CDS and the Liquidity Provision in the Bond Market

Transcription

1 CDS and the Liquidity Provision in the Bond Market Massimo MASSA Lei ZHANG 2012/114/FIN

2 CDS and the Liquidity Provision in the Bond Market Massimo Massa* Lei Zhang** * Professor of Finance The Rothschild Chaired Professor of Banking Co-Director of the Hoffmann Research Fund at INSEAD, Boulevard de Constance Fontainebleau Cedex, France. Massimo.massa@insead.edu ** Nanyang Business School, Nanyang Technological University, Division of Banking and Finance, 50 Nanyang Avenue, Singapore zhangl@ntu.edu.sg Corresponding author A Working Paper is the author s intellectual property. It is intended as a means to promote research to interested readers. Its content should not be copied or hosted on any server without written permission from publications.fb@insead.edu Find more INSEAD papers at

3 Abstract We study the effect of credit default swap (CDS) on the corporate bond market. We argue that CDS, by reducing the need of investors to liquidate the bonds in the face of credit deterioration of the issuer, reduces fire sale risk and provides bond liquidity. Given that bond investors are segmented by regulation i.e., insurance companies, banks, and pension funds are subject to the risk-based capital requirements and in principle only hold high quality assets, we expect the liquidity provision role of CDS to be concentrated among investment grade bonds. We test this hypothesis using a comprehensive sample of US corporate bonds with CDS contracts information over the period. We show that the presence of CDS reduces yield spreads and increases liquidity for investment grade bonds. This effect is stronger during the financial crisis period. We provide a proper instrumental variable identification that pins down the need for CDS contracts by exploiting the level of loan concentration of the banks which the bond issuer borrows from. We also examine two events that have been shown to trigger forced sales by bond investors: bond rating downgrades from investment grade to high yield, and the selling pressure of property insurance companies following Hurricane Katrina. In both cases, the presence of CDS contracts lowers the impact of fire sales by reducing the drop in bond liquidity and the rise in yield spreads. Our results have important normative implications, as they suggest that at least for the class of investment grade bonds CDS may actually help to reduce risk contagion around financial crises. Keywords: CDS, bond liquidity, yield spreads, fire sales, crisis JEL Classification: G10, G15, G21

4 Introduction The recent financial crisis has brought to the fore the role of credit derivatives and its implications for the financial markets. The markets for credit derivatives, in particular, credit default swaps (CDSs) have been found controversial in the spread of the financial crisis (Allen and Carletti, 2006, Stulz, 2010). Chris Wolf, a hedge fund manager, stated that CDS has become essentially the dark matter of the financial universe. 1 George Soros has called for most or all trading in credit default swaps to be banned or strictly regulated. 2 The rapid rise of such a market has made the problem even more acute: the total notional amount of the CDS market has grown from $6 trillion to $41 trillion from 2004 to The indictment is based on the way credit insurance would affect the debtor-creditor relation in the case of distress of the borrower, turning creditors into empty creditors (Bolton and Oehmke, 2011). 4 Indeed, CDS contracts may affect the incentives of the lenders and increase the risk of bankruptcy of the borrowing firm. The lenders, if protected in the case of distress, would have lower incentives to restructure the debt, as they can benefit from their CDS positions ( the empty creditor problem ). The lower risk in the presence of distress, would also reduce the lenders incentives to effectively monitor the borrowers (Thompson, 2007, Parlour and Winton, 2008). Regardless of this almost consensual theoretical dark view, the empirical evidence on the impact of CDS on the underlying firm is mixed at best. On the one hand, the presence of CDS contracts appears to reduce loan quality, increases bankruptcy risk and borrowing costs, and lowers the efficiency of the bond market (Ashcraft and Santos, 2007, Purnanandam, 2011, Subrahmanyan et al., 2011, Kalimipalli and Nayak, 2011). On the other hand, CDS seems to stimulate bank credit supply and improve the borrowing terms e.g., maturity and spreads of the firms for which such an 1 The $55 trillion question, by Nicholas Varchaver, Fortune Magazine, Sep. 30, One way to stop bear raids, Wall Street Journal, Mar. 24, This figure is from the statistics provided by the Bank of International Settlements. 4 The use of derivatives can in general allow a decoupling of voting and cash-flow rights in common equity through the judicious use of derivatives to hedge cash-flow risk (Hu and Black, 2006, 2007, Kahan and Rock, 2007). 1 Electronic copy available at:

5 insurance contract exists, enabling them to have higher leverage and borrow at longer maturities (Hirtle, 2008, Saretto and Tookes, 2011). In this paper, we focus on one important yet undocumented role of CDS: the reduction in forced sale (fire sale) risk in the underlying bonds. The recent financial crisis has brought to attention the transmission mechanism related to the need of financial intermediaries i.e., mutual funds, hedge funds, insurance companies etc. to meet the withdrawals from their investors. This phenomenon has been alternatively termed as fire sales (Coval and Stafford, 2007, Shleifer and Vishny, 2011), financial run (Bernardo and Welch, 2004), or forced sales due to regulatory pressure (Ellul et al., 2011). The idea is that, if a shock induces some financial intermediaries to liquidate an asset, the forecast of such a sale will lead the other players in the market to try to preempt it by selling as well. The ensuing massive sales can drastically negatively affect the asset price and its liquidity. This effect will be especially relevant in the bond market, given that most corporate bonds are traded over-thecounter with high search costs, and the liquidity in this market arises from the bond dealers committing risk capital to market making. More importantly, regulatory pressures may force some large institutional bond investors to sell in the presence of a drop in market value or a downgrade in bond ratings, and thus create significant fire sale or liquidity risk for the bonds. For example, consider the largest class of investors in the corporate bond market: the insurance companies 5. By regulation insurance companies can only hold investment grade bonds, and the amount of risk-based capital required by the state regulator is based on the credit quality of their asset holdings (Ellul et al., 2011). A negative shock to the bonds they hold, such as a rating downgrade, will require them to post additional equity capital, unless they choose to sell the bonds. However, if such insurance companies were to buy CDS protection, the need to find new capital would be largely reduced. In other words, CDS contracts allow investors who are required by regulation to hold high quality bonds to defer the sale in the case the bonds get downgraded or even 5 At the end of the second quarter of 2005, insurance companies hold $574 billion of publicly issued corporate bonds, among which $484 billion are held by life insurance companies and $90 billion are held by property and reinsurance companies. 2

6 lose the coveted investment grade status. More importantly, this would lower the other players preemptive incentives to front-run the insurance companies, and effectively induces a lower need to rush to sell the bonds in the market. Moreover, the presence of CDS may induce arbitrageurs to provide liquidity when bonds are subject to such fire-sale related liquidity shocks. Indeed, if bonds are temporarily underpriced because of liquidity shocks, the CDS basis arbitrage strategy i.e., buy bonds and buy CDS protection would facilitate liquidity provision in the bond market (Choudhry, 2006). In other words, CDS reduces the limit of arbitrage (Shleifer and Vishny, 1997) in the bond market. This suggests that bonds issued by firms with CDS contracts would suffer less fire sale risk. The lower risk of fire sale should increase bond liquidity and reduce bond yield spreads. This intuition provides a new and novel angle unexplored in the literature till now on the role of CDS in the corporate bond market. We hypothesize that the presence of CDS contracts lowers the yield spreads of the bonds and increases their liquidity. Such effect should concentrate among investment grade bonds the ones that may experience fire sale risk due to regulatory pressures. High yield bonds, already held by investors who are not subject to regulatory constraints in terms of the quality of the assets they hold e.g., hedge funds and high yield mutual funds should be less subject to fire sale risk. We test this hypothesis using data on a comprehensive sample of US corporate bonds with CDS contracts information over the 2001 to 2009 period. In the first part of our analysis, we focus on the overall relationship between CDS contracts and both corporate yield spreads and bond liquidity. We document a strong negative relationship between yield spreads and the availability of CDS contracts in the case of investment grade firms. The presence of CDS contracts reduces the bond yield spread by 22 bps. There is instead no relationship for high yield bonds. The link between CDS contracts and yield spreads is stronger during the financial crisis period (the dotcom crash, the subprime crisis), in line with our hypothesis of a liquidity provision role of CDS. 3

7 Then, we focus on bond liquidity. We follow Bao et al. (2011) and define bond illiquidity as the implied bid-ask spread based on the auto-covariances of bond price changes. We find that, in the case of investment grade bonds, the existence of CDS contracts lowers bond illiquidity by 9% compared to the unconditional mean. In contrast, in the case of high yield bonds, the presence of CDS contracts actually increases bond illiquidity. These effects are even stronger during the financial crisis. That is, the presence of CDS contracts increases liquidity for investment grade bonds and reduces it for high yield bonds. To provide a causal interpretation of our analysis, we consider an instrumental variable identification that exogenously pins down the need for CDS contracts in the market. This is based on the level of loan concentration of the lending banks, from which the bond issuer borrows its bank debt. The intuition is that banks tend to use credit derivatives to hedge their loan positions. The less diversified their loan portfolio is, the higher the incentive they have to purchase CDSs for hedging purposes. We directly provide evidence of this claim, by explicitly relating the degree of concentration of the loan portfolios of the banks across different industries and geographical regions to their use of credit derivatives for hedging purposes. We show a strong positive relationship between the use of credit derivatives and the degree of concentration of the loan portfolios, both in the case of notional amount and in the case of notional amount standardized by the assets of the bank. 6 The economic significance is also sizable. One standard deviation more concentrated loan portfolio is related to 59% higher use of credit derivatives for hedging purposes. The instrumental variable analysis confirms the previous findings, displaying a strong negative relationship between the presense of CDS contracts and yield spreads and a positive relationship between CDS presense and bond liquidity for investment grade bonds. One standard deviation increase in the instrumented presence of CDS contracts lowers yield spread by 26 bps and reduces illiquidity by 8%. No effect is there for high yield bonds. 6 There is instead no relationship between loan concentration and the bank s use of derivatives to hedge interest rate risk and foreign exchange risk. This implies that the degree of loan concentration is directly related to the bank s hedging of credit risk but not some general hedging purposes. 4

8 Several robustness checks confirm our results. In particular, we exploit a proxy of depth in the CDS market, as measured by the number of CDS quotes provided by the CDS dealers (Qiu and Yu, 2012). Consistently with the previous results, we find that the CDS depth reduces bond yield spreads and increases bond liquidity for investment grade bonds. No effect is there for high yield bonds. In the second part of our analysis, we focus on two events that have been shown in the literature to trigger forced sales by bond institutional investors: bond rating downgrades from investment grade to high yield grade, and the selling pressure of property insurance companies following Hurricane Katrina. The downgrade from investment grade to high yield triggers forced sales as insurance companies liquidate their holdings in the downgraded bonds ( fallen angels ) to comply with the NAIC s regulatory constraints (Ellul et al., 2011) 7. We argue that the presence of CDS contracts lowers the impact of the forced selling pressure of insurance companies upon such downgrade. And indeed, we find that bonds without CDS contracts experience a 150% higher drop in institutional ownership than the bonds with CDS contracts. This provides a direct evidence in support of our hypothesis, as the lower drop in institutional ownership can only be attributed to the lower need to sell assets generated by the presence of CDS. Consistently, the presence of CDS contracts lowers the impact of fallen angels in terms of both yield spreads and bond liquidity. Bonds without CDS contracts experience a 200% (150%) higher increase in yield spread (bond illiquidity) than bonds with CDS contracts. Next, we examine an even more exogenous event: the shock to property insurance companies provoked by Hurricane Katrina. Hurricane Katrina (23-30, August, 2005) is the costliest natural disaster in the history of the United States with an insured damage of over $40 billion. The selling pressure of Katrina-exposed property insurance companies, driven by the need to meet redemption claims, generated a selling of the bonds held by those investors (Massa and Zhang, 2011). These 7 Ellul et al. (2011) show that the selling pressure of insurance companies generates significant price drops on the bonds downgraded from investment grade to high yield. Indeed, consistent with Ellul et al. (2011), we find a significant 4% decrease in bond institutional ownership around the quarter of such downgrade, which is much higher than that of an average 1% decrease in bond institutional ownership around other rating downgrades not crossing the investment grade/high yield threshold. 5

