Identifying patterns of the dynamic credit risk of banks customers and financial institutions: case study- an Iranian bank

Credit risk assessment has always been one of the most important concerns of banks. Widely used models such as financial models have been used to assess credit risk so far. But increasing non-performing loans indicates that today these models cannot assess the credit risk of customers. Inconstant and uncertain environmental, social and political factors affect customer behavior and change customer behavior over time so that low risk customers become high risk customer. Hence a dynamic model had better to design for customers’ credit risk evaluation. Recognizing patterns of changes of customer behavior is a key factor in the bank's success in assessing and managing customer credit risk. Previous research has used conventional models such as financial and empirical data models. In this study, considering the dynamics of customers over time in different segments of low risk, medium risk and high risk and using big data analysis methods, dynamic customer behavior patterns have been extracted. In this study, the loans information of individual customers of three banks in a sample of 284,000 customers between 1389 and 1396 were examined and 30 variables were used to assess credit risk. The results showed that the model with a sensitivity of 92.1% and a detection of 89.1% has a high efficiency and performs better and has a higher efficiency compared to the probit, logit and neural network models. The result of this research is the recognition of dynamic patterns of customers' credit risk and using it. The bank has a more accurate assessment of customers' credit risk. Also in this method there is no need to retrain the data and run the algorithms again, therefore, this method saves time and money. Most importantly, the results of this study are significantly different from the results of the current bank model (static method).