銀行的流動性風險與績效之關係—跨國之實證分析

全文

(1)國立高雄大學金融管理學系碩士論文. Bank Liquidity Risk and Performance: A Cross-Country Analysis 銀行的流動性風險與績效之關係—跨國之實證分析. 研究生:葉顓儀撰指導教授:陳怡凱博士、高蘭芬博士. 中華民國九十八年六月.

(2) 致謝詞. 研究所兩年的時間過得真快，轉眼之間研究生的生活即將結束。研究所在學期間承蒙吾師陳怡凱博士與高蘭芬博士細心與耐心的教誨，使本篇論文得以順利的完成，在此致上最誠摯的謝意。在論文的研究過程當中，感謝陳怡凱老師的教導與指正，讓我在專業的領域上能夠更加的精進，同時也激發了我的學習潛能。陳怡凱老師面對學生的態度輕鬆、隨性但不隨便，所以每次的 meeting 不必壟罩在嚴肅的氣氛當中，就能夠使我穫益良多。. 口試的期間，承蒙林靜香博士及王澤世博士在百忙之中，給予詳細的審閱與指導，並提供許多寶貴的意見，使本篇論文的內容可以更加的詳實與嚴謹。此外，感謝高大金管所各位老師這兩年來對學生的照顧與付出，給予我專業上的知識與獨立思考的能力，在此致上最大的謝意。. 在研究所這段期間，感謝同學們彼此扶持、彼此幫助。課外之餘的活動也替我的生活更增添了一份色彩，使我的研究所生活更多采多姿。此外，也謝謝系辦助理給予行政方面的協助。. 最後，最感謝的就是我的家人，爸爸、媽媽在我這二十多年的求學生涯中，不斷對我的付出與關懷，讓我可以無後顧之憂地完成我的學業。謹以此論文獻給我的父母及師長們。. 葉顓儀謹誌於高雄大學金融管理學系中華民國九十八年.

(3) Bank Liquidity Risk and Performance: A Cross-Country Analysis Advisor: Professor Yi-Kai Chen Department of Finance, National University of Kaohsiung Advisor: Professor Lan-Feng Kao Department of Finance, National University of Kaohsiung Student: Chuan-Yi Yeh Department of Finance, National University of Kaohsiung Abstract The aim of this study is to employ alternative liquidity risk measures besides liquidity ratio, and to investigate the causes of liquidity risk (causes of liquidity risk model), using an unbalanced panel dataset of 12 advanced economies commercial banks over the period 1994-2006. Besides, we estimate the causes of liquidity risk model through the fixed effects regression. In addition, we further investigate the relationship between bank liquidity risk and performance (bank liquidity risk and performance model). In our study, we regard liquidity risk as an endogenous determinant of bank performance. Thus, we apply panel data instrumental variables regression, using two stage least squares (2SLS) estimators to estimate bank liquidity risk and performance model. We find that liquidity risk is the endogenous determinant of bank performance. The causes of liquidity risk include components of liquid assets and dependence on external funding, supervisory and regulatory factors and macroeconomic factors. Besides, we also find that liquidity risk may lower bank profitability (return on average assets and return on average equities) because of higher cost of fund, but increase bank’s net interest margins. Besides, we classify countries as bank-based or market-based financial system. The empirical results show that liquidity risk is negatively related to bank performance in market-based financial system; however, it has no effect on bank performance in bank-based financial system.. Key Words: Liquidity Risk, Fixed Effects, Performance, Instrumental Variables, Financial System I.

(4) 銀行的流動性風險與績效之關係—跨國之實證分析指導教授：陳怡凱博士國立高雄大學金融管理學系指導教授：高蘭芬博士國立高雄大學金融管理學系學生：葉顓儀國立高雄大學金融管理學系. 摘要使用 1994-2006 年 12 個已開發國家商業銀行的橫縱資料，本研究的主要目的為使用其他非流動比率的方法來衡量銀行的流動性風險，並且探討銀行流動性風險的成因(流動性風險成因模型)。此外，我們藉由固定效果迴歸來估計流動性風險成因模型。不僅如此，本研究進一步探討銀行的流動性風險與績效之間的關係(銀行的流動性風險與績效之關係模型)。且本研究將流動性風險視為銀行績效的內生性決定因素。因此，我們利用工具變數迴歸且使用兩階段最小平方法來估計銀行的流動性風險與績效之關係模型。實證結果發現流動性風險為銀行績效的內生性決定因素。且流動性風險的成因包含資產的流動性、對外融資依賴度的要素，政府監管要素與總體經濟條件。此外，本研究也發現流動性風險較大的銀行會因為有較高的融資成本，因而降低了獲利。但是流動性風險對於銀行的淨利息邊際卻存在著流動性溢酬。此外，本研究將樣本國家區分為以銀行為基礎或以金融市場為基礎的金融體系。實證結果發現，以金融市場為基礎的金融體系之下，流動性風險會降低銀行的績效。然而，以銀行為基礎的金融體系之下，流動性風險對於銀行的績效則不存在影響性。. 關鍵字：流動性風險、固定效果、績效、工具變數、金融體系. II.

(5) Contents. Abstract ......................................................................................................................... I Abstract (in Chinese) .................................................................................................. II Contents ..................................................................................................................... III List of Tables ............................................................................................................... V List of Figures ........................................................................................................... VII Chapter 1 Introduction................................................................................................ 1 1.1 Background and Motivations ........................................................................... 1 1.2 The Purposes and Major Findings ................................................................... 3 1.3 Research Framework ....................................................................................... 7 Chapter 2 Literature Review ...................................................................................... 9 2.1The Profile of Liquidity Risk ............................................................................ 9 2.2 Quantitative Measures of Liquidity Risk ....................................................... 11 2.2.1 Liquidity Ratios .................................................................................. 12 2.2.2 Maturity Ladder .................................................................................. 13 2.2.3 Sources and Uses of Liquidity ............................................................ 13 2.2.4 Peer Group Ratio Comparisons .......................................................... 14 2.2.5 Liquidity Index.................................................................................... 14 2.2.6 Financing Gap and the Financing Requirement.................................. 15 2.2.7 Liquidity Planning .............................................................................. 15 2.2.8 Balance Sheet Liquidity Analysis ....................................................... 16 2.2.9 Cash Capital Position .......................................................................... 16 2.3 Qualitative Measures of Liquidity Risk ......................................................... 17. III.

(6) Chapter 3 Data and Methodology ............................................................................ 19 3.1 Sample Selection and Data Description ......................................................... 19 3.2 Variable Description....................................................................................... 20 3.2.1 Liquidity Risk Measures ..................................................................... 20 3.2.2 Causes of Liquidity Risk ..................................................................... 22 3.2.2.1 Bank-specific Risk Causes ....................................................... 22 3.2.2.2 Supervisory and Regulatory Risk Causes ................................ 24 3.2.2.3 Macroeconomic Risk Causes ................................................... 25 3.2.3 Determinants of Bank Performance .................................................... 26 3.2.3.1 Bank-specific Performance Determinants ............................... 26 3.2.3.2 Market Structure Performance Determinants .......................... 29 3.2.3.3 Supervisory and Regulatory Performance Determinants ......... 29 3.2.3.4 Macroeconomic Performance Determinants............................ 30 3.3 Econometric Specification ............................................................................. 31 3.3.1 Panel Unit Root Tests ......................................................................... 31 3.3.2 Causes of Liquidity Risk Model ......................................................... 33 3.3.3 Bank Liquidity Risk and Performance Model .................................... 35 3.3.4 Subsample Analysis ............................................................................ 37 Chapter 4 Empirical Results ..................................................................................... 39 4.1 Descriptive Statistics ...................................................................................... 39 4.2 Panel Unit Root Tests and Collinearity Test Results ..................................... 39 4.3 Regression Results ......................................................................................... 40 4.4 Regression Results in Different Financial System......................................... 45 4.5 Robust Tests ................................................................................................... 48 Chapter 5 Conclusions............................................................................................... 50 Reference .................................................................................................................... 54 Appendix A: Net Funding Requirement and Net Liquidity Statement ................ 88 Appendix B: Balance Sheet Liquidity Analysis and Cash Capital Position ......... 89 Appendix C: Survey Questions of Bank Regulation and Supervision .................. 90 IV.

