Optimal Sourcing with Ambiguously Correlated Suppliers

We present a comprehensive study for the optimal sourcing strategy of a firm where the market selling price is endogenously determined by the total output and suppliers face correlated disruptions that are difficult to estimate. We develop a distributionally robust (DR) model for a single group of correlated suppliers that maximizes the firms' expected profit under the worst-case disruption distribution. Our model only requires the marginal disruption probabilities and accommodates both yield and capacity uncertainty. We derive simple expressions that allow the firm to eliminate less reliable suppliers on the combined basis of cost and reliability when yield is uncertain. In the case capacity uncertainty, we show the cheapest supplier influences the orders placed with more reliable suppliers, if the cheapest supplier receives no order. We incorporate the DR model in a more general model where the available suppliers are grouped based on common sources of vulnerability. This model accommodates cases where correlations exist between all or no suppliers as special cases. The model allows decision-makers to tailor the sourcing strategy based on their risk tolerance. We introduce new measures for supplier efficiency that accounts for both supplier costs and reliabilities. These measures are incorporated in a solution procedure that provides the optimal supplier selection and order allocation. The measure of efficiency serves as an extension of the commonly cited insight that "cost is an order qualifier and reliability is an order winner" for the case of correlated suppliers. We end the paper by studying the impact of buyer competition on the optimal sourcing strategy. We present equilibrium solutions for cases of symmetric information and study the information sharing behavior of a better-informed firm under information asymmetry.