Mario Jametti

RISK SELECTION IN NATURAL DISASTER INSURANCE

It is widely recognized that market failure prevents efficient risk sharing in natural disaster insurance, leading to several public-private partnership arrangements across the globe. We argue that risk selection, a situation where the public partner insures the majority of high risk agents, is potentially an important issue. To illustrate our concerns we build a simple model of reinsurance in a natural disaster insurance market. We show that risk selection is a likely equilibrium outcome and discuss the policy options available. The model is based on the French institutional setup and describes well the stylized facts. The policies implemented by the French government correspond to the ones we identify to alleviate risk selection. We also present two alternative public-private partnership setting that deal effectively with risk selection; hurricane insurance in Florida and catastrophe insurance in Spain.

Elections and de facto Expenditure Decentralization in Canada

This paper empirically investigates the underlying determinants of expenditure decentralization, based on the predictions of a new political economy model of partial decentralization. The analysis is based on an agency model, in which two levels of government are involved in the provision of a public good and voters are imperfectly informed about each governments contribution to the good, creating a shared accountability problem. Under shared expenditure responsibility, the degree of decentralization is endogenous and depends on the relative political conditions prevailing at each level of government. Consistent with the models predictions, empirical results from a panel of Canadian provinces show that decentralization in a province increases with the electoral strength of the provincial government and decreases with the electoral strength of the federal government, in addition to being affected significantly by the partisan affiliation of both levels of government. A series of alternative empirical specifications, including an IV regression exploiting campaign spending data, are presented to assess the robustness of these results.