On the Optimal Size and Composition of Customs Unions: An Evolutionary Approach.

Customs unions enable countries to freely access each other's markets, which is thought to increase intra-regional trade and economic growth. However, accession to a customs union also comes with the condition that all members need to consent to a common external trade policy. Especially if countries feature different economic structures, this may act as a force against the creation of large customs unions. In this paper, we propose a new mathematical approach to model the optimal size and composition of customs unions in the form of a bi-objective combinatorial non-linear problem. We also use a multi-objective evolutionary algorithm (NSGA-II) to search for the best (non-dominated) configurations using data on the trade flows and economic characteristics of 200 countries. Our algorithm identifies 445 different configurations that are strictly preferable, from a global perspective, to the real-world landscape of customs unions. However, many of these non-dominated configurations have the feature that they improve outcomes for the world as a whole, on average, but not for all individual countries. The best configurations tend to favour the creation of a few large customs unions and several smaller ones.