Understanding mortgage choice

Abstract

The home is the largest asset for many U.S households and these purchases are almost always financed by a mortgage. Prior research shows United States borrowers to be unusual in their high use of fixed rate mortgages (FRMs). In other industrialized countries borrowers predominantly utilize adjustable rate mortgages (ARMs). The prior literature also argues that borrowers in the United States would benefit from greater use of adjustable-rate mortgages. For example, 5/1 ARMs are the most common variety of ARM with a low (relative to FRMs) fixed interest rate for the first five years that then adjusts annually. Given that the average borrower keeps the same mortgage for about five years, 5/1 ARMs offer payment advantages of roughly 15 percent per month relative to fixed rate mortgages. Despite this, U.S. borrowers overwhelmingly opt for fixed-rate mortgages. Prior work demonstrates broad consensus regarding the factors that should influence mortgage choice - the decision to take a fixed rate or adjustable rate mortgage. However, attempts to test the theory with empirical data have been limited. Some studies used small, regional borrower samples or national samples that lacked information about borrowers. Coulibaly and Li (2009) use national data from the Survey of Consumer Finances that includes borrower data such as moving expectations and income volatility. However, Coulibaly and Li (2009) have relatively weak proxies for risk aversion and financial literacy - two factors that are likely to influence mortgage choice. This dissertation consist of three related essays on topic of mortgage choice. My first essay (Essay A) builds on the literature by using data from the Panel Study of Income Dynamics (PSID) to examine the factors that influence mortgage choice. Like Coulibaly & Li (2009) this work uses a national sample and incorporates both pricing and borrower data. However, the PSID allows me to better measure risk aversion and financial literacy. I confirm that many of the factors hypothesized to influence mortgage choice have the expected effect on mortgage decisions. These include pricing variables such as FRM-ARM rate differential and yield curve steepness as well as borrower characteristics such as likelihood of moving and borrowing constraints. However, other findings (or lack thereof) in areas like the impact of risk aversion, education and income volatility point the way towards future work. My second essay (Essay B) adds to both the mortgage choice and the intergenerational linkages literature by exploring a potential intergenerational linkage on mortgage choice. Intergenerational transmission of knowledge, wealth and preferences has long received attention in the labor economics literature. That literature finds that earnings, education, IQ/ability, health, wealth, behaviors and attitudes tend to persist from one generation to the next. Using data from the PSID and controlling for other price and borrower factors, I find that adult children whose parents have ever had an ARM are significantly more likely to choose ARMs themselves. To my knowledge, this is the first time this factor has been documented and adds to our understanding of how borrowers in the United States make mortgage choice decisions. Coulibaly and Li (2009) and my own work utilize national datasets to demonstrate that borrowers react as expected to many of the factors hypothesized to influence mortgage choice. However, both Coulibaly and Li (2009) and I are unable to control for behavioral attributes except risk aversion, we both use proxy measures of financial literacy, and suffer from unobserved heterogeneity in the mortgage choice faced by borrowers due to differences in timing and life circumstances. My third essay (Essay C) builds on the literature by presenting individuals with a hypothetical standardized mortgage choice. I do this by clearly defining two, fairly typical mortgages for the borrower to choose between - one a FRM and the other an ARM. Within the scenario, the borrower is given an expected holding period, payments for each option, and payment differentials. By presenting the borrower with a specific hypothetical mortgage choice that has fixed assumptions and a clear optimal choice, I am better able account for the heterogeneity that has confounded prior studies. To better understand which factors may influence this decision, I collected additional information on borrowers' patience, risk aversion, and impulsiveness, and their financial literacy related to their knowledge of debt. The debt-related questions I utilize were developed by Gathergood and Weber (2017) after Van Ooijen and Van Rooij (2016) showed that generic financial literacy measures were not necessarily correlated with debt-specific financial literacy. Like past work, I find that financial literacy is alarmingly low. Not surprisingly then, when asked to pick the optimal mortgage loan in a well-described and unambiguous situation, respondents were only able to do so 52 percent of the time. I find that higher levels of debt-specific financial literacy is a key determinant of respondents' ability to make an optimal mortgage choice. However, experience-related variables, such as age, homeownership, and being head of household are not. To my knowledge, this is the first work in the U.S. to field these debt-specific financial literacy questions, to examine their impact on a clearly defined mortgage choice scenario, and to test to what extent the language used to label the mortgage options in the question impacts responses. It is also the first time that the role of patience and impulsiveness are examined as they relate to mortgage choice. Finally, the risk aversion data here is superior to previous work. It is more concurrent with the mortgage decision than the PSID data I used in my first essay. (The PSID only collected this data in 1996 and I applied it to borrower decisions made from 1994 to 2013.) And, it's based on a more proven risk aversion assessment than Coulibaly & Li (2009). Together, these three papers add to our understanding of the factors that influence mortgage choice in the United States. The PSID allows me to use a better proxy of financial literacy (education) and measurement of risk aversion than prior studies. My custom-designed survey improves on this by measuring debt-specific financial literacy and collecting information on borrowers' attitudes concerning risk aversion, patience and impulsiveness. Finally, I demonstrate an intergenerational linkage in mortgage choice, a significant factor that had to date, not been documented.