Regulatory Capital and Bank Lending: The Role of Credit Default Swaps *

Transcription

1 Regulatory Capital and Bank Lending: The Role of Credit Default Swaps * Susan Chenyu Shan Shanghai Advanced Institute of Finance, SJTU cyshan@saif.sjtu.edu.cn Dragon Yongjun Tang The University of Hong Kong yjtang@hku.hk Hong Yan Shanghai Advanced Institute of Finance, SJTU hyan@saif.sjtu.edu.cn November 5, 2014 * We thank Viral Acharya, Tim Adam, Edward Altman, Thorsten Beck, Allen Berger, Chun Chang, Jaewon Choi, Greg Duffee, Phil Dybvig, Lijing Du, Rohan Ganduri, Todd Gormley, John Griffin, Jean Helwege, Paul Hsu, Grace Hu, Victoria Ivashina, Dimitrios Kavvathas, Dan Li, Feng Li, Jay Li, Chen Lin, Tse-Chun Lin, Jun Liu, Christian Lundblad, Spencer Martin, Ronald Masulis, Ernst Maug, Greg Niehaus, Neil Pearson, Francisco Pérez-González, QJ Jun Qian, Stephen Schaefer, Philipp Schnabl, Sascha Steffen, Philip Strahan, RenéStulz, Sheridan Titman, Cong Wang, Tan Wang, Yihui Wang, John Wei, Andrew Winton, Deming Wu, Yu Yuan, Haoxiang Zhu, and seminar participants at the University of Hong Kong, Australian National University, University of Melbourne, Institute for Financial Studies of Southwestern University of Finance and Economics, Shanghai Advanced Institute of Finance, Central University of Finance and Economics, Renmin University of China, University of South Carolina, Zhejiang University, Chinese University of Hong Kong, Wuhan University, Shanghai University of Finance and Economics, George Mason University, the Office of the Comptroller of the Currency (OCC), the 2014 NUS RMI Symposium on Credit Risk, the 2014 Fixed Income Conference, the 2014 Conference on Financial Markets and Corporate Governance, the 2014 CICF, the 2014 C.R.E.D.I.T Conference, the 2014 FMA, and the 2014 TCFA Best Paper Symposium for comments and suggestions. We acknowledge the support of the National Science Foundation of China (project # ).

2 Regulatory Capital and Bank Lending: The Role of Credit Default Swaps Abstract We link regulatory capital requirements to bank lending by examining the role of credit default swaps (CDS), which can be used for capital relief. Although CDS-using banks appear to be as well-capitalized as non-cds-using banks, we document that the regulatory capital ratios of CDS-using banks are effectively lower when the selection of CDS usage is accounted for by instrumentation. These deteriorated capital positions result partly from increased risky lending because CDS-using banks issue more and larger loans, particularly to borrowers whose debt is referenced in CDS contracts. Moreover, banks that actively used CDS at the onset of the credit crisis raised their capital and reduced lending to a greater extent during the crisis than non-cds-using banks. Overall, our findings suggest that banks that take advantage of CDS-based regulatory capital relief provide more credit to risky borrowers and are more procyclical.

3 1. Introduction Capital plays a pivotal role in banking and bank regulations. Regulatory requirements for capital adequacy are the subject of a current debate focused on their effects on financial stability and on the supply of credit to the economy. 1 However, identifying the causal effect of regulatory capital on bank lending is a difficult empirical task because observed bank capital levels are typically above regulatory minimums. In addition, the implications of bank capital management for banks risk-taking behavior remain unclear. In this paper, we investigate the role of credit default swaps (CDS) in affecting banks capital ratios and lending practice, as CDS-using banks take advantage of the capital relief allowed in the capital regulation with the usage of CDS to hedge credit risk exposures. The CDS market has become globally important with trillions of dollars in notional value outstanding, but the role of CDS in banking is controversial. On the one hand, CDS may be regarded as innovative financial derivatives providing a new or alternative venue through which banks can transfer credit risk (Acharya and Johnson, 2007; Oehmke and Zawadowski, 2014). On the other hand, CDS can create perverse incentives that induce banks to become empty creditors (Bolton and Oehmke, 2011). For the most part, prior research studies the impact of CDS on borrowing firms (e.g., Saretto and Tookes, 2013; Subrahmanyam, Tang and Wang, 2014). Our study instead focuses on the role of CDS in banks capital management and examines both banklevel and loan-level evidence to illustrate the link between regulatory capital and bank lending. Holding capital levels above their regulatory minimums is useful for banks seeking to attract business in a competitive banking industry (Allen, Carletti, and Marquez, 2011), to increase bank value (Mehran and Thakor, 2011), and to survive a banking crisis (Berger and 1 For instance, while Admati and Hellwig (2013) advocate for stronger capital regulations, DeAngelo and Stulz (2014) argue that high leverage is optimal for banks. Thakor (2014) reviews the related issues and notes that the socially efficient capital level may exceed individual banks privately optimal levels. 1

