Why does misallocation persist?

Recent papers argue that the misallocation of resources can explain large cross-country TFP differences. This argument is underpinned by empirical evidence documenting substantial dispersion in the marginal products of resources, particularly capital, in developing countries. But why does misallocation persists? That is, why don't distortions disappear on their own? This is particularly true for capital misallocation, a point we illustrate in a simple model of capital accumulation with credit constraints. We distinguish between misallocation on the intensive and the extensive margin, and show that the former should disappear asymptotically under general conditions, while the latter may persist. We conclude by discussing possible theories of persistent misallocation.