9 forced sales can only be attributed to supply-side shocks i.e., Katrina-exposed property insurance companies as opposed to firm-specific shocks such as rating downgrades 8. We document that the presence of CDS contracts reduces the impact of forced sales in terms of both the drop in bond prices as well as the drop in liquidity. Specifically, in the face of the selling pressure of Katrina-exposed property insurance companies, bonds without CDS contracts experience a 45% higher increase in yield spreads than that of bonds with CDS contracts. In line with the previous results, the increase on bond illiquidity concentrated among bonds without CDS contracts. Our study contributes to several strands of literature. First, our paper relates to the emerging literature on the impact of CDS contracts on the underlying firm. The theoretical literature has mostly focused on the effects of CDS on renegotiation between debtors and creditors, and the associated costs and benefits (e.g., Arping, 2004, Hu and Black, 2008a, b, Yavorsky, 2009, Bolton and Oehmke, 2010, Subrahmanyam et al., 2011, Gormley et al., 2011), as well as the ensuing managerial incentive in risk taking (Thompson, 2007, Parlour and Winton, 2008) 9. The empirical literature (e.g., Hull et al., 2004, Longstaff et al., 2005, Norden and Wagner, 2008, Norden and Weber, 2009, Chen et al., 2010, Ismailescu and Phillips, 2011, Kim 2011, Nashikkar et al., 2011) does not provide unequivocal evidence. On the one hand, CDSs are found to reduce loan quality, increase bankruptcy risk and increase the probability of credit rating downgrade (Ashcraft and Santos, 2007, Peristiani and Savino, 2011, Purnanandam, 2011, Subrahmanyan et al., 2011). On the other hand, Bedendo et al. (2011) find that CDS contracts do not significantly increase the probability of bankruptcy when the firm is already in distress. CDSs are also found to stimulate bank credit supply and improve borrowing terms, and enable firms to maintain higher leverage and borrow at longer maturities (Hirtle, 2008, Saretto and Tookes, 2011). We contribute to this literature by providing 8 In this case, the concern of a potential spurious correlation between the presence of CDS contracts and firm-specific shocks in the proximity of rating downgrades can be ruled out. 9 On the bright side, Duffee and Zhou (2001) argue that CDS allows for the decomposition of credit risk into components that have different sensitivities to information, thus potentially helping banks overcome a lemon problem when hedging credit risk. 6

10 evidence of the bright side of CDS contracts on the bond market and showing how this is directly related to bond liquidity. Second, we contribute to the literature on fire sales and financial crisis (Shleifer and Vishny, 2011, Bernardo and Welch, 2004, Coval and Stafford, 2007) and on fire sales in the bond market in particular (Da and Gao, 2009, Ellul et al., 2011). We contribute by focusing on the way CDS contracts reduces the market impact of such selling pressures on the bond market, improving liquidity and reducing yield spread. Third, we contribute to the literature on the impact of bond illiquidity on corporate yield spreads. Bao et al. (2010) find that in the cross-section, bond illiquidity explains the individual bond yield spreads with large economic significance. Friewald et al. (2012) and Nielsen et al. (2012) confirm that illiquidity explains a large part of the variation in yield spreads across bonds after accounting for credit risk, and the yield spread contribution from bond illiquidity increased dramatically during the period of the subprime crisis. We contribute to this literature by showing that the presence of CDS contracts provides liquidity to investment grade bonds by reducing fire sale risk, which directly translates into lower yield spreads, and consistently, this effect is significantly stronger during the crisis period. Fourth, we contribute to the literature on the financial effects of large natural disasters (Sprecher and Pearl, 1983, Dividson et al., 1987, Shelor et al., 1990, Aiuppa et al., 2003, Ewing et al., 2006, Blau et al., 2008). We show how the existence of CDS contracts may muffle such effects. Our findings have important normative implications as well. Indeed, they show that the presence of CDS contracts at least for the class of investment grade bonds does in fact reduce the effect of fire sales and this benefit is especially stronger during the period of financial crisis. This evidence suggests that, contrary to the general media perception, CDS may actually help to reduce credit risk transfer and contagion around financial crises (Stulz, 2010). The remainder of the paper is organized as follows. In Section II, we describe the data and the construction of the main variables. In Section III, we present evidence on the link between the presence of CDS contracts and both yield spreads and bond liquidity. In Section IV, we use an 7

11 instrumental variable specification to establish a causal relationship. In Section V, we examine the role of CDS in two fire-sale related events. A short conclusion follows. II. Construction of Data and Variables We use data from multiple sources. The data on monthly bond yield spreads come from Bank of America-Merrill Lynch Corporate Master Index Compositions. The BofA-Merrill data have been used in previous studies (Schaefer and Strebulaev, 2008, Acharya et al., 2009). These data cover most rated US publicly issued corporate bonds (Acharya et al., 2009) and provide bond-level information on the option-adjusted yield spread, coupon rate, duration, face value, and credit ratings. We require that each bond must be included in the index for over 24 months. We obtain information on a number of bond characteristics such as the offering date, the maturity date, offering amount, seniority, callability, fungibility and credit enhancement from the Mergent Fixed Income Securities Database (FISD) 10. This database reports the characteristics of nearly all US fixed income securities. We merge the BofA-Merrill data with the Mergent FISD using bond CUSIPs. We obtain information on the tick-by-tick bond transactions from the Trade Reporting and Compliance Engine database (TRACE) from 2002 to TRACE is the Financial Industry Regulatory Authority (FIRA) over-the-counter (OTC) corporate bond market real-time price dissemination service. TRACE consolidates transaction data for all eligible corporate bonds - investment grade, high yield and convertible debt. It provides detailed records on the time of trade execution, price, yield and some information on trading volume. We get information on CDS contracts from the Markit CDS database. This provides daily firmlevel data on CDS spreads for the period from 2001 through The CDS spread is the periodic fee that the protection buyer pays to the protection seller in a credit default swap contract until the contract matures or a credit event occurs, in which case the protection buyer delivers defaulted bonds to the 10 The FISD data used in our analysis are based on the 2009 edition of the FISD database. 8

12 seller in exchange for the face value of the issue in cash (physical settlement) or the protection seller directly pays the difference between the market value and face value of the issue to the protection buyer (cash settlement). The Restructuring Clause of a CDS contract specifies the credit events that trigger settlement. Typically, Markit reports a composite daily CDS spread, which is an average across all the quotes provided by market makers after a series of data cleaning tests. The Markit database also provides identifying information on the reference entity (such as firm name and ticker), and the terms of the CDS contract (maturity, currency denomination and restructuring clauses). 11 We focus on the spreads of all the CDS contracts written on US firms and denominated in US dollars. Our combined sample includes 158,122 bond-month observations (3,468 firm-year observations) from January 2001 to December We provide the descriptive statistics in Table I. For each variable, we report the data frequency, source, number of observations, mean and stand deviation. The detailed definitions of each variable can be found in the Appendix. In our sample, 83 percent of the bonds are investment grade bonds, 64% of the bonds are callable bonds, and 9% of the bonds have credit enhancement. On average, among the bond issuers, 74% of the bond issuers have CDS contracts outstanding during the sample period. This fraction is higher among investment grade issuers (78%) than high yield issuers (61%). III. CDS, Yield Spreads and Bond Liquidity A. Yield Spreads We begin by relating the presence of CDS contracts to corporate yield spreads. We estimate a pooled specification in which the bond yield spread is regressed on a CDS presence dummy, and a set of control variables. The dependent variable is the option-adjusted bond yield spread. It is defined as the number of percentage points that the treasury spot curve must be shifted in order to match discounted 11 Specifically the maturity of CDS contracts ranges from 6 months up to 30 years, and there are four major restructuring clauses (full restructuring, modified restructuring, modified-modified restructuring and no-restructuring). A detailed discussion of different restructuring clauses can be found in Packer and Zhu (2005). 9

13 cash flows to the bond s price. The CDS presence dummy ( CDS presence ) is a dummy variable equal to 1 if the issuing firm has quoted CDS contracts on its bonds in the previous month and 0 otherwise. The set of control variables includes major bond characteristics such as coupon rate, duration, offering amount, callability, fungibility and credit enhancement and important firm characteristics such as equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment as well as industry fixed effects at the two-digit SIC level, time fixed effects at the monthly level and credit rating fixed effects at the issue level. We also consider specifications with time (monthly) credit rating fixed effects. All firm-year (month) variables are taken at the end of the previous year (month). We always cluster the errors at the firm level 12. We report the results in Table II. In Panel A, we report the overall results, while in Panel B, we consider the results around the period of financial crisis. We start with the overall results. Columns (1)-(3) are for the subsample of investment grade bonds, while columns (4)-(6) are for the subsample of high yield bonds. In column (1), we control for bond characteristics including coupon rate, duration, offering amount, callability, fungibility and credit enhancement, time fixed effects at the monthly level, and credit rating fixed effects at the issue level. In column (2), we control for time credit rating fixed effects. In column (3), we control for firm characteristics including equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. Columns (4)-(6) follow the same specifications as in columns (1)-(3), respectively. The results display a significantly negative relationship between yield spread and the availability of CDS contracts in the case of investment grade issuers. The results hold across different specifications and are economically sizable. The presence of CDS contracts reduces the yield spread by 22bps, which represents a 14% increase relative to the unconditional mean. However, there is no relationship in the case of high yield bonds. 12 All of our results are consistent and statistically more significant if we cluster the errors at the bond level. 10

14 The results on the control variables are largely in line with expectations. For both investment grade and high yield bonds, the coupon rate is positively related to bond yield spreads, supporting a tax-based explanation. Consistent with Campbell and Takslar (2003), even after controlling for credit rating time fixed effects, equity volatility and firm leverage are strongly positively related to bond yield spreads 13. Equity beta (firm profitability) is positively (negatively) related to bond yield spreads but only so in a significant way for investment grade bonds. Interestingly, bond duration is positively related to yield spreads for investment grade bonds while negatively related for high yield bonds 14. In Panel B, we interact the variable of CDS presence with a dummy variable indicating periods of financial crisis: the dotcom crisis ( ) and the subprime crisis (2008 and the first half of 2009). Columns (1-(3) are for the subsample of investment grade bonds. Columns (4)-(6) are for the subsample of high yield bonds. We follow the same specifications as in Panel A. We always include time (monthly) or credit rating time fixed effects and therefore the crisis period dummy is omitted from the regressions. Also in this case, the results show a significantly negative relationship between the presence of CDS contracts and yield spreads for investment grade and no relationship for high yield bonds. More interestingly, we see that the link between CDS presence and yield spreads is significantly stronger during the crisis period, in line with our hypothesis of a fire-sale risk related role of CDS. B. Bond Illiquidity Next, we relate the presence of CDS contracts to bond illiquidity. Following Bao et al., (2011), we define bond illiquidity as the implied bid-ask spread based on the auto-covariances of bond price changes: Bond Illiquity 2 γ (0 if γ 0, where γ Cov p, p, and p is the log price at time t. The focus variable ( CDS presence ) as well as the control variables are defined as in the 13 We find that the inclusion of credit rating time fixed effects renders firm size insignificant in explaining bond yield spreads. 14 This finding can be explained as follows. Bond issuers with longer duration bonds face higher interest rate risk but lower short-term refinancing risk. In the case of high yield bonds, the benefit due to lower refinancing risk may dominate so that bond duration is negatively related to yield spreads. 11

