Microsoft Word - 14157-new Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 272 ajfa.macrothink.org Determinants of Systemic Risk of Banks in India Mihir Dash School of Business, Alliance University Received: Jan. 2, 2019 Accepted: June 5, 2019 Published: June 5, 2019 doi:10.5296/ajfa.v11i1.14157 URL: https://doi.org/10.5296/ajfa.v11i1.14157 Abstract This study examines the determinants of systemic risk for banks in India. The independent variables considered for the study include the sector, bank size, return on assets, beta, leverage, capital adequacy, non-performing assets, price to book value, deposits, loans & advances, investments, net interest income, and non-interest income. A mixed panel regression model was applied, with bank fixed effects and year random effects. The results of the study indicate that public sector banks have a much higher level of systemic impact than private sector banks. Further, the determinants of systemic impact are different for public sector and private sector banks. The systemic impact of public sector banks was positively related with size and negatively related with price to book value ratio and investments to total assets ratio, while the systemic impact of private sector banks was negatively related with return on assets and positively related with beta and net interest income to total funds ratio. Keywords: systemic risk, determinants, public sector banks, private sector banks. Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 273 ajfa.macrothink.org Introduction Systemic risk represents the impact that the failure of a bank or financial institution would have on the entire financial system and/or economy, through its network of interlinked financial intermediaries. The failure of an institution leads to financial stress on institutions that have lent money to it, which in turn may lead to failure of some of these institutions. this leads to a kind of domino or ripple effect, and spreads across the entire financial system. The recent experience of the global financial crisis of 2008-09 and the subsequent Euro-zone crises of 2010-11 has demonstrated the importance of measuring the level of systemic risk associated with different financial institutions and understanding the factors contributing to systemic risk. The collapse of some of the most prominent banks in the world, including the Lehman Brothers and Washington Mutual Bank, along with several near-failures which had to be bailed out of crisis by the U.S. Government, highlighted the significance of understanding, measuring, and monitoring systemic risk. Several economists have suggested that undercapitalisation of large financial institutions can result in financial instability, particularly when the entire financial system is undercapitalised. This is leads to the concept of “too big to fail” (TBTF), i.e. that large financial institutions are so systemically important that they cannot be allowed to fail. A similar concept is that of “too interconnected to fail” (TICTF), i.e. that financial institutions that are highly inter-connected with other institutions are very systemically important and so cannot be allowed to fail. A question that several authors have posed is: which financial institutions should be bailed out in the event of a solvency/liquidity crisis (e.g. Acharya et al, 2012)? This logically requires identifying which institutions are critical to stability of the financial system, i.e. “systemically important.” According to the Basel Committee on Banking Supervision (BCBS), the concept of systemic importance should be measured in terms of the potential impact of the failure of a bank on the global financial system and wider economy, rather than just the risk that a failure can occur (Moore and Zhou, 2014). There are many theories suggesting that large and complex banks contribute to systemic risk. A possible root for the systemic importance of large, inter-connected banks is moral hazard; as regulators are reluctant to close or unwind large and complex banks, this leads banks to take on excessive risks in the expectation of government bailouts (e.g., Farhi and Tirole, 2012). Another possibility is that of agency effects, i.e. that poor governance of large and complex banks can lead to bank managers engaging in non-traditional risky activities (for example, trading) and tend to be financed more through short-term debt, making them more vulnerable to liquidity shocks and market failures (e.g. Laeven and Levine, 2007; Boot and Ratnovski, 2012). The Indian banking system, which was initially hailed to be unaffected by the crises, was affected indirectly, mainly on account of growing trade and financial integration with the global economy. Though Indian banks were not pushed to the point of insolvency, monitoring of systemic risk has become important in the dynamic banking environment in India in order to avoid potential system failure. This study examines the determinants of systemic risk for Indian Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 274 ajfa.macrothink.org banks. The Indian banking industry has two important segments, public sector banks and private sector banks. Public sector banks are owned and controlled by the government, and are subjected to political interference and constraints. Many studies have argued that private sector banks outperform public sector banks due to professional, efficient management, and better customer focus and