Populous communities often prefer more government involvement than less populous communities, but does community size per se affect citizens' preferences for government? Endogeneity commonly prevents testing for causal effects because (i) people can select into communities while (ii) government structures can affect community size (e.g., by en- or discouraging migration and fertility decisions). This paper studies a plausibly exogenous setting from post-WWII Baden-Württemberg (located in Southern Germany), in which the French occupation zone prevented the entry of German expellees after 1945, whereas the U.S. occupied zone did not. Consequently, municipalities on the U.S. side, just across the border from the French zone, experienced large and relatively homogenous population shocks. Studying voting patterns in the 1949 national- and 1952 state-level elections for 828 municipalities, we find more populous municipalities systematically preferred the SPD (Sozialdemokratische Partei Deutschlands), the party advocating for greater government involvement in virtually all areas of policymaking, over the CDU (Christlich Demokratische Union), the major conservative party that emphasized free markets. Our results hold when accounting for a host of potential confounding factors, county-fixed effects, pre-WWII vote shares, employing fractional response models and alternative instrumental variable specifications. Our benchmark estimates imply that a one standard deviation increase in population size (equivalent to ≈4,000 citizens) raised the SPD vote share by more than 11 percentage points.