15 previous specification. All the firm-year (month) variables are taken at the end of the previous year (month). We report the results in Table III. In Panel A, we consider the overall results, while in Panel B, we consider the results around the period of financial crisis. The layout of the columns is the same as in Table II. We find that, in the case of investment grade rating, bonds of firms with CDS contracts are more liquid than those of firms with no CDS contracts. This result holds across the different specifications and is economically significant. The existence of CDS contracts lowers bond illiquidity by 9% compared to the unconditional mean. In contrast, CDS presence increases illiquidity in the case of high yield bonds by 8% relative to the unconditional mean. If we focus on the crisis period, in line with the findings on yield spreads, we see that the impact of the presence of CDS contracts on bond liquidity is significantly stronger. That is, the presence of CDS contracts increases liquidity for investment grade bonds especially during the financial crisis period. This evidence consistently supports the liquidity provision role of CDS on the bond market. IV. An Instrumental Variable Identification A. Main Results The previous results document a significant relationship between the presence of CDS contracts and bond liquidity and yield spreads, providing strong support for our hypothesis. However, it may not be enough to establish a causal relationship. Indeed, it may be possible that CDS contracts exist in the very firms characterized by unobserved liquidity and risk features that also determine yield spreads and bond liquidity. To address this issue, we provide an instrumental variable specification. We use as instrument the level of loan concentration of the lending banks which the bond issuer borrows from. The intuition is that banks use credit derivatives to hedge their loan positions. The less diversified their overall loan portfolio is, the higher is the incentive they have to purchase CDSs for hedging purposes. To provide evidence of this claim, we link the degree of concentration of a bank s 12

16 loan portfolio across different industries and geographical regions to its use of credit derivatives, foreign exchange derivatives and interest rate derivatives (for hedging purposes). The analysis is done at the bank level. First, for each bank, we define a measure of loan concentration based on the bank loan data from LPC Dealscan 15. We focus on all of the loan transactions in the US. For each bank-year, we classify its existing loans into different industries (two digit SIC)-states pairs. We then calculate the herfindal index as the proxy for loan concentration. We expect that banks whose loans are concentrated in a specific region and industry would face a higher credit risk and have greater incentives to purchase credit derivatives for protection. Next, we link (by name matching) LPC Dealscan with the Bank Regulatory database, which contains balance sheet and off-balance sheet information of US banks 16, and more importantly, detailed information on the banks use of interest rate, foreign exchange and credit derivatives specifically for hedging purposes. We collect the notional amounts of such derivatives positions. Finally, we regress the use of credit derivatives, foreign exchange derivatives and interest rate derivatives (for hedging purposes) to the degree of concentration of a bank s loan portfolio across different industries and geographical regions. We report the results in Table IV, Panel A. The dependent variable in columns (1)-(2) is the log value of the notional amount of credit derivatives (RCFDA535). The dependent variable in columns (3)-(4) is the notional amount of credit derivatives divided by the bank s total asset (RCFDA535/RCFD2170). The dependent variable in column (5) is the notional amount of foreign exchange derivatives divided by the bank s total asset (RCFD8726/RCFD2170), while the dependent variable in column (6) is the notional amount of interest rate derivatives divided by the bank s total asset (RCFD8725/RCFD2170). Bank size is the log value of total asset (RCFD2170). We provide a detailed description of each data item in the Appendix. The results show a significantly positive relationship between the use of credit derivatives for hedging and the degree of concentration of the loan portfolio of the bank, both in case of notional 15 Dealscan is a comprehensive dataset that contains detailed information relating to the start and expiration dates of loan deals along with the names of the lending banks, loan amounts, and terms and conditions of the loans. 16 We require the bank s total amount of commercial and industrial loans (RCON1766) to be larger than $100 million. This requirement makes sure that the banks we use are commercial banks actively involved in the corporate loan market. 13

17 amount and in case of notional amount standardized by bank asset. The economic significance is sizable. One standard deviation more concentrated portfolio is related to 59% (133%) higher use of credit derivatives in notional amount (notional amount standardized by the bank asset), compared to the unconditional mean. The last two columns of Panel A (Table IV), provide some placebo tests. They show that such a relationship does not exist in the case of interest rate derivatives and foreign exchange derivatives. This significantly increases our confidence in the results for the credit derivatives, suggesting that the concentration of the loan portfolio is directly related to credit risk, but not to interest risk and foreign exchange risk, which are arguably more systemic than credit risk and are less likely to be managed by loan diversification. These findings make us confident to exploit the concentration of the loan portfolio of the banks as a proxy for the market demand for credit protection i.e., as an instrument for the presence of CDS contracts. Therefore, we proceed to link the presence of CDS contracts to bond yield spread and bond illiquidity, by instrumenting the CDS presence with the degree of loan concentration of the lending banks. We proceed as follows. First, for each bank, we calculate the degree of concentration of its loan portfolio using the herfindhal as a measure of concentration. Second, we aggregate the degree of loan concentration at the issuer level, by taking the value (loan amount)-weighted herfindal among all the banks from which the issuer borrows in the prior 5 years. Then, we link the degree of weighted average concentration of the lending banks to the existence of a CDS contract for the specific firm. We estimate a probit regression of the CDS presence dummy on the loan herfindal and a set of control variables. We report the results in columns (1) and (2) of Table IV, Panel B. At the bottom of these columns, we also report the F-test to provide the Staiger and Stock (1997) test of weakness of instrument on the loan herfindal variable. The results show that the presence of CDS contracts is strongly positively related to the degree of concentration of the lending banks. This means that, the more concentrated the loan portfolios of the banks are, the higher the probability that the firm has a CDS on its bonds. Moreover, the F-test 14

18 delivers a Staiger and Stock (1997) statistic of weak instruments higher than 5. This comfortably allows us to trust the strength of our instrument. Then, we estimate an instrumental variable specification using the degree of loan portfolio concentration of the lender as instrument 17. We report the results in Table IV, Panel B, columns (3)-(6). In columns (3)-(4), we use as dependent variable the bond yield spread, while in columns (5)-(6), we use the bond illiquidity. In Columns (3) and (5), we break down the analysis for the subsample of investment grade bonds, while in columns (4) and (6) we focus on the subsample of high yield bonds. In all the specifications, we include industry, time and credit rating fixed effects, and cluster the errors at the issuer level. The results support the previous ones, displaying a negative relationship between CDS presence and yield spread and a positive one between CDS presence and bond liquidity for investment grade bonds. In particular, one standard deviation increase in the instrumented CDS presence dummy lowers yield spread by 26 bps and reduces illiquidity by 8%. No effect is there for high yield bonds. As a robustness check, we also consider an alternative instrument based on the average loan herfindal among issuers with the same industry (two-digit SIC code), region (state level) and basic rating category (investment grade/high yield). That is, the instrument is not based on the set of banks that lend to the specific firm, but on all the banks that are lending to similar firms in terms of geography, rating and industry. This alleviates any residual concern on endogeneity induced by the fact that the previous instrument was based on the banks that had chosen to lend to the specific firm. The results are reported in Panel C, Table IV. For brevity, we mute the control variables and only report the variables of interest. The findings are consistent with the previous ones in Panel B. In fact, the statistical significance is even stronger. The F-test in the first-stage regression delivers a Staiger and Stock (1997) statistic of weak instruments over 25. These results display a causal link between 17 We follow Wooldrige (2001) and use the fitted value from the probit regression (column (2), Panel B, Table IV) as the instrumental variable. 15

19 CDS and both yield spreads and bond liquidity, i.e., the presence of CDS contracts lowers yield spreads and increases bond liquidity for investment grade bonds. B. Robustness Checks We now consider some robustness checks to the previous results in Table IV. We start by considering an alternative measure of CDS presence in the market. We redefine CDS presence as a dummy variable equal to 1 if the issuing firm has the most liquid CDS contracts in the previous month and 0 otherwise i.e., CDS contracts with 5-year maturity and MR restructuring clause 18. This follows the literature (Zhang, et al., 2008, Cao, et al., 2010, Elkamhi, et al., 2010) that not all the CDS contracts are equally liquid and the effect of the CDS should be concentrated among the most liquid ones. We report the results in Table V. Panel A and Panel B follow the same specifications as in Panel B and Panel C of Table IV, respectively. In the interest of brevity we only report the interested variables. We find that the results are very much similar to the previous ones. Next, we focus on the subsample of bonds issued by firms with CDS contracts trading in the market. In this case, we measure the CDS presence using a proxy of depth in the CDS market. We use the number of dealers providing CDS quotes as a measure for the depth of CDS contract. 19 This measure has been used by Qiu and Yu (2012) who document that the CDS depth is higher for investment grade bonds than for the high yield ones, and it is significantly related to the endogenous liquidity provision by informed financial institutions. We therefore focus on the instrumental variable specification instead of a simple OLS regression, with the CDS depth instrumented by the loan concentration of the lending banks. We construct the instrument in the same way as in Table IV. We report the results in Table VI. In Panel A, columns (1) and (2), we run OLS regressions of the CDS composite depth on the loan herfindal. We perform the F-test to identify the weakness of the loan herfindal variable. Then, we use it as the instrument for the CDS composite depth in columns (3)-(6). 18 Under the MR (modified restructuring) clause, the restructuring agreements are counted as a credit event, but the deliverable obligation against the contract has to be limited to those with a maturity of 30 months or less after the termination date of the CDS contract or the reference obligation that is restructured (regardless of maturity). 19 The Markit data only provide information on the number of dealers in the 5-year maturity contracts. Therefore, we define CDS composite depth as the log number of dealers in the CDS contracts with 5-year maturity. 16

20 The dependent variable in columns (3)-(4) is the bond yield spread, while dependent variable the dependent variable in columns (5)-(6) is the bond illiquidity. Columns (3) and (5) are for the subsample of investment grade bonds. Columns (4) and (6) are for the subsample of high yield bonds. In all of the specifications, we control for industry, time and credit rating fixed effects, and cluster the standard errors at the issuer level. In Panel B, we use as instrument is the average loan herfindal among issuers in the same industry (two-digit SIC code), region (state level) and basic rating category (investment grade/high yield). For brevity, we only report the interested variables in the table. The results show that the loan portfolio concentration also explains the degree of depth of the CDS market. This evidence clearly shows that the demand for CDS contracts is indeed related to the banks need to rely on credit derivatives for hedging purposes. The Staiger and Stock (1997) tests of weak instrument are comfortably passed. Then, when we focus on the impact of CDS depth on bonds. The results display a negative relationship between the CDS depth and both bond yield spreads and bond illiquidity. One standard deviation higher instrumented CDS composite depth is related to a 28 bps lower yield spread and 6% lower bond illiquidity for investment grade bonds. Consistent with the previous findings, no effect is there for high yield bonds. V. Event-based Analysis We now focus on two events that may affect the behavior of institutional investors holding corporate bonds: the rating downgrade from investment grade status to high yield, and the selling pressure of Katrina-exposed property insurance companies following Hurricane Katrina. A. Falling Angels We begin by focusing on the bonds experiencing rating changes. We are interested in bonds that are downgraded from investment grade to high yield ( fallen angels ). Ellul et al. (2011) show that such downgrade triggers the forced sales of insurance companies, and generates large negative liquiditydriven effect on bond prices. In line with their findings, we find that in our sample, there is a 17

21 significant 4% decrease in bond institutional ownership 20 around the quarter of such downgrade (9% decrease relative to the ownership before the downgrade). In direct contrast, the average drop in bond institutional ownership around other rating downgrades not crossing the investment grade/high yield threshold 21 is only 1%. Our hypothesis predicts that bond issuers without CDS contracts should experience a greater drop in institutional bond ownership, a higher increase in bond yield spreads and bond illiquidity, upon such rating downgrade from investment grade to high yield. Our test is structured as follows. We regress the changes in bond ownership, the change in bond yield spreads and the changes in bond illiquidity around the month of rating changes on a fallen angel indicator, its interaction with a no CDS indicator and a set of control variables. For a given bondmonth (bond quarter), 22 we define the fallen angel indicator as a dummy taking the value of 1 in the month (quarter) in which the bond is downgraded from investment grade to high yield and zero otherwise. The no CDS indicator equals 1 if the bond issuer has no CDS contracts in the previous month (quarter) and 0 otherwise. The variable of focus is the interaction term. We expect it to be negatively related to the changes in bond institutional ownership, while positively related with both the changes in bond yield spreads and the changes in bond illiquidity. We report the results in Table VII. In Panel A, we focus on the changes in bond institutional ownership around rating changes, and in Panel B and Panel C on the changes in bond yield spreads and the changes in bond illiquidity, respectively. Columns (1)-(3) are based on the full sample of rating changes including both rating downgrades and rating upgrades. In column (1), we only interact the fallen angel dummy with the No CDS dummy. In column (2), we add the interaction terms of fallen angel dummy with bond characteristics including bond duration, offering amount and bond age. In column (3), we add additional interaction terms of fallen angel with risk characteristics such as 20 We derive the data on quarterly institutional holdings of corporate bonds from Lipper s emaxx fixed income database from the first quarter of 2001 to the second quarter of It contains details of fixed income holdings for U.S. and European insurance companies, U.S., Canadian and European mutual funds, and leading U.S. public pension funds. 21 The threshold rating category that defines investment grade and high yield status is BBB-, i.e., above BBB-, investment grade; below BBB-, high yield. 22 The test on the change in bond institutional ownership is at the bond-quarter level given the quarterly frequency in the institutional holdings data. The tests on the change in bond yield spreads and the change in bond liquidity are at the bondmonth level. 18

