It has often been argued that the size of the public sector tends to increase with the number of parties in government. This is allegedly so because the partisan fragmentation of the cabinet encourages each party to channel public funds to narrow segments of the society as parties can externalise most of the costs of targeted spending. Moreover, individual parties are not held accountable for the overall state of the public economy but for the amount of targeted spending they deliver to their target groups.
In my paper, I reassess the empirical relevance of such theoretical assertions. In particular, I argue that the growth of the public sector is driven by the interaction of corruption and partisan fragmentation rather than the number of parties as such. In other words, the GDP share of public spending can be expected to depend on the number of government parties if corruption is sufficiently prevalent. When corruption pervades the institutions of representative politics, clientelistic strategies based on the exchange of material favours for political support can be expected to gain importance at the cost of programmatic goals, and this manifests itself as a positive effect of coalition size on spending. However, a discernible effect should not exist if corruption is (almost) absent.
This hypothesis is tested using Western European time-series cross-sectional data from 1984 to 2010. The results suggest that coalition size has no robust effect on government spending. Instead, a positive effect is discernible only at sufficiently high levels of corruption, but no statistically significant effect exists when there is little corruption.
This observation has implications for the evaluation of some policy recommendations that have been put forward in previous literature. It is especially questionable whether restrictions on proportional representation would increase fiscal discipline in every case.
