Computing corporate bond returns: a word (or two) of caution

Abstract We offer several suggestions for researchers using corporate bond return data. First, despite clear instructions from older papers (e.g., Bessembinder et al., The Review of Financial Studies 22:4219–4258, 2009) about ways to compute credit excess returns, a lot recent research simply subtracts Treasury Bill return. show that this imprecision is likely contaminate inferences, as the rate component returns negatively correlated spread component. This problem all looking at especially time series analysis and safer bonds investment grade). provide simple approach Wharton Research Data Services (WRDS) data remove interest returns. Second, we note significant differences in coverage across Trade Reporting Compliance Engine (TRACE) platform typical indices. some rules who are TRACE select subset closest those contained inside indices used by institutional investors. Third, differential quality prices hence between Corporate provided (i) correctly estimate (ii) synchronous entire set bonds, allowing consistent cross-sectional comparability, (iii) suffer less stale pricing issues. Due these issues, should try, where possible, source multiple sources ensure robustness their results.