perspective regarding the move to freer world trade. To gauge whether or not global free trade
can be achieved through the formation of FTAs, one needs to understand whether FTAs serve as
“building blocks” or “stumbling blocks” of global free trade. Kemp and Wan can be interpreted
as proving the existence of a path of Pareto-improving customs unions which could be used in
principle to implement global free trade. Likewise, our results imply that there exists a sequence
of enlarging FTAs which are Pareto-improving, and that could be used to implement global free
trade, provided that one is willing to implement restrictions on the aggregate trade flows between
all FTA members and their partners not involved in the new FTA. See the conclusion of this
paper for details.
A considerable number of papers have investigated the building-block stumbling-block ques-
tion, some of which we touch on here. For example, Yi (2000) applies a n-country coalition
formation game where welfare-concerned governments decide whether or not to form FTAs. He
shows under two different FTA formation rules (“open regionalism" and “unanimous regional-
ism") that the possibility of implementing global free trade by forming FTAs may not materialize
due to free-riding problems if the number of countries is sufficiently large. Then, we can conclude
that Yi (2000) indicates that FTAs serve as “stumbling blocks" to free trade. On the contrary,
Furusawa and Hideo (2007) and Goyal and Joshi (2006) use a n -country strategic network11
model to examine the incentives of countries to form FTAs. Both papers conclude that global
free trade emerges as a stable outcome in the case of symmetric models where governments
maximize welfare.12 In this case, FTAs serve as building blocks to global free trade.13
6 Conclusion
This paper develops a methodology to study welfare-enhancement in the presence of free
trade areas. The methodology is general and is able to identify the corresponding bound for
restrictiveness of the rules of origin compatible with welfare gains. We also apply the methodology
11Strategic network equilibria allow for FTAs where member countries can be part of more than one preferential
agreement. In this sense, it is more realistic than coalition formation games that do not allow these cases in
equilibrium.
12In Furusawa and Hideo (2007), the symmetric model means that countries have the same level of indus-
trialization (i.e., same measure of consumers and same measure of firms). Goyal and Joshi define symmetric
economies based on market size and costs of production.
13Several papers on the political economy of trade have also investigated this question. Krishna (1998) and
Ornelas (2005) conclude that FTAs are stumbling blocks to free trade for different reasons. The former finds that
member countries would lose interest in multilateral openness after forming an FTA, and the latter finds that
member countries’ external tariff reductions would decrease the interest of non-members in global free trade.
11