In 2004, Housing Policy Debate brought together a diverse set of articles examining the growing subprime mortgage market grouped under the theme “Market Failures and Predatory Lending.”Footnote1 At the time, a scholarly and policy debate concerning the benefits and costs of subprime mortgage lending was in full swing. Authors considered questions such as the efficiency and equity of subprime mortgage pricing and terms, geographic and racial disparities in subprime mortgage lending, and the linkages between capital markets and inner-city home finance. One recurring theme was the paucity of publicly available mortgage market data, and the shortcomings and limitations of the primary public reports mandated by the Home Mortgage Disclosure Act (HMDA).
Rapid and dramatic changes in the US mortgage market preceded and precipitated the global financial crisis of 2008. Researchers faced continuing challenges in analyzing these market changes and the regulatory responses. Information about the kinds of loan terms offered, the pricing of mortgage loans, and especially the proliferation of fees, prepayment penalties, adjustable-rate provisions and other loan terms and features remained difficult to obtain. This special issue brings together six articles that take advantage of a variety of new data sources to provide new perspectives on pricing, regulation and sustainability of home mortgage financing.
The Federal government's role in gathering data on the mortgage market has been largely limited to monitoring race discrimination in approvals and denials. The Home Mortgage Disclosure Act of 1975 provides a valuable, yet limited data tool for research on the mortgage market. HMDA records are compiled annually from virtually all mortgage lenders and include the applicant's race and gender, the lender's decision to approve or deny a mortgage, and the home's geographic location. Prior to 2004, HMDA included no price information whatsoever. After 2004, HMDA data included limited price information (the difference between the annual percentage rate and a market index rate only), and only for a small segment of the market at the high end of the pricing scale. HMDA data have been useful in looking at race and gender disparities in mortgage approvals and, since 2004, in the segregation of some borrowers into the higher-priced subprime market segment, but could not provide detailed information about loan characteristics and pricing components such as lender and broker fees.
More detailed data are gathered in private proprietary data sets, notably McDash/LPS and LoanPerformance, but those data are very expensive for academic researchers, and because of their proprietary nature, restrictions are imposed on the types of analyses that may be performed. The gaps in available information on mortgage lending and home financing meant important policy questions were difficult to answer empirically. For example, did mortgage brokers facilitate the functioning of the market by making lower-cost loans available to subprime borrowers, or did they use their market knowledge to steer borrowers into higher-cost and less advantageous loans? Does the annual percentage rate fairly reflect the true cost of a mortgage loan, given the proliferation of loan fees, penalties and other pricing elements? Are advance pricing disclosures required by federal law achieving their goal of facilitating comparison-shopping, or are they used to bait and switch mortgage applicants?
The National Mortgage Data Repository project was launched by the National Consumer Law Center, working with several scholars and with the support of the Ford Foundation, to close this data gap. Mortgage loan records were coded directly from copies of original loan documents, including loan applications, estimated disclosures, and final closing documents. The availability of complete loan documents permitted the inclusion of unprecedented detail concerning the terms not only of final loans but also of original loan applications and interim lender cost estimates. The level of detail also includes each and every individual fee itemized on “good faith estimates” and settlement statements, as well as information about prepayment penalties and other loan characteristics. Borrower demographics including age, race, gender and geography are included as well. Four of the articles in this issue exploit the NMDR's depth of detail to explore important mortgage market policy questions.
The NMDR, regrettably, suffers from several limitations. First, the number of usable records (700 observations) was small. Second, the sample is one of convenience, not one selected as representative of any particular loan population. Records were obtained from state regulators, enforcement agencies, nonprofit loan counselors and public-interest attorneys. The loan files covered a long time window, from 1994 to 2008, and given their source, are loans most likely to involve claims of fraud and other borrower disputes. Massachusetts loans are overrepresented. Loan amounts and incomes of borrowers are smaller than national averages, and minority borrowers are overrepresented. On the other hand, the availability of detailed loan and borrower information about the troubled segment of the subprime market provides unique research opportunities, as illustrated in the first four articles of this special issue.
