An Evaluation of HUD’s Accuracy in Calculating Income Limits

An Evaluation of HUD's Accuracy in Calculating Income Limits

Jess Remington, Economist

Every year, HUD’s Program Parameters and Research Division (PPRD) calculates Income Limits based on annual estimates of Median Family Income (MFI). Income Limits are used to determine family eligibility for many HUD programs, such as the Housing Choice Voucher, Section 8 Project-Based Rental Assistance, and Public Housing programs. In addition, Income Limits are used to determine the maximum possible rents that may be charged to families in some programs, including properties financed with low-income housing tax credits.

HUD's Office of Policy Development and Research (PD&R) has periodically reviewed the accuracy of the Income Limits methodology, although it has not done so in more than a decade. This analysis aims to update PD&R's understanding of the accuracy of the Income Limits methodology and produce recommendations for improving accuracy. When HUD underestimates Income Limits, families are inadvertently excluded from participation in voucher programs or other programs that rely on Income Limits for eligibility. By contrast, when HUD overestimates Income Limits, it expands eligibility to families with relatively higher incomes. Because the annual budget established by Congress for most means-tested programs is fixed, expanded eligibility does not increase the amount of available resources; it simply reduces the likelihood that lower-income households can secure benefits. Therefore, estimating Income Limits accurately is important to ensure that allocated funds effectively reach the intended recipients.

In this article, I analyze the accuracy of HUD's calculations of Income Limits from fiscal year 2008 through 2022. Income Limit and Median Family Income estimates should closely track that year's actual reported income from the American Community Survey (ACS), with an understanding that some divergence may be intentional and desirable for policy reasons. For example, Income Limits are designed to moderate the effects of unusually high economic volatility and assist residents of areas with high housing costs.

This analysis produced four main findings:

• The estimates for MFI and Very Low-Income Limits (VLILs) are roughly accurate; historically, they have fallen within 10 percent of actual income, and their accuracy seems to have improved over time.

• Inaccuracies in MFI estimates likely result from differences between annual income growth and HUD's sources for its inflation factors (past income growth and projections of overall inflation).

• When compared with actual income, HUD's estimates of VLILs tend to exceed the statutory target of 50 percent of median family income. This finding is largely due to the intentional policy choice to adjust these estimates for high housing costs.

• Areas subject to an income cap are more likely to overestimate income. Without the cap, this overestimation would be even larger. This finding supports HUD's decision to implement a new Income Limits methodology for fiscal year (FY) 2024, which limits annual increases to 10 percent.

Background

Median Family Income: The U.S. Census Bureau measures MFI annually as part of the ACS. MFI represents the middle point of the income distribution for families. Because the ACS does not publish its estimates until the end of the year following each annual survey, HUD calculates an estimated MFI — upon which it bases its annual Income Limits — using data from previous ACS releases and trends it forward using inflation factors.

Income Limits: Income Limits are set at 30 percent (Extremely Low-Income), 50 percent (Very Low-Income), and 80 percent (Low-Income) of the projected median family income for an area.

MFI calculations: HUD produces its MFI estimates using a prior year’s 1-year or 5-year ACS MFI estimates and checks them for statistical reliability. (For example, HUD used 2022 ACS estimates to produce the FY 2024 calculations.) HUD then forecasts these ACS data with an inflation factor calculated by dividing the Congressional Budget Office’s (CBO’s) projection of the national Consumer Price Index (CPI) for the fiscal year of the MFI estimate by the annual CPI of the ACS base year.

Other Considerations

Very Low-Income Limit Adjustments

VLILs are defined as 50 percent of the area's Median Family Income, although HUD may adjust this percentage. Possible adjustments include the high housing cost adjustment and low housing cost adjustment, which increase or decrease VLILs based on the ratio of housing costs to area median income; the state non-metro median family income adjustment, which ensures that VLILs do not fall below 50 percent of the state's nonmetropolitan area MFI; and ceilings and floors, which ensure that VLILs do not excessively increase or decrease from the previous year.

