Various versions of the following basic equation will be estimated:
∆π𝑖,𝑡 = 𝛼1∗ 𝐾𝑡−1 + 𝛼2 ∗ 𝑋𝑡−1+ 𝜆𝑡+ 𝜇𝑖 + ℇ𝑖,𝑡 (1)
Whereby ∆πi,t represents the percentage change in profitability of a bank i in year t, measured as the change in ROE. Kt-1 represents averaged lagged bank capital, the variable of interest. Xi,t−1 represents a set of control variables, λt and μi represent the time fixed effects and bank fixed effects, and the latter term, ℇi,t, represents the error term. By regressing performance on lagged explanatory variables, the empirical set-up mitigates endogeneity concerns for reverse causality.
As the recent financial crisis impacted banks worldwide, its effects cannot be ignored when investigating the relation between bank capital and profitability. Therefore, an additional equation investigates how prior capital relates to bank profitability during financial crises by analyzing the recent financial crisis. The following ordinary least squares (OLS) regression will be examined:
∆π𝑖,𝑝𝑟𝑒−𝑡 = 𝛼1∗ 𝐾𝑖,𝑝𝑟𝑒−𝑡 ∗ 𝑛𝑜𝑛 − 𝑐𝑟𝑖𝑠𝑖𝑠 + 𝛼2∗ 𝐾𝑖,𝑝𝑟𝑒−𝑡 ∗ 𝑐𝑟𝑖𝑠𝑖𝑠 + 𝛼3∗ 𝑋𝑖,𝑝𝑟𝑒−𝑡+ 𝜆𝑡+ 𝜇𝑖+ ℇ𝑖,𝑡
(2)
10 variables of eq. (1). Two additional dummy variables are included, namely non-crisis and crisis. Wherenon-crisisis a dummy variable equal to one for the years preceding and after the financial crisis, and zero otherwise. Crisis is a dummy variable taking the value of one during the years in which the financial crisis was unfolding, and zero otherwise. The years 2007, 2008, and 2009 are classified as crisis years, which is in line which several researches (see, e.g., Demirguc‐Kunt et al., 2013).
Prior to the actual regressions, some additional tests and adjustments are made to correct for frequently present econometric threats. Bank and time fixed effects are introduced into the equation, represented by 𝜆𝑡 and 𝜇𝑖, to reduce concerns about possible omitted variables. Bank fixed effects control for time-invariant omitted variables that affect profitability cross-sectionally (e.g. quality of bank regulation and supervision). Time fixed effects, control for macroeconomic trends that vary over time but not cross-sectionally (e.g. interest rates and inflation). To minimize the effect of endogeneity, the control variables are lagged by one period. To check for possible multicollinearity, a correlation matrix is constructed of all main variables, which will be discussed in section 3.2. To control for the possible influence of outliers, all balance sheet variables are winsorized at the 1st and 99th percentile. The main dependent variable, ∆πi,t , is winsorized at the 3rd and 97th percentile.5 For normalization purposes all variables are based on ratios, whenever possible.
3.2 Variables
The main dependent variable of this research is ∆π𝑖,𝑡 , calculated as ∆ROE. ROE is measured as the net income divided by the total equity. ROE is a frequently used profitability measure as it accounts for both the on- and off-balance sheet activities of banks (see, e.g., Petria et al. (2015); Lee and Hsieh (2013)). A bank’s percentage change in profitability, ∆πi,t , is defined as the bank’s ROE minus the bank’s one year lagged ROE divided by its one year lagged ROE (Eq. 3). ∆πi,t is winsorized at 3% in both tails.
∆π𝑖,𝑡 =
𝑅𝑂𝐸𝑖,𝑡 − 𝑅𝑂𝐸𝑖,𝑡−1 𝑅𝑂𝐸𝑖,𝑡−1
(3)
The main explanatory variable of this research is bank capital. As stated before, this paper differentiates between three types of capital ratios; the equity-asset ratio, the tier 1 capital
5 In unreported robustness checks different levels of winsorizing (e.g. 1%, 3%, and 5%) are applied. This,
11 ratio, and the total capital ratio. All capital ratios represent two-year averages before time period t (i.e. the average value of ROEt-1 and ROEt-2). By using lagged estimates, it can be determined whether higher capital in prior years influences performance in years thereafter. This is particularly interesting when entering into a financial crisis, as it is not known beforehand when a crisis will strike. Furthermore, problems of endogeneity are reduced, because it is not likely that lagged capital and current performance are jointly determined.
The first examined measurement of bank capital is the traditional measurement of bank capital, equity over total assets. The second and third capital ratios are based on the Basel III capital ratios, for which Basel III sets regulatory minimums. As a second capital measure, the tier 1 ratio is examined. Tier 1 capital contains shareholders’ funds and perpetual, noncumulative preference shares, which are divided by a bank’s risk-weighted assets. The third capital ratio is the total capital ratio. The total capital ratio is obtained by the sum of tier 1 and tier 2 capital divided by risk-weighted assets and off-balance sheet exposures. Tier 2 capital also considers additional bank capital, such as revaluation reserves, undisclosed reserves, hybrid instruments, and subordinated term debt.
Basel III minimum requirements will be implemented in several phases over time. The first phase of Basel III implementation started in 2013, whereas the last phase is expected to be fully implemented in 2019 (see table 1). A more extensive overview of all Basel III requirements and their implementation phases is presented in appendix A. To reduce the impact of outliers, all bank’s capital ratios are averaged over two lagged years and wisorized at the 1st and 99th percentile.
