Reducing documentation burdens to accelerate and broaden access to emergency rental assistance

View the PDF Version

Key findings

We found that simplifying income documentation reduced application processing times and increased application approval rates, but did not significantly increase the number of new applications received for Emergency Rental Assistance (ERA).

Agency priority

At the outset of the COVID-19 pandemic, housing advocacy groups warned that 20% of renting households would be at risk of eviction by the end of 2020 if the government did not act.1 Housing instability was especially high among renters with low incomes and renters who identified as people of color.2

The U.S. Department of the Treasury (Treasury) administered the Emergency Rental Assistance (ERA) programs, which collectively provided over $46 billion to state, local, territorial and Tribal governments (grantees) to prevent eviction and housing instability in the wake of the pandemic.3 The ERA grantees provided direct cash assistance to renters, landlords, and utility providers to assist with rent, utilities, and other housing-related expenses.

Individuals were eligible for ERA assistance if their household income fell below an area-specific threshold (in addition to being at risk of housing instability and experiencing hardship due to the pandemic). To establish that ERA applicants met the income criteria to be eligible to receive ERA, grantees often required W2s or paystubs to verify their income. Locating and uploading this documentation was time-consuming and difficult for many applicants, as was verification and follow-up by administrators.4

Treasury encouraged grantees to incorporate program design flexibilities that enabled more efficient delivery of ERA assistance to eligible renters in communities disproportionately impacted by the pandemic. This evaluation examined one such flexibility, the fact-specific proxy (FSP), which enables grantees to use an alternative data source (e.g., average incomes of households based on their geographic area) to verify an applicant’s income to determine eligibility to receive ERA assistance, without potentially burdensome follow up requirements. In evaluating the FSP, this effort contributed to the American Rescue Plan Equity Learning Agenda and Treasury’s Office of Capital Access Learning Agenda questions on the effectiveness of efforts to allocate American Rescue Plan Act (ARP) funds in a manner that advances equitable outcomes.

Program change description

Virginia’s Department of Housing and Community Development (VA DHCD) was an early adopter of program flexibilities, simplifying income documentation requirements through an FSP. Introduced on June 10, 2021, partway through VA DHCD’s Rent Relief Program (RRP)5, the FSP used the applicant’s ZIP code as a proxy to verify income eligibility. Renter households with three or fewer members residing in the 500 ZIP codes whose median income fell below the federally-defined low income threshold for Virginia ($66,950 for a three-person household) were able to corroborate their eligibility through a written attestation included in the application form.

Evaluation design

We partnered with VA DHCD to conduct a retrospective impact evaluation to understand if streamlining the requirement for providing individual income eligibility documentation improved access to assistance for low-income renters in the state.

This evaluation took advantage of the fact that estimated median income is the only determinant of whether a ZIP code is qualified for the FSP. Because ZIP code-level median income was the only determinant of receiving FSP, the ZIP codes that did not qualify for the FSP can serve as a comparison group for the ZIP codes that did, after controlling for area-level income. See the analysis plan for a full explanation of the design. We note unplanned analyses below and explain these departures in detail in the technical appendix.

Analysis of existing data

Using data on over 98,000 tenant-initiated applications processed by a single vendor from 2/21/2021 to 12/1/2021, we created three datasets for the analysis. The first dataset aggregates the total number of applications and the total amount of ERA assistance spent at the ZIP code level.6 The second and third datasets are restricted to individuals who applied prior to the implementation of FSP, because their decision to apply could not have been affected by the FSP.7 We measured the days it took to process those individuals’ applications as well as the rate of approval, both at the ZIP code level (planned) and at the individual-level (unplanned, but greater statistical power with a larger sample size). To estimate the impact of the FSP, we used a regression discontinuity (RD) and an ordinary least squares (OLS) regression that adjusts for the sole confounding variable (median income). Figure 1 reports all estimates. We focused on the OLS results because of the RD estimates’ imprecision.

Finally, we note that much of RRP went to underserved communities: Black households and extremely low income households accounted for 59% and 71% of approved households, respectively.8 Because we are missing demographics for around 15% of applicants however, our subgroup analyses are imprecise. We report them in the technical appendix alongside descriptive statistics and a comparison of demographics in FSP vs non-FSP ZIP codes.