Elizabeth Renuart and Michael Collins exploit the NMDR's comprehensive detail of mortgage closing fees and costs to evaluate the effectiveness of mortgage cost disclosures mandated by the 1969 Truth in Lending Act (TILA) and the 1974 Real Estate Settlement Procedures Act (RESPA). TILA and RESPA disclosures are currently under review by the new Consumer Financial Protection Bureau. The avowed purpose of RESPA was to reduce settlement costs in home financing transactions, and to eliminate unnecessary fees, through mandatory disclosure. Comparing the fees and costs on NMDR loans to closing costs reported in earlier studies, Renuart finds that both the amount of fees and costs as a percentage of the mortgage loan, and the number of distinct fees and charges, have increased significantly since RESPA was enacted. While this finding is tempered by the fact that NMDR oversamples high-cost subprime loans, it nevertheless undermines the Congressional intent that mere disclosure of loan fees and costs would bring competitive pressures to bear either on their amount or the complexity. Renuart also finds that early estimated disclosures are not particularly accurate when compared to final closed transactions, and that fees are often higher at closing, supporting the claim that early disclosures are sometimes used to bait and switch borrowers. Finally, she concludes that future cost disclosure regulation cannot expect to achieve its goals without adequate government data collection on actual market practices, and, in particular, collection of the type of detailed mortgage cost and fee data that the NMDR contains.
In contrast, Michael Collins finds that the early estimated mortgage cost disclosures are reasonably accurate when compared to the final closing documents, and on average over-estimate total costs, although 39 percent of borrowers did pay closing costs that exceeded the early estimate. He also finds that broker fees are systematically under-estimated in early disclosures. Using the estimated disclosure amounts together with a set of loan and borrower variables in a regression model, he finds that the final costs are not significantly driven by borrower demographics, with the possible exception of age (older borrowers are more likely to see cost increases at closing.)
Elvin K. Wyly engagingly revisits the racial disparities in the subprime mortgage market through the lens of critical race and critical feminist theory. Combining the breadth of data available from HMDA reports with the depth of information offered by the NMDR, he identifies the distinct and cumulative impacts of race, gender and age on the mortgage borrowing experience of elderly African-American women. His approach resists the dominant method in “fair lending” research, which calls for isolating race from the long list of endogenous borrower and market characteristics, so as to isolate possible bias of mortgage underwriters from the racial and gendered impact of discrimination based on home values, wealth, employment, and education.
In a series of regression models using the NMDR data, Wyly finds that (1) age intensifies race and gender inequalities in the subprime market, i.e. elderly subprime borrowers are more likely to be female and African-American; (2) Black women pay higher interest and higher fees among subprime borrowers; and (3) elderly Black women are more likely to experience lending processes characterized as predatory, including bait-and-switch tactics and high total loan fees. The race and particularly the gender and age aspects of the subprime crisis and their interaction, he concludes, are the overlooked narrative of the global financial crisis.
Sonia Garrison and Wei Li probe the NMDR data for evidence of “no-benefit” refinancing, i.e. mortgage loans whose economic benefits to the borrower are significantly outweighed by the economic costs, a practice that some state laws attempt to curtail. Contesting the traditional assumption that cash-out refinancing transactions must be wealth-maximizing for borrowers who choose them, the authors propose a straightforward net present value test and apply it to the loans in the NMDR. As their models show, evaluating the net benefit of a refinancing transaction used to pay off credit card or other high-rate debt involves many moving parts. Borrowers, and even sophisticated credit counselors, could find it difficult to accurately calculate the net economic benefit of a refinancing transaction.
