National economic growth for the first quarter of 1997 was very strong based on advanced estimates. The annualized growth rate of 5.6 percent was the highest it has been in a decade. This buoyant national economy provides the landscape upon which to view the condition of the U.S. housing market.

In 1997, housing has made a solid debut, although there is some concern that the unusually warm weather has borrowed some starts from later in the Spring. The first quarter of 1997 saw modest gains in housing production levels from the last quarter, but the levels are quite high and rival last year's excellent performance.

  • Permits in the first quarter totalled 1,422,000 on a seasonally adjusted annual basis, up 2 percent from last quarter and 1 percent from the first quarter of 1996. Single-family permits, at 1,048,000, were up 3 percent from last quarter but down 3 percent from the first quarter a year ago.

  • Starts totalled 1,441,000, a statistically insignificant 2 percent above last quarter and down a statistically insignificant 2 percent from a year ago. There were 1,153,000 single-family starts, 6 percent above last quarter but down a statistically insignificant 1 percent from a year ago.

Marketing is also off to a good start. This can be said even though the quarter's performance is being compared with 1996's record-setting pace (the third best year for new home sales and the best year ever for existing home sales).

  • New home sales in the first quarter, at 824,000 units, are 8 percent ahead of last quarter and 12 percent ahead of the first quarter of 1996.

  • Sales of existing homes are up 2 percent from last quarter and up 3 percent from the first quarter of 1996.

  • The median price of a new home declined slightly while the median price of an existing home rose 2 percent during the first quarter.

Inventories of new homes declined to 297,000 units, a 4.5 months' supply, whereas the inventory of existing homes increased to 2,350,000 homes, a 6.9 months' supply. The first-quarter inventory decline of new homes was 8 percent, while the increase in inventory of existing homes was a seemingly large 52 percent.

Interest rates rose slightly -- 10 basis points during the quarter and about 50 basis points from a year ago -- but have not reduced housing market activity. However, these rates are still below 8 percent, a level that is supportive of housing markets. Recent Federal Reserve actions are not yet reflected in mortgage interest rates although not much of a change in housing is expected as a result of the quarter-point change.

Such robust housing activity may be the direct result of housing affordability, which continues at a high level, although there were slight declines from last quarter. Furthermore, builders (as reported by the National Association of Home Builders) seem upbeat, especially about future sales expectations.

The multifamily production and multifamily rental markets are somewhat mixed. Permits were off slightly from last quarter but were up substantially from the first quarter of 1996. Starts were up 3 percent from last quarter but down 11 percent from a year ago. The market absorption of newly completed apartments continues at a good level, with 72 percent of apartments being leased up within 90 days of completion, although the number of apartments completed was down significantly. Rental vacancy rates improved, with a decline to 7.5 percent from 1996's record high of 7.8 percent.

Regional Perspective

HUD's field economists report that home sales and single-family permit activity during the first quarter continued the strong performance of 1996. Builders and REALTORS® in much of the country report that buyer confidence is at a high level and are looking forward to a healthy sales housing market this Spring.

Sales housing markets in the New England and the New York/New Jersey regions have improved, with first-quarter building permits up more than 12 percent compared with the first quarter of 1996. In the Washington, D.C. area, consumer surveys predict a strong Spring market. In the Southeast new home sales have been especially strong in the Atlanta suburbs. Single-family permits in the Midwest were down 10 percent in the first quarter, but this was still one of the strongest first quarters in the past 10 years for the region. The Chicago area's sales market remains very strong. Home shows in the Minneapolis-St. Paul area drew record numbers of prospective buyers.

The Denver area has seen a surge in the production of homes in the $125,000 to $150,000 price range, while sales of higher priced homes have slowed slightly. California housing markets show steady improvement; a February survey reported that the annual sales volume was up almost 11 percent over 1996. Single-family permits in the San Francisco Bay Area, Orange County, and Los Angeles County were significantly above first-quarter 1996 levels. First-quarter home sales in the Seattle area were up more than 20 percent over the first quarter of 1996, and soaring land prices are causing builders to look to the condominium market.

