Do You Have a Room for Us in Your IT? An Economic Analysis of Shared IT Services and Implications for IT Industries
研究了政府部门共享IT服务的决策，通过博弈模型分析三种共享机制（成本分摊、利润中心、协调）对政府和供应商的影响，发现共享不一定带来好处，协调可能降低社会福利。
We are witnessing an interesting and unique phenomenon in enterprise information technology (IT) adoption and management in public-sector organizations: shared IT services. Instead of implementing separate IT services, governments come together to pool their IT resources into one single IT service. In this study, we develop a game-theoretic model to analyze the governments’ decision to share IT services and understand how the introduction of shared services transforms the strategic interactions between the governments and the vendor. We study three common regimes used in the adoption of shared IT services: (1) a cost-sharing regime where costs are split proportionally, (2) a profit-center regime where one government charges a surplus-maximization price to the other, and (3) a coordination regime where governments coordinate their decisions to maximize aggregate surplus. Our analyses generate several intriguing findings. First, although charging a surplus-maximizing price seems to be a lucrative option, we find that a government does not always benefit by acting as a profit center. Second, the cost-sharing regime does not always incentivize the shared service adoption despite being often viewed as a fairer and more convenient arrangement. Third, we find that there can be significant under-utilization of shared services in the absence of proper coordination, in a sense that the governments may choose not to share their IT services even if doing so would increase their aggregate surplus. Finally, even though coordination promotes the adoption of shared IT services, it can sometimes be inefficient from the social welfare’s perspective because the increase in governments’ surplus can be outweighed by the decrease in the vendor’s profit. We also present a range of extensions to our model to show that our main takeaways carry over when some model assumptions are relaxed.