Reports - Nationwide Summary Statistics for
2003 HMDA Data
Fact Sheet (July 2004)

The following nationwide summary statistics are based on data compiled by the Federal Financial Institutions Examination Council (FFIEC) for institutions covered by the Home Mortgage Disclosure Act. 1

The Data

For 2003, the FFIEC prepared 65,808 disclosure statements for 8,121 lenders, with a separate statement for each metropolitan statistical area in which a lender had an office or could be deemed to have an office under Regulation C, based on the number of applications and loans (table 1). The data reflect the lending activity of 4,050 commercial banks; 638 savings associations; 1,903 credit unions; and 1,530 mortgage companies (of which 1,186 were independent mortgage companies).

Applications Received and Loans Made

In 2003, lenders covered by HMDA reported a total of about 42 million loans and applications that did not result in an origination (table 1). The total volume of reported home loan applications and purchases increased in 2003 by 33 percent from 2002, primarily due to a significant increase (41 percent) in refinancing activity. The 2003 data show that lenders acted on approximately 34 million applications for home purchase loans, home improvement loans, and refinancings (compared with approximately 26 million in 2002), and they purchased about 7 million loans (compared with 5 million in 2002). The number of applications for home purchase loans in 2003 increased from 2002 levels by 10 percent. Requests for refinancings accounted for 72 percent of all home loan applications (data not shown in tables). This proportion is about 9 percent higher than in 2002, when applications for refinancings accounted for about 66 percent of all applications. The increased volume and share of applications for refinancings resulted primarily from a sharp drop in interest rates on mortgage loans.

Loan Programs and Changes in Lending Volume by Race and Income

Lending institutions tend to specialize in different types of home loans. For example, in the case of home purchase loans, mortgage companies tend to do most of the government-backed lending, including FHA and VA loans, accounting for about 83 percent of all such originations in 2003. Depository institutions, particularly commercial banks, do most of the home improvement (84 percent) and multifamily lending (90 percent). (Data not shown in tables.)

Lower-income households rely more heavily on government-backed loans than do higher-income households (table 2). The low downpayment requirements and the ability to finance closing costs make FHA and VA loans particularly attractive to lower-income households and first-time homebuyers, who are likely to have fewer financial resources than other homebuyers. Dollar limits on the amount of FHA loan insurance or VA loan guarantees make these government-backed loans unavailable or less attractive to households seeking to buy more expensive properties.

Applications for different types of home purchase loans vary across racial and ethnic groups and income categories (table 2). For example, in 2003, 19 percent of Hispanic applicants and 21 percent of black applicants for home purchase loans applied for government-backed mortgages; the comparable rates for Asians, whites, and Native Americans were 4 percent, 12 percent, and 15 percent, respectively. Moreover, 21 percent of home purchase loan applicants with incomes less than 80 percent of the median family income for their metropolitan statistical areas applied for government-backed loans; by contrast, 5 percent of applicants with incomes of at least 120 percent of the metropolitan statistical area median applied for such loans (derived from table 2, government-backed memo item). In addition, loan applicants seeking to buy a home in low- or moderate-income neighborhoods were more likely to apply for government-backed loans than those seeking to buy homes in upper-income neighborhoods. In central cities and neighborhoods with greater proportions of minority residents, applications for government-backed loans accounted for a higher share of all home purchase loan applications, than did such applications in non-central cities and neighborhoods with smaller proportions of minority residents.

For conventional and government-backed home purchase lending, lending to Asians, Hispanics, blacks, and whites was up about 16 percent, 18 percent, 15 percent, and 11 percent, respectively, from 2002 to 2003; Native Americans experienced a 5 percent decline in such lending from 2002 to 2003 (table 7).

Home purchase lending increased for all income categories from 2002 to 2003. Applicants with incomes 120 percent or more above the median for the metropolitan statistical area experienced the greatest expansion, 13 percent; over the same period, home purchase lending increased by 10 percent for applicants with incomes 100-119 percent of the median, 8 percent for applicants with incomes 80-99 percent of the median, and 6 percent for applicants with incomes less than 80 percent of the median (table 7).

Denial Rates

The overall denial rate for conventional home purchase loans in 2003 was 14 percent, a rate unchanged from 2002, but well below the rate in 2001 (21 percent). A large portion of this decline can be attributed to a significant drop in 2002 and in 2003 of the number of applications for manufactured home loans, which have very high denial rates.2 (Data not shown in tables). In 2002, the roughly two-dozen reporters categorized by HUD as manufactured home loan specialists denied about 60 percent of all the applications they received for conventional home purchase loans, and in 2003, they denied about 61 percent of those applications. This proportion represents a marginal decline from earlier years, when these lenders denied about two-thirds of all their applicants. In 2002 and in 2003, there was a sharp contraction from 2001 in reported manufactured home loan activity. (Data not shown in tables.) The number of applications for conventional home purchase loans submitted to these lenders in 2002 dropped 65 percent from 2001, and in 2003 fell 37 percent from 2002, reflecting difficulties in the manufactured housing market. If the manufactured home purchase loan applications and denials reported by these lenders are excluded from both the 2002 and 2003 data, the denial rates for conventional home purchase loan applications would have been 11.5 percent in 2002 (instead of 14 percent) and 12.4 percent in 2003 (instead of 14 percent).

The HMDA data show that denial rates for home purchase loan applications vary with income. For example, in 2003 the denial rate for conventional home purchase loans for lower-income applicants was 19 percent as compared with 9 percent for higher-income applicants (derived from table 3).

