It is generally accepted that regional labor markets are characterized by strong interdependencies. However, only few studies include spatial elements to their estimations. Using the model framework proposed by Cliff and Ord (1973, 1981) and the estimation technique proposed by Kelejian and Prucha (1998), we estimate a spatial time series model for the Swiss cantonal unemployment rates on a quarterly level. Our model contains a spatial lag in the level and in the error term, as well as further exogenous explanatory variables. While both spatial lags turn out to be significant in our estimations, the dependency in the error term seems to be even stronger than the one in the level.