Impact of Credit Default Swaps on. Firms Investment Decisions, Financing Preferences, Cash Holdings and Risk Profiles

Transcription

1 Impact of Cred Default Swaps on Firms Investment Decisions, Financing Preferences, Cash Holdings and Risk Profiles By Kathleen P. Fuller, Serhat Yildiz*, and Yurtsev Uymaz This version September 23, 2014 Contact Information: Kathleen P. Fuller Holman 321, School of Business, Universy of Mississippi, Universy, MS 38677, Uned States, Phone: Serhat Yildiz (Contact Author) Holman 226, School of Business, Universy of Mississippi, Universy, MS 38677, Uned States, Phone: Yurtsev Uymaz Room 0.04, Hen Goleg, Bangor Business School, Bangor Universy, Bangor, Gwynedd, LL57 2DG, Uned Kingdom, Phone:

2 Abstract This study examines the impact of cred default swaps (CDS) on firms financing preferences, investment decisions, cash holdings, and idiosyncratic, market, and total risks. By examining non-financial S&P 500 firms from 2002 to 2012, we find that CDS firms decrease their investment by 1.7%, long term debt issue and equy issue around 1.8% annually but increase their short term debt issue by 0.6% and cash holdings. We also find having CDS traded at least once decreases firms total risk and especially market risk. Among of investment graded firms; CDS firms decrease their investment more, issue more equy and less short term debt compared to non-cds firms. 2

3 1. Introduction Introducing market frictions to Miller and Modigliani s (1958, 1963) model makes capal structure and investment policies matter. Dominant capal structure theories such as pecking order and trade-off theory mainly focus on demand side frictions of debt capal. In pecking order information asymmetry and debt capacy have important roles in firm s debt demand, and in trade-off tax advantage and financial distress costs of debt determine corporation s demand of debt. However, recently scholars emphasize that not only demand side but also the supply side frictions such as cred ratings, cred supply shocks, cred supply uncertainty and access to debt market influence a firm s capal structure, financing decisions, and investment decisions and qualy 1. An important innovation in capal supply markets related to supply frictions is the cred default swaps (CDS). Dramatically fast growing CDS market 2 has important effects on supply, cost, amount and matury of debt, and also cred risk of reference firm 3. By decreasing debt capal supply side frictions CDS enable firm to hold more debt and for a longer period of time (Sarretto and Tookes, 2013) however CDS also increase reference firm s likelihood of financial distress (Subrahmanyam et al., 2014). Since supply frictions matter for firms capal structure, financing, and investment policies and CDS alleviate these frictions, follows that firms wh CDS may have not only higher levels of debt but also different investment policies, capal financing, and cash holdings. Moreover, CDS have increase a firm s likelihood of financial distress, thus, they may affect idiosyncratic, 1 Sufi, 2009; Lemmon and Roberts, 2010; Morellec, 2010; Fualkender and Petersen, 2006; Harford and Uysal, The CDS market has grown from around USD 2 trillion in 2002 to nearly USD 60 trillion in 2007 (Deutsche Bank, 2009). 3 Hirtle, 2009; Ashcraft and Santos, 2009; Subrahmanyam et al. 2014; Sarretto and Tookes,

4 systematic, and total risk of firm as well. Using a sample of non-financial firms wh CDS traded from 2002 to 2012, we complement Saretto and Tookes (2013) and Subrahmanyam et al., s (2014) studies by examining CDS impact on the firm s investment policies, capal financing, cash holdings, and also idiosyncratic, market and total risks. Interestingly, we find that CDS firms decrease their investments by 1.7%, long term debt issue and equy issue around 1.8% annually but increase their short term debt issue by 0.6% and cash holdings. While having CDS trading on their debt has no impact on the risk profiles of firms, having CDS traded at least once decreases firms total risk and especially market risk. Among of investment graded firms; CDS firms decrease their investment more, issue more equy and less short term debt compared to non-cds firms. Debt capal supply side frictions play important roles in firm s capal structure and investment decisions. Having access to public debt markets increases firm s leverage ratio (Fualkender and Petersen, 2006) and has real impacts on firm s investment abily and qualy (Harford and Uysal, 2014). Saretto and Tookes (2013) find that lessening frictions wh CDS have significant impacts on capal structure, but they don t examine how CDS affect the investment policies of firms. However, examining CDS impact on investment decisions is important because, such an examination will reveal real consequences of CDS trading and increase our understanding of this fast growing derivatives market. Our study complements Saretto and Tookes study by examining impact of CDS on firms investment policies. CDS can benef firms investment policies by the alleviating capal supply frictions (as in Saretto and Tookes, 2013) or they may harm the investments through increased cost of capal (as in Ashcraft and Santos, 2009) and increased likelihood of financial distress (as in Subrahmanyam et al., 2014). Hence, the exact impact of CDS on firms investment decisions is an empirical question. Consistent wh 4