Denial rates for conventional home purchase loans continue to vary among applicants by race or ethnicity and by income. In 2003, 24 percent of black applicants, 24 percent of Native American applicants, 18 percent of Hispanic applicants, 12 percent of white applicants, and 11 percent of Asian applicants were denied conventional home purchase loans (table 3). These rates of denial are little changed from 2002: denial rates for conventional home purchase loans in 2002 were 26 percent for blacks, 23 percent for Native Americans, 18 percent for Hispanics, 12 percent for whites, and 10 percent for Asians. The denial rates from 2002 to 2003 follow the general pattern of declining denial rates since 1998. This pattern is in contrast to the experience over the longer period from 1993 to 1998, when denial rates for all ethnic and racial groups increased substantially.

Differences in the income levels of the racial or ethnic groups account for some of the differences among them in denial rates for 2003. However, other factors are more important given that for all income groups, white and Asian applicants experienced lower rates of denial than Native American, black, or Hispanic applica01/15/2009 11:04 AM to which racial discrimination may account for remaining differences in denial rates across racial and ethnic lines cannot be determined solely from the HMDA data reported by lenders.

Many lenders report reasons for denial of loan applications; in 2003, as in prior years, the reason most frequently cited for the denial of a single-family home loan application, regardless of the applicant's race or ethnic status, was poor or no credit history (data not shown in tables). This factor was cited in 36 percent of the denials for Native Americans, 33 percent of the denials for blacks, 30 percent of the denials for whites, 28 percent of the denials for Hispanics, and in 19 percent of the denials for Asians.

The Impact of the 2000 Decennial Census on assessing changes in lending by census tract characteristics, 2002 to 2003

Changes reflected in the 2000 Decennial Census complicate 2002 to 2003 comparisons of lending in census tracts grouped by racial or income characteristics. In the 2000 Decennial Census, many census tracts experienced changes in physical boundaries and in demographic characteristics. Lenders were required to use these changed census tract boundaries to report property location for loans and applications for the first time in 2003. In 2002, lenders were required to use the 1990 Decennial Census tract boundaries and demographics.

Between the 1990 and the 2000 Decennial Census, substantial changes occurred in neighborhood demographic characteristics (table 8, memo item). In 1990, 43 percent of all census tracts had a population that was less than 10 percent minority (derived from data shown in the 1990 column); in 2000, only 28 percent of the census tracts had a population that was less than 10 percent minority (derived from data shown in 2000 column). Moreover, the number of census tracts that were over 80 percent minority increased by about 50 percent from 1990 to 2000. Changes in the distribution of census tracts by relative median family income were much less significant.

To facilitate comparison of the 2003 data with previous years' data, the 2003 data were adjusted by assigning to the 2000 census tract boundaries their 1990 census tract demographic characteristics (table 8). The adjusted 2003 data show that from 2002 to 2003, census tracts with 80 to 100 percent minority population experienced the greatest increase in home purchase lending, 15 percent; such lending increased 9 percent for census tracts with less than 10 percent minority population; 8 percent for tracts with 10-19 percent minority population and 8 percent for tracts with 20-49 percent minority population; and 10 percent for tracts with 50-79 percent minority population.

Home purchase lending also varies by the income level of census tracts (table 8). Adjusted 2003 data show that low- and moderate-income census tracts taken together experienced the largest increase, 16 percent, in home purchase lending. Such lending for middle- and upper-income census tracts increased by 9 percent, respectively, from 2002 to 2003, according to the adjusted 2003 data.

Missing Information on Race and Ethnicity

Until recently, lenders were not required to collect information on an applicant's race or ethnicity and sex for applications taken entirely by telephone. From 1993 to 2001, the incidence of applications of all types lacking race or ethnicity data grew from 8 percent to 30 percent. However, as of January 1, 2003, lenders are required to ask applicants for race, ethnicity, and sex information in telephone applications. For applications taken in person or by mail or electronic means (such as by facsimile or the Internet), a lender must request the information. In all cases, an applicant has the option not to provide the information.

The 2003 HMDA data suggest that the change in reporting requirements may have reduced the incidence of applications and loans reported without data on race or ethnicity. The proportion of applications of all types lacking race or ethnicity data fell to 17 percent in 2003, down from 28 percent in 2002 (table 9). For home purchase loan applications, the proportion of applications missing race or ethnicity data is lower than the proportion for other application types, but follows the same general pattern. Applications for home purchase loans lacking information on race or ethnicity grew from about 4 percent in 1993 to about 18 percent in 2001, involving roughly 1.4 million loan applications in 2001. In 2002, the percentage of such applications fell to 15 percent, but involved roughly 1.1 million applications; in 2003, the percentage fell to 12 percent and involved roughly 1 million applications. The same pattern is observed for home purchase loans originated. In 1993, 3 percent of such loans lacked information about race or ethnicity; by 2001, 13 percent lacked this information. In 2002, the proportion of home purchase loans missing the information fell to 12 percent, and in 2003 it fell to 10 percent.


Tables 1 through 9

Tables are in Portable Document Format (PDF).


1.The FFIEC has also compiled insurance data submitted by seven major private mortgage insurance (PMI) companies under the auspices of the Mortgage Insurance Companies of America. These data, which relate to application decisions made by the PMI companies during 2003, show about 2.9 million applications for PMI (about 1.5 million for home purchase loans, and 1.4 million for refinancings). These data are available from the individual companies, and-in the same formats as the HMDA data-at the central depositories and from the FFIEC.

2. These calculations are based on the list of lenders identified by HUD as manufactured home lending specialists in 2002, which was the most recent list available at the time of this publication.