统一的人口与劳动力统计数据

Federal Reserve Board, Washington, D.C.ISSN 1936-2854 (Print)ISSN 2767-3898 (Online) Harmonized Population and Labor Force Statistics John Coglianese, Seth Murray, and Christopher J. Nekarda 2025-057 NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminarymaterials circulated to stimulate discussion and critical comment.The analysis and conclusions set forthare those of the authors and do not indicate concurrence by other members of the research staff or theBoard of Governors. References in publications to the Finance and Economics Discussion Series (other thanacknowledgement) should be cleared with the author(s) to protect the tentative character of these papers. Harmonized Population and Labor Force Statistics* John Coglianese†, Seth Murray‡, and Christopher J. Nekarda§ Board of Governors of the Federal Reserve SystemJuly 2025 Abstract The official labor force statistics often exhibit discontinuities in Jan-uary, when updated population estimates are incorporated into theCurrent Population Survey (CPS) for the current year but are not re-vised backward through history. We construct harmonized popula-tion estimates spanning five decades and produce new weights for theCPS microdata that are benchmarked to these estimates. Using theseweights, we estimate harmonized labor force statistics that reflect thelatest available information about the population and its character-istics. The harmonized labor force series are free from the disconti-nuities in the historical data and show a notably larger labor forceshortfall in the post-pandemic period. JEL codes: C8, E24 Keywords: population, labor force, employment, unemployment, im-migration, CPS 1. Introduction The Bureau of Labor Statistics (BLS) publishes timely statistics about theU.S. labor market that receive wide attention each month. Many of theseindicators, including the unemployment rate and the labor force participa-tion rate (LFPR), are derived from the Current Population Survey (CPS),a monthly survey of about 60,000 households. Statistics from this surveycan capture the overall labor market because individual survey responsesare weighted to be representative of the demographic and geographic com-position of the U.S. population. However, since the population cannot becounted in real time (except for once every 10 years in the decennial cen-sus), the weights are constructed to match populationestimates, which canrevise when underlying source data are updated. Each January, the BLS incorporates updated population estimates into theCPS for the current year, but does not revise the official household surveyestimates back in history. When the revisions shift the composition of thepopulation across demographic groups whose labor market outcomes differ,such a shift can result in large discontinuities between the updated labormarket statistics for the current year and the out-of-date statistics for pre-vious years, potentially confounding statistical analyses and assessments ofthe labor market. This issue affects not only the official statistics publishedby the BLS, but also statistics that researchers calculate from CPS micro-data. In this paper, we introduce a methodology for estimating CPS–based laborforce statistics that are “harmonized” — made comparable over time — overfive decades to consistently reflect the latest available data on the popula-tion. Our approach involves assembling harmonized population data at thedemographic group level, reweighting CPS microdata to match these tar-gets, and then computing time-series estimates from the reweighted micro-data. Since our method closely follows the BLS’s estimation process, ourharmonized labor force statistics can be interpreted as close approximationsof the values that the BLS would have produced if it had been able to usethe latest population data when it originally published its statistics.1We provide harmonized estimates for the unemployment rate, the LFPR, andother labor force statistics, along with harmonized microdata weights thatresearchers can use to reproduce any CPS statistic adjusted for populationrevisions. We plan to update both the microdata weights and the harmo-nized time series annually to reflect each new vintage of population esti-mates from the Census Bureau. By not revising the historical time series of labor force statistics to reflectnew population data, the BLS’s standard practices often introduce discon-tinuities in the time series between each December to January. These dis-continuities are evident in the published series for the civilian noninstitu-tionalized population (CNP) aged 16 years or older, shown by the solid blueline in figure 1. For example, in January 2012, the CPS incorporated pop-ulation estimates derived from the 2010 Census, which resulted in a jumpof 1.7 million from December 2011 to January 2012, even though the typi-cal pace of change before and after that period was only about 200,000 per month. Similar