A Study On Threshold Effects In Panel Data Model And Its Application In Local Debt Problems

Economists and policy makers generally believe that massive fiscal deficits and bank bailouts can effectively mitigate the recession.With the outbreak of the global financial storm in 2008,the public debt of various countries increased sharply,and then many new economic problems such as the Euro sovereign debt crisis appeared.Therefore,economists began to evaluate the effect of using rapid expansion of public debt to fight against economic crisis,and found that there is a “critical point”(debt threshold)for the impact of debt on economic growth.Beyond this critical point,debt significantly inhibits economic growth.However,due to the different models used or the different data sets analyzed,researchers have no agreement on what the value of debt threshold is.At the same time,some scholars have noticed that the external environment and national policies have common impacts on the local economic system.These common impacts and their spillover effects among regions not only lead to strong crosssectional correlation among a large number of macro variables,but also become the root of many endogenous problems.In recent years,the widely studied relationship of local debt and economic growth in China,is also affected by these common effects and has some severe problems such as the endogeneity of threshold variable and cross-sectional dependence.Therefore,either from the view of academic research or practical implication,it is of particular importance to find a suitable model to study the impact of local public debt on economic growth for China,a country in rapid transformation.The content of this thesis includes:Firstly,we review the development process of Chinese local debt and its relationship with economic growth through extensive and detailed descriptive statistical analysis.We find that national policies and external economic environment not only have an impact on local economic growth,but also significantly affect the expansion of local debt.This proves that there are some common factors underlying local debt and economic growth in China.In addition,because these factors and their related spillover effects are not observable,they can not be represented by certain economic variables,which makes it necessary to add unobserved common factors into the usual error structure in our model.Secondly,we combine the common correlation effects in the error term with the standard neoclassical growth model,and adopt the heterogeneous linear model,heterogeneous asymmetric model with a given threshold value and heterogeneous symmetric model with the quadratic term of debt,respectively,to study the relationship between local debt and economic growth in China.1.In the heterogeneous linear model,we find obvious individual heterogeneity in the impact of local debt on economic growth,and some preliminary evidences of debt threshold effect through the fitted graphs.2.From the heterogeneous asymmetric model,we find that there is a “critical point”for the impact of local debt on economic growth in the average sense,but the debt threshold effect is not statistically significant due to the heterogeneity.3.The heterogeneous symmetric model has arrived similar conclusions as the asymmetric model,which in some sense shows that the findings of the asymmetric model are robust.Thirdly,we overcome the drawbacks of different models aforementioned,study a panel data model with an unknown threshold in the framework of common correlated effects structure,and provide the estimation method and testing statistic for the threshold effects.Monte Carlo simulation are carried out to investigate the finite smaple performance for the estimator and testing statistic.In empirical analysis,we found an inverted U-shaped relationship between local debt and economic growth,but the relationship is not statistically significant;we also found obivous difference in the Ushaped relationship between high-income and low-income cities.Fourthly,we take the influence of macroeconomic variables on the threshold parameter into consideration using a functional threshold panel data model with common correlation effects error structure.The estimation and test statistics for thershod effect are provided.Monte Carlo simulation shows that the estimator and testing statistics have good small sample properties.In the empirical part,we introduce the lag of debt or inflation rate of the same period as the decisive variables of threshold parameters,respectivley.Finally,it is found that the threshold effect is significant no matter which variable is added to the threshold function.The main contributions of this thesis are summarized as follows: 1)Theoretical contribution.In the framework of common correlation effects,the constant threshold model and the function threshold panel model are studied,respectively,and the statistics for the slopes and testing the threshold effect are both given.Through a large set of Monte Carlo simulations,we find our estimator and testing statisitcs have very good finite sample performance.2)Empirical findings.Firstly,we confirm that there are some external factors jointly affect the local debt and economic growth and then justify the use of our models;secondly,our study is the first one to consider the impact of external shocks and spillover effects in the literature of Chinese local debt problem,and we adopt various empirical model setups from different perspectives;thirdly,this thesis relaxes the homogeneity assumption among different cities,and studies the problem of local debt in China for the first time in the framework heterogenous slopes;finally,the assumption of constant debt threshold is relaxed and the relationship between debt and growth is investigated in the framework of functional threshold,which provides a different perspective for policy makers to formulate related policies for local government debt.Our empirical studies do not find sufficient evidence to support the existence of a common debt threshold among the different cities in China,but support the existence of time-varying and heterogeneous debt threshold effect.Considering the recent situation of Chinese economic development,some policy suggestions can be provided according to our findings.First of all,the government should fully understand the non-linear relationship between the scale of local debt and economic growth,and be alert to the negative impact of too fast issuance of local debt.Secondly,the debt threshold is timevarying and heterogeneous across different cities.To regulate the local debt scale,government department should formulate policies for different regions at different period according to the real economic situation.