Competition and Technological Complexity in Procurement: An Empirical Study of Dual Sourcing

The role of competition in defense procurement has long been controversial due to the extraordinary demand for technological advance, the relatively small production quantities involved and the importance of learning by doing. Since the early 1980s, Defense Department policy has encouraged the use of competition in the production phase of procurement where possible, particularly for relatively simple technologies. Recent theory, however, suggests that splitting production between two bidders ("dual sourcing") produces strong incentives for collusion unless the bidders are unsure of each other's costs, e.g. for sophisticated technologies in the early phases of production. Furthermore, dual sourcing may help discipline contractors in settings where contractual incompleteness is a particular problem. For both these reasons, dual sourcing may be more valuable for complex, rather than simple, technologies. To date, however, there have been no attempts to investigate empirically how technological complexity affects the viability of competition in procurement. I explore the effects of dual sourcing using a panel dataset comprising 14 missile systems with an average of 12.5 years of production history per system. Each missile's complexity is categorized based on the nature of its guidance and control system. Simple missiles enjoy greater scale economies than complex missiles, but the learning curves for the two missile types are not significantly different. The effects of dual sourcing, though, depend importantly on the nature of the technology involved: it significantly speeds learning and interferes with scale economies for complex missiles, but has no significant effects for simple ones. Dual sourcing thus produces no apparent savings for simple missiles, but for the average complex missile dual sourcing lowers unit costs by the seventh year of dual sourcing. Whether these potential savings justify the costs of transferring technology to the second source and the possible weakening of R&D incentives remains an open question.