Working Paper Series no. 860: Credit Default Swaps and Credit Risk Reallocation

Credit default swaps are derivative instruments that impact individual risk of investors. Except a few contributions, current research remains silent on the distributional consequences of trading credit default swaps on individual credit risk for two reasons. First, CDS are a zero-sum game in aggregate and payoffs are merely transfers inside the financial system. However, recent contributions such as Gabaix (2011) stress how individual shocks may affect aggregate outcome. Second, lack of granular data on both derivatives and debt holdings have prevented these studies.

In this paper, we use granular quarterly data on both debt and CDS exposures by French investors on non-financial corporations (NFC) and Euro-Area (EA) banks on French NFCs.^[1] We provide new answers to how CDS reallocate investors exposure to credit risk. Essentially, we show that CDS increase investors' portfolio risk, with a stronger effect for dealers and investment funds than for banks.

To guide our empirical investigation, we build a methodology to disentangle and characterize investor strategies by reference and period. This methodology exploits the sign, ratio, and timing of matched debt and CDS positions. There are broadly three motives behind CDS trading: arbitrage, hedging, and speculation. Arbitrageurs take offsetting positions in CDS and debt to benefit from relative price discrepancies. As Figure 1 depicts, this strategy is anecdotal and represents 2% of CDS purchasers (dark red bar), and 0.02% of CDS sellers. Hedgers use CDS as an insurance product to downsize corresponding debt exposures, either in reaction to shocks, or to maintain lending relationships. Hedging represents a mere 13 to 19% of CDS purchase (red bar), with other types of offsetting CDS purchases adding to 8%. Finally, speculators use CDS as an alternative venue to amplify debt exposures or to gain exposure without holding the underlying debt. The distribution of strategies already implies that the CDS effect on individual portfolio risk mainly depends on speculative strategies pursued, the light blue and blue bars in Figure 1.

We first find that CDS decrease exposure concentration, with hedgers purchasing CDS to cover their largest exposures, and speculators selling CDS when they hold relatively little underlying debt. However, since hedgers only represent a small fraction of CDS purchases, their impact on aggregate portfolio risk remains moderate. Overall, CDS decrease investors exposure concentration across investor sectors, as measured by the Herfindahl-Hirschman Index (HHI) and the Gini coefficient. Short and long speculative strategies contribute to this decline, while hedging does not potentially due to decreasing hedging ratios at the intensive margin. CDS-induced diversification may also have a limited effect on return diversification since country and sector concentration increase.

Second, we investigate whether CDS tilt investors relative risk exposures. We find that the use of CDS for short speculation increases with reference risk for banks and investment funds. Similarly, they shed off their riskiest exposures for hedging, which should be beneficial for their portfolio risk. The relative insensitivity of dealers to reference risk is consistent with their intermediation role in transactions, responding to demand rather than driving it. However, banks and dealers incentives to sell CDS increase with reference risk, a pattern that we do not observe for funds. One explanation for dealers is that the higher demand for hedging on riskier references reflects in higher risks borne by CDS selling counterparties. Since the margin advantage of debt increases with reference risk, our results imply that banks and dealers have higher incentives to benefit from it than funds.

Overall, accounting for CDS may have an ambiguous effect on portfolio risk. Investors might use CDS to diversify their exposures to references but they can also use them to gain exposures on riskier references. In the last section, we show that CDS translate into higher portfolio risk for all sectors, the increase being stronger for dealers and investment funds. The rise in portfolio risk is driven by speculating strategies, both on the long and short sides. Hedging strategies mitigate it but represent a small share of strategies.