We propose a probabilistic voting model to discuss how campaign spending affects electoral outcomes. In our model, private donations are restricted and campaigning is funded by a statutory allowance provided by the central government and by party contributions. Campaign spending increases exposure and thus the popularity of the candidates. We show that, when the variance of voter ideology is high, candidates that are ex-ante more popular or candidates with lower restrictions in the access to party contributions spend more in campaigning and win with a higher probability. 

To provide empirical support for our analysis, we compile a novel dataset about campaign spending covering all candidates (12 parties, independent candidates and coalitions) in 306 municipalities for 3 elections (2005, 2009 and 2013). Our identification strategy relies on the rules that define the statutory allowance provided by the central government to fund campaigning expenditures. Our results show that although the spending of the winner (often the incumbent) is not significant, campaign spending accounts for 8pp of the vote share of the runner-up in the election. We also show that spending of the two biggest parties in Portugal accounts for 6 to 9pp of their vote share.