Second best and new institutional economics: is a synthesis possible?
The study of institutions and policies in network industries, especially in developing countries, is dominated by two different theoretical frameworks. One is the second best approach explained in a book by the late Jean Jacques Laffont in 2005, "Regulation and Development". Laffont argues, using theoretical models (mostly based on asymmetric information) and some empirical evidence, that developing countries face many constraints that make it impossible to apply policies that come close to first-best solutions. For example, the social cost of tax distortions is much higher than in developed countries, which makes it much more difficult to use subsidies. The general idea is that if reform has to be introduced in network industries in a context where much else is not reformed, then regulatory reform will necessarily be piecemeal and not a big bang transformation. The other theoretical framework has in economist Pablo Spiller its intellectual leader. Spiller's approach is rooted in the transaction cost economics pioneered by Nobel Prize winners Ronald Coase and Oliver Williamson (who themselves have work on regulated industries). This school, much less formal in its approach but equally based on rigorous empirical methods, argues that the main issue in institutional choice is to minimize transaction costs, so that social interactions can find the quickest possible route to efficiency. In network industries the main source of transaction costs is the commitment problem faced by investors: by anticipating opportunistic regulations, investors will be reluctant to invest in sunk assets in the first place, unless institutional mechanisms exist that make private investors confident that opportunism will not happen. Then Spiller argues that these institutional mechanisms will be different across countries, and that to be useful they will have to adapt to the institutional endowment. For example, the way to achieve commitment in Chile (detailed legislation) will be different from the way to achieve commitment in the United Kingdom (a license system with independent regulators). Both approaches have developed with real cases and real problems in mind, but without much dialogue (or so it seems to me). But in essence they are not that different: Laffont emphasizes the importance of local constraints, as Spiller does. The second best approach is more static, more formal, more normative, and the new institutional approach is more dynamic, more multidisciplinary, more positive. But they are not incompatible, and perhaps a synthesis is overdue before we move on to new complementary approaches that for example take into account behavioral issues and social norms.