5 increased risk of financial distress and higher cost of capal findings, we find that CDS firms actually decrease their annual investments about 1.7 % and decrease occurs even among investment grade firms. In addion to capal structure and investment policies capal supply frictions affect firms decisions to issue equy and debt. Specifically, capal supply uncertainty increases probabily of equy issuance while decreases probabily of debt issuance (Massa, Yasuda and Zhang, 2009) and capal supply shocks cause firms to decrease net debt and equy issuance (Lemmon and Roberts, 2010). Since CDS decrease capal supply frictions (Sarretto and Tookes, 2013) these contracts may impact firms decisions to issue equy and debt. We examine the CDS s abily to reduce friction when a firm issue equy, short term debt and long term debt. We find that while CDS firms increase their short term debt issue, they decrease their long term debt issue and equy issue. Among of investment graded firms; CDS firms issue more equy and less short term debt compared to non-cds firms. CDS not only increase the amount and matury of debt (Saretto and Tookes, 2013) but they may also increase the amount of capal available to firms (Hirtle, 2009). Given the increase in capal, we examine how firms utilize this extra capal. Firm can use the capal in investments or can keep as cash to meet financial obligations. If CDS friction reduction works similar to obtaining a bank loan rating, we expect firms to increase their investments (as in Sufi, 2009). However, if CDS firms are close to be financial distressed (as in Subrahmanyam et al. 2014), then we expect firms to keep capal as cash to meet their financial obligations instead of investing. Supporting to second view, we find that CDS firms increase their cash holdings. This finding is consistent of Subrahmanyam et al. which find that, in 2 years following CDS inception, firm s cred rating declines and likelihood of bankruptcy more than doubles. Our findings show that perhaps CDS 5

6 firms are aware of increased likelihood of financial distress and increase their cash holdings as precaution. We also expect CDS to have implications for risk profile of firm. Instefjord (2005) theoretically shows that CDS may enable banks to approve more risky loans. Wh increased risk in the loans provided, is natural to expect that CDS may increase the riskiness of the borrowers. In addion Subrahmanyam et al. (2014) find that cred risk of reference firm increases upon inception of CDS. Then we expect that riskiness of firm may increase following CDS introduction. However, the introduction of CDSs may decrease the information asymmetry between the firm and market participants (Sufi, 2009; Subrahmanyam et al., 2014). Lessened information asymmetry wh CDS trading may decrease the riskiness of firm. We examine impact of CDS trading on firm s overall risk, market risk and idiosyncratic risk. We find that having CDS trading on their debt has no impact on the overall risk, market risk and idiosyncratic risk of firms, however, firms wh CDS trading occurs at least once have lower market risk and total risk. This finding supports the views that CDS decrease information asymmetry between market participants and firm. Another contribution of our study is that there is an ongoing debate about the roles of CDSs in financial markets; our analysis contributes this debate by revealing real impacts of CDS trading on firm s financing decisions, investment policies, and risk profiles. From 2002 to 2007 the gross notional amount of outstanding cred default swaps (CDS) grew from around USD 2 trillion to nearly USD 60 trillion (Deutsche Bank, 2009). Though the CDS market is a large component of the financial markets, s role is still debated. Warren Buffet claims CDS are "financial weapons of mass destruction 4. Stout (2009) considers the Commody Futures Modernization Act in 2000, 4 A nuclear winter?, The Economist,Sep 18th

7 which turned derivative contracts into legally enforceable ones, as one reason for the 2007 financial crisis. Moreover, CDS trading increases a firm s cred risk and likelihood of bankruptcy (Subrahmanyam et al., 2014). CDS trading also increases the cost of debt financing for riskier firms and firms wh high information asymmetry (Ashcraft and Santos, 2009). Yet, Stulz (2010) argues that CDS worked well during the financial crisis, and eliminating over-thecounter trading of CDS could reduce social welfare. In addion, Sarretto and Tookes (2013) show that CDS decrease debt capal supply side frictions and enable firm to hold more debt especially when cred constraints become binding. By revealing CDS impacts on firms risk profiles, investment policies, and financing decisions, our study increase our understanding of CDS market. 2. Lerature 2.1 Capal supply frictions and firms financing, and investment decisions The relation between capal supply frictions and firms financing, and investment decisions are studied both theoretically and empirically. Recent studies 5 show that capal supply markets frictions matter for the capal structure of the firm. Morellec (2010) theoretically models role of cred supply uncertainty in the firm s investment and financing decisions and finds that cred supply uncertainty has important consequences not only for current capal structure but also for corporate investment. Wh the cred market access, the firm has fewer restrictions on current investment, however this access also reduces the opportuny costs of the investment (i.e. waing becomes less risky). These two effects have oppose impacts on the investment decision. While 5 Morellec 2010; Faulkender and Petersen, 2006; Harford and Uysal, 2014; Lemmon and Roberts, 2010; Massa, Yasuda and Zhang, 2009; Leary,