(7) List of Tables. Table 1 Empirical Results of the Relationship between Bank Liquidity Risk and Performance ..................................................................................................... 61 Table 2 Bank Observations in Each Country and Year ................................................ 62 Table 3 Variable Description ........................................................................................ 63 Table 4 Descriptive Statistics ....................................................................................... 65 Table 5 Descriptive Statistics in Each Country ............................................................ 66 Table 6 Panel Unit Root Test Results Using Balanced Panel Data ............................. 70 Table 7 Panel Unit Root Test Results Using Unbalanced Panel Data ......................... 71 Table 8 Collinearity Test .............................................................................................. 72 Table 9 Causes of Liquidity Risk Results Using FGAPR to Measure Liquidity Risk 74 Table 10 Bank Liquidity Risk and Performance Results Using FGAPR to. Measure. Liquidity Risk ................................................................................................ 75 Table 11 Causes of Liquidity Risk Results in Different Financial System (Dependent variable: FGAPR) .......................................................................................... 76 Table 12 The Relationship Between Financial System and Bank Performance Using FGAPR to Measure Liquidity Risk ............................................................... 77 Table 13 Bank Liquidity Risk and Performance in Different Financial Systems Using FGAPR to Measure Liquidity Risk (Dependent Variable: ROAA) .............. 78 Table 14 Bank Liquidity Risk and Performance in Different Financial Systems Using FGAPR to Measure Liquidity Risk (Dependent Variable: ROAE) .............. 79. V.

(8) Table 15 Bank Liquidity Risk and Performance in Different Financial Systems Using FGAPR to Measure Liquidity Risk (Dependent Variable: NIM) ................. 80 Table 16 Causes of Liquidity Risk Results Using Alternative Liquidity Risk Measures ........................................................................................................ 81 Table 17 Bank Liquidity Risk and Performance Results Using NLCS to Measure Liquidity Risk ................................................................................................ 82 Table 18 Causes of Liquidity Risk Results in Different Financial System Using Alternative Liquidity Risk Measures (Dependent Variable: NLCS) ............. 83 Table 19 The Relationship Between Financial System and Bank Performance Using NLCS to Measure Liquidity Risk .................................................................. 84 Table 20 Bank Liquidity Risk and Performance in Different Financial Systems Using NLCS to Measure Liquidity Risk (Dependent Variable: ROAA) ................. 85 Table 21 Bank Liquidity Risk and Performance in Different Financial Systems Using NLCS to Measure Liquidity Risk (Dependent Variable: ROAE) ................. 86 Table 22 Bank Liquidity Risk and Performance in Different Financial Systems Using NLCS to Measure Liquidity Risk (Dependent Variable: NIM) .................... 87. VI.

(9) List of Figures. Figure 1 : Flowchart of Research Framework ............................................................... 8. VII.

(10) Chapter 1 Introduction. 1.1 Background and Motivations Since August 2007, the U.S subprime mortgage crisis has not only threatened to the U.S. economy into a recession, but affected the global financial system. Furthermore, it brings a huge challenge to short-term and long-term development for global banking industry. Because the crisis has caused banks and other financial institutions become nervous about lending to other banks, banks generally lack of liquidity following the subprime mortgage crisis. Especially, banks depend heavily on the short-term money market or purchased funds market will be more likely to suffer liquidity problem in the future, and the Northern Rock is an example. The funding sources of Northern Rock, one of the five largest British mortgage lenders, mostly relied on wholesale money markets and securitization of mortgages instead of customer deposits. So it is extremely lack of stable funding sources. After subprime mortgage crisis, Northern Rock was unable acquire funding from money market because of credit freeze. In September 2007, Northern Rock was influenced by magnitude liquidity squeezes, and forced to a bailout from the Bank of England. It consequently suffered the bank run crisis. From Northern Rock crisis we can realize the importance of bank liquidity and diversified funding sources, though liquidity risk was rarely mentioned in the past. Swary (1986) also provided an explanation of the failure of the Continental Illinois National Bank in the U.S, which only had a small part of core deposit on its liability side. Thus, it is worth of discussing for bank’s liquidity risk.. 1.

(11) According to the definition of the Basel Committee on Banking Supervision (1997), liquidity risk arises from the inability of a bank to accommodate decreases in liabilities or to fund increases in assets. When a bank has inadequate liquidity, it cannot obtain sufficient funds, either by increasing liabilities or by converting assets promptly, at a reasonable cost, thereby affecting profitability. Besides, Decker (2000) indicated that liquidity risk can be divided into funding liquidity risk and market liquidity risk. Comparing to credit risk, there are fewer literature to discuss with liquidity risk. Basel I Accord (Basel Committee on Banking Supervision, 1988) set out regulatory standards for credit risk and market risk. Besides, Basel II Accord (Basel Committee on Banking Supervision, 2004) even takes operational risk into account. However, they seldom mention the liquidity risk. Landskroner and Paroush (2008) also indicated that there has been extensive academic and regulatory discussion of the different major banking risks: credit risk, market risk and even operation risk. However relative little attention has been paid to liquidity risk that has become one of the major risks faced by banks and other financial institutions in recent years. Previously, the related literatures of liquidity risk mainly focus on bank run or failures (Diamond and Dybvig, 1983). Besides, previous empirical studies were mainly to investigate the determinants of bank profitability or net interest margins. ( e.g. Bourke, 1989; Molyneux and Thornton, 1992; Demirgüç-Kunt and Huizinga, 1999; Shen et al., 2001; Barth et al., 2003; Demirgüç-Kunt et al., 2003; Kosmidou et al., 2005; Athanasoglou et al., 2006; Pasiouras and Kosmidou, 2007; Athanasoglou et al., 2008; Kosmidou, 2008; Naceur and Kandil, 2009 ). They usually use liquidity ratios to measure bank liquidity, and regarded liquidity risk as exogenous variable. However, there are seldom studies to discuss the causes of liquidity risk. Furthermore,. 2.

(12) previous empirical evidence showed that the effect of liquidity risk on bank profitability is mixed. Some studies found out the positive effect (e.g. Molyneux and Thornton, 1992; Barth et al., 2003); others found out the negative effect (e.g. Bourke, 1989; Demirgüç-Kunt and Huizinga, 1999; Kosmidou, 2005; Kosmidou, 2008). Besides, previous studies found that banks with high liquidity have lower net interest margins. (e.g. Demirgüç-Kunt and Huizinga, 1999; Shen et al., 2001; Demirgüç-Kunt et al., 2003; Naceur and Kandil, 2009). Regulator have strict request to the bank in credit risk and operational risk in the past, but do not focus on liquidity risk. However, we can found that liquidity risk will cause severe consequence to banks following the subprime mortgage crisis. Besides, the credit crunch of 2007 reminded many banks of the importance of liquidity risk (Matz, 2008). Thus, it is important for banks to strengthen liquidity risk management, and liquidity risk will be an important issue in the future.. 1.2 The Purposes and Major Findings Generally, liquidity risk measures can be calculated from balance sheet positions. In the past, better practices for liquidity risk measures focused on the use of liquidity ratios. However, Poorman and Blake (2005) indicated that it was not enough to measure liquidity just using liquidity ratios and it was not the solution. Beyond mere liquidity ratios, banks must develop a new view of liquidity measurement. Recently, there are many methods provided to assess bank liquidity risk besides traditional liquidity ratios. Therefore, the purpose of this study is to employ alternative liquidity risk measures besides liquidity ratio. In our study, we use financing gap measures provided by Saunders and Cornett (2006) to assess bank liquidity risk. In normal 3.

