How should we measure the performance of corporate bond mutual funds? Evaluating model quality and impact on inferences

Abstract

The performance of corporate bond mutual funds tends to be estimated using models with limited empirical validation. We survey the literature to determine the models in use and develop a representative set of models. Testing that set of models, we find considerable variation in quality, with the most effective models sharing common traits. We recommend, among the tested models, the four-factor model proposed by Jones and Mo (2021). Regarding the stylized facts of corporate bond fund performance, our recommended model produces notable deviations from the prior literature and other models, including less evidence of positive alphas not attributable to luck.