|
Reforming the GSE Housing Goals
Jonathan Brown
Essential Information
AEI Conference on GSE Housing Goal Regulation
June 9, 2003
I. GSE Housing Goal Framework
The methodological framework developed by HUD for evaluating GSE housing
goal performance has three components:
A. the definition of the relevant mortgage market for evaluating GSE performance;
B. the definition of the underserved sectors of the mortgage market that are the focus of attention in evaluating GSE performance; and
C. the performance standard that the GSEs must satisfy.
The first component, the definition of the relevant market, raises the most complex issues and requires policy-makers to come to terms with major data availability constraints. In many respects, this is the most important component of the overall GSE housing goal framework.
HUD has defined the relevant market broadly to include all conventional home purchase and refinancing loans, with the exception of the B&C lending portion of the subprime market. We generally agree with HUD's approach, except for HUD's failure to recognize that a portion of the FHA-insured mortgage market could be underwritten on a conventional basis.
On a number of occasions, Fannie Mae and Freddie Mac have contended that HUD should limit the relevant market for GSE housing goal purposes to the types of conventional mortgage loans for which the GSEs have achieved a high level of market penetration or, using our terminology, high market shares. Such contentions were aggressively advanced by the GSEs in HUD's 2000 GSE housing goal rulemaking:
Both GSEs provided numerous comments concerning the types of mortgages that HUD should exclude from the definition of the single-family owner market, both when HUD is evaluating the GSEs' performance relative to the conventional conforming owner market (i.e., determining whether the GSEs' lead or lag the market for single-family- owner mortgages) and when HUD is calculating the overall market shares for each housing goal (as described in Appendix D). Fannie Mae stated that it ''can only purchase or securitize mortgages that primary market lenders are willing to sell'' and that certain types of products (such as ARMs) ''are particularly difficult to structure for sale to the secondary market''. Fannie Mae added that ''HUD fails to adjust for those housing markets that are not fully available to Fannie Mae and Freddie Mac''. Freddie Mac stated that it ''has not achieved, and is unlikely to achieve in the near term, the same penetration in the subprime and manufactured housing segments of the market as it has achieved in the conventional, conforming market'' and therefore HUD should not include these segments in its market definition. According to the GSEs, markets that are ''not available'' to them or where they are not a ''full participant'' should be excluded from HUD's market definition. In addition to the subprime and manufactured housing markets, examples of market segments mentioned by the GSEs for exclusion included: low-down payment mortgages (those with loan-to-value ratios greater than 80 percent) without private mortgage insurance or some other credit enhancement; loans financed through state and local housing finance agencies; below-market interest-rate mortgages; specialized CRA mortgages; and portions of depository portfolios that are not available at mortgage origination for purchase by the GSEs. 65 FR 65090 (October 31, 2000).
HUD vigorously rejected the argument that sectors of the conventional loan market that traditionally have not been extensively served by the GSEs, such as subprime loans and manufactured housing loans, should be excluded from the definition of the relevant market when evaluating GSE housing goal performance. The only major exception to this inclusive approach involves very high-risk subprime loans, which HUD has decided to exclude from the relevant market definition. According to HUD:
In general, HUD disagrees with the comments offered by the GSEs about excluding those market segments that they haven't yet been able to penetrate fully. Congress stated that HUD was to estimate the size of the conventional conforming mortgage market, not the market that the GSEs perceive as available for them to purchase. However, with respect to the subprime market, HUD believes that the risky, B&C portion of that market should be excluded from the market definition for each of the housing goals. Thus, HUD includes only the A-minus portion of the subprime market in its overall estimates of the goals qualifying market shares.
Excluding other important segments of the lower-income mortgage market, as the GSEs recommend, would render the resulting market benchmark useless for evaluating the GSEs' performance. The loans that the GSEs would exclude are important sources of lower-income credit and, in fact, are among the very loans the GSEs are supposed to be funding. A recent report by the Department of Treasury demonstrated the targeting of CRA-type loans to lower-income and minority families. Numerous studies have shown that the manufactured home sector is an important source of low-income housing. In many of these markets, a more active secondary market would encourage lending to traditionally underserved borrowers. While HUD recognizes that some segments of the market may be more challenging for the GSEs than others, the data reported in Tables A.7a and A.7b of this Appendix show that the GSEs have ample opportunities to purchase goals-qualifying mortgages. As market leaders, the GSEs should be looking for innovative ways to pursue this business, rather than suggesting that it is not available to the secondary market. 65 FR 65091 (October 31, 2000).
