Masters in Statistics Oral Defense: "Mortgage Transition Model Based on Loan Performance Data"

Abstract: The unexpected increase in loan default on the mortgage market is widely considered to be one of the main cause behind the economic crisis. To provide some insight on loan delinquency and default, I analyze the mortgage performance data from Fannie Mae website and investigate how economic factors and individual loan and borrower information affect the events of default and prepaid. Various delinquency status including default and prepaid are treated as discrete states of a Markov chain. One-step transition probabilities are estimated via multinomial logistic models. We find that in general current loan-to-value ratio, credit score, unemployment rate, and interest rate significantly affect the transition probabilities to different delinquency states, which lead to further default or prepaid events.

Key words: Loan performance, Markov chain, Mortgage, Multinomial logistic model, Transition probability, Transition Matrix.

Host: Jimin Ding