Summary

Housing production and sales continued at a strong pace in the third quarter of 1997. All the relevant measures revealed high levels of activity, and all geographical regions reported strong or improving housing markets, even regions that had previously lagged behind. The favorable conditions of recent years culminated in the third quarter with the attainment of an all-time high homeownership rate of 66.0 percent, breaking a 17-year-old record.

Housing production in the third quarter compared favorably with high levels from 1996:

Housing permits, at 1,419,000 on a seasonally adjusted annual rate (SAAR), were unchanged from the past two quarters. If the rate continues into the fourth quarter, the 1997 annual rate will rival 1996's 8-year high.
Housing starts, at 1,452,000 (SAAR), were up slightly* from last quarter, and if the pace of the first three quarters continues, 1997 will be within 2 percent of 1996's 8-year high.
Shipments of manufactured homes, at 361,000 (SAAR), were near 1996's record-setting pace.

Marketing of housing in the third quarter continued at very high levels for conventionally built and existing homes, while placements of manufactured homes declined:

Sales of new single-family homes, at 808,000 (SAAR), were ahead* of those for last quarter. If the fourth quarter continues at the rates set in the first three quarters, 1997 will have the highest sales in the past 18 years.
Sales of existing homes, at 4,270,000 (SAAR), continued at record-breaking levels. If the fourth quarter continues the pace set in the first three quarters, 1997 will have the highest sales ever.
Placements of manufactured homes, at 264,000 (SAAR), were down 20 percent from last quarter and 12 percent from the same quarter last year.
Inventories of new homes at the end of the quarter -- 286,000 -- were down slightly* from the second quarter and represented 4.4 months of sales at the current pace. Inventories of existing homes, at 2,010,000, were down 9 percent and represented 5.6 months of sales at the current rate.
Builders' views of housing market activity were up overall and up for current sales, prospective sales, and buyer traffic.

Low interest rates, favorable housing affordability, and a strong economy have led to an all-time record high quarterly homeownership rate. Commitment interest rates for conventional, fixed-rate, 30-year loans averaged 7.47 percent in the third quarter, down 23 basis points from the second quarter. Affordability, an index that combines the interest rate, median home price, and family incomes, was slightly ahead of last quarter, with a family earning the median income having 126.6 percent of the income needed to purchase a median-priced existing home. The U.S. homeownership rate reached a historic high of 66.0 percent in the third quarter, 0.3 percentage point* ahead of the second-quarter rate. Homeownership has increased or remained stable in 12 of the past 13 quarters, reversing the decline that occurred from 1980 to 1985.

The picture for multifamily housing (5+ units) was mixed during the third quarter, with permits up slightly, absorptions up, vacancies unchanged, and starts down:

Permits were issued for 308,000 multifamily units on a seasonally adjusted annual basis, up 2 percent* from the last quarter and 5 percent from the third quarter of 1996. If the pace of the first three quarters continues, 1997 will be the highest year since 1989.
Starts were begun on 276,000 multifamily units (SAAR), down 6 percent* from last quarter. However, if the pace of the first three quarters continues, 1997 could become the best year since 1989.
Absorptions (within 90 days) rose to 78 percent of the 44,000 apartment units completed in the second quarter. The increase in apartments completed was 12 percent, while the increase in the absorption rate was 6 percentage points.
Vacancies were 7.9 percent of the rental stock in the third quarter, unchanged from the second quarter of 1997 but down* from the third quarter of 1996.

*This change is not statistically significant.

Regional Perspective

HUD's field economists reported that job market conditions remained favorable through the third quarter of 1997. All regions reported employment growth, ranging from slow improvement in the New York/New Jersey region to the highest improvement of the 1990s in the Mid-Atlantic region. The Pacific region rebounded sharply, reflecting the improvement in California, and the Northwest region continued to benefit from the activity in aerospace and high-technology industries.

Homebuilding, as measured by single-family building permit activity, was down modestly throughout much of the Nation. The New York/New Jersey and Pacific regions were the only regions to record increases during the first three quarters of 1997.

