Default risk of Icelandic consumer loans : implementation of a credit risk model

Abstract

Online consumer lending in Iceland has seen significant growth in the past years. With online applications and live data monitoring comes more access to useful data that has the possibility of improving the assessment of default risk. If assessment of default risk can be improved, the opportunity for offering a broader range of interest rates on these loans come available, resulting in a more accurate pricing of each loan based on the estimated default risk of each borrower. The results indicate that a logistic regression model can improve the assessment of credit risk. Using alternative data sources, in particular loan application data, can also prove to be of additional benefit when assessing the probability of default. A reduction of type II errors, lowering the number of defaulted borrowers in the portfolio of a credit lender does however come at the cost of forgoing borrowers that would otherwise be able to pay off their loans, that is an increase of type I errors.