The History of the Unemployment Rate: How America Started Counting the Jobless
Before 1940: Counting by Decade, Arguing by Month
Through the worst years of the Great Depression the United States had no regular measurement of unemployment. The decennial census asked about work status, and a special unemployment census was attempted in 1937, but between those snapshots the country relied on estimates produced by trade unions, private research organizations, and state agencies using incompatible definitions. Published figures for the same year could differ by millions of people, and the disagreements were not merely technical: whether relief programs were working was a live political question that the available statistics could not settle. Retrospective estimates put the peak unemployment rate of the Depression at roughly a quarter of the labor force in 1933, but nobody at the time had a defensible monthly number.
1940: The Monthly Household Survey Begins
The breakthrough was methodological. Rather than counting everyone, statisticians applied the emerging science of probability sampling to interview a representative sample of households each month and infer the national picture from it. In March 1940 the Work Projects Administration launched the Sample Survey of Unemployment on exactly that basis. Crucially, the survey defined its terms operationally: a person was counted as unemployed only if they were without a job, available for work, and actively looking for work — a definition that separated the jobless from those not participating in the labor market at all. That activity-based definition, refined but not abandoned, is still the one used today, and it is the source of most public confusion about the statistic.
1942-1959: From the WPA to the Census Bureau and the BLS
Responsibility for the survey moved to the Census Bureau in 1942, and in 1948 it was renamed the Current Population Survey, the name it still carries. In 1959 the Bureau of Labor Statistics took over responsibility for analyzing and publishing the labor force data, with the Census Bureau continuing to conduct the interviews — the division of labor that persists today. The postwar period established the monthly rhythm that dominates economic news: a survey week containing the twelfth of the month, and a national release in the first days of the following month. The household survey was joined by a separate establishment survey of employers, which produces the payroll employment figure released in the same report from a completely different sample.
1962 and 1979: Two Reviews That Reshaped the Statistic
As the unemployment rate became politically consequential, it attracted formal scrutiny. In 1962 the President's Committee to Appraise Employment and Unemployment Statistics, chaired by the economist Robert Aaron Gordon and generally known as the Gordon Committee, delivered Measuring Employment and Unemployment, endorsing the basic approach while recommending a larger sample and improved procedures. In 1979 the National Commission on Employment and Unemployment Statistics, chaired by Sar Levitan, conducted another wide-ranging review. Both reviews confronted the same underlying tension: a single headline number cannot capture discouraged workers, involuntary part-time work, and underemployment, yet the public and the press demand one number. The answer both times was to keep the headline rate and publish supplementary measures alongside it.
1976 and 1994: The Alternative Measures and the CPS Redesign
In 1976 the BLS began publishing a family of alternative unemployment measures, then labelled U-1 through U-7, to show how the rate would look under broader or narrower definitions. In January 1994 a major redesign of the Current Population Survey introduced computer-assisted interviewing and revised questionnaire wording, tightening the definitions of discouraged and marginally attached workers. The alternative measures were reorganized at the same time into the current U-1 through U-6 series, with U-3 designated the official unemployment rate and U-6 the broadest. Since then the notable readings in the series include a postwar peak of 10.8 percent in late 1982, 10.0 percent in October 2009, and roughly 14.7 percent in April 2020 during the pandemic shutdowns — the highest monthly figure in the survey's history.