Voter turnout is an essential indicator of the quality of democracies (Schmitter 2004). However, voter turnout has declined worldwide from 76% in the late 1980s to 66% in 2011 and is even lower in local elections (Solijonov 2016). Since budget transparency increases participation (Benito and Bastida 2009), understanding the effects of partisanship on local public expenditures may invigorate voter turnout.

Existing literature has shown mixed evidence of partisanship effects on local budgets. Some authors emphasize the role played by political parties (Pettersson-Lidbom 2008; Gerber and Hopkins 2011; de Benedictis-Kessner and Warshaw 2016), while others highlight the lack of partisanship effect on public expenditures (Ferreira and Gyourko 2009; Leigh 2008). Ferreira and Gyourko (2009) and de Benedictis-Kessner and Warshaw (2016) reach contradictory results analyzing U.S. mayoral elections using different lower bounds on city size. If the impact of political parties differs by city size, heterogeneity may explain the conflicting results shown in the literature.

In this paper, I investigate the heterogeneous influence of city size to clarify its effects on partisanship. First, I analyze the impact of political parties on public expenditures on French municipalities with a regression discontinuity design. I find strong effects of partisanship; left-wing mayors invest 65 euros more per capita than their right-wing counterparts. Second, I estimate partisanship effects on two clusters of municipalities built with a K-means algorithm. I find similar partisanship effects on capital outlays in small and large municipalities and impacts on current expenditures only in small cities.