Senator Warren Gets Taken In by a False Analysis
Monday, March 17, 2014
Last week, in a speech to the National Community Reinvestment Coalition, Senator Elizabeth Warren noted that an academic paper published by the St. Louis Federal Reserve was “proof” that the affordable housing goals did not contribute to the subprime mortgage boom. We have seen this paper before and advised the authors that it was based on a faulty understanding of how the housing market worked during the period they covered. However, the paper was not modified in any substantial respect, and Senator Warren — who was no doubt delighted to find an academic paper that supported her position — has been taken in.
The title of the paper is “Did Affordable Housing Legislation Contribute to the Subprime Securities Boom?” researched and written by Andra C. Ghent, Ruben Hernandez-Murillo, and Michael T. Owyang. As we will show, the affordable housing goals created substantial demand for subprime loans, a high percentage of which — as the paper notes — met the technical requirements of the goals. For this reason, Fannie and Freddie, the two GSEs were, by far, the largest buyers of securities backed by such loans. Indeed, as shown in the chart below, the GSEs bought more than 40 percent of all these securities issued in 2004, a dramatic increase from their purchase share only three years earlier. To achieve this rise in market share, the expansion of GSE purchases in 2002, 2003, and 2004 accounted for more than half of the growth in the market for private mortgage-backed securities (PMBS) backed by subprime mortgages.
Sources: Overall PMBS market data are from Inside Mortgage Finance. Purchase data are from OFHEO’s “Mortgage Markets and the Enterprises” annual reports. Actual for 2002-2005, estimated for 2001 based on subsequent trend.
As Countrywide and other aggregators realized that they could sell PMBS directly to the market, bypassing the GSEs, the GSEs’ incremental contribution to the PMBS market’s growth declined. Nevertheless, the GSEs continued to be the largest individual purchasers of PMBS until the entire market collapsed in 2007. Accordingly, the answer to the question in the title of the St. Louis Fed paper can only be a resounding yes; the GSEs’ purchases gave the market its start and then sustained it to the end.
The flaws in the St. Louis Fed paper
As noted above, soon after we saw this paper we advised Andra Ghent, as the lead author, that she and her colleagues did not understand the process by which mortgages were originated and made their way through the various levels of the housing finance system to the GSEs. Nevertheless, subsequent drafts of the initial paper have not taken account of the points we made. Now that the paper has become a news item, cited by a senator in a major public speech, we will restate these points, responding to the latest draft of the paper we have seen, dated October 29, 2013.1 We will describe why the theory in the paper is wrong — largely because the authors do not understand how the mortgage market worked during the years covered in their analysis.
The paper’s theory is that if the affordable housing goals stimulated subprime mortgage lending, there would be a statistically significant difference between (i) the number of mortgages originated in census tracts where mortgages were eligible for affordable housing credit and (ii) the number of mortgages originated in census tracts where mortgages were generally not goals-eligible. In the theory advanced in the paper, more mortgages should be made in those tracts where the mortgages were goals-eligible. “An increase in the number of originations,” the paper states, “would suggest that the lenders made a conscious attempt to make loans to borrowers in the target group, which could have led to the subprime securities boom” [emphasis added]. Since the authors found no difference, they concluded that the affordable housing goals were not influential in encouraging subprime lending.
The GSEs continued to be the largest individual purchasers of PMBS until the entire market collapsed in 2007.
The reference to lenders in the authors’ statement above points to a flaw that ultimately discredits their theory. Lenders or other mortgage originators had virtually no role in deciding whether a mortgage met the complicated income and location standards that qualified these loans for the affordable housing goals. What the lenders knew was that there was a new and greater demand for subprime loans — that traditional underwriting standards had been relaxed, and now they could sell to the GSEs directly, or to the large aggregators, mortgages for which there had previously been only limited demand.
The location of a borrower’s home within a particular metropolitan statistical area, or the borrower’s income as a percentage of the median income in that location, were not relevant to the originating lender, who was concerned solely with finding subprime borrowers. The aggregators’ function was to buy, on a wholesale basis, the mortgages that were made by lenders and other originators at the local level; the aggregators would then create separate mortgage pools consisting largely or entirely of subprime mortgages, issuing PMBS based on these pools. For the pools to be sold to Fannie and Freddie, the aggregators — principally the large subprime lenders like Countrywide, but also the large commercial and investment banks — were assisted by the GSEs themselves, which furnished them with detailed descriptions of the loans they needed in order to meet the goals. Eventually, with the guidance of the GSEs, the aggregators were able to create pools in which 100 percent of the subprime mortgages were goals-eligible. In a letter to Fannie Mae, dated September 30, 2005, responding to a Fannie Mae request, HUD described exactly how the GSEs could obtain goals credit for mortgages in PMBS pools where they have not purchased an entire pool.
