2007 American Community Survey: A Comparison to Selected Housing and Financial Characteristics for the United States from the 2007 American Housing Survey

Mary B. Schwartz

The American Community Survey (ACS) is a new approach to collecting reliable, timely information needed by local communities. It collects much of the same data previously collected in the decennial census sample and is a critical element in the U.S. Census Bureau'€™s 2010 Decennial Census Program. The ACS is a major innovative step designed to meet the nation’s need for the kind of in formation that has previously been available only once every ten years from the decennial census sample.

The ACS was fully implemented in 2005. It is the largest household survey in the United States, with an initial sample size of about 3 million housing unit addresses throughout the country. Release of annual estimates for the fully implemented sample from the ACS began in 2006 for geographic areas with populations of 65,000 or more. In 2006, group quarters, although not relevant to this comparison report as the estimates compared here are for households only, were added to the ACS sample. In 2008, the ACS began producing estimates for all geographic areas and subpopulations of at least 20,000. In 2010, release of 5-year period estimates will start for all areas down to block groups. All estimates, including 3-year and 5-year period estimates, will be updated every year.

Following the 1970 census, the Subcommittee on Construction Statistics of the Cabinet Committee on Construction recommended that the Department of Housing and Urban Development (HUD) sponsor a yearly housing survey, to be conducted by the Bureau of the Census. In 1972, HUD received money that paved the way for the Annual Housing Survey. The Annual Housing Survey consisted of two parts: a national sample of housing units from urban and rural areas to be examined every year, and metropolitan area samples from 60 selected Standard Metropolitan Statistical Areas (SMSAs), including the largest and many of the smaller, fast growing, with one-third of them to be examined in detail every third year.

The Census Bureau began collecting the national data in 1973 and continued annually until 1981. Beginning in 1982, because of budget restrictions at HUD, the national sample was interviewed only in odd numbered years and the survey was appropriately renamed the American Housing Survey (AHS) in 1984. It is still this way today. Over time, the sample size has remained relatively stable; and in 2007 out of the approximately 60,000 housing units sampled, 52,850 were eligible to be interviewed. The Census Bureau began collecting metropolitan area data in this survey in 1974 and continued data collection annually through 1996. Beginning in 1997, because of budget restrictions at HUD, the metropolitan sample was interviewed only in even-numbered years. In a cost containment measure in 1985, the number of metropolitan areas was reduced to 44 (interviewed on a rotating basis every six years) and the sample sizes were reduced from 5,000 (smaller metropolitan areas) and 15,000 (larger metropolitan areas) to around 4,250 for all metropolitan areas. In 1999, in another cost containment measure, the AHS began producing estimates for six metropolitan areas for which data were particularly robust in the national sample in every other year in which the national survey was done, that is, in every other odd-numbered year. Thus, the national sample was supplemented by several thousand cases - just enough to produce re liable estimates for New York, Northern New Jersey, Philadelphia, Chicago, Detroit, and Los Angeles in 1999 and 2003. For example, in 2003, the national sample was supplemented by 7,700 cases (for a total size of 63,400 sampled housing units) in order to produce reliable estimates for the six areas.

The year 2006 marked another landmark in the metropolitan areas produced by the AHS. The metropolitan areas would be interviewed on the same schedule as the national sample, so 2004 marked the last even numbered year for any AHS estimates. Also, the number of metropolitan areas surveyed in 2007 was reduced to seven: Baltimore, MD; Boston, MA; Houston, TX; Miami, FL; Minneapolis, MN; Tampa, FL; and Washington, DC.

The purpose for taking each of the surveys is different. The ACS is a national survey that collects basic demographic, socioeconomic, and housing information that has traditionally been collected once every ten years in the decennial census. The ACS collects this information on a continuous basis throughout the decade and provides it at various levels of geography depending on population size. Additionally, the ACS draws a new sample every year; therefore, the ACS is not a longitudinal survey measuring changes to the housing units and the households living in them over time.

Federal agencies rely on the data to administer and evaluate government programs. For example, HUD uses the ACS data to set income limits to determine whether a household is eligible for housing assistance under various housing programs. HUD also uses data on rents to set the maximum that a landlord ca n charge in order to participate in low- income housing programs, like Section 8. The US Department of Veterans Affairs uses ACS data on characteristics of veterans to evaluate the need for educational, employment, and health care programs for those who served in the military. Local governments'€™ planning for new roads, hospitals, schools, senior centers, and affordable housing depend on the timely information provided by the ACS.

The AHS is designed specifically to address housing policy issues, like housing affordability or discrimination in housing and mortgage markets, and differences in the effectiveness in housing and mortgage delivery systems based on race, age, household type, or so forth. The AHS collects data on the Nation's housing, including apartments, single-family homes, mobile homes, vacant housing units, household characteristics, income, housing and neighborhood quality, housing costs, equipment and fuels, size of housing unit, and recent movers. Also, the AHS is a longitudinal survey measuring the changes to particular housing units over time. In fact, many of the homes in the AHS have been in sample since 1985, although new construction is added to the sample with each survey.

Since the ACS was designed to look at the cross section of the population and how we are housed, the ACS provides reliable estimates for 35 housing-related variables for nearly all geographic areas and subpopulations as small as 20,000 in 2007 (and even smaller in 2010). In contrast, the AHS provides estimates for a nearly 500 housing- 3 related variables but only for larger areas. The smallest areas for which data are available are "zones" (population of 100,000 or more) within select, and increasingly fewer, metropolitan areas.

The two surveys attempt to measure many of the same concepts and characteristics. Indeed, interest in the user community for comparing results from these surveys has been growing. This report will attempt to address many of the data users’ concerns. However, differences in emphasis, sample de sign, collection techniques, and processing methods described later, may make the comparison of results from these two surveys more difficult to explain.

While the AHS metropolitan area sample design provided estimates in 2007, there are only seven metro areas available. Since the ACS provides estimates for all metropolitan areas, a comparison of the estimates of these seven areas between the ACS and the AHS would be possible. However, this report is limited to a comparison of results at the national level.

November 26, 2008
April 14, 2009