22 equity volatility and equity beta. 23 Columns (4)-(6) follow the same specifications as columns (1)-(3), except that they are only based on the subsample of rating downgrades. In the interest of brevity, in columns (3) and (6), we don t report the results on firm-level controls. We always include time credit rating fixed effects, industry fixed effects at the two-digit SIC level, and cluster the errors at the issuer level. The results show a significantly negative coefficient on the interaction term between the no CDS dummy and the fallen angel dummy for the changes in bond institutional ownership, while a significantly positive one for both the changes in yield spreads and the changes in bond illiquidity. This holds across the different specifications. The effect is not only statistically significant, but also economically relevant. Bonds without CDS contracts experience a 150% higher drop in institutional ownership than that of bonds with CDS presence. A similar effect is there in the case of yield spread and bond illiquidity. Bonds without CDS contracts experience a 200% (150%) higher increase in yield spread (bond illiquidity) than that of bonds with CDS presence. These findings strongly support our hypothesis that the presence of CDS contracts reduces the fire sale effect for investment grade bonds upon rating downgrades. B. Hurricane Katrina The second experiment is based on Hurricane Katrina (August 23-30, 2005) and the Katrina-exposed property-casualty insurance companies. Hurricane Katrina is the costliest natural disaster in the history of the United States, with a total property damage estimated at $81 billion (2005USD) and almost $40.6 billion of insured losses (Knabb et al., 2005) 24. It represents a large exogenous shock to the property insurance and reinsurance industry, especially for companies with large insurance exposure to Katrina 25. Given that insurance companies are the largest corporate bond holders, this provides an 23 The purpose of including these additional interaction terms is to eliminate concerns that the result on the interaction of the fallen angel dummy and the no-cds contracts dummy may be driven by other bond or firm characteristics. 24 A special report by Towers Perrin Co. (2005 studying the impact of Hurricane Katrina on the insurance industry estimates the range of privately insured loss to be between 40$ and 55$ billion. 25 Here by large exposure, we mean those insurance companies that have large market share of insurance business in the Gulf region (state of Mississippi, Alabama, and Louisiana). For example, State Farm Insurance, which has the largest market share in the Gulf region (26.62%), states the following words on its website: In a typical year, State Farm receives between 19

23 ideal experiment in which the selling pressure of the bonds held by exposed insurance companies is not related to firm specific characteristics (e.g., credit risk), but is driven by the market concerns on the forced sales by the affected insurance companies to meet redemption claims. Previous research in the finance and insurance literature has studied the effect of natural disasters on the stock prices of insurance companies (e.g., Sprecher and Pearl, 1983, Dividson et al., 1987, Shelor et al., 1990, Aiuppa et al., 2003, Blau et al., 2008). The empirical evidence suggests that insurers stock prices decline in response to the loss effect of hurricanes and the effect is particularly strong for insurers with more regional exposure (Lamb, 1995, 1998). Massa and Zhang (2011) show that the selling pressure of Katrina-exposed insurance companies induced an increase in the shortselling on the bonds held by those investors, and led to significant price drops up to seven months after the hurricane. In this context, we test how the presence of CDS contracts may help to reduce such impact on bond yield spreads and bond liquidity. We use the pre-katrina exposed insurance bond ownership to proxy for the selling pressure of the bonds after the hurricane 26. First, we identify the set of property & casualty insurance and reinsurance companies that are considered to have high exposure to Hurricane Katrina, using data from the Holborn Corporation (2005) Hurricane Katrina report 27. The Holborn Report lists the names of property & casualty (reinsurance) companies along with their 2004 market shares in the states of Louisiana, Mississippi, and Alabama, and whether they have rating or outlook changes immediately after the hurricane. We include the top ten property insurance companies by their market shares (including both personal and 600, ,000 catastrophe claims. In 2005, we received that number in a six week period immediately following Katrina. Since then, nearly 100 percent of all claims have been resolved. In total, State Farm has paid more than $3.1 billion in claims as a result of Katrina, which does not include payments to policy holders from the National Flood Insurance Program. 26 We focus on the pre-katrina property insurance ownership for the following reasons. First, it is exogenous with respect to the changes in bond liquidity and yield spreads given the total unexpectedness of insured damages. Second, the economic rationale can be explained with a simple example. Suppose that Start Farm Insurance has 10 billion bond holdings before Katrina, invested in two bonds, 8 billion in bond A and 2 billion in bond B. For each bond, the total issue outstanding is 100 billion. Therefore, before Katrina, State Farm s ownership in bond A is 8% and ownership in bond B is 2%. After Katrina, State Farm needs to immediately liquidate 5 billion to deal with insurance claims. Ideally, it would want to liquidate its bonds across the boards and keep the portfolio balanced. In this case, it should sell 4 billion in bond A and 1 billion in bond B. As a result, State Farm's ownership in bond A would drop from 8% to 4% and the ownership in bond B would drop from 2% to 1%. Therefore, the forced liquidation will have a much bigger impact on bond A than on bond B because of higher pre- Katrina ownership. In other words, higher exposed property ownership implies higher forced selling pressure after Katrina. 27 The Holborn report is publicly available at the URL: 20

24 commercial lines) and eight reinsurance companies with negative rating outlook changes. The names of those insurance companies are provided in the Appendix. 28 Then, we define the pre-katrina exposed insurance bond ownership as the par amounts held by property and reinsurance companies with high exposure to hurricane Katrina at the end of the second quarter of 2005 divided by the amount of bond issue outstanding. Non-exposed bond ownership is defined as the difference between total institutional ownership minus the exposed insurance ownership. In our sample of corporate bonds, the pre-katrina exposed property insurance ownership ranges from 0% (1-percentile) to 12% (99-percentile) of bond issue outstanding, with a mean of 1.3% and a standard deviation of 2.4%. Next, we regress the changes in bond yield spreads and the changes in bond liquidity around Katrina, on the pre-katrina exposed insurance ownership, a no CDS dummy as defined before, and the interaction term between them. Our variable of interest is the interaction term. Our hypothesis predicts a positive relationship between the interaction term and both the changes in yield spreads as well as the changes in bond illiquidity around Katrina. We report the results in Table VIII. In columns (1)-(3), the dependent variable is the change in bond yield spreads from Aug 23, 2005 to Sep 9, 2005 (the two weeks during which Hurricane Katrina formed and fully dissipated). In columns (4)-(6), the dependent variable is the difference of bond illiquidity between Sep 2005 and Aug In column (1), we only interact the pre-katrina exposed insurance ownership with the no CDS dummy. In column (2), we add the interaction term of nonexposed institutional bond ownership with the no CDS trading dummy. In column (3), we interact the exposed insurance ownership with bond characteristics including bond duration, offering amount and bond age. Columns (4)-(6) follow the same specifications as in columns (1)-(3), respectively. The results show a significantly positive coefficient on the interaction term between the no-cds contracts dummy and the pre-katrina exposed insurance ownership. This finding holds for both the 28 We obtain the data on institutional holdings of corporate bonds from Lipper s emaxx fixed income database. It is worth mentioning that we exclude those bond issuers that may be directly affected by the hurricane, which include life, property insurance and reinsurance companies, and firms headquartered in the states of Louisiana, Mississippi, and Alabama. 21

25 changes in yield spreads and the changes in bond illiquidity, and it is robust across different specifications. In the face of the selling pressure of Katrina-exposed property insurance companies, bonds without CDS contracts experience a 45% higher increase in yield spreads than that of bonds with CDS presence. Consistently, the impact of pre-katrina exposed insurance ownership on bond illiquidity is concentrated among bonds without CDS contracts. Again, this analysis confirms our hypothesis that CDS contracts reduces the fire sale effect both in terms of bond yield spreads and bond liquidity. Conclusion In this paper, we study the effect of credit default swap (CDS) on the corporate bond market. We argue that CDS, by reducing the need of investors to immediately sell the bonds in the face of credit deterioration of the bond issuer, reduces fire sale risk and provides bond liquidity. This effect should translate into lower bond yield spreads and higher bond liquidity. Given that bond investors are segmented by regulation, e.g., insurance companies, banks, and pension funds are subject to the riskbased capital requirements and in principle only hold investment grade bonds, we expect the liquidity provision role of CDS to be concentrated among investment grade bonds. We test these hypotheses using a comprehensive sample of US corporate bonds with CDS contracts information in the period. We show that the presence of CDS contracts reduces yield spreads and increases liquidity of investment grade bonds. This effect is stronger during the financial crisis period. We provide an instrumental variable identification that pins down the need for CDS contracts by exploiting the level of loan concentration of the banks from which the bond issuer borrows its bank debt. We find consistent results that the presence of CDS contracts reduces yield spreads and increases liquidity for investment grade bonds. No such effect is there for high yield bonds. We also examine two events that have been shown to trigger forced sales by bond investors: bond rating downgrades from investment grade to high yield, and the selling pressure of property 22

26 insurance companies following Hurricane Katrina. In both events, the presence of CDS contracts lowers the impact of fire sales by reducing the drop in bond liquidity and lowering the rise in yield spreads. Our findings provide a novel view of the role of credit derivatives on the underlying bond market, not limited to the monitoring or restructuring incentives of the lenders, but related to fire sale risk and bond liquidity. Our results also have important normative implications, as they suggest that at least for the class of investment grade bonds the presence of CDS contracts may actually help to reduce risk contagion around financial crises. 23

27 References Aiuppa, T. A., R. J. Carney, and T. M. Krueger, 1993, An Examination of Insurance Stock Prices Following the 1989 Loma Prieta Earthquake, Journal of Insurance Issues and Practices, 16:1-14. Allen, F., and E. Carletti, 2006, Credit Risk Transfer and Contagion, Journal of Monetary Economics, 53, Arping, S., 2004, Credit protection and lending relationships, Working paper, University of Amsterdam. Ashcraft, A.B., and J.A.C. Santos, 2009, Has the CDS Market Lowered the Cost of Corporate Debt?, Journal of Monetary Economics 564, Bao, J., Pan, J., Wang, J., 2011, The illiquidity of corporate bonds, Journal of Finance 66, Beyhaghi, M., and N. Massoud, 2011, Why and how do banks lay off credit risk? The choice between loan sales and credit default swaps, Working paper, York University. Bernardo, A., and I. Welch, 2004, Liquidity and Financial Market Runs, Quarterly Journal of Economics 119-1, Berndt, A., R. A. Jarrow, and C. Kang, 2006, Restructuring Risk in Credit Default Swaps: An Empirical Analysis, Working Paper, Carnegie Mellon University. Blanco, R., S. Brennan, and I.W. Marsh, 2005, An Empirical Analysis of the Dynamic Relation between Investment-Grade Bonds and Credit Default Swaps, Journal of Finance 605, Blau, M. B., R.A. Van Ness, and C. Wade, 2008, Capitalizing on Catastrophe: Short Selling Insurance Stocks around Hurricanes Katrina and Rita, Journal of Risk and Insurance, 75: Boehmer, E., S. Chava, and H.E. Tookes, 2010, Capital Structure, Derivatives and Equity Market Quality, SSRN working paper. Bolton, P., and M. Oehmke, 2010, Credit Default Swaps and The Empty Creditor Problem, Review of Financial Studies, Forthcoming. Cao, C., F. Yu, and Z. Zhong, 2010, The Information Content of Option-Implied Volatility for Credit Defaul Swap Valuation, Journal of Financial Markets 13, Campbell, J. and G. B.Taksler, 2003, Equity Volatility and Corporate Bond Yields, The Journal of Finance, 58, Chen, R. R, Fabozzi, F., and Sverdlove, R., 2010, Corporate Credit Default Swap Liquidity and its Implications for Corporate Bond Spreads, The Journal of Fixed Income, 202, Choudhry, M., 2006, The Credit Default Swap Basis: Illustrating Positive and Negative Basis Arbitrage Trades, market research working paper. Coval, J. and E. Stafford, 2007, Asset Fire Sales (and Purchases) in Equity Markets, Journal of Financial Economics 86, Damodaran, A., and M.G. Subrahmanyam, 1992, The effects of derivative securities on the markets for the underlying assets in the United States: A survey, Financial Markets, Institutions and Instruments 1, Da, Z., and P. Gao, 2009, Clientele Change, Persistent Liquidity Shock, and Bond Return Reversals After Rating Downgrades, Working Paper. Das, S., M. Kalimipalli, and S. Nayak, 2011, Did CDS contracts improve the market for corporate bonds, Working paper, Santa Clara University and Wilfrid Laurier University. Duffee, G. R., and C. Zhou, 2001, Credit derivatives in banking: Useful tools for managing risk?, Journal of Monetary Economics, 48(1), Duffie, D., and H. Zhu, 2009, Does a Central Clearing Counterparty Reduce Counterparty Risk?, Working Paper. Ericsson, J., and O., Renault, 2006, Liquidity and Credit Risk, Journal of Finance 61, ; 24

