UNCERTAINTY, LEARNING AND HETEROGENEITY IN INTERNATIONAL ENVIRONMENTAL AGREEMENTS

In an earlier paper (Kolstad and Ulph, 2006) we studied the formation of International Environmental Agreements under uncertainty about the damages that might be caused by climate change and different models of learning (complete learning, partial learning or no learning), in which better information about these damage costs becomes available. Our results were generally pessimistic: the possibility of either complete or partial learning generally reduces the level of global welfare that can be achieved from forming an IEA relative to no learning. In that paper we assumed that uncertainty regarded a parameter common to all countries, so that countries were identical ex ante as well as ex post. In this paper we extend our analysis to the case where there is no correlation between damage costs across countries; each country is uncertain about a particular parameter (in our case the benefit-cost ratio) drawn from a common distribution but, ex post, each country’s realized parameter value is independently drawn. Consequently, while countries remain identical ex ante, they may be heterogeneous ex post. We show that this change reinforces our negative conclusions about the effects of partial learning on international environmental agreements, but, under certain conditions, moderates our negative conclusions about the effects of complete learning.