Table 1: Basel III phase-in capital requirements
2006-2012 2013 2014 2015 2016 2017 2018 2019
Tier 1 ratio 4.00% 4.50% 5.00% 5.50% 6.00% 6.50% 7.00% 7.50% Total capital ratio 8.00% 8.00% 8.00% 8.00% 8.00% 8.00% 8.00% 8.00% Table 1 reports minimum requirements for the tier 1 and the total capital ratio and their years of implementation. All dates are as of 1 January. Source: Basel Committee for Banking Supervision (BCBS), 2010.
12 To isolate the effect of capital on profitability, the equation controls for time-specific and bank-specific characteristics that might influence profitability. Furthermore, we also allow for the possibility that the impact of equity on profitability might depend on other important variables, such as liquidity. Including these variables, which relate to both profitability and capital, resolves the issue of a potential omitted variable bias. In this section, the choice of these control variables is motivated and their calculations will be discussed. Control variables are based on previous academic literature in explaining bank profitability and bank equity (i.e. Berger and Bouwman (2013); Beltratti and Paladino (2015) and Demirguc‐Kunt et al.
(2013)). The following control variables are included in the equations: deposits, income diversity, liquid assets, bank size, loans, and GDP growth. Each control variable is lagged over one year before time period t. An overview of all variables is presented in table 2.
Deposits represent the ratio of deposits to total assets. Banks have deposits that are insured to protect depositors and to ensure financial stability. It is expected that a higher deposit ratio is positively related to bank performance, as deposit financing is not subject to runs in case of deposit insurance, but money market funding is (Gorton, 2010). Furthermore, Gropp and Heider (2010) argue that to mitigate the potential horal-hazard problem of the insurance, bank have to hold a minimum amount of capital on their balance sheet. This indicates a positive relationship between the deposits ratio and bank capital.
Income diversity measures the extent to which a bank’s activities are diversified away from traditional banking. In recent years, banks have moved along the spectrum from pure commercial banking to more specialized banking, such as asset management and commission paying services (Aebi et al., 2012). Leaven and Levine (2007) observe that diversification within financial institutions is negatively related to firm performance. The measure of income diversity is based on Laeven and Levine (2007) and is defined as follows:
Income diversity
= 1 − | 𝑁𝑒𝑡 𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 − 𝑂𝑡ℎ𝑒𝑟 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝐼𝑛𝑐𝑜𝑚𝑒
𝑇𝑜𝑡𝑎𝑙 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑖𝑛𝑐𝑜𝑚𝑒 |
13 Table 2 Variable definition
Variable Description Source Calculation
∆π Percentage change in return on equity
Computation on Bankscope data
(ROEt - ROEt-1) / ROEt-1, where
ROE = Net income / total assets NIM Net interest margin Bankscope Net interest income / Average
earning assets Equity-asset ratio Average of two year lagged
equity-asset ratio
Computation on Bankscope data
Book equity /total assets Tier 1 ratio Average of two year lagged tier 1
ratio
Computation on Bankscope data
- Total capital ratio Average of two year lagged total
capital ratio
- Capital buffer Averaged of two year lagged
capital buffer
Computation on Bankscope data
Capital - capital requirement Non-crisis Dummy which takes the value of 1
during normal times, 0 otherwise
Empirical literature Dummy with value of one in the years 2006, 2010-2014, zero otherwise.
Crisis Dummy which takes the value of 1 in crisis years, 0 otherwise
Empirical literature Dummy with value of one in the years 2007 -2009, zero
otherwise. Deposits Deposits to total assets Computation on
Bankscope data
Deposits/total assets Income diversity Income diversity ratio Computation on
Bankscope data
1 - ((Net interest income - other operating income) / total operating income)) Liquid assets Liquid assets are trading assets and
loans and advances with a maturity of less than three months
Computation on Bankscope data
Liquid assets/total assets
Bank size Natural logarithm of total assets Computation on Bankscope data
LN(total assets) Loans Gross loans to total assets Computation on
Bankscope data
Loans /total assets GDP growth Annual percentage change of
Gross Domestic Product
Eurostat (Real GDPt / Real GDPt-1) - 1
This table provides the definitions, descriptions, sources, and calculations of the independent and dependent variables. Dependent variables are winsorized with 3% in both tails, all other variables are winsorized at the 1st and 99th percentile.
14 In order to control for bank size, the natural logarithm of total assets is used. Larger banks are assumed to have better risk diversification opportunities and lower cost of funding, which is positively related to bank profitability. Furthermore, supported by the “too-big-to-fail” argument, larger banks are expected to benefit from a guarantee that decreases their cost of funding and allows them to invest in riskier assets (Iannotta et al., 2007). According to Pasiouras and Kosmidou (2007), bank size is positively related to profitability, because larger banks have a higher degree of product diversification and loan diversification. Furthermore, they should benefit from economies of scale. However, Berger et al. (1987) claim that large banks often experience scale inefficiencies, by which size is negatively related to bank profitability. Empirical theories offer contradicting predictions about the relation between size and the capital position. According to Ahmad et al. (2008), large banks have relatively easy capital market access and depend on possible bailout policies of the government, which might cause large banks to hold on to a relatively low capital ratio. Indicating a negative relation between size and bank capital. On the contrary, higher earnings might lead to greater diversification and more investment opportunities, lowering the cost of capital. This provides incentives to large banks to raise capital, thereby avoiding risk. In additional robustness checks, subsamples will be created for medium and large banks. Thereby, separate analyses are conducted to analyze the effect of capital on profitability, based on bank size.
