We develop a structural bond pricing approach and implement it on a large panel of US industrial bonds using an efficient maximum likelihood methodology. We evaluate the model's ability to predict yield spread levels and changes out-of-sample. Errors are smaller and distinctly less variable than...