Not to be outdone by other Secretaries who are gaining a lot more public attention, on March 26, Commerce Secretary Wilbur Ross said “Hold my beer…” then announced that he was going along with Attorney General Jeff Sessions’ request to add a question about citizenship to the 2020 Census. The decision was announced despite concerns about the threat of a population undercount voiced by previous Census directors, the scientific and voting rights communities, and leaders in the public and private sectors.
Most people understand the threat of an undercount for the primary purpose for which the U.S. Census was designed: the apportionment of seats to the U.S. House of Representatives. The first Census, mandated by the Constitution and engineered by James Madison and Thomas Jefferson, also included some demographic questions, but it was the total population data that Jefferson used to apportion seats, of about 30,000 persons per seat.
According to estimates from Election Data Services, the 2020 Census data will impact over a dozen states by changing their Congressional delegations. As the table below shows, under current population projections, and with the size of the House capped at 435, seven states look to gain at least one seat, and Texas and Florida could gain more. Nine states would lose seats, based on relative population changes.
| States Gaining Districts (7) | States Losing Districts (8 or 9) | |
| Arizona +1 (from 9 to 10) | Alabama -1 (from 7 to 6) | |
| Colorado +1 (from 7 to 8) | Illinois -1 (from 18 to 17) | |
| Florida +2 (from 27 to 29) | Michigan -1 (from 14 to 13) | |
| Montana even or +1 (from At-large to 2) | Minnesota -1 or even (from 8 to 7 or no change) | |
| North Carolina +1 (from 13 to 14) | New York -1 (from 27 to 26) | |
| Oregon +1 (from 5 to 6) | Ohio -1 (from 16 to 15) | |
| Texas +2 or +3 (from 36 to 38 or 39) | Pennsylvania -1 (from 18 to 17) | |
| Rhode Island -1 (from 2 to 1) | ||
| West Virginia -1 (from 3 to 2) |
Apportionment is obviously very consequential for democratic representation (and partisan control of Congress), but there are additional, economic consequences of an undercount that are less well understood.
Andrew Reamer, Research Professor at George Washington University, has analyzed the fiscal impact of Census undercounts to states, and is producing a series of reports, Counting Dollars for 2020: The Role of the Decennial Census in the Geographic Distribution of Federal Funds. In this series, Reamer breaks down the various types of programs that receive federal funds, and estimates what per capita and overall costs would result from undercount estimates.
For example, in a recent analysis of five Health and Human Service programs that rely on the Federal Medical Assistance Percentage (FMAP) funding formula, Reamer showed how an additional 1% population undercount for Texas in 2010 would have resulted in a nearly $300 million loss in funding for these programs. That’s more than $1000 per person.
In upcoming reports, Reamer is working to analyze approximately 300 programs and $800 billion of funding that could be affected by a Census undercount. Moreover, he pointed out that there would be losses in both the public and private sectors that would be very difficult to quantify, given the importance of Census data that is linked to so many organizational decisions throughout the U.S. economy.
For example, consider all of the geographically specific demographic data linked to the identification of medically underserved areas, used to allocate support for medical programs, including physician training and funding for doctors to serve specific services. Errors in Census data, which are used to weight and design the sampling surveys that provide medical supply and demand information, could have far-reaching rippling effects throughout the economy.
“An accurate Decennial Census has a substantial impact on how the American economy functions,” noted Reamer, to “identify business opportunities, determine where to locate, and what to sell.” In addition to all of the market segmentation analysis that public-facing companies like Target and Walmart rely on to understand population characteristics and anticipate consumption patterns, businesses rely on Census-derived information (information that is not taken directly from the Census, but that builds on the basic population data provided by the Census) to make employment decisions, in order to invest in places with people who have the skills that they need to operate.
Small businesses use Census-derived data to make location and purchasing decisions, even if they don’t know it. Market analysis vendors repackage and augment this data when they provide consumer reports, consumer ratings, and the like to private and media clients. In short, there are very few areas of social or economic organization that are not impacted by the quality of Census data.
Professor Reamer’s next report will be a more complete listing of all federal financial assistance programs that rely on Census-derived data, hopefully released early this summer. You can follow the series at the George Washington Institute of Public Policy.