Journal of Real Estate Finance and Economics, 23:2, 185±211, 2001
# 2001 Kluwer Academic Publishers. Manufactured in The Netherlands.
Competing Risks of Mortgage Termination: Who
Re®nances, Who Moves, and Who Defaults?
ANDREY D. PAVLOV
Faculty of Business Administration, Simon Fraser University, British Columbia, Canada
Why, when, and who terminates their mortgages? The primary reasons for mortgage termination are re®nancing,
selling of the property, and default. This article is the ®rst to explicitly model these competing risks within a
uni®ed conceptual framework and provide a link between theoretical value-maximizing mortgage-termination
models and empirical estimation. I ®nd, for instance, that the re®nancing risk is highly sensitive to interest-rate
changes and other variables capturing the value of the mortgage. On the other hand, the necessity to relocate,
either through sale of the property of default, is sensitive to the local economic conditions but largely independent
of the value of the mortgage. Furthermore, I explicitly model the spatial distribution of the mortgage-termination
risks. This approach captures striking spatial patterns of mortgage termination. It also mitigates, at least partially,
one of the biggest obstacles to mortgage termination estimation: omitted variables.
Key Words: mortgage prepayment, re®nancing, default, household mobility, mortgage-backed securities
The volume of outstanding mortgage-backed securities in 1997 approached $1.7 trillion
and encompassed nearly half of all outstanding United States' mortgage loans.
Homeowners can substantially impact the holders of these securities by exercising their
right to prepay or default. If an investor buys a security well above (or below) par, an
unanticipated return of principal can result in a substantial loss (or gain). Understanding
when, why, and who terminates their mortgages can help investors better assess
prepayment and default risks and ultimately reduce the required rate of return on
This article makes several important contributions to the literature investigating
mortgage-termination behavior. First, I explicitly model the various causes of mortgage
termination as competing risks and thus follow more closely the implications of the related
theoretical utility-maximizing frameworks. Utilizing mortgage and transaction data from
Los Angeles, California, I demonstrate that the re®nancing risk is highly sensitive to
interest-rate changes and other variables capturing the value of the mortgage. The
necessity to relocate, however, is sensitive to the local economic conditions but largely
independent of the value of the mortgage. Furthermore, the traditional noncompeting risks
estimation of the overall mortgage-termination risk is misleading, especially for mortgage
characteristics and economic conditions away from the mean.