Housing

• "The Subprime Crisis: How Much Did Lender Regulation Matter?" by Robert B. Avery and Kenneth P. Brevoort. August 2010. SSRN #1726192.
• "Complex Mortgages," by Gene Amromin, Jennifer Huang, Clemens Sialm, and Edward Zhong. November 2010. SSRN #1714605.
• "Endogenous Gentrification and Housing-Price Dynamics," by Veronica Guerrieri, Daniel Hartley, and Erik Hurst. July 2010. SSRN #1657176.
• "What Are the Social Benefits of Homeownership? Experimental Evidence for Low-Income Households," by Gary V. Engelhardt, Michael D. Eriksen, William G. Gale, and Gregory B. Mills. February 2011. SSRN #1752381.
• "The Inefficiency of Refinancing: Why Prepayment Penalties Are Good for Risky Borrowers," by Christopher J. Mayer, Tomasz Piskorski, and Alexei Tchistyi. December 2010. NBER #16586.

Ever since the collapse of the housing bubble and the start of the ensuing recession, economic researchers have been at work trying to determine what caused the bubble, why it collapsed, and how the chaos in real estate was transmitted to the broader financial market and economy. This review looks at some recent papers in that literature.

What is to blame? | Why did the availability of credit for housing increase in the 2000s and what effect did that increase have on housing price appreciation and subsequent decline? Many market-oriented analysts argue that the federal Community Reinvestment Act and the affordable housing investment goals for the government-sponsored enterprises Fannie Mae and Freddie Mac recklessly increased the availability of credit for housing. In a recent paper, Robert Avery and Kenneth Brevoort, economists with the Federal Reserve, test whether the regulations that encouraged more lending to lower-income people and poor areas led to worse loan performance. If the regulations induced lending that would not otherwise have taken place, presumably loan defaults would be higher in areas with more CRA-regulated institutions or more people that fulfill GSE lending goals.

Their research design exploits differences in the geographic density of institutions governed by CRA regulations (commercial banks and thrifts) relative to those not governed by them (mortgage companies and credit unions). CRA-regulated institutions get credit for loans made to homebuyers living in low- and moderate-income census tracts within the counties of their branch locations. Low- and moderate-income census tracts are defined as having less than 80 percent of the median income of the broader area in which the bank is located. GSE regulations define low- and moderate-income areas as census tracts with less than 90 percent of median income of the area in which the bank is located.

The authors test whether census tracts with greater density of CRA-regulated lending suffered worse loan performance in the current financial crisis. They also examine loan performance in census tracts that just qualified for CRA (just below 80 percent of median income) or GSE (just below 90 percent of median income) attention relative to loan performance in census tracts just above 80 or 90 percent of the median income. The assumption in such a research design is that nothing of substantive importance varies with the presence or absence of legal qualification for extra loan attention, and thus the comparison of outcomes between such census tracks is a good approximation of an experiment.

Surprisingly, tracts with greater CRA lender density had lower 2008 loan delinquency rates. However, in regressions estimated only on states that experienced very large growth in real estate loans during 2000–2007 (the so-called "sand states" of Arizona, California, Florida, and Nevada), those tracts with greater CRA lender density had higher 2008 loan delinquency — but the delinquency rate was even higher outside the CRA-relevant assessment area. That is, loans issued to people outside the county in which the bank was located, which do not count toward CRA ratings, performed worse than those that did count toward the CRA rating. Loans generally did not do well in the sand states, but the variation within those states is not consistent with CRA compliance pressure (because CRA-compliant areas had better, not worse, performance).

Complex mortgage users were not only higher-income but also had higher ratios of house value to income and they purchased houses 20 percent more expensive.

The authors also estimate regressions for loan quality (debt-to-income ratio and no-income-documentation loans) in 2004–06, the peak of the housing boom. They find that loan quality was higher, not lower, in census tracts with greater CRA lender density.

In the discontinuity tests, the authors find no evidence of any threshold effects. Loan quality and performance increase with census tract income, but this is true of all loans and not just those favored by the regulations. There are no discontinuous worse loan results in census tracts that just qualify as CRA- or GSE-compliant relative to tracts that do not.

Another paper, by Gene Amromin et al., also casts doubt on the role of affordable housing mandates in the recent housing bubble. The paper analyzes data on what the authors term "complex mortgages," which they define as being interest-only, have negative amortization, or have teaser rates. Two hypotheses exist about the use of such products: The first is that they are a rational capital market response for those consumers whose incomes and house values are expected to rise. Such products are also rational in environments where incomes and home prices are volatile because they allow homeownership with a default option if incomes or prices decline. The second hypothesis is simply that unsophisticated consumers were duped by lenders into using these complex products.

