Using bank transactions to assess credit risk

Abstract: A lack of proper credit risk assessment can have catastrophic consequences, not only for individual borrowers or lending banks, but for society as a whole, which was the case in the financial crisis of 2008. Therefore it is essential that any risk assessment process has as much information as possible at hand so that the delinquency predictions are as accurate as they can be. This study uses real loan data to investigate whether incorporating information about gambling, making collection payments, and being dishonest in reporting income in risk assessment models improves their power to predict delinquency of loan payments. Prior literature has considered both hard and soft information in credit risk assessment, and this study contributes to that by using a unique dataset which provides the opportunity to test whether the behavioural patterns mentioned above, which have not previously been used in credit risk assessment, is associated with delinquency. The results shows that there is no significant connection between the researched factors and delinquency of loans, but that borrowers with a history of gambling or collection payments still tend to have a higher perceived risk which is reflected in the interest rate of their loans.