Identification And Inference Of Heterogeneous Policy Effects: From The Perspective Of Conditional Parameters Of Dependent Variables

In micro-econometrics,the Oaxaca-Blinder Decomposition and Difference-inDifference are two classical empirical research methods.However,the two methods also have limitations in that they can only measure their average effects and cannot reflect the heterogeneity of empirical content.Heterogeneity analysis is an important research direction in empirical research in economics,which can reflect the parameter differences between different groups,and has important application value in empirical research.First of all,in order to overcome the OB decomposition,only the limitation of the difference in mean between groups can be investigated.Many scholars have proposed a variety of heterogeneous decomposition methods,such as JMP1993,DFL,MM and FFL decomposition method.Secondly,in view of the limitation of the mean value of treatment effects in the policy evaluation of the difference in difference,some scholars have also proposed heterogeneous difference in difference methods.For example,the quantile treatment effect(QTE)is the most common way to characterize policy heterogeneity.However,the corresponding methods proposed by the abovementioned scholars all have some limitations.For example,the JMP1993,DFL,and MM methods in describing the heterogeneity of OB decomposition cannot decompose the effect into the individual contribution of each variable;the QTE method of describing heterogeneity includes at least two periods of observations,which are defined and identified on this basis QTE is more complex than the single-section data model.This article proposes a method of decomposing the sub-populations based on the value of the dependent variable at different levels,so it is called the outcome conditioned(OC)decomposition.JMP1993,DFL,and MM are unable to decompose the effect into the individual contribution of each variable.The goal of OC decomposition is to subpopulations whose dependent variable values are at different levels between the two groups and have the ability to decompose the difference of the dependent variable into each explanation variable contribution.Similarly,unlike the quantile treatment effect(QTE)in the existing DID model,we make full use of the important feature that the DID implementation environment contains at least two periods of observation and compare the policy effect with the initial value of the target variable(Y~0),and propose a parameter called Outcome Condition(OC)-DID.OC-DID can identify the heterogeneous effects of policies on these sub-populations in different positions of the dependent variable distribution before the implementation of the policy.The second chapter of this article first adopts the OB decomposition method and the RIF quantile decomposition method to empirically investigate the wage differences caused by the “dual-track” employment in the public sector in my country.The study found that there is a significant difference in the wages of employees inside and outside the organization,and the wages of regular employees are significantly higher than those of contracted employees.The organization of wages differs significantly,and the wage gap caused by the endowment difference is greater than the wage gap caused by the coefficient difference.There is heterogeneity in the establishment wage difference of the public sector,which is mainly manifested in that the establishment wage difference of low-wage personnel is greater than that of high-wage personnel.The third chapter of this article proposes a new “counterfactual” decomposition method to analyze the income growth and distribution trends and influencing factors of people with different income levels.Unlike the traditional Oaxaca-Blinder(1973)method,which decomposes the average income difference of the overall population,the new method can decompose the income difference of the population within a given range,so it is called the dependent variable condition(outcome conditioned,OC)break down.Compared with the RIF decomposition widely used in recent years,the OC decomposition can distinguish the income changes of individuals in the same income class from the mobility effects of individuals between different income classes by controlling the “counterfactual” income levels of individuals.The article uses CHIP data from 1995 to 2013 to examine the evolutionary trends and heterogeneity characteristics of income growth of different income groups brought about by the rapid economic development during this period.The study found that if the decomposition method that does not control the income class is adopted,the income growth rate of people with different income levels is relatively consistent;when the OC decomposition is used to control the income class of the individual,the income growth rate is closely related to the income level of the individual,which shows that Different income growth rates have led to individuals moving between different income groups.The article further analyzes the role of individual education level in it.The empirical results show that compared with high-income groups,the rate of return to education of the middle-income group is lower,while the improvement of the education level of the low-income group has a significant effect on their income.Shows the heterogeneous effect of education on class mobility.The fourth chapter of this paper uses the DID method to evaluate the policy effect of rural drinking water safety projects on rural residents' income increase.The study found that the rural drinking water safety project has significantly improved the probability of safe drinking water and the income level of rural residents.Among them,the drinking probability of tap water has increased by 40%,and the average income has increased by nearly 7%;the income increase effect of the rural drinking water safety project has regional and educational differences.Among them,the income-increasing effect of rural residents in the central provinces and junior high school and below is the most significant.The fifth chapter of this article proposes a new outcome conditioned-DID method,which can identify sub-populations in different positions of the dependent variable distribution before the implementation of the policy,and the heterogeneous effects of the policy on these sub-populations.It can reveal the dynamic characteristics of the policy's influence on the sub-populations with different initial values of the target variable over time.The identification of OC-DID does not rely on the “common trend”assumption,while allowing policies to be endogenous.We give the steps of identifying,estimating and inferring parameters when the policy is exogenous and endogenous.Using this new method,this article evaluates the heterogeneous characteristics of the labor protection role of the new “Labor Contract Law” implemented in 2008 for subgroups of migrant workers with different initial values of various welfare variables.We found that the new labor law has the greatest increase in the income of migrant workers with lower initial income;the longer the work week before the implementation of the new law,the more their working hours will decrease after the implementation of the new law.The research in this article provides a further targeted answer to the new labor law's protection of labor rights and interests of disadvantaged groups represented by migrant workers.At the same time,our method provides a new way to identify the net effect of endogenous policies in the absence of instrumental variables,and also provides a new analytical tool for evaluating whether other policies can be applied to potential target groups in the future.The micro-econometric method is an important tool in the empirical research of economics,which is directly related to the possibility of the exploration direction of economic empirical research and the richness of the exploration content.In this paper,the empirical research is carried out more rationally through continuous improvement of innovative measurement methods.At the same time,combined with specific empirical examples,the paper uses improved measurement methods to investigate the practical significance of class issues and labor protection issues.