A deconstruction of Cox’s proportional hazards model and an inquiry into its ability to predict the outcome of Chapter 11 bankruptcy proceedings

Abstract

Set against the backdrop of Chapter 11 proceedings in the United States, a specification of Cox’s proportional hazards model proposed by Partington, Stevenson, Russel and Torbey (2001) was examined for its stability over time. Faced with findings of instability consistent with those of Wong, Partington, Stevenson and Torbey (2007), the model was respecified to only include two firm-specific covariates (capturing firm size and earnings) and a time-dependent market-wide covariate. A ‘calendar-time model’ was then introduced to enable analysis of the firm-specific covariates in abstraction from the systematic effects of time. With such effects controlled for, the suggestion was made that any remnant instability in the model was a result of non-systematic factors reflecting the changing nature of firms which filed for Chapter 11 protection during the period examined.