Skip to main content
off the charts

House Votes to Eliminate Local Data

Republicans in the House inexplicably voted this week to defund the American Community Survey (ACS), the nation's main source of state and local data on affordable housing, household income, poverty, race, state-to-state migration, immigration, education level of the workforce, types of disabilities of local residents, and scores of other major topics.

The federal government uses the data to distribute more than $400 billion in federal formula funds each year, and the information helps communities and businesses decide where to build new roads, bridges, schools, homes, and stores.  Business groups including the U.S. Chamber of Commerce, National Retail Federation, International Council of Shopping Centers, and National Association of Home Builders consider the data vital.

The House proposed no alternative to collecting these data.

The ACS replaced the traditional long form of the decennial census.  Several of the questions date back to the 1850s.

Some who voted to end the ACS funding objected to it as an example of government intrusion because its questions go beyond the Constitution's required enumeration of inhabitants.  But the census has always asked more than that, all the way back to 1790.  Federal agencies have determined that every ACS question is necessary to carry out legislation that Congress itself has passed.

The key difference between the ACS and the old decennial census is timing.  Instead of blitzing households all at once every ten years, the ACS continuously contacts different households — about 3 million a year — and releases updated findings annually.  To get down to the neighborhood level, the Census Bureau averages together five years of ACS data to generate a sample size that is almost as large as the one the old census long form collected.

Some proponents of the ACS cut cited a desire to save money.  But if the Census Bureau has to return to the old long-form approach to collect the same data, the effort may end up costing more, partly because Census would need to hire and train a new, inexperienced staff every ten years.

The bigger cost by far would come from losing the information entirely.