Although government performance evaluation has undergone extensive scrutiny in academia, the existing literature largely hinges upon single-case studies, and considers the evaluation procedure a set of ‘complete’ contracting actions. Yet, there remains unclear about how power configurations within an organization have an impact on the process and outcomes of government performance evaluation. To take up this research question, this article draws on comparative case studies, and conducts a series of empirical investigations into the performance evaluation processes in three governmental organizations——The Municipal Work Committee of City A, The Organization Department of District B, and The Local Taxation Bureau of County C. Through a theoretical lens of control rights, this article develops a more sophisticated analytical framework, which consists of three dimensions, i.e., ‘target setting power’, ‘evaluation power’, and ‘incentive distribution power’, thereby unraveling different power configurations in these three cases. On this basis, we dissect how different types of power relations have led to varying attitudes and behavior of different stakeholders. Finally, we conclude three distinctive patterns of government performance evaluation, which include the ‘interactive’, ‘dominant’ and ‘direct-control’ modes. 
   This article attempts to make the following theoretical contributions: first, the comparative analysis across the aforementioned three cases substantiates the conceptual value of residual control rights in the research of government performance evaluation. Although these three cases are very much different in their functions, levels, goals, etc., they consistently show that performance evaluation is a dynamic and interactive process. In the first place, the planning of performance evaluation needs to be achieved through different forms of interaction and negotiation. Even after the evaluation plan is formulated, it is subject to any adjustment according to the changing contexts. There is also room for flexibility in the form of inspection and acceptance, as well as in the way of interpretation and announcement of results. The analytical framework of threefold dimensions i.e., ‘target setting power’, ‘evaluation power’, and ‘incentive distribution power’ helps us make classification among different cases, which provides the possibility for further abstraction and theorization of government performance evaluation.
    Second, in the process of performance evaluation, the residual control rights held by the evaluator often determines the primary configuration of evaluation, therein lies different types of interrelations between evaluators and evaluatees. The less the residual control rights held by the evaluator, the higher scientificity and objectivity of the evaluation design are required (e.g., the evaluation project organized by the Municipal Work Committee of City A); on the contrary, the more emphasis is placed on the evaluation results and the corresponding incentive distribution, the focus on evaluation procedures and methods will tend to be reduced (e.g., the evaluation project undertaken by the Organization Department of B District). Nonetheless, excessive residual control rights held by the evaluator will lead to out-of-balance power relations, which will render the objectives of performance evaluation ineffective (e.g., the case of the Local Taxation Bureau of C County).
    Third, these three residual control rights—i.e., ‘target setting power’, ‘evaluation power’, and ‘incentive distribution power’ are not at the same degree of importance, or does not work in the same way. More specifically, ‘incentive distribution power’ appears to be more decisive, which have an impact on the other two to a great extent. Besides, both the ‘target setting power’ and the ‘evaluation power’ are more likely to be transferred to the third parties through outsourcing. For example, the evaluator tends to assign part of the responsibility (such as the design of performance plans, and the inspection tasks, etc.) to a third institution. Outsourcing has become a common phenomenon in contemporary practices of government performance evaluation. However, the involvement of the third-party power may change the distribution of residual control rights, and thus would cause any uncertainty. Yet, viewed from the above three cases, the role of the third parties varies according to the difference in the allocation of residual control rights between the evaluator and the evaluatee. Under this circumstance, the third parties tend not to assume a decisive role.
     Taken together, this article suggests that government performance evaluation is fundamentally an incomplete contracting process, which is embedded (and thus is mirrored) in the processes of target setting, implementation, measurement, and incentive provision. While Zhou and Lian′s seminal work (2012) employ the notion of ‘residual control rights’ to delves into the relatively static power relations among different levels of governments, this article endeavors to shed new light by indicating the variegated power configurations in government performance evaluation, which are formed by different stakeholders at multiple levels and positions. In these processes, residual control rights held by all parties affect the logic of their actions directly, which in turn make them internalized into different modes of government performance evaluation.