A TIME-SERIES ANALYSIS OF DISAGGREGATE UNITED-STATES UNEMPLOYMENT

This paper addresses an issue that has been neglected in the vast array of literature focusing on the determination of and differences in regional unemployment rates. The issue of whether causal relationships exist between disaggregated unemployment rates is considered. A multivariate vector autoregression model is estimated separately for unemployment rates of ten states and nine regions. The resulting impulse response functions indicate that significant causal relationships exist between unemployment rates of the states and regions analyzed. The results suggest that a mechanism allowing for causal relationships between disaggregate unemployment rates should be incorporated into future analyses. © 1991 Louisiana State University Press.