8 the first one increases the investment, the second one decreases. He shows that the second effect generally dominates the first one. Accordingly, negative shocks to the cred supply may reduce the investment, even if firms have sufficient funds to finance their investment opportunies internally. Faulkender and Petersen (2006) investigate the link between where firms obtain their capal (private or public debt) and their financing decisions (debt use). They argue that access to capal markets constraints the debt use of the firm and consistently find that firms that can access to public debt markets use more debt. Harford and Uysal (2014) complement the Faulkender and Petersen (2006) s study by examining the effect of debt markets access on investment decisions. By proxying debt market access wh having a debt rating, they find that compared to non-rated firms, rated firms are more likely to undertake acquisions, pay higher premiums for their targets, receive less favorable market reaction to acquision announcements. Overall, they show that access to debt markets affects firm s abily to invest as well as qualy of investment. By studying the 1961 emergence of the market for certificates of depos, and the 1966 Cred Crunch, Leary (2009) finds that following an expansion (contraction) in the availabily of bank loans, leverage ratios of bank-dependent firms significantly increase (decrease) relative to firms wh bond market access. Above mentioned studies show that frictions in capal supply have an impact on the firm s capal structure. Massa, Yasuda and Zhang (2009) study the impact of capal supply uncertainty (CSU) on leverage and matury. CSU is defined as the volatily of potential supply of capal by firm s 8

9 bond investors. They find that CSU has negative and significant effects on the firm s probabily of issuing bonds and commercial paper. In contrast, CSU has posive and significant effects on the firm s probabily of issuing equy and borrowing from banks. Yasuda and Zhang suggest that the firm responds to an increase in CSU of s investor base by substuting away from bonds and into equy and bank loans. By studying the introduction of syndicated bank loan ratings by Moody s and S&P in 1995, Sufi (2009) examines whether the introduction of loan ratings increases firms use of debt and investment. Sufi finds that introduction of loan ratings increases the supply of debt. Firms that obtain a loan rating experience an increase in their leverage ratio, asset growth, cash acquisions, and investment in net working capal. Overall, Sufi finds that loan ratings enable informationalopaque borrowers to reach capal of uninformed investors. In turn, this leads more investment. Lemmon and Roberts (2010) study how shocks to cred supply affect the firms financing and investment decisions. By focusing on the collapse of Drexel Burnham Lambert, Inc., Lemmon and Roberts find that debt issuance and investment decrease wh this shock, but firms leverage ratios remained relatively stable. Overall, the lerature finds that capal supply frictions affect not only financing policies of firms but also the investment policies. 2.2 CDS and firms financing, and investment decisions The studies in section 2.1 support the view that capal supply frictions matter for financing and investment decisions of firm. Accordingly, we expect that tools that decrease these frictions may affect the financing and investment decisions of firm. CDS may be considered as such tools. According to Saretto and Tookes (2013) capal suppliers can hedge the cred risk of a borrower using CDS, and the CDS market can relax the firm s cred supply constraints and enable firms 9

10 to increase their leverage and extend their debt matury. Specifically, they find that debt ratios for CDS firms increase between 6% and 22% of mean leverage and debt matury increases between 0.68 and 1.79 years, impact is greatest when cred supply constraints are most binding. The impact of CDS on the cost of debt is studied by Ashcraft and Santos (2009). They find no supportive evidence that the average firm wh a traded CDS benefs from a reduction in the cred spreads pays to issue in the bond market or the spreads pays to borrow from banks. Surprisingly, Ashcraft and Santos find that the onset of CDS trading increased the cost of debt financing in capal markets for the riskier firms as well as those that have more information asymmetry. Gamba and Saretto (2012) theoretically model, and empirically document, that investments are negatively related to changes in CDS spread. They focus on determinant of CDSs use rather than impact of CDSs on firms investment policies. 2.3 CDS and risk profiles Subrahmanyam et al. (2014) argue that the CDS posion can affect the credor-borrower relation and thus the borrower s cred risk. CDS may also increase capal supply and financial flexibily of reference firm because CDS enable credor to hedge the cred risk. Yet, wh hedging abily of CDS lender may decrease the monoring of borrowers, and be tougher during financial distress (e.g. debt negotiation) and increase likelihood of bankruptcy. Consistent wh their second argument, they find that the likelihood of a rating downgrade and the bankruptcy increase after CDS trading begins. Instefjord (2005) theoretically shows that wh risk sharing advantages of CDS, banks tend to take more risk and increase their riskiness. Hirtle (2009) find some evidence that use of cred derivatives (all derivatives not just CDS) increases the supply of bank cred. 10