Sales of both new and existing homes, however, have remained very strong. The Washington, D.C., metropolitan area is expected to have the best year for home sales of the 1990s. The Midwest region should have one of the best years of the past 15 years. Denver area existing home sales are poised to set a record in 1997.

Sales housing markets in Orange County, San Diego, and the San Francisco Bay Area had large production increases during the first three quarters of the year. Las Vegas remains a very strong market, and Phoenix may set another record in 1997.

Rental housing market conditions remained strong throughout the country. New England is on track to have its most productive year for apartment construction since 1990, particularly in the Boston area and in southern New Hampshire. Manhattan's rental market is very tight, although almost 8,000 new rental units will be completed in 1998 and 1999.

More than 78,465 multifamily units were permitted in the Southeast during the first 9 months of 1997. Construction was up significantly in almost every major market in Florida, where strong demand attracted large amounts of investment capital. Midwest rental housing markets reported apartment occupancy in the 93- to 96-percent range, and construction activity continued at a healthy pace.

The Southwest apartment boom continued, with all States reporting increased building permit activity during the first 9 months of 1997. Texas recorded the largest percentage increase, 40 percent, with more than 32,700 units. The Dallas-Fort Worth area led the way with a 57-percent increase and what may prove to be the best year of the 1990s. In the Rocky Mountain region, Denver may also have its best year of the 1990s.

Multifamily housing permit activity in California was up 39 percent. Seattle is expected to match or slightly exceed last year's level, but apartment production is still below the level needed to meet projected demand.

THE 1996 HMDA DATA:
A CLOSER LOOK

This article discusses the importance of separating out lending for manufactured housing (mobile homes) and "B&C" (subprime) lending when analyzing Home Mortgage Disclosure Act (HMDA) data.¹ It examines conventional home purchase mortgage applications but does not address refinance applications. In the past researchers looking at conventional lending patterns did not attempt to separate manufactured home and B&C applications from all home purchase applications. But the rapid increase in the number of manufactured home and B&C applications reported in the HMDA data, combined with their unique characteristics, make separate analysis advisable. For example, this article shows that the recent increase in mortgage rejection rates can be largely explained by the greater share of these loans in the HMDA database. This distinction does not change the fact that rejection rates are substantially higher for minorities.

The article discusses the HMDA database and its coverage of manufactured home and B&C loans. It discusses the growth in HMDA reporting of manufactured home and B&C applications, noting that reporting of these applications has increased by more than 250 percent since 1993. It examines the effect of manufactured home and B&C applications on HMDA rejection rates. Rejection rates for all racial and ethnic groups decrease significantly after controlling for manufactured home and B&C applications. For example, the rejection rate for black borrowers falls from 49 percent to 25 percent after controlling for these loans, while the rejection rate for white borrowers falls from 24 percent to 11 percent. The article also examines the borrower and locational characteristics of manufactured home and B&C lending, comparing these loans with conventional prime and FHA loans.

HMDA Data

HMDA requires most lenders who originate loans in metropolitan areas to submit information on the applicant (for example, income and race) and on the census tract location of the property. In addition to this basic information, lenders must report whether the loan was approved, rejected, or withdrawn by the applicant. The lender is not required to report on whether the applicant seeks a mortgage for a manufactured home or whether the application is considered subprime.²

Researchers and policymakers have used HMDA data to examine many mortgage lending questions. Many of these studies have dealt with neighborhood redlining and the persistent disparity between minority and white rejection rates. Others have used HMDA data to study the affordable lending behavior of portfolio lenders and Government-sponsored enterprises (Fannie Mae and Freddie Mac). The absence of a manufactured home variable and a B&C variable has typically resulted in all applications being grouped together in these studies.

