Capital Flows

In this paper, we reexamine the question “Why doesn’t capital ßow from rich to poor countries?” posed by Lucas (1990). Our Þndings suggest that even if capital ßows freely it will ßow to middle income countries rather than to poor countries. We start with the stylized fact that poorer countries have high intermediation costs, and develop a simple disaggregation of the neoclassical production function based on a model of costly intermediation of funds between safe and risky projects when there is ex ante heterogeneity of project potential. When intermediation costs are ignored, the model behaves much like the neoclassical model in terms of capital returns. However, when intermediation costs are considered, the return for a given amount of capital can be non-monotonic in costs. Therefore, the combination of capital and cost differences across countries gives rise to a rich variation of returns, one that suggests a tendency for capital to ßow to middle income countries, as seen in data. Indeed, when we embed the static return function in a two-country dynamic model, there is capital outßow from a poor country that removes capital controls and becomes open. We Þnd that even though the closed economy dominates in terms of capital employed in production, it is the open economy that dominates in terms of income, consumption and welfare. ∗Finance and Business Economics, Marshall School of Business, HOH 701, USC, Los Angeles, CA 90089-1427. e-mail: [email protected], [email protected]. We are grateful to Matthias Doepke, Charles Engel, Doug Joines, Robert Lucas, Aris Protopapadakis, Jan Zabojnik and the research lunch participants at USC, seminar participants at the University of Washington, Universidad Torcuato Di Tella, Stanford University, the 2001 winter meetings of the Econometric Society in New Orleans, and the 2001 annual meeting of the Society for Economic Dynamics in Stockholm, Sweden, for helpful comments.