4 Bouwman, 2013). Bank managers may prefer higher capital ratios to avoid regulatory scrutiny, but holding too much capital may not be in the best interest of bank shareholders who want to maximize their return on capital. Bank regulations, including the Basel II Capital Accord, allow banks to apply a lower risk weight and thus effectively hold less capital against their assets when they use CDS to hedge credit risk. 2 However, banks face impediments in employing CDS to hedge individual loan exposures (Minton, Stulz, and Williamson, 2009); hence banks make a deliberate decision whether to use CDS for capital relief. Using data on U.S. banks over the period, we find that the capital ratios of CDS-using banks and non-cds-using banks are indistinguishable at first glance. However, by recognizing CDS usage as a bank s endogenous choice and using instrumental variables to address the endogeneity issue, we demonstrate that CDS-using banks have lower effective capital ratios than their non-cds-using peers. This result is similar for the risk-weighted Tier 1 capital ratio and for the Tier 1 leverage ratio. The finding that instrumentation unveils the significant relation between CDS usage and regulatory capital is consistent with the anticipation effect that poorly capitalized banks tend to use CDS to boost their capital ratios, following the logic highlighted in Edmans, Goldstein, and Jiang (2012). Moreover, we document that the quality of capital, as measured by the ratio of Tier 1 capital to total capital, is lower for CDSusing banks. This finding likely ensues because Tier 3 capital, as opposed to higher quality Tier 1 capital, can be used to cover the market-risk capital required for CDS trading positions. Whereas CDS-using banks can maintain the appearance of adequate capital ratios, the effective leverage ratios of these banks are increased, which leads to an incentive to increase lending (Inderst and Mueller, 2006). Increased risky lending is also predicted by the hedge- 2 This possibly reflects the belief of bank regulators in the risk management benefits of CDS. As the former chairman of the Federal Reserve Board Alan Greenspan stated, CDS facilitate the development of a far more flexible, efficient, and hence resilient financial system (Greenspan 2004). 2

5 more/bet-more result of Simsek (2013) because CDS allow a pure play of credit risk with low collateral requirements (Shen, Yan, and Zhang, 2014). Indeed, we find that, ceteris paribus, CDS-using banks extend more commercial and industrial (C&I) loans than non-cds-using banks. C&I loans, as opposed to mortgage loans, are more often kept on the banking books instead of being securitized. After establishing that CDS-using banks tend to have lower (real) capital ratios and make more loans than non-cds-using banks, we examine the characteristics of loans made at the individual loan level. Using data for syndicated loans, we find that the average size of a loan (relative to the borrower s assets) is larger when the lead bank is an active user of CDS and the borrower has outstanding CDS contracts referencing its name. However, these loans are of lower quality at origination than loans made by a lead bank not engaged in the CDS market. Moreover, it appears that CDS-using banks are aware of the risk inherent in these loans because CDS-using banks set aside larger loan loss provisions than their non-cds-using counterparts. Furthermore, given that loss provisions can be added back into Tier 2 capital, a larger loss provision can thus raise the total capital ratio. This result helps us understand the appearance of comparable levels of capital ratios for CDS-using and non-cds-using banks and the lower capital quality of CDSusing banks. The CDS market was significantly disrupted by the bankruptcy of Lehman Brothers and the collapse of AIG that led to a global financial crisis in We find that banks that used CDS to manage capital ratios before the crisis were negatively impacted during the crisis. These banks were forced to raise additional capital and to restrict their lending more than their non- CDS-using peers during the crisis period. We also document that while CDS-using banks enjoyed larger gains in stock prices before the crisis, they suffered larger declines in their stock 3

6 prices during the crisis than non-cds-using banks. This finding sheds light on the motives for banks CDS usage: to maintain lower effective capital levels and achieve higher returns on capital, which are both beneficial for shareholders. However, when disasters strike, these banks will suffer and may produce negative externalities in the financial system that damage depositors, borrowers, and taxpayers. Our empirical analysis substantiates the theoretical model of Bolton and Oehmke (2011), which focuses on a specific effect of CDS for creditors, and helps us better understand the link between bank capital and lending behavior. These findings are consistent with the theoretical argument of Thakor (1996), who examines the effect of the risk-based capital requirement ( Basel I ) on bank lending and finds that relaxing capital requirements can lead to more bank lending. Our study indicates that banks engage in riskier lending to enhance their profitability when aided by the regulatory capital relief from CDS usage and in erstwhile compliance with regulatory requirements. Bank leverage and associated vulnerability, are thus masked by the reading of satisfactory risk-weighted capital ratios. This regulatory allowance appears to have made these banks more procyclical, which is counter to regulatory objectives and possibly contributed to and exasperated the credit crisis of Although bank usage of CDS for trading purposes is addressed by the Dodd-Frank Act (i.e., Section 619, the Volcker Rule ), Basel III still allows banks to use CDS for capital relief. Our results illustrate an important consequence of the role of CDS in capital regulation. Our findings demonstrate a real effect of CDS through the angle of capital relief in the banking context. Prior studies have focused on the hedging role of CDS for borrowers and 3 Regulatory failures concerning CDS and other financial securities have been regarded as a major cause of the financial crisis in 2008 (e.g., page 18 of the final report of the Financial Crisis Inquiry Commission). Some prominent figures such as Rajan (2005) expressed concerns about the potential risks of credit derivatives, such as CDS, to the financial system before the crisis. 4