(13) condition, banks seldom face the liquidity crisis, and liquidity risk may vary with overall economic environment. Besides, previous studies seldom focused on the causes of liquidity risk. Thus, another purpose of this study is to investigate the causes of liquidity risk (causes of liquidity risk model), using an unbalanced panel dataset of 12 advanced economies commercial banks over the period 1994-2006. Besides, we estimate the causes of liquidity risk model through the fixed effects regression. In this model, we use each bank’s financing gap ratio (FGAPR) as the dependent variable, and divide the causes of liquidity risk into internal and external factors as independent variable. The empirical results indicate that large banks have incentive to hold more loans thus have larger financing gap ratio. However, over the limit point the effect of size becomes negative. Thus, the effect of size on liquidity risk is non-linear. Banks with much less risky liquid assets and risky liquid assets can reduce their liquidity risk. Besides, banks depend heavily on the external funding face more severe liquidity problem. Thus banks should diversify their funding sources to reduce liquidity risk. In regulation and supervision, we can find that countries with greater official power, higher restrictiveness make their banks suffer less liquidity risk. However, we find no evidence that regulatory empowerment of private monitoring of banks has significantly impact on liquidity risk. Thus, we can find that direct government supervision and regulation of bank activities could reduce bank liquidity risk. About macroeconomic environment, the results indicated that banks run down their liquidity buffer in boom because they increase their loans but attract less customer deposits in this period. In addition, we further investigate the determinants of bank performance in terms of the perspective of the bank liquidity risk (bank liquidity risk and. 4.

(14) performance model). Previous studies regarded liquidity risk as exogenous determinant of bank performance. However, from the causes of liquidity risk model, we can find that bank liquidity risk may affected by another factors. In our study, we thus regard liquidity risk as an endogenous determinant of bank performance. Besides, we apply panel data instrumental variables regression, using two stage least squares (2SLS) estimators to estimate the determinants of bank performance model. In this model, we use return on average assets (ROAA), return on average equities (ROAE) and net interest margins (NIM) as the dependent variable. Besides, we divide the determinants of bank performance into internal and external factors as independent variable. The empirical results show that liquidity risk may lower bank profitability (ROAA and ROAE). Banks with larger gap lack stable and cheap fund, and thus they have to use liquid assets or much external funding to meet the demand of fund, increase their cost of funding. It consequently decreases their performance. However, liquidity risk will increase bank’s net interest margins. It indicated that banks with high levels of illiquid assets in loans may receive higher interest income. The effect of size provides evidence of economies of scale in banking. However, over the optimum point the effect of size becomes negative due to bureaucratic. Thus, the effect of size on bank profitability is non-linear. The results also indicated that capital strength of banks has a positive impact on their performance. We can find that increase exposure to credit risk will lower their profitability (ROAA and ROAE). However, credit risk has the positive effect on bank’s net interest margins. It provides that credit risk requires banks to apply a risk premium implicitly in the interest rates charge. About market structure, the effect of concentration is positive using ROAA and ROAE as the dependent variable, which provides evidence to support the structure-conduct-. 5.

(15) performance (SCP) hypothesis. Turning to supervision and regulation, the results support that greater official power, greater regulatory empowerment of private monitoring of banks, higher restrictiveness can increase bank’s performance. About macroeconomic environment, the results indicated that economic boom has significantly positive effect on bank performance. The relationship between inflation annual percent change of last year and bank performance is significantly positive. There. are. large. differences. in. financial. systems. across. countries.. Demirgüç-Kunt and Levine (1999) constructed conglomerate index of financial structure and produces two categories of countries: bank-based and market-based system. Besides, the financing behavior is very different between bank-based and market-based financial system. In our study, we classify countries as bank-based or market-based system, and investigate the difference of causes of liquidity risk in different financial systems. The empirical results indicated that the bank-specific variable has the same effect on bank liquidity risk in two financial systems. About supervision and regulation, it provides that greater official power, higher activity restrictiveness will diminish bank liquidity risk in market-based financial system. However, we find that greater regulatory empowerment of private monitoring of banks will increase bank liquidity risk in market-based financial system. Regarding macroeconomic environment, the results indicates that economic boom make banks in market-based financial system run down their liquidity buffer, but macroeconomic has no effect on bank liquidity risk in bank-based financial system. We also investigate the effect of financial system on bank performance. The empirical results show that market-based system has the positive effect on bank performance. This indicated that stock market development may improve bank 6.

(16) performance. Besides, we further investigate the determinants of bank performance in different financial systems. We find that liquidity risk is negatively related to bank performance in market-based financial system; however, it has no effect on bank performance in bank-based financial system. Specifically, we explore how to measure bank liquidity risk, what are the causes of liquidity risk, what is the relationship between liquidity risk and bank performance. Although liquidity risk is not the major risk in banks, it may cause banks to go into bankruptcy, thus we can’t ignore it as before. The contribution of this study is to use another liquidity risk measures besides to liquidity ratio, and we are the first study to investigate the causes of liquidity risk. Besides, we find that liquidity risk is an endogenous determinant of bank performance. In subsample analysis, we further classify countries as bank-based or market-based system, and investigate the difference of causes of liquidity risk in different financial systems. Besides, we further investigate the effect of liquidity risk on bank performance in different financial systems.. 1.3 Research Framework To achieve these purposes, this study is structured as follows. The first section deals with the background and motivations, purposes and major findings of this study. Literature review offers the liquidity risk measures will be shown in the second section. The third section includes sample selection and data description, variable description, and econometric specification. Finally, the empirical results and conclusions are showed in the fourth and fifth section. Besides, the flowchart of research framework of our study is shown in Figure 1:. 7.

(17) Introduction. Literature Review. The Profile of Liquidity Risk. Quantitative Measures of Liquidity Risk. Qualitative Measures of Liquidity Risk. Data and Methodology. Sample Selection and Data Description. Variable Description. Empirical Results. Conclusions. Figure 1 : Flowchart of Research Framework. 8. Econometric Specification.

(18) Chapter 2 Literature Review. 2.1The Profile of Liquidity Risk The risk of bank failures or insolvency is caused primarily by a shortage of liquidity (or liquidity risk). Saunders and Cornett (2006) indicated that liquidity risk may be from the liability side of the bank’s balance sheet, when deposit drains are abnormally large and unexpected. Besides, it may also arise from the asset side, when borrower largely exercises their loan commitments and other credit lines. Comparing to credit risk, there are fewer literature to discuss with liquidity risk. Basel II Accord published by Basel Committee on Banking Supervision (2004) set outs regulatory standards for credit risk, market risk and operational risks but says little about liquidity risk. Landskroner and Paroush (2008) also indicated that there has been extensive academic and regulatory discussion of the different major banking risks: credit risk, market risk and even operation risk. However relative little attention has been paid to liquidity risk that has become one of the major risks faced by banks and other financial institutions in recent years. Previously, the related literatures of liquidity risk mainly focus on bank run or failures. Diamond and Dybvig (1983) developed a model to explain why banks choose to issue deposits that are more liquid than their assets .They specifically investigated bank liquidity and found out that lack of it may lead to a bank run. A bank run is the sudden and unexpected increase in bank deposit withdrawals. Besides, the model has been widely used to understand bank runs and other types of financial crises, as well as ways to prevent such crises. Bank runs are traditionally assumed to take place among retail depositors. However, large wholesale depositors are increasingly more important, because banks 9.

