What drives governments with similar revenues to publicly provide very different amounts of goods for which private substitutes are available? Key examples are education and health care. This paper compares spending by Brazilian municipalities on pre-primary education--a good that is also provided privately--with spending on public infrastructure like parks and roads, which lacks private substitutes. Panel data from 1995-2008 reveal how the distribution of income affects public investment. Revenue is endogenous to investment outcomes, and the analysis addresses this problem by exploiting a 1998, nationwide education finance reform and several revisions to the policy. The author constructs a variable that captures exogenous variation in revenue generated by nonlinearities of the law to instrument for observed revenue. Municipalities with higher median income and more inequality are less likely to allocate revenue to education or to expand pre-primary enrollment. They are more likely to allocate revenue to public infrastructure. There is suggestive evidence that this occurs for two reasons, hypothesized in two separate literatures. In rich and unequal municipalities, fewer total people support public education spending (the collective choice channel), and also, any given poor person wanting public education has less influence over policymakers there (the political power channel). Appendices include: (1) Appendix Tables; (2) Institutional Detail: Pre-Primary Education in Brazil, FUNDEF, and FUNDEB; (3) Mathematical Appendix; and (4) Construction of Simulated Instrumental Variables. (Contains 15 tables, 5 figures and 45 footnotes.)