11 3. Hypotheses 3.1 Hypothesis development Wh CDS a lender can hedge the cred risk of reference firm and so they may increase capal supply available to the firm (Subrahmanyam et al., 2014). Also, Hirtle (2009) find some evidence that capal supply to the firm may increase wh derivatives such as CDS. Saretto and Tookes (2013) find CDS firm may hold more debt for longer period of time. Accordingly, we argue that the increased capal supply may increase the cash holdings of the firms. Formally we test the following hypothesis: Hypothesis 1: The introduction CDSs will increase the cash holdings of the firms. Massa, Yasuda and Zhang (2009) find capal supply frictions may affect the financing choice of firms. Lemmon and Roberts (2010) show that capal supply shocks may cause firms to decrease net debt and equy issuance. Since CDS may decrease the frictions of capal supply, CDS firm can increase debt and equy issues of the firms. On the other hand, CDS may increase capal available to the firm (e.g., argument of hypothesis 1), and so CDS firms may need to issue less compared to non-cds firm. In this case, we expect CDS firms to decrease their debt and equy issues. Wh these competing arguments, we test following two hypotheses. 6 Hypothesis 2a: CDS firms will decrease their equy, long term and short term debt issues. Hypothesis 2b: CDS firms will increase their equy, long term and short term debt issues. The relation between capal supply frictions and firms investment decisions is shown in several studies (Sufi, 2009; Lemmon and Roberts, 2010; Morellec, 2010). Since CDS may decrease 6 Different investment grade firms financing and investment policies respond to supply frictions differently (Lemmon and Roberts, 2010), hence in our financing and investment policies analyses we divide firms into investment grade and non-investment grade groups. 11

12 capal supply frictions, we expect that they may increase firms investment. On the other hand if CDS trading signal that firms are closer to financial distress (as in Subrahmanyam et al., 2014), then we expect CDS firms to decrease their investments. Thus, we test following two hypotheses. Hypothesis 3a: Investment grade firms wh CDSs will increase their investments. Hypothesis 3b: Investment grade firms wh CDSs will decrease their investments. Instefjord (2005) theoretically shows that wh risk sharing advantages of CDS, banks tend to take more risk. If wh increased risk appete banks approve more risky loans, this may increase riskiness of the firms. Also Subrahmanyam et at.(2014) find that CDS trading increase the cred risk and likelihood of financial distress of the firms. Deteriorating cred rating and increased financial distress may affect firms total, idiosyncratic, and market risk. So we expect CDS may increase firms three types of risks. Yet, the introduction of CDS may decrease the information asymmetry between the firm and market participants (Subrahmanyam et al., 2014; Sufi, 2009). Accordingly, markets view CDS users as less risky than the non-cds users. Since the expectations are mixed, we test the following two hypotheses. Hypothesis 4a: Total, idiosyncratic, and market risks of CDS firms will increase. Hypothesis 4b: Total, idiosyncratic, and market risks of CDS firms will decrease. 2. Data and methodology Similar to Saretto and Tookes (2013), we collect CDS data from Bloomberg. Our sample of firms consists of all non-financial S&P 500 firms from 2002 to To be included our sample; firms must also be listed on Compustat and CRSP databases. We follow Lemmon and 12

13 Roberts (2010) to measure secury issue and net investment variables. 7 Net equy issue and investment decisions are considered as a function of cash flows, market to book, log(sales), Altman s Z-score, term spread, and equy market return. Adopting a similar approach to Fung et al. (2012) the riskness of the firm is defined as total risk, market risk, and idiosyncratic risk. Ferreira and Laux (2007) find several firm characteristics that impact the riskiness of firm; specifically, market to book, size, dividend, zscore, term spread and market return all impact the risk. Following in Opler et al. (1999) models, the change in cash is considered as a function of cash flow, market to book, size, leverage and capal expendure. We define two binary variables related to CDS to capture the impact of CDS on investments, financing preferences, risk proxies and cash holding. First, CDS trading is an indicator variable equal to one if there is traded CDS on the firm s debt during the year t. The firms wh CDS on their debts and whout CDS on their debts may be different due to unobservable factors; the variation in timing of introduction of CDSs may be helpful to migate this concern. To controls for time-invariant unobservable difference between firms that uses CDS and non-cds firms we define CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. The main variable of interest is CDS trading variable. This methodology is similar to the methodologies of Saretto and Tookes (2013) and Ashcraft and Santos (2009). 3. Results 3.0 Descriptive statistics and univariate comparisons 7 Detailed definions and calculation procedures of the variables are given in the appendix. 13

14 Table 1 provides the summary statistics for the all firms, CDS firms and non-cds firms. Both CDS firms and non-cds firms have posive net investments but non-cds firms investments are more volatile. A similar pattern exists in financing decisions as well, i.e. non- CDS firms net equy issue, long term debt issue and short term debt issues have higher standard deviations compared to those of CDS firms. CDS firms have higher sales than the non-cds firms and CDS firms cash flows are more stable. Though CDS firms are more profable, standard deviation of their ROEs are higher. In terms of riskiness CDS firms have lower risk in three different proxies which are market risk, idiosyncratic risk and total risk. Consistent wh Saretto and Tookes (2013) CDS firms have higher leverage and larger compared to non-cds firms. {Insert Table 1 about here} Figure 1 illustrates the change in the number of firms that have CDS trading on their debt in our sample. There is a significant increase in the number of firms that have CDS on their debt from 80 firms in 2002 to 221 firms in The highest increase is around 8.3% between 2005 and {Insert figure 1 about here} Table 2 provides univariate comparison of three different samples. Non-CDS sample consists of firms that CDS have never been traded on their debts. CDS traded is the sample of the firms that CDS are traded on firms debts at least once and CDS trading is the sample of the firms that CDS are traded on the firms debts during the given year. Surprisingly, non-cds firms investment is around 3% higher than both CDS samples. The high t-score implies that the 14

