There a large belief in the development community that institutions matter. This has emerging from a large number of cross-country regressions, regressions that one should always take with a grain of salt because of methodological issues and data quality. However, this result has emerged so frequently that it must indeed matter. But the precise mechanism through which institutions matter of the course of development is still rather unexplored.
Ines Lindner and Holger Strulik come up with an interesting theory. When the economy is fragmented into small regions, entrepreneurial behavior is governed by local and informal enforcement. But as economies become more integrated, through specialization and/or the reduction in transportation costs, "connectivities" go from local to global, and informal enforcement is not sufficient. Formal institutions need to arise, and if they are weak, entrepreneurship will be weak. I wonder, though, how such a theory of networks could be tested in the data.