How to (Properly) Compute Credit Default Swap Returns

This paper proposes empirical methods to measure Credit Default Swap (CDS) return and explores its factor structure. We find that approximated CDS returns deviate significantly from actual returns based on the upfront fee, computed with protection sellers' cash flows. Past CDS returns and the skewness positively predict CDS returns, suggesting that CDS buyers have lottery preferences and CDS sellers are either overconfident or speculative in trading. Conventional pricing factors have weak explanatory power. Corporate bonds have spillover effects on future CDS returns in that a zero-cost portfolio sorted by statistical moments of past bond returns explains the CDS cross section.