Competition and Relational Contracts in the Rwanda Coffee Chain

How does competition affect market outcomes when formal contracts are not enforceable, and parties’ resort to relational contracts? Difficulties with measuring relational contracts and dealing with the endogeneity of competition have frustrated attempts to answer this question. We make progress by studying relational contracts between upstream farmers and downstream mills in Rwanda's coffee industry. First, we identify salient dimensions of their relational contracts (unenforceable provision of services in both directions before, during and after harvest) and measure them through an original survey of mills and farmers. Second, we take advantage of an engineering model for the optimal placement of mills to construct an instrument that isolates geographically determined variation in competition. Conditional on the suitability for mills within the catchment area, we find that mills surrounded by more suitable areas: (i) face more competition from other mills; (ii) use fewer relational contracts with farmers; and (iii) exhibit worse performance. In contrast to conventional wisdom, an additional competing mill also (iv) makes farmers worse off; (v) reduces the aggregate quantity of coffee supplied to mills by farmers; and (vi) conditional on the farmer's distance from the mill, lowers relational contracts more for farmers close to the competing mill, suggesting that competition directly alters farmers temptation to renege on the relational contract. The finding that increased competition downstream leaves all producers -- including upstream producers -- no better-off suggests a potential role for policy in a second-best environment in which contracts are hard to enforce.