We propose a new source of cross-sectional variation that may identify causal impacts of government spending on the economy. We use the fact that a large number of federal spending programs depend on local population levels. Every ten years, the Census provides a count of local populations....