Use of Alternative Data to Supplement Low- and Moderate-Income Summary Data in the Community Development Block Grant Program
The CDBG program’s primary national objective is to provide benefit to low- and moderate-income (LMI) persons. Program activities that benefit all residents of a particular area demonstrate compliance with this objective when at least 51 percent of the population can be shown to be LMI. HUD currently provides grantees several options for documenting the percentage of LMI persons in a CDBG activity’s service area. HUD produces LMISD, drawn from custom tabulations of American Community Survey (ACS) data, at several geographic levels and releases the data every five years. Grantees may also conduct local income surveys to demonstrate eligibility. LMISD’s sample-based estimates can be imprecise, especially for areas with small populations. The report investigates potential alternative methods for determining LMI eligibility. Specifically, it explores the use of several publicly available administrative datasets and new uses of ACS data that could potentially supplement, challenge, or replace current LMISD estimates. The administrative datasets that were utilized in this analysis include federal income tax data from the Internal Revenue Service (IRS), home price indices produced by the Federal Housing Finance Agency (FHFA), residential vacancy rates from the US Postal Service (USPS), and jobs data from the US Census’ Longitudinal Employer-Household Dynamics Program (LEHD). The new uses of ACS data that were explored include allowing for the application of LMISD estimates outside of the current five-year cycle and the calculation of LMI estimates from standard (as opposed to custom) ACS tables.
The analysis highlights the advantages and disadvantages of these options. It suggests that methods involving administrative datasets likely have too many limitations to be feasible options in the short-term. Administrative datasets suffer from issues of missing data and incomplete geographic coverage. Perhaps more importantly, alternative estimates of economic distress from the administrative datasets show very low levels of correlation with LMI percentage estimates from LMISD. Adopting such measures may thus conflict with CDBG’s statutory intent, with its focus on targeting benefits to the LMI population. New uses of ACS show more promise as potential supplements to LMISD, although further research is needed to understand the implications of such changes.