7 elaborated on the effects of CDS usage on borrowing firms cost of debt (Ashcraft and Santos, 2009), credit supply availability (Saretto and Tookes, 2013), and bankruptcy risk (Subrahmanyam, Tang, and Wang, 2014). The results in this study suggest that active engagement in the CDS market allows banks to hold less capital and assume greater risk. Although this finding is contrary to the perceived role of CDS in managing banks credit risk exposure, it is consistent with the implications of a theoretical model developed by Yorulmaezer (2013), which predicts that banks take excessive risk when they utilize capital relief tied to CDS. Our study therefore provides a new perspective on bank risk taking (e.g., Stulz, 2014) and addresses the question raised by Levine (2012) that asked why the Fed did not prohibit banks from reducing regulatory capital via CDS : banks did what regulators allowed and achieved satisfactory regulatory indicators that masked their real vulnerability to shocks. The rest of this paper proceeds as follows: Section 2 provides the motivational background and places our study in a relevant context. Section 3 describes our datasets and sampling procedure. Section 4 lays out the empirical results on the effect of banks CDS usage on their regulatory capital ratios. Section 5 presents both bank-level and loan-level evidence on the link between CDS usage and corporate lending. Section 6 discusses how CDS usage affects banks performance during the 2008 credit crisis. Section 7 concludes. 2. Background The incorporation of capital relief induced by credit derivatives is an important development in bank capital regulation. After completing the first CDS deal in 1994, JPMorgan soon communicated with receptive U.S. regulators about allowing banks to reduce their capital reserves by hedging lending-based credit risk exposure through CDS protection at a time when 5

8 U.S. bank regulators were calling for revisions to the 1988 Basel capital accord. In August 1996, the Federal Reserve Board issued a statement suggesting that banks should be allowed to reduce capital reserves by using credit derivatives. 4 In June 1997, the Federal Reserve Board released a document providing guidance on how credit derivatives held in the trading account should be treated under the market risk capital requirement that was approved by the Basel Committee a year earlier. 5 In December 1997, JPMorgan marketed the Broad Index Secured Trust Offering ( Bistro ), a synthetic collateralized loan obligation (CLO) structured in three tranches. When JP Morgan failed to move the super senior tranche it kept on its trading book, it received permission from the Federal Reserve in early 1998 to use a much lower risk weight on the security that remained on its banking books protected by CDS. 6 Other banks followed suit. Meanwhile, the International Swaps and Derivatives Association (ISDA) also advocated for credit derivatives to be included in capital regulations in a March 1998 white paper, entitled Credit Risk and Regulatory Capital. Consequently, in June 1999, with the confluence of US regulatory actions, it was proposed that credit derivatives be counted as credit exposure hedges that were similar to guarantees either in full or in part in the first Basel II consultative paper, which recognized that the recent development of credit risk mitigations such as credit derivatives has enabled banks to substantially improve their risk management. 7 The proposal eventually became a part of Basel II that was approved in Supervisory Guidance for Credit Derivatives, Division of Banking Supervision and Regulation, Board of Governors of the Federal Reserve System, August 12, 1996: For a detailed historical account of how credit derivatives became part of bank capital regulations, see Tett (2009). 5 Application of Market Risk Capital Requirements to Credit Derivatives, June 13, 1997: 6 The Fed indicated that such transactions allow economic capital to be more efficiently allocated, resulting in, among other things, improved shareholder returns. See Capital Treatment for Synthetic Collateralized Loan Obligations, November 17, 1999: 7 See 6

9 Basel II is rather flexible in recognizing CDS as a hedge for banks. For example, a mismatch between the underlying obligation and the reference obligation under CDS is permissible if the reference obligation is junior to the underlying obligation. In other words, bond CDS can be counted as a hedge for loan risk. Basel II also allows a maturity mismatch and partial hedging (for credit event definitions and coverage). The role of CDS in bank capital regulation is maintained in Basel III, albeit with certain modifications. Banks usage of CDS for capital relief is indirectly confirmed by statements from protection sellers. For example, AIG disclosed in its 2007 annual report that 72% of the CDS protection it sold was used by banks for capital relief. 8 It is necessary for protection sellers to make such claims for credit derivatives to be counted for capital relief for protection buyers. However, CDS are not regulated as insurance policies under the U.S. Commodity Futures Modernization Act of Therefore, although banks can obtain capital relief using CDS contracts, sellers of these contracts are not required to hold additional capital to provide the protection because they are typically nonbank financial institutions, such as insurance companies (e.g., AIG), and are thus outside of bank regulators reach. The historical development of capital rules leading up to Basel II suggests that capital relief can be an important reason for banks to take CDS positions. 9 Theoretical models, including that of Parlour and Winton (2013), suggest that the cost of holding capital can be a motive for banks to use CDS to transfer credit risk. Because banks facing a higher cost of capital tend to use CDS to boost capital ratios, the endogenous selection One recent striking example of banks active management of capital ratios using CDS is demonstrated by JPMorgan s London Whale trading incident in early See, e.g., JP Morgan and the CRM: How Basel 2.5 beached the London Whale, by Michael Watt, Risk Magazine, October 5,