28 Ellul, A., C. Jotikasthira, and C. T. Lundblad, 2011, Regulatory Pressure and Fire Sales in the Corporate Bond Market, Journal of Financial Economics, 2011, 101, Elkamhi, R., R. S. Pungaliya, and A. M. Vijh, 2010, Do Firms Have a Target Leverage? Evidence from Credit Markets, Working Paper. Ewing, B. T., S. E. Hein, and J. B. Kruse, 2006, Insurer Stock Price Responses to Hurricane Floyd: An Event Study Analysis Using Storm Characteristics, Weather and Forecasting, 21: Fitch, Inc, 2004, CDS Market Liquidity: Show Me the Money, FitchRatings; Gopalan, R., V. Nanda, and V. Yerramilli, 2011, Does poor performance damage the reputation of financial intermediaries? Evidence from the loan syndication market, Journal of Finance, forthcoming. Gormley, T., N. Gupta, and A. Jha, 2011, Corporate bankruptcy and creditor incentives, Working paper, University of Pennsylvania, Indiana University and Texas A&M International University. Han, S., and Zhou, H., 2008, Effects of Liquidity on the Nondefault Component of Corporate Yield Spreads: Evidence from Intraday Transactions Data, Working paper, Federal Reserve Board. Hirtle, B, 2008, Credit Derivatives and Bank Credit Supply, Working Paper, Federal Reserve Bank of New York. Holborn Corporation, 2005, Katrina: Market Insured Losses. Hu, H. T. C., and B. Black The New Vote Buying: Empty Voting and Hidden (Morphable) Ownership. Southern California Law Review 79: Hu, H. T. C., and B. Black Hedge Funds, Insiders, and the Decoupling of Economic and Voting Ownership: Empty Voting and Hidden (Morphable) Ownership. Journal of Corporate Finance 13: Hu, H. T. C., and B. Black. 2008a, Debt, Equity, and Hybrid Decoupling: Governance and Systemic Risk Implications. European Financial Management 14: b. Hu, H. T. C., and B. Black. 2008b, Equity and Debt Decoupling and Empty Voting II: Importance and Extensions. University of Pennsylvania Law Review 156: Hull, J., Predescu, M., and White, A., 2004, The Relationship between Credit Default Swap Spreads, Bond Yields, and Credit Rating Announcements, Journal of Banking and Finance, 28, ; Ismailescu, I., and B. Phillips, 2011, Savior or Sinner: Credit Default Swaps and the Market for Sovereign Debt, Working paper, University of Waterloo. Kahan, M., and E. B. Rock, 2007, Hedge Funds in Corporate Governance and Corporate Control, University of Pennsylvania Law Review, 155(5), Kim, G.H., 2011, Credit Default Swap, Strategic Default, and the Cost of Corporate Debt, Working Paper. Kwan, S, 1996, Firm-Specific Information and the Correlation between Individual Stocks and Bonds, Journal of Financial Economics 401, Lamb, R. P., 1995, An Exposure-Based Analysis of Property-Liability Insurer Stock Values Around Hurricane Andrew, Journal of Risk and Insurance, 62: Lamb, R. P., 1998, An Examination of Market Efficiency Around Hurricanes, Financial Review, 33: Lando, D., and M.S. Nielsen, 2010, Correlation in corporate defaults: Contagion or conditional independence?, Journal of Financial Intermediation 19, Longstaff, F.A., K.G., S. Schaefer, and I. Strebulaev, 2011, Corporate bond default risk: A 150-year perspective, Journal of Financial Economics, forthcoming. Longstaff, F.A., Mithal, S. and Neis, E., 2005, Corporate Yield Spreads: Default Risk or Liquidity? New Evidence from the Credit Default Swap Market, Journal of Finance, 60, Massa, M., and L. Zhang, 2011, The Spillover Effects of Hurricane Katrina and the Choice Between Bank Loans and Bonds, Working Paper. Minton, B.A., R.M. Stultz, and R.Williamson, 2009, How much do banks use credit derivatives to hedge loans?, Journal of Financial Services Research 35,

29 Morrison, A.D., 2005, Credit derivatives, disintermediation, and investment decisions, Journal of Business 78, Nashikkar, A., M.G. Subrahmanyam, and S. Mahanti, 2011, Liquidity and arbitrage in the market for credit risk, Journal of Financial and Quantitative Analysis 46, Norden, L., and W. Wagner, 2008, Credit Derivatives and Loan Pricing, Journal of Banking and Finance 3212, Norden, L., and M. Weber, 2009, The Co-movement of Credit Default Swap, Bond and Stock Markets: An Empirical Analysis, European Financial Management 153, Parlour, C.A., and G. Plantin, 2008, Loan Sales and Relationship Banking, Journal of Finance, 63(3), Parlour, C. A., and A. Winton, 2008, Laying off Credit Risk: Loan Sales versus Credit Default Swaps, Working Paper, UC Berkeley. Peristiani, S., and V. Savino, 2011, Are Credit Default Swaps Associated with Higher Corporate Defaults? New YorK Fed Staff Report, 494. Purnanandam, A., 2011, Originate-to-distribute model and the subprime mortgage crisis, Review of Financial Studies 24, Qiu, J., and F. Yu, 2012, Endogenous Liquidity in Credit Derivatives, Journal of Financial Economics 103, Saretto, A., and H. Tookes, 2011, Corporate leverage, debt maturity and credit default swaps: The role of credit supply, Working paper, University of Texas at Austin and Yale University. Shelor, R. M., D. C. Anderson, and M. L. Cross, 1992, Gaining From Loss: Property- Liability Insurer Stock Values in the Aftermath of the 1989 California Earthquake, Journal of Real Estate Research, 59: Shleifer, A., 1986, Do Demand Curves for Stocks Slope Down?, Journal of Finance, 41, Shleifer, A., and R.W. Vishny, 1997, The Limits of Arbitrage, Journal of Finance, 52, Shleifer, A., and R.W., Vishny, 2011, Fire Sales in Finance and Macroeconomics. Journal of Economic Perspectives 25, Sprecher, C. R., and M. A. Pertl, 1983, Large Losses, Risk Management, and Stock Prices, Journal of Risk and Insurance, 50: Subrahmanyam, M.G., D.Y. Tangz, and S.Q. Wang, 2011, Does the Tail Wag the Dog? The Effect of Credit Default Swaps on Credit Risk, Working Paper. Sufi, A., 2009, The real effects of debt certification: Evidence from the introduction of bank loan ratings. Review of Financial Studies 22: Tang, D.Y., and H. Yan, 2011, What moves CDS spreads?, Working paper, University of Hong Kong, University of South Carolina and Shanghai Advanced Institute of Finance. Towers Perrin Corporation, 2005, Hurricane Katrina: Analysis of the Impact on the Insurance Industry. Yavorsky, A., 2009, Analyzing the Potential Impact of Credit Default Swaps in Workout Situations, Special Comment, Moody.s Investor Services. Wooldrige, J., 2001, Econometric Analysis of Cross Section and Panel Data. The MIT Press. Zhang, B. Y., H. Zhou, and H. Zhu, 2008, Explaining Credit Default Swap Spreads with Equity Volatility and Jump Risks of Individual Firms, Review of Financial Studies 22,

30 Appendix: Variable Definitions Option-adjusted yield spreads: the number of percentage points that the fair value of the treasury spot curve is shifted to match the present value of the discounted cash flows to the bond s price. For securities with embedded options, such as callability, a log normal short interest rate model is used to evaluate the present value of the securities potential cash flows. In this case, the option-adjusted spread is equal to the number of percentage points that the short interest rate tree must be shifted to match the discounted cash flows to the bond s price. Bond illiquidity: for each bond-month, it is defined as the implied bid-ask spread based on the auto-covariances of bond price changes: 2 γ (0 if γ 0), where γ Cov p, p ) and p is the log bond price ( clean price ) at time t. We use the tick-by-tick transaction data from TRACE to calculate the price changes. We require the number of bond transactions to be larger than 10 for each bond-month. Coupon rate: the interest rate paid on a bond as a percentage of the issuing amount (par value). Duration: the average maturity of a bond s cash flows. Offering amount: the dollar amount of bond issuing outstanding. Callability: a dummy variable equal to 1 if the bond is callable. A callable bond gives the issuer the right to early redemption at a given price (redemption price) or a given date (call date). Fungibility: a dummy variable equal to 1 if the bond is fungible. Fungible bonds can be reopened in the future by increasing the total amount outstanding of the issue. Credit enhancement: a dummy variable equal to 1 if the bond has credit enhancements, e.g., guarantees, letters of credit, etc. Bond age: the number of years since the bond issuing date. Bond rating fixed effects (issue level): 21 credit rating dummies, each corresponding to the current month composite rating (simple averages of ratings from Moody s, S&P and Fitch) from AAA to CCC3. The rating correspondences are detailed below. Numeric Composite Moody's S&P Fitch 1 AAA Aaa AAA AAA 2 AA1 Aa1 AA+ AA+ 3 AA2 Aa2 AA AA 4 AA3 Aa3 AA- AA- 5 A1 A1 A+ A+ 6 A2 A2 A A 7 A3 A3 A- A- 8 BBB1 Baa1 BBB+ BBB+ 9 BBB2 Baa2 BBB BBB 10 BBB3 Baa3 BBB- BBB- 11 BB1 Ba1 BB+ BB+ 12 BB2 Ba2 BB BB 13 BB3 Ba3 BB- BB- 14 B1 B1 B+ B+ 15 B2 B2 B B 16 B3 B3 B- B- 17 CCC1 Caa1 CCC+ CCC+ 18 CCC2 Caa2 CCC CCC 19 CCC3 Caa3 CCC- CCC- 20 CC Ca CC CC 21 C C C C Investment grade: a dummy variable indicating that the bond credit rating is above or equal to BBB3. High yield: a dummy variable indicating that the bond credit rating is below BBB3. CDS presence: a dummy variable equal to 1 if the issuing firm has quoted CDS contracts on its bonds in the previous month and 0 otherwise. 27