(19) are increasingly dependent on wholesale funding. Thus, the interbank market becomes a key point of runs, such as the failure of the Franklin National Bank in 1974 and Continental Illinois National Bank in 1984. Swary (1986) provided an explanation of the failure of the Continental Illinois National Bank in Chicago in the United States. The bank only had a small part of core deposit on its liability side. It relied heavily on large deposits from other domestic banks, foreign deposits, and on interbank lines of credit for the funding of an undiversified loan asset portfolio. When Japanese banks became nervous about the oil producing loan exposures of the bank, they withdrew their lines of credit. This caused other banks to do the same and caused bank run. Consequently the bank was unable to fund its assets, and the bank eventually be taken over by the Federal Deposit Insurance Corporation. After the subprime mortgage crisis, people gradually realized the importance of liquidity risk. Furthermore, Decker (2000) indicated that liquidity risk can be divided into funding liquidity risk and market liquidity risk. Funding liquidity risk is the risk that bank will be unable to meet its obligations as they come due because of inability to liquidate assets or obtain adequate funding sources. Freixas et al. (2000) showed that liquidity may dry up for a solvent bank in the interbank market if there is imperfect information, or if there is market tension which reduces lending banks’ excess liquidity and reduces their scope to diversify. The interbank market as a whole may face liquidity problems if each bank refuses to lend to others because it cannot be confident of borrowing itself to address its own liquidity shortages. However, market liquidity risk is that banks cannot easily unwind or offset specific exposures without significantly lowering market prices because of inadequate market depth or market disruptions.. 10.

(20) Banks hoarded liquidity in order to provide sufficient funding for their ongoing business following the subprime crisis. Thus, it causes the whole credit squeeze. Furthermore, the shortage of aggregate liquidity brings about contagious failures in the banking system. Davis (2008) indicated that the understanding of the liquidity problems in 2007-8 require theory to go beyond the Diamond and Dybvig (1983) concept of bank funding liquidity risk to encompass market liquidity risk. Besides, he also point out that banks accelerate shift from holding loans on balance sheet to relying on securitization in the period 2000-7. This may cause banks to reduce the incentive to monitor loans. Besides, banks held increasingly low levels of liquid assets on balance sheet, given low interest rates, and they undertook aggressive wholesale liability management to maintain funding levels. One consequence of these problems was the failure of the solvent UK mortgage bank Northern Rock. Because it had an aggressive wholesale funding ratio and had been relying on securitizing assets, which was no longer feasible. We can realize that banks are vulnerable if they hold low level of liquid assets, overly rely on short term wholesale funding, and depend on securitization. Besides, we also realize that bank liquidity risk is important, and it is more and more valued.. 2.2 Quantitative Measures of Liquidity Risk Generally, liquidity risk measures can be calculated from balance sheet positions. In the past, better practices for liquidity risk measures focused on the use of liquidity ratios. However, there are another ways can be used to assess bank liquidity risk besides traditional liquidity ratios. Besides, they can be divided into quantitative measurement and qualitative measurement. About quantitative measurement, Basel Committee on Banking Supervision (2000) proposed maturity laddering method for. 11.

(21) measuring liquidity risk. Saunders and Cornett (2006) indicated that banks can use sources and uses of liquidity, peer group ratio comparisons, liquidity index, financing gap and the financing requirement, and liquidity planning to measure their liquidity exposure. Besides, Matz and Neu (2007) also indicated that banks can apply balance sheet liquidity analysis, cash capital position and maturity mismatch approach to assess liquidity risk. We will briefly discuss each liquidity risk measures below:. 2.2.1 Liquidity Ratios In the past, better practices for liquidity risk measures focused on the use of liquidity ratios. The ratios previous studies used include liquid assets to total assets ratio (e.g. Bourke, 1989; Molyneux and Thornton, 1992; Barth et al., 2003; Demirgüç-Kunt et al., 2003), liquid assets to deposits ratio (Shen et al., 2001) and liquid assets to customer and short term funding (Kosmidou et al., 2005). The higher value of liquidity ratio makes bank more liquid and less vulnerable to failure. Besides, some studies use loans to total assets ratio (e.g. Demirgüç-Kunt and Huizinga, 1999; Athanasoglou et al., 2006), net loans to customer and short term funding ratio (e.g. Pasiouras and Kosmidou, 2007; Kosmidou, 2008; Naceur and Kandil, 2009) to assess bank’s liquidity risk. The higher the value of these ratios, the more liquidity risk the banks will suffer. The liquidity ratios and the empirical results of the relationship between bank liquidity risk and performance are shown in Table 1. We can find that the effect of liquidity risk on bank profitability is mixed. Some studies found out the positive effect (e.g. Molyneux and Thornton, 1992; Barth, 2003); others found out the negative effect (e.g. Bourke, 1989; Demirgüç-Kunt and Huizinga, 1999; Kosmidou, 2005; Kosmidou, 2008). Besides, previous studies found that banks with high liquidity have lower net. 12.

(22) interest margins. (e.g. Demirgüç-Kunt and Huizinga, 1999; Shen et al., 2001; Demirgüç-Kunt et al., 2003; Naceur and Kandil, 2009).. 2.2.2 Maturity Ladder Basel Committee on Banking Supervision (2000) proposed maturity laddering method for measuring liquidity risk, and getting net funding (financing) requirements. This method assesses all cash inflows against its outflows. Besides, the example is shown in Table I of Appendix A. The maturity ladder method allows a comparison of cash inflows and outflows on a day-to-day basis or over a series of specified time periods. Daily and cumulative net funding requirements can be determined from this measurement. The appropriate time frame will depend on the nature of bank’s business. Banks that rely on short-term funding concentrate primarily on managing their liquidity in the very short term. Banks that are less dependent on short-term funding might actively manage their net funding requirements over longer period. While liquidity is typically managed under normal conditions, the BIS cautions that banks must also be prepared to manage liquidity under abnormal conditions.. 2.2.3 Sources and Uses of Liquidity When banks suffer liquidity problem, they need to meet fund demands through liquidating assets or borrowing funds. Thus, bank manager must measure its liquidity position on a daily basis. Saunders and Cornett (2006) indicated that the net liquidity statement that lists sources and uses of liquidity is a useful tool. It provides measure of banks net liquidity position. The example is shown in Table II of Appendix A. It shows that banks can obtain liquid funds in three ways. First, they can sell their liquid. 13.

(23) assets such as T-bills immediately with little price risk and low transaction cost. Second, they can borrow funds in the money or purchased funds market up to a maximum amount. Third, they can use any excess cash reserves over and above the amount held to meet regulatory imposed reserve requirements. Comparing the sources of liquidity with the uses of liquidity of bank in Table II of Appendix A, the bank has a positive net liquidity position. All banks report their historical sources and uses of liquidity in their annual and quarterly reports. As bank’s manager deals with liquidity risk, historical sources and uses of liquidity statements can assist the manager in determining where future liquidity issues may arise.. 2.2.4 Peer Group Ratio Comparisons Saunders and Cornett (2006) indicated that banks can compare key ratios and balance sheet features with those of banks of a similar size and geographic location to determine their liquidity exposure, such as loans to deposits, borrowed funds to total assets, and commitments to lend to assets ratio. A high ratio of loans to deposits and borrowed funds to total assets means that bank relies heavily on the short-term money market rather than on core deposits to fund loans. This could mean future liquidity problems if bank is at or near its borrowing limits in the purchased funds market. A high ratio of loan commitments to assets indicates the need for a high degree of liquidity to fund any unexpected execution of these loans. Thus, banks with high commitment often face more liquidity risk exposure.. 2.2.5 Liquidity Index Liquidity index is developed by Jim Pierce at the Federal Reserve, this index measures the potential losses a bank could suffer from a sudden or fire-sale disposal of assets compared with the amount it would receive at a fair market value established 14.

