Recent harvest declines in the Western United States have focused attention on the question of economic impacts at the community level. The impact of changing timber-related economic activity in a given community on other local activity and the general economic health of the community at large has been a persistent and often contentious issue in debates surrounding forest policy decisions. The economic base hypothesis, in which changes in local export-related economic activity are assumed to cause changes in economic activity serving local demand, is a common framework for understanding impacts of forest policy decisions and forms the basis of models commonly used to provide estimates of expected local impacts under different policy options. This study uses community-specific, time-series employment data to test the economic base hypothesis in the small, semi-isolated communities of southeast Alaska. Estimates were derived for each of 15 communities. Export-related activity was not found to cause changes in economic activity serving local demand for the average community. However, the results indicated statistically significant differences among communities in their response to shocks in export related activity. The implications of these results for policy, and for the theory and practice of modeling economic impacts at small spatial scales, are explored in the final sections of this study. Specifically, secondary economic impacts cannot be taken as a foregone conclusion in policy analysis, and the fundamental assumptions of static impact modeling approaches deserve greater scrutiny.