service, particularly in terms of Management Soundness and Earnings and Profitability (Dash and Das, 2013; Dash et al, 2015). In view of this, the determinants of systemic risk would be expected to differ between public sector and private sector banks. Literature Review Measurement of Systemic Risk There are many definitions of systemic risk and systemic importance advocated in the literature, and many more approaches proposed for their measurement. Adrian and Brunnermeier (2008) was one of the first authors to suggest a measure for systemic risk, viz. the conditional value-at-risk (CoVaR), which focuses on the tail distribiution. They were able to identify the contribution of each bank to systemic risk using this measure. Acharya et al (2010a, 2010b) proposed the concept of systemic expected shortfall (SES), i.e. the amount by which a bank is undercapitalised in a systemic event in which the entire financial system is undercapitalised, to measure systemic risk. Acharya and Steffan (2012) extended the framework by introducing the concepts of marginal expected shortfall (MES), which measures the performance of a bank when the market return as a whole experiences its worst 5% trading days within a year, and the bank’s market leverage ratio (LVG), the market value of assets divided by the market value of equity. Brownlees and Engle (2012, 2017) and Acharya et al (2012) suggested the SRISK index, which estimates the expected capital shortage of a bank during on a substantial market meltdown, as a measure for systemic risk. Hautsch et al (2013, 2015) used a parsimonius econometric approach to measure systemic risk, the realised systemic risk beta, viz. the total effect of a bank’s VaR on the VaR of the entire financial system, taking into account the bank’s network relationships. Suh et al (2013) proposed a method for estimating systemic risk using credit default swaps. Their method had the added advantage of being able to measure systemic risk contributions in both directions, i.e. the overall effect of systemic risk on individual credit risks and vice versa. Karimalis and Nomikos (2014) proposed a methodology for estimating the CoVaR, i.e. the Value-at-Risk of the financial system conditional on the failure of a financial institution based on copula functions, and extended this approach to estimate other conditional risk measures such as Conditional Expected Shortfall (CoES). Moore and Zhou (2014) proposed the expected system loss (ESL), viz. the expected loss to the financial system as a whole given that a particular bank fails, which they estimated using multivariate extreme value theory, as a measure of systemic importance of the bank. Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 275 ajfa.macrothink.org Hattori et al (2014) pointed out that systemic risk measures are essentially a form of scenario analysis, as they analyse the impact of certain types of assumed trigger events on the financial system, based on past patterns of failure; however, this may not be an indicator for robustness against future, unprecedented modes of failure. Also, they argued that most market-based estimates of systemic risk may overestimate the importance of short-term changes. They suggested combining different systemic risk measures together with macro-stress testing scenarios, providing a wider range of potential sources of failure. van Oordt and Zhou (2015) analysed bank systemic risk into two dimensions, the level of bank tail risk and the linkage between the level of bank tail risk and severe financial shocks to the system. Determinants of Systemic Risk Several studies have analysed the determinants of systemic risk and systemic importance of banks. Stolbov (2012) examined macro-determinants of systemic risk for some major economies. He found that gross government debt to GDP, state fragility index, EU membership, and world gross GDP share are key determinants of systemic risk for the sovereign CDS prices, while stock market total value traded to GDP, state fragility index, and financial openness index are the key determinants of systemic risk in the stock market. Moore and Zhou (2014) found that size and non-traditional banking activities were the significant determinants of systemic importance of US banks in the period 2000-10; in particular, they found that banks above a certain size have equal systemic importance. Bostandzic et al (2014) found that banks with higher levels of Tier 1 capital had lower exposure and contribution to global systemic risk. Further, they found that bank size and interconnectedness are positively related to global financial fragility. They also found that deposit insurance schemes that require banks and depositors to bear more financial risk are associated with greater vulnerability and contribution to a crisis of the financial sector. van Oordt and Zhou (2015) found that banks with higher non-performing loan ratios and lower profitability ratios tended to have higher tail risk, while larger banks, with higher trading revenue, and higher non-interest income tend to have higher systemic risk. Laeven et al (2016) found that systemic risk increases with bank size and is inversely related with bank capital; in particular, low capital in large banks is the key driver of systemic risk. Further, they found that market-based activities and country characteristics have moderating effect on these relationships. Anghelache and Oanea (2016) found that financial