推动政务数据部门间共享是释放公共数据价值的关键环节。大数据管理机构是负责政务数据共享的主管部门，在推动政务数据部门间共享中面临“业务部门弱激励”和“部门协同难”的困境。本文研究问题是：（1）大数据管理机构推动政务数据部门间共享的协同模式有哪些？（2）为什么有些大数据管理机构可以推动协同，而有些却难以推动协同？（3）不同协同模式的效果如何？既有文献从部门协同、数据治理和制度性集体行动框架等理论探究政务数据部门间共享的影响机制与解释框架。上述成果各有优势，但研究视角缺乏有效整合。多数研究的理论预设是，大数据管理机构是既有协同情境的被动接受者，协同情境的改善只能依赖于制度环境的完善，而忽略大数据管理机构作为协同主体的能动性及其对协同情境的重塑。通过整合上述理论，本文构建起大数据管理机构推动政务数据部门间共享的部门协同理论框架，基于因果过程观测路径的案例研究方法，探讨大数据管理机构推动政务数据部门间共享的四种协同模式及其因果机制。研究结果发现：（1）大数据管理机构所构建的主要协同模式，可划分为激励相容的双向赋能型协同、压力传导的政治主导型协同、常规运作的谈判式协同、利益驱动的需求主导型协同四类。（2）大数据管理机构推动部门间数据共享的模式选择取决于政务数据复杂性、上级注意力和技术整合度等协同情景，其协同能力和协同激励的构建过程直接影响协同结果。大数据管理机构在选择协同策略、构建与业务部门的协同模式上具有能动性。（3）不同的协同模式是大数据管理机构在既定情境下的现实选择。双向赋能型协同最为理想，能够实现数据逻辑与科层逻辑的兼容；谈判式协同的数据逻辑较为滞后，有一定的协同成本；而政治主导型协同则对上级干预的依赖性较高。本研究贡献：（1）打破了既有研究对于大数据管理机构面临“权小责大”等困境而难以施展作为的刻板印象，阐明其能够根据协同情境构建协同能力与协同激励、推动政务数据部门间共享的理论依据，（2）弥合了既有研究的解释框架，发展出大数据管理机构推动政务数据部门间共享的四种协同模式，影响因素及其效果。本文的政策启示：基于大数据管理机构推动政务数据部门间共享的协同模式及其效度与限度，未来仍需加强政务数据的确权定责，健全大数据管理机构促进业务高效协同、数据按需共享的体制机制，完善自主性部门协同的实现路径。

Promoting the sharing of government data among government departments is a key step to release the value of public data. The big data management organization is the main department responsible for government data sharing, and faces the dilemma of "weak incentive of business departments" and "difficult collaboration of departments" in promoting government data sharing among departments. The research questions of this paper are as follows: (1) What are the collaborative modes of big data management organizations to promote sharing among government data departments? (2) Why are some big data management organizations able to promote collaboration while others are not? (3) What are the effects of different collaborative modes?Existing literature explores the influence mechanism and explanatory framework of inter-departmental sharing of government data from the perspectives of departmental collaboration, data governance and institutional collective action framework. Each of the above results has its own advantages, but the research perspectives lack effective integration. The theoretical presupposition of most studies is that big data management organizations are passive recipients of existing collaborative situations, and the improvement of collaborative situations can only rely on the perfection of institutional environment, while ignoring the initiative of big data management organizations as collaborative subjects and their reshaping of collaborative situations. By integrating the above theories, this paper builds a theoretical framework of departmental collaboration for big data management organizations to promote sharing among government data departments. Based on the case study method of causal process observation path, four collaborative modes and their causal mechanisms for big data management organizations to promote sharing among government data departments are discussed.The research results show that: (1) The main collaborative models constructed by big data management organizations can be divided into four categories: bidirectional enabling synergy compatible with incentives, political oriented synergy guided by pressure conduction, negotiation synergy in conventional operation, and interest-driven demand-oriented synergy. (2) The mode selection of big data management institutions to promote inter-departmental data sharing depends on the cooperation scenarios such as the complexity of government data, the attention of superiors and the degree of technology integration, and the construction process of their collaborative ability and collaborative incentive directly affects the collaborative results. Big data management organizations have the initiative in selecting collaborative strategies and constructing collaborative modes with business departments. (3) Different collaborative modes are realistic choices for big data management organizations under given situations. Bidirectional enabling collaboration is ideal, which can realize the compatibility of data logic and hierarchical logic. The data logic of negotiation collaboration is relatively backward and has a certain cost of collaboration. However, the politically-dominated synergy has a high dependence on superior intervention.Contribution of this research: (1) It breaks the stereotype of existing researches that big data management institutions are difficult to perform their duties due to the dilemma of "little power and great responsibility", and clarifies the theoretical basis of their ability to build collaborative ability and collaborative incentive according to collaborative situations and promote sharing among government data departments. (2) It bridges the explanatory framework of existing researches. Develop big data management organizations to promote the sharing of government data departments of four collaborative modes, influencing factors and their effects.The policy implications of this paper are as follows: based on the collaborative mode and validity and limits of big data management agencies promoting sharing among government data departments, it is still necessary to strengthen the power and responsibility determination of government data in the future, improve the system and mechanism of big data management agencies promoting efficient business collaboration and on-demand data sharing, and improve the realization path of autonomous department collaboration.