  1. Simplifying income documentation reduced application processing time by two weeks. In non-FSP ZIP codes, the average time to process an application as approved, rejected, or inactive was 180 days. We estimate a statistically significant fifteen-day reduction in that time.9
  2. Simplifying income documentation increased the approval rate by 11 percentage points.10 This increase is from a baseline approval rate of one-third of applications in non-FSP ZIP codes.
  3. Simplifying income documentation did not significantly increase the total number of new applications received for ERA. In non-FSP ZIP codes, the average number of applications over the period we studied was 45. Our estimate implies an increase of roughly one application, which is not statistically significant.11
  4. Simplifying income documentation may have increased the total amount of rental assistance paid out. The average non-FSP ZIP code received a total of $175,931 in rental assistance. Our estimate implies an increase of $1,400 in FSP ZIP codes compared to non-FSP zip codes, which is not statistically significant. Our other planned analysis finds a significant increase, but the result is imprecise as noted above.12

Figure 1. Impact of simplifying income documentation on days spent processing applications, approval rate, total number of applications, and total amount of rental assistance paid to approved applicants  Impact of simplifying income documentation on days spent processing applications, approval rate, total number of applications, and total amount of rental assistance paid to approved applicants.


We found VA DHCD’s efforts to simplify income documentation using an FSP made getting out much-needed emergency rental assistance faster, and that individuals facing housing instability were more likely to receive such assistance during the COVID-19 pandemic. Importantly, this finding holds for individuals who applied before FSP was implemented or publicized, and who therefore could not have been encouraged to apply for ERA because of the use of an FSP.

We found no evidence that reducing barriers through FSP brought substantially more people into the ERA program. For an oversubscribed program such as ERA, this may constitute good news: there is no evidence that program staff were overwhelmed by an influx of new applications for ERA due to the use of an FSP. This finding suggests FSP could be a complement to more traditional outreach-focused approaches if the goal is to increase the number of applications for assistance.

Our findings contribute to the growing evidence that an FSP can be useful in programs where application volume is high, when specific application criteria may be slow to verify, when it may be difficult for some applicants to upload supporting documentation, and when applications for assistance are potentially time-sensitive. For example, settings such as applications for financial assistance after storm-related disasters could be suitable for the use of an FSP. Further, implementing an FSP can help reduce the likelihood that application reviewers face large backlogs and reduce the chance of eligible applicants missing out on much needed assistance.


  1. See Aspen Institute report, 20 Million Renters Are at Risk of Eviction (Jun. 2020).
  2. See Consumer Financial Protection Bureau, Housing Insecurity and the COVID-19 Pandemic (Mar. 2021)
  3. The ERA1 program is authorized by the Consolidated Appropriations Act, 2021, and provided $25 billion for assistance to eligible households. The ERA2 program is authorized by the American Rescue Plan Act of 2021 and provides $21.55 billion for assistance to eligible households.
  4. See, for example, the National Low Income Housing Coalition’s report on fact-specific proxies (Impacts, pg. 11).
  5. VA DHCD’s Rent Relief Program (RRP) started in June 2020 and ended in May 2022, with funding from the state’s Housing Trust Fund, the Coronavirus Relief Fund, ERA1, and ERA2.
  6. We depart from the analysis plan by taking the log (plus 1) of these outcomes to reduce the skew created by a large number of 0s, which biased estimates toward positive and significant findings. See simulation study in the technical appendix.
  7. This group’s outcomes are therefore unaffected by “post-treatment bias” (see Rosenbaum (1984) and analysis plan).
  8. See here for demographics reported by VA DHCD and Executive Order 13985 for definition of underserved.
  9. The (planned) estimate of FSP on the ZIP-level average spent processing an application was -14.84 for the OLS (p = 0.042, 95% CI [-29.13, -0.56]) and 8.69 for the RD (p = 0.679, 95% CI [-32.49, 49.88]). We also report the results of an unplanned analysis at the individual-level (clustering standard errors by ZIP), which helps the RD in particular by providing many more data points for its extrapolation. Estimated impacts on (winsorized) days spent processing at the individual level are -15.23 for the OLS (p <0 .001, 95% CI [-24.27, -6.19]) and -15.85 for the RD (p<0.001, 95% CI [-19.44, -12.27]).
  10. This analysis was unplanned and can be estimated at the ZIP- or individual-level, so we report both here. The estimates for the ZIP-level approval rate are .11 for the OLS (p = 0.007, 95% CI [0.03, 0.20]) and 0.16 for the RD(p = 0.196, 95% CI [-0.08, 0.39]). For the individual-level probability of approval, the estimates are 0.12 for the OLS (p < 0.001, 95% CI [0.07, 0.17]) and 0.20 for the RD (p < 0.001, 95% CI [0.10, 0.30]).
  11. Estimates are 0.11 for the OLS (p = 0.191, 95% CI [-0.05, 0.26]) and 0.36 for the RD (p = 0.056, 95% CI [-0.01, 0.74]).
  12. Estimates are 0.75 for the OLS (p = 0.106, 95% CI [-0.16, 1.65]) and 2.651 for the RD (p = 0.025, 95% CI [0.34, 4.96]).