15 difference is statistically significant. Non-CDS firms issue more long term debt than CDS firm (around 1.5%). Non-CDS firms have higher total risk, idiosyncratic risk and market risk. Especially market risk is 11%-8% higher for non-cds firms. Overall, Table 2 provides preliminary support the views that CDS firms and non-cds firms may differ in their investment and financing decisions. Risk profiles of the CDS firms and non-cds firms are also different. {Insert Table 2 about here} 3.1 CDS impact on the change in cash holdings The regression in Table 3 examines the impact of CDS on change in cash holdings of the firms. Note that by definion of dependent variable increases in cash and equivalents are presented as posive numbers and decreases are presented as negative numbers. The model is an OLS model, t-stats and p-values are calculated using firm and year clustered standard errors. The results of test are summarized in Table 3. The estimated equation is Y i 1 CDS traded 2CDS _ ' _ trading X, where Y is the change in cash for firm i in year t, CDS _ traded is a binary equal to one if there is a CDS market for firm i s debt in any year t between 2002 and 2012, CDS _ trading is an indicator variable equal to one if there is a CDS market on the firm i s debt in year t, and X 15

16 is a vector of control variables, including cash flow, market to book, size, leverage and capal expendure. 8 {Insert Table 3 about here} In Table 3 results, our main variable of interest, CDS trading variable has a posive and significant coefficient. This coefficient implies that firms that have CDS trading on their debts increase their cash holdings by 0.84% on average; this result is significant at 1% level. Thus, the increased capal supply which is caused by CDS affects cash holdings of firms posively. We can interpret this finding in two different ways. As a posive impact of CDS, the results indicate that reduced frictions wh CDS (as in Sarretto and Tookes, 2013) enable firm to increase their cash holdings. The increase in cash may help firm to meet their short term liquidy obligations easier and thus firms become financially more flexible. On negative side of CDS, if CDS trading increases likelihood of financial distress (as in Subrahmanyam et al., 2014) and then firms increase their cash holdings as a cushion for expected financial distress. Table 3 also shows that size and market to book are also important determinants of cash holdings of the firm. Figure 2 illustrates the trend of change in cash holdings of CDS firm, from the year before the CDS iniations to following two years wh CDS. The trend supports our finding in Table 3. Cash holdings of CDS firms were decreasing by 5.2% the year before the CDS iniation. However, wh the iniation of CDS cash holdings start increasing. One year after the CDS iniation increase in cash holdings reach their peaks. In year two the increase slowed yet cash is still higher than the year before CDS iniation and first year wh CDS. This figure shows that 8 Detailed definions of variables are given in the appendix. 16

17 CDS increase cash holdings of firms. Wh reduced friction firms can reach capal markets easier and this helps them to increase their short term liquidy by holding more cash. {Insert figure 2 about here} 3.2 CDS impact on the equy and debt issue decisions The regressions in Table 4 test whether and how CDS affect the financing policies of firms. The models are constructed adopting Lemmon and Roberts (2010) approach. The t-statistics and p- values are calculated wh firm and year clustered standard errors. The independent variables are net equy issue (model 1), long term debt issue (model 2) and short term debt issue (model 3). The estimated equation is Y i 1 CDS traded 2CDS _ ' _ trading X, where Y is the net equy, long term debt or short term debt issue for firm i in year t respectively, CDS _ traded and CDS _ trading are as defined section 3.1, and X is a vector of control variables, including cash flow, market to book, size, Z-score, term spread, and market return. Lemmon and Roberts (2010) find that different investment grade firms financing and investment policies respond to supply frictions differently, thus in our financing and investment policies analyses we divide firms into investment grade and non-investment grade groups wh invest dummy variable. Invest. Grade is defined following S&P, and equals to one if firm i is rated BBB- or higher, zero if firms are rated BB+ or lower in year t. {Insert Table 4 about here} 17

18 Table 4 models show that firms wh CDS trading on their debt decrease their net equy issue and long term debt issue around 1.8% and increase their short term debt issue around 0.6%. These results imply that CDS affect short term and long term financing decisions of firms differently. The significant interaction terms in three models show that investment grade firms wh CDS on their debts issue more equy and less short term debt compared to the benchmark group. Specifically, investment grade firms wh CDS issue 0.8% more equy and 0.6% less short term debt compared to investment grade firms whout CDS. The results regarding to the long term is economically less significant. 3.3 CDS impact on net investment decisions This section examines the impact of CDS on firms investment decisions. Table 5 summarizes our findings. Model 1 tests CDS s impact in overall sample, while model 2 separates the impact according to investment grade of firm. In both models dependent variable (net investments) and controls are calculated according to Lemmon and Roberts (2010). T-statistics and p-values are calculated wh firm and year clustered standard errors. The estimated equation is Y i 1 CDS traded 2CDS _ ' _ trading X, where Y is the net investment for firm i in year t, CDS _ traded and CDS _ trading are as defined section 3.1, Invest. Grade is as defined in section 3.2, and X is a vector of control variables, including cash flow, market to book, size, Z-score, term spread, and market return. {Insert Table 5 about here} In Table 5 model 1 CDS trading variable has a negative and significant coefficient. The coefficient implies that CDS trading decreases annual net investments around 1.7%, on average. 18