31 Equity volatility: for each stock-month, it is the standard deviation of daily stock returns in the month. Equity beta: for each stock-month (i,t), we estimate the factor loadings by running the following regression: ris, rf, s ait, 1 it, 1 ( rms, rf, s) is,, where we use the previous 180 days as the estimation period, and we require a minimum of 90 observations for each regression. The dependent variable is the daily return of firm i at day s less the risk-free rate r f, s. The independent variable is the excess return of market portfolio over the risk-free rate ( r m, s rf, s ). Market value of assets: stock price (data199) * shares outstanding (data25) + short term debt (data34) + long term debt (data9) + preferred stock liquidation value (data10) deferred taxes and investment tax credits (data35). Market-to-Book Ratio: market value of assets/book assets (data6) Total debt: long term debt (data9) + short term debt (data34) Book leverage: total debt/book assets (data6) Firm size: log (book assets) (data6) Profitability: income before extraordinary items (data20)/book assets (data6) Cash holding: cash and short-term investments (data1/data6) Dividend payer: a dummy variable equal to 1 if the firm pays cash dividends in the year Industry fixed effects: the two-digit SIC industry dummies Banks use of derivatives for hedging: the Bank Regulatory database contains off-balance sheet data on the banks use of derivatives for hedging purposes from the CALL reports. Operationally, we use the following data items: credit derivatives (RCFDA535, notional amount of credit derivatives on which the reporting bank is the beneficiary); FX derivatives (RCFD8726, notional amount of foreign exchange derivative contracts marked to market, with purposes not trading); Interest rate derivatives (RCFD8725, notional amount of interest rate derivative contracts marked to market, with purposes not trading.). Bank size is defined as the log value of total assets (RCFD2170). We require the bank s total amount of commercial and industrial loans (RCON1766) to be larger than $100 million. Exposed insurance companies to Hurricane Katrina: we follow the methodology of Massa and Zhang (2011) to identify Katrina-exposed property insurance and reinsurance companies. The set of property & casualty insurance and reinsurance companies that are considered to have high exposure to Hurricane Katrina is identified using data from the Holborn Corporation (2005) Hurricane Katrina report, publicly available at the URL: The Holborn Report lists the names of property & casualty (re)insurance companies along with their 2004 market shares in the states of Louisiana, Mississippi, and Alabama, and whether they have rating or outlook changes immediately after the hurricane. We include the top ten property insurance companies by their market shares (including both personal and commercial lines) and eight reinsurance companies with negative rating (outlook) changes that can be identified in Lipper/EMAXX as managing firms. These firms are: State Farm Insurance Company, Allstate Insurance Co Group, Progressive Casualty Group, Alfa Insurance, Mississippi Farm Bureau Casualty Insurance, United Services Automobile Association, Nationwide Assurance, American Modern Home Insurance, American International Insurance, St. Paul Travelers Companies, Ace American Reinsurance, Alea North America Insurance, Endurance Reinsurance Corp of America, Odyssey America Reinsurance, Olympus Insurance, Partner Reinsurance United States, Transatlantic Reinsurance United States. 28

32 Table I Summary Statistics In this table, we present summary statistics of the major variables used in later analysis. Our data come from multiple sources. The data on bond yields spread, duration and issue-level credit ratings come from Bank of America-Merrill Lynch Corporate Master Index Compositions. Additional bond characteristics including coupon rate, bond offering amount, callability, fungibility, credit enhancement come from Mergent FISD. We obtain information on bond transactions from the Trade Reporting and Compliance Engine database (TRACE). We get the information on CDS contracts from the Markit CDS database. Firm-level stock return and accounting information come from CRSP and Compustat. We include equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. Our combined sample include bond-month observations (3468 firm-year observations) from January 2001 to December For each variable, we report the data frequency, source, number of observations, mean and stand deviation. The detailed definitions of each variable can be found in the Appendix. Bond Characteristics Frequency Source N Mean Std. Dev. Option-adjusted spread (%) Month BofA-Merrill Lynch Duration Month BofA-Merrill Lynch Investment grade Month BofA-Merrill Lynch Log(offering amount) Month Mergent FISD Coupon rate Month Mergent FISD Callability Month Mergent FISD Fungibility Month Mergent FISD Credit enhancement Month Mergent FISD Bond age Month Mergent FISD Bond illiquidity Month TRACE Firm Characteristics CDS presence Month Markit Equity volatility Month CRSP Equity beta Month CRSP Book size Year Compustat Market-to-book Year Compustat Book leverage Year Compustat Profitability Year Compustat Cash holding Year Compustat Dividend payer Year Compustat

33 Table II CDS Presence and Bond Yield Spread In this table, we link the presence of CDS contracts to corporate yield spreads. The dependent variable is the option-adjusted (OA) yield spread, defined as the number of percentage points that the treasury spot curve must be shifted in order to match discounted cash flows to the bond s price. Our interested variable is CDS presence, a dummy variable equal to 1 if the issuing firm has quoted CDS contracts on its bonds in the previous month and 0 otherwise. In Panel A, columns (1)-(3) are for the subsample of investment grade bonds, while columns (4)-(6) are for the subsample of high yield bonds. In column (1), we control for bond characteristics including coupon rate, duration, offering amount, callability, fungibility and credit enhancement, time fixed effects at the monthly level, and credit rating fixed effects at the issue level. In column (2), we control for time credit rating fixed effects. In column (3), we control for firm characteristics including equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. Columns (4)-(6) follow the same specifications as in columns (1)-(3), respectively. The detailed definitions of each variable can be found in the Appendix. All firm-year (month) variables are taken at the end of the previous year (month). We control for industry fixed effects at the two-digit SIC level, and always cluster the standard errors at the firm level. In Panel B, we interact the variable of CDS presence with a dummy variable indicating the financial crisis period. The crisis period includes year 2001, 2002, 2008 and the first half of Columns (1)-(3) are for the subsample of investment grade bonds. Columns (4)-(6) are for the subsample of high yield bonds. We follow the same specifications as in Panel A. We always include time (monthly) fixed effects therefore the crisis period dummy is dropped out of the regression. ***, ** and * represent significance levels at 1%, 5% and 10% respectively using heteroscedasticity robust standard errors with t-statistics given in parentheses. 30

34 Table II (Cont d) Panel A: Main Regression Dep: Bond Yield Spread Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence *** *** *** ** (-3.40) (-3.28) (-5.43) (1.01) (2.25) (0.52) Controls Coupon rate 0.060*** 0.068*** 0.057*** 0.243*** 0.196*** 0.177*** (5.00) (6.26) (6.02) (2.65) (2.68) (2.70) Duration 0.022*** 0.024*** 0.026*** *** *** *** (7.56) (8.33) (10.69) (-4.23) (-3.99) (-4.42) Log(offering amount) (-0.05) (-0.40) (-1.02) (-0.10) (-0.13) (-0.55) Callability (-1.07) (-0.38) (0.60) (-0.47) (-0.50) (-0.19) Fungibility ** ** ** (-2.21) (-2.30) (-2.05) (1.37) (1.60) (1.34) Credit enhancement (-0.93) (-0.65) (-0.75) (-0.52) (0.04) (0.38) Bond age (-1.37) (-1.28) (-0.16) (-0.75) (0.08) (-0.01) Equity volatility *** *** (11.99) (10.07) Equity beta 0.172*** (3.96) (1.18) Book size (-1.35) (-0.97) Market-to-book ** *** (-2.06) (-3.29) Book leverage 0.565*** 2.470*** (3.03) (3.21) Profitability ** (-2.24) (-1.63) Cash holding (-1.50) (0.51) Dividend payer 0.160** (2.45) (-0.23) Industry (two-digit SIC) FE Y Y Y Y Y Y Time (monthly) FE Y - - Y - - Credit rating (issue level) FE Y - - Y - - Time Credit rating FE - Y Y - Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 131, , ,202 26,920 26,920 26,920 R-squared Panel B: Interaction with Financial Crisis Period Dep: Bond Yield Spread Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence * *** (-0.83) (-1.65) (-3.18) (0.52) (1.07) (0.13) CDS presence * Crisis period ** ** *** (-2.52) (-2.12) (-3.19) (0.63) (1.36) (0.51) Same Specifications as Panel A Y Y Y Y Y Y Number of Obs. 131, , ,202 26,920 26,920 26,920 R-squared

35 Table III CDS Presence and Bond Illiquidity In this table, we link the presence of CDS contracts to bond illiquidity. The dependent variable is the implied bid-ask spread based on the auto-covariances of bond price changes: 2 γ (0 if γ 0), where γ Cov p, p ) and p is the log price at time t. Our interested variable is CDS presence, a dummy variable equal to 1 if the issuing firm has quoted CDS contracts on its bonds in the previous month and 0 otherwise. In Panel A, columns (1)-(3) are for the subsample of investment grade bonds, while columns (4)-(6) are for the subsample of high yield bonds. In column (1), we control for bond characteristics including coupon rate, duration, offering amount, callability, fungibility and credit enhancement, time fixed effects at the monthly level, and credit rating fixed effects at the issue level. In column (2), we control for time credit rating fixed effects. In column (3), we control for firm characteristics including equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. Columns (4)-(6) follow the same specifications as in columns (1)-(3), respectively. The detailed definitions of each variable can be found in the Appendix. All firm-year (month) variables are taken at the end of the previous year (month). We control for industry fixed effects at the two-digit SIC level, and always cluster the standard errors at the firm level. In Panel B, we interact the variable of CDS presence with a dummy variable indicating the market crisis period. The crisis period covers year 2001, 2002, 2008 and the first half of Columns (1)-(3) are for the subsample of investment grade bonds. Columns (4)-(6) are for the subsample of high yield bonds. We follow the same specifications as in Panel A. We always include time (monthly) fixed effects therefore the crisis period dummy is dropped out of the regression. ***, ** and * represent significance levels at 1%, 5% and 10% respectively using heteroscedasticity robust standard errors with t-statistics given in parentheses. 32

36 Table III (Cont d) Panel A: Main Regression Dep: Bond Illiquidity Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence *** *** *** 0.327*** 0.279*** 0.184* (-3.02) (-2.65) (-3.00) (2.86) (2.66) (1.76) Controls Coupon rate *** *** *** ** ** ** (-4.22) (-3.71) (-4.05) (-2.36) (-2.49) (-2.12) Duration 0.144*** 0.146*** 0.146*** 0.128*** 0.134*** 0.135*** (26.70) (26.63) (26.25) (8.25) (7.62) (7.53) Log(offering amount) *** *** *** *** *** *** (-6.24) (-6.91) (-8.65) (-3.29) (-2.79) (-3.39) Callability 0.068** 0.079** 0.092*** (2.13) (2.55) (2.78) (0.90) (0.77) (1.25) Fungibility * (-0.28) (-0.41) (-0.01) (1.24) (1.50) (1.67) Credit enhancement (-0.92) (-1.23) (-1.36) (0.25) (-0.10) (0.17) Bond age 0.061*** 0.061*** 0.062*** 0.081*** 0.080*** 0.076*** (13.18) (12.81) (13.22) (4.24) (5.29) (4.77) Equity volatility *** *** (8.31) (5.18) Equity beta (-1.11) (0.40) Book size (0.03) (1.36) Market-to-book ** (-2.17) (-0.40) Book leverage 0.392*** (2.79) (1.22) Profitability (0.59) (-0.70) Cash holding (1.03) (0.42) Dividend payer (1.33) (0.85) Industry (two-digit SIC) FE Y Y Y Y Y Y Time (monthly) FE Y - - Y - - Credit rating (issue level) FE Y - - Y - - Time Credit rating FE - Y Y - Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 64,024 64,024 64,024 9,356 9,356 9,356 R-squared Panel B: Interaction with Financial Crisis Period Dep: Bond Illiquidity Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence (-0.89) (-0.86) (-1.34) (0.95) (1.48) (0.87) CDS presence * Crisis period ** *** *** 0.607** 0.336* (-2.49) (-3.27) (-3.74) (2.39) (1.75) (1.24) Same Specifications as Panel A Y Y Y Y Y Y Number of Obs. 64,024 64,024 64,024 9,356 9,356 9,356 R-squared