10 of banks using CDS may depend on other activities in which these banks are engaged, and these activities may result in deteriorating capital positions and worsening capital quality. Figure 1 illustrates the expected structural changes in a bank s on and off balance sheet items after it begins to trade CDS. The bank s on and off balance sheet activities may both increase, as CDSs may expand both trading and banking assets and facilitate securitization that results in off balance sheet activities. The component on the right-hand side of the balance sheet affected by CDS is the core capital ratio. U.S. bank regulations consist of three main elements: the Tier 1 risk-weighted capital ratio, the total risk-weighted capital ratio, and the Tier 1 unweighted capital ratio (the leverage ratio ). The leverage ratio is based on adjusted non-riskweighted assets on the balance sheet and does not include off balance sheet assets. 10 CDS usage affects capital ratios through both the denominator and the numerator. It impacts the denominator by changing risk weights and by modifying the quantity of credit risk as well as the positions of market risk, as CDS facilitate the process of moving assets off the balance sheet without risk transfer, effectively increasing bank size without increasing regulatory capital. 11 CDS also affect the numerator the total capital amount because Tier 3 capital can be counted against the market risk of CDS trading positions. Moreover, loan loss provisions, possibly associated with CDS-related loans, can be counted as Tier 2 capital. One implication of this illustration is that under seemingly identical capital ratios, banks may have completely changed their capital quality and risk components in their loan portfolios by using CDS. Capital ratios aided by CDS may not be equivalent to true capital adequacy especially if counterparty risk ratings of protection sellers prove to be unreliable, as in the case of 10 Leverage ratio is not in the Basel Capital Accord; therefore, it falls into the category of supervisory review. 11 The denominator for the risk-based capital ratio also includes the credit equivalent amount of off-balance sheet items. CDSs may help banks move assets off the balance sheet and obtain a lower credit equivalent amount. 8

11 AIG. Moreover, the quality of bank capital may also be affected because CDS trading may generate substantial market risk exposure, which may be covered by Tier 3 capital. Given this regulatory background, it is puzzling that Minton, Stulz, and Williamson (2009) find little CDS usage by banks for hedging purposes in an earlier sample. In addition to the difference in time periods and data sources, our reconciliation is as follows. First, banks tend to use basket or index CDS to satisfy capital requirements more effectively. These banks may first securitize loans to generate CDO tranches and then buy CDS by referencing the pool of loans to obtain capital relief because they may have to retain those tranches or provide implicit guarantees to outside investors on those tranches, as demonstrated by Acharya, Schnabl and Suarez (2013). However, bank statements at the time may only contain single-name CDS on individual loans. Second, banks have no incentive to publicize their usage of CDS for capital relief because such information may lead to negative perceptions of their capital adequacy. Whether a higher capital level benefits or hurts bank lending remains an unresolved issue in the literature. From the monitoring incentive perspective, higher capital ensures skin in the game and thus improves banks incentives to hold efficient loan portfolios. Hence, higher capital is associated with more lending (Mehran and Thakor, 2011). The alternative view is that higher leverage leads to more discipline and thus more lending (Calomiris and Khan, 1991; Diamond and Rajan, 2002). Although theories conflict regarding the causal effect of bank capital on lending, the important role of CDS in bank capital management affords us a unique perspective on this link. In this paper, we illustrate a new channel for increased credit supply by banks after their CDS usage. We argue that it is banks CDS usage that drives the increase in credit supply because capital requirements are easier to meet when CDS are in place. This channel is different 9

12 from the hedging channel studied by Saretto and Tookes (2013) because, as we will demonstrate, when the lending bank does not use CDS, its credit supply to a firm is not affected whether there are CDS contracts available in the firm s name. Hence, the loan-level analysis we conduct in this paper helps us highlight the unique and important role of CDS-using banks play in driving the CDS effect on borrowing firms which is the focus of other studies because nonbanks such as bond investors and hedge funds can also trade CDS. 3. Data and Sample Description We employ three main datasets on banks, syndicated loans, and corporate borrowers. The first dataset includes bank CDS positions, regulatory capital ratios and other characteristic variables, and stock prices for publicly listed banks. The second dataset contains information on individual syndicated corporate loans with loan contract terms at origination, including loan size, interest rate, maturity, and lender identities. The third dataset provides CDS transaction information for individual U.S. publicly listed corporate borrowers Bank CDS Positions Our primary source of bank CDS position data for the period is the Federal Reserve Consolidated Financial Statements for Holding Companies ( FR Y-9C ). 12 Banks with more than $150 million in assets are required to file FR Y-9Cs (the threshold increased to $500 million in 2006). For the main analysis, we focus on banks that have acted as syndicate lead arrangers in the Loan Pricing Corporation s Dealscan database. We manually match an RSSD ID in the bank dataset to the name of a lead lender in Dealscan to identify the list of lending banks 12 Our sample does not include thrifts, which are regulated differently from bank holding companies in the U.S. 10