In investigating the banks’ business model, the ratio of net loans to total assets is measured. In banking literature this is used as a measure of lending specialization and credit risk. A relatively high loan ratio might suggest that a bank is specialized in lending, whereby it benefits from information advantages, which reduce intermediation costs and improve profitability (Goddard et al., 2010). However, Goddard et al. (2010) also indicate that loans are generally regarded as riskier and less liquid compared to other assets on a bank’s balance sheet. A high loan/asset ratio might indicate a greater risk of failure, which can be negatively related to profitability. Furthermore, a high loan ratio could be an indicator of high risk. Modigliani and Miller (1995) state that higher capital reduces the risk on equity. Banks might increase their capital ratio to compensate for high risk, measured by a high loan ratio. This indicates a positive relation between capital and bank loans.
15
3.2 Data
This study analyses a panel data set of European countries over the period 2006-2014. Annual bank-level data of the European Union 28 region (EU28) banks is used for the period from 2004 to 2014.6 Data is obtained from the Bureau Van Dijk’s BankScope database. To select the sample, all active banks in the EU28 region with assets in excess of $10 billion in at least one of the selected years are selected. These criteria are in line with Iannotta et al. (2007), who also investigate European bank performance. The sample includes all commercial, savings, and co-operative banks. This results in a panel data set in which a total of 462 banks are included. Not all banks enter the sample in every year, therefore the sample is unbalanced. All EU 28 countries are represented in the sample, except for Bulgaria.7 For a full overview of the composition of banks per country, see appendix B.
Unfortunately, the data availability of the tier 1 ratio and the total capital ratio is not as exhaustive compared to the equity-asset ratio. In order to adequately compare the different capital ratios, only observations are investigated for which all three variables are available.
The sample only consists of European banks for several reasons. Firstly, these countries have a high degree of regulatory convergence. In addition, they are all supervised by the European Central Bank (ECB), and have therefore a high degree of comparability. Secondly, accurate financial and accounting information is easily accessible for these countries. Third, the scope of this research involves understanding the potential effects of Basel III’s capital requirements, therefore it is logical to examine countries that already have implemented Basel II. The European Union region is known as the most Basel II-convergent region, as it has the best implementation of Basel I and II, and has minimum requirements for the tier 1 ratio and total capital ratio (Chalermchatvichien et al., 2014). Furthermore, only relatively large banks are analyzed. As already mentioned, these banks are of major importance to the stability of the system as a whole, and therefore, particularly interesting to investigate.
6
This paper analyzes the years 2006 – 2014. However, because lagged variables are used for explaining bank profitability, data is obtained for the year 2004 and onwards.
7 Bureau Van Dijk’s BankScope does not report any Bulgarian banks with assets in excess of $10 billion for the
16 Table 2: Descriptive statistics
Mean Median Maximum Minimum Std. Dev. Observations Dependent Variables
ROE 0.048 0.067 0.248 -0.346 0.126 1371
∆ROE% -0.268 -0.111 3.242 -5.063 1.429 1371
Net interest margin 0.018 0.016 0.044 0.002 0.010 1371 ∆Net interest margin 0.014 -0.001 0.587 -0.440 0.186 1371 Main independent variables
Equity-Asset ratio 0.065 0.061 0.263 0.007 0.033 1371
Tier 1 ratio 0.114 0.106 0.324 0.058 0.042 1371
Total capital ratio 0.139 0.129 0.353 0.081 0.042 1371 Control variables Deposits 0.676 0.696 0.971 0.136 0.168 1371 Income diversity 0.428 0.403 1.348 -0.322 0.242 1371 Liquid assets 0.216 0.179 0.926 0.013 0.163 1371 Bank size 17.952 17.612 21.522 14.256 1.603 1371 Loans 0.555 0.602 0.921 0.009 0.203 1371 GDP growth 0.983 1.500 11.100 -14.800 3.053 1371
Table 2 provides descriptive statistics of the main regression variables. The sample includes 462 banks, from 27 countries in the European Union and the sample period represents the years 2006-2014. All variables are defined in table 2. All balance sheet variables are winsorized at the 1st and 99th percentile. The dependent variables, ROE and ∆ROE, net interest margin and ∆net interest margin are winsorized at the 3rd and 97th percentile.
17 Figure 1. Evolution of capital ratios, yearly. This figure shows the evolution of the three capital ratios investigated in this research; the equity-asset ratio, the tier 1 capital ratio and the total capital ratio.
When analyzing the performance indicator (ROE), the results of the results of Tunay et al. (2015) are used for comparison. Tunay et al. (2015) investigate bank performance of 270 European banks for the years 2005-2014. The average ROE of their research is higher compared to this paper’s sample, namely 8% compared to 4.8%. However, the results of Berger and Bouwman (2013) also show a higher ROE, in non-crisis periods as well as in banking or market crises, although they solely examine U.S. banks in their research. Figure 1 shows a more detailed investigation of ROE and its evolution in the years of the sample period. As shown, ROE decreases rapidly in the years 2008 and 2009, the years in which the financial crisis was unfolding in Europe. Furthermore, it can be observed that average ROE is considerably low in the years after the financial crisis, compared to pre-crisis levels.
Focusing on the control variables, Beltratti and Stulz (2012), who investigate the financial crisis, find similar estimates as this paper’s sample for loans, deposits, income diversity, and liquid assets. Furthermore, the estimates of the control variables are also in line with Aebi et al. (2012), Beltratti and Paladino (2015) and Demirguc‐Kunt et al. (2013). Therefore, it can be assumed that the control variables of the sample are reliable, and do not deviate much from previous studies. 6 8 10 12 14 16 18 2006 2007 2008 2009 2010 2011 2012 2013 2014 Equity-as s et ratio Tier 1 ratio
Totcal capital ratio Year P er ce nt ag e
18 -5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 2006 2007 2008 2009 2010 2011 2012 2013 2014 Me an of GDP growth Year G D P gr ow th in %
Figure 2. Evolution of the mean of Return on Equity. This figure shows the evolution of the return on equity ratio (ROE) over the years 2004-2014. ROE is calculated as net income divided by total assets.
