Cumulative accuracy profile in banks' credit risk assessment: accounting based models and market based models

This study examine the Merton structural model based on market data and the discriminant analysis model based on accounting data in banks during 1386 to 1398. Due to the different structure of banks' balance sheets, for the first time, using the transformed data maximum likelihood estimation method and other liablilty with an adjustment and calculate the market value of assets and their volatility Using the stock price, we calculated the distance to default and the probability of default with the modified Merton model. Then, with the discriminant analysis model and Wilkes lambda index, we introduced a model based on accounting data to measure credit risk in banks. Among variables, inactive credits to total credits, total credits to main deposits and reserves to inactive credits have the most impact on determining the credit risk of banks, respectively, which is determined by z coefficients. The lower  z score, the greater credit risk and vice versa. Finally, using the cumulative accuracy profile and accuracy ratio, which is a new method in determining an efficient model for credit risk, Merton's structural model is compared with z-score model, and finally Merton's structural model with an accuracy ratio of 70.97 as an efficient model for measuring credit risk in banks