The researchers had access to data on 10 million mortgages from nine out of the 10 largest mortgage servicers in the country over the period 2003–2007. In the frequency distribution of complex mortgages by metropolitan statistical areas, the top 25 percent of cities that used these mortgages experienced the greatest appreciation and depreciation in the recent housing cycle. Complex loans were used by higher-income people ($141,000 average income, versus $88,000 for fixed-rate mortgage holders and $101,000 for adjustable-rate mortgage consumers) with high credit scores (only 7 percent below 620) in nonrecourse-loan states (where foreclosure results only in the loss of the home and not other assets) in MSAs with previously higher home price appreciation and higher population growth. Complex mortgage users were not only higher-income but also had higher ratios of house value to income and they purchased houses 20 percent more expensive at every income level. Put simply, complex mortgage users had the same credit scores as those with fixed-rate mortgages, but purchased much more expensive houses. For example, complex mortgage use exceeded 40 percent in multiple counties in California, Nevada, Colorado, and Florida, but only 5 percent in Albany, NY.

In short, complex mortgages were used by the affluent to stretch their budgets to afford more expensive homes. Complex mortgage use was higher in areas with high population growth and no price decline in the last 10 years, and less in low-income areas.

Why not more housing? | Why did increased flows of credit into housing in the 2000s result in rising home prices rather than more housing supply with little or no price appreciation? A common explanation is that supply was constrained by natural barriers and/or land-use regulation (see "Zoning's Steep Price," Fall 2002). But a new paper by Veronica Guerrieri et al., economists at the Federal Reserve Bank of Cleveland, argues that housing prices can increase in a city even though housing supply for the city in the aggregate is completely elastic.

The mechanism that causes this result is gentrification. That is, an increase in the income of some city residents (or a decrease in the price of credit that increases the ability to bid for housing) increases the price of housing nearest the housing occupied by the most affluent city residents. People bid up the price of this housing so as to experience the positive spillovers created by the affluent (better services, shopping, and other neighborhood amenities). The data show that the greatest price appreciation from a positive income shock occurs in the neighborhoods nearest an existing affluent neighborhood, rather than in the most affluent neighborhood itself.

This price appreciation can occur in the context of an overall-elastic city housing supply. The authors find that even after controlling for housing supply elasticities, the elasticity of housing prices to income changes was 0.96 from 2000 to 2006.

These results do not undermine the argument to reduce or eliminate regulatory barriers to housing supply. But housing supply deregulation will not eliminate housing price increases during economic booms.

Importance of homeownership? | Subsidies for homeownership are often justified by economists because of the positive externalities that allegedly result: local physical amenities and higher civic involvement. Simple regressions support this hypothesis, but those results are suspect because homeowners differ from renters in ways not observable to the researcher. Thus, homeownership itself may not be causal; it may be simply a surrogate for characteristics of homeowners not specified in the regressions that make them more likely to be civic minded. Researchers have attempted to solve this problem through the use of instrumental variables that are correlated with homeownership but not the unobserved characteristics of homeowners, but scholars are not convinced that these measures do the job.

A recent paper by Gary Engelhardt et al. tackles the issue through random assignment. Low-income renters in Tulsa, OK were recruited and randomly assigned to a treatment or control group. Those in the treatment group were eligible for a savings program in which up to $750 in annual savings would be matched on a two-for-one basis when used as a down payment for home purchase. The control group was not eligible for the savings subsidy. After four years, homeownership rates were 9.2 percent higher in the treatment group. But neighborhood volunteering was actually significantly less in the treatment group and other measures of civic involvement showed no difference between the two groups. Those in the treatment group were more likely to spend money on housing maintenance, but only for the interior and not exterior of their homes, and thus there were no positive spillovers. Justifications of homeownership subsidies based on positive externalities would appear to be invalid.

Prepayment | One of the characteristics of nontraditional mortgages that upsets consumer groups is prepayment penalties — an extra charge for paying off the balance of a mortgage before the end of the loan. The Federal Reserve issued regulations during 2008 to restrict prepayment penalties on adjustable-rate loans, and the Dodd-Frank financial reform bill included similar provisions for adjustable- and high-interest-rate loans.

In a new paper, Christopher Mayer from Columbia Business School and two colleagues argue that prepayment penalties prevent adverse selection among the riskiest borrowers. The prepayment penalty binds a pool of borrowers to a lender. Without the prepayment penalty, those who have good ex post outcomes will refinance out of the pool, leaving only the borrowers more likely to default. In the absence of the prepayment penalty, lenders have to charge higher rates that, in turn, make default more likely among those borrowers who experience negative shocks and exclude some borrowers from the market in the first place.