CAPÍTULO 5. CONCLUSIONES Y DISCUSIÓN
5.1. Discusión
Overall, the results outline that the credit extension that as due to the availability of CDS contracts is less pronounced (and potentially countered) for CDS firms when their creditors on average use CDS contracts for hedging purposes relative to when their CDS creditors are net sellers of CDS contracts. However, as the findings furthermore suggest a clear separation between the credit amount provided by CDS and non-CDS creditors, the results may also raise concern about that the results are driven by firm characteristics rather than creditors’ CDS trading. On the one hand, creditors may to a larger extent become net CDS buyers when the credit risk of the firm is high andtherefore provide less credit to these firms. At the same time, non-CDS creditors of the same firm may provide even less credit due to their lack of hedging towards firms with, potentially, higher credit risk. Similar, creditors’ may only sell credit protection on borrowers that are relatively safe firms. On the other hand, if the variation in the aggregated credit exposures of a firm is not driven by firm characteristics then the CDS holdings by a firm’s creditors may imply unanticipated financial constraints to the firm. In particular, this may explain why firms may choose to switch to new lenders after the inception of CDS trading as
25In particular, I separate the positive observations of the CDS coverage ratio based upon the Q25
suggested in Shan, Tang, and Yan (2016). To provide more insights about the issue of unobserved firm heterogeneity, I next investigate the role of firm characteristics for the effect on firm-specific credit exposures.
As CDSs hedge the buyer against the credit risk of the reference firm, I start by examining to what extent lagged values of measures that reflect a firm’s financial health explain creditors’ CDS holdings. In particular, I use proxies for the firm’s credit risk (Z-score and Rating), income (Profitability), debt financing (Leverage and Short-term Debt), as well as financial constraints (Firm Size). The analysis is conducted in the sample of CDS firms only and the results are presented in Table 6, Panel A.
- Table 6 -
In Models (1) to (3), I focus on the role of credit risk measures in explaining the
presence of CDS creditors by regressing the dummy variableCDS Outstanding, as well as
the dummy variables where I condition on creditors’ net CDS positions (CDS Outstanding
× Net Buyers and CDS Outstanding × Net Sellers), on the set of measures of firms’ financial health and industry-fixed effects. The results show that the presence of CDS creditors is positively correlated with firms’ leverage ratios and that firms with lower (higher) ratings, i.e., lower (higher) repayment probabilities, are more likely to have CDS creditors that are net protection buyers (sellers). Although the statistical significance is weak, the results furthermore suggest that firms are more (less) likely to have net selling (buying) CDS creditors when they are more (less) profitable and/or have less
(more) short-term debt. In Models (4) to (5) I likewise report the determinants of CDS
Coverage, conditional on creditors’ net CDS positions, in order to test whether the credit risk measures also explain the variation in the firm-specific CDS coverage ratios. Although the statistical significance is not as strong as before, the results show that higher levels of credit risk of the firm leads to higher positive CDS coverage ratios and, thus, reveal that firms’ credit risk and financial constraints positively relate to creditors’ CDS holdings.
Next, I utilize the results obtained from the investigation of the determinants of firms’ CDS coverage in a robustness test. Specifically, I use the CDS coverage residual to capture the CDS coverage impact and isolate the effects due to firm’s credit risk. In order to distinguish between the effect of net protection-buying creditors and net
protection-selling creditors I define CDS Coverage × Net Buyers Residual, respectively
CDS Coverage × Net Sellers Residual, to be equal to the average firm-specific CDS coverage residual value obtained from the determinants of the CDS coverage model given by Model (4), respectively Model (5), of Panel A in Table 6. Then I investigate the impact of the CDS coverage residual on firms’ total credit ratios, as well as the credit ratios held by CDS and non-CDS creditors to test whether creditors’ CDS holdings has a affect on firms’ overall debt financing. As the residual measure should only capture the effect of
creditors’ CDS holdings that is not driven by the credit risk of the firm, I only expect to find a significant effect for CDS creditors’ credit ratios to the extent that theystrategically change the credit exposures to their borrowers. In the regression on non-CDS creditors’ credit ratio I expect only a significant CDS coverage residual coefficient to the extent that non-CDS creditors off-set the changes in the credit supply by CDS creditors. The analysis is again conducted in the sample of CDS firms only and the results are presented in Table 6, Panel B. Models (1), (3) and (5) show the results in case of net protection- buying creditors and outline that the result of lower credit supply by CDS creditors is robust when controlling for the firm’s financial health while the effect on firms’ total credit ratio or the credit supply by non-CDS creditors is insignificant. As the coefficient in the model for firms’ total credit ratio is also negative, the results indicate that CDS firms have less credit available when their creditors are protection buyers. However, the effect may be minor in terms of their aggregate credit exposure, probably because non-CDS creditors partly off-set the reduction in credit supply. In contrast, Models (2), (4) and (6) outline robust results regarding an increase in firms’ total credit ratio and the credit supplied by CDS creditors in case of net protection selling creditors while the effect on the credit ratio of non-CDS creditors is insignificant. Overall, the robustness test provides evidence that CDS firms may benefit in terms of a credit extension from having creditors that are net protection sellers, but also may experience that credit is withdrawn when the creditors are net protection buyers.
In terms of the economic significance of these findings, it is clear that the individual firm will be more exposed to changes in the credit supply caused by creditors’ CDS holdings the more creditors of the firm hold (or are able to hold) CDS contracts on the firm’s debt. Furthermore, one would expect that the effect on the firm-specific credit
exposures is only pronounced for firms with a high share of CDS creditors if the effect
indeed stems from creditors’ CDS holdings. In other words, if the effect is purely driven by firm characteristics, then there should be no difference between the credit amount provided by CDS creditors and non-CDS creditors, respectively. I test this hypothesis by utilizing differences in the presence of CDS creditors across CDS firms. Specifically, I
separate CDS firms into the samples ofHigh CDS Creditor Ratio and Low CDS Creditor
Ratio, where High CDS Creditor Ratio (Low CDS Creditor Ratio) refers to firms where the ratio of CDS creditors to total creditors is above (below) the median value. The results are presented in Appendix Table A3 and provide evidence that both the results of lower credit supply in the case of net CDS-buying creditors and the results of higher credit supply in the case of net CDS-selling creditors are driven by the sample of firms that have a relative high share of CDS creditors. Hence, the results suggest that the effect is indeed driven by the CDS trading of a firm’s creditors and moreover that the
economic significance is largest for firms with a high share of CDS creditors.
Before turning to the discussion of the significance of these results relative to the liquidity in CDS markets, the next section studies whether creditors’ CDS holdings also affect firms’ debt financing more generally.