leverage, size, risk, and market to book value had a significant impact on systemic risk contribution of Romanian commercial banks. Methodology The objective of the study is to analyse the determinants of systemic risk for banks in India. Due to the wide differences in performance between public sector and private sector banks, the Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 276 ajfa.macrothink.org determinants of systemic risk would be expected to differ between public sector and private sector banks. The study was conducted using sample of thirty-one Indian banks, including twenty-one public sector banks, and ten private sector banks. The list of sample banks is given in the table below. Public sector banks Private sector banks Allahabad Bank Axis Bank Ltd Andhra Bank Federal Bank Ltd Bank of Baroda HDFC Bank Ltd Bank of India ICICI Bank Ltd Bank of Maharashtra IndusInd Bank Ltd Canara Bank Jammu & Kashmir Bank Ltd Central Bank of India Karnataka Bank Ltd Corportaion Bank Karur Vysya Bank Ltd Dena Bank Kotak Mahindra Bank Ltd IDBI Bank Ltd Yes Bank Ltd Indian Overseas Bank Punjab & Sind Bank Punjab National Bank State Bank of Bikaner & Jaipur State Bank of India State Bank of Mysore State Bank of Travancore Syndicate Bank United Commercial Bank Union Bank of India Vijaya Bank The data pertaining to bank characteristics was collected from the Capitaline database1. The SRISK estimates were collected from NYU Stern’s V-Lab database2. The study period was 2007-16. The dependent variable considered for the study is the measure of systemic risk proposed by Brownlees and Engle (2012), SRISK. This index measures the expected capital shortage faced by a bank during a period of system distress when the market declines substantially. It is estimated as 𝑆𝑅𝐼𝑆𝐾 , = 𝑘𝐷 , − (1 − 𝑘)𝑊 , (1 − 𝐿𝑅𝑀𝐸𝑆 , | 𝐶 | ), where k is the minimum fraction of capital (as a ratio of total assets) each bank needs to hold, Di,t and Wi,t are the book value of its debt (total liabilities) and the market value of its equity, 1 www.Capitaline.com 2 https://vlab.stern.nyu.edu/analysis/RISK.WORLDFIN-MR.GMES Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 277 ajfa.macrothink.org respectively, and the long-run marginal expected shortfall LRMES is defined as the tail expectation of the firm’s equity return conditional on a market decline 𝐿𝑅𝑀𝐸𝑆 , | = −𝐸 𝑅 , | 𝑅 , | < 𝐶 . Note that SRISK can take negative values. A bank with negative SRISK represents a well- capitalised bank with large enough capital buffers to easily absorb systemic shocks. The total systemic risk in the financial system is measured by aggregating the positive SRISK contributions of different financial institutions. The independent variables considered for the study ae discussed in the following. The most common determinant for systemic risk is that of bank size, and the commonly-used proxy for size is the logarithm of the bank’s total assets (see for example, Laeven et al, 2014). The systemic risk of a bank would be expected to increase with bank size. This reflects the “too big to fail” hypothesis, that the failure of a large bank would have too a great impact on the entire financial system, so that government should intervene to prevent such a failure. Another common determinant is capital adequacy (Laeven et al, 2014). The measure for capital adequacy used for the study is the Capital Adequacy Ratio. It is expected that higher levels of capital adequacy would be associated with a lower systemic impact. Non-performing loans is an important determinant (van Oordt and Zhou, 2015), and would be expected to play a role in increasing systemic risk particularly for public sector banks. The measure considered in the study is the Net Non-Performing Loans to Net Advances. Two other important determinants are beta and leverage (Anghelache and Oanea, 2016). These have also been included in the present study. Both would be expected to be positively related with systemic impact. Bank profitability may also be related with systemic impact. In the present study, it is measured by the return on assets of the bank. Non-interest income has been found to be a significant determinant of systemic impact in several studies (Moore and Zhou, 2013; van Oordt and Zhou, 2015), positively related with systemic impact. This was measured in the present study using the Non-Interest Income to Total Funds ratio. Along with this, the Net Interest Income to Total Funds ratio is also considered. Laeven et al (2014) have also considered deposits to total assets and loans & advances to total assets in their analysis. These have also been included in the present study, along with investments to total assets. Bostandzic et al (2014) have also considered the valuation ratios as potential determinants of systemic impact. The price to book value ratio has been considered in the present study. The study used a mixed panel regression model for explaining systemic risk, formulated as follows: Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 278 ajfa.macrothink.org 𝑆𝑅𝐼𝑆𝐾 , = 𝑎 + ∑ 𝑏 (1 + 𝑆)𝑥 , + ∑ 𝑐 𝐷 + ∑ 𝑑 𝐷 + 𝜖 , , where the dependent variable on the LHS is the SRISK of the ith bank at time point t, xi,t are the independent variables for the ith bank at time point t, S represents a dummy variable for public sector banks (S = 1) against private sector banks (S = 0), the Di represent the individual bank dummies, in order to capture the bank fixed effect, and the Dt represent the year