Figure 2. Evolution of the mean of the GDP growth rate. This figure shows the evolution of the real GDP growth rate over the years 2006-2014 in the European Union. Data is collected from Eurostat. The variable is defined in table 2. 0 2 4 6 8 10 12 14 16 2006 2007 2008 2009 2010 2011 2012 2013 2014
Average Return on Equity
19 Table 3: Correlation matrix
ROE ∆ROE% Net interest margin ∆Net interest margin Equity-Asset ratio Tier 1 ratio Total capital ratio Deposits Income diversity Liquid assets
Bank size Loans GDP growth
ROE 1.000
∆ROE% 0.283 1.000
Net interest margin 0.210 0.052 1.000
∆Net interest margin 0.005 0.008 0.041 1.000
Equity-Asset ratio 0.117 0.049 0.537 -0.067 1.000
Tier 1 ratio 0.008 0.013 -0.071 0.070 0.321 1.000
Total capital ratio -0.024 -0.015 -0.110 0.069 0.253 0.884 1.000
20
In order to get more insights into the years of the financial crisis, the average real GDP growth rate of the countries in the sample is analyzed. Figure 2 indicates that major declines in the GDP growth rate took place in the years 2008 and 2009. In addition, 2012 is also characterized by a major decline in the real GDP growth rate. This paper follows most previous research by investigating the years from 2007 to 2009 as the financial crisis years (see. e.g. Demirguc‐Kunt et al.,2013). However, based on the GDP growth rate, the years 2008 and 2009 might cover a more suitable time horizon for examining the financial crisis. In a robustness check, an additional investigation will be made by investigating this alternative time period.
To investigate the possible presence of multicollinearity a correlation matrix is constructed, presented in table 3. According to Belsey et al. (2005) a value equal or larger than 0.7 or – 0.7 can indicate multicollinearity. Table 4 shows that the correlation between the capital ratios is quite high (0.884 between tier 1 capital and total capital). However, these variables are never used in the same regression, but are solely used for comparability purposes. Therefore, they do not have to be excluded from the sample. Loans are also negatively correlated with liquid assets, with a coefficient of -0.760. Unreported robustness checks indicated that excluding these variables from the model did not influence the outcomes.
4. Results
In this section, the main regression results are presented. In order to check whether the results are robust, additional regressions will be performed in which the same variables as used in the main regressions are included, with some exceptions noted below.
4.1 Empirical results from the baseline model
21 Table 4: Regression results for analyzing the effect of capital on profitability, dependent variable ∆ROE
(1) (2) (3) (4) (5) (6)
Capital Ratios
Equity-Asset ratio 1.487 -1.131
Tier 1 ratio -2.802 -4.522**
Total capital ratio -1.968 -3.265
Control Variables Deposits 1.185 1.23 1.207 Income diversity -0.317 -0.31 -0.321 Liquid assets 0.302 0.232 0.226 Bank size -0.44 -0.522** -0.487** Loans -0.414 -0.421 -0.408 GDP growth 0.059** 0.063** 0.062** Adjusted R² 0.005 0.006 0.006 0.013 0.017 0.015 Observations 1371 1371 1371 1371 1371 1371 Durbin-Watson stat 2.475 2.472 2.472 2.496 2.5 2.498
Bank and time fixed effects Yes Yes Yes Yes Yes Yes
This table shows the OLS regression estimates of ∆ROE and the explanatory variables. All variables are defined in table 2. All explanatory variables are winsorized at the 1st and 99th percentile. ∆ROE is winsorized at the 3rd and 97th percentile. ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively.
Column 1 of table 4 suggests that the equity-asset ratio positively influences bank profitability. However, no significant results can be observed. By adding the control variables the coefficient becomes negative (column 4). Thereby, the regression results of the baseline model do not demonstrate a significant effect in the relation between the equity-asset ratio and bank profitability. This contradicts studies of Berger and Bouwman (2013) and Beltratti and Paladino (2015), who observe the equity-asset ratio to be positively influencing bank performance and residual income. Regarding the one-on-one relation with risk-adjusted capital ratios, both the tier 1 and the total capital ratio signs are negative, but not significant. This negative effect is still observable when control variables are included. By including the control variables, the tier 1 capital ratio is significantly, negatively influencing the ∆ROE.
22 Bouwman (2013) also observe bank size to be significantly negatively related to bank performance, especially for small and medium banks. To get a better understanding of the role of size in explaining the relationship between capital and profitability, an additional robustness check will be performed which differentiates between medium and large sized banks, presented in section 4.3.1.
The results of table 4 indicate that for the full sample period, all three capital ratios have a negative effect on bank profitability, however, only the tier 1 ratio has a significant impact. This indicates that only for the tier 1 capital ratio an actual relationship with bank profitability can be observed. An increase in the tier 1 capital ratio of 1% is expected to lower the change in profitability with more than 400%, which is a large effect. Section 2.2 stressed that the sign of the capital coefficient in relation to bank profitability could not be predicted, as theory provides conflicting predictions this connection. Summarizing above findings, it can be stated that the tier 1 capital ratio negatively affects bank profitability.
