Comparing Models of Corporate Bankruptcy Prediction : Distance to Default vs. Z-Score

This paper examines the performance of two commonly applied bankruptcy prediction models, the accounting ratio-based Altman Z-Score model, and the structural Distance to Default model which currently underlies Morningstar's Financial Health Grade for public companies (Morningstar 2008). Specifically, we tested the following:1. The ordinal ability of each model to distinguish companies most likely to file for bankruptcy from those least likely to file for bankruptcy as measured by the Accuracy Ratio2. The cardinal ability of each model to predict bankruptcy as measured by the bankruptcy rate of healthy-scored companies and the average rating before bankruptcy3. The decay of the ordinal performance of each model over time as measured by the cumulative percentage change in Ordinal Score4. The stability of the ratings of each model as measured by the Weighted Average Drift DistanceWe are cognizant that the Z-Score was not intended to be used on non-manufacturing companies (Altman 2002). However, in practice we find it is commonly used to gauge the financial health of all companies. For our purposes, we found it more relevant to include we did include non-manufacturing companies in our testing universe. We found that Distance to Default has superior ordinal and cardinal bankruptcy prediction power within our universe. It also has a more durable bankruptcy signal, but it generates less stable ratings than the Z-Score