This article revisits the sectoral shifts hypothesis by examining unemployment fluctuations for 48 U.S. states over the period 1990:M01–2011:M12. We develop a panel approach that incorporates dynamics, parameter heterogeneity, aggregate factors, and cross‐sectional dependence (CSD). Our findings provide support for a positive and significant effect of the employment dispersion index on unemployment. This outcome is robust under alternative specifications and measures of employment dispersion. The empirical evidence corroborates the presence and relevance of CSD and heterogeneity among states. The results show that, once unobserved common factors and cross‐state heterogeneity are taken into account, labor reallocation has a significant effect on unemployment that is half the size of the estimate when cross‐sectional dependence is not taken into account. (JEL E24, E32, J21, R23, C23)