Comparing Bankruptcy Prediction Models in Iran

This paper compares the predictive power of a number of previous research models on bankruptcy prediction in Tehran Stock Exchange (TSE) from 2001 to 2009. To compare the predictive power of these models, the adjusted R2 (as In-sample metric) and root mean squared error (RMSE), mean absolute error (MAE) and mean absolute percent error (MAPC) (as out-of-sample prediction metrics). Finally, based on the estimation results of previous models in Iran, we present a final version logit type model that has higher performance than other models. The empirical results of In-sample and out-of-sample prediction power indicate that our presented model has higher adjusted R2 and lower RMSE, lower MAE and lower MAPC, too.