HUD's proposed housing goal rules for Fannie Mae and Freddie Mac (the GSEs) and the many public announcements of Fannie Mae and Freddie Mac on their housing commitments have fostered the general impression that the GSE housing goals have been a success, although some adjustments are needed to strengthen the goals.  Looked at from a local  community reinvestment perspective and a broad civil rights perspective, however, there are major problems in the implementation of the GSE housing goals.   In our view, these problems stem in large part from serious flaws in the structure of the goals. 

          Our conclusion is based on an extensive review of GSE performance in individual metro areas (MSAs) across the nation.  This review indicates that in many MSAs in 1998 the GSE market share of 1-4 family mortgage loans in low and moderate income and minority neighborhoods was roughly one-half the GSE market share of such loans in upscale, non-minority neighborhoods. 

            In conducting this analysis, we measured GSE market share as the ratio between Fannie Mae and Freddie Mac's combined purchases of 1-4 family mortgages and the total number of conventional 1-4 family mortgage loan originations reported under HMDA with loan amounts below  $227,000 -- the 1998 GSE loan amount ceiling for mortgages on 1-unit homes.  GSE purchases are based on the Census Tract File in the GSE Public Use Database.  This file does not distinguish between home purchase loans and refinancing loans, so we consolidated these two types of mortgage loans in computing GSE market shares. 

            Implicit in the GSE housing goal concept is the idea that the GSEs should, at minimum, support the mortgage market in low and moderate income and minority neighborhoods to the same extent that they support the mortgage market in affluent, non-minority neighborhoods.  Clearly, this minimum standard of support is not being met where the GSE market share of mortgage loans in low and moderate income or minority neighborhoods is much less than the GSE market share in affluent or non-minority neighborhoods.  Similarly, the GSEs fall short of this minimum standard when the GSE market share of mortgage loans to low or moderate income or minority borrowers is well below the GSE market share of loans to more affluent or non-minority borrowers.

         Attached are four maps that provide stark examples of how the GSE market share tends to decline sharply in inner-city neighborhoods.   The 1998 Chicago(.pdf) map, Cleveland(.pdf) map, and Akron(.pdf) map were submitted to HUD pursuant to the pending rulemaking on GSE housing goals by Illinois ACORN and the Metropolitan Strategy Group of Ohio.  The 1996 Chicago(.pdf) map was presented by Essential Information at a 1998 conference on the role of the GSEs.

             The maps contain a diagonal overlay to identify minority census tracts.  In the case of the Chicago maps, the diagonal overlay identifies census tracts in which non-White and non-Asian persons comprise 70% or more of the census tract population.   In the Cleveland and Akron maps, the diagonal overlay indicates census tracts in which minority persons comprise 40 % or more of the census tract population.  As shown in the maps, there is a very strong correlation between  the racial composition of census tracts and the GSE market share.   Also, comparison of the 1996 and 1998 Chicago maps show that the overall pattern of GSE market share disparity did not changed very much during this two-year period.      

            To provide further insight into the nature of GSE housing goal performance in the Chicago PMSA, we prepared a set of statistical tables on 1998 GSE market shares by census tract race and income categories and by borrower race and income categories.  These tables are based on GSE loan purchase data and HMDA loan origination data for the nine-county Chicago PMSA. 

            Table 1 provides GSE market share data by census tract racial category.  Table 1 shows a decline in GSE market share from 75.10% in predominantly non-minority census tracts to only 33.28% in census tracts in which minorities comprise 75% or more of the census tract population.  In the Chicago PMSA, these two census tract categories represent almost 60% of the PMSA's total of 1,776 census tracts. 

            Table 2 provides GSE market share data by census tract income category.  This table shows a decline in GSE market share from 77.54% in upper income census tracts to only 39.01% in low income census tracts.  The census tract income related decline in GSE market share shown in Table 2 is similar to the census tract race related decline shown in Table 1, except that not as many census tracts are pushed down to the lowest market share interval.  This is consistent with our general experience that a sharper focus on GSE market share disparities can often be obtained by looking at census tract race rather than census tract income. 

            Table 3 compares the GSE market share in minority census tracts versus non-minority census tracts for separate census tract income intervals.  This table show that while the GSE market share in non-minority census tracts ranges from 61.27% to 69.62% for census tract within various moderate and middle income intervals, it never rises above 42.61% for minority census tracts in comparable census tract income intervals.  While such census tract-based analysis does not by itself establish disparate treatment or disparate impact in violation of the federal non-discrimination laws, it does indicate that there are major disparities between neighborhoods along racial lines in accessing the GSE-operated secondary mortgage market.  This disparity needs to be addressed directly in implementing the GSE housing goals.    