19 Though firms wh CDS hold more debt and for a longer period of time (Saretto and Tookes, 2013), they don t increase their investment spending. This finding is in-line wh argument of Subrahmanyam et al. (2014) that CDS trading may affect decision markers incentives and induce suboptimal real decisions on reference firms. To see the impact of CDS on investment decision more clearly, we separate the sample into two by using investment grade dummy variable in model 2. In the model 2 the interaction variable (Inv. Grade*CDS trading) has a posive and significant coefficient. This coefficient implies that investment grade firms that have CDS traded on their debts invest less compared to the benchmark group. This analysis also support above argument that CDS may induce suboptimal real decisions to reference firms. 3.4 CDS impact on firms risk profiles We examine risk profile analysis we use three different proxies: market risk, idiosyncratic risk, and total risk. These proxies are calculated following Fung et al. (2012) paper. Specifically, market risk is proxied wh betas from CAPM regression. The standard deviation of residuals from the CAPM regression represents the idiosyncratic risk. The standard deviation of the daily returns for an individual firm represents total risk. T-stats and p-values are calculated using firm and year clustered standard errors. The results of examination are summarized in Table 6. The estimated equation is Y i 1 CDS traded 2CDS _ ' _ trading X, where Y is standard deviation of residuals, betas from CAPM regression or standard deviation of the daily returns for firm i in year t, CDS _ traded and CDS _ trading are as defined 19

20 section 3.1, and X is a vector of control variables, including market to book, size, Z-score, dividend, term spread, and market return. The control factors in the models are chosen following Ferria and Laux (2007). {Insert Table 6 about here} Table 6 shows that having CDS trading in a given year has no impact on risk profiles of firms. However, the significant CDS traded variable means CDS inception decreases the firm s total risk and market risk. Thus trading CDS at least once will decrease a firms total risk and market risk by approximately 9.01%. This is consistent wh Sufi (2009) and Subrahmanyam et al. (2014) that the iniation of CDS reduces the information asymmetry between market and firm. Since the information is obtained by the markets at iniation, and incorporated to the firm s risk profile, continuing to trade CDS (CDS trading) carries no further information about the riskiness of the firm. Accordingly, we find that CDS trading is insignificant. 4. Conclusion In this study we find that CDS markets are not sideshows, they have real consequences on reference firms investment policies, financing decisions, cash holdings and risk profiles. CDS firms decrease their investment spending, and increase their cash holdings. In addion, CDS affect firms financing policies; CDS firms increase their short term debt issue but decrease their long term debt and equy issues. We also find markets may consider having CDS traded once as an information event, and this event decreases market risk of the firm. Overall, our examination contributes to a better understanding of CDS market and s impact on the reference firms. We do 20

21 not argue that CDS are necessarily harmful or beneficial, we only reveal how this important derivative markets affect the reference firm from different aspects. 21

22 Appendix Variable definions Investment and secury issue variables are defined following Lemmon and Roberts (2010); Cred Rating = senior long-term debt rating (280). Net Investment = (capal expendures (128) + acquisions (129) sale PPE (107) + increase in investments (219) sale of investments (109)) start-of-period assets (6). Net LT Debt Issues = long-term debt issues (111) long-term debt reduction (114) start-ofperiod assets. Net ST Debt Issues = change in current debt (301) start-of-period assets. Net Equy Issues = sale of common and preferred stock (108) purchase of common and preferred stock (115) start-of-period assets. Total Debt = long-term debt (9) + short-term debt (34). Book Leverage = total debt book assets (6). Market Value of Assets = stock Price (199) shares outstanding (25) + short-term debt + longterm debt + preferred stock liquidation value (10) deferred taxes and investment tax creds (35). Firm Size = log(sales) Market-to-Book = (market equy + total debt + preferred stock liquidating value deferred taxes and investment tax creds) book assets. Z-Score = 3.3 pre-tax income (170) + sales (12) retained earnings (36) (current assets (4) current liabilies (5)) book assets. Cash Flow = income before extraordinary ems (123) lagged book assets. Term Spread = the yield spread between the 1- and 10-year Treasury bonds. Equy Market Return = CRSP annual value-weighted return. Return on Equy = operating income before depreciation (data13) book equy (data60). Change in Cash = change in cash equivalents (274) start-of-period assets. (Increases in cash and equivalents are presented as posive numbers. Decreases are presented as negative numbers) Cred default swaps variables are defined following Saretto and Tookes (2013);. 22