Differences in the outcomes of tier 1 capital ratio and the total capital ratio might be explained by the tier 2 capital ratio. The total capital ratio is calculated by the sum of the tier 1 and the tier 2 capital divided by risk weighted assets, whereas the tier 1 capital ratio is calculated in a similar way but excluding tier 2 capital from the numerator. As discussed before, the tier 2 capital ratio accounts for additional bank capital, such as revaluation reserves, undisclosed reserves, hybrid instruments and subordinated term debt. Demirguc‐Kunt et al. (2013), who
analyze capital in relation to the stock market performance over the financial crisis, argue that the differences between tier 1 and tier 2 can be explained by the suggestion that investors focus more on the component of capital that is available to absorb losses (the tier 1 ratio), while the bank continues as a going concern. The suggest indicate that the tier 2 capital ratio is not related to bank profitability, thereby weakening the relation between the total capital ratio and profitability.
4.2 Empirical results for examining the financial crisis
23 The results in table 5 contrast earlier results of eq. (1), and reveal that interpretation of the outcomes of table 4 need to be nuanced. By investigating both time periods separately, a significantly positive relationship can be observed between the equity-asset ratio and profitability, during both non-crisis as well as crisis periods. The findings are consistent with Berger and Bouwman (2013) and Beltratti and Paladino (2015), who also observe a significant positive relation between the equity-asset ratio and bank performance. The results indicate that the observed negative connection between capital and profitability of table 4 can be explained by the differences within the time periods. Regarding the equity-asset ratio, the results of column 1 and 4 indicate that hypothesis 1 can be rejected, assuming that capital will negatively affect bank profitability during non-crisis periods. In addition, hypothesis 2 can be accepted, as the equity-asset ratio is positively influencing bank profitability during crisis periods.
During non-crisis periods, risk-adjusted capital ratios negatively influenced the ∆ROE. In contrary to the equity-asset ratio, the relationship between the risk-adjusted capital ratios and profitability changed markedly during the crisis period, as risk-adjusted capital is now positively influencing profitability. The total capital ratio shows significant coefficients, both during non-crisis and crisis periods. This is consistent with the expected bankruptcy costs theory, which explains that capital is particularly beneficial for profitability during the event of a financial crises, in which expected bankruptcy costs increase unexpectedly. These results coincide with findings from Demirguc‐Kunt et al. (2013), who find that an increase in the
risk-based capital ratio increases quarterly stock returns during crisis periods. Outcomes of column 6 confirm hypotheses 1 and 2 for the total capital ratio, which predicted a negative effect of capital on profitability during non-crisis periods and a positive effect during crisis periods.
24 Table 5: Analyzing the effect of capital on profitability during crisis and non-crisis periods.
(1) (2) (3) (4) (5) (6) Capital Ratios Equity-Asset ratio*non-crisis 8.184*** 8.221** Equity-Asset ratio*crisis 13.204*** 9.254* Tier 1 ratio*non-crisis -0.276 -3.060 Tier 1 ratio*crises 3.609 3.042
Total capital ratio*non-crisis -1.701 -4.830**
Total capital ratio*crisis 6.021* 7.522**
Control Variables Deposits 0.762 0.284 0.165 Income diversity -1.268*** -1.213*** -1.212*** Liquid assets -0.052 0.461 0.468 Bank size -0.363 -0.703*** -0.845*** Loans -0.707 -0.415 -0.729 GDP growth 0.021 0.027 0.028 Adjusted R² 0.072 0.065 0.068 0.084 0.082 0.090 Observations8 1346 1346 1346 1346 1346 1346 Durbin-Watson stat 2.159 2.144 2.142 2.174 2.169 2.173
Bank and time fixed effects Yes Yes Yes Yes Yes Yes
This table shows the OLS regression estimates of ∆ROE and the explanatory variables. The effect of pre-crisis capital on the performance of banks during a crisis is examined. All variables are defined in table 2. All explanatory variables are winsorized at the 1st and 99th percentile. ∆ROE is winsorized at the 3rd and 97th percentile. ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively.
As risk-adjusted capital enhances profitability only during crisis periods, it can be stated that risk-adjusted capital is a main line of defense against negative shocks. The main goal of the Basel III capital requirements is to provide a more resilient banking sector and prevent future financial system breakdowns. The results imply that higher capital is indeed advantageous for the profitability of banks surrounding a crisis. As bank profitability is of great importance in assuring the stability of the economy, the new capital requirements will enhance the stability of the economy. Whereby, banks are expected to be less affected by future economic shocks. However, during non-crisis periods, higher prior risk-adjusted capital deteriorates subsequent bank performance. Hence, this might explain why bankers and investors oppose to regulations that aim at increasing the capital requirements.
8
25 Furthermore, different capital measures reveal different patterns for bank profitability. Hence, authorities should realize that using only one single capital ratio in assessing bank quality may result in wrong regulatory policies. Therefore, the results promote the appropriateness for the use of multiple variables in capital regulation. Only Investigating the equity-asset ratio, may not explain the full relationship between capital and profitability. This suggests that the current emphasis of Basel III on strengthening multiple capital requirements is broadly appropriate.
4.3 Robustness Checks
This section provides a number of robustness checks. The findings are qualitatively similar to the main results. The same dependent and independent variables as used in the main regressions are included, with some exceptions noted below.
