The purpose of this paper is to propose an alternative method of bankruptcy prediction considering corporate financial situations changing over time. The paper suggests an econometric approach to bankruptcy prediction by incorporating macroeconomic variables in a simple hazard model of Shumway(2001).
The study examines 367 non-financial firms listed in KSE at the time of 1990; among the 367 firms, 127 went bankrupt over the period 1990-2000. To investigate the forecasting accuracy in bankruptcy prediction models, this study compares three different econometric models: a logit model, a simple hazard model (not including baseline hazard rates), and a discrete-time hazard model including macroeconomic variables as baseline hazard rates. For this purpose, three macroeconomic variables - the volatility of exchange rates, prior bankruptcy rates and the volatility of interest rates - are considered instead of the usual time-dummy variables as baseline hazard rates. The empirical results show the superiority of the discrete-time hazard model including macroeconomic variables as baseline hazard rates to other models.