14th journées Louis-André Gérard-Varet

Font Size: 
More for everyone. The effect of local interests on spending on infrastructure
Askill Harkjerr Halse

Last modified: 2015-03-14


The model of distributive politics by Weingast et al. (1981) predicts that a legislature with elected representatives from different areas will spend too much on local (`particularistic') public goods because each member does not fully take into account the alternative cost. The incentive to exploit the common pool is stronger the smaller is the geographic constituency of each representative. Previous empirical tests of this hypothesis are based on legislature size as a measure of fragmentation, and measures of public spending which do not necessarily capture particularistic spending. In my study, I look at spending on local road construction and maintenance by Norwegian regional governments. As measure of geographical government (de)fragmentation, I use the population share of the municipality to which the average council member belongs, exploiting that the electoral system gives considerable time-variation in geographic representation. Consistent with the model by Weingast et al., I find that spending on roads decreases when the more populated municipalities occupy more seats. I do not find the same effect for other spending purposes. Spending on roads is also lower when more of the municipalities are represented in the council, which could indicate that particularistic spending is more attractive when it does not have to be shared among everyone.


Distributive politics; Pork-barrel; Concentrated gains; Law of 1/n