4.3.1 Influence of Bank size
In line with the research of Berger and Bouwman (2013) and Demirguc‐Kunt et al. (2013), an
additional analysis of bank capital and profitability is performed, by creating subsamples for medium and large sized banks. Berger and Bouwman (2013) investigate small, medium, and large banks. This research, however, only discriminates between medium and large banks, as the entire sample focusses only on relatively large banks (total assets exceeding $10 billion). Following Demirguc‐Kunt et al. (2013), large banks are defined as banks with assets in excess
of $50 billion.9 The medium sample includes a total of 732 banks and the large sample considers a total of 614 banks. Demirguc‐Kunt et al. (2013) motivate their investigation of
large banks by the assumption that these are generally more sophisticated institutions that operate on a global scale with complex balance sheets. Furthermore, these banks may have more opaque assets and may be better able to skirt capital regulation through regulatory arbitrage. The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 defines systemically important banks as banks with total assets exceeding $50 billion.10 Next to that, so called system-banks are essential to the stability of the financial system as a whole, making them especially important to examine. By analyzing medium and large banks separately, an examination is made on whether estimates for medium and large banks influence the baseline results. Previous research has predominantly observed bank size to be positively related to
9 An unreported robustness examines the influence of different cut-off points, based on the median of total assets
($44.5 billion). This alternative cut-off point shows similar results. These results are available upon request.
10 The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 is accessable via
26 profitability. However, the results so far have not confirmed this effect. Table 6 contains the results of analyzing eq. (1) and (2) for medium and large bank subsamples. Columns 1-6 consider medium sized banks, whereas the results of columns 7-12 apply to large banks.
Three main results are found. Firstly, a higher equity-asset ratio helps medium sized banks to improve their profitability at all times (i.e. both during non-crisis as well as during crisis periods), which can be observed in columns 1-6. This applies to all the three capital ratios investigated, by which the equity-asset ratio is always significant and the total equity ratio is marginally significant during crisis years. This is in line with Berger and Bouwman (2013), who observe market shares of small banks to be positively influenced by a higher equity-asset ratio, at all times. Secondly, higher risk-adjusted capital deteriorates large bank profitability during a non-crisis period, but improves profitability during a crisis period. However, the significance is marginal. Thirdly, results confirm the results presented in table 5, as the equity-asset ratio is positively related to bank profitability, during both non-crisis and crisis periods, and for medium as well as large banks.
According to Berger and Bouwman (2013), relatively small banks face shocks more often and have limited and relatively costly access to the financial market in the event of unanticipated needs, compared to large banks. As a result, higher capital is beneficial for relatively small banks during non-crisis as well as crisis periods. For large banks equity is less crucial as they can rely on access to the financial and interbank market, in addition to their on-balance-sheet capital. Allowing them to absorb unexpected negative shocks. Therefore, during non-crisis periods, their profitability is negatively influenced by risk-adjusted capital. However, crisis periods enhance stress for all banks by which even large banks might not have sufficient access to financial markets and interbank lending. This might explain their profitability being positively related to bank capital during crisis periods.
27 Table 6 Regression results for medium and large banks, dependent variable ∆ROE%
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Capital Ratios Equity-Asset ratio*non-crisis 9.221** 10.209** 8.601 10.456* Equity-Asset ratio*crisis 9.086* 10.949* 17.780** 11.920 Tier 1 ratio*non-crisis 3.641 3.542 -2.954 -9.162* Tier 1 ratio*crises 3.498 3.868 15.441* 0.007
Total capital ratio*non-crisis 1.504 1.195 -4.419 -10.249
Total capital ratio*crisis 4.524 5.948* 24.682** 12.206
Control Variables Deposits 0.136 -0.130 -0.236 1.527 0.852 0.955 Income diversity -0.605 -0.481 -0.551 -1.247** -1.130** -1.150** Liquid assets 1.121 1.655* 1.674* -1.640 -1.007 -0.777 Bank size 0.211 0.001 -0.160 -1.652*** -2.166*** -2.206*** Loans 1.080 1.503 1.243 -5.675*** -4.790*** -4.460*** GDP growth 0.041 0.040 0.042 0.002 0.030 0.029 Adjusted R² 0.102 0.097 0.096 0.103 0.097 0.099 0.089 0.088 0.096 0.134 0.136 0.147 Observations 732 732 732 732 732 732 614 614 614 614 614 614 Durbin-Watson stat 2.176 2.170 2.163 2.175 2.176 2.167 2.201 2.171 2.178 2.264 2.264 2.274
Bank and time fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
28
4.3.2 Measure influence of excess capital on profitability
The data demonstrate that banks hold far more capital than is required by regulators, indicating that banks hold a significant amount of their capital as a capital buffer (see Figure 1 section 3.3). This raises the question on how the level of excess capital influences profitability. Table 7 presents the results of an additional regression examining eq. (1) and (2), but with bank capital buffers as the main explanatory variable of interest. Capital buffers are measured in absolute terms, i.e. the difference between the banks’ i capital in year t and the minimum official capital requirement set by regulators.11 Capital buffers are lagged averages over two years before year t, thereby examining whether prior capital buffers influence subsequent bank profitability. The minimum tier 1 capital ratio requirement is 4.0% for the years 2006-2012, 4.5% for the year 2013, and 5.0% as of 2014. Regarding the total capital ratio, the minimum is 8.0% for the whole sample period.
The results reported in table 7 are consistent with previous sections, investigating the capital ratios, not the capital buffers. Columns 1 and 2 provide results when eq. (1) is examined. The outcome indicates that capital buffers are negatively related to bank profitability, by which solely the tier 1 capital buffer is significant on a 5% level. This is in line with the earlier discussed results of table 4, in which only a negative significant effect could be observed for the tier 1 capital ratio. Columns 3 and 4 provide results when eq. (2) is considered. Again, differentiating between non-crisis and crisis periods reveals that analyzing the entire sample period needs some nuance, as risk-adjusted capital ratios during both periods are characterized by a different sign when explaining bank profitability. The total capital ratio is significantly negatively influencing bank profitability during non-crisis periods, and significantly positively during crisis periods, which is in line with the results of table 5.