Rental markets remain firm throughout much of the Nation. Multifamily housing permit activity was up considerably in the first quarter in the tight Boston, New York, and Northern New Jersey markets. Last year's improved levels of apartment construction are being sustained in much of the Midwest, Southeast, and Southwest regions. There is still concern, however, about the capacity of some markets to absorb the increased supply of rental housing. Suburban Chicago submarkets have become more competitive. Apartment absorption has slowed in Denver, Austin, and Phoenix, where competition has increased in submarkets with concentrations of luxury apartments. First-quarter multifamily permits dropped in Denver, Salt Lake City, Colorado Springs, Phoenix, and Las Vegas as these markets adjust to the large volume of apartments still coming on line.


The Department of Housing and Urban Development has released to the public extensive information on the mortgage purchases of Fannie Mae and Freddie Mac, two Government-sponsored enterprises (GSEs) that HUD oversees. The GSEs are secondary market institutions that purchase conventional loans originating in the United States. The data include information on every single-family and multifamily mortgage purchased by the GSEs from 1993 to 1995.1 This release marks another major step in HUD's democratization of data efforts.

The information is intended to help mortgage lenders, planners, researchers, and housing advocates study the flow of mortgage credit and capital in America's communities. It will also help people understand where Fannie Mae and Freddie Mac are focusing their affordable homeownership efforts and how their affordable lending performance compares with those of other lenders in local market areas.

History of GSE Legislation and Housing Goals

The Federal Housing Enterprises Financial Safety and Soundness Act of 1992 (the 1992 GSE Act) required HUD to create a database on mortgage purchases by the GSEs and make it available to the public. In 1992, Congress felt there was an "information vacuum" on the type of mortgages the GSEs were purchasing, the neighborhoods from which they were buying the mortgages, and the borrowers they were serving. 2 Because Fannie Mae and Freddie Mac receive significant benefits from their Government-sponsored status, Congress wanted to assure that the GSEs were benefitting all types of borrowers and communities.3

As required by the 1992 GSE Act, the Department has established income-based and geographically targeted housing goals for the purchase of mortgages by each GSE. The housing goals ensure that an appropriate portion of each GSE's mortgage purchases are targeted to lower income families and neighborhoods underserved by the mortgage market. By regulation HUD has established three specific goals for 1996 and 1997-99: a low- and moderate-income goal that focuses on families with less than median incomes, a geographically targeted goal aimed at low-income and high-minority census tracts, and a special affordable housing goal that focuses on very-low-income families.4 (See the following box for definitions of the goals.)

Low- and Moderate-Income Housing Goal

The low- and moderate-income goal is defined in terms of purchases of mortgage loans made to families with incomes at or below the area median income (AMI). At least 40 percent of the dwelling units in properties whose mortgages were purchased by the GSEs in 1996 had to be for such families, with this goal rising to 42 percent for 1997-99.

Geographically Targeted Housing Goal

The geographically targeted goal requires the GSEs to purchase mortgages on properties located in census tracts within metropolitan areas where either (a) the median income of families does not exceed 90 percent of the AMI or (b) minorities make up 30 percent or more of the residents and the median income of families does not exceed 120 percent of the AMI. A similar definition applies in nonmetropolitan areas. In 1996, 21 percent of mortgages purchased by the GSEs had to be located in these areas, with this goal rising to 24 percent for 1997-99. HUD determined that income and race are appropriate statistical indicators of underservice.

Special Affordable Housing Goal

The special affordable housing goal is both income and geographically based. The GSEs are directed to purchase mortgages on units occupied by low-income owners and renters in low-income areas and on units in any area occupied by very-low-income owners and renters. The special affordable housing goal is set at 12 percent and 14 percent, respectively, of the total number of dwelling units financed by each GSE's mortgage purchases for 1996 and 1997-99. A further requirement is that, among its mortgage purchases that meet total special affordable housing goal, each GSE must annually purchase qualifying multifamily mortgages in an amount at least equal to 0.8 percent of its total dollar volume of mortgages purchased in 1994. This minimum multifamily dollar volume is $1.29 billion for Fannie Mae and $988 million for Freddie Mac.

