This paper develops a model to estimate the implied default probabilit
y of corporate bonds. The model explicitly considers the risk averse b
ehavior of investors to provide a more precise framework for estimatin
g the implied default probability. A Kalman filter method is used to e
stimate time-varying risk premium associated with the investor's risk
aversion. The results of nonlinear regressions indicate that previous
risk-neutrality models consistently overestimate the implied default r
ates of corporate bonds. The results also suggest that investors may h
ave been adequately compensated for investment in risky bonds.