The very high-risk subprime loans which HUD excluded from the relevant market for GSE housing goal purposes are commonly referred to as B, C, and D loans or, using HUD's terminology, the B&C portion of the subprime market. In recent years, B, C, and D loans have comprised roughly 25%-30% of the overall subprime market. For example, Inside B&C Lending reports that during the first nine months of 2002 B, C, and D loans represented 26% of all subprime loans. We do not exclude B, C, and D subprime loans from the conventional mortgage market in our computation of GSE market shares because there is no publicly available data on the percentage of subprime loans that are originated by such lenders by borrower income or racial category or by census tract income or racial catergory or even by individual MSA.
Our inclusion of B&C lending as part of the relevant conventional market does to some extent overstate the size of the relevant conventional loan market as defined by HUD. However, there are other factors in our methodology for defining the size of the conventional loan market that work to cast the relevant market too narrowly and we believe that, for the most part, these factors offset the overstatement resulting from the inclusion of B&C lending. First and most important, a significant percentage of FHA-insured loans could have been underwritten as conventional prime loans and these loans should be viewed as a market that the GSEs can and do, in fact, penetrate. Second, in using HMDA data to determine the size of the conventional mortgage market for GSE housing goal purposes, we must exclude mortgage loans made on 2-4 unit properties that have loan amounts that are below the GSE loan ceilings for 2-4 unit properties, but higher than the GSE loan ceilings for 1-unit properties.
The traditional view has been that there is little overlap between the FHA-insured mortgage market and the conventional mortgage market because FHA-insured mortgage loans generally have loan-to-value (LTV) ratios that are so high that they could not have been underwritten as conventional loans that the GSEs would purchase. In recent years, however, the GSEs have adopted more flexible underwriting standards for home purchase loans and presently buy a considerable number of home purchase loans with high LTV ratios.
For example, the LTV ratios of home purchase loans to borrowers with incomes below the median family income for their local areas that were purchased by Fannie Mae in 2000 demonstrate this pattern. Among these Fannie Mae loan purchases, 20.2% of the loans had LTV ratios in the 90%-95% range and 7.3% had LTV ratios greater than 95%. In our opinion, a reasonable assumption as to the size of the overlap between FHA-insured loans and coventional loans is that roughly 15% to 20% of FHA-insured borrowers could qualify for prime conventional mortgage loans.
Any significant overlap between FHA-insured loans and conventional loans means that defining the relevant market for GSE housing goal purposes to be comprised only of conventional loans results in a definition of the relevant loan market that is excessively narrow. As HUD has recognized in other contexts, the relevant market should include all loans that the GSE could reasonably be expected to purchase and there are many FHA-insured loans that fall within this category.
The second component, the definition of underserved sectors, involves determining which borrower categories or loan categories are not adequately served by the secondary market for prime mortgage loans - i.e., the secondary market constituted by the GSEs. Since barriers in accessing the secondary market for prime loans typically mirror barriers in accessing the underlying origination market for prime loans, this second component can be viewed as identifying the underserved sectors of the mortgage loan origination market. Underserved sectors of the mortgage loan origination market are usually characterized by low origination rates for loan applications or the prevalence of
high-cost loans, such as subprime loans. As measured in the loan origination market and the GSE secondary market, the most underserved sectors of the 1-4 family mortgage market are clearly low income borrowers, African-American borrowers, Hispanic borrowers, and borrowers in low-income or minority neighborhoods.
The GSE housing goal regulations mandated by FHEFSSA and established by HUD define underserved sectors very broadly. In our view, there is a need for more narrowly defined categories within these broader definitions of underserved sectors. Unless this is done, the most underserved sectors will continue to remain underserved by the GSE secondary mortgage market.
The third component, the GSE performance standard, is based on the FHEFSSA provisions stating that the GSE have an affirmative obligation to support the financing needs of affordable housing and should exercise their ability to lead the market in making mortgage credit available for affordable housing purposes. HUD's GSE housing goal regulations have interpreted this statutory language to mean in general that Fannie Mae and Freddie Mac's market share of financing of underserved sectors should be equal to their market share of financing of the conventional mortgage loan market as a whole.
