The U.S. census kicks into full gear on April 1, 2020, and states are paying close attention. Not only will the census determine the distribution of congressional seats, but the data collected will have a profound impact on state budgets. How? Because of the role that the census plays in creating the datasets and statistical indicators used by many federal grant programs—such as Medicaid, the State Children’s Health Insurance Program, and the Women, Infants and Children nutrition program (WIC)—to apportion funding among states, local governments, and other grantees. Such programs made up 32% of state revenues in 2017.
According to Dr. Andrew Reamer, who leads the Counting for Dollars initiative at the George Washington Institute for Public Policy, 316 federal programs relied on data derived from the census to distribute some or all of their funds in 2017. Approximately two-thirds of these programs flow at least partially to state governments.
Many states—such as Colorado and Illinois—are investing serious dollars to make sure their residents get counted. In fact, according to The New York Times, 26 states are spending close to a third of a billion dollars to promote accurate response rates when the census count begins in April. “We’re seeing more and more states encouraging their residents to participate in the census,” said Wendy Underhill, director of elections and redistricting at the National Conference of State Legislatures. “This push is much more about ensuring that each state gets its fair share of federal funding—much of which is based on formulas that use census data–than it is about whether a congressional seat could be at stake.”
But understanding how the census count will affect federal funding to any given state—much less predicting the amount of funding that a state might gain or lose—is difficult and requires a nuanced understanding of how each federal grant uses demographic data to allocate funds.
How the Census Affects Grant Funding
The census collects data on population and housing that directly affect only a relatively small number of federal grants. But these data are also used to create over 52 datasets and statistical indicators, such as income measures, poverty thresholds, and the consumer price index. These indicators are used in turn by many federal grant programs that employ statutory formulas to distribute funding. Importantly, the impacts of the count will be felt for years to come. Every year, many federal grant programs allocate money based on indicators that rely on updated annual population estimates that use the census as the baseline and then adjust for factors such as reported births and deaths. A person not counted in 2020 is a person “lost” for the next ten years.
The census can affect funding in various ways, depending on the structure of the program. For some programs, a statutory formula uses census data to determine the program’s overall annual funding level. For others, a formula distributes program funding among the grantees. Still others use census data to determine federal matching rates, reimbursement rates, or other program features that affect funding to states.
There are hundreds of federal grant programs, and many do not use formulas to distribute funding at all. But because the largest grants are all formula-based, the share of total federal grant funding potentially affected by the census is significant. Ninety-four percent of federal grant dollars to state and local governments in 2017 were allocated by formula, according to the nonprofit research organization Federal Funds Information for States (FFIS). Of all the programs tracked by FFIS, Pew found that those using census data provided $580 billion to states in 2017, or 95% of the total.
Nothing is Simple
But not all formula grants are affected by the census to the same degree. For some programs, the entire funding for a given year may be determined by census-derived formula factors; for others, census data may only affect a small portion of funding. For example, the amount of funding a state receives for the WIC program is unlikely to change much based on the census count. That’s because WIC only uses census data to distribute funds when total appropriations increase from one year to the next—and even then, only the increase is distributed using census data.
By contrast, a state’s entire apportionment for the Social Services Block Grant, a flexible program that allows states to tailor social services to the needs of their residents, is based on the state’s share of the national population, as determined annually by the Census Bureau.
Census-related changes to Medicaid, which is larger than all other federal grants combined, could have the most profound impact on state budgets. Medicaid funding is based on reimbursements for costs incurred; but the federal government’s reimbursement rate is determined by an indicator called the Federal Medical Assistance Percentages (FMAP), which uses state and federal per capita income calculated using the census-derived population count. The FMAP formula compares each state’s per capita income relative to the US per capita income (over a three-year average), providing higher reimbursement rates to lower-income states and lower reimbursement rates to higher-income states. Thus, a population undercount in a state could result in per-capita income being overrepresented for a state, which in turn could distort a state’s FMAP and cause a state to lose federal Medicaid dollars.
A 2018 analysis by Dr. Reamer found that an undercount of one percent of a state’s population in the could decrease its federal Medicaid reimbursement rate by as much as one dollar for every $100 spent by the state in most states.
Ultimately, the wide variety of ways in which the census can affect federal funding precludes an accurate forecast of its effect on any particular state’s budget; but it also gives testament to just how far-reaching that impact may be.
Anne Stauffer is a director, Rebecca Thiess is an associate manager, and Laura Pontari is an associate with The Pew Charitable Trusts’ fiscal federalism initiative.