23 CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. CDS trading is an indicator variable equal to one if there is a traded CDS on the firm s debt during the year t. Risk related variables are defined following Fung et al. (2012); The firm s market is estimated as follows. I first run the following capal asset pricing model (CAPM). r i,t r f,t = α i,t + β i,t (r m,t r f,t ) + u i,t, where r i,t is firm i s daily return at time t, r m,t is a value-weighted market return, and r f,t is the risk-free rate. The regression coefficient, β, represents the market risk. The standard deviation of residuals from the CAPM regression represents the idiosyncratic risk. The standard deviation of the daily returns for an individual firm represents total risk. 23

24 Table 1 Descriptive Statistics This Table presents summary statistics for the sample of non-financial firms in the S&P500 index during the period. Net investment is defined as capal expendures plus acquisions minus sale PPE plus increase in investments minus sale of investments divided by start-of-period assets. Net equy issue is defined as sale of common and preferred stock minus purchase of common and preferred stock divided by start-of-period assets. Net LT Debt Issue is calculated as longterm debt issues minus long-term debt reduction divided by start-of-period assets. Market risk, is the beta coefficient of the firm from CAPM model. The standard deviation of residuals from the CAPM regression represents the idiosyncratic risk. The standard deviation of the daily returns for an individual firm represents total risk. Return on Equy is operating income before depreciation divided by book equy. Cash Flow equals to income before extraordinary ems divided by lagged book assets. Market-to-Book is sum of market equy, total debt, preferred stock liquidating value minus deferred taxes and investment tax creds divided by book assets. Z-Score is sum of (3.3 pre-tax income), sales, (1.4 retained earnings), (1.2 (current assets current liabilies) divided by book assets. Term Spread is the difference the yield spread between the 1- and 10-year Treasury bonds. Equy Market Return is CRSP annual value-weighted return. Size is log of sales. Book Leverage is total debt divided by book assets. Table 1 Descriptive Statistics S&P 500 S&P 500 CDS traded S&P 500 no-cds traded St. St. St. Variable Mean Median Dev. N Mean Median Dev. N Mean Median Dev. N Net investment Net equy issue Long term debt issue Short term debt issue Market risk (betas) Idiosyncratic risk (std. residuals) Total risk (std. returns) ROE Cash flow Market to book Altman's Z-score Market return Term spread Leverage (book leverage) Size (log(sales))

25 Table 2 : Univariate comparisons This Table provides univariate comparisons of firms wh no-cds traded on their debts wh firms that have CDS traded on their debts at least once or currently CDS being traded on their debts. CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. CDS trading is an indicator variable equal to one if there is a traded CDS on the firm s debt during the year t. Net investment is defined as capal expendures plus acquisions minus sale PPE plus increase in investments minus sale of investments divided by start-of-period assets. Net equy issue is defined as sale of common and preferred stock minus purchase of common and preferred stock divided by start-of-period assets. Net LT Debt Issue is calculated as longterm debt issues minus long-term debt reduction divided by start-of-period assets. Market risk, is the beta coefficient of the firm from CAPM model. The standard deviation of residuals from the CAPM regression represents the idiosyncratic risk. The standard deviation of the daily returns for an individual firm represents total risk. Return on Equy is operating income before depreciation divided by book equy. Cash Flow equals to income before extraordinary ems divided by lagged book assets. Market-to-Book is sum of market equy, total debt, preferred stock liquidating value minus deferred taxes and investment tax creds divided by book assets. Z-Score is sum of (3.3 pre-tax income), sales, (1.4 retained earnings), (1.2 (current assets current liabilies) divided by book assets. Change in cash equals to change in cash equivalents divided by start of period assets. T-stats are in parentheses. No CDS Traded vs. CDS traded Net investment (6.22) LTD issue (4.53) Market Risk (9.36) Idiosyncratic risk (11.52) Total Risk (10.07) Equy issue (-0.91) Change in Cash (8.87) Z-score (0.08) No CDS Traded vs. CDS trading (7.86) (4.90) (6.89) (11.74) (7.24) (-1.55) (6.82) (0.21) 25

26 Table 3: CDS impact on change in cash holdings The model is an OLS regression wh year and firm clustered standard errors. Dependent variable is change in cash holdings of firm, cash equivalents divided by start of period assets, control variables are determined following Opler et al.(1999) paper. CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. CDS trading is an indicator variable equal to one if there is a traded CDS on the firm s debt during the year t. Cash Flow equals to income before extraordinary ems divided by lagged book assets. Market-to-Book is sum of market equy, total debt, preferred stock liquidating value minus deferred taxes and investment tax creds divided by book assets. Size is log of sales. Leverage is total debt divided by book assets. Capal expendures equals to capal expendures divided by total assets of the firm. *, **, *** are significant at 10%, 5%, and 1% levels respectively. Variable Coefficient t-stat. P-value Intercept *** CDS traded * CDS trading *** Cash flow Market to book *** Size *** Leverage Capal expendure Year clustered SE X Firm clustered SE X R-squared N