Our basic standard for evaluating GSE affordable housing goal performance is that a GSE's market share of the various underserved segments of the conventional loan market should, at minimum, be the same as the GSE's market share of the higher income and White borrower segments of the conventional mortgage loan market. At first glance, this may appear to be a very different methodology than that used by HUD in setting the housing goal targets, but closer inspection shows that our methodology is essentially the same as HUD's.
HUD sets the numeric housing goal target for each affordable housing goal category based on the percentage of total housing units financed by the conventional mortgage market that fall within that particular affordable housing goal category. For example, in the 2000 housing goal rulemaking, HUD estimated that housing units eligible to satisfy the Low-and Moderate Income Goal would comprise 50-55% of the housing units financed by the conventional mortgage market. HUD then set the Low-and Moderate Income Goal requirement for Fannie Mae and Freddie Mac to require that 50% of the units financed by their loan purchases must satisfy this requirement.
Similarly, HUD estimated that housing units eligible to satisfy the Geographically Targeted Goal requirement would comprise 29-32% of the housing units financed by the conventional mortgage market. Accordingly, HUD set the Geographically Targeted Goal requirement for Fannie Mae and Freddie Mac to require that 31% of the units financed by their loan purchases must satisfy this requirement.
By setting these two GSE housing goal targets within the numeric range estimated for each housing goal category's percentage of the entire market, HUD, in effect, required that the distribution of GSE loan purchases with respect to each housing goal category match that category's distribution within the underlying conventional loan market. Another way of saying that the GSE distribution should match the market distribution is to say that the GSE market share of loan originations (or units financed) that could qualify for a housing goal category should match the GSE market share of loan originations (or units financed) that could not qualify for the housing goal category.
We prefer our market share methodology because it provides a more effective way to measure and compare GSE performance in serving the various mortgage market subsectors within and outside each housing goal category. Such focus on subsectors of the broadly defined housing goal categories is vital in order to identify the more underserved subsectors of the mortgage market that are not receiving the full benefit of GSE housing goal regulation.
While HUD has accepted in principle the equalization of market shares as the basic standard for GSE performance, the Special Affordable Housing Goal requirement that it established in 2000 falls short of this mark. HUD estimated that housing units eligible to satisfy the Special Affordable Housing Goal requirement would comprise 23-26% of the housing units financed by the conventional mortgage market. However, HUD set the Special Affordable Housing Goal requirement for Fannie Mae and Freddie Mac to require that only 20% of the units financed by their loan purchases must satisfy the requirement. This requirement covers loans to low-income borrowers regardless of location and loans to low and moderate income borrowers in low and moderate income neighborhoods.
The only major difference between our methodology and HUD's approach is that HUD generally measures the size of the relevant mortgage market and its various subsectors and also GSE loan purchases in terms of the number of units financed, while we generally measure the relevant market and GSE performance in terms of the number of loan originations in the primary mortgage market and number of loans purchased by the GSEs. When focusing on GSE performance in serving the financing needs of 1-4 family housing, the distinction between measuring GSE performance in terms of number of loans rather than number of units financed is very limited. The vast majority of mortgage loans made on 1-4 family properties, both owner-occupied and rental, involve 1-unit properties. With 1-unit properties, the number of loans and the number of units financed are identical. With 2-4 units properties, the number of loans will be less than the number of units financed, which means that loans on 2-4 unit properties should be given extra weight when measuring GSE performance on a per loan basis.
II. Flawed Structure of the Current Housing Goal Regulations.
A. Overbroad Housing Goal Categories with No Subgoals.
The affordable housing needs categories for which HUD has established housing goal targets are definded so broadly that the GSEs have wide latitude in using above-average perfomance in serving the least disadvantaged sectors within a broad target category as a set-off to compenstate for poor performance in serving the most disadvantaged sectors. If GSE housing goal regulation is to become an effective means to direct or at least encourage the GSEs to adequately support the most seriously underserved segments of the mortgage market, the broad housing goal categories need to be supplemented with more specific goals or subgoals that focus on these underserved segments.
