Do firm effects drift? Evidence from Washington administrative data

We study the time-series properties of firm effects in two-way fixed model popularized by Abowd et al. (1999) (AKM) using two approaches. The first—the rolling AKM approach (R-AKM)—estimates models separately for successive two-year intervals. second—the time-varying (TV-AKM)—is an extension original that allows unrestricted interactions year and indicators. apply to both approaches leave-out methodology Kline (2020) correct biases estimated variance components. Using administrative wage records from Washington State, we find, first, hourly rates are highly persistent with autocorrelation coefficient between 2002 2014 0.74. Second, R-AKM reveals cyclicality worker–firm sorting. During Great Recession variability increased, while degree sorting decreased. Third, misspecification standard resulting restricting be over time appears minimal.