Commodity Procurement with Demand Forecast and Forward Price Updates

Commodities, ranging from natural gas to memory chips, can be procured both by trading on the date in spot markets and in advance in forward markets. Transaction costs, such as brokerage fees, are typically higher in spot markets than in forward markets. Moreover, the forecast of a ¯rm's commodity requirement (demand) for a given future date typically changes in an uncertain fashion over time. Thus, although the dynamics of forward and spot prices are notoriously uncertain, firms that procure commodities face the dilemma of choosing between early and possibly less expensive commitments with residual demand uncertainty and late and possibly more expensive sourcing of the exact amount needed. We investigate this issue by developing and analyzing a model of commodity procurement for a single future date. Our model generalizes models available in the real options and operations management literature, by simultaneously considering correlated demand forecast and forward price updates in a setting characterized by multiple forward transactions and a single spot transaction. We derive the structure of the optimal procurement policy and discuss its computation in cases of practical interest. In a numerical study, based on applying our model to natural gas data, we offer managerial insights on the effects that demand forecast and forward price updates, both in isolation and combined, have on the value of a firm's procurement policy. We also assess the sensitivities of these effects to parameters of interest and the potential managerial relevance of the combined effect. Our model and results have significance beyond the specific application.