Association for Public Policy Analysis & Management

The substantial literature on housing quality demonstrates the importance and the challenges of conceptualizing and measuring housing quality. The precariousness of the low-rent unassisted housing stock and persistent concerns about housing quality of the assisted stock make this an opportune time to revisit housing quality.

After reviewing the relevant literature, we develop alternative housing quality indices, test their validity, and apply them to both the assisted and comparable unassisted housing stock. We focus on indicators of physical integrity or housing systems, and exclude measures that are more likely to reflect the resident’s housekeeping or behavior.

Because no consensus exists about the housing features that should be included in the definition of a dwelling’s quality and how each should be weighted in determining overall quality, we rely on three external criteria first suggested by the Experimental Housing Allowance Program: market value, consumer rating, and normative standards. We test the market value criterion with a hedonic approach. The consumer rating criterion identifies the dwelling features most associated with the resident’s assessment of the dwelling as a good place to live regardless of the market price of these features, and reflects the renewed interest in measuring happiness and subjective well-being. The normative standards criterion reflects community concerns and policy decisions such as building codes. Our main data are the 2011 and 2013 national AHS. Both provide augmented national samples and assisted housing samples, with assisted housing identified by address matches to HUD administrative data, not self-report.

Despite its intuitive appeal, the market value criterion performs poorly and we drop it from further analysis. We construct one consumer rating index and three indices using the normative standards criterion: a weighted index based on a previous analysis for HUD-PDR; an unweighted index; and an index with weights derived from a factor analysis. For the consumer rating index, we use the odds ratios from ordered logistic regressions as the weights. Both the convergent and predictive validity of these four indices are strong.

Consistent with past research, the prevalence rate of almost all housing quality problems is very low, with most dwelling units having no problems. The quality of assisted housing is comparable to the quality of unassisted housing, the incidence of housing problems persisting over the 2011-2013 period is very low, and repairs are made promptly.

Heterogeneity in housing quality exists, however. Geographically, assisted housing quality in central cities and in the Northeast is considerably lower than in other regions. Demographically, the nonelderly disabled, non-whites, and large households emerge as more likely to live in lower-quality assisted units. The analysis also supports the 40^th percentile of rents definition of the Fair Market Rent.

Policy implications include support for a shift to biennial inspections in the voucher program and biennial and triennial inspections for standard and high performers, respectively, in the public housing program, and to serious consideration of proposals to streamline inspections to encourage participation in the voucher program by private owners of rental properties. We also offer suggestions for future research.