We study the nature of relational contracts among managers of production lines in a large Indian ready-made garments exporter. Manufacturing firms in this context often suffer from persistent worker absenteeism, and large absenteeism shocks are frequent. We show that managers rely on relational contracts through which they lend and borrow workers to mitigate the negative effect of absenteeism on productivity. Managers have clear primary trading partners, and within these relationships, switch often between being net lenders (principals) and net borrowers (agents), depending on the relative size of their shocks. We show that the dynamics of borrowing and lending workers across manager pairs closely follows predicted behavior from a model of relational contracting that incorporates state uncertainty, heterogeneous manager types, and learning. Partners who are physically closer on factory floors trade more, consistent with lower monitoring costs in the presence of moral hazard. Managers lend more workers the smaller their own shock and the larger their partner’s. Uncertainty about the reliability of partners resolves over time, making partners trade more intensively as relationships mature. To address endogeneity concerns, we exploit idiosyncratic variation in absenteeism driven by the (lunar calendar) dates of major cultural festivals in the states of workers’ origins.

Written with Jean-Francois Gauthier (Boston College), Anant Nyshadham (Boston College), and Jorge Tamayo (Harvard Business School)