Modeling prepayment for mortgage-backed security in Korea

The financial crisis of 2007-2009 exposed the credit risks of mortgage-backed securities (MBS). Nevertheless, prepayment remains one of the most important risks of MBS for MBS issuers and purchasers. When mortgages are paid off early, the flow of interest income that MBS purchasers were expecting is terminated. The risk of prepayment must be managed. MBS sellers or purchasers often obtain insurance to protect against this risk. The purpose of this study is to present a model of prepayment rates that is more accurate and precise than other models currently in use. The new model explains the prepayment rate by using the method of panel-corrected standard errors (PCSE). With this method, refinancing incentive, burnout, seasoning, prepayment penalty, and changes in house prices have a substantial effect on the prepayment rate. This study also shows that a feasible generalized least squares (FGLS), which often results in overconfidence for variables, can erroneously identify certain variables as significant. Such variables include the age of mortgage loans with burnout, seasonality, and unemployment rate. Another major finding of this study is that the development of dummy variables for the early months of mortgage pool can provide a better mechanism than the mortgage loan age variable to estimate borrowers’ prepayment characteristics. In addition, this study reveals that, contrary to expectation, the global financial crisis did not lead to a significant increase in the refinancing incentive sensitivity after controlling mortgage loan variables.