dummies, in order to capture the year random effect. Findings The descriptive statistics for the variables are presented in Table 1 below. Table 1. descriptive statistics of SRISK and its determinants private sector public sector Mean St. Dev. Min Max Mean St. Dev. Min Max SRISK ($ m) -2841.41 5079.93 -25319 3100 1940.70 2120.98 -122 14521 ln(Total Assets) 13.63 1.11 11.62 15.80 14.32 0.86 12.50 16.93 Return on Assets 1.39 0.40 0.34 2.02 0.72 0.51 -1.25 2.50 Beta 0.84 0.27 0.24 1.57 0.80 0.22 0.22 1.41 Leverage 8.50 6.05 1.89 27.68 29.31 15.92 7.83 103.85 Capital Adequacy Ratio 14.78 2.33 11.03 22.46 11.92 1.05 9.44 15.00 Net Non-Performing Assets to Net Advances 0.83 0.81 0.00 4.31 1.99 1.77 0.15 11.89 Price to Book Value Ratio 2.47 1.79 0.46 9.58 0.87 0.42 0.26 2.70 Deposits to Total Assets 0.76 0.11 0.52 0.90 0.84 0.05 0.42 0.91 Loans & Advances to Total Assets 0.58 0.04 0.47 0.68 0.62 0.03 0.51 0.70 Investments to Total Assets 0.30 0.04 0.20 0.43 0.26 0.03 0.16 0.34 Net Interest Income to Total Funds 3.15 0.86 1.07 5.62 2.46 0.52 0.59 3.66 Non-Interest Income to Total Funds 1.61 0.52 0.52 2.63 0.97 0.27 0.45 1.83 The private sector banks had a negative average SRISK and a negatively-skewed distribution of SRISK, while the public sector banks had a positive average SRISK and a positively-skewed distribution of SRISK. Private sector banks also had higher return on assets, capital adequacy, price to book value ratios, net interest income to total funds, and non-interest income to total funds than public sector banks, while public sector banks had higher leverage and net non- performing assets to net advances than private sector banks. There was not much of a difference between public and private sector banks in terms of size, beta, deposits to total assets, loans & advances to total assets, and investments to total assets. The results of the panel regression are presented in Tables 2 and 3 below. Table 2 presents the summary of statistical tests for groups and covariates, while Table 3 presents the parameter estimates and significance. Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 279 ajfa.macrothink.org Table 2. Tests of Between-Subjects Effects Dependent Variable: SRISK ($ m) Model I Model II Source F Stat p-value F Stat p-value Intercept 7.078 0.008 16.165 0.000 bank 8.586 0.000 15.359 0.000 year 4.310 0.000 5.710 0.000 ln(Total Assets) 9.330 0.003 11.150 0.001 sector * ln(Total Assets) 28.447 0.000 78.262 0.000 Return on Assets 14.857 0.000 25.459 0.000 sector * Return on Assets 13.317 0.000 15.552 0.000 Beta 34.197 0.000 46.760 0.000 sector * Beta 21.806 0.000 33.223 0.000 Leverage 1.954 0.163 1.941 0.165 sector * Leverage 4.057 0.045 4.391 0.037 Capital Adequacy Ratio 0.591 0.443 sector * Capital Adequacy Ratio 0.820 0.366 Net Non-Performing Assets to Net Advances 0.000 0.993 sector * Net Non-Performing Assets to Net Advances 1.084 0.299 Price to Book Value Ratio 14.025 0.000 17.062 0.000 sector * Price to Book Value Ratio 11.546 0.001 15.256 0.000 Deposits to Total Assets 0.892 0.346 2.323 0.129 sector * Deposits to Total Assets 5.737 0.017 5.098 0.025 Loans & Advances to Total Assets 3.574 0.060 sector * Loans & Advances to Total Assets 1.233 0.268 Investments to Total Assets 3.091 0.080 sector * Investments to Total Assets 0.082 0.775 Net Interest Income to Total Funds 8.432 0.004 6.576 0.011 sector * Net Interest Income to Total Funds 9.893 0.002 6.955 0.009 Non-Interest Income to Total Funds 0.913 0.340 sector * Non-Interest Income to Total Funds 0.978 0.324 Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 280 ajfa.macrothink.org Table 3. Parameter Estimates Dependent Variable: SRISK ($ m) Parameter Coeff t Stat p-value Coeff t Stat p-value Intercept 4397.226 0.257 0.797 -10578.643 -0.852 0.395 Allahabad Bank -76113.340 -4.481 0.000 -70040.480 -6.933 0.000 Andhra Bank -74703.558 -4.447 0.000 -68478.262 -6.895 0.000 Axis Bank Ltd -3637.858 -2.230 0.027 -3469.793 -2.294 0.023 Bank of Baroda -79996.195 -4.608 0.000 -73454.769 -6.868 0.000 Bank of India -78701.873 -4.537 0.000 -72351.973 -6.820 0.000 Bank of Maharashtra -73313.242 -4.409 0.000 -67396.971 -6.906 0.000 Canara Bank -78607.887 -4.521 0.000 -72674.861 -6.863 0.000 Central Bank of India -76482.713 -4.487 0.000 -70938.551 -6.917 0.000 Corporation Bank -75233.326 -4.459 0.000 -69225.172 -6.928 0.000 Dena Bank -73311.008 -4.425 0.000 -66986.663 -6.895 0.000 Federal Bank Ltd 1974.807 1.447 0.149 1783.790 1.481 0.140 HDFC Bank Ltd -14842.406 -7.445 0.000 -13858.932 -7.273 0.000 ICICI Bank Ltd -9110.329 -3.916 0.000 -8845.393 -4.288 0.000 IDBI Bank Ltd -76973.519 -4.515 0.000 -71788.620 -7.102 0.000 Indian Overseas Bank -76908.534 -4.523 0.000 -70738.764 -6.956 0.000 IndusInd Bank Ltd -456.462 -0.469 0.639 -31.621 -0.037 0.971 Jammu & Kashmir Bank Ltd/The 4129.962 2.244 0.026 5258.616 3.245 0.001 Karnataka Bank Ltd/The 3002.203 2.006 0.046 3046.984 2.278 0.024 Karur Vysya Bank Ltd/The 6283.394 3.750 0.000 5914.837 4.157 0.000 Kotak Mahindra Bank Ltd -6245.489 -4.194 0.000 -6393.232 -4.480 0.000 Punjab & Sind Bank -71849.270 -4.363 0.000 -65983.535 -6.852 0.000 Punjab National Bank -79229.726 -4.530 0.000 -73249.639 -6.838 0.000 State Bank of Bikaner & Jaipur -72529.804 -4.385 0.000 -65696.653 -6.836 0.000 State Bank of India -78764.352 -4.310 0.000 -72869.284 -6.336 0.000 State Bank of Mysore -71165.576 -4.319 0.000 -64565.665 -6.790 0.000 State Bank of Travancore -72212.815 -4.357 0.000 -66164.736 -6.863 0.000 Syndicate Bank -76269.478 -4.495 0.000 -70422.623 -6.922 0.000 United Commercial Bank -75788.491 -4.486 0.000 -70006.842 -6.934 0.000 Union Bank of India -77423.677 -4.495 0.000 -71365.530 -6.860 0.000 Vijaya Bank -74417.073 -4.475 0.000 -68716.606 -7.024 0.000 Yes Bank Ltd 0a 0a [year=2007] 6722.166 4.062 0.000 7065.554 4.861 0.000 [year=2008] 6388.592 4.283 0.000 6702.933 5.156 0.000 [year=2009] 5575.231 4.254 0.000 5793.970 5.238 0.000 [year=2010] 4680.356 4.150 0.000 4689.801 4.862 0.000 [year=2011] 4913.351 4.730 0.000 4849.746 5.649 0.000 Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 281 ajfa.macrothink.org [year=2012] 4049.743 4.963 0.000 3671.839 5.644 0.000 [year=2013] 2428.967 3.468 0.001 1952.050 3.460 0.001 [year=2014] 1573.958 2.660 0.008 1002.343 2.099 0.037 [year=2015] 566.039 1.070 0.286 412.203 0.874 0.383 [year=2016] 0a 0a ln(Total Assets) 5066.337 4.252 0.000 5462.618 5.254 0.000 [sector=0] * ln(Total Assets) -4332.851 -5.334 0.000 -5020.016 -8.847 0.000 [sector=1] * ln(Total Assets) 0a 0a Return on Assets -131.243 -0.242 0.809 -683.160 -1.723 0.086 [sector=0] * Return on Assets -3871.904 -3.649 0.000 -2838.175 -3.944 0.000 [sector=1] * Return on Assets 0a 0a Beta 2302.747 2.292 0.023 2342.717 2.377 0.018 [sector=0] * Beta 6221.938 4.670 0.000 7214.696 5.764 0.000 [sector=1] * Beta 0a 0a Leverage -17.468 -1.138 0.256 -19.551 -1.294 0.197 [sector=0] * Leverage 131.409 2.014 0.045 129.460 2.096 0.037 [sector=1] * Leverage 0a 0a Capital Adequacy Ratio 10.027 0.070 0.944 [sector=0] * Capital Adequacy Ratio -165.643 -0.905 0.366 [sector=1] * Capital Adequacy Ratio 0a Net Non-Performing Assets to Net Advances 256.942 1.512 0.132 [sector=0] * Net Non-Performing Assets to Net Advances -509.354 -1.041 0.299 [sector=1] * Net Non-Performing Assets to Net Advances 0a Price to Book Value Ratio -2492.467 -3.717 0.000 -2732.728 -4.170 0.000 [sector=0] * Price to Book Value Ratio 2353.533 3.398 0.001 2657.621 3.906 0.000 [sector=1] * Price to Book Value Ratio 0a 0a Deposits to Total Assets 5194.245 1.046 0.297 2474.266 .540 0.589 [sector=0] * Deposits to Total Assets -17527.892 -2.395 0.017 -15445.471 -2.258 0.025 [sector=1] * Deposits to Total Assets 0a 0a Loans & Advances to Total Assets -6438.208 -0.856 0.393 [sector=0] * Loans & Advances to Total Assets -16477.730 -1.110 0.268 [sector=1] * Loans & Advances to Total Assets 0a Investments to Total Assets -15264.478 -1.902 0.058 [sector=0] * Investments to Total Assets 3935.368 0.286 0.775 [sector=1] * Investments to Total Assets 0a Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 282 ajfa.macrothink.org Net Interest Income to Total Funds 1.879 0.004 0.997 27.788 0.057 0.954 [sector=0] * Net Interest Income to Total Funds 2338.908 3.145 0.002 1827.424 2.637 0.009 [sector=1] * Net Interest Income to Total Funds 0a 0a Non-Interest Income to Total Funds 1199.327 1.346 0.180 [sector=0] * Non-Interest Income to Total Funds -1161.044 -0.989 0.324 [sector=1] * Non-Interest Income to Total Funds 0a a. This parameter is set to zero because it is redundant. There were several significant factors in the model. The bank fixed effects were found to be significant, indicating that there were significant differences in systemic impact between the banks. In particular, the banks with highest systemic impact were Karur Vysaya Bank, Jammu and Kashmir Bank, and Karnataka Bank (all of which are private sector banks), while the banks with least systemic impact were Bank of Baroda, Punjab National Bank, and State Bank of India (all of which are public sector banks). The year random effects were also found to be significant, indicating significant differences in systemic impact over time. Of course, systemic impact was highest in the crisis period of 2007-09, and there was found to be a significant decrease in systemic impact in 2015-16 as compared with previous years. This could be the result of tightening of capital regulations with the implementation of the Basel III norms from 2013. Bank size was found to be significant and positively related with systemic impact; however, for private sector banks, the relationship was not significant. Return on assets was found to be not significant; however, for private sector banks, return on assets was significant and negatively related with systemic impact. Beta was found to be significant and positively related with systemic impact, and was more influential for private sector banks than for public sector banks. Leverage was found to be not significant; however, for private sector banks, leverage was significant and positively related with systemic impact. Price to Book Value Ratio was found to be significant and negatively related with systemic impact; however, for private sector banks, the relationship was not significant. Deposits to Total Assets was found to be not significant; however, for private sector banks, deposits to total assets was significant and negatively related with systemic impact. Net Interest Income to Total Funds was found to be not significant; however, for private sector banks, net interest income to total funds was significant and positively related with systemic impact. Finally, Capital Adequacy Ratio, Net Non-Performing Assets to Net Advances, Loans & Advances to Total Assets, Investments to Total Assets, and Non-Interest Income to Total Funds were found to be not significant. The results of the panel regression for public sector banks are presented in Tables 4 and 5 below. Table 4 presents the summary of statistical tests for groups and covariates, while Table 5 presents the parameter estimates and significance. Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 283 ajfa.macrothink.org Table 4. Tests of Between-Subjects Effects (public sector banks) Dependent Variable: SRISK ($ m) Source F Stat p-value Intercept 9.594 0.002 bank 4.349 0.000 year 2.749 0.005 ln(Total Assets) 12.073 0.001 Return on Assets 0.423 0.517 Beta 2.679 0.104 Leverage 2.274 0.134 Capital Adequacy Ratio 0.002 0.965 Net Non-Performing Assets to Net Advances 0.000 0.985 Price to Book Value Ratio 30.831 0.000 Deposits to Total Assets 1.376 0.243 Loans & Advances to Total Assets 0.283 0.595 Investments to Total Assets 4.275 0.040 Net Interest Income to Total Funds 0.578 0.448 Non-Interest Income to Total Funds 1.328 0.251 Table 5. Parameter Estimates (public sector banks) Dependent Variable: SRISK ($ m) Parameter Coeff t Stat p-value Intercept -71514.348 -3.165 0.002 Allahabad Bank -1636.940 -1.490 0.138 Andhra Bank -498.757 -0.662 0.509 Bank of Baroda -5312.641 -2.351 0.020 Bank of India -3961.561 -1.796 0.075 Bank of Maharashtra 885.432 1.439 0.152 Canara Bank -3998.239 -1.922 0.056 Central Bank of India -1851.295 -1.293 0.198 Corporation Bank -789.810 -0.917 0.361 Dena Bank 1155.592 1.930 0.055 IDBI Bank Ltd -2325.580 -1.312 0.191 Indian Overseas Bank -2119.980 -1.557 0.122 Punjab & Sind Bank 2290.483 2.409 0.017 Punjab National Bank -4745.687 -1.930 0.055 State Bank of Bikaner & Jaipur 1528.728 1.893 0.060 State Bank of India -4157.103 -0.987 0.325 State Bank of Mysore 3082.076 3.305 0.001 Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 284 ajfa.macrothink.org State Bank of Travancore 1802.566 2.713 0.007 Syndicate Bank -2135.242 -1.753 0.082 United Commercial Bank -1231.812 -1.160 0.248 Union Bank of India -2984.208 -1.779 0.077 Vijaya Bank 0a [year=2007] 5795.819 2.747 0.007 [year=2008] 5619.692 2.895 0.004 [year=2009] 3876.008 2.379 0.019 [year=2010] 3913.768 2.838 0.005 [year=2011] 3228.041 2.742 0.007 [year=2012] 2342.197 2.710 0.007 [year=2013] 1449.163 2.124 0.035 [year=2014] 326.312 0.604 0.547 [year=2015] -415.003 -0.830 0.408 [year=2016] 0a ln(Total Assets) 5198.187 3.475 0.001 Return on Assets -250.612 -0.650 0.517 Beta 1290.967 1.637 0.104 Leverage -16.780 -1.508 0.134 Capital Adequacy Ratio -4.500 -0.044 0.965 Net Non-Performing Assets to Net Advances -2.493 -0.019 0.985 Price to Book Value Ratio -2920.455 -5.553 0.000 Deposits to Total Assets 4212.420 1.173 0.243 Loans & Advances to Total Assets -2880.443 -0.532 0.595 Investments to Total Assets -11938.625 -2.068 0.040 Net Interest Income to Total Funds 322.384 0.760 0.448 Non-Interest Income to Total Funds 747.846 1.153 0.251 a. This parameter is set to zero because it is redundant. For the public sector banks, the bank fixed effects were again found to be significant, indicating that there were significant differences in systemic impact between the banks. In particular, the banks with highest systemic impact were State Bank of Mysore, Punjab & Sind Bank, and State Bank of Travancore, while the banks with least systemic impact were Bank of Baroda, Punjab National Bank, and State Bank of India. The year random effects were also found to be significant, again indicating a significant decrease in systemic impact in 2015-16 as compared with previous years. Further, for the public sector banks, bank size was found to be significant and positively related with systemic impact, while price to book value ratio and investments to total assets ratio were found to be significant and negatively related with systemic impact. The other variables were not significantly related with systemic impact. The results of the panel regression for private sector banks are presented in Tables 4 and 5 below. Table 4 presents the summary of statistical tests for groups and covariates, while Table 5 presents the parameter estimates and significance. Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 285 ajfa.macrothink.org Table 6. Tests of Between-Subjects Effects (private sector banks) Dependent Variable: SRISK ($ m) Source F Stat p-value Intercept 0.024 0.877 bank 10.667 0.000 year 2.884 0.006 ln(Total Assets) 1.661 0.202 Return on Assets 8.509 0.005 Beta 18.668 0.000 Leverage 0.020 0.888 Capital Adequacy Ratio 0.146 0.704 Net Non-Performing Assets to Net Advances 0.411 0.524 Price to Book Value Ratio 0.000 0.994 Deposits to Total Assets 1.134 0.291 Loans & Advances to Total Assets 0.024 0.877 Investments to Total Assets 10.667 0.000 Net Interest Income to Total Funds 2.884 0.006 Non-Interest Income to Total Funds 1.661 0.202 Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 286 ajfa.macrothink.org Table 7. Parameter Estimates (private sector banks) Dependent Variable: SRISK ($ m) Parameter Coeff t Stat p-value Intercept -7473.603 -0.273 0.785 Axis Bank Ltd -5928.249 -2.223 0.030 Federal Bank Ltd 2719.082 1.295 0.200 HDFC Bank Ltd -17370.329 -5.311 0.000 ICICI Bank Ltd -12577.122 -3.244 0.002 IndusInd Bank Ltd -895.387 -0.555 0.581 Jammu & Kashmir Bank Ltd/The 5842.883 1.985 0.051 Karnataka Bank Ltd/The 4870.375 2.077 0.042 Karur Vysya Bank Ltd/The 8118.748 3.035 0.003 Kotak Mahindra Bank Ltd -6581.768 -2.924 0.005 Yes Bank Ltd 0a [year=2007] 8448.572 2.765 0.007 [year=2008] 7777.578 2.960 0.004 [year=2009] 9369.069 3.805 0.000 [year=2010] 6195.154 2.901 0.005 [year=2011] 7651.405 3.710 