ELIMINATING CREDIT BARRIERS TO INCREASE HOMEOWNERSHIP How Far Can We Go?

Using data from the 1998 Survey of Consumer Finances (SCF), this paper estimates homeownership rates that would prevail in the United States if borrowing constraints were eliminated. These estimates are obtained by first identifying a group of unconstrained households in the SCF based on a unique set of survey questions. Housing tenure preferences are then estimated over unconstrained households controlling for sample selection effects, and results are used to forecast owneroccupancy rates that would prevail in the absence of borrowing constraints. If all borrowing constraints were removed, ceteris paribus, the owner-occupancy rate among nonfarm families in the United States would increase just over 4 percentage points. In contrast, borrowing constraints have a comparatively small effect on the number of current renters who expect to own a home in the next ten years. This pattern suggests that borrowing constraints depress owner-occupancy rates in part by delaying homeownership rather than by permanently excluding families from owner-occupied housing. Additional analysis confirms that household social and financial stability, ability to care for a property, and wealth are all very important determinants of whether families prefer to own. On balance, therefore, evidence suggests that there is room for further relaxation of borrowing constraints to expand access to homeownership. However, such efforts are likely to be more fruitful if attention is given to mortgage products that alleviate borrowing constraints in the early years of the mortgage and if mortgage product innovation is coupled with policies designed to enhance the social and financial stability of families. I . I N T R O D U C T I O N Few symbols of personal economic success loom larger in the minds of Americans than owning one’s own home. Norms favoring homeownership have been further buttressed by the belief that because homeownership is a site-specific investment homeowners take better care of their neighborhoods and, therefore, make good citizens (e.g., Rohe, McCarthy, and Van Zandt 2000; DiPasquale and Glaeser 1999). Although evidence on whether homeowners make better neighbors is still tentative, it is certain that as a society we value homeownership for both personal and social reasons. This is clear from a long history of tax policies that encourage homeownership by reducing the cost of housing for owner-occupiers (e.g., Rosen 1979, 1985). In addition, the last decade has witnessed a series of government and industry efforts designed to reduce mortgage borrowing constraints for low-income families and other disadvantaged groups. These efforts have focused primarily on relaxing wealth requirements and have culminated in the recent introduction of zero and near-zero down payment mortgages for eligible borrowers. E L I M I N A T I N G C R E D I T B A R R I E R S T O I N C R E A S E H O M E O W N E R S H I P 2 Innovations in affordable mortgage lending—along with the dramatic economic expansion of the 1990s—have helped raise U.S. homeownership rates to historic levels, from 64 percent in 1989 to just over 67 percent by 2000. Against that backdrop, an important question for both government and the financial industry is as follows: To what extent can further relaxation of borrowing constraints boost homeownership rates, both for the U.S. population overall and for families of different age, race, ethnicity, and financial status? This paper seeks to answer that question by estimating homeownership rates that would prevail for various subgroups of the population if all borrowing constraints were eliminated but households otherwise abide by their budget constraints. Addressing these issues is difficult because of three fundamental problems that all studies on the impact of borrowing constraints face. Researchers must identify which families are credit constrained; they must evaluate how those families would behave if borrowing constraints were relaxed, ceteris paribus; and they must control for the myriad possible constraints that lenders impose on prospective borrowers. In the context of homeownership, such constraints include down payment standards, as well as house payment-to-income and total debt payment-to-income ratios, and the various ways these standards are applied for different types of loans (e.g., fixed versus variable rate mortgages). In addition, to the extent that nonmortgage debt such as auto and consumer loans can be used as substitutes for mortgage debt, borrowing constraints outside of the mortgage market potentially affect housing tenure decisions as well. The solution to these problems offered here is to apply sample selection methods to unusually rich data from the 1998 Survey of Consumer Finances (SCF). At the core of the estimation strategy is a unique set of survey questions that enable one to identify a priori a group of households that almost certainly hold as much mortgage debt as they would like given prevailing market rates and the menu of existing mortgage products. These families are characterized as not credit constrained (unconstrained) in the analysis to come, while all other households are characterized as possibly credit constrained for reasons that will become apparent later in the paper. Housing tenure preferences—to own or to rent—are then estimated over just the unconstrained families, controlling for the endogenous selection of families into the unconstrained group through a three-celled bivariate probit model. The resulting housing tenure coefficients reflect the impact of household demographic and financial characteristics on preferences for owning when families are subject to their budget constraints but not subject to binding borrowing constraints. Those coefficients are used to predict homeownership rates that would prevail for the entire population—unconstrained and possibly creditconstrained families. Comparing predicted with actual homeownership rates permits one to evaluate how much higher homeownership rates would likely rise if all borrowing constraints were eliminated, ceteris paribus. Is there opportunity for industry and government to expand homeownership through further relaxation of borrowing constraints? Results from the analysis suggest a qualified