Changes to calculations

HUD has changed aspects of the Income Limits methodology over the years. Through FY 2014, HUD based the inflation factor on CPI and a multiyear growth rate in national income. Beginning in FY 2015, the inflation factor calculation changed to instead use CBO projections of the CPI for the upcoming fiscal year.

Another notable change was the introduction of the ceilings and floors adjustment in FY 2010. HUD eliminated its "hold harmless" policy that year, which had stipulated that Income Limits could not fall from the previous year, and in its place established a floor (limits cannot fall to a level that is more than 5 percent less than that of the previous year) and a ceiling, or cap. The cap methodology stipulated that Income Limits cannot increase by more than 5 percent or by twice the percentage change in the national median family income, whichever is greater. Starting in FY 2022, HUD switched from the inflated national MFI to the uninflated national MFI to mitigate the effects of large year-to-year changes in inflation rates in calculating the cap. In FY 2024, HUD modified the determination of the cap to be either 5 percent or twice the change in the national median family income, whichever is greater, but the increase cannot exceed 10 percent.

Methodology

I collected 1-year ACS MFI values from 2008 to 2022. ACS 1-year estimates are available for geographic areas with a minimum population of 65,000. I adjusted the income data to the fiscal year with a deflation factor adjustment, which is the average quarterly CPI for the fiscal year divided by the annual CPI for the calendar year. I then calculated the annual mean percentage error by measuring the difference between projected median income (HUD's MFI values for each fiscal year) and actual median income (the deflated ACS MFI values). I also calculated the annual mean percentage error by measuring the difference between the projected Very Low-Income Limit (HUD's calculated VLIL) and the "actual" Very Low-Income Limit (50 percent of the deflated ACS MFI). I then weighted the resulting means by population. Finally, I analyzed the differences in accuracy between areas that received the high housing cost (HHC) adjustment and the income cap versus those that did not.

Results

1. HUD’s MFI and VLIL estimates are roughly accurate: they typically fall within 10 percent of actual household income and have increased in accuracy over time.

Since FY 2008, the average percentage error in HUD’s MFI calculations for all HUD-defined Metropolitan Fair Market Rent (FMR) areas ranged from –8 percent to +7 percent. In FY 2022, HUD underestimated MFIs by an average of 1% percent.

The average annual percentage error for VLIL estimates ranged from –2 percent to +12 percent. In FY 2022, HUD overestimated VLILs by an average of 2 percent.

Notably, discrepancies between actual and estimated VLILs were greater during years with higher volatility in price levels, such as the periods following the Great Recession and the COVID-19 pandemic. CBO's CPI estimates were also less accurate during these years. The average annual percentage error appears to have declined since FY 2013, although it is unclear whether this trend will persist during future periods of economic volatility.

2. Inaccuracies in HUD's MFI estimates are likely due to divergences between actual annual income growth and HUD's sources for its inflation factors.

Historically, MFIs have followed a trend of overestimating incomes in years during which incomes fell relative to either past income growth or to projections of overall inflation, and underestimating incomes in years during which incomes rose faster than inflation. In the aftermath of the Great Recession, from FY 2009 to FY 2012, HUD overestimated annual MFI by an average of 6.2 percent or $3,845, as the annual rate of growth for incomes fell relative to the average growth rate from the previous 5 years. The trend started to reverse as the income growth rate increased in FY 2013 and FY 2014. In FY 2015, HUD changed the MFI calculation methodology to use future projections rather than the average of previous years and switched from using the growth rate in national income to calculate the inflation factor to using CBO projections of the CPI, which captures the rising costs of goods and services. From FY 2015 to FY 2021, HUD underestimated annual MFI by an average of 5.5 percent, or $4,608, as the annual rate of income growth increased relative to overall inflation. This divergence began closing in FY 2022 in the aftermath of the COVID-19 pandemic as inflation outpaced income growth.

