Providing Incentives for Mortgage Originators

We are in the process of looking at the McDash database on loan-level performance. We want to match the variables analyzed in our model to observable cross-sectional heterogeneity in firm characteristics (servicer practices, age of the firm), or variation across periods in the characteristics of the industry (access to credit, house prices). Using our model’s implications on the pooling versus sorting decisions of firms as the identifying assumptions, we want to estimate the repayment probabilities of pooled and sorted portfolios. These differences may be both in level and in their evolution over time. What is the expected cost of default in the next few years, for a given prediction of house price evolution? Do bad loans default earlier, and what does this imply for the timing of defaults in the coming years? The calibration of our model should allow us to perform counterfactual exercises and shed some light on these questions. Conclusions from the data analysis can also be used in future research on the design of optimal servicing of loans scheme, i.e., the provision of incentives for mortgage repayment.