Comments

Francisco Rodríguez: In "Coordination Failures, Clusters, and Microeconomic Interventions," Andrés Rodríguez-Clare presents a coherent case in favor of a new type of industrial policy for Latin America. He argues that the region's disappointing growth performance calls for renewed thinking about the type of strategies that could complement Washington Consensus policies. He contends that many of the microeconomic interventions presently in vogue in the region do not have strong theoretical or empirical support, and he concentrates on the case for one particular type of alternative intervention, centered on the creation of conditions that will facilitate the emergence of clusters based on economies of agglomeration.

Rodríguez-Clare's case is supported by an elegant stylized model in which all sectors have the potential to reap productivity gains from clustering. Whether they do so depends on their ability to solve coordination failures. The first-best solution can be achieved by addressing coordination failures in sectors that already have a revealed comparative advantage. Relative price signals will then correctly steer the economy toward the sector in which it has a natural comparative advantage and will generate clustering in that sector. There is no need to pick winners: as long as the coordination problem is solved, the economy will pick them by itself.

I have three sets of observations about this paper. The first relates to the significance of models of multiple equilibria, such as the one proposed in this paper, to understanding the problems of developing countries. The second has to do with to the relevance of knowledge-intensive clusters in a world of rapidly declining costs of transmitting and processing goods and information. The last addresses whether it is actually possible to avoid picking winners in the design of industrial policy.

The Relevance of Multiple Equilibrium Models for Developing Countries

Rodríguez-Clare's key argument is premised on the ideas that coordination failures are pervasive in LDCs and that solving them can generate substantial productivity gains. These failures can only be important in the context of multiple equilibria. The premise is that the economy is stuck in a suboptimal equilibrium, as a result of either the failure of agents to coordinate or the government's failure to provide the type of incentives that would generate cooperation. The existence of multiple equilibria and the belief that LDCs are stuck in the bad ones are thus vital parts of Rodríguez-Clare's case.

Models of multiple equilibria have experienced a renaissance of sorts in development economics. Although the basic intuition behind these models goes back at least to Myrdal's application of the idea of cumulative causation to underdeveloped regions, their acceptance by the mainstream had to wait for formalizations via models such as that of Murphy, Shleifer, and Vishny.¹ These models are very attractive for development economics because they enable analysts to explain how economies that do not differ in their fundamentals may end up in very different situations. Indeed, it is safe to say that one can always come up with a sensible model of multiple equilibria to account for any fact that cannot be explained based on observables. An added attraction is that they offer an intuitive role for policy interventions in which the government acts as a coordinating actor that helps the economy move to the higher-level equilibrium.

The main attraction of these models is also their main drawback. Since they explain differences that are not based on fundamentals, it is very difficult to test them. Indeed, one can't do meaningful comparative statics with these models because comparative statics analysis requires taking derivatives with respect to small changes around the equilibrium--and in this neighborhood, models of multiple equilibria look just like models with a unique equilibrium. To test a model of multiple equilibria, one needs to be able to evaluate the effect of large shocks that drive the economy from one equilibrium to another. Theory gives us few clues as to how large these shocks need to be.

Davis and...