13 that use CDS in a given quarter. We refer to a field in Dealscan called Lead Arranger Credit, which takes values of either Yes or No for every bank, to identify syndicate lead arrangers. We ensure that the match is made in the same year to account for bank name changes. Finally, we restrict the sample to the period because Dealscan only began providing relatively complete loan information in 1994 and because our borrower CDS transaction dataset ends in 2009, when a substantial change occurred in the CDS market. FR Y-9C filers include 7,646 banks, and 121 banks act as syndicate lead lenders in Dealscan. CDS position data for foreign banks are not available from FR Y-9C filings. We collect additional bank CDS position data from the Quarterly Report on Bank Derivatives prepared by the Office of the Comptroller of the Currency (OCC) that includes U.S. subsidiaries of large foreign banks. The OCC reports list the top banks with the largest credit derivative positions every quarter beginning in 1998 whether the bank is domiciled in the U.S. Both the FR Y-9C filings and the OCC reports provide aggregate CDS positions and positions held by banks as beneficiaries ( bought ) or guarantors ( sold ). We crosscheck the CDS position data covered by the two datasets and find that they are consistent with one another. Based on the quarterly CDS positions held by banks reported in the FR Y-9C and OCC reports, we define banks that have a nonzero CDS position in a given quarter, either a long position or a short position, as CDSusing banks. 13 Banks with no CDS positions are denoted as non-cds-using banks. To assure consistency between our bank-level and loan-level analyses, we restrict our sample banks to Dealscan syndicate lead lenders, which can be matched with bank identifiers in Compustat. Other bank-level control variables are extracted from the Compustat and FR Y-9C 13 The banks act as the beneficiary for long positions, which are specified by the variable BHCKC969 in the FR Y- 9C report and the CDS bought column in the OCC report. The banks act as the guarantor for the short positions, which are specified by the variable BHCKC968 in the FR Y-9C report and the CDS sold column in the OCC report. 11

14 reports. Our base sample includes 84 banks with complete financial information 14, 43 of which took nonzero CDS positions at some point during the sample period Corporate Loans At the loan level, we are interested in the effects of CDS trading on the size of the loans issued by firms whose debt is referenced in CDS. We sum the loan amount, take a simple average of the all-in-drawn spread and maturity to aggregate tranches (also called facilities) from the same loan deals (also called packages), and conduct our analysis at the deal level. We use other deal-level information in Dealscan, including the security of the issue, loan type, loan purpose, and the number of syndicate lenders as control variables. We merge Compustat/CRSP with Dealscan loan records by using borrower identifiers in Compustat to obtain borrowing firms financial data. 15 This matching procedure produces a dataset of 67,747 loan deals during the period. Of those, 47,247 are syndicated loans (i.e., distribution method is syndication ), and the rest consist of club deals and sole-lender loans. In our multivariate analysis, we exclude firms with missing loan characteristics (such as loan amount, spread, maturity, security, loan type, loan purpose, and lender information) and those with missing firm financial data in the quarter prior to loan initiation (such as total assets, cash-to-total assets ratio, book leverage, market-to-book ratio, sales-to-total assets ratio, tangible assets, and Altman s Z-score). Our base regression sample thus contains 15,546 syndicated loans. In robustness checks, we also use the combined sample of syndicated loans and sole-lender loans, totaling 17,268 observations. 14 By restricting the sample banks to those acting as syndicate leaders, we confine our base sample to relatively large banks because CDS-using banks are typically large and may not be comparable to small banks. 15 We appreciate the Dealscan-Compustat link file provided by Chava and Roberts (2008). 12

15 3.3. CDS Transactions Referencing Individual Borrowing Firms We determine whether CDS contracts referencing the borrowers debt exist at the time of loan issuance by using two major datasets on the sources of CDS transactions: CreditTrade and GFI Group. The CreditTrade data cover the period from June 1997 to March 2006; the GFI data cover the period from January 2002 to April The overlap of the two datasets allows us to perform a crosscheck to ensure data accuracy. We further validate the data by using Markit quotes. Following Subrahmanyam, Tang, and Wang (2014), we use the first CDS transaction record for the issuer appearing in the data as the CDS introduction date. We identify 921 U.S. firms with debt referenced in CDS trades and quotes from June 1997 to April 2009, which accounts for 8.1% of the total number of unique borrowers during the same period. We include all borrowing firms in our sample, whether they are large or small, whereas Saretto and Tookes (2013) restrict their sample to S&P 500 firms. Among the 47,247 Dealscan syndicated loans, 9,341 of them are made to 867 firms that have CDS referencing their debt at some time during the sample period ( CDS firm ), and 6,641 loans are made to firms with CDS trading at the time of loan origination ( CDS trading ) Overview of the Sample Our base sample primarily consists of large banks that are required to file quarterly reports with the Federal Financial Institutions Examination Council. This is expected because lead arrangers of syndicated loans are typically large banks, as the average book asset of our sample banks is $331.6 billion. The mean total risk-weighted capital ratio (Tier 1, Tier 2, and Tier 3 capital combined and divided by total risk-weighted assets), Tier 1 risk-weighted capital ratio, and Tier 1 leverage ratio (Tier 1 capital divided by non-risk-weighted total assets less intangible 13