As shown in table 1, the GSEs' activity under the goals has improved since 1993.5 For example, Fannie Mae's low- and moderate-income goal performance improved from 34 percent in 1993 to 43 percent in 1995. This improved performance is consistent with several trends in the market. From 1993 to 1995, the market shifted from a predominantly refinance market to a home purchase market. Coinciding with this shift, the GSEs, along with private lenders and mortgage insurers, were changing their underwriting standards to treat lower income families more equitably. In addition, these changes were complemented with new mortgage products, such as 97-percent loan-to-value (LTV) ratio mortgages, which are designed to assist borrowers with insufficient savings for a downpayment. Taken together, all of these changes led to more mortgages for lower income and minority homebuyers and neighborhoods.

The GSE Public Use Data Base provides further insight into how the affordable housing goals are being met. The reporting of data by the GSEs is designed to track their performance in meeting the housing goals and fill the earlier information vacuum on the GSEs' overall activities.

Overview of the GSE Public Use Data Base

Beginning with 1993 the GSEs have provided HUD annually with loan-level data on each single-family and multifamily mortgage they acquire. The single-family component of the database consists of three loan-level files, one of which includes census-tract identifiers. The Census Tract File has loan-level information on the census-tract location of the properties that secure mortgages purchased by the GSEs. This file allows analysis of the GSEs' mortgage purchases by groups interested in local communities, counties, and cities. Since the Census Tract File contains information on the race, gender, and income of the borrower(s), community groups can analyze borrower characteristics as well as neighborhood characteristics of the GSEs' purchases. Additional variables on this file are the unpaid principal balance of the mortgage and whether the mortgage was purchased by a first-time homebuyer or a repeat buyer.

Because the GSEs operate in a competitive market, HUD had to omit certain variables that were deemed proprietary by the Secretary. In addition, it was necessary to recategorize some of the variables into ranges so that they were no longer proprietary.

To further fill the "information vacuum," yet continue to protect the proprietary interests of the GSEs, HUD created additional file structures for the dissemination of important mortgage data, such as the LTV ratio and whether the loan was a current or prior year origination. To accomplish this objective, two additional files were released that removed geographic information.6 A similar method was used to release the GSEs' multifamily data. 7 These modified data files are available to the public through the GSE Public Use Data Base, with lists of the available data elements presented in table 2.

The GSE Public Use Data Base data files that are available for 1993-95 contain information on 10.8 million single-family units and 750,000 multifamily units. The 1996 loan-level data, expected to be available by the end of Summer 1997, will add another 3 million single-family units and 372,000 multifamily units.

GSE Public Use Data Base and HMDA database. In some ways the GSE Public Use Data Base is similar to the Home Mortgage Disclosure Act (HMDA) database. Both databases provide loan-level information on mortgages, including the census-tract location of the property, unpaid principal balance of the mortgage, and the borrower's race, gender, and income. In addition, both databases offer information on the type of institution (thrift, bank, mortgage company) that sold the mortgage into the secondary market.

There are also differences between these databases. For example, the GSE Public Use Data Base provides information on every GSE acquisition for 1993-95 whereas HMDA underreports the GSEs' mortgage purchases.8 In addition, the GSE Public Use Data Base has information on prior-year originations and first-time homebuyers. It provides information on a unit-by-unit basis, while HMDA's information is at the property level. HMDA does link information on whether mortgages are home purchase or refinance mortgages to census-tract identifiers, whereas the GSE Public Use Data Base does not. The GSE database includes information on LTV ratios, but with only census-tract characteristics, not census-tract identifiers. HMDA provides information on loan applications, including applications that have been denied or withdrawn; the GSE database is based on originations. Finally, HMDA has broader coverage of the mortgage market by including originations by banks and thrifts. Overall, the GSE Public Use Data Base and the HMDA database are complementary, with HMDA providing a good source for market data and the GSE data including details on the GSEs' secondary market activities.