(24) under normal market conditions. The greater the differences between immediate fire-sale asset prices (Pi) and fair market prices (Pi*), the less liquid is bank’s portfolio of assets. The liquidity index I is defined as:. N  I   wi  i 1  .  Pi  *  Pi.   . (1). where I is liquidity index; Pi is fire-sale asset prices; Pi* is fair market prices. wi is the N. percent of each asset in the bank’s portfolio, and. ∑w. i. =1.. The liquidity index will. i =1. always lie between 0 and 1. The larger the discount from fair value, the smaller the liquidity index or higher the liquidity risk the bank faces. Besides, the liquidity index for this bank could also be compared with similar indexes calculated for a peer group of similar bank.. 2.2.6 Financing Gap and the Financing Requirement Saunders and Cornett (2006) indicated that banks can measure liquidity risk exposure by determining their financing gap. Bank managers often regard the average core deposit as stable source of funds, thus it can permanently fund a bank’s average loans. So the financing gap is defined as the difference between a bank’s average loans and average core deposits. If the financing gap is positive, the bank must fund it by using its cash, selling liquid assets and borrowing funds in the money market. A widening financing gap can warn of future liquidity problems for a bank.. 2.2.7 Liquidity Planning Liquidity planning allows managers to make important borrowing priority decisions before liquidity problems arise. Such planning can lower the cost of funds. 15.

(25) and can minimize the amount of excess reserves that a bank needs to hold. A liquidity plan includes four components. The first component is delineation of managerial details and responsibilities. Responsibilities are assigned to key management personnel should a liquidity crisis occur. The second component is to detailed list of fund providers who are most likely to withdraw, as well as the pattern of fund withdrawals. The third component is to identify the size of potential deposit and fund withdrawals over various time horizons in the future as well as funding sources to meet such withdrawals. The fourth component is to set internal limits on separate subsidiaries’ and branches’ borrowings.. 2.2.8 Balance Sheet Liquidity Analysis The balance sheet liquidity approach sets different balance sheet items on the asset side and the liability side into relation, depending on whether assets are liquid or illiquid, and on whether their funding is stable or volatile. The balance sheet liquidity analysis is shown in Figure I of Appendix B. Intangibles, strategic investments and equity capital are not included because of not for liquidation. To secure an appropriate balance sheet structure with respect to liquidity risk, sticky assets should be funded by stable liabilities, and liquid assets can be funded by volatile liabilities. However, this method misses time dimension. The balance sheet liquidity analysis characterizes balance sheet positions only as liquid or illiquid. There are no statements about in which time frame positions can be liquidated or become due.. 2.2.9 Cash Capital Position Moody’s originally invented the cash capital concept to analyze the liquidity structure of a bank’s balance sheet as part of its external rating process. Its intention is to measure the bank’s ability to fund its assets on a fully collateralized basis assuming 16.

(26) that the access to unsecured funding has been lost. In Figure II of Appendix B, the cash capital can be calculated as: Cash capital = collateral value of unencumbered assets – (short-term inter-bank funding + non-core deposits) = (core deposits + long-term funding + equity +contingency funding capacities) – (firm-wide haircuts + contingent outflows + illiquid assets). (2). The unencumbered assets are defined as assets that are available to be used as collateral.. 2.3 Qualitative Measures of Liquidity Risk Matz and Neu (2007) indicated that qualitative assessment of liquidity risk is at least as important as a quantitative measurement based on models. Besides, they provided some qualitative liquidity risk measures besides quantitative measures. The first is that is diversified funding sources established? They indicated that a bank should not only rely on unsecured inter-bank funding, non-bank deposits and long-term own-bond issues, and they should have established alternative funding sources such as commercial paper, secured funding via repos, and securitization. Besides, its own bond issues should be diversified on maturities, investors, and structure. Besides, bank should not heavily rely on liquidity backup lines from other banks, because such sources will cease to be available in a crisis. The second is that what are the current long term and short term rating and what is the outlook? They indicated that it is the greatest danger for the short-term solvency of a bank to downgrade bank’s short-term rating to a non-prime status. In this case,. 17.

(27) generally, unsecured funding sources will be significantly reduced and must be replaced by alternative, generally secured, funding sources. The third is that does the bank measure liquidity risk under different environments, including stress levels? Does the bank only consider fixed cash flows or does it model cash flows by taking into account behavioral adjustment? They indicated that scenario analyses, in particular, at stress levels are the key to proper liquidity management. The impact of various ―what-if‖ scenarios on the liquidity must be analyzed to bring the subject to the appropriate attention at senior management level. In such scenarios non-fixed cash flows that need to be modeled by behavioral adjustments are affected most. The fourth is that does a liquidity contingency plan exist that addresses responsibilities of each unit and the measures to be taken? They indicated that management usually does not have much time to react in a crisis situation. It is inevitable that each unit knows a priori what to do.. 18.

(28) Chapter 3 Data and Methodology. 3.1 Sample Selection and Data Description In our study, we use annual bank level, market structure, supervisory and macroeconomic data from 12 advanced economies (Australia, Canada, France, Germany, Italy, Japan, Luxembourg, Netherlands, Switzerland, Taiwan, United Kingdom and United States) over the period 1994-2006.1 The data was initially collected from 1993-2007, but was modified to include 1994-2006 due to large amounts of missing data in 1993 and 2007. Besides, we focus on commercial bank and delete the unavailable and incomplete observations. Finally, we yield an unbalanced panel data consisted of 14360 observations. The panel is unbalanced because it contains banks entering or dropping out of the market during the sample period (e.g. due to mergers). However, unbalanced panels are more likely to be the norm in typical economic empirical settings (Baltagi, 2005). Bank observations in each country and year are shown in Table 2. We can find that the number of banks is decreasing because of merger and acquisition. The data for calculation of bank-specific and market structure variables are available from Bankscope database.2 Besides, all variables available from Bankscope. 1. We choose advanced economies as our sample because of their transparent information, and well financial system. According to the definition of World Economic Outlook published by IMF, there are 31 advanced economies. For data completeness, we choose 12 countries. 2 In our study, we use unconsolidated bank statements (Bankscope consolidation codes U1, U2) where such statements are available. U* statements were used only if no other unconsolidated statements existed. If no unconsolidated statements were available, we used consolidated statements (C1, C2, C*). Banks with a consolidation status of A1 were dropped.. 19.

(29) database are adjusted for inflation. The data unit of each bank in a given year is million U.S dollars. Supervisory and regulatory variables are available from Barth et al. (2004). 3 Macroeconomic variables are available from International Monetary Fund’s (IMF) World Economic Outlook (WEO) Database.. 3.2 Variable Description In the causes of liquidity risk model, we consider two liquidity risk measures, bank-specific variables, supervisory and regulatory variables, and macroeconomic conditions. In the bank liquidity risk and performance model, we additionally consider the market structure variables. Table 3 provides a description of all the variables used in this study.. 3.2.1 Liquidity Risk Measures So far, there is no specific standard for measuring bank liquidity risk. Banks usually use a variety of methodologies to measure current and future liquidity, because no one metric provides a comprehensive view. In the past, better practices for liquidity risk measures focused on the use of liquidity ratios. However, Poorman and Blake (2005) indicated that measure liquidity just using liquidity ratios was not enough and it was not the solution. 4 Beyond mere liquidity ratios, banks must develop a new view of liquidity measurement. Thus, there are a large number of ways. 3. Barth et al. (2004) conducted a survey of national regulatory agencies and obtained information on numerous bank regulations and supervisory practices in 107 countries. However, they conduct pure cross-country regressions because information on regulations and supervisory practices is available only for one point in time. They also indicated that they were able to collect historical data for a few variables, however, and found very little change over time. Moreover, controlling for any changes does not alter their findings. Besides, Barth et al. (2001) describe the survey questions and data collection process in detail. 4 A large regional bank, Southeast Bank, used over 30 liquidity ratios for liquidity measurement. However, it finally failed due to liquidity risk.. 20.