To separate out manufactured home and B&C loans, HUD used trade publications to compile a list of large lenders that specialize in manufactured home and B&C lending. HUD also identified some specialized lenders by finding high rejection rates and then confirming with these lenders that they mainly originate manufactured home or B&C loans.³ HUD was unable to separate out the manufactured home and B&C loans of lenders that do not specialize in these loans.⁴

HMDA's coverage of manufactured home and B&C loans is hard to estimate given the limited information on these loans. Judging from trade publications that report manufactured home and B&C lending volume, HMDA's coverage of these loans is improving, especially for manufactured home loans. A few large firms dominate the manufactured home loan market, and each of these firms reports HMDA data. The B&C market, however, is composed of many lenders, each with a small share of the overall B&C market. Many of these lenders do not report HMDA data.

Growth in Manufactured Home and B&C Reporting

Manufactured home loan applications reported to HMDA increased for all racial and ethnic groups between 1993 and 1996. Table 1 reports that manufactured home loan applications increased from 18,801 to 72,494 applications (a 286-percent increase) for Hispanic borrowers; increased from 42,585 to 155,078 applications (a 264-percent increase) for black borrowers; and increased from 368,620 to 964,402 applications (a 162-percent increase) for white borrowers.

This trend is also occurring in the B&C market. Table 1 reports that the number of applications for B&C loans increased from 34,146 to 449,130 between 1993 and 1996 (a 1,215-percent increase). In 1993 B&C applications accounted for 1 percent of conventional HMDA applications, and in 1996 this percentage increased to 8.7 percent of applications.⁵

Black borrowers account for a disproportionately large share of manufactured home and B&C loan applications. In 1996 black borrowers accounted for 7.5 percent of all HMDA applications but received 12.1 percent of all manufactured home loans and 13.6 percent of all B&C loans. Hispanic and white borrowers have a proportional share of manufactured home loan applications. Hispanic borrowers have a disproportionately high share of B&C loan applications, while white borrowers have a disproportionately low share. Hispanic borrowers applied for 5.6 percent of all applications, 5.7 percent of manufactured home loan applications, and 8.4 percent of B&C loan applications. White borrowers applied for 75.7 percent of all applications, 75.5 percent of manufactured home loan applications, and 61.5 percent of B&C loan applications.

The share of manufactured home and B&C applications, as identified by HUD, grew from 14.4 percent of all home purchase applications reported in the 1993 HMDA data to 33.4 percent in 1996. This remarkable growth was due to a number of factors:

B&C lending increased because there was an increasing number of borrowers with impaired credit histories, and lenders and investors were seeking higher yielding assets.⁶
Manufactured housing is growing in importance. Between 1993 and 1996, the number of mobile home placements rose from 286,000 to 320,000. In The State of the Nation's Housing 1997, the Harvard University Joint Center for Housing Studies concluded that the popularity of manufactured housing in the past 2 years contributed to higher homeownership rates because "manufactured homes are now much more likely to be good-quality, multi-section units that are permanently sited on individual lots or in planned communities."⁷
Large specialized lenders are gaining a greater share of the manufactured home and B&C markets at the expense of small specialized lenders. Trade publications indicate that consolidations and acquisitions have been an important trend in recent years.
Specialized lenders who report HMDA data are gaining a greater share of the manufactured home and B&C markets at the expense of small specialized lenders who do report HMDA data.

HUD is unable to determine the relative importance of these factors. If the last factor is important, Table 1 overestimates the growth in manufactured home and B&C applications.

Table 1. Manufactured Home and B&C Conventional Home Purchase Applications from 1993 to 1996

Table

Mortgage Rejection Rates

Mortgage rejection rates for manufactured home and B&C loans are much higher than the mortgage rejection rate for prime loans. In 1996 the 55.1-percent rejection rate for manufactured home loan applications and the 54.2-percent rejection rate for B&C loan applications were more than four times the 12.7-percent rejection rate for conventional prime loan applications and more than five times the 10.0-percent rejection rate for Federal Housing Administration (FHA) loan applications.

