The Experimental Methods Of Development Policy Evaluation And The Application

All along,development policy runs through economic development,and evaluating the impact of policy is one of the three core economic tasks(Heckman and Vytlacil,2005).The study of policy efficacy is merely about answering "does it work?",but more about answering "when does it work and why it works?"(Rodrik and Rosenzweig,2010).As methods of dedicating to causal identification,randomized controlled trials,natural experiments and quasi-experiments have promoted the transformation of policy evaluation methods from traditional structural methods to treatment-effect methods.These experimental methods also bring about the credibility revolution in empirical economics(Angrist and Pischke,2010).At present,the policy evaluation based on natural experiments and quasi-experiments is accelerating in China,but overall,there are still many deficiencies.On the one hand,there are numerous problems in the policy evaluation based on natural experiments and quasi-experiments in China,which makes the credibility of the research questionable.On the other hand,Chinese scholars do not understand the randomized controlled experiments well enough,and there are some deviations in cognition.Therefore,this dissertation aims to systematically introduce the development and application of randomized controlled trials,and detailly discuss the controversy and frontier development of randomized controlled trials,as well as the correlation between randomized controlled trials and natural experiments,quasi-experiments.On one hand,it can deepen the understanding of randomized controlled trials in Chinese academics,in order to better conduct research and design;on the other hand,it provides new research perspectives and ideas for non-experimental researchers to carry out policy evaluation.Further,taking the policy of China's mandatory corporates to disclose Corporate Social Responsibility as a pointcut,this thesis examines the practical application of natural experimental methods in China's policy evaluation.The main contents and conclusions of this thesis are as follows:First,reviewing and teasing out the rise and application progress of randomized controlled trials.Taking Jameel Poverty Action Lab(J-PAL)as the research platform,the application fields of randomized controlled trials has extended to many development issues involving agriculture,education,finance,health,labor market,corruption governance,political participation;the research contents are also more abundant,in addition to evaluating specific poverty alleviation policies,it has also been used for theoretical discussion and mechanism analysis.At the same time,domestic and foreign scholars have also started randomized controlled trials in China and focus on policy or project interventions to improve children's health,nutrition and education.Second,systematically discussing the controversy and frontier development of randomized controlled trials.This paper is based on the framework of internal validity and external validity to discuss this topic.The controversy of randomized controlled trials mainly focuses on the following aspects:(1)internal validity issues;(2)external validity issues;(3)general equilibrium problems;(4)lack of theory and mechanism;(5)ethical issues.Corresponding to the controversy of randomized controlled trials,the frontier development of randomized controlled trials is mainly reflected in the following aspects:(1)more perfect identification methods;(2)improvement of external validity;(3)combination of randomized controlled trials and theory;(4)the attempt of general equilibrium effect and long term effect.Third,explaining the interrelationship between natural experiments,quasi-experiments and randomized controlled trials.On the one hand,there is a complementary relationship between them.Based on the causal identification framework of randomized controlled trials,natural experiments and quasi-experiments can well solve the causal identification problems caused by non-randomization in observation data under certain conditions.On the other hand,natural experiments,quasi-experiments and randomized controlled trials are closely related to each other and mutually promote.Randomized controlled trials make natural experiments and quasi-experiments more mature in research design and method application,and more similar to randomized controlled trial;natural experimental research and quasi-experimental research make randomized controlled trials more perfect in identification method.Fourth,evaluating the impact of the policy that forces corporates to disclose social responsibility on corporate performance.Overall,the policy of mandatory disclosure of corporate social responsibility results in a significant decline in corporate performance.Compared to firm performance measured by ROA,the policy has a greater impact on firm performance measured by Tobin Q.Further research finds that:(1)The policy significantly reduces corporate performance in the long term,whether measured by ROA or Tobin Q.The long term effects on ROA are "inverted U-shaped" and long term effects on Tobin Q are "U-shaped."(2)State-owned nature further deepens the negative impact of the policy on corporate performance,whether measured by ROA or Tobin Q.;(3)Higher levels of marketization and political connections will significantly mitigate the negative effect of the policy on corporate performance(measured by ROA),but have no significant effect on the Tobin Q.Fifth,to evaluate the impact of the policy that forces corporates to disclose social responsibility on corporate financing constraints.Overall,the policy of mandatory disclosure of corporate social responsibility results in a significant decline in the financing constraints of the sample corporations.Further research shows that:(1)The policy of mandatory disclose corporate social responsibility reduces corporate financing constraints in the long term,and the dynamic impact in long term is "inverted U-shaped";(2)Although,state-owned nature and higher marketization level mitigate the financing constraints in a certain extent,but it's not significant;(3)The political connection can not reduce the financing constraints,but it has an inhibitory effect on them.This dissertation systematically reviews and discusses the experimental methods for the development policy evaluation,and evaluates the policy of China's mandatory disclosure of corporate social responsibility based on the natural experimental methods.It is of great significance to strengthen the understanding of randomized controlled trials,to conduct better research design for China's policy evaluation based on natural experiments and quasi-experiments,and to enhance the credibility of China's policy evaluation.