HUD calculates median incomes as the basis of its income limits that are used to determine eligibility for various HUD programs. Medians are not directly used in HUD programs and are calculated at the family level only, not the per person level as is done for income limits. The average family size is over 3, so, by convention, HUD equates the median family income for an area with a four-person family for the purposes of calculating income limits.

HUD uses the Section 8 program's Fair Market Rent (FMR) area definitions for its median incomes, which means that medians are developed for each metropolitan area, parts of some metropolitan areas, and each non-metropolitan county. For FY 2019, there are two changes to the geographic area definitions used last year. First, the two counties (islands) that comprise the Kahului-Wailuku-Lahaina, HI will no longer have medians calculated separately by county; the names of these areas will change from the island names (Kalawao and Maui County) to the metropolitan in FY 2019. Second, Enid, OK is a new metropolitan area, formed out of nonmetropolitan Garfield County.

HUD uses median family income data (as opposed to median household income data) from the American Community Survey (ACS) for all areas in the United States. For Puerto Rico an annual survey is also conducted and is called the Puerto Rico Community Survey (PRCS). The FY2019 median incomes use survey data from the 2016 ACS and PRCS. The 2016 data are inflated using a Consumer Price Index (CPI) forecast from the Congressional Budget Office (CBO) through the midpoint of FY 2019. Island areas (Guam, American Samoa, Northern Marianna Islands, and the Virgin Islands) use income data from a census conducted in 2010 of the previous year's (2009) income data, augmented by the change in the national income between 2009 and 2016 (from the ACS). The same CBO forecast is then applied from mid-2016 to the mid-point of the fiscal year, April 2019.

As mentioned above, the FY 2019 median incomes incorporate the 2016 ACS data into the calculation process. Specifically, for each metropolitan area, subarea of a metropolitan area, and nonmetropolitan county, HUD determines if a statistically valid one-year ACS income estimate is available. If one-year data is not available, then statistically valid five-year ACS data (data collected from 2012 through 2016) is used. There are cases where statistically valid five-year ACS data is not available. In those cases, an average of at least two of the past three years of income estimates is used. If at least two years of statistically valid income data are not available, the one-year state nonmetro median is used¹.

Footnotes:

¹For metropolitan subareas without statistically significant five-year ACS data there is no averaging with past years; the larger metropolitan area is used, which is one-year ACS data.

For the FY 2019 median incomes, HUD requires that the margin of error be less than half of the estimate and that the survey median is based on at least 100 responses (as identified by a count indicator value of 4 or more in HUD's special tabulations of ACS data). If the current year estimate does not meet both conditions, the previous years' estimates must meet the margin of error condition to be used in averaging.

HUD uses a CPI forecast from CBO to inflate the 2016 ACS data to the mid-point of FY 2019. The CBO projection of fiscal year CPI, published in January 2019, is used to inflate the 2016 data.

Median family incomes start with the development of median incomes for the nation (with national metropolitan and nonmetropolitan median incomes), for each state and territory (again including national metropolitan and nonmetropolitan median incomes), and for each metropolitan area and nonmetropolitan area using the FMR area definitions for the United State and its territories.

The major steps for calculating medians² are detailed below:

HUD uses 2016 ACS or PRCS median family incomes as the basis for FY 2019 medians for all areas designated as Fair Market Rent areas in the US and Puerto Rico. In areas where there is a statistically valid survey estimate using 2016 one-year ACS or PRCS data, that is used. If not, statistically valid 2016 five-year data is used. Where statistically valid five-year data is not available, HUD will average the valid income estimates from the previous three years of ACS or PRCS data or for two of these three years. This data from the current 2016 five-year data will be considered valid if the margin of error of the estimate is less than one-half of the estimate.

This same test will be applied to the 2015 five-year data and the 2014 five-year data, which will be inflated to 2016 using the change in national CPI calculated between 2014 or 2015 and 2016.

Metropolitan subareas, HUD Metro FMR Areas (HMFAs), do not use averaged data from the current and two previous years (adjusted to the current year) if the five-year data is not statistically valid. These subareas use the 2016 ACS data for the larger metropolitan area.

For all places in the US and Puerto Rico:

All estimates (using either one-year data or five-year data) are then trended from 2016 to April 2016, the midpoint of FY 2019.

For the non-Puerto Rico Insular Areas of the United States,³ which currently lack the annual survey of ACS or PRCS, 2010 Decennial Census data were used for the first time in the FY 2016 medians and income limits. This continues to be the basis of the FY 2019 medians and income limits. National ACS income changes are used to update 2010 Decennial Census data to 2016 and then the same CPI forecast trend factor is applied to bring the data forward to the midpoint of FY 2019.

Attachment 1 shows the distribution of changes in median incomes between FY 2018 and FY 2019 for each state and the United States, overall. The distribution of changes is also shown separately for metropolitan and nonmetropolitan portions of each state and the United States.

Attachment 2 shows the median incomes by state and for the United States and the State Metropolitan median incomes and the State Nonmetropolitan medians.

Footnotes:

²Family refers to the Census definition of a family, which is a householder with one or more other persons living in the same household who are related to the householder by birth, marriage, or adoption. The definition of family excludes one-person households and multi-person households of unrelated individuals. ³The areas without ACS coverage are the U.S. Virgin Islands, Guam, American Samoa, and the Northern Marianas Islands. Puerto Rico is covered by the ACS-equivalent Puerto Rico Community Survey.