In connection with the sale of PMBS to the GSEs, the aggregators would provide the GSEs with “tapes” that contained loan-level data on every mortgage in the pool. Using this data, the GSEs could establish exactly how many mortgages in each pool met the affordable housing goals. These were then reported to HUD, along with the other goals-eligible mortgages that they had also acquired during the year.
At a recent conference, Ghent averred that the GSEs did not have loan-level data on the subprime mortgages they acquired, and thus could not have been acquiring PMBS in order to meet the goals. This again is incorrect. The linked letter is proof that the GSEs had this data. Otherwise, there would have been no point in requesting the ruling from HUD.
Lenders or other mortgage originators had virtually no role in deciding whether a mortgage met the complicated income and location standards that qualified these loans for the affordable housing goals.
Further, here is an analysis prepared by Fannie Mae that shows the contribution of subprime, Alt-A, and, significantly, PMBS toward the affordable housing goals over several years. This table was provided by Fannie Mae to the Financial Crisis Inquiry Commission (FCIC) in response to an FCIC request, but was not used or referred to in the FCIC’s majority report.2 The term PLS used in the table refers to Private Label Securities, which are what we refer to as PMBS and the Ghent paper refers to as PLMBS. The bolded numbers show where the particular acquisition — a subprime loan, an Alt-A loan, or subprime PMBS — met one or more of the goals. The analysis shows beyond question that the GSEs had access to the loan level data in the PMBS they acquired from various mortgage aggregators, allowing them to claim goals credit.
Moreover, in the paper itself, the authors cite Part 81.16 of Title 24 of the Code of Federal Regulations, which specifically authorizes the GSEs to count their purchase of Real Estate Mortgage Investment Conduits (REMICs) as eligible for goals credit:
A GSE’s purchase or guarantee of all or a portion of a REMIC shall be treated as a mortgage purchase and receive credit toward the achievement of the housing goals, provided:…(B) The GSE has the information necessary to support counting the dwelling units financed by the REMIC, or that part of the REMIC purchased or guaranteed by the GSE, toward the achievement of the particular housing goal. 3[emphasis added]
This regulation would have been unnecessary if the GSEs did not have access to the “information necessary to support counting the dwelling units financed by the REMIC.”
Thus, the GSEs actively managed their acquisition of subprime loans that met the affordable housing goals. Lenders at the originator level needed to know nothing of these goals. What they knew was that there was a market for mortgages where the borrowers could not meet traditional mortgage standards and were therefore charged somewhat higher interest costs. Since the technical requirements for goals credit were unknown to and unnecessary for lenders, they made loans wherever they could find them; under these circumstances, there would be no statistically significant difference between the number of mortgages originated in different census tracts.
For this reason, Senator Warren was misled, and misled her audience.
Peter J. Wallison is a senior fellow and Edward J. Pinto is a resident fellow at the American Enterprise Institute.
FURTHER READING: Wallison and Pinto also coauthor “How the Government Is Creating Another Housing Bubble.” Alex J. Pollock contributes “The Housing Bubble and the Limits of Human Knowledge” and “What to Do With Fannie and Freddie.” Arnold Kling writes “Your Mortgage, Their Rent.”
Image by Dianna Ingram / Bergman Group
1. The paper discusses both the Community Reinvestment Act (CRA) and the affordable housing goals, but uses different methods to draw its conclusions for each. This memorandum, however, addresses only the affordable housing goals and the methodology used in the paper to argue that the affordable housing goals did not affect the GSEs’ acquisition of subprime mortgages.
2. The data for PLS in 2004 were missing in the original document.
3. The information necessary was occupancy status (primary owner, second home or rental); loan purpose (purchase or refinance); street address (including city, state, zip); income; and loan amount in dollars. This allowed the GSEs to determine where the borrower was located, the borrower’s income as a percent of the metropolitan statistical area median, whether it was a purchase or a refinance, and whether the home was a single family or owner-occupied multi-family rental.