«Urban Problems and sPatial methods VolUme 17, nUmber 1 • 2015 U.S. Department of Housing and Urban Development | Office of Policy Development and ...»
and (2) employer reports giving establishment-level data, also known as the Quarterly Census of Employment and Wages. The coverage is roughly 96 percent of private, nonfarm wage and salary employment. LEHD is in the process of integrating data on self-employed individuals and independent contractors that are not covered in the UI files but that are available from the Census Bureau Business Register, which contains the universe of all businesses, including all sole proprietorships, on an annual basis (whether the sole proprietor has employees or is a nonemployer). In addition, the LEHD project has acquired personnel records from the U.S. Office of Personnel Management so that federal workers are now also tracked in LEHD. See also Abowd, Haltiwanger, and Lane (2004).
Partially synthetic public use datasets have already been developed for selected Census Bureau surveys (the Survey of Income and Program Participation, the Longitudinal Business Database, the LEHD dataset of AR, and ACS group quarters data) to permit release of additional microdata without jeopardizing the confidentiality of respondents.
For other examples and discussions of partially synthetic data, see Abowd and Lane (2004); Abowd and Woodcock (2004, 2001); Kennickell (1997); Little (1993); Little, Liu, and Raghunathan (2004); Reiter (2005a, 2005b, 2005c, 2004a, 2004b, 2003); and Reiter and Mitra (2009).
Many AHS demographic, economic, and housing variables also appear in ACS and can therefore be used for modeling the joint distributions. Because the precise location of the unit is known, when operating within a confidential environment, one can also link census tract characteristics from the 2000 long form and ACS. Those additional variables could help in developing imputation models.
It appears likely that enough exact-matched cases are available to enable researchers to form decent imputation models. From 2000 through 2012, 16.6 million ACS interviews were completed, mostly at unique addresses.12 In 2011, approximately 132 million housing units were in the United States; thus, ACS cases represent approximately one-eighth of all housing units (although a small fraction of ACS housing units will have been demolished by 2011). One-eighth of the 2011 AHS sample size of 180,000 units yields more than 20,000 units in the 2011 AHS that had been in the 2000–2012 ACS. Of course, not all these matches are possible, because some AHS cases are added each survey year to account for new construction, so previous AHS sample sizes were smaller (although one can use both the national AHS files and the metropolitan area AHS files), and some matching variables will be missing.
Analysts can derive models for selected AHS characteristics that are not included in ACS by examining the relationships between AHS variables that are also in ACS and those that are not. These models can then be used to create the synthetic AHS variables for ACS households—both those in the exact match universe and those only in ACS. Multiple imputations are typically done to reduce potential bias from any one draw from the joint distribution. The validity of the imputations can be tested by reference to the actual values for the exact matches, and it seems wise to focus on imputing only a few critical housing measures.
Some Data Issues The previous section discussed some promising techniques for enhancing existing datasets for housing research. Some problems remain, however. One of these problems is undercoverage of housing units by the sampling frame that the Census Bureau uses for its household surveys. To the extent that units at the edge of habitability, or more likely units that are the result of conversion or subdivision (such as a converted garage) are missed, the statistics we use to describe our housing stock will be biased. A second type of problem is the development of new types of living quarters, such as congregate housing. As the population ages, more and more people are living in developments that cater to their needs, including housing units that do not have kitchens (because meals are provided centrally). How we should measure their growth depends on how we allow the definition of habitable housing to vary, and whether we can adequately add them to our sampling frame.
Another problem is that we measure the same concept in multiple ways and in multiple surveys.
A prime example is the vacancy rate. On AHS, the definition of a vacant housing unit is quite similar to that used in the Housing Vacancy Survey.
A housing unit is vacant if no one is living in it at the time of the interview, unless its occupants are only temporarily absent. In addition, housing units where all the occupants have a usual residence elsewhere are grouped with vacant units. … For vacant housing Duplicates can be treated as independent observations, because they will be matched to different AHS observations.
units that are not intended by their current owners for year round use (seasonal and migratory), the respondent was asked whether the construction and heating of the housing unit made it suitable for the unit to be occupied on a year-round basis. A housing unit is suitable for year-round use if it is built as a permanent structure, properly equipped, insulated, and heated as necessitated by the climate.13 This definition differs from the definition of a vacant housing unit used in the American Community Survey, however, because of the ACS residence rules.
