It is costly to learn about market conditions elsewhere, especially in developing countries. This paper examines how such information frictions affect trade. Using data on regional agricultural trade in the Philippines, I first document a number of observed patterns in trade flows and prices that suggest the presence of information frictions. I then incorporate information frictions into a perfect competition trade model by embedding a process whereby heterogeneous producers engage in a costly sequential search process to determine where to sell their produce. I show that introducing information frictions reconciles the theory with the observed patterns in the data. Structural estimation of the model finds that information frictions are quantitatively important: roughly half the observed regional price dispersion is due to information frictions. Furthermore, incorporating information frictions improves the out-of-sample predictive power of the model.

We develop a general equilibrium framework to determine the spatial distribution of economic activity on any surface with (nearly) any geography. Combining the gravity structure of trade with labor mobility, we provide conditions for the existence, uniqueness, and stability of a spatial economic equilibrium and derive a simple set of equations that govern the relationship between economic activity and the geography of the surface. We then use the framework to estimate the topography of trade costs, productivities and amenities in the United States. We find that geographic location accounts for at least twenty percent of the spatial variation in U.S. income. Finally, we calculate that the construction of the interstate highway system increased welfare by 1.1 to 1.4 percent, which is substantially larger than its cost.