Developing a Proxy for Identifying Family Developments in HUD’s LIHTC Data: Using Information on the Distribution of Bedroom-Sizes

The only existing national database on projects in the Low-Income Housing Tax Credit (LIHTC) Program has limited data on which developments serve families, a population of considerable interest to policymakers and researchers. To fill this gap, we use existing data on the size distribution of units in LIHTC projects to develop a proxy for family developments. We supplement this work with data on occupants of LIHTC developments in six states to test how well this proxy works. We estimate that this proxy would capture 92 to 96 percent of units in family developments.

Assessments of housing programs frequently distinguish how well such programs serve families (Khadurri, Buron, and Claminco, 2006; Khadurri, Buron, and Lam, 2004; Newman and Schnare, 1997). Although not always stated explicitly, a focus on families and their environments might arise out of heightened concern for the children they may contain or out of recognition that issues related to working-age adults may be of particular interest in housing programs. (Housing programs generally apply a loose definition of family, encompassing any household composition that operates as a unit, further distinguishing families from elderly families or populations requiring special services. For our purposes, we take family to mean a multiple-person household operating as one unit, which may or may not contain children and which would not be classified as an elderly household.) Assessing the largest federal supply-side program (the Low-Income Housing Tax Credit [LIHTC] Program) is hampered by our limited ability to identify which LIHTC developments serve (or house) families. No national data currently exist on tenants of LIHTC housing. The one existing national database on LIHTC projects includes some information on whether states report that a development “targets” specific populations, including families, but those data are fairly incomplete, even among newer projects. In addition, states vary on whether family is a “targeted population” in their allocation process, or if families are generally served in developments that do not target other specific groups, such as the elderly4 or those with special needs.

In the absence of good national data on which developments serve families (whether targeted or as a remainder category), researchers have either collected the data needed for a particular state (Kawitzky et al., 2013;5 Pfeiffer, 2009) or used proxies, such as units with at least two bedrooms (Ellen and Horn, 2012; Khadurri, Buron, and Claminco, 2006). This second method focuses on units rather than family developments as a whole, which may be more appropriate for some policy questions than others. This article develops and tests a method for identifying family developments within the national LIHTC stock, using publicly available data. We first develop this categorization scheme using the U.S. Department of Housing and Urban Development’s (HUD’s) LIHTC data. We then assess its performance through a combination of HUD’s LIHTC data on projects and data we have collected on LIHTC tenants in six states.