Table 1. Overview of the GSEs' Activity Relative to Final Rule Goal Definitions 1


Table 2. GSE Public Use Data Base -- Types of Available Information


Studying GSEs' Purchase Activities

The GSE database can be used to measure the GSEs' performance in meeting their housing goals, address major policy issues at the national level, characterize borrowers and identify neighborhoods from which the GSEs are purchasing mortgages, and examine local trends in the GSEs' purchases. Several variables in the GSE Public Use Data Base can be used by analysts to assess and monitor the GSEs' activities and performance.

Borrower characteristics. An important reason for the 1992 GSE Act's requirement that Fannie Mae and Freddie Mac submit loan-level data was to determine what types of borrowers the GSEs were serving, particularly to determine the shares of mortgage lending that helped lower income families become homeowners or that provided affordable rental housing. In 1995, 9.3 percent of Fannie Mae's purchases were mortgages for very-low-income borrowers, while 6.8 percent of Freddie Mac's purchases were in this category. As displayed in figure 1, both GSEs increased their purchases of mortgages for very-low-income borrowers during the 1993-95 period, with increases of 6.8 percent to 9.3 percent and 6.2 percent to 6.8 percent for Fannie Mae and Freddie Mac, respectively.

Income is only one borrower characteristic. A second characteristic that is analyzed is the race of the borrower. As shown in figure 2, from 1993 to 1995, mortgages made to minority borrowers increased from 9.8 percent to 14.6 percent of Fannie Mae's total purchases and from 9.0 percent to 10.5 percent of Freddie Mac's total purchases. The data in figures 1 and 2 show the GSEs' acquisitions at the national level. These same comparisons can be done for a particular metropolitan area or group of neighborhoods.

Figure 1. GSEs' Mortgage Acquisitions From Very-Low-Income Borrowers 1993-95

Figure 1

Figure 2. GSEs' Mortgage Acquisitions From Asians, African Americans, and Hispanics 1993-95

Figure 2

Neighborhood characteristics. From 1993 to 1995, Fannie Mae's purchases in low-income census tracts exceeded Freddie Mac's activity, as shown in figure 3. In addition, from 1993 to 1994, these purchases jumped from 6.8 percent to 8.7 percent of Fannie Mae's total purchases, while Freddie Mac's activity was essentially constant at 6.5 percent and 6.6 percent of its total purchases over the same period. Another way to characterize the GSEs' activities in low-income tracts is to look at the minority concentration of the census tracts. As shown in figure 4, in 1995, 55.5 percent of the GSEs' combined activity in low-income census tracts (0-80 percent) was also in predominantly minority census tracts (30-100 percent); the remaining 44.5 percent of the activity was in census tracts that were less than 30 percent minority. In contrast the GSEs' activities in higher income census tracts were concentrated in low-minority neighborhoods. This same type of analysis can also be done at the local level.

The analysis of neighborhood characteristics highlights some of the issues that HUD addressed in finalizing the GSE rule. One of the significant changes in the final rule was the change in the Geographically Targeted Goal from an Office of Management and Budget (OMB) central-cities-based definition to a tract-based definition that focuses on low-income and high-minority census tracts. The 1992 GSE Act required the GSEs to reach a 30-percent target level by the end of 1994 for purchasing mortgages from OMB-defined central cities. This definition gave HUD time to study the characteristics of areas that were not served adequately by the mortgage market.

Figure 3. GSEs' Acquisitions of Home Purchase Mortgages in Low-Income Census Tracts 1993-95

Figure 3

During the 1993-95 transition period, HUD studied the central-cities-based definition and concluded that there were portions of central cities that could not be characterized as underserved. The portions of central cities experiencing market problems, such as high denial rates, were emphasized by the new geographic goal. HUD chose a more targeted approach, redefining the goal to focus on lower income and minority census tracts. The change in the definition is illustrated by the map of the Washington, D.C., metropolitan statistical area (MSA), shown in figure 5. OMB central cities in the Washington, D.C., MSA include Washington, D.C., and Arlington, Virginia, which are outlined by the diamond-shaped box in the center of the map. 9

A portion of the wealthier northwest quadrant of Washington (in white) is classified as served in the new framework, while the poorer minority neighbor-hoods in northeast and southeast Washington (in green) are classified as underserved. Underserved areas outside of the Washington and Arlington central cities were not specifically targeted under the old OMB central-cities-based definition, but are included in the new geographic goal. The new tract-based definition is a better approach to targeting the credit needs of underserved areas.

