摘要：Recent nonparametric studies find that the regional distribution of unemployment rates is more dependent on geographic location (neighborhood effects) than on national factors (state effects). These results, however, do not control for stochastic dependency between neighborhood and state effects (joint effects) and do not provide any measure of their relative size. In this paper, we construct a simple measure that is invariant in space and across time and compare state, neighborhood, and joint effects in Europe and North America. We find that neighborhood effects are stronger than state effects in Europe, whereas in North America the size of the joint effect is such that one cannot say which effects prevail.