16 assets) are 13.2%, 10.2% and 8.2%, respectively, which are all higher than regulatory minimums. 16 In addition to specifying the minimum risk-weighted capital ratios complying with the Basel capital accords, the US banking regulators also set minimum non-risk-weighted Tier 1 capital leverage. 17 Among all types of loans, including commercial and industrial loans (C&I loans), home mortgages, farm loans, and consumer loans, among others, C&I loans account for the largest percentage 19.8% for the average bank. Other bank characteristics are comparable to those reported in Loutskina (2011). Panel B of Table I presents the year-by-year summary of the bank sample. The first instance of a bank reporting CDS positions occurred in 1997 after the August 1996 regulatory announcement by the Federal Reserve Board and the OCC. Banks enter and exit the CDS market over time. The maximum number of CDS-using banks in any given quarter in our sample is 20. The size of bank total assets grew steadily during the sample period. The total amount of new loans increased from $ billion in 1994 to $4.56 trillion in 2007 and then declined to $2.66 trillion in 2008 and to $2.12 trillion in Panel C of Table I summarizes the syndicated loans in our sample by year. Approximately 20% 9,341 loans are made to 867 CDS-referenced firms. The largest number of syndicated loans was issued in 2005 at 3,828, whereas 2007 witnessed the largest average loan size in our sample ($ million). Although CDS firms account for less than 10% of our entire sample of borrowers, they account for 43% of the syndicated loan volume in dollar terms. The average loan 16 Basel II requires an 8% minimum total risk-weighted capital ratio and a 4% minimum Tier 1 risk-weighted capital ratio. Basel III increases the minimum Tier 1 capital ratio to 6% (the minimum common equity capital ratio is 4.5%). For U.S. banks, the requirements are 10% and 6%, respectively, during our sample period. The level of equity capital measures the extent to which a bank is prepared to internalize the cost of bank failure rather than to rely extensively on deposit-based financing (Allen, Carletti, and Marquez, 2011). 17 The Federal Reserve Board, the Federal Deposit Insurance Corporation (FDIC), and the OCC proposed that Tier 1 capital leverage ratio be tightened: Bank holding companies with more than $700 billion in consolidated total assets or $10 trillion in assets would be required to maintain a Tier 1 capital leverage buffer of at least 2 percent above the minimum supplementary leverage ratio requirement of 3 percent, for a total of 5 percent ( 14

17 size for CDS firms ($868 million) is more than twice as large as the average loan size for non- CDS firms. 4. Banks CDS Usage and Their Regulatory Capital Ratios Capital relief is a major objective for banks to use CDS given the paramount importance and substantial cost of bank capital. If banks exclusively use CDS for hedging and do so effective, their net credit exposures will be reduced. Consequently, the risk weights, as well as the size of risk-weighted assets, are reduced, leading to higher capital ratios with the same amount of capital. Conversely, if CDS-using banks also extend riskier loans, it may result in larger risk-weighted assets and lower capital ratios. 18 Therefore, the net impact of CDS on riskweighted capital ratios depends on the relative strength of these counteracting effects Unconditional Results with Observed Banks CDS Usage Table I shows that all banks in our sample maintain capital ratios higher than the minimum regulatory requirements. We now analyze regulatory capital ratios in a multivariate framework. Our baseline specification for bank capital ratios is as follows: Bank Regulatory Capital Ratio it α βcdsusage 2 it γ Bank Characteristics 1 t 3 it 1 γ Year Fixed Effects γ Bank Fixed Effects ε i it (1) We examine both the total risk-weighted capital ratio and the Tier 1 risk-weighted capital ratio. The key independent variable is the indicator CDSUsage, which equals one if the bank has a nonzero CDS position in a given quarter and zero otherwise (see the variable definitions in the 18 Banks can also use CDS for dealer activities without exact matching positions. In such cases, more trading assets will appear on banks balance sheets (or off-balance sheet), possibly resulting in larger risk-weighted assets. 15

18 Appendix for details). 19 To control for other differences that may systematically drive capital ratios between CDS-using and non-cds-using banks, we control for bank fixed effects. We include in the regressions bank characteristics that may affect bank regulatory capital ratios that are identified in the literature (e.g., Ellul and Yerramilli, 2013), including bank total assets and its square, sales growth rate, deposits-to-assets ratio, loans-to-assets ratio, and market share in bank deposits. These variables are lagged one quarter when entering the regressions. To account for the possibility that banks with different funding strategies or sources of revenue may hold different levels of capital, we also control for the deposits-to-liabilities ratio and the noninterest income-to-total operating income ratio. These variables describe bank operating strategies ( business model ) and act as controls for bank types. In all the specifications, we control for year fixed effects to isolate time trends in average regulatory capital ratios. The estimation results of the baseline regressions are presented in Panel A of Table II, which indicates that banks total and Tier 1 regulatory capital ratios decline slightly when a bank uses CDS. However, the coefficient estimates for CDSUsage are not statistically significant. This finding is consistent with Minton, Stulz, and Williamson (2009), although it conflicts with the expectation that using CDS helps improve bank regulatory capital ratios. The estimation results for the control variables conform to the results documented in the literature We use the dummy to represent CDS-using banks rather than the quantity of CDS positions held by banks in the baseline regression because CDS positions are highly skewed across banks. The top two CDS-using banks, Bank of America and J.P. Morgan Chase & Co, hold CDS positions far exceeding those of other banks. We focus on the extensive rather than the intensive margin. In Internet Appendix Table IA1, we demonstrate consistent results with an alternative sample excluding the largest banks. 20 For example, capital ratios increase with the deposits-to-assets ratio and decrease with the loans-to-assets ratio. Banks use capital as a complementary funding source to deposits, which is consistent with the view that high-level bank capital signals bank creditworthiness to potential depositors (e.g, Allen, Carletti, and Marquez, 2011; Demirgüc-Kunt and Huizinga, 2010). 16

