A demand partitioning framework to reserve production for small enterprises

Abstract The reservation of goods to be produced in the micro, small, and medium enterprises (MSME) sector, in the early years after India's independence, addressed the dual needs of development of the industrial sector and production of goods. However, these industrial policies created an incentive for firms to remain small so that they can continue to avail of the benefits provided by the Government. On the positive side, the MSMEs typically employ more labor intensive production processes and consequently contribute significantly to the provision of employment opportunities, generation of income, and poverty reduction. But, on the negative side, the policies have also partly facilitated the creation of a divide in terms of productivity between the MSMEs and large sized firms. In particular the policy raises important questions for a firm auctioning supply contracts among suppliers with a significant cost differential. In this paper we propose an idea to allocate supply contracts wherein a manufacturing firm partitions the stochastic demand into mutually exclusive portions and awards each portion to a different supplier. We characterize such an optimal procurement mechanism when there are two types of suppliers and an arbitrary number of demand portions. We show that the optimal procurement may require the manufacturer to intentionally withhold some demand portion, and this arises when one type of supplier is considerably inefficient in serving a demand portion. We extend our analysis to the cases with multiple types with two suppliers and two types with multiple suppliers. The optimal partition is composed of at most six contiguous demand portions, and it may include a detrimental demand portion that only generates a negative expected payoff to both supplier types. Our demand partitioning mechanism leads to a strictly higher manufacturer's expected payoff than the conventional winner-take-all case unless one supplier type completely dominates the other. We present numerical experiments that indicate when such a mechanism holds the greatest advantage for the buyer.