The Evaluation Methodologies And Empirical Applications Of Causal Effect In Multiple Policies

Economic policy is a vital field of economic research.The formulation and implementation of macroeconomic to regulate macroeconomic operation and microeconomic policies to regulate the behavior of microeconomic agent are important means for the government governs the country.The effectiveness of policy depends on the level of policy formulation,in order to improve the effectiveness of policy,so it needs to be evaluated by scientific methods.Policy evaluation is an important basis for determining policy selection,enhancing policy quality and optimizing policy combination.The core of policy evaluation is the evaluation on causal effect between policy intervention and policy consequence.Generally,the causal effects of policy intervention are evaluated based on observational data,and the main difficulty to carry out the conclusion of cause and effect by using observational data is that the presence of confounders distorts the cause and effect between policy intervention and policy consequence.The random assignment mechanism can balance the distribution of confounders in each group.Therefore,statistical inference based on random sampling can reveal causal relationships and estimate causal effects.Causal effect assessment of economic policy is usually based on observational data under natural or quasi-experimental conditions,which does not meet the conditions of random assignment.At this time,causal relationships can not be revealed by the traditional statistical inference methods,so it is necessary to develop causal inference methods for observational research.At present,there are some methods of causal effects evaluation in natural experiments which have similar ideas:on the basis of randomized experiments,and under the counterfactual framework,the hidden allocation mechanismis are restored by virtue of "study design" to simulate the analysis mechanism of randomized experiments,and the corresponding estimation methods are developed by gradually relaxing the determination conditions.Most of the existing theoretical researches have focused on a single policy evaluation at the aspect of policy practice,but in practice,it is more common for multiple policies to be implemented at the same time.In order to select the optimal policy or policy combination,it is necessary to separate the causal effects of each policy and compare the size of the causal effects.Compared with single policy assessment,the establishment,identification and estimation of multiple policy causal models are complicated by the increasing of policy processing variables,and further research is needed.In view of the fact that the multiple policy causal effects evaluation needs the support of relevant theoretical methods,but the relevant research is scarce,this research carries out the research as follows based on the causal effects evaluation method for observational research by predecessors:First,the theoretical research on evaluation methodologies of causal effects in multiple pilicies:The research,on the basis of sorting out the development of non-binary treatment causal inference,studied the causal effects evaluation strategies of several independent binary treatments variables,and utilized Monte Carlo Simulation Method to analyze the finite sample nature of causal effects estimators.(1)Based on the Potential Causal Model,defined the causal effects of each policy under the double policies,and given the corresponding identification hypothesis,namely the unconfoundedness hypothesis or the conditional unconfoundedness hypothesis.Furthermore,under the hypothesis of conditional unconfoundedness,put forward the specific estimation methods:matching,inverse probability weighting,difference-in-differences and estimation based on generalized propensity score.In addition,simulated and analyzed the statistical properties of each estimator.Each estimator has good properties of finite samples except the inverse probability weighting method with the generalized propensity score,which is demonstrated by the simulation results.(2)Spread the assessment strategies of causal effects of double policies into multiple policies processing,put forward the causal effects identification hypothesis in multiple policies(unconfoundedness and conditional unconfoundedness),the corresponding estimation methods,including matching estimates based on binary propensity score,as well as inverse probability weighting estimation,regression estimation based on the difference-in-differences,and matching,inverse probability weighting estimation,regression adjusting estimation based on the generalized propensity score.(3)Paid special attention to the comparison of the causal effects of various policies in multiple policies processing,and given two matching estimation method for the comparison of causal effects.This research expanded the existing theoretical research on causal inference through in-depth and systematic research on the issues related to the causal effects of multiple policies processing.Second,the empirical application of causal effects evaluation in multiple policies:Scientific and technological innovation is the driving force of economic development.In order to promote the innovation of science and technology,and to achieve economic growth with high quality,the Chinese government has formulated the incentive policies to promote the development of new and high technology industries,regards the recognition of high-tech enterprises and establishment of national high-tech industrial development zone as a policy orientation.The research focused on the analysis of the impact of high-tech enterprise recognition policy on the economic performance of private enterprises,and eliminated the impact of high-tech zones by right of difference-in-differences under the evaluation of multiple policies,and obtained the net influenceof high-tech enterprise recognition policy on the economic performance of private enterprises.The following conclusions are drawn:(1)In general,the policy of high-tech recognition policy plays a significant role in promoting the economic performance of private enterprises,and the policy effect rises with the time fluctuation.(2)The policy effect is closely related to the period of recognition.On the one hand,it reflects the rapid impact of policies on enterprises,and on the other hand,it also shows that enterprises are strongly dependent on the policy.(3)The further mechanism analysis demonstrates that the high-tech recognition policy effectively relieves the financing constraints faced by enterprises,and the credit market financing is a more effective financing channel,as compared with the capital market financing.Finally,corresponding policy suggestions have been came up based on the empirical conclusion.The empirical conclusion of this research provides a more scientific theoretical reference for the government to formulate effective innovation incentive policies and implement dynamic and precise management.