Probability of default in collateralized credit operations

The goal of this paper is to identify the major determinants of the probability of default in a mortgage credit operation, which is backed by collateral. We use an exclusive data set with 268,036 loan contracts and apply logistic regression and Cox proportional hazards model in the estimation. The discriminatory power of the estimated models is analyzed by several accuracy indicators. The inclusion of time-dependent macroeconomic variables in addition to covariates representing characteristics of the contract and individuals improved the overall performance. Logistic regression showed a higher discriminatory power than Cox proportional hazards model according to all accuracy indicators. It is worth mentioning the negative relationship between the probability of default and the economy base interest rate. Decreases in the base interest rate lead banks to lose revenue from treasury operations and expand credit operations to compensate the loss. This strategy brings individuals with a higher probability of default to the financial market.