The basic idea behind the ACS current residence concept is that everyone who is currently living or staying at a sample address is considered a current resident of that address, except for those staying there for only a short period of time. For the purposes of the ACS, the Census Bureau defines this short period of time as less than 2 consecutive months (often described as the 2-month rule). Under this rule, anyone who has been or will be living for 2 months or less in the sample unit when the unit is interviewed (either by mail, telephone, or personal visit) is not considered a current resident. This means that their expected length of stay is 2 months or less, not that they have been staying in the sample unit for 2 months or less. In general, people who are away from the sample unit for 2 months or less are considered to be current residents, even though they are not staying there when the interview is conducted, while people who have been or will be away for more than 2 months are considered not to be current residents. The Census Bureau classifies as vacant [a housing unit] in which no one is determined to be a current resident. (U.S. Census Bureau, 2009: chapter 6) The implications for measuring vacancy rates derive from the ACS interview methodology. First, a questionnaire is mailed out and, starting in January 2014, a solicitation to respond is sent via the Internet as well. This questionnaire is followed about 1 month later by a telephone followup (if a phone number can be obtained), and then, for a sample of nonrespondents, about 1 more month later by a personal visit. A unit that might have been vacant at the original date of interview (month 1) may well become occupied at month 3, when the household is interviewed, yielding a lower vacancy rate than might otherwise be recorded. Thus, it is likely that vacancy rates measured by ACS differ from those measured by household surveys and from the decennial census (which uses yet a different methodology).14 Data References Appendix American Community Survey (ACS): http://www.census.gov/acs/www/; see also U.S. Census Bureau, American FactFinder.
American Housing Survey (AHS): http://www.census.gov/housing/ahs/ and http://www.huduser.
org/portal/datasets/ahs.html (1997–2011 AHS); see also U.S. Census Bureau, American FactFinder.
Quoted from http://www.census.gov/housing/ahs/files/, Appendix A: 31–32.
See Clark (2012), Fish (2013), and Griffin et al. (2004) for comparisons with the decennial census.
Black Knight Financial Services, Inc.: https://www.lpsdefault.com/fnds/home.asp.
Census of Construction, 2002 and 2007 (2012 forthcoming): http://www.census.gov/econ/census07/;
see also U.S. Census Bureau, American FactFinder.
Construction statistics (monthly, quarterly, and annual surveys): http://www.census.gov/econ/ construction.html.
Construction statistics (New Residential Construction, Manufactured Homes Survey, New Residential Sales, Residential Improvements, Characteristics of New Housing, Construction Price Indexes, and Value of Construction Put in Place): http://www.census.gov/mcd/.
CoreLogic, Inc.: http://www.corelogic.com/solutions/property-information-analytic-solutions.aspx.
Decennial Census of Housing, 1940–2000 (tabulations): http://www.census.gov/housing/census/data/.
Decennial Census of Population and Housing, selected historical decennial census population and housing counts, 1790–1990: http://www.census.gov/population/www/censusdata/hiscendata.html.
Decennial Census of Population and Housing, 1990: http://www.census.gov/main/www/cen1990.
Decennial Census of Population and Housing, 2000 and 2010: http://factfinder2.census.gov.
Fannie Mae, National Housing Survey (NHS): http://www.fanniemae.com/portal/research-andanalysis/housing-survey.html.
Federal Financial Institutions Examination Council, Community Reinvestment Act (CRA), Distressed and Underserved Tracts, http://www.ffiec.gov/cra/distressed.htm.
Federal Financial Institutions Examination Council, Home Mortgage Disclosure Act (HMDA):
Federal Housing Finance Agency, government-sponsored enterprise mortgage purchases: http:// www.fhfa.gov/Default.aspx?Page=137.
Federal Reserve Board of Governors, Flow of Funds Accounts, Net Change in Mortgages Outstanding and of Total Outstanding Mortgages: http://www.federalreserve.gov/releases/z1/about.htm.
Freddie Mac, House Price Index (FMHPI): http://www.freddiemac.com/finance/fmhpi/.
Freddie Mac, Primary Mortgage Market Survey (PMMS): http://www.freddiemac.com/pmms/.
Housing Vacancy Survey (HVS): http://www.census.gov/housing/hvs/.
