Duration Models in Loan Management

The purpose of this study is to estimate the future duration of a loan contract on the basis of several factors. The main methodology consists of a brief explanation of a survival analysis and a thorough application of a survival analysis in loan management. A real dataset from a credit institution (situated in Varna) is used. All contracts were signed for 30 days but some contracts were ended earlier, others - later. The main research question concerns the following statement. We may try to predict future loan duration by making an econometric model describing the dependency between the loan duration (as a dependent variable) and several independent variables. The dataset is analysed by calculating life tables, applying the Kaplan-Maier method and using Cox regression within SPSS. It is has been proved that the main covariates affecting loan duration are the variables: born in the region, month of birth and age. The formulated conclusions are valid for the analysed credit institution. This work provides a methodology for adapting duration models in credit institutions. The presented methodology (in this paper) may be applied over the dataset of other credit institutions (including banks) for loan duration prediction.