(30) provided to assess bank liquidity risk besides liquidity ratios in recent years.5 Saunders and Cornett (2006) indicated that banks can measure liquidity risk exposure by determining their financing gap. Bank managers often regard the core deposit as stable source of funds, thus it can permanently fund a bank’s loans. Besides, core deposits are lower cost funding sources. The financing gap is defined as the difference between a bank’s average loans and average core deposits:. Financing gap = Average loans - Average core deposits. (3). If the financing gap is positive, the bank must fund it by using its cash, selling liquid assets and borrowing funds in the money market. So the relationship is:. Financing gap = -Liquid assets + Borrowed funds. (4). It can be written as:. Financing gap + Liquid assets = Financing requirement (Borrowed funds). (5). Following Saunders and Cornett (2006), we measure liquidity risk by computing bank’s financing gap. In fact, extrapolating from recent liquidity events, Gatev and Strahan (2006) also indicated that retail liabilities are a more stable source of bank financing than wholesale funds. Besides, Demirguc-Kunt et al. (2006) showed that in contemporary times, aggregate bank deposits do not significantly decline during times of financial distress, especially in developed countries and even in many. 5. See Saunders and Cornett (2006) and Matz and Neu (2007).. 21.

(31) less-developed economies. This suggests that currency is not withdrawn from the banking system in any critical amount in modern economies. Thus, financing gap is defined as the difference between bank’s loans and customer deposits in our study.6 Besides, we divided financing gap by total assets to standardize, finally get the ratio of financing gap to total assets. Besides, we also use net loans to customer and short term funding ratio to measure bank liquidity risk. In our study, liquidity risk is represented by two alternative measures: financing gap ratio and liquidity ratio.. 3.2.2 Causes of Liquidity Risk In this model, we consider liquidity risk measures, bank-specific variables, supervisory and regulatory variables, and macroeconomic conditions. In our study, liquidity risk is represented by two alternative measures: financing gap ratio and liquidity ratio. Following Saunders and Cornett (2006), we measure liquidity risk by computing bank’s financing gap. In our study, it is defined as the difference between a bank’s loans and customer deposits. Besides, we divided financing gap by total assets to standardize, finally get the ratio of financing gap to total assets (FGAPR). Banks with higher financing gap ratio must use its cash, selling liquid assets and much external funding to fund this gap, and face larger liquidity risk. Besides, we also use net loans to customer and short term funding ratio to measure bank liquidity risk. 3.2.2.1 Bank-specific Risk Causes Bank-specific causes of liquidity risk include size, square of size, less risky 6. Saunders and Cornett (2007) indicated that core deposits are generally defined as demand deposits, negotiable order of withdrawal (NOW) accounts, money market deposit accounts (MMDAs), other saving accounts, and retail certificates of deposit (CDs). In our study, we use customer deposit to proxy core deposits.. 22.

(32) liquid assets, risky liquid assets, and external funding dependence. We use natural logarithm of bank’s total assets (SIZE) to proxy size, and their square (SIZE 2) to capture the non-linear relationship. Because of too big to fail argument, large banks would benefit from an implicit guarantee, thus decrease their cost of funding and allows them to invest in riskier assets (Iannotta et al., 2007). So we expect that large banks usually hold more loans and thus have larger financing gap ratio. However, the largest banks will face less liquidity risk because of too big to fail argument. Thus, the effect of size on bank liquidity risk is non-linear. The credit crunch of 2007 reminded many banks of the importance of liquidity risk management. Although liquidity risk may cause bank failures, Davis (2008) indicates that banks can protect against liquidity risk. On the asset side, it can be done by holding a significant proportion of liquid assets. Cash can be used immediately to meet liquidity needs, while government securities can be used readily as collateral. On the liability side, banks should ensure enough diversified funding sources to reduce liquidity risk. Because banks can sell or collateralize its liquid assets to obtain liquid funds, holding liquid assets can reduce bank’s liquidity risk. However, banks may difficult to sell or collateralize their liquid assets because of credit freeze. For this reason, we divided liquid assets into less risky liquid assets and risky liquid assets. Besides, we divided less risky liquid assets and risky liquid assets by total assets separately to standardize, finally get less risky liquid assets to total assets ratio (LRLA) and risky liquid assets to total assets ratio (RLA). Banks can sell their less risky liquid assets such as treasury bills with little price risk and low transaction cost, but they may difficult to collateralize their risky liquid assets like trading securities because of credit freeze to get liquid funds. Thus we expect that LRLA has negative and RLA. 23.

(33) has positive effect on the liquidity risk. We use the ratio of external funding to total liabilities (EFD) to proxy the external funding dependence. External funding are sum of money market funding and other funding. Banks rely on the short-term money market rather than on core deposits to funds loans may face liquidity problem in the future (Saunders and Cornett, 2006). The larger funds they need to borrow in the money markets and the greater liquidity problems from such reliance they will face. Thus we expect that EFD and bank’s liquidity risk have the positive relationship. 3.2.2.2 Supervisory and Regulatory Risk Causes After subprime mortgage crisis we realized that government regulation and supervisory practices are important for banking. We use official supervisory power index (OSP), private monitoring index (PMI), and overall bank activities and ownership restrictiveness (BAR) to proxy government regulation and supervisory practices. The official supervisory power index aggregates information on whether bank supervisors can take specific actions against bank management, bank owners, and bank auditors both in normal times and times of distress, and larger number indicates greater power. Supervisory agencies can use these powers to improve the governance of banks. The private monitoring index includes information on the degree to which bank regulations force banks to disclose accurate information to the public and induces private sector monitoring of banks. Besides, larger number indicates greater regulatory empowerment of private monitoring of banks. The index of overall bank activities restrictions measures the degree to which banks face regulatory restrictions on their activities in securities markets, insurance, real-estate, and owning shares in. 24.

(34) non-financial firms. Besides, larger number indicates higher restrictiveness. In our study, we use interactive terms to examine the effects of supervisory and regulatory variables.7 The interactive terms include the interactions between annual percent change of GDP and official supervisory power index (GDPC×OSP), interactions between annual percent change of GDP and private monitoring index (GDPC×PMI), interactions between annual percent change of GDP and overall bank activities and ownership restrictiveness (GDPC×BAR). However, powerful government will ask their banks to increase liquidity hoard. Banks forced to disclose accurate information to the public will increase their liquidity hoard. Francisco González (2005) indicated that relaxing restrictions on banking activities may encourage bank risk-taking by expanding a bank’s range of activities. In this situation, we expect that strict restrictiveness on bank activities will make them decrease risk-taking and increase liquidity hoard. Yet relaxing restriction may also increase opportunities for bank diversification, and thereby reduce risk-taking. In this situation, strict restrictiveness on bank activities has the opposite effect. 3.2.2.3 Macroeconomic Risk Causes In order to capture the effect of the macroeconomic environment, the two macroeconomic variables used are annual percent change of GDP (GDPC) and annual percent change of inflation (INF). Besides, we further add GDP annual percent change of last year (GDPCt-1) and inflation annual percent change of last year (INFt-1) to capture the lagged effects. Aspachs et al. (2005) indicated that banks hoard. 7. Barth et al. (2004) conducted pure cross-country regressions because information on regulations and supervisory practices is available only for one point in time. In our study, we finish our model estimator by using interactive terms.. 25.

