We study differences in markups earned by Bangladeshi garment exporters across buyers with different sourcing strategies and make three contributions. First, we distinguish buyers with a relational versus a spot sourcing strategy and show that a buyer’s sourcing strategy is correlated across products and origins. Buyer fixed effects explain most of the variation in sourcing strategies, suggesting that these depend on organizational capabilities. Second, we use novel data that match quantities and prices of the two main variable inputs in the production of garments (fabric and labor on sewing lines) to specific export orders. We derive conditions under which these data allow measurement of within exporter–product–time differences in markups across orders produced for different buyers. Third, we show that exporters earn higher markups on otherwise identical orders produced for relational, as opposed to spot, buyers. A sourcing model with imperfect contract enforcement, idiosyncratic shocks to exporters, and buyers that adopt different sourcing strategies trading off higher prices and reliable supply rationalizes this and other observed facts in the industry. We discuss alternative explanations and policy implications.

There is a consensus that global value chains have aided developing countries' growth. This essay highlights the governance complexities arising from participating in such chains, drawing from lessons we have learned conducting research in the coffee and garment supply chains. Market power of international buyers can lead to inefficiently low wages, prices, quality standards, and poor working conditions. At the same time, some degree of market power might be needed to sustain long-term supply relationships that are beneficial in a world with incomplete contracts. We discuss how buyers’ market power and long-term supply relationships interact and how these relationships at the export-gate could be leveraged to enhance sustainability in the domestic part of the chains. We hope that the lessons learned by combining detailed data and contextual knowledge in two specific chains—coffee and garments—have broader applicability to other global value chains.