            Table 4 provides GSE market share data by borrower income category.  This table indicates that borrower income related disparities in GSE market shares are not as great as neighborhood related disparities.  Further, the GSE market share disparity is only substantial for low income borrowers and not much of a factor for borrowers with incomes above 80% of MSA median family income.     

            Table 5 provides GSE market share data by borrower racial category.  This table shows that the GSE market share is dramatically lower for African-American borrowers and moderately lower for Hispanic borrowers.   

            The failure of the GSE housing goals to provide residents of many minority neighborhoods and low and moderate income neighborhoods and, more generally,  minority borrowers and low and moderate income borrowers with the same access to the secondary mortgage market enjoyed by most Americans is rooted in a number of major structural flaws in the 1992 GSE housing goal legislation.  

            First, the legislation used unduly broad definitions to determine the focus of the GSE housing goals with the result that they are not properly targeted on underserved sectors of the mortgage market.  Much of this inadequate targeting or dilution is associated with misapplication of terms commonly used in CRA and community development activities.  For purposes of the Low and Moderate Income Housing Goal, low and moderate income borrowers are defined to include not only borrowers with incomes less than 80% of MSA median family income (as they are for CRA purposes), but also borrowers with incomes between 80% and 100% of MSA median family income.  Yet, Table 4 shows that in the Chicago PMSA,  borrowers with incomes in the 80-100% of MSA MFI range (lower-middle income borrowers) have a GSE market share (69.03%) that is the same as the average GSE market share for the PMSA as a whole (68.92%).  Why are GSE housing goals needed to encourage GSE purchases from this middle subsector of the mortgage market?  Moreover, if the GSE housing goals were successful in encouraging the GSEs to change underwriting and other policies to stimulate more GSE purchases of loans to low and moderate income borrowers, would it not follow naturally that borrowers in the 80-100% of MSA MFI range would also benefit from such changes? 

            Similarly, for purposes of the Special Affordable Housing Goal, the GSE legislation defined low income borrowers to mean borrowers with incomes less than 80% of MSA median family income, rather than the 50% of MSA median family income standard used for CRA purposes.  Yet, as Table 4 shows, it is borrowers with incomes below 50% of MSA  MFI  who have the most difficulty in accessing the GSE secondary market and clearly the most need for a ³special affordable housing goal.²           

           In addition, the GSE legislation established a geographic GSE housing goal that included all central cities, rural areas, and other underserved areas as defined by HUD.  Clearly, this was an excessively broad definition of geographic areas with restricted access to the GSE-operated secondary market, although HUD has to some extent narrowed its scope by means of its regulatory authority to define underserved geographies.

            Second, the GSE legislation prohibits HUD from establishing enforceable subgoals for the Low and Moderate Income Housing Goal and the geographic area goal.  Subgoals are a  logical tool to ensure that the most underserved sectors of the mortgage market are given adequate consideration by the GSEs.  For example, census tracts with minority population percentages above 50%, or even 75%, would be good candidates for an enforceable subgoal. 

            Third, and perhaps most important of all, the GSE legislation failed to establish separate GSE housing goals (or sets of housing goals) for 1-4 family mortgage loans and multifamily mortgage loans.  As result, in implementing all three statutory housing goals, HUD has devised a housing goal scoring system that counts the number of housing units financed by GSE loan purchases and combines 1-4 family units and multifamily units.  This consolidation of 1-4 family and multifamily units has had a major impact in diluting the impact of the Special Affordable Housing Goal, the one statutory GSE housing goal that made some effort to achieve proper targeting. 

             Because a very high percentage of multifamily housing loans qualify for the Special Affordable Housing Goal, the GSEs can significantly reduce their obligation to purchase 1-4 family housing loans under the Special Affordable Housing Goal by increasing their purchases of multifamily mortgage loans.  According to HUD, roughly 50% of the multifamily mortgage loan purchases of the GSEs qualify for the Special Affordable Housing Goal.  In 1997, multifamily loans accounted for 44% of Fannie Mae's Special Affordable Housing Goal units and 31% of such units for Freddie Mac.  In short, the commingling of 1-4 family housing loans and multifamily housing loans within the same Special Affordable Housing Goal basket enables the GSEs to trade-off performance on the 1-4 family side versus performance of the multifamily side.  

            Reinforcing these statutory weaknesses, HUD set the Special Affordable Housing Goal too low in the 1995 GSE rulemaking.  Although HUD estimated that Special Affordable Housing units comprised 20%-23% of the market, it set the Special Affordable Housing Goal for the 1997-98 period at only 14% of GSE loan purchases measured on a housing unit basis.   

