Search Costs, Intermediation, and Trade: Experimental Evidence from Ugandan Agricultural Markets (with Craig McIntosh and Meredith Startz) Search costs may be a barrier to market integration in developing countries, harming both producers and consumers. We present evidence from the large-scale experimental rollout of a mobile phone-based marketplace intended to reduce search costs for agricultural commodities in Uganda. We find that market integration improves substantially: trade increases and excess price dispersion falls by 20% between treated markets. This reflects price convergence across relative surplus and deficit markets, with no change on average. Interpreting the experimental variation through the lens of a model with fixed costs of search, we estimate that the marketplace caused a 5% reduction in total trade costs between treated markets. Contrary to the stated goals of the marketplace, but consistent with the existence of economies of scale in search or other trade costs, almost all activity on the platform is among larger traders, with very little use by smallholder farmers. Nevertheless, the benefits of improved arbitrage by traders appears to pass through to farmers in the form of higher revenues in surplus markets, as trader entry increases and measured trader profits decrease in response to falling search costs.