U.S. Department of the Treasury

To receive HAF assistance, homeowners must demonstrate both a COVID-related financial hardship as well as a qualifying income level. Some income verification methods that rely heavily on producing specific documentation, such as income tax records, could create a barrier to access for certain homeowners. An individual's circumstances can change over time. Job loss, income fluctuations due to self-employment or independent-contract work, destruction of property from natural disasters, or other life circumstances can make income verification challenging. Therefore, providing flexibility for applicants to demonstrate their income can be critical in serving some homeowners.

While homeowners with a mortgage typically were required to provide income documentation when they obtained their loan, an individual's income and situation can change. Years after taking out the loan, homeowners applying for HAF assistance may not have easy access to current income documentation and may have difficulties navigating detailed documentation requirements. Moreover, the HAF program covers a variety of expenses beyond mortgage assistance, including delinquent property taxes; utilities; internet; homeowner's insurance; HOA fees; repairs to maintain a home's habitability; and assistance to obtain clear title to property. Many of these expenses are typically incurred without the homeowner first providing extensive homeowner documentation.

Fortunately, HAF participants can verify an applicant’s income for eligibility through a variety of methods. To simplify the HAF applicant’s eligibility process, HAF participants may elect to use reasonable fact-specific proxies to determine the approximate income level of a given household based on external data sources. This both provides more flexibility to serve applicants who may have difficulty producing certain documentation and can reduce administrative burden for HAF program staff. Using fact-specific proxies can increase program efficiency in determining an applicant’s eligibility based on income.

Determining Eligibility with Fact-Specific Proxies

EXAMPLES

The State of Indiana’s HAF Program allows applicants to demonstrate their income eligibility through documentation and self-attestation. The Indiana Housing and Community Development Authority (IHCDA) realized that income verification created a bottleneck for processing applications for a variety of reasons: varying formats across applicants’ submitted documentation; receiving illegible copies of documentation which required additional outreach to applicants; barriers to applicants’ access to technology; and applicants’ lack of knowledge about using technology. By collaborating with Indiana’s Emergency Rental Assistance (ERA) team, Indiana’s HAF team learned about the benefits of using an applicant’s proof of receiving federal benefits as a fact-specific proxy to verify the applicant’s income. Applicants who receive federal benefits such as, SNAP, WIC, or TANF can provide a copy of their approval letter from the federal program as income verification documentation. By using applicants’ federal benefits as a fact-specific proxy when determining their income eligibility, the State of Indiana has found that this practice streamlined the applicant eligibility screening processes for staff and reduced the documentation burden for applicants.

The State of South Carolina’s HAF Program uses fact-specific zip code proxies along with the homeowner’s attestation of income to determine a homeowner’s eligibility for HAF assistance based on income. By utilizing this proxy, South Carolina’s HAF Program staff finds that it relieves the burden from homeowners by reducing the amount of documentation required to apply for HAF assistance. Additionally, this practice has sped up the application review process which in turn serves both the program and the homeowner by getting the much-needed HAF assistance applied to the appropriate account quicker.

Determining Income for Public Benefit Recipients

Eligibility based off an applicant’s other state and/or federal public benefits information is a useful proxy for determining whether an applicant’s income qualifies him or her for assistance under the HAF Program. When an applicant can demonstrate his or her income eligibility by providing proof of receiving public benefits, the HAF program staff can verify the applicant’s income based on the public benefit income requirements. This practice can reduce the time needed to process applications and therefore reduce the burden on the program’s staff.

Key Considerations for Determining Income Using Homeowners’ Public Benefits Information

• Provide a list of public benefit programs to help applicants demonstrate their qualifying income level. A few examples of federal public benefits include:

  • Temporary Assistance for Needy Families (TANF) block grant
  • Supplemental Security Income (SSI)
  • Supplemental Nutrition Assistance Program (SNAP)
  • State-financed General Assistance (GA)
• Share how homeowners can obtain copies of their benefit certification or benefits approval letter by providing links to government websites and identifying public libraries’ locations for those that don’t have access to a computer to, as applicable, download, print, or upload such documentation for submission with their application for HAF assistance
• Provide staff with a clear list of income requirements for the other State and Federal programs to help with verifying income by cross referencing these resources. This list should also include examples of acceptable proof of benefits such as, an approval letter or examples of other acceptable forms of documentation.

Guidelines to Identifying Other Fact- Specific Proxies

Treasury prepared the guidelines below to help HAF participants identify alternative methods for verifying applicants’ income through other fact-specific proxies such as, area median income. HAF participants can verify the applicant’s information independently by cross-referencing the applicant’s household size and income against a proxy data source on the backend.

