This paper documents, using county level data, some geographical features of the US business cycle over the past 30 years, with particular focus on the Great Recession. It shows that county level unemployment rates are spatially dispersed and spatially correlated, and documents how these characteristics evolve during recessions. It then shows that some of these features of county data can be generated by a model which includes simple channels of transmission of economic conditions from a county to its neighbors. The model suggests that these local channels are quantitatively important for the amplification/muting of aggregate shocks.(This abstract was borrowed from another version of this item.)