| The panel data model can increase the amount of information in the data,and can describe a variety of heterogeneity,and is more and more widely used in empirical research in the field of economics.However,panel data models that are widely used in empirical research usually assume that the model is linear and that there is no inter-individual correlation between random error terms.However,these two assumptions are inconsistent with many observations in reality.For example,at the micro level,due to the education,living environment and cultural habits,there will be related relationships among individuals;at the macro level,due to macro policies,spillover effects,and externalities,it will lead to correlations between countries and regions.relationship.These individual correlations cannot be eliminated by adding individual heterogeneity variables(Chudik et al.,2011).The linear hypothesis of panel data ignores the existence of a large number of non-linear economic variables and phenomena in reality.For example,in financial data analysis,it is often found that financial variables have asymmetry,leverage effects,and herding effects;at different stages of the economic cycle,economic growth also produces different transformations,and these effects usually make the variables appear non-Linear and asymmetric features.At this time,linear models are used to characterize these variables,which will lead to the wrong choice of models and result in biased results.Therefore,the research object of this paper is the estimation and test of the panel smooth transition model based on the logical function under cross-section correlation.There are two main research lines in this paper.One is the estimation problem of the smooth panel transition model.In this problem,this article will focus on how to estimate the static and dynamic panel smooth transition models with cross-section correlation,and deal with the unbalanced panels.Problems and endogenous problems.The other is the test problem of the panel smooth transition model.In this problem,this article will try to solve the problem of nonlinear unit root test and co-integration test.The main content of the full text is summarized as follows:(1)This paper examines the estimation of the static panel smooth transition model under cross-section correlation.The CCE estimator proposed by Pesaran(2006)is used to estimate the panel smooth transition model with cross-section correlation.Based on the CCE estimator,A nonlinear Wald test is proposed to test the smooth transition model of the panel.In this paper,a Monte Carlo simulation is performed on the Wald test statistics and the panel smooth transition model PSTR-CCE estimator.The limited sample properties show that the Wald test statistics are more powerful than the LM statistics that do not consider cross-section correlation,and Has a relatively stable inspection level.After comparing the standard PSTR estimator,PSTR-CD estimator,PSTR-CCEP estimator,and the PSTR-CCEMG estimator proposed in this paper,we can find that when the sample size is small and you want to get the exact estimated value of the conversion parameters,use The CD estimator is better,and when the number of sections is large and a more accurate position parameter estimate is desired,the PSTR-CCEMG estimator is more appropriate.Then this article applies the PSTR-CCEMG estimate to the empirical content and examines the relationship between population mortality and income.Because this paper uses micro data,there is the problem of this unbalanced panel.In order to solve this problem,this paper uses the Asymptotic Least Squares estimator estimator to obtain the estimated coefficient of the unbalanced panel smooth transition model,and bootstrap The corrected standard error of the estimated coefficient is obtained.The empirical results in this article show that for different income groups in different economic cycles,the relationship between population mortality and income is different.For low-income groups,during the economic boom,income and mortality are positively correlated,which is the procyclical phenomenon of mortality pointed out by Ruhm(2000);during economic recession,the relationship between income and mortality is negatively correlated.Yes,the procyclicality of mortality has disappeared.For high-income groups,there is no significant correlation between income and mortality during the economic downturn.However,during the economic downturn,the increase in income of high-income groups will result in increased mortality and a procyclical rate of mortality.phenomenon.(2)The estimation of the dynamic panel smooth transition model in the presence of cross-section correlation is examined.Based on the form of the logical smooth transition function,this paper proposes the DCCE estimator of the panel smooth transition model under the cross-section correlation and dynamic assumption.The asymptotic properties of the DCCE-2SLS estimator in the presence of endogenous explanatory variables are presented.In order to investigate the limited sample nature of the estimator,a Monte Carlo simulation is performed in this paper.The simulation results show that when the CCE-2SLS estimator is used to estimate the dynamic model,the estimator is biased,and the DCCE-2SLS is used to estimate the model.Good limited sample nature.In the empirical application section,the impact of government debt as a percentage of GDP on income inequality is examined.In this paper,a nonlinear test and a cross-section correlation test are performed on the benchmark model.The test results show that the benchmark model should be estimated using a nonlinear model and a DCCE-2SLS estimator with cross-section correlation.This paper uses an endogenous dynamic panel smooth transfer model to study the relationship between government debt ratios and income inequality.The estimated results of the model show that for more developed OECD countries,when the government debt ratio exceeds 26%,An increase in the government debt ratio will increase the degree of income inequality;for developing emerging market countries,when the