Developing a Proxy

To develop and use a proxy, a HAF participant will need to do the following^[1]:

1. Use population-level income data to identify which geographic areas align with the HAF program’s priorities such as, low-income households earning at or below 150% area median income (AMI). This will create the “fact-specific proxy.”
2. Develop a process for looking up addresses of applicants within the defined geographic areas.
3. Integrate the new fact-specific proxy within the application review process to help program staff determine applicant eligibility.

AN EXAMPLE PROXY

HAF participants can use many sources of data to create a fact-specific proxy (or proxies) that can alleviate applicant documentation burden while ensuring the population served aligns with the HAF Program’s goals.

For this example, the following federal data and resources can be used and adopted by a HAF participant:

• Geographic area: Use census tracts. The tract level provides a good tradeoff between specificity and data quality.
• Data Sources: Use median homeowner data from U.S. Census American Community Survey (ACS) 2019 5-Year estimates table B25119 for applicants, where data exist. These data are specific to the HAF program population and will best reflect applicant incomes.
  • Where data do not exist for a tract, backfill with median overall incomes for tracts from ACS 2019 5-Year estimates table B19013.
  • Margin of error strategy: Accept the ACS estimate values as provided, without doing further analysis on margin of error, in order to avoid penalizing lower-population areas.
• Address lookups: For all areas covered by the U.S. Department of Transportation (DOT) National Address Database (NAD), perform lookups to determine address coordinates. Using these coordinates, perform geospatial queries to locate address within a census tract using Census TIGER line/shapefiles.
  • For addresses not covered by the NAD, query the Census Geocoder for tract information.

Using the above resources, the median income of owner-occupied houses for a given address's census tract can be looked up. Compare this value to the income limit set by the program (equal to or lesser than 150% of the AMI limit defined on page 5 of the HAF Guidance) for the larger area where an address falls in. If the median owner-occupied household income falls sufficiently below the program’s income threshold for eligibility, the proxy would show this fact and the program can rely on it. For example, if the proxy indicates a median income in the tract is 80% of County AMI, then it suffices to establish eligibility under a cap of 100% of the median income for the United States.

Key Considerations for Developing a Proxy

• The specific design of a data proxy can be adjusted to the needs of the specified location; while the previously stated Federal data sources and systems could work generically for any locality, Treasury also recognizes that there are many ways to develop a reasonable fact-specific income proxy.

DETERMINING A GEOGRAPHIC AREA FOR A PROXY

The geographic area chosen for the fact-specific proxy will determine which data is available and what processing you will need to do on applicants’ addresses to determine income facts. For instance, the use of the U.S. Census Bureau’s data may provide access to many geographic areas. Census tracts offer a good tradeoff between size, data availability, and data quality. The census offers smaller units, such as block groups, but the sample size of these units makes further subdivisions less meaningful. In general, programs should aim to strike a balance between specificity, data quality, data availability, and ease of implementation.

Considering ZIP codes

For some participants, census tracts may be technologically burdensome to use, since locating households within particular census tracts requires more advanced geocoding. In these cases, U.S. Postal Service (USPS) ZIP Codes (or a related geographic area) can make integrating the proxy simpler with an application system. Using Census ZIP Code Tabulation Areas (ZCTAs) is a reasonable choice in areas where ZCTAs afford adequate precision for the HAF Program. A participant may also do their own analysis of census-tract data to local ZIP Codes, depending on the team's technical capacity. For example, a HAF participant could choose to use census tracts for analysis but generate a list of qualifying USPS Zip Codes based on analyzing an intersection of tracts and ZIP Codes in the HAF participant’s jurisdiction. This hybrid approach would allow for a simpler application system in exchange for more data pre-processing when developing the proxy, which may be an attractive tradeoff.

Alternative geographies

Geographies other than those used by the U.S. Census Bureau are also possible and reasonable to use (for example, school districts, state legislative districts, etc.), assuming the program administrator is familiar with locally used geographic areas and has access to data reliably collected for them.

Making tradeoff decisions

In picking a geographic area for the fact-specific proxy, the key is that program administrators understand the tradeoffs between the size of the population in the areas, the income diversity of the population within the area, and the implications of the choices for launching a fact-specific proxy. Once a choice is made, document it, including provide a justification for choosing the fact-specific proxy, and apply it consistently for all applicants.

FINDING DATA SOURCES

Data should come from government agencies (federal or local) or other reputable entities (e.g., universities, utility companies, etc.) and be reasonably recent.

The U.S. Census Bureau’s American Community Survey (ACS) data

As with geographic areas, the U.S. Census Bureau provides a variety of valuable sources such as, detailed estimate tables from its 1- and 5 - year ACS data. In our example, we use median owner-occupied incomes from the 2019 5-year estimates from the ACS data to verify that most households in a geographic area fall within the program’s AMI limit for a household. The ACS Data Tables can offer a fact-based information about most residents in a certain area and, when compared to the program’s income limits, provide assurance that applicants from a given geographic area will likely qualify for assistance under the HAF program and thus can self-attest to their income without further income documentation. Other similar U.S. Census Bureau tables are also acceptable such as, the overall area median income (for all residents) from ACS 2019 5-year estimates.

