Study On SVR-based Quantitatively Method For Incentive Contracts

Incentive contract optimization models include three types of optimization models:moral hazard model, adverse selection model and signaling problem model. Over the past decade, researches on incentive contract optimization models basically are limited in analytical analysis based theoretical methods; therefore the applications of the models are just stay at the qualitative analysis level. That makes it impossible for these models to be put into practical applications in a quantitative manner. In order to address this issue, this thesis proposes emulation methods to solve incentive and contract optimization models includes one moral hazard model (the model in the example of linear payment of basic salary plus the portion of the income), two adverse selection models (one is the model in which the principal designs the contract for the purpose that he can incent the agent work hard no matter the agent is in good or bad work condition; the other is the model in which the principal designs the contract for the purpose that he can also incent the agent to select the option reflecting the agent's type) and two signaling problem models( one is the model in which the principal designs incentive contracts to make the agent deliver his characteristic signal; the other is the model in which the principal designs incentive contracts to deliver his own characteristic signal to the agent). The key of the incentive contracts for the optimization model is to use the non-analytical expressing method for the utility function. By using SVR to modeling the utility functions encountered in incentive contract models in my emulation, the problem is properly solved that the utility function can not be quantitatively expressed by analytic functions, and based on this method the relevant quantitative computational models for incentive contract optimization models are constructed respectively.The main works of this thesis are as follows:First of all, a quantitative-calculation-based method is proposed for emulating moral hazard contract model. Being able to calculate quantitatively, we use SVR to express the utility function and use equilibrium analysis to solve the optimized model with multi-extreme points. Using our method, we quantitatively analyzed the moral hazard optimized model with the payment of basic salary plus the portion of the income.Secondly, a SVR-based quantitative calculation method is proposed for solving adverse selection incentive contract mode. Being able to calculate quantitatively, the support vector regression (SVR) is used to express the utility function. Based on that, the gradient expression of adverse selection model is derived for both good natural condition and bad condition, and that of adverse selection model is also derived for both high efficient agent and low efficient agent, respectively. Then, the relevant gradient descent algorithm is given. Using the method, the two adverse selection incentive contract models above are quantitatively analyzed, and the effects of parameters'varying to the changing trends of adverse selection model equilibriums are observed.Finally, A SVR-based quantitative calculation method is proposed for solving signaling incentive contract model. Being able to calculate quantitatively, the support vector regression (SVR) is used to express the utility function. Based on that, the gradient expression of signaling model is derived for both high efficient agent and low efficient agent to signal to the principal designing the contract, and that of signaling model is also derived for both the principal providing high productivity with heavy workload and the principal supplying low productivity with easy workload, respectively. Then, the relevant gradient descent algorithms are given. Using our method, the two signaling incentive contract models above are quantitatively analyzed, and the effects of parameters'varying to the changing trends of the two signaling models'equilibriums are observed.Through the quantitative analysis of the three models of incentive contract, the study shows that quantitative analysis results are consistent with the existed theoretical results and contain more information, and confirmed the validity of the quantitative analysis methods. The rationality of the results also shows that the quantitative analysis method for incentive contract model is feasible.