Political economists contest whether voters can restrain government spending via the ballot box, or whether constitutional constraints are required to restrain spending. We engage this debate by studying a heretofore unclassied type of constitutional limit: a limit placed by a city on its own ability to tax or spend. We pose a theoretical model that explains the existence of such limits as a response to both a population divided in taste for public goods, and the absence of institutional substitutes. Our survey results show that at least 1 in 8 cities a limit on its ability to tax or spend. Consistent with our model, we nd that when the non-wealthy proportion of the population is suciently large relative to the wealthy proportion, limits are more likely to be adopted. In addition, we find empirical support for the importance of institutional substitutes: cities with less restrictive state regimes, and cities in metropolitan areas with fewer jurisdictions are both more likely to adopt a limit. Our model suggests that these limits weakly constrain revenues, and we find empirically that municipal revenue growth declines after limit adoption by 16 to 22 percent.