0.000 [year=2012] 7039.089 4.045 0.000 [year=2013] 3936.757 2.454 0.017 [year=2014] 3357.982 2.443 0.017 [year=2015] 1173.581 1.014 0.314 [year=2016] 0a ln(Total Assets) 1813.567 1.289 0.202 Return on Assets -3963.675 -2.917 0.005 Beta 9623.341 4.321 0.000 Leverage -16.043 -0.142 0.888 Capital Adequacy Ratio -68.630 -0.382 0.704 Net Non-Performing Assets to Net Advances 457.091 0.641 0.524 Price to Book Value Ratio -2.145 -0.007 0.994 Deposits to Total Assets -8651.957 -1.065 0.291 Loans & Advances to Total Assets -31624.452 -1.568 0.122 Investments to Total Assets -24602.327 -1.280 0.205 Net Interest Income to Total Funds 1980.436 2.270 0.027 Non-Interest Income to Total Funds 596.284 0.460 0.647 a. This parameter is set to zero because it is redundant. For the private sector banks, the bank fixed effects were again found to be significant, indicating that there were significant differences in systemic impact between the banks. In particular, the banks with highest systemic impact were Karur Vysaya Bank, Jammu and Kashmir Bank, and Karnataka Bank, while the banks with least systemic impact were HDFC Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 287 ajfa.macrothink.org Bank, ICICI Bank, and Kotak Mahindra Bank. The year random effects were also found to be significant, again indicating a significant decrease in systemic impact in 2015-16 as compared with previous years. Further, for the private sector banks, return on assets was found to be significant and negatively related with systemic impact, while beta and net interest income to total funds ratio were found to be significant and positively related with systemic impact. The other variables were not significantly related with systemic impact. Discussion The results of the study have identified some banks with relatively high systemic impact, viz. Karur Vysaya Bank, Jammu and Kashmir Bank, and Karnataka Bank. These banks must be monitored more carefully, and perhaps may be required to hold more capital or liquid assets to avert crisis. The results of the study also suggest that systemic risk of Indian banks has been declining significantly from 2013. This is perhaps the result of higher capital controls by the RBI with the phased implementation of Basel III norms in India. The results of the study indicate that public sector banks have a much higher level of systemic impact than private sector banks. Further, the determinants of systemic impact are different for public sector and private sector banks. The systemic impact of public sector banks was positively related with size and negatively related with price to book value ratio and investments to total assets ratio, while the systemic impact of private sector banks was negatively related with return on assets and positively related with beta and net interest income to total funds ratio. The presence of a size effect for systemic impact in the case of public sector banks suggests that consolidation for public sector banks may increase instability of the financial system. This is not the case for private sector banks, so that private sector bank mergers may be beneficial for systemic risk. This would, however, need to be studied in greater detail. Several of the findings are similar to those in the literature. Bank size was found to be significant and positively related with systemic impact for public sector banks, as suggested by several authors (Moore and Zhou, 2014; Laeven et al, 2016). Return on assets was found to be significant and negatively related with systemic impact for private sector banks, which is related to the findings of van Oordt and Zhou (2015). Beta was found to be significant and positively related with systemic impact, which is related to the findings of Anghelache and Oanea (2016). Leverage was found to be significant and positively related with systemic impact for private sector banks, as suggested by Anghelache and Oanea (2016). Price to book value ratio was found to be significant and negatively related with systemic impact for public sector banks, as suggested by Anghelache and Oanea (2016). Some findings have not been discussed previously in the literature. For example, deposits to total assets was found to be significant and negatively related with systemic impact for private sector banks; on the other hand, loans & advances to total assets and investments to total assets were not significant. Also, net interest income to total funds was found to be significant and positively related with systemic impact for private sector banks. On the other hand, some of the findings are contrary to the literature; Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 288 ajfa.macrothink.org for example, non-interest income to total funds were found to be not significant, contrary to the finding of Moore and Zhou (2014). Finally, two important variables, capital adequacy ratio and net non-performing assets to net advances, were found to be not significant, contrary to the findings of Laeven et al (2016) and van Oordt and Zhou (2014), respectively. The results of the study suggest that capital adequacy does not have much of an effect on systemic impact, which is contrary to economic logic. This would have to be investigated further to understand the interlinkage between capital adequacy, leverage, liquidity, and other relevant variables. In particular, this would have important policy implications for the regulation of bank capital and leverage. There are