Table 2 reports the 1996 rejection rates for manufactured home, B&C, conventional prime, and FHA loans by racial and ethnic group. After controlling for manufactured home and B&C loan applications, the rejection rate for conventional applications for all racial and ethnic groups fell dramatically. The rejection rate for black applications fell from 48.7 percent to 25.3 percent; the rejection rate for Hispanic applications fell from 34.4 percent to 19.6 percent; and the rejection rate for white applications fell from 24.1 percent to 10.9 percent.

The differential in rejection rates between minority and white applicants is highest for black borrowers for all loan products. The differential between the black and white rejection rates is 16.1 percentage points for manufactured home loan applications; 11.1 percentage points for B&C loan applications; 14.4 percentage points for conventional prime loan applications; and 7.1 percentage points for FHA loan applications.

After controlling for manufactured home and B&C loan applications, the trend in conventional prime rejection rates becomes clearer. Rejection rates for all racial and ethnic groups are at their 1993 levels after a 2-year period of moderately lower rates. For example, the conventional prime loan rejection rate for black borrowers was 24.9 percent in 1993 and 25.3 percent in 1996; the rejection rate for Hispanic borrowers was 21.4 percent in 1993 and 19.6 percent in 1996; and the rejection rate for white borrowers was 10.9 percent in both 1993 and 1996.

Table 2. 1996 Denial Rates by Racial and Ethnic Group

Table

Table 3. Share of Originations by Loan Product and Borrower and Neighborhood Characteristics

¹For example, 22.9 percent of all loans originated for American Indians are manufactured home loans.

Affordability Characteristics

Manufactured home loans, B&C loans, and FHA loans are important sources of affordable lending for minority and low-income borrowers and their neighborhoods. These markets serve borrowers that do not meet the underwriting standards of the conventional prime market. In 1996 manufactured home, B&C, and FHA loans combined accounted for 29.6 percent of all home purchase loans. Manufactured home and B&C loans accounted for 8.7 percent and 2.7 percent of home purchase loans, respectively, and FHA accounted for 18.2 percent of home purchase loans. Table 3 reports the shares of 1996 originations by loan product and borrower and neighborhood characteristics.

Borrower Race. Black borrowers rely more on manufactured home and B&C loans than Hispanic or white borrowers. Black borrowers are 1.3 times more likely to have a manufactured home loan and 2.0 times more likely to have a B&C loan than white borrowers. Hispanic borrowers are 0.7 times less likely to have a manufactured home loan and 1.4 times more likely to have a B&C loan than white borrowers.

Blacks and Hispanics rely heavily on FHA loans, which account for 38.2 percent of black loans and 42.0 percent of Hispanic loans. The combined share of manufactured home, B&C, and FHA loans is 54.5 percent for black borrowers and 51.3 percent for Hispanic borrowers. The conventional prime market provides less than 50.0 percent of home purchase loans to black borrowers and to Hispanic borrowers.

Borrower Income. Low-income borrowers are more likely to have a manufactured home than moderate- or high-income borrowers;⁸ they are 4 times more likely to have a manufactured home loan than high-income borrowers. The share of loans accounted for by B&C loans does not vary significantly by borrower income. B&C loans account for between 2 and 3 percent of loans originated for low-, moderate-, and high-income borrowers. This result may be indicative of the sample of B&C lenders that report under HMDA. However, there is evidence that B&C loans are originated primarily for borrowers with impaired credit histories, independent of their income level.⁹

Neighborhood Characteristics. Manufactured home, B&C, and FHA loans account for a larger share of loans for properties located in low-income and minority neighborhoods.¹⁰ A borrower in a low-income neighborhood is 3.3 times as likely to have a manufactured home loan, 1.8 times as likely to have a B&C loan, and 2.4 times as likely to have an FHA loan than a borrower in a high-income neighborhood. Manufactured home loans account for 8.6 percent, B&C loans account for 3.8 percent, and FHA loans account for 30.2 percent of loans originated in low-income neighborhoods. A borrower in a minority neighborhood is 1.1 times as likely to have a manufactured home loan, 1.9 times as likely to have a B&C loan, and 1.8 times as likely to have an FHA loan than a borrower in a nonminority neighborhood. Manufactured home loans account for 6.3 percent, B&C loans account for 4.0 percent, and FHA loans account for 32.0 percent of loans originated in minority neighborhoods.

