Abstract: The Principal-Agent Theory is a research on incentive contract to promote the agent to fulfill the purpose of the principal in lack of perfect information based on the main tools of the static incentive model and the dynamic incentive model. This paper focuses on how the leaders will motivate the workers to do long-run innovation and improve the policy incentives in the enterprise's dynamic mechanism in the innovation management of the enterprise in lack of some information by light of the principal-agent dynamic model. This paper construct respectively the continuous-time models for the distribution of the employee's efforts in the conventional work and the innovative work under the situations where the moral hazard exists only, and where both the adverse selection and the moral hazard and coexist, and discusses the changes of the optimal incentive contracts under such two circumstances. It is also found out through comparison that the effort level of employees will remain unchanged when the past in irrelevant to the future and they will select the directions of efforts according to their own interests ignoring the innovation of the enterprise. Thus, it is proposed that in the dynamic mechanism of policy motivation in the enterprise innovation, both the past performance and the effort costs of the employees should be included in the performance evaluation system to motivate the workers to adjust their effort levels on innovation at different periods.

Key words: Principal-agent theory, dynamic incentive model, enterprise innovation dynamic mechanism, policy incentive, performance evaluation system