government debt ratio exceeds 51.67%,the increase in the government debt ratio will have the effect of expanding income inequality.At the same time,this article also examines the effect of government debt ratios around the threshold on factors affecting income inequality,and the results show that when debt levels are low,for emerging market countries,to reduce income inequality,While ensuring economic growth and employment stability,we should increase the debt level moderately and reduce the proportion of investment in the medical and health sector;when the debt level has reached a certain level,we must consider the reasonable degree of income distribution,reduce the debt level and increase The proportion of government investment in the medical and health field can narrow the income gap.For the more developed OECD countries,when the debt level is low,in addition to a modest increase in debt level,it is also necessary to increase the proportion of government investment in the medical and health field to reduce income inequality;when the debt level exceeds a threshold,focus on Increasing the rate of economic growth,reducing the proportion of government investment in health care,and reducing debt levels will help reduce income inequality.(3)Based on the second-generation panel unit root test method proposed by Pesaran(2007),this paper proposes a LSTAR process that assumes a linear unit root process under the cross-section-dependent condition and an alternative hypothesis that the non-linear logic function smoothly transfers the overall stable process.-NPU panel unit root test,constructed the individual tNLCIPSistatistics and panel tNLCIPS statistics,given the asymptotic distribution of the statistics,and given the critical values of the corresponding statistics by Monte Carlo simulation.The results of Monte Carlo simulation show that when the data generation process is subject to a non-linear process,the second-generation panel unit root test using Pesaran(2007)will cause serious distortion of the test level and a decrease in test power;when the data generation process follows the logic Functional form,and using the exponential non-linear panel unit root test method(ESTAR-NPU)to test,it will cause distortion of the test level and decrease the test power.When the data generation process is in the form of a logical function,the logical panel non-linear panel unit root test method(LSTAR-NPU)proposed in this chapter is used to test,and the correct test level is obtained.The test efficiency is better,and it is generally higher than the error.The power of the panel unit root test.(4)This paper examines the estimation and testing problems of the panel smooth transition co-integration model under cross-section correlation.In this paper,the FMOLS estimator is used to estimate the panel smooth transition co-integration model under cross-section correlation.The panel tFMOLS statistic and the panel LM statistic are constructed to test whether the non-linear co-integration relationship exists.Near nature.After Monte Carlo simulation,it is found that the use of FMOLS estimators has a great advantage in dealing with endogenous and heterogeneous models.Then in the empirical part,we examine the problem of international capital flows by examining the mystery of Feldstein-Horioka.This paper uses data from 21 OECD countries and 10 emerging market countries from 1996 to 2018,selects 6 variables to characterize the level of government governance,and uses it as a transfer variable for the long-term relationship between investment and savings rates.Tested.The empirical results show that the level of government governance is indeed an important factor affecting the relationship between the investment rate and the savings rate.There is a stable non-linear relationship between the investment rate and the savings rate.When the level of government governance is improved,the The positive relationship will weaken,thereby promoting the international flow of capital,and the positive relationship between the savings and investment rates of emerging market countries will be strengthened,which partially explains the mystery of Feldstein-Horioka.The innovations of this paper are as follows:(1)A non-linear panel inspection method based on cross-section correlation is proposed.In the presence of unbalanced panels,an easy-to-apply and easy-to-estimate method for unbalanced panels is given.(2)Existing literatures rarely study dynamic panels,endogenous explanatory variables,and cross-section correlations in the framework of panel smooth transition models,and these three problems often appear simultaneously in empirical evidence.In this paper,endogenous,dynamic panel and cross-section correlation are unified in a framework,and an estimator is proposed to solve the estimation problem in the state of non-linear panel,which provides a theoretical method for empirical research.(3)In the framework of a panel smooth transition model based on a logic function,a non-linear panel unit root and co-integration test statistics under cross-section correlation are proposed.On the premise that the second-generation panel unit root and co-integration tests can gradually improve the power of the test in the existing literature,the non-linear panel unit root test and co-integration test are closer to reality,but the existing literature does not consider the two.The corresponding test statistic proposed in this paper takes into account the situation of this combination,and has a stable test level and good test power.(4)The empirical work in this paper also adds a new perspective to the existing literature.This paper supports the theory of purchasing power parity through a non-linear panel unit root test,and expounds the hypothesis of international capital flows through a non-linear panel co-integration test.The relationship between government debt and income inequality is studied through the estimates presented in this paper.It shows the extent to which government debt affects income inequality,and provides empirical evidence for policymakers. |
PDF Full Text Request