(35) liquidity during periods of economic downturn, when lending opportunities may not be as good and they run down liquidity buffers during economic expansions when lending opportunities may have picked up. Thus we expect that higher economic growth make banks run down their liquidity buffer and induce banks to lend more. However, banks will attract less deposit during economic expansions, consequently increasing their financing gap.. 3.2.3 Determinants of Bank Performance In this model, we consider performance measures, liquidity risk measures, bank-specific variables, market structure variables, supervisory and regulatory variables, and macroeconomic conditions. This study use return on average assets (ROAA), return on average equities (ROAE) and net interest margin (NIM) to evaluate bank performance. ROAA reflects the ability of a bank’s management to generate profits from the bank’s assets. ROAE indicates the return to shareholders on their equity. Average assets and equities are being used in order to capture any differences that occurred in assets and equities during the fiscal year (or season effects). NIM measures the gap between what the bank pays savers and what the bank receives from borrowers. Thus, NIM focuses on the traditional borrowing and lending operations of the bank. 3.2.3.1 Bank-specific Performance Determinants In our study, we use the ratio of financing gap to total assets (FGAPR) to proxy liquidity risk, and we regard liquidity risk as an endogenous determinant of bank performance. Banks with higher financing gap ratio must use its cash, selling liquid assets and much external funding to fund this gap. It consequently increases their cost of funding and reduces profitability. However, Demirgüç-Kunt et al. (2003) indicated. 26.

(36) that banks with high levels of liquid assets in cash and government securities may receive lower interest income than banks with less liquid assets. If the market for deposits is reasonably competitive, then greater liquidity will tend to be negatively associated with interest margin. Thus, the proportion of liquid assets increases will decrease bank liquidity risk, leading to a lower liquidity risk premium of the net interest margin (Angbazo, 1997; Shen et al., 2001; Drakos, 2003). Thus, we expect that FGAPR has negative relationship with ROAA and ROAE and positive relationship with NIM. We also consider another factors affect bank performance besides liquidity risk. Besides, we divide these factors into bank-specific factors, market structure factors, supervisory and regulatory factors, and macroeconomic conditions. Bank-specific determinants of performance include size, square of size, capital, and credit risk. Bank size is generally used to measure economies or diseconomies of scale in the banking industry. The cost differences may cause a positive relationship between size and bank performance, if there are significant economies of scale (Bourke, 1989; Molyneux and Thornton, 1992; Goddard et al., 2004). In addition, as Short (1979) argues, size is closely related to the capital adequacy of a bank since relatively large banks tend to raise less expensive capital and, hence, appear more profitable. In previous studies, some studies have found scale economies for large banks (e.g. Berger and Humphrey, 1997; Altunbaş et al., 2001; Athanasoglou et al., 2006; Kosmidou, 2008) while others have found diseconomies for larger banks (e.g. Kosmidou et al., 2005; Pasiouras and Kosmidou, 2007). However, Eichengreen and Gibson (2001) indicated that the effect of a growing bank’s size on profitability may be positive up to a certain limit. Beyond this point the effect of size could be negative due to bureaucratic. Thus, the relationship may be expected to be non-linear. As. 27.

(37) previous studies, we use natural logarithm of bank’s total assets (SIZE) to proxy size, and their square (SIZE2) to capture the non-linear relationship. We use the ratio of equity to assets (ETA) to proxy the capital strength. Banks with high capital-asset ratios are considered relatively safer in the event of loss or liquidation. Besides, increase in capital may raise expected earnings by reducing the expected costs of financial distress (Berger, 1995). The lower risk increases banks creditworthiness and consequently reduces the cost of funding. Previous studies that use capital ratios as an explanatory variable of bank profitability found a positive relationship (e.g. Demirgüç-Kunt and Huizinga, 1999; Barth et al., 2003; Kosmidou et al., 2005). Thus, banks with higher equity to assets ratio will have lower needs of external funding and therefore higher profitability. The loan loss provisions to loans ratio (LLPL) is used to proxy the credit risk. Changes in credit risk may reflect changes in the health of the bank’s loan portfolio (Cooper et al., 2003), which may affect bank performance. Besides, Miller and Noulas (1997) indicated that the more financial institutions are exposed to high-risk loans, the higher the accumulation of unpaid loans and the lower the profitability. However, riskier loans should produce higher interest income. Maudos and Fernández de Guevara (2004) indicated that the risk of non-repayment or default on a credit (credit risk) requires the bank to apply a risk premium implicitly in the interest rates charged for the operation. Banks that assume greater credit risk present higher interest margins. Doliente (2005) also indicated that the impact of credit risk may reflect the additional risk premium charged by banks for the financial costs of forgone interest revenue. Thus, we expect that LLPL has negative relationship with ROAA and ROAE and positive relationship with NIM.. 28.

(38) 3.2.3.2 Market Structure Performance Determinants Regarding market structure variables, we use three-bank concentration ratio (CON), CR3. CON is calculated as the total assets held by the three largest commercial banks divided by the total assets of all commercial banks in each country. The higher the value is, the lesser competition they have. According to the structure-conduct-performance (SCP) hypothesis, banks in highly concentrated markets tend to collude and thus earn monopoly profits (Short, 1979; Gilbert, 1984; Molyneux et al., 1996).8 3.2.3.3 Supervisory and Regulatory Performance Determinants We use official supervisory power index (OSP), private monitoring index (PMI), and overall bank activities and ownership restrictiveness (BAR) to proxy government regulation and supervisory practices. In our study, we use interactive terms to examine the effects of supervisory and regulatory variables. The interactive terms include the interactions between annual percent change of GDP and official supervisory power index (GDPC×OSP), interactions between annual percent change of GDP and private monitoring index (GDPC×PMI), interactions between annual percent change of GDP and overall bank activities and ownership restrictiveness (GDPC×BAR). Barth et al. (2004) indicated that strong supervision can help prevent banks from engaging in excessive risk-taking behavior and thus improve bank development, performance and stability. However, powerful supervisors may use their powers to benefit favored constituents, attract campaign donations, and extract bribes (Djankov. 8. Previous studies indicated that collusion may cause higher interest rates spread (higher interest rates being charged on loans and less interest rates being paid on deposits) and higher fees being charged (e.g. Goldberg and Rai, 1996 ; Goddard et al., 2001).. 29.

(39) et al., 2002; Quintyn and Taylor, 2002). Under these circumstances, powerful supervision will be positively related to corruption and will not improve bank development, performance and stability. Barth et al. (2003) indicated that it is possible that the wider the range of activities the greater will be profit opportunities for banks. However, banks may systematically fail to manage well a diverse set of financial activities beyond traditional banking, and hence profitability would be lower. 3.2.3.4 Macroeconomic Performance Determinants In order to capture the effect of the macroeconomic environment, the two macroeconomic variables used are annual percent change of GDP (GDPC) and annual percent change of inflation (INF). Besides, we further add GDP annual percent change of last year (GDPCt-1) and inflation annual percent change of last year (INFt-1) to capture the lagged effects. GDP is a measure of total economic activity within an economy. Higher economic growth encourages banks to lend more and permits them to charge higher margins, and improving the quality of their assets. Previous studies found that economic growth has positive effect on bank’s performance (e.g. Kosmidou et al., 2005; Pasiouras and Kosmidou, 2007; Athanasoglou et al., 2008; Kosmidou, 2008). Thus, GDPC and GDPCt-1 are expected to have a positive impact on bank performance. The relationship between inflation and performance is ambiguous. Perry (1992) indicated that the relationship between inflation and performance depends on whether inflation expectations are fully anticipated. An inflation rate fully anticipated by the bank’s management implies that banks can appropriately adjust interest rates to increase their revenues faster than their costs and thus acquire higher economic profits.. 30.