To summarize, when examining the full sample period at once the empirical results reveal that increasing the tier 1 ratio will deteriorate banks’ profitability. However, discriminating between non-crisis and crisis periods indicates that during crisis periods bank profitability increases by a higher total capital ratio, whereas, the opposite can be observed for the non-crisis periods. This implies that a larger capital buffer is beneficial in case of unexpected negative shocks, confirming hypothesis 2 of a positive relationship of capital and profitability during crisis periods. In addition, these findings are also in line with the expected bankruptcy
11 Capital buffers are only examined for the tier 1 and total capital ratio, not for the equity-asset ratio. Basel III
29 costs hypothesis discussed in section 2 and confirm the outcomes of Berger and Bouwman (2013), who observe that a stronger capital position is positively related to bank performance. However, during non-crisis periods a larger capital buffer deteriorates profitability, which confirms hypothesis 1.
Table 7 Regression results with capital buffer as main explanatory variable
(1) (2) (3) (4)
Capital buffer
Tier 1 buffer -5.485**
Total capital buffer -3.265
Tier 1 buffer*non-crisis -0.027
Tier 1 buffer*crisis 0.032
Total capital buffer*non-crisis -4.830**
Total capital buffer*crisis 7.522**
Control Variables Deposits 1.245 1.207* 0.324 0.165 Income diversity -0.300 -0.321 -1.207*** -1.212*** Liquid assets 0.266 0.226 0.474 0.468 Bank size -0.535** -0.487** -0.681*** -0.845*** Loans -0.375 -0.408 -0.359 -0.729 GDP growth 0.065*** 0.062** 0.028*** 0.028*** Adjusted R² 0.016 0.015 0.079 0.090 Observations 1371 1371 1346 1346 Durbin-Watson stat 2.505 2.498 2.174 2.173
Bank and time fixed effects Yes Yes Yes Yes
This table shows the OLS regression estimates of capital buffers and the explanatory variables. All variables are defined in table 2. All explanatory variables are winsorized at the 1st and 99th percentile. The dependent variable, ∆ROE, is winsorized at the 3rd and 97th percentile. ***, **, * denote statistical significance at the 1%, 5%, and
30
4.3.3 Use an alternative cut-off point for the financial crisis period
An alternative cut-off point for the pre-crisis and crisis years might be more appropriate, as suggested by the investigation of the real GDP growth rate (see figure 2). In this section eq. (2) will be analyzed, whereby the crisis period is adjusted to the years 2008 and 2009. Results are reported in table 8.
Table 8 Regression results, with an alternative crisis period
(1) (2) (3) (4) (5) (6) Capital Ratios Equity-Asset ratio*non-crisis 5.226** 5.957** Equity-Asset ratio*crisis 7.556** 5.446 Tier 1 ratio*non-crisis -1.783 -3.760* Tier 1 ratio*crises 2.973 2.928
Total capital ratio*non-crisis -2.470 -4.361**
Total capital ratio*crisis 5.507* 5.607*
Control Variables Deposits 0.537 0.295 0.198 Income diversity -1.549*** -1.464*** -1.416*** Liquid assets -0.009 0.327 0.373 Bank size -0.558 -0.797*** -0.844*** Loans -0.532 -0.396 -0.537 GDP growth 0.009 0.009 0.010 Adjusted R² 0.074 0.071 0.075 0.095 0.096 0.100 Observations12 1363 1363 1363 1363 1363 1363 Durbin-Watson stat 2.115 2.103 2.106 2.146 2.141 2.145
Bank and time fixed effects Yes Yes Yes Yes Yes Yes
This table shows the effect of pre-crisis capital on the performance of banks during a crisis. 2008 and 2009 are defined as crisis years. All variables are defined in table 2. ∆ROE is winsorized at the 3rd and 97th percentile. ***, **, * denote statistical significance at the 1%, 5%, and 10% level, respectively.
In order to indicate whether the adjusted crisis period is a more suitable period when investigating the recent financial crisis, table 5 and table 8 have to be compared. When examining 2008 and 2009 as the crisis period, similar patterns can be observed. Again, the equity-asset ratio positively influences bank profitability at all times. Although with less significance, the risk-adjusted capital ratios are negatively influencing profitability during
12
31 non-crisis periods and positively during crisis periods. The control variables perform in a similar way, as both income diversity and bank size are significantly negatively influencing ∆ROE.
It can be concluded that by analyzing the years 2008 and 2009 as the crisis period leads to similar results as the 2007 to 2009 crisis period. Hereby, the results are robust to this alternative crisis period.
4.3.4. Use an alternative measure for profitability
As an alternative measure for profitability, the ∆net interest margin (∆NIM) is examined. Net interest margin is defined as net interest income divided by total assets, where net interest income is calculated as the interest received minus interest paid. Net interest income drives a wedge between interest rates received by savers on deposits and interest paid by on loans by lenders. The net interest margin has been used in many studies of bank performance (see, e.g., Dietrich and Wanzenried (2011) and Lee and Hsieh (2013)). As for ∆ROE, ∆NIM is winsorized with 3% at both tails.
The results are presented in table 9. First, the one-on-one relation between the capital ratios and bank NIM is examined displayed in columns 1-3. Next, eq. (1) is examined by including bank-level and country-level controls to the regression, displayed in columns 4-6. Columns 7-9 consider examination of eq. (2). The results contrast with earlier findings, as table 7-9 reveals that in the baseline model, all capital ratios except the equity-asset ratio, are positively related to bank profitability, measured as the ∆net interest margin, although, without any significance. Regarding the examination of eq. (2), both the tier 1 and total capital ratio are positively influencing bank profitability during all examined periods. Hereby, the tier 1 capital ratio is marginally significant in the non-crisis period and the total capital ratio is significant on a 1% level during the crisis period. Because the results contradict earlier findings, the sample is not robust for having ∆net interest margin as an alternative measure for bank profitability.