Integrated Public Use Microdata Series (IPUMS): https://www.ipums.org.
Manufactured Homes Survey (MHS): http://www.census.gov/construction/mhs/mhsindex.html.
Mortgage Bankers Association, Weekly Applications Survey: http://www.mortgagebankers.org/ ResearchandForecasts/ProductsandSurveys/WeeklyApplicationSurvey.
MPF Research: http://www.realpage.com/apartment-market-research/rental-market-trends.
National Association of Home Builders: http://www.nahb.org/.
National Association of Realtors®: http://www.realtor.org/.
National Council of Real Estate Investment Fiduciaries (NCREIF), Data and Products Guide: http:// www.ncreif.org/public_files/NCREIF_Data_and_Products_Guide.pdf.
New Residential Construction (Building Permits Survey, Quarterly Starts and Completions by Purpose and Design, Annual Characteristics of New Housing, Length of Time from Authorization to Start and from Start to Completion, Construction Price Indexes): http://www.census.gov/ construction/nrc/.
New York City Housing and Vacancy Survey (NYCHVS): http://www.nyc.gov/html/hpd/html/pr/ vacancy.shtml and http://www.census.gov/housing/nychvs/about/.
Panel Study of Income Dynamics (PSID): http://psidonline.isr.umich.edu/.
Property Owners and Managers Survey (POMS): http://www.census.gov/housing/poms/.
Real Capital Analytics Inc.: https://www.rcanalytics.com/Public/rca_cppi.aspx.
RealtyTrac® Inc.: http://www.realtytrac.com/.
Rental Housing Finance Survey (RHFS): http://www.census.gov/hhes/rhfs/.
Residential Energy Consumption Survey (RECS): http://www.eia.gov/consumption/residential/.
Residential Finance Survey (RFS): http://www.census.gov/housing/rfs/.
Survey of Consumer Finances (SCF): http://www.federalreserve.gov/econresdata/scf/scfindex.htm.
Survey of Market Absorption (SOMA): http://www.census.gov/hhes/www/housing/soma/overview.
Survey of Residential Alterations and Repairs (SORAR): http://www.census.gov/construction/c50/ c50index.html.
United Nations Statistics Division, Compendium of Human Settlements Statistics/Compendium of Housing Statistics: http://unstats.un.org/unsd/demographic/sconcerns/housing/housing2.htm.
U.S. Bureau of Labor Statistics, Consumer Price Index (CPI): http://www.bls.gov/cpi/.
U.S. Census Bureau, American FactFinder (2000 Census of Population and Housing, 2010 Census of Population and Housing, 2005–2012 ACS, 2011 AHS, 2002–2012 Census of Construction [sector 23]): http://factfinder2.census.gov.
U.S. Census Bureau, Housing Unit Estimates (vintage 2012): http://www.census.gov/popest/ data/housing/totals/2012/index.html and its methodology, at http://www.census.gov/popest/ methodology/2012-hu-meth.pdf.
U.S. Census Bureau, Longitudinal Employer-Household Dynamics (LEHD): http://lehd.ces.census.gov.
U.S. Census Bureau, Research Data Centers (RDCs): http://www.census.gov/ces/rdcresearch/index.
U.S. Department of Housing and Urban Development (HUD), Homelessness Data Exchange:
HUD, National Low-Income Housing Tax Credit (LIHTC) Database: http://LIHTC.huduser.org.
HUD, Public and Indian Housing Information Center (PIC) system and Tenant Rental Assistance Certification System (TRACS): aggregate data (A Picture of Subsidized Households) at http://www.
huduser.org/portal/datasets/assthsg.html; a 5-percent microdata sample from those systems for researchers at http://www.huduser.org/portal/pumd/index.html.
U.S. Postal Service, Vacancies: Aggregate data quarterly at the census tract level for government and not-for-profit researchers from HUD at http://www.huduser.org/portal/datasets/usps.html.
Zillow Home Value Index: http://www.zillow.com/howto/api/APIOverview.htm.
Acknowledgments An earlier version of this article was prepared for the January 2014 conference of the Weimer School of Advanced Studies in Real Estate and Land Economics. The author thanks Arthur Cresce, George Galster, Henry Pollakowski, Mark Shroder, and Dav Vandenbroucke for their comments and suggestions.