This paper evidences collusion among employers in the textile and clothing manufacturing industry in India. I develop a simple comparative static test to distinguish collusion from standard forms of imperfect competition, showing that firm-specific demand shocks predict opposite employment effects at unshocked competitors who operate independently (↓ employment) versus firms that were previously colluding but whose collusion breaks due to the shock (↑ employment). Next, I argue that large employers in the garment industry organize into industry associations to pay workers exactly the local minimum wage. Small demand shocks leave wages and employment at association members unchanged, suggesting that firms are willing to forego opportunities to sustain collusion. However, when a large demand shock leads affected members to deviate from the minimum wage, unaffected non-members respond as in oligopsony (↑ wage, ↓employment), but unaffected members respond as if their collusion dismantles (↑ wage, ↑ employment). Imposing specific models of labor supply and production, the “ full-IO” approach rejects oligopsony in favor of the breakdown of collusion. Collusion spurs substantial losses even compared to firms exercising their independent but not their

collective market power, reducing the average worker’s wage by 9.6% and employment by 17%.