3. VLILs tend to overestimate income, in part because of intentional policy choices such as the HHC adjustment and the state nonmetropolitan median adjustment.

HUD overestimated VLILs in 11 of the past 15 years. In FY 2022, HUD overestimated VLILs by an average of 2.4 percent. The average overestimate for all years was 3.3 percent.

This overestimation is partly by design. HUD adjusts the VLIL to improve the usefulness of the Income Limits program. One such adjustment is the income floor, which mandates that Income Limits decrease by no more than 5 percent from the previous year's limit. This mandate provides stability for families receiving housing assistance by reducing the number of families that will lose their eligibility each year. Although the income floor does contribute to VLIL overestimates (particularly during years of income deflation), it does not appear to be a significant contributor.

Another adjustment is the HHC adjustment, which adjusts VLILs upward for areas with housing costs that are unusually high relative to median income. This adjustment is intended to address geographic differences in affordability, in which families that earn incomes above the typical low-income threshold might still struggle to afford housing in certain areas. The HHC adjustment explains a significant proportion of the discrepancy between actual and estimated incomes. Since FY 2008, HUD has overestimated VLILs in areas with high housing costs by an average of 20 percent. This rate is significantly higher than the 3 percent overestimation rate for the nation as a whole and the 0.9 percent overestimation rate for areas without high housing costs. The impact of the HHC adjustment is particularly significant because many of the most populous U.S. metropolitan areas qualify as areas with high housing costs.

4. Areas that receive a cap are more likely to overestimate income. This finding supports HUD's decision to apply an additional 10 percent cap for FY 2024.

HUD is significantly more likely to overestimate income for areas that receive a cap on year-to-year changes in Income Limits than it is for areas that do not receive a cap. Since FY 2010 — the first year the cap was applied — HUD overestimated capped areas by 7.6 percent compared with 1.2 percent for noncapped areas. This result may be expected because capped areas are more likely to receive the HHC adjustment.

The cap, which stipulates that Income Limits cannot increase by the higher of 5 percent or twice the change in national median family income, has become increasingly relevant. From FY 2010 (which is the first year that the cap was implemented) to FY 2018, no more than 20 percent of HUD FMR areas received the income cap. This percentage increased dramatically following the COVID-19 pandemic because of significant inflation in housing costs, consumer goods, and the national median family income. In FY 2022, 55 percent of HUD FMR areas received the income cap, and in FY 2023, 85 percent of HUD FMR areas received it. The cap amount also was higher than usual during this period, at 11.89 percent in FY 2022 and 5.92 percent in FY 2023.

Considering the continued high rate of inflation, the income cap for FY 2024 would have reached a record high of 14.8 percent under the FY 2010 methodology. HUD's modification to the income cap adjustment for FY 2024 capped the increase at 10 percent. We can reasonably expect that the 10 percent cap will help improve HUD's accuracy in estimating Income Limits because the VLILs for capped areas would be even higher without the cap applied, capped areas constitute an increasingly large percentage of the United States, and capped areas continue to significantly overestimate income.

I also investigated whether implementing a 10 percent cap would significantly limit the number of families that are eligible for vouchers or public housing. I found that such a cap would not limit eligibility, because 98 percent of voucher and public housing households earn incomes that are substantially lower than the current Low-Income Limit for their area. Specifically, only 1.3 percent of households earn incomes within 5 percent of Low-Income Limit eligibility, and only 1.6 percent of households earn incomes within 10 percent of Low-Income Limit eligibility. This finding is true for both tenant-based and project-based voucher households as well as recipients of public housing.

Recommendations

• HUD could consider using an alternative data source to calculate the inflation factor for MFI. Rather than using the CPI, HUD might consider using a source that more closely approximates income growth, such as CBO's projections for the Employment Cost Index.

• HUD could follow up on the implementation of the 10 percent cap on Income Limit increases with an assessment of how the cap affects the accuracy of calculated Income Limits.