Abolish the census
August 15, 2018
This article originally appeared in the Washington Examiner on August 13, 2018.
In less than two years, the federal government will embark on a time-honored, decennial ritual: the U.S. census. One could be forgiven by thinking, that, 230 years and nearly two dozen censuses later, Washington, D.C. knows how to count people. Yet, despite a declining number of individuals per household and the rise in low-cost digital correspondence, the cost of the census is rising far above the rate of inflation. According to the Government Accountability Office, “the average cost for counting a housing unit increased from about $16 in 1970 to around $92 in 2010.” The report further notes that, over the past three years, the U.S. Census Bureau has underestimated how much it will cost to conduct the 2020 census.
The U.S. Census Bureau now estimates that the undertaking will cost $15.6 billion, up from their 2015 estimate of $12.3 billion, a 27 percent increase. Given the bureau’s cost escalation and well-documented accuracy problems, it may be time to look to another approach to count the nation’s citizens. By ditching a dedicated decennial census, the federal government can save taxpayers billions of dollars while obtaining vital information elsewhere.
In 1790, the Revolutionary War was in the rearview mirror and the U.S. Constitution had recently been ratified. To ensure that “a more perfect union” could have adequate representation, the framers inserted a clause in the document mandating “The actual Enumeration shall be made within three Years after the first Meeting of the Congress of the United States, and within every subsequent Term of ten Years.” Appointed marshals were tasked with going district by district, household by household, and counting persons (excluding, of course, Indians not taxed).
Since the ratification of the Constitution and the subsequent 1790 census, the “enumeration” has been dutifully carried out every ten years with increases in the number and scope of questions asked. While this requirement hasn’t been challenged in Congress and the courts, other countries in recent decades have begun to see a dedicated census as redundant. A United Nations survey on the enumeration methods of different countries notes that more than a dozen European countries, including the Netherlands, Norway, Germany, and Switzerland, link their already-existing administrative records to create census-type records.
This sounds like a reasonable approach, if the quality of the data is thorough enough to provide decent record-keeping. In 1995, the National Research Council considered an administrative records-based system as an alternative to the current process, noting that a “high proportion of the U.S. population is included in one or more existing administrative records…the coverage… may well expand in the future.” The alleged problem is the allowable scope, since “no federal or state administrative system now possesses race and ethnicity data for the U.S. population.”
But including questions on race and ethnicity merely undermine the census’s original purpose of determining governmental apportionment and representation.
But are birth, death, and tax records up to the task of substituting for a household-by-household enumeration? Ideally, IRS data can be used to track state residence/migration and combined with SSA data to keep tabs on births and deaths. Using tax data can prove tricky, since a large portion of the population has no tax liability. The IRS, however, has information documents on non-filers in their Information Returns Master File. IRS researchers have previously used sampling and merging techniques to produce state-by-state population estimates, with encouraging results. They find, “the overall correspondence between census and administrative records data is extremely good,” with results for a majority of states either within three percentage points of the census estimate or closer to adjusted census estimates than the original, official count.
In important respects, an administrative records-based census might actually run smoother than a census. Taxpayers can look forward to saving tens of billions of dollars each decade on a needlessly intrusive process. While constitutional questions may arise over “enumeration,” counting individuals via SSA and IRS records should pass muster as just another way of counting individuals and households.