           Several basic reforms are needed to improve the working of the GSE housing goals.

First, with respect to the 1-4 family side of the mortgage market, underserved sectors should be defined by looking at GSE market shares.  Such market share analysis provides the most efficient means to identify underserved census tract categories, borrower income categories, and borrower racial categories.   Obviously, such GSE market share data was not available to Congress in 1992 when it enacted the GSE housing goal legislation.  Now that the data is available, Congress should examine it closely and restructure the statutory GSE housing goals to provide for better targeting. 

            In implementing the GSE housing goal legislation,  HUD has generally relied on loan origination and loan denial rates derived from HMDA data to identify underserved segments of the mortgage market.  While this approach has some value and we have employed it ourselves, our experience has been that GSE market share data provide a much sharper lens for identifying market segments that have restricted access to the GSE-operated secondary mortgage market. We recommend that HUD use this approach in its current rulemaking to bring as much targeting to the GSE housing goal rules as possible within the statutory framework. 

            Second, the GSE housing goal legislation needs to be restructured to provide separate housing goals or sets of housing goal for 1-4 family housing and multifamily housing.   This reform is needed  to prevent the GSEs from trading off improved housing goal performance on the multifamily side for a reduced housing goal obligation on the 1-4 family side.  As noted above, most of this trade-off takes place within the Special Affordable Housing Goal.

            Multifamily housing loans and 1-4 family housing loans targeted toward underserved sectors are the most important elements of any strategy to address affordable housing needs. Thus, GSE housing goals should be structured in a manner that encourages the GSE to take affirmative measures on both fronts.  

            Third, better targeted GSE housing goals for 1-4 family housing loans are needed to encourage the GSEs to purchase more CRA-type mortgage loans originated by banks and savings associations.  One of the main reasons why the GSE market share is so low in most minority and low and moderate income census tracts is that many of the mortgage loans originated in such tracts are CRA-type loans that often are not sold to the GSEs.  In some cases, these loans have below-market rate features and the GSEs will not purchase such loans without discounting them to market rates.  In our view, given the large indirect subsidies received by the GSEs and their substantial resources, including strong earnings streams, the GSE should share more equally with banks the various costs of effective CRA implementation.  For example,  in the case of  CRA-type loans with below-market rates, the GSEs should absorb some of the cost of marking these loans to market.  This would make banks more willing to sell CRA-type loans to the GSEs and this would, in turn, significantly expand the capacity of banks to originate CRA-type loans. 

            Fourth, HUD should establish a process for administrative review of GSE performance at the local level when the GSE market share data indicate that the GSEs have significantly underserved  minority census tracts or low and moderate income census tracts within a particular MSA or rural area.  Local government officials, community groups, and lenders should be encouraged to participate in this administrative review.  The goal of this administrative process should be to encourage the GSEs to make commitments to improve their housing goal performance in the local community. 

            Review of GSE performance within individual MSAs or rural areas is important because if the GSEs are evaluated only in terms of national targets they can substantially underserve individual MSAs and still meet these national targets.  Also, it is very difficult to set uniform targeting standards that are appropriate for all MSA or rural areas in the nation.  In particular, the  relationship between minority percentage of census tract population and GSE market share varies dramatically in different parts of the nation.  In Brooklyn, neighborhood access to the GSE-operated secondary mortgage market may not be greatly restricted until the minority percentage rises well above 50%.  By contrast, in Iowa, such restricted access may be seen in census tracts with a minority percentage of only 15%.  

          Fifth, with respect to 1-4 family mortgage loans, separate and enforceable subgoals should be established for home purchase loans and refinancing loans.   The GSE housing goals could be implemented much more effectively if for evaluation purposes the home purchase loan market were separated from the volatile and increasingly controversial refinancing loan market. 

          Finally, some comment on whether the GSEs should be given housing goal credit for the purchase of subprime loans is appropriate.  In many minority neighborhoods of the nation's larger cities, more than 40% or even 50% of the refinancing loans originated in 1998 were made by mortgage lenders that HUD has classified as subprime lenders.  If the GSEs were to purchase a substantial share of these loans and were given housing goal credit for such purchases, this would be booked for evaluation purposes as a substantial improvement in GSE housing goal performance in these neighborhoods.  In our view, however, the primary mission of the GSEs with respect to housing goals should be to bring as many borrowers as possible into the mainstream of prime mortgage credit financing or provide a strong secondary market for legitimate CRA-type lending.  At this point, it would be counterproductive to this underlying GSE mission to encourage the GSEs to purchase subprime loans by giving them housing goal credit for such purchases.