37 Table IV Instrumental Regression In this table, we perform an instrumental regression analysis to examine the impact of CDS presence on bond yield spread and bond illiquidity. We intend to use the level of loan concentration of the lending banks, from which the bond issuer borrows its bank debt, to identify the demand for CDS contracts, i.e., as an instrument for the presence of CDS contracts. In Panel A, to justify the loan concentration-based instrument, we examine the concentration of a bank s loan portfolio across different industries and geographical regions, and relate it to the bank s use of credit derivatives, foreign exchange derivatives and interest rate derivatives for hedging purposes. The analysis is done at the bank level. First, we construct banks loan concentration based on the data from LPC Dealscan. We only focus on bank borrowers in the US. For each bank-year, we classify its existing loans into different industry (two digit SIC) state pairs. We then calculate the herfindal across those pairs as the concentration of the bank s loan portfolio. Next, we perform name matching by bank names to link LPC Dealscan with the Bank Regulatory database, which contains the off-balance sheet data on banks use of derivatives for hedging purposes. We require the bank s total amount of commercial and industrial loans (RCON1766) to be larger than $100 million. The dep. var. in columns (1)-(2) is the log value of the notional amount of credit derivatives (RCFDA535). The dep. var. in columns (3)-(4) is the notional amount of credit derivatives divided by the bank s total asset (RCFDA535/RCFD2170). The dep. var. in column (5) is the notional amount of foreign exchange derivatives divided by the bank s total asset (RCFD8726/RCFD2170), while the dep. var. in column (6) is the notional amount of interest rate derivatives divided by the bank s total asset (RCFD8725/ RCFD2170). Bank size is the log value of total asset (RCFD2170). Detailed descriptions of each data item are given in the appendix. In Panel B, we link the presence of CDS contracts to bond yield spread and bond illiquidity, with CDS presence instrumented by the loan concentration of the lending banks. We proceed as follows. First, at the issuer level, we calculate the loan herfindal as the value (loan amount)-weighted bank herfindal among all the banks from which the issuer borrows in the last 5 years. In columns (1) and (2), we run a probit regression of the CDS presence dummy on loan herfindal. We perform the F-test to identify the weakness of the loan herfindal variable. Then, we calculate the fitted value from column (2) and use it as the instrument for CDS presence in columns (3)-(6). The dep. var. in columns (3)-(4) is the bond yield spread, while the dep. var. in columns (5)-(6) is the bond illiquidity. Columns (3) and (5) are for the subsample of investment grade bonds, while columns (4) and (6) are for the subsample of high yield bonds. In all the specifications, we include industry, time and credit rating fixed effects, and cluster the standard errors at the issuer level. Panel C follows the same specifications as in Panel B, except that the instrument is the average loan herfindal among issuers with same industry (two digit SIC), region (state) and basic rating category (investment grade/high yield). ***, ** and * represent significance levels at 1%, 5% and 10% respectively using heteroscedasticity robust standard errors with t-statistics given in parentheses. Dep. var.: Panel A: Banks Loan Concentration and the Use of Derivatives for Hedging Credit Derivatives (log of notional amount) Credit Derivatives (notional amount/assets) FX (notional amount/assets) Interest Rate (notional amount/assets) (1) (2) (3) (4) (5) (6) Loan herfindal 3.318*** 3.318*** 0.051*** 0.051*** (8.97) (4.70) (4.39) (3.06) (0.20) (0.08) Bank size 2.695*** 2.695*** 0.029*** 0.029*** 0.001*** 0.082*** (22.26) (9.27) (4.98) (3.13) (2.99) (3.28) Year FE Y Y Y Y Y Y Clustering - Bank - Bank Bank Bank Number of obs R-squared

38 Table IV (Cont d) Panel B: Instrumental Regression (Loan Herfindal) Dep.: CDS Presence Dep.: Bond Yield Spread Dep.: Bond Illiquidity First Stage (Probit Regression) Investment Grade High Yield Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence (instrumented by *** *** fitted value from first stage) (-3.81) (1.14) (-2.71) (1.35) Controls Coupon rate *** *** 0.301** *** *** (-3.29) (-1.23) (3.49) (2.59) (-4.06) (-2.68) Duration *** *** 0.135*** 0.118*** (0.96) (-0.12) (7.34) (-3.78) (21.70) (6.76) Log(offering amount) 0.383*** * *** *** (4.09) (0.39) (-1.81) (0.57) (-7.84) (-3.01) Callability *** * ** (-3.26) (-1.43) (-1.71) (-1.13) (1.99) (-0.44) Fungibility *** (0.55) (0.36) (-0.99) (2.71) (-0.19) (0.92) Credit enhancement *** *** ** * (-4.68) (-4.52) (-2.12) (-0.68) (-1.84) (0.24) Bond age 0.039** *** 0.074*** (2.33) (-0.40) (-0.95) (1.65) (11.93) (4.76) Equity volatility *** *** *** *** (0.77) (9.84) (8.03) (6.45) (4.65) Equity beta 0.275** 0.174*** (2.49) (3.29) (1.31) (-1.64) (-0.41) Book size 0.435*** (5.72) (0.15) (-0.59) (-0.23) (0.34) Market-to-book ** (0.21) (-1.32) (-2.07) (-1.35) (0.05) Book leverage 1.624*** 0.730*** * (2.67) (3.67) (0.55) (1.79) (-0.86) Profitability * *** (-1.76) (-3.33) (0.04) (-0.94) (0.96) Cash holding (0.96) (-0.45) (0.80) (-0.27) (0.23) Dividend payer 0.477** (2.55) (1.57) (-0.82) (0.43) (-0.93) Loan herfindal 2.028*** 2.074*** (4.81) (4.08) Industry, Time, Credit rating FE Y Y Y Y Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 136, , ,712 21,674 56,391 7,460 F-test Panel C: Instrumental Regression (Average Loan Herfindal) Dep. Dep.: CDS Presence Dep.: Bond Yield Spread Dep.: Bond Illiquidity First Stage (Probit Regression) Investment Grade High Yield Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence (instrumented by fitted value from first stage) Average loan herfindal 2.567*** 2.659*** (4.87) (5.01) *** *** (-3.80) (1.13) (-3.76) (1.31) Same Specifications as Panel B Y Y Y Y Y Y Number of Obs. 136, , ,712 21,674 56,391 7,460 F-test

39 Table V Robustness Check I: Instrumental Regression Using 5-year CDS Contracts In this table we perform a robustness check to the IV results in Table IV. In particular, we redefine CDS presence as a dummy variable equal to 1 if the issuing firm has the most liquid CDS contracts trading in the previous month and 0 otherwise, i.e., CDS contracts with 5-year maturity and MR restructuring clause. Panel A and Panel B follow the same specifications as in Panel B and Panel C of Table IV, respectively. For brevity we only report the interested variables. Panel A: Instrumental Regression (Loan Herfindal) Dep.: CDS Presence Dep.: Bond Yield Spread Dep.: Bond Illiquidity First Stage (Probit Regression) Investment Grade High Yield Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence (instrumented by fitted value from first stage) Loan herfindal 1.797*** 1.882*** (4.69) (4.03) *** *** (-3.73) (0.77) (-2.74) (1.29) Same Specifications as in Table Y Y Y Y Y Y IV, Panel A F-test Panel B: Instrumental Regression (Average Loan Herfindal) Dep. Dep.: CDS Presence Dep.: Bond Yield Spread Dep.: Bond Illiquidity First Stage (Probit Regression) Investment Grade High Yield Investment Grade High Yield (1) (2) (3) (4) (5) (6) CDS presence (instrumented by fitted value from first stage) Average loan herfindal 2.278*** 2.423*** (4.65) (4.89) *** *** (-3.82) (0.77) (-3.89) (1.26) Same Specifications as in Table Y Y Y Y Y Y IV, Panel B F-test

40 Table VI Robustness Check II: Instrumental Regression Using CDS Composite Depth In this table, we perform another robustness check, by focusing the subsample of bonds issued by firms with CDS contracts trading in the market. Specifically, we link the depth of CDS contracts to bond yield spreads and bond illiquidity. Following Qiu and Yu (2012), we use the number of dealers providing CDS quotes as a proxy for the depth of CDS contract. The Markit data only provide information on the number of dealers in the 5-year maturity contracts. Therefore we define CDS composite depth as the log number of dealers in the CDS contracts with 5-year maturity. Qiu and Yu (2012) show that CDS depth is significantly related to the endogenous liquidity provision by informed financial institutions. We therefore rely on the instrumental regression instead of a simple OLS regression, with CDS composite depth instrumented by the loan concentration of the lending banks. We construct the instrument based on the bank herfindal in the same way as described in Table IV. In Panel A, in columns (1) and (2), we run OLS regressions of the CDS composite depth on the loan herfindal. We perform the F-test to identify the weakness of the loan herfindal variable. Then, we use it as the instrument for the CDS composite depth in columns (3)-(6). The dep. var. in columns (3)-(4) is the bond yield spread, while the dep. var. in columns (5)-(6) is the bond illiquidity. Columns (3) and (5) are for the subsample of investment grade bonds. Columns (4) and (6) are for the subsample of high yield bonds. In all of the specifications, we control for industry, time and credit rating fixed effects, and cluster the standard errors at the issuer level. Panel B follows the same specifications as in Panel A, except that the instrument is the average loan herfindal among issuers in the same industry (two digit SIC), region (state) and basic rating category (investment grade/high yield). ***, ** and * represent significance levels at 1%, 5% and 10% respectively using heteroscedasticity robust standard errors with t-statistics given in parentheses. CDS composite depth (instrumented by loan herfindal) Panel A: Instrumental Regression (Loan Herfindal) Dep.: CDS Composite Depth Dep.: Bond Yield Spread Dep.: Bond Illiquidity First Stage (OLS Regression) Investment Grade High Yield Investment Grade High Yield (1) (2) (3) (4) (5) (6) *** ** (-3.50) (-0.53) (-2.44) (-0.15) Loan herfindal 0.438*** 0.427*** (4.69) (3.35) Other controls Y Y Y Y Y Y Industry, Time, Credit rating FE Y Y Y Y Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 117, , ,531 16,858 51,527 6,491 F-test CDS composite depth (instrumented by average loan herfindal) Panel B: Instrumental Regression (Average Loan Herfindal) Dep.: CDS Composite Depth Dep.: Bond Yield Spread Dep.: Bond Illiquidity First Stage (OLS Regression) Investment Grade High Yield Investment Grade High Yield (1) (2) (3) (4) (5) (6) *** *** (-3.12) (-0.33) (-3.22) (0.44) Average loan herfindal 0.490*** 0.505*** (4.87) (3.56) Same Specifications as Panel B Y Y Y Y Y Y Number of Obs. 117, , ,531 16,858 51,527 6,491 F-test