Other Policy Issues

The GSE data can also be used to analyze other national policy issues, leading to interesting and counterintuitive findings. One issue has been the role of the GSEs in affordable lending with low-downpayment mortgages. Conventional wisdom indicates that lower income borrowers would make smaller downpayments and thus would have higher LTV ratios but, as shown in figure 6, the GSE loan-level data reveal that 71 percent of the GSEs' 1995 mortgages to borrowers with very low incomes (0-60 percent of AMI) had LTV ratios at or below 80 percent. This percentage is greater than the corresponding percentages for the other two borrower incomes classes. 10 Further research into findings such as these can be done with the GSE Public Use Data Base.

Availability of the Database

The 1993-95 single-family and multifamily loan-level data sets are available from HUD USER for all GSE acquisitions or by year. In addition, if an analyst is interested in obtaining information about a particular MSA or State, or for all nonmetropolitan areas or multifamily units, the data are available on diskette.11 As an alternative to obtaining data from HUD USER, analysts may acquire data on the GSEs' mortgage activities in particular MSAs or States by accessing the Unison Institute's Web page.12 At that Web site, analysts can select the type of data desired including location of the property, year of acquisition, and one or both GSEs.

Figure 4. GSEs' Mortgage Acquisitions by Census Tract Minority Concentration and Income Level 1995

Figure 4

Figure 5. Washington, DC Metropolitan Area Underserved Areas

Figure 5

Figure 6. GSEs' Mortgage Acqusistions with 80% or Less LTV by Borrower Income 1995

Figure 6

As part of the democratizing data effort, the GSE data are being made available through as many types of media as possible. At present, there are three Web sites that are providing information on the GSEs' loan-level purchases and on HUD's implementation of the 1992 GSE Act. As shown in figure 7, the three Web sites are linked yet are providing different avenues for the dissemination of information on the GSEs. The HUD Web site provides information on general GSE regulatory activities and the GSEs' performances relative to their housing goals, the HUD USER Web site makes available loan-level data and basic national aggregations on the GSEs' performances, and the Unison Institute's Web site is intended to provide information about particular localities whether tabular or at the loan level. 13

The Unison Institute's efforts will be an important avenue for community groups and analysts to access data on the GSEs' activities in local areas, making data extracts and standard reports readily accessible to all types of analysts. Unison's standard report component allows users to generate tables depicting the GSEs' activities for an MSA, a county, or even a group of census tracts. Unison's planned efforts include displaying mortgage market data, as derived from HMDA, alongside GSE data. This combination is a powerful tool that will allow local groups and analysts to see tables on the income, race, and gender of borrowers, and whether the GSEs' purchases were above or below the level of activity by banks and thrifts. 14 Unison has met with several focus groups to develop forms and reports that are user friendly, flexible, and informative. The report formats will be added to Unison's GSE Web site over the next few months.

Figure 7. Evolving Organizational Structure For Public Dissemination of Information
on GSE Activities Via the Internet

Figure 7

Housing Finance Working Paper Series

In addition to the work on the GSE Public Use Data Base, the Office of Policy Development and Research has initiated a Housing Finance Working Paper Series. This series is intended to cover topics related to housing finance, including the GSEs' activities. The first working papers completed are "The GSEs' Funding of Affordable Loans," by Harold L. Bunce and Randall M. Scheessele (December 1996) and "The Multifamily Secondary Mortgage Market: The Role of Government-Sponsored Enterprises," by William Segal and Edward J. Szymanoski (March 1997). 15

The Bunce and Scheessele paper illustrates how HMDA can be used to create a market reference for GSE purchases. 16 The paper compares the borrower and neighborhood characteristics of mortgages purchased by Fannie Mae and Freddie Mac with all mortgages originated in the conventional conforming loan market, including originations retained in portfolio by banks and thrift institutions.