1. Low and Moderate Income Housing Goal.
The low and moderate income housing goal requirement covers all owner borrowers with incomes less than or equal to the local area median family income. This expansivly defined underserved sector encompasses borrowers that under commonly accepted definitions are classified as low income borrowers, moderate income borrowers, and lower-middle income borrowers. Yet, as shown in Table 1, the level of GSE support varies significantly between these three borrower categories, with low income borrowers receiving the lowest level of GSE support and lower-middle income borrowers receiving the highest level of support.
In the absence of more specific goals or subgoals, GSE performance in meeting the low and moderate income goal requirement is measured only in terms of average GSE performance for this broadly defined underserved sector as a whole. In essence, this regulatory framework gives the GSEs a license to trade off poor performance in serving low income borrowers for strong performance in serving lower-middle income borrowers. Invariably, trade-offs of this nature work against the most disadvanteged segments of a broadly-defined underserved sector. By contrast, if the housing goal that HUD establishes for the entire underserved sector were also adopted as a subgoal that applied to each of the three subsectors, this trade-off could not take place.
2. Geographically Targeted Housing Goal.
A similar trade-off process is at work in the implementation of the underserved
geographies goal requirement. The underserved geographies sector is broadly defined to include not only neighborhoods with high concentrations of minority persons and low income neighborhoods, but also the vast majority of integrated neighborhoods, all moderate income neighborhoods, and many lower-middle income neighborhoods. Specifically, GSE loan purchases within metropolitan areas qualify for the Geographically Targeted Goal if they are secured by properties located in census tracts in which (i) the median family income is less than or equal to 90% of the metropolitan area median family income or (ii) the minority population percentage is greater than or equal to 30% and the median family income is less than or equal to 120% of the metropolitan
area median family income.
As shown in Table 2, Table 3, and Table 4, there is wide variation in the extent to which the GSEs serve the diferent segments of the broadly defined Geographically Targeted Goal. These tables indicate that GSE performance in serving neighborhoods with high concentrations of minority persons, especially African-American neighborhoods, and low income neighborhoods is comparatively weak, while GSE performance in serving integrated neighborhoods and lower-middle income neighborhoods is comparatively strong. Table 4 is especially disturbing because it shows that the wide disparity in GSE market shares between white neighborhoods and neighborhoods with concentrated minority populations persists even after controlling for differences in borrower incomes.
Once again, the GSEs, in meeting the underserved geographies requirement, may engage in trade-offs that are adverse to effective GSE housing goal implementation. This can only be prevented if HUD establishes subgoals for the most underserved segments of the expansively defined underserved geographies sector.
The GSE Maps section of our GIS Website ( http://www.public-gis.org ) provides an extensive set of GSE market share maps that document GSE underperformance in minority and low income neighborhoods. Many of these maps were filed by local community groups in HUD's GSE housing goal rulemaking in 2000.
3. Combining 1-4 Family and Multifamily Loans.
In measuring GSE housing goal performance, the current regulatory scheme consolidates GSE purchases of 1-4 family loans and multifamily loans. Multifamily loans are far more likely than 1-4 family loans to qualify for one of the GSE housing goal requirements. For example, in 2000, only 23.83% of the 1-4 family mortgage loans purchased by the GSEs were made to low and moderate income borrowers (borrowers with incomes less than 80% of local area median family income). By contrast, in the same year, 81.86% of the rental units financed by GSE purchases of multifamily loans were affordable to low and moderate income renters.
Given this mismatch in the GSE housing goal eligibility of 1-4 family loans versus multifamily loans, consolidating 1-4 family loans and multifamily loans in measuring GSE performance enables the GSEs to use increased purchases of multifamily loans to offset weak performance in serving low income borrowers in the 1-4 family mortgage market. This trade-off dynamic is one of the primary reasons why the GSEs have been able to satisfy the GSE housing goal requirements even though they are underserving low income borrowers.
The GSE housing goal system should be restructured so that it has separate
housing goal requirements for 1-4 family loans and multifamily loans. These two loan
markets have little in common and their underserved segments cannot be properly measured on the same scale.
