Changes in Panel Bias in the U.S. Current Population Survey and its Effects on Labor Force Estimates

Gregory Erkens


The U.S. Current Population Survey (CPS) is a rotating panel survey of U.S. households that measures the labor force statuses of the non-institutional civilian population. Each panel is a random sample of households that rotates into the survey for 4 consecutive months, rotates out of the sample for the next 8 months, and then rotates back into the survey for a final 4 months. CPS microdata undergo several weight adjustments so that estimated totals for various demographic characteristics match projected population totals. The final adjustment for demographic information is called the Second Stage adjustment, and it adjusts for age, race, gender, and ethnicity. The final stage of weighting--Composite estimation--employs the correlation in overlapping panels between adjacent months to reduce the variation of key labor force estimates. In the Second Stage adjustment each panel can produce an estimate of employment and unemployment, and each panel exhibits certain tendencies. This paper looks at how those tendencies have changed over time, their effects on Composite estimation, and a possible method to adjust for those effects.