The U.S. Department of Housing and Urban Development (HUD) data

Alternatively, a HAF participant could construct a proxy using HUD data such as, HUD's Comprehensive Housing Affordability Strategy (CHAS) for 2015-2019 data or HUD's Qualified Census Tracts and Difficult Development Areas for 2023 data to determine whether an adequate percentage of the population in a given geographic area qualifies for assistance. For example, a HAF participant could reasonably develop a proxy that uses the fact that an applicant lives in a HUD Qualified Census Tract to determine that applicants from a qualified tract will likely qualify for assistance under the program and thus can self-attest to their income without further income documentation.

Other sources

HAF participants are not limited to federal data in designing a proxy. If the state has data assets or products that give insight into incomes of households for an area or allow for modeling of the probable income or financial hardship for an applicant on the basis of information provided in the application, these sources are permissible.

Margin of error strategy

If a HAF participant is using a data source based on sampling, such as ACS detailed tables, there will also be sample-error estimates included with the data. Because the error inversely correlates with sample size (that is, smaller samples will tend to have higher error), and any fact-specific proxy will rely on estimated data, it is acceptable to disregard the sample error in the interest of equity; when looking at all census tracts across the U.S., taking the margin-of-error into account when building a proxy will systematically disadvantage lower-population areas. HAF participants are welcome to do analysis on their specific geographic areas and determine an approach to error that is reasonable and equitable.

Combining multiple data sources

It is acceptable to combine multiple data sources for a variety of reasons, as long as it is done in a statistically valid manner. ACS does not have data for median owner-occupied household income for some areas. It is reasonable to fall back to a larger section of the population such as, overall area median income. A HAF participant may also combine income data, such as HUD Qualified Census Tracts, with other signs of area economic distress from local utilities, tax collections, or other locally available data sources, to determine geographic areas that are experiencing acute financial hardship.

Similarly, a HAF participant could combine multiple census tables to gain access to features beyond just location to estimate incomes for applicants based on other data in their applications. While the design of a more sophisticated proxy using multiple tables is beyond the scope of this document, a HAF participant designing such a model will need to be carefully tested for systematic bias along any protected categories and carefully weigh the additional predictive power against the risk of model bias compared to a simpler model based purely on location.

As with picking a geographic area, there are a variety of possible designs that are acceptable, from the relatively simple one in the example to more sophisticated models based on multiple data sources. HAF participants are free to tailor their data sources and methodology to their program, provided their methodology follows the spirit of these guidelines and that it is documented and applied consistently.

The precise mechanism for accessing the proxy data will depend on the application itself. Some options include:

• Building a lookup table of addresses or ZIP Codes for the application to directly access proxy facts while the homeowner is entering their data into the application.
• Making a web service that can ingest an address, perform geocoding and lookups, and return a fact to the application through integration
• Creating lookup tables in spreadsheets for telephone operators, case workers, and application reviewers to use while entering or approving applications

PERFORMING ADDRESS LOOKUPS

Once a HAF participant has selected a geographic area and data source(s), they will need to devise a method for determining which geographic area a given applicant falls into in its jurisdiction.

If a HAF participant is using ZCTA geographic areas, this can be done reasonably by comparing a ZIP Code to a table of data indexed by ZIP Code. Similarly, if a HAF participant is using a local geography that maps directly to some aspect of a postal address such as, city or town, the lookups can be done directly against the fact data.

If a HAF participant is using census tracts or a geography that cannot be directly extracted from the applicant's address, the HAF participant will need to geocode addresses. A variety of commercial and open-source products exist to geocode addresses, and all have tradeoffs in terms of accuracy, cost, and regional performance. The Federal Government provides two products that may help:

• The U.S. Census Bureau provides a Census Geocoder that returns census geographies for U.S. addresses.
• The U.S. Department of Transportation provides the National Address Database, which provides latitude and longitude coordinates for more than 61 million addresses; these coordinates can be located within a variety of geographies to lookup facts for addresses.

INTEGRATING THE PROXY WITHIN THE APPLICATION

The combination of the data collected and the technique for determining facts from the data about applicants by using their addresses creates a fact-specific proxy to use in the application process. Although the specific integration will depend on the proxy design and the application itself, generally a HAF participant will look up the income for the applicant based on where they live and compare it to the AMI limits prescribed by the program.

Using our example proxy from the beginning, we would:

1. Use an applicant's address to locate their census tract, then
2. Find the median income of homeowners in that tract.
3. If the area median income for the homeowner in the tract falls below the program’s AMI limit for the larger area, we can say that the proxy has returned that fact. This fact will support the use of self-attestation, and the applicant should be presented the option to provide a self-attestation for his or her income without providing additional documentation.