What Drives Rising per Unit Cost In the Housing Choice Voucher Program?

Introduction and Purpose:

The per unit cost (PUC) of the Housing Choice Voucher (HCV) program has increased rapidly in recent years, which has become an important constraint on the program’s expansion. While total program funding has increased significantly, the number of families served has remained relatively flat. Frustratingly, little is known about the factors driving this increase. Theoretically, the public cost of a housing voucher reflects the following: market rents, the size and location of subsidized housing, the demographics of subsidized families, the frequency of mobility, and the relative bargaining power of landlords and public housing authorities (PHAs). This paper examines the main drivers of the recent increase in PUC by decomposing the influence of these various factors. Through this, we aim to disentangle the relative contributions of market, compositional, and institutional factors to PUC growth and to inform policy responses.

Methods:

To analyze factors driving PUC increases in the HCV program, we constructed a panel dataset at PHA level using longitudinal data spanning 2015–2019 and 2021–2024 (pandemic-era data was excluded due to unreliability). This file combined programmatic data on the HCV program with the American Community Survey (ACS). We employ a mixed-effects hierarchical linear model to predict year-over-year changes in PUC. We examine the influence of median rents, household size, neighborhood characteristics, and other factors. This method allows us to determine which factors are empirically associated with PUC.

Results:

Results indicate that the rent growth rate and the lagged rent growth rate were positively associated with the PUC growth rate (all p<.001), a relationship that was consistently confirmed across all models. In addition, changes in % new move-ins (p<.01), people per unit (p<.001), and % disabled households (age 62+) (p<.001), and the distribution of bedroom sizes (p<.001) showed significant positive associations with the PUC growth. As expected, the neighborhood poverty rates (p<.001) and household rent contributions (p<.001) were negatively associated with PUC. This pattern was maintained in the random-slope model and the extended model, including year fixed effects, and model fit was significantly improved (p<.001).

Despite these multiple significant associations, only rent and lagged-rent growth showed substantial median growth over the analytic time-period. Demographic and institutional factors increased in some jurisdictions while falling in others, resulting in largely flat trends. Thus, while our research confirms that a number of factors influence public costs, only rising rents can be identified as causing the large increase in PUC.

Conclusions and Implications:

These results expand our understanding of PUC increases by showing that the program’s efficiency depends on market forces beyond its control. Despite policies that cap rents in properties subsidized by vouchers and requirements that rents be reasonable relative to market comparatives, program costs appear to reflect market forces, albeit at a lag. This finding contributes to a broad literature related to the tradeoffs associated with neoliberal governance. Indeed, it appears that the program’s growth is hamstrung by the specific dynamics that demand its expansion: as rent increases increase demand for subsidies, the voucher program necessarily serves fewer people.