Wharton GIS Lab


Default/Delinquency Models

Community Reinvestment and Credit Risk; An analysis of loan performance in relation to credit history scores and neighborhood variables.

Empirical Model:
Property price information;
Borrowed data;
Credit history;

Tract aggregate mortgage lending activity is very good predictor of loan performance. Properties that are unusually expensive for neighborhood are source of risk. The credit score is strongly predictive of loan performance.

The model finds that the probability of being 60 days or more delinquent is five times as large in tracts with mortgage activity near the sample mean than in those tracts with activity levels about twice the mean. Also, it is four times greater for borrowers with credit scores near the low end of the score range as compared to borrowers with scores near the high end.