Turning to the actual cash-out refinance transactions in the NMDR, Garrison and Li find that nearly half of the cash equity extracted by borrowers is used to pay transaction costs. Depending on the duration of the refinance loan, they find that 75 percent to 87 percent of cash-out refinancing loans resulted in negative net present value, with a median loss of more than $8,000. In other words, the combination of transaction costs and higher interest rates on prior mortgage debt considerably outweighed the benefit of using equity borrowing to pay high-rate credit card debt. They conclude with the suggestion that their net present value calculation could be made available in the form of on-line calculators for homeowners and counselors to evaluate cash-out refinancing preventively, as well as to enforce state law anti-flipping provisions.
Complementing the four articles on the subprime market represented in the NMDR, McKernan reports on the experience of mostly low-income and minority mortgage borrowers participating in Individual Development Account (IDA) savings programs to accumulate down payments. IDA programs offer associated pre-purchase counseling, financial education and support. These programs sometimes include lower-priced, fixed rate loan programs, or restrict the terms of mortgages that participants may obtain.
Using a constructed data set of 831 IDA participants, and a matched set of non-IDA home buyers drawn from HMDA and NeighborWorks loan performance data, the study finds that IDA participants were more likely to obtain FHA mortgages, and less likely to obtain higher-cost subprime mortgages. IDA participants were also significantly less likely to lose their home to foreclosure than mortgage borrowers with similar demographic characteristics generally.
Finally, Jonathan Spader examines the refinancing behavior of homeowners with Community Advantage Program (CAP) mortgages, which are low-cost, fixed-rate loans purchased by Self-Help, a North Carolina community development financial institution, through a special funding program enabling banks to offer mortgages they otherwise could not resell. The loans are targeted to first-time, low-income homebuyers. Most subsequent refinancing was done to obtain a lower, usually fixed, interest rate, but a significant minority of refinancing activity was motivated by equity extraction. The majority of cash-out refinancing was also into fixed-rate products with improved rates, but a significant minority of borrowers refinanced into higher-cost, adjustable-rate subprime loans. Those who did extract equity usually sought to pay off credit card debt or finance home improvements. Although Spader is cautious in interpreting the results, it appears that CAP mortgage borrowers were generally prudent in their refinancing behavior (especially compared with those described by Garrison and Li in the NMDR) and may have been less susceptible to the push marketing of the worst subprime lenders than low-income homeowners generally.
Although the NMDR, as well as the specialized mortgage data sets constructed from IDA and CAP programs by McKernan and Spader, are by design not representative of the entire mortgage market, each of these articles offers important insights into the complex and varied borrowing experiences of low- and moderate-income and minority homeowners in different market segments. We can see at the same time the problematic pricing and disclosure practices in the troubled subprime sector represented by the NMDR, and the potential for better outcomes for similar borrowers participating in well-designed financial support programs exemplified by the IDA and CAP programs. Narratives of the subprime crisis too often simplistically blame the foreclosure tsunami on the very idea of extending mortgage credit to minority and low-income homebuyers. The reality is that mortgage product design and choice can have a huge impact on the homeownership experience. Hopefully this research can help policy makers, lenders and funders in the continuing effort to steer the housing finance system towards the goal of sustainable homeownership.
Acknowledgments
Thanks are due to many individuals and institutions who helped to make this special issue possible. In particular, Jessica Hanff deserves recognition for her patient editorial work. Thanks also to Patricia McCoy, Elizabeth Renuart, Jen Douglas, and Stephen Ross and their colleagues for their efforts in designing and creating the National Mortgage Data Repository, reviewing the numerous proposals submitted in response to a call for papers, and helping to organize the March 2010 research conference at Valparaiso that led to this special issue. The Ford Foundation, the National Consumer Law Center, and Valparaiso University provided essential support for the database and the conference. Finally, I wish to express my personal appreciation for the hard work of the authors and the anonymous reviewers in seeing these projects through to successful completion.
Notes
1 Housing Policy Debate 15, no. 3 (2004).