The Credit Spread Puzzle - Myth or Reality?

Abstract

Many papers find that standard structural models predict corporate bond spreads that are too low compared to actual spreads, givin rise to the so-called credit spread puzzle. We show that the puzzle derives in large part from strong biases and low statistical power in commonly adopted approaches to testing the models. The biases are due to Jensen's inequality and arise when tests are carried out on a representative firm rather than on individual firms. Using data on individual firms during 2002-2012 we quantify the size of the bias in spread predictions and find it to be particularly severe for high-quality firms and short-maturity bonds. The problem of low statistical power arises because ex-post realized default frequency - often used to calibrate models - is a very poor estimate of ex-ante default probability. Finally, we test the Merton model via a bias-free approach using around 400,000 transactions in the period 2002-2012. We find that the Merton model captures both the average level and time series variation of the long-term BBB-AAA spread.