Measuring and explaining the CDS-bond basis term-structure shape and dynamics

The CDS-bond basis quantifies the difference in risk premia between credit default swap (CDS) and bond markets. It is hard to measure at the individual firm level given substantial missing-value problems (30%-100%) in either or both markets, even for highly liquid blue-chip financial firms. We propose a novel imputation approach to obtain full historical firm-level basis term-structures across all maturities. Our approach can accommodate different term-structure interpolation methods, including Nearest-Neighbor, spline, and Nelson-Siegel interpolation. Using the new methodology, we construct the full history of the 2011-2021 JP Morgan (JPM) basis term-structure and use it to analyze its empirical determinants. We find that factors like market liquidity, funding liquidity, counterparty risk, and the default premium all impact the basis term-structure, though not all at the same moment in time. All factors are statistically significant during the Covid-19 pandemic. The various empirical limits-to-arbitrage proxies correlate differently with different parts of the basis term-structure, stressing the need to model the full basis term-structure rather than assuming it to be flat. The results are robust for other blue-chip financials, each time requiring the full basis term-structure imputation approach as proposed in this paper.