April 2000 - If urban overconcentration really is an issue, it ought to affect economic growth rates in a robust, consistent fashion. And it does. Not only is there an optimal degree of urban concentration that varies with country income, but departures from optimal concentration result in substantial growth losses. Overconcentrated countries can reduce concentration by investing in interregional transport infrastructure - in particular, increasing the density of road networks. Henderson explores the issue of urban overconcentration econometrically, using data from a panel of 80 to 100 countries every 5 years from 1960 to 1995. He finds the following: At any level of development there is indeed a best degree of national urban concentration. It increases sharply as income rises, up to a per capita income of about ,000 (Penn World table purchasing parity income), before declining modestly. The best degree of concentration declines with country scale. Growth losses from significantly nonoptimal concentration are large. Those losses tend to rise with level of development, peaking at a very high level (about 1.5 annual percentage points of economic growth). Results are very robust; In a group of 72 countries in 1990, roughly 30 have satisfactory urban concentration, 24 have excessive concentration, and 5 to 16 countries have too little; The list of countries with highly excessive concentration includes Argentina, Chile, Costa Rica, and Panama (in Latin America); the Republic of Korea and Thailand (in Asia); Congo (in Africa); and Greece, Ireland, and Portugal (in Europe). Many of these countries have explicitly unitary governments or federal structures have traditionally been severely constrained; The list of countries with too little urban concentration includes Belgium (a small, split country) and special cases such as Czechoslovakia and the former Yugoslavia; Urban concentration declines with national scale. It initially rises with income, then peaks at a per capita income of about ,000, before declining. If the largest city in a country is a port, increased trade leads to increased urban concentration. Otherwise, increased trade leads to deconcentration as markets in the hinterland open up to trade. But trade effects are modest. Similarly, more political decentralization (or increased federalism) only modestly reduces urban concentration. However, interregional transport infrastructure - especially dense road networks - significantly reduce urban concentration, an effect that rises with income. This paper - a product of Infrastructure and Environment, Development Research Group - is part of a larger effort in the group to analyze the role of economic geography and urbanization in the development process, particularly as influenced by infrastructure investment and political decentralization. The author may be contacted at vernon_henderson@brown.edu.