41 Table VII CDS Presence and the Impact of Fallen Angels In this table, we focus on a particular sample of bond-months (bond-quarters) during which the bond experiences rating changes. We examine how the presence of CDS contracts may alter the impact of fallen angels on bond institutional ownership, yield spreads and bond illiquidity. For a bond-month (bond-quarter), it is defined as a fallen angel if the bond is downgraded from investment grade to high yield. Our interested variable is the interaction term between the fallen angel dummy and the No CDS dummy (1- CDS presence dummy). Panel A: Change in Institutional Bond Ownership In Panel A, we focus on the changes in bond institutional ownership around the quarter of rating changes. The data on quarterly institutional bond holdings are from Lipper s emaxx fixed income database. Bond institutional ownership is calcualted as the total institutional holdings divided by the bond issue outstanding. The dependent variable is the change in bond institutional ownership relative to the previous quarter. Columns (1)-(3) are based on the full sample of rating changes including both rating downgrades and rating upgrades. In column (1), we only interact the fallen angel dummy with the No CDS presence dummy. In column (2), we add the interaction terms of fallen angel with bond characteristics including bond duration, offering amount and bond age. In column (3), we further interact the fallen angel with risk characteristics such as stock volatility and beta. Columns (4)-(6) follow the same specifications as in columns (1)-(3) but based on the subsample of rating downgrades. For brevity, in columns (3) and (6), we don t report the firm-level controls, which include equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. We always include time (quarterly) credit rating fixed effects, industry fixed effects at the two-digit SIC level, and cluster the standard errors at the issuer level. Dep: Change in bond ownership around rating changes Full Sample on Rating Change Sub-sample on Rating Downgrade (1) (2) (3) (4) (5) (6) Fallen angel *** ** (-4.03) (-0.48) (-0.25) (-2.36) (-0.92) (-0.40) Fallen angel * No CDS ** ** ** *** *** *** (-2.27) (-2.21) (-2.24) (-2.93) (-2.84) (-2.97) Controls No CDS * (-1.57) (-1.56) (-1.80) (0.27) (0.27) (-0.08) Coupon rate (0.54) (0.58) (0.47) (0.33) (0.34) (0.34) Duration 0.001** 0.001** 0.001** 0.001* 0.001* 0.001* (2.42) (2.42) (2.42) (1.67) (1.74) (1.80) Log(offering amount) (1.10) (1.13) (1.28) (1.57) (1.47) (1.39) Callability (0.71) (0.70) (0.50) (1.02) (1.01) (0.95) Fungibility (0.66) (0.65) (0.53) (0.42) (0.41) (0.24) Credit enhancement (0.64) (0.62) (0.72) (-1.34) (-1.35) (-1.00) Bond age *** *** *** * * * (-3.10) (-3.09) (-2.97) (-1.85) (-1.87) (-1.76) Fallen angel * Duration (-0.23) (-0.12) (-0.27) (-0.25) Fallen angel*log(offering amt) (-0.23) (-0.61) (0.32) (-0.13) Fallen angel * Bond age (0.46) (0.10) (0.41) (-0.17) Fallen angel * Equity volatility * (-1.47) (-1.67) Fallen angel * Equity beta 0.023** 0.029** (2.03) (2.29) Firm-level controls - - Y - - Y Industry, Time Credit rating FE Y Y Y Y Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 3,340 3,340 3,340 2,234 2,234 2,234 R-squared

42 Table VII (Cont d) Panel B: Change in Bond Yield Spread In Panel B, we focus on the change in bond yield spread around the month of rating changes. The dependent variable is the change in bond yield spread relative to the previous month. Columns (1)-(3) are based on the full sample of rating changes including both rating downgrades and rating upgrades. In column (1), we only interact the fallen angel dummy with the No CDS dummy. In column (2), we add the interaction terms of fallen angel with bond characteristics including bond duration, offering amount and bond age. In column (3), we further add the interaction terms of fallen angel with risk characteristics such as equity volatility and equity beta. Columns (4)-(6) follow the same specifications as columns (1)-(3), except that they are based on the subsample of rating downgrades. For brevity, in columns (3) and (6), we don t report the results on firm-level controls, which include equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. We always include time (monthly) credit rating fixed effects, industry fixed effects at the two-digit SIC level, and cluster the standard errors at the firm level. Dep: Change in yield spread around rating changes Full Sample on Rating Change Sub-sample on Rating Downgrade (1) (2) (3) (4) (5) (6) Fallen angel 0.497** (2.13) (-0.64) (-1.46) (1.28) (-0.20) (-1.22) Fallen angel * No CDS 1.618*** 1.679*** 1.943*** 1.848** 1.884*** 2.385*** (2.65) (2.74) (3.47) (2.58) (2.63) (3.61) Controls No CDS (-0.51) (-0.54) (-0.09) (0.05) (-0.00) (0.37) Coupon rate (1.00) (0.86) (1.07) (1.58) (1.38) (1.60) Duration ** ** ** *** *** (-1.52) (-2.07) (-2.16) (-2.18) (-2.72) (-2.82) Log(offering amount) (1.13) (0.75) (0.29) (0.97) (0.62) (0.44) Callability (-1.32) (-1.34) (-1.36) (-1.43) (-1.45) (-1.57) Fungibility * * (-1.27) (-1.30) (-1.55) (-1.63) (-1.66) (-1.90) Credit enhancement 0.095* 0.096* 0.115** 0.193** 0.190** 0.213** (1.86) (1.88) (2.11) (2.09) (2.04) (2.16) Bond age * * (-1.30) (-1.13) (-1.59) (-1.86) (-1.50) (-1.97) Fallen angel * Duration 0.057** 0.062*** 0.070*** 0.073*** (2.40) (2.59) (2.68) (2.93) Fallen angel*log(offering amt) (0.92) (1.12) (0.73) (0.94) Fallen angel * Bond age (-0.52) (-0.22) (-0.99) (-0.55) Fallen angel * Equity volatility (0.02) (-0.79) Fallen angel * Equity beta 0.604** 1.049*** (1.98) (2.59) Firm-level controls - - Y - - Y Industry, Time Credit rating FE Y Y Y Y Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 4,972 4,972 4,972 3,443 3,443 3,443 R-squared

43 Table VII (Cont d) Panel C: Change in Bond Illiquidity In Panel C, we focus on the change in bond illiquidity around the month of rating changes. The dependent variable is the change in bond illiquidity relative to the previous month. Columns (1)-(3) are based on the full sample of rating changes including both rating downgrades and rating upgrades. In column (1), we only interact the fallen angel dummy with the No CDS dummy. In column (2), we add the interaction terms of fallen angel with bond characteristics including bond duration, offering amount and bond age. In column (3), we further add the interaction terms of fallen angel with risk characteristics such as equity volatility and equity beta. Columns (4)-(6) follow the same specifications as columns (1)-(3), except that they are based on the subsample of rating downgrades. For brevity, in columns (3) and (6), we don t report the results on firm-level controls, which include equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. We always include time (monthly) credit rating fixed effects, industry fixed effects at the two-digit SIC level, and cluster the standard errors at the firm level. Dep:Change in bond illiquidity around rating changes Full Sample on Rating Change Sub-sample on Rating Downgrade (1) (2) (3) (4) (5) (6) Fallen angel ** (1.42) (0.33) (0.90) (2.41) (0.50) (1.35) Fallen angel * No CDS 3.381*** 3.662*** 3.566*** 3.955*** 4.475*** 6.323*** (4.15) (4.23) (3.85) (4.00) (4.34) (4.88) Controls No CDS (-0.83) (-0.80) (-0.28) (1.00) (1.06) (0.41) Coupon rate (-0.92) (-0.86) (-1.05) (-0.83) (-0.77) (-0.56) Duration * 0.029* 0.031** (1.29) (1.28) (1.44) (1.94) (1.95) (2.06) Log(offering amount) * (-1.11) (-1.11) (-1.40) (-1.43) (-1.65) (-1.72) Callability (0.83) (0.83) (0.88) (0.23) (0.21) (0.19) Fungibility (-1.04) (-0.95) (-0.95) (-0.19) (-0.09) (-0.08) Credit enhancement (0.09) (0.05) (-0.20) (-0.28) (-0.33) (-0.49) Bond age (0.01) (-0.16) (-0.16) (0.40) (0.17) (-0.03) Fallen angel * Duration (-0.59) (-1.27) (-0.74) (-1.20) Fallen angel*log(offering amt) (0.08) (0.00) (1.03) (0.46) Fallen angel * Bond age (1.07) (0.87) (1.50) (1.24) Fallen angel * Equity volatility *** *** (-3.39) (-2.64) Fallen angel * Equity beta (1.47) (-0.03) Firm-level controls - - Y - - Y Industry FE Y Y Y Y Y Y Time Credit rating FE Y Y Y Y Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 2,143 2,143 2,143 1,422 1,422 1,422 R-squared

44 Table VIII CDS Presence and the Impact of Exposed Insurance Ownership around Hurricane Katrina In this table, we examine how the presence of CDS contracts may alter the impact of Hurricane Katrina on bond yield spreads and bond illiquidity, through the channel of exposed property insurance and reinsurance companies and their bond holdings. Our interested variable is the interaction term between the pre-katrina exposed insurance bond ownership and the No CDS dummy (1- CDS presence dummy). The data on institutional holdings of corporate bonds are from Lipper s emaxx fixed income database. We exclude those bond issuers that may be directly affected by the hurricane, which include life and property (re) insurance companies, and firms headquartered in the states of Louisiana, Mississippi, and Alabama. First, we identify the set of property & casualty insurance and reinsurance companies that are considered to have high exposure to Hurricane Katrina, using data from the Holborn Corporation (2005) Hurricane Katrina report, publicly available at the URL: The Holborn Report lists the names of property & casualty (re)insurance companies along with their 2004 market shares in the states of Louisiana, Mississippi, and Alabama, and whether they have rating or outlook changes immediately after the hurricane. We include the top ten property insurance companies by their market shares (including both personal and commercial lines) and eight reinsurance companies with negative rating (outlook) changes that can be identified in Lipper/EMAXX as managing firms. The names of those insurance companies are provided in the appendix. Then, we define pre-katrina exposed insurance bond ownership as the par amounts held by property and reinsurance companies with high exposure to hurricane Katrina at the end of the second quarter of 2005 divided by the amount of bond outstanding. Non-exposed bond ownership is defined as the difference between total institutional ownership minus the exposed insurance ownership. In columns (1)-(3), the dependent variable is the change in bond yield spread from Aug. 23, 2005 to Sep 9, 2005 (the two weeks during which Hurricane Katrina formed and fully dissipated). In columns (4)-(6), the dependent variable is the difference of bond illiquidity between Sep 2005 and Aug In column (1), we only interact the pre-katrina exposed insurance ownership with the No CDS dummy. In column (2), we add the interaction term of non-exposed institutional ownership with the No CDS dummy. In column (3), we interact exposed insurance ownership with bond characteristics including bond duration, offering amount and bond age. Columns (4)-(6) follow the same specifications as in columns (1)-(3), respectively. For brevity, in columns (3) and (6), we don t report the results on firm-level controls including equity volatility, equity beta, market-to-book, book leverage, book size, profitability, cash holding and dividend payment. We always control for credit rating fixed effects (issue level), industry fixed effects at the two-digit SIC level, and we cluster the standard errors at the firm level. ***, ** and * represent significance levels at 1%, 5% and 10% respectively using heteroscedasticity robust standard errors with t-statistics given in parentheses. 41

45 Pre-Katrina exposed insurance ownership Pre-Katrina exposed insurance ownership * No CDS Table VIII (Cont d) Dep: Change in yield spread (Aug 23, 2005 Sep 9, 2005) Dep: Change in bond illiquidity (Aug, 2005 Sep, 2005) (1) (2) (3) (4) (5) (6) 0.194* 0.196* 2.921** (1.83) (1.85) (2.24) (-0.51) (-0.50) (-0.11) 1.522** 1.460** 1.337** *** ** ** (2.56) (2.47) (2.40) (2.96) (2.57) (2.31) Controls No CDS (-1.50) (-1.25) (-0.30) (-0.28) (-0.38) (-0.25) Coupon rate (-1.05) (-1.04) (-0.31) (-0.43) (-0.42) (-0.31) Duration 0.006*** 0.006*** 0.005*** (4.21) (4.21) (4.04) (-0.05) (-0.06) (0.00) Log(offering amount) 0.038*** 0.038*** 0.020*** (2.90) (2.91) (2.62) (-0.54) (-0.53) (-0.67) Callability * 0.115* 0.128** (0.14) (0.16) (1.34) (1.94) (1.95) (2.00) Fungibility (0.66) (0.66) (0.61) (0.14) (0.16) (0.14) Credit enhancement (-0.40) (-0.38) (-0.41) (-0.81) (-0.78) (-0.59) Bond age 0.004* 0.004* (1.90) (1.91) (0.87) (0.56) (0.57) (0.10) Non-exposed ownership ** ** (-2.09) (-2.11) (-0.70) (-0.22) (-0.24) (-0.02) Non-exposed ownership * No CDS Pre-Katrina exposed insurance ownership * Duration Pre-Katrina exposed insurance ownership * Log (offering amt) Pre-Katrina exposed insurance ownership * Bond age (0.39) (0.29) (0.19) (0.24) (-0.30) (-0.33) * (-1.83) (-0.00) (-1.50) (0.90) Firm-level controls - - Y - - Y Industry FE Y Y Y Y Y Y Credit rating FE Y Y Y Y Y Y Clustering Issuer Issuer Issuer Issuer Issuer Issuer Number of Obs. 1,830 1,830 1,830 1,098 1,098 1,098 R-squared

46