19 4.2. Incorporating the Selection of Bank CDS Usage with Instrumental Variables The bank decision to use CDS and the choice of capital ratios can be jointly determined. The observed insignificant and negative relationship between banks usage of CDS and bank capital ratios may reflect the selection of banks into CDS trading. For instance, banks with weaker capital positions are more likely to use CDS with the associated capital relief benefits. Similarly, after suffering a negative shock to its loan portfolio (e.g., a default by a group of borrowers in the portfolio), a bank may need more capital to cover its losses, and at the same time, the use of credit derivatives to hedge can increase. This endogeneity may result in either an underestimation or an overestimation of the relationship between CDS usage and bank capital ratios, depending on the correlation between the unobserved factors for capital ratio and for CDS usage choice. We construct instrumental variables to better identify the effect of CDS usage on bank capital ratios. Our first instrumental variable for bank CDS usage is the fraction of a bank s borrowers that have issued bonds in the past quarter. Borrowers with outstanding bonds indicate exposures and market breadth for transaction opportunities. Bank lenders are more likely to use CDS when more of their borrowers issue bonds. As indicated by Column 1 of Internet Appendix Table IA2 a larger fraction of borrowers that have bond issuance predicts a higher probability of CDS usage by the lender in the next quarter. Therefore, this instrument seems to satisfy the relevance condition. Meanwhile, this instrument also meets the exclusion requirement because a borrower s public bond market activities should not directly affect its bank s choice of regulatory capital ratios. Our second instrumental variable is the bank weather-induced revenue volatility before 1997, which is constructed following Pérez-González and Yun (2013). The relevance of the 17

20 instrument is the following: If a bank s revenue is more dependent on local weather conditions, the bank is more likely to use weather derivatives to hedge. Such banks are also more likely to take positions in CDS contracts because banks that use derivatives to hedge tend to hedge more than one aspect of their business (Saretto and Tookes, 2013). Because weather derivative contracts were introduced in 1997 and because our sample banks also began using CDS in 1997, we construct this instrumental variable using pre-1997 (1994 to 1996) information on weather and bank revenue. In the meantime, the weather derivative, which is a small market, is remote from the bank s main business, and pre-1997 weather-induced revenue volatility is not likely to have a direct impact on banks capital positions after The empirical results of the second-stage IV estimation with instrumented bank CDS usage are presented in Panel B of Table II. In contrast to the non-instrumented results in Panel A, the coefficient for the instrumented variable is significantly negatively related to bank capital ratios in all specifications. This result further corroborates the finding from the baseline regressions that CDS usage does not improve banks capital ratios. Instead, banks capital ratios significantly decline after instrumentation of the indicator for CDS usage. The IV estimation result suggests that the impact of CDS usage on capital ratios is negative and the causal impact is attenuated by the banks endogenous choice to use CDSs. 21 To construct this IV, we first obtained monthly average temperature information from the National Oceanic and Atmospheric Administration (NOAA). Next, we mapped the location of the weather stations that collected the local temperature information with the latitude and longitude of the location of a bank s headquarters. We obtained the weather exposure by estimating the specification below using data on temperature and bank revenue: Revenue/As sets α β *DD γ *ln(assets ) ε, it i i where Revenue/Assets is the bank quarterly revenue-to-assets ratio, and DD is temperature 65 o F, a measure of local demand for cooling or heating energy. β is the weather beta, which measures the sensitivity of revenue to i variation in DD it. We multiply the estimated β i by i, the historical standard deviation of monthly the period 1994 to 1996 to obtain the pre-1997 weather exposure measure. it i it it t DD it, over 18

21 A bank may appear safe in terms of its regulatory capital ratios, as our baseline regressions indicate, because regulations such as Basel II allow banks to use credit derivatives to manage capital ratios and to substitute the asset risk weight with the (lower) insurer risk weight in the calculation of risk-weighted assets. Our analysis implies that the observed capital ratio is higher than the level when only the causal effect of CDS is at work. The selection of CDS usage may have masked the real capital adequacy of the banks as they expand their risky assets. One implication of these findings is that regulators may not be able to detect real differences in capital ratios across banks under the banking regulation that allows banks to take advantage of the capital relief brought by CDS. The components and riskiness of bank assets may have changed substantially under the same capital ratios with CDS in place. 22 From the banks perspective, however, this outcome may not be adverse because they may have increased their lending capacity, which is consistent with their business model and also benefits borrowers who might otherwise not obtain a loan. To check the robustness of our findings, we rerun the analysis using an alternative sample that excludes the largest banks from the baseline sample. Specifically, we exclude banks with deposits exceeding 10% of the total deposits aggregated across all banks in the same quarter, following Houston, Lin, Lin, and Ma (2010). The results presented in Internet Appendix Table IA1 are similar to those in Table II, i.e., there is no significant difference in regulatory capital ratios between CDS-using and non-cds-using banks, but there is a significant decline in capital ratios for banks using CDS when the selection of CDS usage is accounted for by instrumentation. 22 In Internet Appendix Table IA3, we analyze the determinants of risk-weighted capital ratios for CDS-using and non-cds-using banks. The set of determinants appears to be different for CDS-using banks than other banks. 19

22 4.3. Tier 1 Leverage Ratio In addition to risk-weighted capital ratios, non-risk-based capital measures, particularly the leverage ratio, are advocated as a supplementary prudential tool to complement minimum capital adequacy requirements. Basel III now includes leverage ratios among regulatory measures. The U.S. has adopted the simple leverage ratio, expressed as the ratio of Tier 1 capital to the adjusted amount of total average assets (quarterly average total assets less intangible assets that include goodwill, investments deducted from Tier 1 capital and deferred taxes). If banks usage of CDS affects both risk weights and quantity of assets that are used to calculate capital ratios, non-riskweighted leverage ratios may also be different between CDS-using and non-cds-using banks. Because off balance sheet activities, such as securitization, cannot affect the balance sheet leverage ratio, banks that aim to circumvent regulatory capital requirements may fund their longterm assets through off balance sheet vehicles. A substantial amount of CDS are used to facilitate securitization. 23 Therefore, banks with greater demand for removing assets to off balance sheet tend to use more CDS, which leads to improved leverage ratios. If the usage of CDS induces banks to expand their assets through riskier lending, which would result in a lower leverage ratio, such an effect may be masked by the inflated leverage ratio as a result of the endogenous selection of banks into CDS usage. Similar to our finding with risk-weighted capital ratios, the baseline regression results in column 1 of Table III indicate that banks usage of CDS appears to have no impact on their Tier 1 leverage ratio. The coefficient estimates of CDSUsage become significantly negative, however, when the endogenous selection of CDS usage is accounted for by instrumentation, as indicated by columns 2 and 3. A one standard deviation increase in the likelihood of a bank using CDS 23 The Financial Crisis Inquiry Report (2011, Page 132). 20

23 leads to a 9.53% decline in its leverage ratio, which suggests that banks can use CDS to mask their true leverage levels Quality of Bank Capital The 1996 Amendment to the Basel capital accord allows Tier 3 capital to account for the market risk in the trading book. This amendment matters for our setting because many CDS positions are for trading purposes. In this subsection, we analyze the effect of bank CDS usage on capital quality, measured by the ratio of Tier 1 capital to total capital, which consists of Tier 1, Tier 2 and Tier 3 capital. If banks pursue a strategy that controls or limits a specific capital level target and if banks can do so independently of the impact of the strategy on other capital quality measures, such a strategy improve the capital level but also lead to lower capital quality. Table IV presents the regression results using the same specification as in the previous analysis of capital ratios. The coefficient estimates of the CDSUsage indicator in the baseline regressions are negative and significant. This finding suggests that the composition of bank capital tilts toward lower quality capital when a bank is engaged in the CDS market. The ratio of Tier 1 capital to total capital is lower (or 4.26% lower relative to the mean of the ratio) for CDS-using banks than for banks that do not use CDS. This difference is statistically significant at the 1% level when we control for bank fixed effects. The decline in capital quality remains robust in the instrumental variable estimations that use the fitted value of the CDSUsage variable estimated from the instruments, as indicated in columns 2 and 3. The results in Table IV suggest that bank capital regulations may have induced banks to shift from controlling risk to controlling capital ratios during our sample period through risk- 21

24 weighted assets and the composition of bank capital. 24 Although the capital ratios remain unchanged, the growth in risk-weighted assets is supported more by Tier 2 and Tier 3 capital than by core Tier 1 capital such as bank equity. Because there is no obvious trend in capital ratios, regulators might have overlooked the actual risk accumulated by banks. One goal of Basel III is to raise both the quality and quantity of the regulatory capital base. 25 Indeed, regulators have closed the loopholes in previous capital regulations by (1) improving bank capital definitions, specifically abolishing Tier 3 capital, and by (2) including the leverage ratio (i.e., Tier 1 capital to non-risk-based assets, rather than RWA, must be greater than 3%) in the new capital accord. Overall, our bank-level evidence demonstrates that bank capital level and quality are lower for CDS-using banks. Next, we investigate the implication of using CDS for regulatory capital management for bank lending. 5. Effects of CDS Usage on Lending Practice: Bank- and Loan-Level Evidence The lower risk-weighted regulatory capital ratios for CDS-using banks after incorporating banks choice of using CDS suggest that CDS may induce banks to expand their risky assets, i.e. by increasing their loan issuance. Such a lending expansion effect appears to be strong, more than offsetting the hedging effect of CDS. Both the aggregated amount of loan issuance to CDSreferenced firms and the total number of CDS trades increase rapidly from early 2000 until mid The Pearson correlation coefficient of the quarterly volume of syndicated loan issuance and the number of CDS trades in borrower names is 0.59, which is statistically significant at the 24 Sheila Bair, former chairman of the U.S. FDIC, has expressed her concern in the calculation of RWA: The risk weightings are highly variable in Europe and have led to continuing declines in capital levels There s pretty strong evidence that the RWA calculation isn t working as it s supposed to ( news/ /europe-lax-rwa-calculations-bair). 25 See page 2 of Basel III: A global regulatory framework for more resilient banks and banking systems. 22