In a paper that examines the role of Fannie Mae and Freddie Mac in the multifamily mortgage market, Segal and Szymanoski conclude that a secondary market for mortgages on properties affordable to lower income families lags in development relative to a private secondary mortgage market that has developed to address the finance needs of higher end properties. It is found that the GSEs have the potential to enhance the affordable housing segment of the multifamily mortgage market.

Other papers will analyze the GSEs' purchase activities from 1993 through 1995, the GSEs' funding of single-family rental properties, and the LTV ratios of mortgages purchased by the GSEs relative to total originations in the overall market.


1 HUD expects to release 1996 loan-level data by the end of Summer 1997.

2 The legislative history of the Act reflects congressional concern about the lack of complete and accurate data on the GSEs' business activities. S. Rep. No. 102-282, 102d Cong., 2d Sess., p. 39 (1992).

3 See Privatization of Fannie Mae and Freddie Mac: Desirability and Feasibility, A HUD Report (July 1996) on benefits received by the GSEs from their respective charters.

4 For more information on the Secretary's regulation of the GSEs, visit HUD's GSE Oversight Web page at

5 In October 1993 HUD established three transition housing goals for the GSEs for 1993-94; in November 1994 HUD extended these goals to 1995. In December 1995 HUD issued a regulation that revised the structure of the goals and established levels of the goals for 1996-99. Table 1 reflects measurement of the goals under the revised structure.

6 Despite the omission of geographic information from the files, information characterizing the location of the mortgaged properties, such as the affordability of the area and the minority concentration of the census tract, was maintained. For example, using National File A, an analyst could determine whether the GSEs were purchasing most of their high-LTV, single-family, owner-occupied mortgages from predominantly white, middle-class neighborhoods or from minority, lower income neighborhoods.

7 The primary purpose of the multifamily Census Tract File is to release information on the location of the property, general size of the property, and type of institution selling the property to the GSE. The National File was designed to provide information on the number and affordability of units in each property.

8 HMDA only requires lenders with a home or branch office in a metropolitan area and with assets of $10 million or more to submit data. Lenders who originate 100 or more home purchase loans in the prior calendar year must also submit data. For details, see A Guide to HMDA Reporting: Getting It Right!, Federal Financial Institutions Examination Council, May 1993.

9 There are two other OMB central cities in the Washington, D.C., MSA: Frederick, Maryland, and Fredericksburg, Virginia; Alexandria, Virginia, is not an OMB central city.

10 The loan-level data in National File A do not allow users to distinguish between home purchase and refinanced mortgages. However, this general result does hold if only home purchase mortgages are used.

11 Multifamily data, order forms for the GSE data sets, and additional data tables are available from the HUD USER Web site:

12 The Unison Institute's Right-to-Know Network Web page can be found at

13 The Office of Policy Development and Research has compiled a set of tables derived from the GSE Public Use Data Base. For more information about these tables, please contact Robert Scavotto at (202) 401-0388 or (202) 708-1455 (TTY) or by writing to him at: U.S. Department of Housing and Urban Development, Office of Policy Development and Research, 451 Seventh Street SW, Washington, DC 20410 or Robert_J._Scavotto@HUD.GOV .

14 As part of an earlier HUD-sponsored effort at democratizing data, the Unison Institute made HMDA data readily available for local area analysis. The Right-to-Know Network's Web address for HMDA data is:

15 Copies of the Bunce & Scheessele and Segal & Szymanoski papers may be obtained from HUD USER by calling (800) 245-2691, (202) 708-9981 (fax), (800) 927-7589 (TDD), or mailing a check request to HUD USER, P.O. Box 23268, Washington, D.C. 20026-3268. Each working paper is $5.00 per copy. If you wish to discuss the working papers or would like more information about any of the Housing Finance Working Paper Series, please contact the authors by calling (202) 401­0388 or (202) 708­1455 (TTY) or by writing to the author(s) at: U.S. Department of Housing and Urban Development, Office of Policy Development and Research, 451 Seventh Street SW, Washington, DC 20410.

16 The analysis is based on HMDA data on home purchase loans originated in metropolitan areas between 1992 and 1995.

Home | Table of Contents
Summary | National Data | Regional Activity | Historical Data | Subscription Form