27 Table 4: CDS impact on equy, long term and short term debt issue The models are OLS regressions wh year and firm clustered standard errors. Dependent variable, net equy issue equals to sale of common and preferred stock minus purchase of common and preferred stock divided by start-of-period assets. CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. CDS trading is an indicator variable equal to one if there is a traded CDS on the firm s debt during the year t. Following definion of S&P, investment grade dummy equals to one if firms rate BBB- or higher, zero if firms are rated BB+ or lower. Cash Flow equals to income before extraordinary ems divided by lagged book assets. Market-to-Book is sum of market equy, total debt, preferred stock liquidating value minus deferred taxes and investment tax creds divided by book assets. Term Spread is the difference the yield spread between the 1- and 10-year Treasury bonds. Equy Market Return is CRSP annual value-weighted return. Size is log of sales. No rating is a dummy variable if firm has no cred rating at the end of the year. Model 1 examines overall impact of CDS on equy issue, while model 2 separates the impact according to investment grade firms and non-investment grade firms. *, **, *** are significant at 10%, 5%, and 1% levels respectively. Model 1 (Net equy issue) Model 2 (Long term debt issue) Model 3 (Short term debt issue) Coeff. t-stat. P-value Coeff. t-stat. P-value Coeff. t-stat. P-value Intercept *** *** CDS traded * CDS trading *** * * Invest. Grade *** * Inv.Grade*CdsTrading *** * * Cash flow *** Market to book *** Size *** *** * Zscore Term spread *** *** *** Market return *** * No rating *** * Year clustered SE X X X Firm clustered SE X X X R-squared N

28 Table 5: CDS impact on net investments The models are OLS regressions wh year and firm clustered standard errors. Dependent variable, net investment is defined as capal expendures plus acquisions minus sale PPE plus increase in investments minus sale of investments divided by start-of-period assets. CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. CDS trading is an indicator variable equal to one if there is a traded CDS on the firm s debt during the year t. Following definion of S&P, investment grade dummy equals to one if firms rate BBB- or higher, zero if firms are rated BB+ or lower. Cash Flow equals to income before extraordinary ems divided by lagged book assets. Market-to-Book is sum of market equy, total debt, preferred stock liquidating value minus deferred taxes and investment tax creds divided by book assets. Term Spread is the difference the yield spread between the 1- and 10- year Treasury bonds. Equy Market Return is CRSP annual value-weighted return. Size is log of sales. No rating is a dummy variable if firm has no cred rating at the end of the year. Model 1 examines overall impact of CDS on net investments, while model 2 separates the impact according to investment grade firms and non-investment grade firms. *, **, *** are significant at 10%, 5%, and 1% levels respectively. Model 1 Model 2 Variable Coefficient t-stat P-value Coefficient t-stat P-value Intercept *** *** CDS traded CDS trading ** *** Invest. Grade *** Inv.Grade*CdsTrading *** Cash flow * ** Market to book Size *** *** Z-score Term spread *** *** Market return No-rating *** Year clustered SE X X Firm clustered SE X X R-squared N

29 Table 6 : CDS impact on firms risk profiles The models are OLS regressions wh year and firm clustered standard errors. Market risk, is the beta coefficient of the firm from CAPM model. The standard deviation of residuals from the CAPM regression represents the idiosyncratic risk. The standard deviation of the daily returns for an individual firm represents total risk. CDS traded is an indicator equal to one if there is a CDS market for the firm s debt at any time during the sample period. CDS trading is an indicator variable equal to one if there is a traded CDS on the firm s debt during the year t. Market-to-Book is sum of market equy, total debt, preferred stock liquidating value minus deferred taxes and investment tax creds divided by book assets. Z-Score is sum of (3.3 pre-tax income), sales, (1.4 retained earnings), (1.2 (current assets current liabilies) divided by book assets. Equy Market Return is CRSP annual value-weighted return. Size is log of sales. Dividend is a binary variable that equals to one if firm paid dividend during the year t, zero otherwise. *, **, *** are significant at 10%, 5%, and 1% levels respectively. Idiosyncratic risk model Market risk model Total risk model Variable Coeff. t-stat P-value Coeff. t-stat P-value Coeff. t-stat P-value Intercept *** *** *** CDS traded *** *** CDS trading Market to book Size *** *** *** Dividend ** * Z-score *** *** *** Term Spread *** ** *** Market return *** *** Year clustered SE X X X Firm clustered SE X X X R-squared N

30 Figure 1: Number of firms that have CDS trading on their debt Figure 1 presents the number of firms that have CDS trading on their debt in a given year. Sample includes all nonfinancial S&P500 firms from

31 Figure 2: Average of Change in Cash around CDS introduction Figure 2 presents the average of changes in cash holdings of the CDS users one year before (t-1) CDS are started to be traded on firms debt, during CDS inception year (t), and following two years (t+1, t+2). Sample includes all nonfinancial S&P500 firms from CDS users 0 t-1 t t+1 t Time : t-1 t t+1 t+2 Av. Change: