The History of the AHS – 1973 - 2023

The History of the AHS – 1973 - 2023
George R. Carter III, Director, Housing and Demographic Analysis Division, Office of Policy Development and Research, HUD (2021–Present)

From the early days of the American Housing Survey, its data have been used to evaluate HUD's national housing goals.

The author would like to thank Dav Vandenbroucke, Fred Eggers, and Shawn Bucholtz for their contributions to this article. The author would also like to thank former Housing and Demographic Analysis Division directors Duane McGough, Ronald Sepanik, and Shawn Bucholtz for their work in creating and leading the survey at HUD. The author also thanks all the HUD and U.S. Census Bureau staff who have worked on the AHS over the years. Without you, the survey would not be possible.

The American Housing Survey (AHS), the most comprehensive national housing survey in the United States, is as old as HUD’s Office of Policy Development and Research (PD&R). Since 1973, the AHS has collected data on the characteristics, costs, and quality of the nation’s housing stock and demographic characteristics of households. The survey is sponsored by HUD and conducted by the U.S. Census Bureau. The AHS is a longitudinal survey of housing units, meaning that it surveys the same housing units over time. As the U.S. housing stock grows, so does the AHS sample size, with additions to account for new construction. Although the AHS collects data on households, it is not considered a longitudinal household survey, because it does not collect data on households who move out of sampled housing units. The AHS was originally dubbed the Annual Housing Survey because it was conducted annually from 1973 to 1981. Due to budget constraints, the AHS became a biennial survey conducted in odd-numbered years in 1983. AHS samples were redrawn in 1985 and 2015, and a new sample will be drawn in 2025. AHS data provide HUD with an evidence base for policymaking and facilitate the assessment of housing needs and opportunities for households across income levels, ages, and racial and ethnic groups. AHS microdata and housing statistics are publicly available, making them a valuable resource for planners, local policymakers, community stakeholders, private organizations, businesses, researchers, and anyone who wants to learn about housing at the national and subnational levels.

Before the AHS existed, the only data that the U.S. Census Bureau collected on the nation’s housing came from the decennial censuses, one-time sample surveys, and a survey of vacant housing units. A modification to Title 13 of the U.S. Code in 1939 authorized the collection of the 1940 Census of Housing, which the agency conducted along with the population census every 10 years until 2000. Over time, federal agencies realized that they needed more frequent and timely data on the nation’s housing stock. In the Housing Act of 1949, Congress found that “the supply of the Nation’s housing is not increasing rapidly enough to meet the national housing goal” of “a decent home and a suitable living environment for every American family.” Titles VIII and IX of the Civil Rights Act of 1968 (the Fair Housing Act) expanded on previous acts and prohibited discrimination in the sale, rental, and financing of housing based on race, religion, and national origin. In later years, Congress expanded the Fair Housing Act to prohibit discrimination based on sex (including gender identity and sexual orientation), familial status, and disability status.

In the early 1960s, the Housing Division at the U.S. Census Bureau proposed the idea of collecting the AHS to assess the current housing stock and create programs to further HUD’s housing goals. Duane McGough was the first Government Technical Representative assigned to the AHS and later was one of the first directors of the Housing and Demographic Analysis Division to oversee the survey. In an oral history recorded in 1993, he noted that the idea of an AHS

In late 1969 and early 1970, after HUD submitted the first annual report on the national housing goal to Congress, the Nixon administration established a Subcommittee on Construction and Housing Statistics. This subcommittee, which included all cabinet agencies and other major domestic agencies, recommended that the federal government conduct an annual survey of housing in the United States. Title V of the 1970 Housing Act authorized programs of "research, studies, testing, and demonstrations relating to the missions and programs of the Department." As a result, HUD received funding that paved the way for the agency to establish the AHS in 1972 and PD&R, the office responsible for the AHS, in 1973.

From the AHS’ early days, policymakers have used the survey’s data to evaluate HUD’s national housing goals. McGough noted that the seventh of ten annual reports on HUD’s national housing goals included a special chapter on the quality of housing in the United States and the extent of unmet needs. AHS data were used in all subsequent annual reports and in a report to Congress after the 10th report. For many years, HUD has used AHS data to produce the biennial Worst Case Housing Needs Reports to Congress, tracking trends and characteristics of unassisted, very low-income households who either live in severely inadequate housing or spend more than 50 percent of their household income on rent. The Worst Case Housing Needs Reports rely on AHS data on housing affordability and housing adequacy. The survey measures housing affordability through housing cost burden ratios. Housing adequacy measures minimal housing adequacy and draws on concepts of physical adequacy and equipment breakdowns. Before the advent of the AHS, developmental work was undertaken in a five-city survey on housing quality, which asked respondents to classify the quality of their units as good, moderate, or poor. Research by HUD, along with the Office of Management and Budget, the Congressional Budget Office, the Congressional Research Service, and the Urban Institute, refined the AHS’ housing quality questions to classify the physical adequacy of housing units as adequate, moderately inadequate, or severely inadequate. Dav Vandenbroucke, an economist who worked on the AHS for more than 28 years and worked on the Housing Affordability Data System and Worst Case Housing Needs tables, noted, “It’s frustrating that the crisis of affordable housing is just as bad as [it was] when the survey started. It is because of the AHS that we know the state of the crisis. Housing inadequacy has decreased over time, but housing affordability problems have not.”

When it began in 1973, the Annual Housing Survey had a sample size of 60,000 housing units. In 1985 after the sample was redesigned based on the 1980 decennial census, the national sample was approximately 47,000 housing units. In 2005, the national sample was enhanced with additions from the 2000 Census to improve coverage of manufactured/mobile homes as well as assisted living housing units. For the 2015 AHS, a new representative national sample consisting of approximately 85,000 housing units was drawn using the U.S. Census Bureau’s Master Address File (MAF) as the sampling frame, along with additional oversamples for selected metropolitan areas and an oversample of HUD-assisted housing units. The total sample size of the 2015 AHS panel, which is still in use today, is approximately 115,000 housing units.

In 1974, the AHS began using metropolitan area samples, which continue in the present sample design. Originally, these samples were entirely separate from the national sample and consisted of 60 cities surveyed over a 3-year cycle. After 1981, the number of metropolitan areas was cut from 60 to 44. In 1982, after the AHS became a biennial survey, metropolitan areas were surveyed in even-numbered years; starting in 2005, they were surveyed in the same year as the national survey. In the current 2015 AHS panel (2015–2023), the top 15 metropolitan areas are embedded within the national sample, and an additional 20 metropolitan areas are surveyed in alternating groups of 10 that are surveyed every 4 years. The 2023 AHS sampled five fewer metropolitan areas because of budget constraints. Throughout the history of the AHS, the number of metropolitan areas sampled has varied depending on the availability of funding. 

From 1973 to 1992, the U.S. Census Bureau administered the AHS through in-person interviews using a paper questionnaire. Experiments with computer-assisted telephone interviewing (CATI) were conducted in 1987, 1989, and 1991, and the bureau added CATI as a mode along with paper questionnaires in 1993. In 1997, HUD and the Census Bureau eliminated the paper questionnaire and replaced it with computer-assisted personal interviewing (CAPI) instruments for both the in-person and telephone interviews, and in 2009, the bureau added a Spanish version of the CAPI instrument. Since the 2011 AHS, the survey instrument has consisted of a permanent core questionnaire, collecting data on housing characteristics, costs, and quality and household demographic characteristics, and rotating topical supplements chosen by HUD each survey year. In the 2021 AHS, these topical supplements included questions about household pets; secondhand smoke; the housing search process for renters; moving intentions; wildfire risk; and delinquent rent, mortgage, and utility payments and notices. The questions about household pets, which originally were developed for a module on disaster preparedness for the 2017 AHS, are among the most popular collected in the survey. The 2023 AHS, which is being collected between May 2023 and October 2023, includes topical supplements on housing insecurity, perceptions of urbanization, health issues related to homes, power outages, extreme heat risk, first-generation homeowners, parents’ place of birth, and a split-sample test of sexual orientation and gender identity proxy questions. The 2023 AHS also expands AHS core with questions on cooling problems, disaster-related moves, sprinkler systems, and self-reported sexual orientation and gender identity. Ideas for AHS content come from HUD, external stakeholders, and the public, and the questions are developed in collaboration with questionnaire design experts at the U.S. Census Bureau.

Before 1997, AHS data were stored on tape and programmed with Fortran. Along with the switch to the CAPI instrument in 1997, programming switched to the Statistical Analysis System (SAS) format, and the public could access the data files on CD-ROM. Shawn Bucholtz, director of the Housing and Demographic Analysis Division from 2010 to 2020, championed the effort to make all AHS data available in both SAS and ASCII formats. AHS data are available through the U.S. Census Bureau for download. Researchers who want to access the most recent AHS microdata should consult the publication “Getting Started with the PUF: 2015 and Beyond.” Until the 2009 AHS, housing and household characteristics were also available in the form of a printed report and summary tables. Estimates for the 2011 to 2021 AHS are available in the AHS Table Creator.

The AHS is a longitudinal survey of housing units, which makes it possible to analyze changes in housing units over time and changes in the housing stock over time. Components of Inventory Change reports have tracked changes in the housing inventory across adjacent survey years, documenting additions and subtractions from the housing stock and the reasons for those changes. Research by John Weicher, the Assistant Secretary of PD&R from 1989 to 1993; Fred Eggers, former Deputy Assistant Secretary for Economic Affairs in PD&R; and Fouad Moumen used the entire 1985 AHS panel, covering the period from 1985 to 2013, to examine the extent to which housing affordability changes in housing units over time. Other research by Eggers and Alexander Thackeray, although not longitudinal, used the rich housing data collected in the AHS to analyze housing trends between 1973 and 2005.

The AHS has long been an innovative survey. Over the years, HUD and U.S. Census Bureau staff have adapted the survey’s design to ensure that it collects valid and reliable housing data to inform housing policy. Between 1984 and 1997, when respondents did not know the square footage of their housing unit, the interviewer measured the outside dimensions of the housing units using tape measures (see figure 1). As the telephone survey became a more prevalent mode for data collection, interviewer-recorded neighborhood observations and square-footage measurements were removed from the survey, and the survey began relying more on self-reported information from respondents. Other elements of the sample design, such as neighbor clusters, in which several adjacent neighbors were selected into the sample, were removed because HUD and external researchers used them so infrequently.  The AHS has increasingly limited the neighborhood data it contains because of the risk it poses for household reidentification. The risk of disclosure, mostly because of the availability of other publicly available data sources that researchers can merge with the AHS, have led HUD and Census to suppress data that once was released on the AHS public use file. However, because geospatial data are increasingly available from ACS and other sources, researchers with access to the AHS internal use file can link neighborhood characteristics to AHS sample cases to conduct neighborhood analyses. HUD-defined submetro area “zones” once were included as a geographic variable on the Metro public use files, but they also were discontinued over concerns about household reidentification. Paul Burke, an analyst in the Housing and Demographic Analysis Division under Duane McGough, was instrumental in the work on zones — work that later was advanced by Dav Vandenbroucke. At one point, analysts adjusted Fair Market Rents between decennial years with AHS data, although that changed with the development of the American Community Survey. The AHS is constantly evolving to meet current data needs while accounting for data collection and data release constraints.

To this end, throughout the history of the survey, HUD has sought the input of experts within and beyond HUD to ensure that the AHS remains on the cutting edge of both housing and survey research. As part of the study “Rebuilding the Research Capacity at HUD,” a panel of leading housing researchers convened by the National Academy of Sciences (NAS) evaluated the survey’s usefulness and made several recommendations. The panel concluded that the survey was an important source of information on housing costs and conditions, and strongly recommended the additional funding for the survey, particular funding for additional metropolitan samples. The recommendations in this report, along with those in the PD&R report “AHS: Streamlining the American Housing Survey,” motivated and influenced the 2015 redesign of the AHS, which Shawn Bucholtz led. The NAS report influenced the incorporation of a HUD-assisted oversample into the AHS and the matching of AHS survey cases to HUD administrative records to flag cases as HUD-assisted on the public use file and Internal Use File. The report also influenced AHS’ rotating topical module design. The matching work with the AHS led to other HUD-assisted matching work with the ACS and the U.S. Census Bureau’s Household Pulse Survey. Research matching AHS housing units to property tax data resulted in improvements to the imputations of lot size and year built variables. 

In recent years, work on the AHS has tackled the measurement of evictions and housing insecurity and has provided data to provide evidence for policymaking during and after the COVID-19 pandemic. Inspired by the work of Matthew Desmond and colleagues, HUD and the U.S. Census Bureau incorporated an eviction module in the 2017 AHS. Research on the evaluation of those data was published in a recent issue of Cityscape. Research on these data suggested that the AHS was not the best vehicle to capture evictions and informed the development of forced moves questions on the U.S. Census Bureau’s Household Pulse Survey. Work on housing insecurity by the Housing as a Platform Knowledge Collaborative in PD&R, which started in 2016 and was influenced by the U.S. Department of Agriculture’s Food Security Index, inspired and motivated the development and inclusion of the Housing Insecurity Research Module Follow-On survey in 2019, subsequent research on those data and the inclusion of a housing insecurity topical module in the 2023 AHS. HUD’s work on Housing Insecurity measurement will be the focus of a future PD&R@50 article.

Examples of research using the AHS include a bibliography of academic research using AHS data, HUD-funded research on the AHS, and a Cityscape symposium featuring research using the AHS.