Put your money where your forecast is: Supply chain collaborative forecasting with cost-function-based prediction markets

Despite extensive research on the value of collaborative forecasting among supply chain partners, there has been limited attention to the actual mechanisms of forecast sharing. We propose using cost-function-based prediction markets as means of sharing demand forecasts. We consider a decentralized two-stage supply chain with a supplier who sells to a retailer, in which both partners receive demand forecast updates. In the first stage, the supplier and the retailer agree on a collaborative forecasting scheme to form a consensus forecast. We consider two schemes: collaborative forecasting using cost-function-based prediction markets in which partners pay a set-up cost to operate a prediction market, and direct collaborative forecasting in which partners directly share their forecasts with each other and resolve their forecast differences through costly actions. In the second stage, we analyze the game between the supplier and the retailer to determine the retailer's optimal order quantity as well as the supplier's optimal wholesale price. We show that demand forecast sharing using cost-function-based prediction markets results in a Pareto improvement of the partners' profit over the alternative of directly sharing forecasts. Furthermore, we show that this Pareto improvement is monotone with respect to supply chain partners' forecast accuracy. We conduct an extensive numerical study to provide further insights into the effect of the return on forecast sharing investment, forecast accuracy, and level of demand uncertainty on the benefits gained by implementing cost-function-based prediction markets.