Research On Economic Growth And Inflation Forecasting And Its Influence Factors Of China

Since the international financial crisis in 2008,there has been a decline in out put surplus,total factor productivity and a significant decline in global trade growth.In the post financial crisis period,under the background of the global economic re covery is weak,China's economy has entered a new normal since 2012.In the new normal,China's economic structure and development momentum,and so on,have b een a great change.The external and domestic demand show weak at present,and as the troikas of pulling the economy,investment,consumption and net exports are all sluggish,at t he same time,the supply side of the indicators,such as population,capital output r atio and total factor productivity,are not optimistic.In the meanwhile,the adjustme nt of economic structure is at a critical period,whether the economic structure can be smoothly shifted from secondary industry dominate to tertiary industry dominate or not,realizing the transformation and upgrading of manufacturing industry,relyin g more on consumption to promote its economic development,are all important issues of the current that need to be solved.In such a complex economic context,in order to make the economy develop steadily and healthily,it needs to have a correct understanding of the economic development in China,so that policy making departments and other market players make adequate preparations in advance,and design corresponding policies and market strategies based on the future economic situation,so as to achieve the goal of adapt-ting the new norm and leading the new normal.But in the new normal,the former economic forecasting methods are imperfect to adapt the new problems that we face,therefore,in this thesis,we apply the frontier model of econometrics to study the economic forecasting methods in many aspects.According to the development of macroeconomic forecasting methods in recent years,this thesis uses a few new methods or models to predict and analysis the China's macro-economy under the new normal,and tested the prediction accuracy of these models.Specifically,three types of models,which are used in this thesis,are as follows:The first kind of model is VAR.Since C.Sims proposed this model,VARs have been widely used in the empirical analysis and forecasting of macro-economy,because of capturing the dynamic relationship between variables effectively.However,this model contains so many parameters,and the number of parameters which need to be estimated is increasing rapidly with the increase of lags and the number of variables in the model,this leads to the problem of over-parameterization.With the development of econometrics,especially the appearance of Bayesian estimation method,econometricians have proposed many parameter shrinkage methods to restrict the parameters of VAR models,in order to avoid the problem of over-parameterization.Recently,two important methods,Bayesian stochastic variable selection and large Bayesian VARs,have been proposed.For the first method,based on the data set,it calculates the variables that should be included in the model by Bayesian algorithm,and imposing strict restriction on the parameters of variables that should not be included in the model(that is,these parameters are confined to a small area around 0).For the second method,it associates the restriction that imposed on the model with the number of variables that included in the model,that is when the number of variables is increased,it will impose more tightening prior on the model,thus we will avoid the over-parametrization phenomenon.In the empirical research of this thesis,the two methods have achieved good effects in forecasting China's economic growth and inflation.The second kind of model is Mixed-Frequency Data model.In the formulation of macroeconomic policy,it not only has a correct understanding of the future economic trends,but also has a better understanding of the current economic situation.However,owing to the publication lags of statistics,we lack the essential data to comprehend the current economic situation timely.The Mixed-Frequency Data model can be applied to nowcast the economic growth and analysis macroeconomic situation by using high frequency data.This model,more specifically,the Mixed-Frequency Dynamic Factor Model(MF-DFM),can use more indicators and tackle ragged edge data problem,so it has stronger timeliness than other models.Hence this thesis applies the MF-DFM to nowcast China's economic growth.We found that the data released within the quarter can significantly improve the prediction accuracy of GDP growth rate,besides,the variables which have non synchronized publication lags and in different groups contribute differently to improving the prediction accuracy.The third kind of model is Factor-Augmented Auto-Regression Vector model(FAVAR)with time-varying parameters and time variant dimensions.In consideration of financial market plays a more and more important role in the economy,it is necessary to analyze the impact of financial shock on the real economy under the new normal.We apply it to extract Financial Condition Index(FCI),then this index is used to forecast and analysis macroeconomic situation.More specifically,we apply FAVAR model with time-varying coefficients and time variant dimensions to extract the FCI in this thesis.The time variant dimensions refer to the number of variables that used to extract the financial condition index,which varies with time.The time-varying coefficients refer to the factor loading matrix of FAVAR model which is also time-varying.Taking into account of the rapid development of China's financial market and the impact of the financial crisis on intermediary variables in the financial market,which have an effect on the real economy,may make change,and the role of different indicators may also make changes.So,the model used in this thesis is time-varying coefficients and time variant dimensions,and the data can detennine whether the intermediary variables of the financial market can enter the model or not.The empirical research finds that,around the financial crisis,the number of variables,which used to extract FCI,changes significantly,and the probability of inclusion to the final FCI for each variable also changed greatly.Accurate economic forecasts only provide an accurate evidence for making the right economic policy,but how to make the correct economic policy and market strategy also need to have a correct understanding of the relationship between economic variables.However,in the new normal,there are declining in potential growth rate,structural adjustment and driving force conversion and other new features,so the relationship between economic variables is different from the past.If we are still in the light of past experience,and do not take into account the changes in the relationship between economic variables,this will lead to the poor effect of economic policy.Therefore,in order to provide a more accurate evidence for making the correct economic policy,this thesis also uses the new econometric model to study the relationship between economic variables in the new normal.In this thesis,we use the time-varying coefficients FAVAR model to extract macroeconomic common factors from a large panel of macroeconomic indicators,and use the sign restriction to identify the aggregate demand shock,aggregate supply shock and monetary policy shock.Then we study the changes in the effects of macroeconomic shocks on China's economic growth and inflation in new normal.In addition,this thesis also uses a time-varying parameters model to measure the change of China's inflation persistence.Taking into account of the impact of the financial crisis,the relationship between economic variables may exist short-term fluctuations,at the same time,because of changes in the economic structure of our country,the relationship between economic variables may appear permanent changes.In this thesis,we apply the time varying parameters model,and the parameters evolve as multivariate random walks.This form was firstly introduced by Cogley and Sargent(2005).Owing to the coefficients follow random walk process,this model not only can capture the temporary change of parameters,but also can capture the permanent change of parameters.When we apply VAR or FAVAR model to study the macroeconomic issues,we must face the identification problem of macro shocks.However,in previous studies,most scholars have adopted short-term zero restrictions identification scheme(the Cholesky decomposition is most typical method)and long-term zero restrictions identification scheme.But these two kinds of restrictions identification scheme have some disadvantages.For example,according to the certain problem in the research,short-term restrictions identification scheme will set some specific variables have no simultaneous impact on other variables.The relationship between variables in the real economy may influence each other,if we assume that there is only one-way effect between the variables,it is not consistent with the objective reality obviously.It is because of the shortcomings of those identification schemes,that many empirical research conclusions inconsistent with the economic theory,such as prize puzzle,slope puzzle and so on.In this thesis,we adopt a new identification scheme which based on the two classical identification schemes,to identify macroeconomic shocks,that is sign restriction identification scheme.This scheme,proposed by Uhlig(2005)and Canova(2002)in the study of the macroeconomic shocks,don't impose zero-equality restriction,which is a rather strict prior,just impose an inequality restriction,which is a weak prior.And it can reduce the artificial influence factors when we identify structural shocks,so the data plays a more important role.This study finds that since entering the new normal,a positive aggregate demand shock has less impact on promoting economic growth in short term,but has a bigger crowding-out effect in the mid-term,on the other side,a positive aggregate supply shock has bigger impact on promoting economic growth.These results imply that this empirical research provides a quantitative evidence for the ongoing supply-side reform of China.In the new normal,the impacts of monetary policy shocks and demand shocks on the price are reduced,but in 5th periods,the impact of supply shocks on the price is strengthening.