4. Loans on Small Multifamily Rental Properties.
The GSEs have woefully underserved the segment of the multifamily loan market comprised of loans on 5-50 unit rental properties. These small-sized multifamily properties represent a key component of the affordable rental housing stock and should be a priority of GSE housing goal regulation.
Our analysis of the 1991 Residential Finance Survey data indicates that 37.61% of the multifamily units financed in the 1989 to 1991 period were in 5-49 unit properties. Yet, in 2000, only 2.59% of the units financed by GSE multifamily loan purchases were in 5-49 unit properties. Applying HUD projections of the size of the multifamily mortgage market in 2000, we estimate that in 2000 the GSE market share of units financed in large multifamily properties (50 or more units) was 60.46%, compared to a GSE market share of units financed in small multifamily properties (5-49 units) of only 2.47. Clearly, a subgoal is vitally needed for GSE purchases of loans made on small multi-family properties.
B. Failure to Establish Goals or Subgoals for Minority Borrowers.
A major shortcoming of the GSE housing goal regulation has been its failure to focus on the performance of the GSEs in serving the housing finance needs of minority borrowers. Minority borrowers, especially African-American borrowers, encounter barriers in accessing reasonably priced mortgage credit irrespective of the racial composition of the neighborhood in which they are located. Prime-rate conventional mortgage loans are the predominant form of mortgage credit used by the great majority of mortgage borrowers to obtain home purchase loans and refinancing loans at relatively
low interest rates. Given their low-cost and flexible features, prime conventional loans are central to any strategy to provide affordable housing opportunities. Yet, large numbers of minority borrowers, especially African-American borrowers and Hispanic borrowers, have not been able to access the prime conventional mortgage loan market.
In the US as a whole in 2000, prime conventional loans accounted for 74% of the home purchase loans made to white borrowers, but only 50% of such loans made to Hispanic borrowers and only 40% of such loans made to African-American borrowers. The balance of home purchase financing is provided by government-insured loans, subprime loans, and mobile home loans. Compared to prime conventional loans, government-insured loans (primarily FHA-insured loans) have somewhat higher costs, provide less flexibility in refinancing, forego the possibility of borrower participation in one of the many special loan programns that many banks have established for low and
moderate income borrowers, and provide much less incentive for lenders to ensure that the real estate tranactions they finance are condicted on a fair and honest basis. Subprime loans and mobile home loans, even when they do not contain predatory features, represent an extremely high-cost form of mortgage credit. The restricted access of African-Americans and Hispanics to prime conventional home purchase loans means that a disproportionately large share of the home purchase loans made to these borrowers are high-cost or less beneficial forms of mortgage credit.
A similar pattern of racial disparity in accessing low-cost prime conventional mortgage credit is found in the refinancing loan market. In the US as a whole in 2000, prime conventional loans accounted for 80% of the refinancing loans made to white borrowers, but only 65% of such loans made to Hispanic borrowers and only 48% of such loans made to African-American borrowers. The dichotomy between low-cost prime conventional loans and high-cost alternatives is even starker in the refinancing loan market than in the home purchase loan market because in the refinancing loan market there is very little government-insured lending activity to serve as a buffer between low-cost prime conventional loans and high-cost subprime and mobile home loans. For the US as a whole in 2000, subprime loans comprised 17% of the refinancing loans made to White borrowers, but 29% of refinancing loans made to Hispanic borrowers and 46% of refinancing loans made to African-American borrowers.
As shown in Table 5, the level of GSE support for African-American and Hispanic borrowers is substantially below that provided to White borrowers. What is most striking about Table 5 is that the racial disparities in accessing the GSE secondary market persist and, indeed, are almost constant at all income levels. Table 6 indicates that relative GSE support for Hispanic borrowers is much weaker when the total mortgage market, rather than just that conventional loan market, is used as the relevant market for judging GSE performance. This shift stems from heavy reliance on FHA-insurede loans by many Hispanic borrowers.
The great majority of African-American and Hispanic borrowers fall within one or more of the three broadly-defined housing goal categories that are currently employed to implement the GSE housing goal requirements. Yet, the underwriting standards of the GSEs and, more generally, the prime loan origination market as a whole, as well as the lingering effects of historic patterns of discrimination, disadvantage many African-American and Hispanic borrowers to such an extent that simply including them within the broadely-defined housing goal categories does not provide them with adequate access to the GSE secondary market. Under these circumstances, the most effective way to improve GSE performance with respect to African-American and Hispanic borrowers would be to establish a specific subgoals for these borrowers.