Conclusion

Manufactured home loans, B&C loans, and FHA loans are important sources of affordable lending for minority and low-income borrowers and their neighborhoods. These markets serve borrowers that, for the most part, do not meet the underwriting standards of the conventional prime market. Analyses of HMDA data that recognize the importance of manufactured home, B&C, and FHA loans in affordable lending lead to more focused understanding of mortgage market trends and more focused policy implications for increasing homeownership. The main points of these analyses are:

The number of manufactured home and B&C applications that HUD can identify grew 250 percent from 1993 to 1996.
The difference in rejection rates between conventional prime and manufactured home or B&C loans was high. The conventional rejection rate fell from 26.8 percent to 12.7 percent after controlling for these loans. The FHA rejection rate was 10.0 percent. Clearly, the manufactured home and B&C markets are made of very different borrowers than is the conventional prime market.
Black and Hispanic borrowers are more likely to have a manufactured home, B&C, or FHA loan than white borrowers. Low-income borrowers are more likely to have a manufactured home or FHA loan than moderate- and high-income borrowers, but B&C loans account for approximately the same percentage of loans at all borrower income levels.

Notes

1 Subprime loans include a mix of loans, most of which are characterized by imperfection in the borrower's credit or have terms that do not meet the conforming underwriting standards of Fannie Mae or Freddie Mac. For a more complete discussion of the role of manufactured home and B&C applications in HMDA, see Scheessele, Randall M., 1997, "The Impact of Manufactured Home and Subprime Loans on HMDA Rejection and Origination Rates," working paper, Department of Housing and Urban Development, November.

2 HMDA data are the most comprehensive source of annual information on primary mortgage market originations and secondary market loan purchases. For a good overview of the limitations of HMDA data, see Bunce, Harold L. and Randall M. Scheessele, 1996, "The GSEs' Funding of Affordable Loans," Working Paper Series HF-001, Department of Housing and Urban Development, December.

3 For example, Greentree Financial Corporation and Bank of America, FSB, are the largest manufactured home lenders that report home purchase applications under HMDA. Ford Consumer Finance Company and Access Financial Lending Corporation are the largest B&C lenders that report home purchase applications under HMDA.

4 For example, Countrywide Funding Corporation, The Chase Manhattan Bank, and Norwest have entered the B&C market.

5 The B&C share would be higher if refinance applications were included because refinance loans comprise the majority of B&C lender activity. Martin Wahl and Craig Focardi (1997, "The Stampede to Subprime," Mortgage Banking October: 26-36) estimate that there were $90 billion in B&C loans in 1996. The HMDA data report approximately $25 billion for loans originated by lenders that specialize in B&C lending. Possible explanations for the difference between HMDA and Wahl and Focardi's estimate are that Wahl and Focardi's number may include loans that are not reported under HMDA; the HMDA number does not include B&C loans originated by prime lenders; and not all B&C lenders report under HMDA.

6 See Wahl and Focardi (1997).

7 See Harvard University Joint Center for Housing Studies, 1997, The State of the Nation's Housing 1997; and Fields, Ruth G., 1996, "The Mortgage Sleeper is Waking Up in Manufactured Homes," Secondary Marketing Executive October: 32-39.

8 A low-income borrower has an income that is no more than 80 percent of area HUD median family income. Moderate income is defined as between 80 and 120 percent of area HUD median family income, and high income is defined as more than 120 percent of area HUD median family income.

9 For example, see "Low to Moderate Income and High Minority Area Case Studies," Fair Isaac and Company, Inc, discussion paper (October 4, 1996).

10 A low-income neighborhood is a neighborhood where tract median family income is no more than 80 percent of area median family income. Moderate income is defined as between 80 and 120 percent of area median income, and high income is defined as more than 120 percent of area median family income. A minority neighborhood is a tract where minorities comprise more than 30 percent of the tract population.