Since 2010 and before the pandemic hit, the share of households refusing to participate in the Current Population Survey (CPS) tripled. We show that partially-responding households - households that respond to some but not all of the survey's eight panels - account for most of the rise. Leveraging the labor force status of partially-responding households in the months surrounding their non-response, we find that rising refusals artificially suppressed the labor force participation rate and employment-population ratio but had little discernible effect on the unemployment rate. Factors robustly correlated with state-level refusal rates include a larger urban population, a smaller Democratic vote share (our proxy for sentiment towards government), and the economic and social changes brought about by manufacturing decline.
Current Population Survey Non-Response Unemployment Labor Force Participation Employment-Population Ratio