Abstract: The rapid growth of big data has provided tremendous opportunities for enterprises to understand the market and make decisions better. However, taking the significant cost into account, practitioners have also raised questions about the financial returns on the big-data investment. Inter-enterprise sharing of big data may be an effective approach to alleviate the big-data investment pressure by avoiding inefficiently redundant development of big-data resources, nevertheless, partial sharing or non-sharing is more common. In this context, the decision and incentive alignment issues arising from big-data investment in a supply chain that consists of one upstream supplier and one downstream manufacturer are investigated. Both members can invest in big data to obtain accurate demand forecasts. The forecasts can be shared partially not only from the supplier to the manufacturer (i.e., top-down transparency), but also from the manufacturer to the supplier (i.e., bottom-up transparency). As the two-way partial transparency is exogenously given, a theoretical analysis model is established to solve the decision problems of big-data investment and then the impacts of the two-way transparency on the utilization value and investment incentives of big data are analyzed.
The results indicate that the supplier always benefits from two-way transparency, whereas the manufacturer can benefit not only from top-down partial transparency with lower bottom-up transparency, but also from bottom-up complete transparency with lower top-down transparency. Therefore, both members can benefit from two-way increased transparency when the top-down transparency is sufficiently low. In addition, it is most conducive to the feasibility of big-data investment with top-down complete transparency and bottom-up non-transparency. Overinvestment never happens to the manufacturer, while it may happen to the supplier when both the top-down and bottom-up transparency is sufficiently low. To address the incentive alignment issues, a contract scheme based on investment compensation is proposed, which can achieve the optimal investment levels under the centralized investment setting and realize Pareto improvement. Finally, numerical experiments are conducted to obtain more managerial insights, and show that the return on investment can be improved by 5-49% under the contract scheme.
In summary, the decision making and incentive mechanism for big-data investments of multiple enterprises under partial transparency are studied by employing the game theory. The findings in this paper can provide academic and practical insights for sustainable utilization of big data.

Key words: big data investment, supply chain transparency, demand forecast, information sharing, contract design