some limitations inherent in the study. The sample considered for the study was relatively small, and consisted of the relatively larger Indian banks. Also, the global financial crisis and Euro-zone crises had taken place during the study period, possibly contaminating the results. Further, there could be some multicollinearity between the variables, since many of the measures considered are related. For example, capital adequacy has improved in recent years, so that the significance of capital adequacy could have been affected by the year random effect. The results of the study thus need to be tested for robustness. References Acharya, V.V., Brownlees, C., Engle, R., Farazmand, F., & Richardson, M. (2010). Measuring Systemic Risk. in Regulating Wall Street: The Dodd-Frank Act and the New Architecture of Global Finance, edited by Acharya, V.V., Cooley, T., Richardson, M., and Walter, I., John Wiley & Sons. https://doi.org/10.1002/9781118258231 Acharya, V.V., Engle, R.F., & Richardson, M. (2012). Capital Shortfall: A new approach to ranking and regulating systemic risks. Presented at the American Economic Association Meeting held at Chicago, Illinois, USA on Jan 7, 2012. https://doi.org/10.1257/aer.102.3.59 Acharya, V.V., Pedersen, L., Philippon, T., & Richardson, M. (2010). Measuring Systemic Risk. NYU Stern Working Paper. https://doi.org/10.26509/frbc-wp-201002 Acharya, V.V., & Steffen, S. (2012). Analyzing Systemic Risk of the European Banking Sector. NYU Stern Working Paper. in Handbook on Systemic Risk, edt. J.-P- Fouque and J. Langsam. Cambridge University Press Adrian, T., & Brunnermeier, M.K. (2008). CoVaR. Federal Reserve Bank of New York Staff Report No. 348. https://doi.org/10.2139/ssrn.1269446 Anghelache, G.-V., & Oanea, D.-C. (2016). Romanian Commercial Banks’ Systemic Risk and Its Determinants: A CoVaR Approach. International Journal of Academic Research in Accounting, Finance and Management Sciences, 6(3), 96-109. https://doi.org/10.6007/IJARAFMS/v6-i3/2175 Boot, A.W.A., & Ratnovski, L. (2012). Banking and Trading. International Monetary Fund Working Paper No. 12/238. https://doi.org/10.5089/9781475511215.001 Bostandzic, D., Pelster, M., & Weiss, G.N.F. (2014). Systemic risk, bank capital, and deposit Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 289 ajfa.macrothink.org insurance around the world. SSRN Working Paper No. 2438693. https://doi.org/10.2139/ssrn.2438693 Brownlees, C.T., & Engle, R.F. (2012). Volatility, Correlation and Tails for Systemic Risk Measurement. NYU Stern Working Paper. https://doi.org/10.2139/ssrn.1611229 Brownlees, C.T., & Engle, R.F. (2017). SRISK: a conditional capital shortfall model for systemic risk. European Systemic Risk Board Working Paper. https://doi.org/10.1093/rfs/hhw060 Dash, M., & Das, A. (2013). Performance Appraisal of Indian Banks Using CAMELS Rating. IUP Journal of Bank Management, 12(2), 31-42. Dash, M., Kumari, G., & Anand, S. (2015). Comparison of Public and Private Sector Banking Performance using CAMELS Framework. MANAGEMENT TODAY- An International Journal of Management Studies, 5(3), 107-112. https://doi.org/10.11127/gmt.2015.09.02 Farhi, E., & Tirole, J. (2012). Collective Moral Hazard, Maturity Mismatch and Systemic Bailouts. American Economic Review, 12(1), 60-93. https://doi.org/10.1257/aer.102.1.60 Hattori, A., Kikuchi, K., Niwa, F., & Uchida, Y. (2014). A Survey of Systemic Risk Measures: Methodology and Application to the Japanese Market. IMES Discussion Paper Series, No. 2014-E-3. Hautsch, N., Schaumburg, J., & Schienle, M. (2013). Forecasting Systemic Impact in Financial Networks. Syrto Working Paper Series, Paper No. 16/2013. https://doi.org/10.2139/ssrn.2315827 Hautsch, N., Schaumburg, J., & Schienle, M. (2015). Financial Network Systemic Risk Contributions. Review of Finance, 19(2), 685-738. https://doi.org/10.1093/rof/rfu010 Hendricks, D., Kambhu, J., & Mosser, P. (2006). Systemic Risk and the Financial System: Background Paper. NAS-FRBNY Conference on New Directions in Understanding Systemic Risk held at Federal Reserve Bank of New York on May 18, 2006. Karimalis, E.N., & Nomikos, N. (2014). Measuring systemic risk in the European banking sector: A Copula CoVaR approach. Cass Business School Working Paper. Laeven, L., & Levine, R. (2007). Is there a diversification discount in financial conglomerates?” Journal of Financial Economics, 85, 331-367. https://doi.org/10.1016/j.jfineco.2005.06.001 Laeven, L., Ratnovski, L., & Tong, H. (2016). Bank Size, Capital, and Systemic Risk: Some International Evidence. Journal of Banking and Finance, 69(1), S25-S34. https://doi.org/10.1016/j.jbankfin.2015.06.022 Moore, K., & Zhou, C. (2014). The Determinants of Systemic Importance. Systemic Risk Centre, London School of Economics, Discussion Paper No. 19. Stolbov, M. (2017). Assessing systemic risk and its determinants for advanced and major emerging economies: the case of ΔCoVaR. International Economics and Economic Policy, Asian Journal of Finance & Accounting ISSN 1946-052X 2019, Vol. 11, No. 1 290 ajfa.macrothink.org 14(1), 119-142. https://doi.org/10.1007/s10368-015-0330-2 Suh, S., Jang, I., & Ahn, M. (2013). A Simple Method for Measuring Systemic Risk using Credit Default Swap Market Data. Journal of Economic Development, 38(4), 75-100. van Oordt, M.R.C., & Zhou, C. (2015). Systemic Risk and Bank Business Models. De Nederlandsche Bank Working Papers, Paper No. 442. https://doi.org/10.2139/ssrn.2509314