(40) If inflation is unanticipated, banks may be slow in adjusting their interest rates. It results in a faster increase of bank costs than bank revenues that consequently has a negative impact on bank profitability. Most studies found a positive relationship between inflation and bank profitability (e.g. Bourke, 1989; Molyneux and Thornton, 1992; Kosmidou et al., 2005; Athanasoglou et al., 2006; Pasiouras and Kosmidou, 2007; Athanasoglou et al., 2008). However, Kosmidou (2008) found a negative relationship. Besides, Huybens and Smith (1999) develop a theoretical model in which interest margins tend to rise in the presence of inflation. Empirical studies found that inflation has positive effect on bank’s NIM (e.g. Demirgüç-Kunt and Huizinga, 1999; Kosmidou et al., 2005).. 3.3 Econometric Specification First, we use panel unit root tests to investigate the stationary of the data. Then we will discuss two major models in our study and their estimator method respectively. One is the causes of liquidity risk model and the other is the bank liquidity risk and performance model.. 3.3.1 Panel Unit Root Tests In this section, we use panel unit root tests to investigate the stationary of the data. Since the appearance of the papers by Levin and Lin (1992, 1993), the use of panel unit root tests has become very popular among empirical researchers with access to a panel data set. Besides, recent literature also suggests that panel-based unit root test have higher power than unit root test based on individual time series (e.g. Levin et al. 2002; Im et al. 2003). The most common tests in practice are Levin et al. (2002), Im et al. (2003) and Maddala and Wu (1999). Thus our study focuses on the. 31.

(41) three tests. All tests are based on the following Augmented Dickey-Fuller regression:. Δyit = αi + δi t + ρi yi,t-1 + εit ,. i = 1, …, N, t = 1, …, T. (6). where αi and δi allow for fixed and unit specific time trends for each i. εit is error term and εit ~ IID(0,ζ2). The null hypothesis of unit root is:. H0: ρi = 0, ∀i. (7). Thus under the null hypothesis, all series in the panel are nonstationary processes, and the alternative hypothesis is different for each test. Levin et al. (2002) test may be viewed as a pooled Augmented Dickey-Fuller test. Besides, the test considered an alternative hypothesis where the autoregressive coefficient is homogeneous across groups:. HA: ρi = ρ < 0, ∀i. (8). Im et al. (2003) extends the Levin et al. (2002) framework to allow heterogeneity in the value of the autoregressive coefficient. Under the alternative, parts of the series in the panel are assumed to be stationary. This is in contrast to the Levin et al. test, which presumes that all series are stationary under the alternative hypothesis. In this test, it has individual i =1, …, N1 where yit is stationary and individual i = N1+1, N1+2, …, N where yit is nonstationary. The alternative hypothesis is specified as:. HA: ρi < 0, ∀i = 1, …, N1 and ρi = 0, ∀i = N1+1, N1+2, …, N 32. (9).

(42) The formulation of the alternative hypothesis allows ρi to differ across groups, and is more general than the homogeneous alternative hypothesis. Im et al. (2003) proposed a t-bar test statistic. The t-bar statistic is averaged the Augmented Dickey-Fuller statistics and can be shown as:. 1 t = N. N. ∑t. (10). i. i =1. where ti is the individual ADF t-statistics for the unit root test. However, the test assumes that T is the same for all cross-section units. So the test is applied only for balanced panel data. Maddala and Wu (1999), relying on Fischer (1932), proposed the Fisher Pλ test. The test suggests combining the p-value of the test statistic for unit root in each cross-sectional unit. Besides, the alternative hypothesis is same as the Im et al. test. Unlike the Im et al. test, Fisher Pλ test does not require balanced panel data, so T can differ over cross-sections. The statistic of Fisher Pλ test is given by:. N. Pλ = －2∑ln( pi ) ~ χ 2 (2 N ). (11). i =1. where pi is the p-value of the test statistic of the ADF test in each group.. 3.3.2 Causes of Liquidity Risk Model This model provides an economic analysis of the causes of liquidity risk. Besides, we divide the causes of liquidity risk into internal and external factors. In order to examine the relationship between liquidity risk and the bank-specific, supervisory and macroeconomic variables, the panel fixed effect regression model has 33.

(43) been developed:. B. S. M. Lit = ci + ∑λb Π +∑δs Π +∑γm Π mjt +εit b it. b =1. s jt. s =1. (12). m =1. where Lit is liquidity risk of ith bank at time t, with i = 1,…, N, t = 1,…,T. In our study, it is the financing gap ratio (FGAPR) and the ratio of net loans to customer and short term funding (NLCS).. Π bit , Π sjt , Π mjt are bank-specific, supervisory and. macroeconomic variables with b = 1,…, B, s = 1,…, S, m = 1,…, M, respectively. j refers to the country in which bank i operates; c is a constant term; εit is the error term.. Extending equation (12) to reflect the variables, as described in Table 3, the model is formulated as follows:. Lit = ci + λ1 SIZEit + λ2 SIZEit2 + λ3 LRLAit + λ4 RLAit + λ5 EFDit +δ1GDPC jt × OSPjt + δ2 GDPC jt × PMI jt + δ3GDPC jt × BAR jt +γ1GDPC jt + γ 2 GDPC jt-1 + γ3 INF jt + γ 4 INF jt-1 + ε it. (13). Bank-specific variables include size (SIZE), square of size (SIZE2), less risky liquid assets (LRLA), risky liquid assets (RLA) and external funding dependence (EFD). Supervisory and regulatory variables include the interactions between change of GDP and official supervisory power index (GDPC×OSP), interactions between change of GDP and private monitoring index (GDPC×PMI), interactions between change of GDP and overall bank activities and ownership restrictiveness (GDPC×BAR). Macroeconomic variables include change of GDP (GDPC), GDP change of last year (GDPCt-1), change of inflation (INF) and inflation change of last. 34.

(44) year (INFt-1). Equation (13) is estimated through fixed effects regression taking each bank’s FGAPR as the dependent variable. We have been tested with the Hausman test and reject the null hypothesis of random effects is suitable model. Thus, we use fixed effects rather than random effects model.. 3.3.3 Bank Liquidity Risk and Performance Model This model provides an economic analysis of the relationship between bank liquidity risk and performance. Besides, from the causes of liquidity risk model, we can realize that there are many factors may affect bank liquidity risk. In our study, we thus regard liquidity risk as an endogenous determinant of bank performance, and apply panel data instrumental variables regression to estimate this model.9 In the previous studies, determinants of bank profitability were usually divided into internal and external determinants. In order to examine the relationship between bank liquidity risk and performance, the panel instrumental variables regression model has been developed:. B. Pit = c + βl Lit +. K. ∑. θb Χ bit. +. b=1. S. ∑. ωk Χ kjt. +. j =1. ∑. νs Χ sjt. s =1. M. +. ∑η. m m Χ jt. +εit. (14). m=1. the reduced form equation for Lit is. B. Lit = c +. ∑. λb Π bit. b=1. S. +. ∑. δs Π sjt. s =1. M. +. ∑γ Π m. m jt. +εit. (15). m=1. 9. The ordinary least squares estimator will cause biased. However, instrumental variables regression provides a way to obtain consistent parameter estimates (Dunning, 2008).. 35.