32
Table 9 Regression results using ∆net interest margin as dependent variable
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Capital Ratios
Equity-Asset ratio 0.083 -0.510
Tier 1 ratio 0.377 0.128
Total capital ratio 0.379 0.204
Equity-Asset ratio*non-crisis 0.080
Equity-Asset ratio*crisis -1.733**
Tier 1 ratio*non-crisis 0.588*
Tier 1 ratio*crises 0.495
Total capital ratio*non-crisis 0.470
Total capital ratio*crisis 1.248***
Control Variables Deposits 0.025 0.039 0.040 -0.136 -0.122 -0.135*** Income diversity 0.197*** 0.192*** 0.193*** 0.092 0.121 0.122 Liquid assets -0.245 -0.264 -0.262 0.185*** 0.179 0.196* Bank size -0.073 -0.057 -0.056 -0.102*** -0.055 -0.068** Loans -0.345*** -0.371*** -0.373*** -0.135 -0.119 -0.163 GDP growth 0.011*** 0.010*** 0.010*** 0.008** 0.006* 0.006* Adjusted R² 0.003 0.005 0.005 0.020 0.019 0.019 0.247 0.240 0.243 Observations 1371 1371 1371 1371 1371 1371 1346 1346 1346 Durbin-Watson stat -0.020 -0.018 -0.018 2.240 2.247 2.248 1.509 1.497 1.501
Bank and time fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes
33 5. Conclusion
This paper empirically addresses the impact of capital on bank profitability, by performing a panel data analysis on 462 banks located in the European Union over the time period from 2006 to 2014. To account for the impact of the recent financial crisis, years before and during the crisis are examined separately. Three types of capital ratios are analyzed; the equity-asset ratio, the tier 1 ratio, and the total capital ratio. By examining all three capital ratios, a comparison can be made in explaining bank profitability between the equity-asset ratio, the tier 1 ratio and the total capital ratio
The results reveal that differences in initial capital, whether risk-adjusted or not, did not consistently affect subsequent bank profitability. Furthermore, the connection between capital and profitability varies over economic cycles. Therefore, analyzing non-crisis and crisis periods separately and applying multiple capital measurements gives valuable insights on how capital interacts with profitability.
The main results are as follows. Firstly, the equity-asset ratio always positively influences bank profitability, measured as ∆ROE, both during non-crisis as well as crisis periods. Secondly, the total capital ratio, which comprises risk-adjusted capital, enhances profitability during crisis periods but deteriorates profitability during non-crisis periods. Lastly, differentiating between medium sized and large banks reveals that higher capital, regardless of how it is measured, helps medium sized banks during both non-crisis and crisis periods. In addition, higher risk-adjusted capital deteriorates large bank profitability during a non-crisis period, but improves profitability during a crisis period, however with marginal significance.
34 Nonetheless, it should be noted that during non-crisis periods, profitability is negatively influenced by prior risk-adjusted capital, especially for large banks. This might explain why bankers and investors oppose to regulations that aim at increasing bank capital.
35 5. References
Aebi, V., Sabato, G., & Schmid, M., 2012. Risk management, corporate governance, and bank performance in the financial crisis. Journal of Banking & Finance, 36, 3213-3226.
Ahmad, R., Ariff, M., & Skully, M. J., 2008. The determinants of bank capital ratios in a developing economy. Asia-Pacific financial markets, 15, 255-272.
Altunbas, Y., Carbo, S., Gardener, E. P., & Molyneux, P., 2007. Examining the relationships between capital, risk and efficiency in European banking. European Financial Management, 13, 49-70.
Ayuso, J., Pérez, D., & Saurina, J., 2004. Are capital buffers pro-cyclical?: Evidence from Spanish panel data. Journal of financial intermediation, 13, 249-264.
Basel Committee on Banking Supervision., 2010. Basel III: A global regulatory framework for more resilient banks and banking systems. Basel Committee on Banking Supervision, Basel.
Belsley, D. A., Kuh, E., & Welsch, R. E., 2005. Regression diagnostics: Identifying influential data and sources of collinearity. John Wiley & Sons, Hoboken, New Jersey. Bhattacharya, S., & Thakor, A. V., 1993. Contemporary banking theory. Journal of financial
Intermediation, 3, 2-50.
Beltratti, A., & Stulz, R. M., 2012. The credit crisis around the globe: Why did some banks perform better? Journal of Financial Economics, 105, 1-17.
Beltratti, A., & Paladino, G., 2015. Bank leverage and profitability: evidence from a sample of international banks. Review of Financial Economics, 27, 46-57.
Berger, A. N., 1995. The relationship between capital and earnings in banking. Journal of Money, Credit and Banking, 432-456.
Berger, A. N., & Di Patti, E. B., 2006. Capital structure and firm performance: A new approach to testing agency theory and an application to the banking industry. Journal of Banking & Finance, 30, 1065-1102.
Berger, A. N., & Bouwman, C. H., 2009. Bank liquidity creation. Review of Financial Studies, 22, 3779-3837.
Berger, A. N., & Bouwman, C. H., 2013. How does capital affect bank performance during financial crises? Journal of Financial Economics, 109, 146-176.
Berger, A. N., Hanweck, G. A., & Humphrey, D. B., 1987. Competitive viability in banking: Scale, scope, and product mix economies. Journal of monetary economics, 20, 501-520.
Buser, S. A., Chen, A. H., & Kane, E. J., 1981. Federal deposit insurance, regulatory policy, and optimal bank capital*. The Journal of Finance, 36, 51-60.