C. Failure to Establish Subgoals for Local Geographic Areas.
The most dramatic example of broadly-defined GSE housing goal categories lacking subgoals for important subsectors is the fact that all existing housing goal requirements apply only at the national level. There are no subgoals for local geographies, such as individual MSAs or states. This housing goal structure gives the GSEs carte blanche to trade-off strong housing goal performance in some local geographies for weak housing goal performance in others. Under the current system, the performance of the GSEs within individual states or MSAs lies beyond the reach of regulatory control or even regulatory encouragement.
The scope and severity of obstacles in accessing mortgage credit vary greatly between MSAs. In fact, variations in credit access conditions are often greater between MSAs than between the well-served and underserved mortgage market segments of individual MSAs. This assessment is based on tabulating credit access conditions within individual MSAs on a MSA-wide basis and then comparing these MSA-wide averages. This pattern of wide divergence between MSAs indicates that there are a significant number of MSAs with serious credit access problems that effect a very large proportion of their households. Our online HMDA Database ( http://www.public-gis.org ) provides an extensive set of HMDA data tables that examine this issue.
Similarly, tabulating GSE market share ratios on a MSA-wide basis and then comparing GSE market share ratios between MSAs reveals a pattern of substantial divergence between MSAs in the average level of GSE support for the mortgage market. Indeed, the variation in GSE market share ratios between MSAs is greater than the variation between well-served and underserved segments within individual MSAs.
In 2000, Fannie Mae and Freddie Mac together purchased 62.5% of all conventional home mortgage loans originated in US metro areas. However, in 53 metro areas, they purchased less than 40% of such loans and in 28 of these metro areas, they purchased less than 35% of such loans.
This variation in GSE support for the home mortgage loan market from one MSA to the next has a major impact not only on the overall level of GSE purchases in each metro area, but also on GSE purchases of loans to underserved borrowers and underserved neighborhoods within individual metro areas. In 2000, the two GSEs together purchased 45.9% of all conventional home mortgage loans to low-income borrowers in US metro areas. Yet, in 76 metro areas, they purchased less than 25% of such loans. Also, in 2000, the two GSE purchased 35.8% of all conventional home mortgage loans made to African-American borrowers in US metro areas. However, in 47 metro areas, they purchased less than 20% of such loans.
Typically, the performance of the GSEs is weakest in small and medium-sized MSAs. These MSAs tend to be local mortgage markets in which GSE activities are less visible and attract only limited political attention. Table 7 lists the 68 MSAs in which the GSEs in 2000 purchased less than 23% of conventional loans made to low-income borrowers. As indicated in Table 7, these MSAs tended to have small-sized mortgage markets and very low GSE market shares not only for low income borrowers, but for the entire MSA.
These wide swings between MSAs in the level of housing finance access barriers and GSE support for the mortgage market suggest that the failure to establish specific housing goals or subgoals at the MSA level or some other sub-national geographic level is a major concern.
Congress expressly granted HUD regulatory authority to establish subgoals in
implementing the GSE housing goal requirements. In doing so, Congress did indicate some ambivalence about the role of subgoals. The legislative grant of subgoal authority to HUD states that subgoals established by HUD will not be "enforceable". Such ambivalence is typical of legislation that is the product of comprise between sharply contrasting perspectives -- in this case, the desire of affordable housing advocates for strong GSE housing goal requiremenenst on one hand and the GSEs interest in minimizing federal regulation of their activities on the other.
The most reasonable interpretation of the GSE legislation is that if subgoals are needed, HUD should establish them even though they would only operate on a voluntary basis. Even voluntary subgoals are likely to have a significant impact on GSE performance in view of the GSEs great concern with maintaining a favorable public image. On the other hand, substantial disregard of volunrtary subgolas by the GSEs would, in all likelihood, build public suport for legislative changes that would make subgoals enforceable. What is clear is that if HUD never uses its existing authority to establish voluntary sungoals, it will be much harder to develop support for legislation changing the regulatory status of subgolas from a voluntary to a mandatory basis.
|
