Stuart Gabriel, Ryan Vaughn and I have a new working paper where we document some interesting correlations concerning what types of borrowers during the 2003 to 2006 years received “good deals” from a leading subprime bank. A borrower’s Congressional Representative’s attributes matter. Borrowers who live in the Congressional Leadership’s districts received “sweeter deals”. Below the fold, I reproduce the paper’s introduction.
Introduction
Implosion in housing markets figured prominently in the 2007 meltdown in capital markets and the downturn in the global economy.[1] Neither analysts on Wall St. nor regulators in Washington, D.C. anticipated the depth of the crisis, its geographic and asset class contagion, or its adverse effects on household balance sheets. Emblematic to the crisis was the pervasive failure of subprime mortgages. Those loans provided substantially eased credit qualification and homeownership opportunity to low credit-quality borrowers. As shown in Demyanyk and Van Hemert (2007), the quality of originated subprime mortgages began to deteriorate even prior to the housing bust. By 2008, in the wake of downward spiral in house prices, a full 45 percent of subprime borrowers were underwater. Two years later, a similar share of outstanding subprime mortgages were in default.
Early on and prior to the deterioration in subprime loan performance, lenders appeared to understand the controversial nature of their product and the related importance of Congressional support. To that end, lenders may have sought to direct campaign contributions to elected Representatives to generate support for subprime loan products and to assist in easing regulatory oversight. As documented by Mian, Sufi, and Trebbi (forthcoming), lenders became politically sophisticated in making campaign contributions to elected Representatives in the years leading up to the 2000s crisis.
A second potentially complementary strategy for capturing political support was for lenders to offer more credit and at better terms to borrowers in Districts represented by targeted Congressional Representatives. As further suggested by Mian, Sufi, and Trebbi (forthcoming), such strategic interactions and related side-payments could be important in a world in which explicit “quid pro quo” was not politically feasible. To the extent that interests aligned, both direct campaign contributions as well as District-level direction and pricing of mortgage credit could serve the political economic interests of both the lender and the elected official. District-level expansion of mortgage and housing opportunity could be viewed as a political return to Representatives in exchange for expanded subprime lending opportunities.
This study uses the universe of first-time homebuyer residential home loans issued by a major subprime lender to study the role of Congressional political influence in the access to and pricing of subprime mortgage credit. It provides a political economy explanation for the geography of subprime lending. By merging several data sets, we seek to implement the following thought experiment. Consider two identical marginal borrowers called “A” and “B” who live in the same local labor market at the same point in time. Assume that the two borrowers live in comparable but different residential communities. If “A’s” Congressional Representative is liberal, is a member of the Finance Committee, is a leader of the House of Representatives, or if this Representative receives direct campaign contributions from the subprime lending institution, do these Congressional attributes influence the probability that the institution makes a loan to A versus B? Further, are these same attributes associated with higher or lower loan amounts or loan pricing to A versus B?
The loan level data in our study come from the servicing database of the now defunct New Century Financial Corporation (New Century) and the Home Mortgage Disclosure Act (HMDA). Like other studies, we control for the borrower, loan, or locational attributes that influence the allocation and risk-based pricing of mortgage credit. Unlike other studies, those controls are not the focus of this study. Instead, while controlling for a rich set of household attributes, zip code attributes and including state/year/month fixed effects, our goal is to assess how attributes of the local Congressional Representative influenced access to and pricing of subprime mortgage credit. The “politics” hypothesis posits that these attributes mattered because New Century had specific political goals in mind and used implicit subsidies to achieve those objectives.
The eased qualification requirements associated with subprime lending also may have been important to policymakers seeking to attain federally-mandated lending goals related to minority homeownership. Indeed, as evidenced in numerous recent studies (see, for example, Gabriel and Rosenthal (2005, 2011), racial minorities have been largely underrepresented in homeownership attainment. During the years leading up to and including the boom, substantial policy effort was directed at narrowing racial homeownership gaps.[2] Prominent among those efforts were ambitious quantitative goals for financial institution loan origination among minority and other households as embodied in the Community Reinvestment Act (CRA). Accordingly, below we test as well for whether minority borrowers received differential treatment by New Century. In particular, we test whether minority borrowers had a greater probability of receiving subprime loans as a function of their Congressional Representative’s attributes. Also, conditional on receipt of a subprime loan, we test whether minority borrowers received more favorable loan terms (as evidenced in loan size or loan pricing) as a function of their Representative’s attributes.
Assessment of New Century Financial Corporation loan level data from 2003 – 2006 reveals a new political geography of subprime lending. Our findings highlight that New Century was especially active in offering differential treatment to borrowers represented by the Democratic and Republican leadership of Congress. In the case of borrowers residing in the districts of the Speaker of the House and the Majority and Minority Leaders and whips, subprime lenders were less likely to reject loans; further, New Century offered lower mortgage interest rates and large loan amounts, all things equal, to residents of those areas. This fact is especially true for African American borrowers in these districts. Also, borrowers received rate discounts in districts where New Century donated to the local Congressional Representative’s election campaign.
This paper contributes to the literature investigating the causes and consequences of private industry’s political contributions to the Congress. Ansolabehere, De Figueiredo, and Snyder (2003) argue that political contributions appear to yield a very high return and this raises the question of why industry does not increase its contributions to the Congress. Bombardini and Trebbi (2011) document the role of the electoral strength of an interest group that could support a representative’s re-election campaign. Krozner and Stratmann (2005) study repeat contributions of PACs to representatives and document strategic interactions whereby representatives build a reputation for taking certain positions that help industry and are rewarded by special interests. Bronars and Lott (1997) and Stratmann (2002) both report evidence that changes in campaign contributions are correlated with changes in roll call voting by members of Congress. In earlier work, Stratmann (1992) presents evidence documenting how farming PAC contributions are targeted to specific representatives depending on who their constituents’ attributes. Mian, Sufi and Trebbi (2010) document the voting patterns by representatives on key pieces of banking legislation after the crisis of 2008 began. They document that representatives from areas where there were more subprime lending activity were more likely to vote for bailouts.
Unlike these earlier studies, using our detailed loan data we are able to investigate how a major firm directed its campaign contributions to specific Representatives as well as directed its loans to constituents of those same Representatives. Such representatives likely recognized that New Century was pleasing their constituents by allowing them increased access to capital at lower interest rates (see Mian and Sufi 2009). We document the geographic clusters of lending and pricing activity across congressional districts while relying on micro data that allows us to control for a rich set of borrower attributes. By enhancing credit access to underserved borrowers, the lending patterns we document were likely to have increased constituent support and helped to keep the incumbent in power.
The plan of the paper is as follows. The following section describes the data and sample. Section 3 discusses econometric strategy and results of analysis of HMDA and New Century microloan files, including assessment of mortgage origination and pricing. Section 4 provides concluding remarks.
[1] As reported in Shiller (http://www.econ.yale.edu/shiller/data.htm), US national house prices recorded a decline of 31 percent over the 2006 – 2010 period, about on par with the peak-to-trough contraction during the Great Depression.
[2] Both Presidents Clinton and G. W. Bush sought to address racial and ethnic gaps in homeownership. In a statement dated June 18, 2002, President Bush noted that “The goal is that everyone who wants to own a home has a shot at doing so. The problem is we have what we call a homeownership gap in America. Three-quarters of Anglos own their home, and yet less than 50 percent of blacks and Hispanics own homes. That problem signals that something may be wrong in the land of plenty. And we need to do something about it.” See White House News Release from that date.
[Cross-posted at The Reality-Based Community]
