This paper presents a review and synthesis of the past fifteen years of research on mortgage termination risk. Understanding termination risk is fundamental to explaining the workings of the now $13 trillion mortgage market. The review covers theoretical, empirical, and methodological work, including both residential and commercial market segments. In addition to synthesizing existing findings, open questions for further research are identified. The overall objective is to introduce this important topic to journal readers who may not be actively involved in the mortgage field and to stimulate thinking about open issues that deserve additional research.
Mortgage loans are among the most complex of financial contracts due to the uncertainty of their duration. This uncertainty arises because they may terminate early as a result of either default or prepayment. Default occurs when the borrower cannot, or will not, make contractually promised payments. Prepayment occurs when the loan is paid in full prior to maturity. Causes for these two outcomes are varied, adding to the complexity of explaining and, ultimately predicting, the rate of termination for the loans that are either held in portfolio by financial intermediaries or which provide backing by mortgage-backed securities (MBS). While textbook treatments of this topic (e.g., Clauretie and Sirmans, 1998) typically introduce readers to related industry jargon and provide a few examples of how early termination rates can affect values of mortgage-related assets, the underlying research on termination risk is mentioned only briefly.
To provide a sense of the variation in recent termination rates, consider the following stylized facts on residential mortgages as of December 2006.' While the overall rate of serious delinquency2 for prime single-family mortgages was 0.7%, there was considerable geographic variation, ranging from 0.3% in California to 2.3% in Louisiana. In the subprime segment of the market, the national rate of serious delinquency was 7.8% but as high as 14% in Louisiana and Mississippi; both states recently affected by extensive hurricane damage, of course. In terms of prepayment, prime loans showed a 15.7% rate nationally and subprime loans a 33% rate,3 much higher than would result from household mobility alone.4 For comparison, as of fourth quarter of 2006, the delinquency rate for commercial mortgages on bank balance sheets was 1.3% (Federal Reserve, 2006)5 and the delinquency rate on commercial mortgage-backed securities was at an all-time low of 0.27% (Standard and Poor's, 2007).
What economic factors produce these results and how do they vary across loan types and over time? Couching the question in an option-theoretic framework, what factors affect exercise of these options and how may those processes best be modeled? 1 review relevant research, identify major themes, and attempt a synthesis: What do we know and what do we not know? In the process, I will also identify open questions and industry trends likely to produce new answers, and probably new questions, in the future.
I focus on the literature from 1992 to 2007 for several reasons. First, the journal published a previous review (Dickinson and Heuson, 1994), which surveys the early research up to 1992, though limited in focus to prepayments in the residential mortgage market. Second, two frequently cited reviews of default risk were published at about this same time (Quercia and Stegman, 1992; and Vandell, 1993). Finally, the period 1992-2007 has been a particularly dynamic period, with the development of commercial mortgage-backed securities, creation and growth of the subprime residential segment, wide adoption of automated underwriting, and generally low interest rates and robust property values, at least up until the housing market downturn in late 2006.
Why should we be interested in mortgage termination rates? The main objective has been to explain the pricing of mortgages and mortgage-related assets. …