Study On Measuring Economic Uncertainty And Financial Uncertainty In China And Its Application

China’s economy has shifted from a stage of high-speed development to a stage of highquality development,and the domestic and international environment China’s economy and society are facing has also undergone significant changes in this new stage,with economic fluctuations caused by uncertain events becoming more frequent as a distinctive feature of the new stage.On the one hand,China is in the transformation of the development mode,optimizing the economic structure,and changing the growth momentum,cyclical,structural,and institutional factors give economic operations downward pressure.On the other hand,the global economic recovery is still constrained by contraction in demand,trade protectionism,economic and trade frictions between China and the United States,the gathering of asset bubbles,as well as high global inflation and other factors that deepen external uncertainty.The combination of these internal and external uncertainties is highly likely to cause economic fluctuations beyond expectations,which undoubtedly becomes a major challenge for China’s high-quality development stage.Based on this background,this dissertation takes uncertainty measurement as the core issue,tries to improve the existing domestic measures of economic uncertainty and financial uncertainty,and gives the monthly indicators of economic uncertainty and financial uncertainty in China in the past 20 years,based on which the similarities and differences of their stage characteristics and their impact on China’s economic volatility and their mechanisms of action are discussed.The main work of this dissertation and its marginal contributions are summarized as follows.First,the study on uncertainty measurement methods.In this dissertation,by comparing the single indicator volatility method,the keyword frequency method,the dispersion degree or prediction deviation method and the conditional volatility method of high-dimensional data,the conditional volatility method of high-dimensional data is selected as the method of measuring uncertainty in this dissertation.This method is in accordance with the unpredictable characteristics of uncertainty,uses high-dimensional objective data,and combines with the DSGE model to measure uncertainty,which has the characteristics of objectivity and high accuracy.Nevertheless,the method based on the constant coefficient SV model cannot reflect the structural and cyclical characteristics of China’s economic development process and neglects the mean effect of uncertainty,so this dissertation uses the TVP-SVM model to extend the original SV model.On this basis,the parameter estimation problem of TVP-SVM is introduced in detail with the estimation method proposed by Chan(2017),and the model selection methods of DIC and marginal likelihood function values of the TVP-SVM model family are introduced in detail with the model comparison method proposed by Chan&Eisenstat(2018).Second,the uncertainty indicator set is studied.This dissertation summarizes the existing uncertainty indicator systems commonly used at home and abroad,finds the following shortcomings,and improves them while retaining their ideas.(1)There are mixed indicators in the set of economic indicators and the set of financial market indicators.Briefly,the set of economic indicators contains unnecessary financial indicators,resulting in economic uncertainty reflecting the fluctuations of the real economy while overlaying some financial market fluctuations,and the root causes of the fluctuations cannot be clarified when analyzing their phase characteristics,while the set of financial market indicators ignores these indicators.The set of economic indicators in this dissertation is based on the set of indicators commonly used in domestic and international research,excluding indicators that are not in the economic monitoring system but represent the financial market.(2)Most of the financial market indicators are mainly those reflecting the stock market.As a branch of the financial market,the stock market cannot reflect the overall operation of the financial market,so this dissertation expands the coverage of the existing financial market indicators based on the definition of the financial market.The new set of financial market indicators covers the money market,capital market,and foreign exchange market.On this basis,we also exclude industry and stock portfolio return data to balance the proportion of stock market indicators,thus reflecting the overall operation of the financial market in a more objective manner.Third,the study of data pre-processing.In the research of this dissertation,we find that the existing uncertainty measurement literature commonly uses the software default X-13 or Tremo-Seats method for seasonal adjustment,while the NBS uses the NBS-SA method.This software,although not publicly available,takes into account the moving holiday effect,fixed holiday effect,monthly trading day effect,and leap year effect that are distinctive to our economic data.In this dissertation,we introduce the NBS-SA method in detail and use the searching method instead of the subjective setting method in the treatment of the moving holiday effect to finally construct the seasonally adjusted variables.And then take the industrial added value as an example to demonstrate the seasonal adjustment process,and combine it with statistical indicators to find that the seasonally adjusted data quality is significantly better than commonly uesd methods.In addition,due to the use of highdimensional data,data dimension reduction is another important pre-processing step required in this dissertation.In this dissertation,we apply the robust sparse principal component method(ROSPCA)to improve the existing traditional principal component method(PCA)to overcome the problems of its principal components being sensitive to outliers,poor data information compression ability,and vague economic significance,and elaborate the process of data dimension reduction based on the uncertainty indicator set constructed,and we found that ROSPCA can identify data outliers while principal components can compress information significantly better than PCA.Fourth,the measurement and’ characterization of economic uncertainty and financial uncertainty in China.Combined with the previous study about the selection of indicators,data processing,and uncertainty measurement methods,economic uncertainty and financial uncertainty in China are constructed.The research is summarized as follows:(1)The optimal model is given for all indicators and it is found that most of the price index indicators do not support the benchmark model;and there are a few cases where the indicators support the existence of random fluctuation mean effect ’term.This indicates that the model selection method adopted in this dissertation is correct to discriminate the optimal models for different indicators in a targeted manner.In actual modeling,a single model should not be used to model all indicators indiscriminately,which not only affects the accuracy of individual uncertainty estimation but also affects the accuracy of the final economic uncertainty and financial uncertainty.(2)The estimation process of individual uncertainty is demonstrated by taking industrial added value as an example,and the analysis of the parameter estimation results reveals that the optimal time-varying parameter model captures the structural changes in the data more effectively than the benchmark model(constant coefficient SV model),more fully exploits the effective information contained in the historical data,reduces the estimation bias of the unpredictable part of the model,and improves the accuracy of individual uncertainty.(3)When synthesizing economic uncertainty and financial uncertainty,we find that the-mean method contains more abnormal individual uncertainty information than the median method,resulting in greater fluctuations in economic uncertainty and financial uncertainty and less,robustness.Therefore,it is more effective to choose the median method instead of the mean method.(4)This dissertation analyzes the stage characteristics of economic uncertainty and financial uncertainty,and it is found that the uncertainty constructed in this dissertation can effectively reflect the impact of major events,such as SARS in 2003,the global financial crisis in 2008,the stock market crash in 2015 and the COVID-19 in early 2020.By comparing and analyzing the financial uncertainty constructed in this dissertation with the original financial uncertainty,we find that the new financial uncertainty better reflects the overall uncertainty level of the financial market,while the original financial uncertainty reflects the volatility of the stock market,and the two differ in time sequence and diverge in trend during the COVID-19.This indicates that the high uncertainty in the financial market does not necessarily originate from the high uncertainty in the stock market,and justifies the reasonable financial uncertainty construction in this paper.Fifth,the impact of economic uncertainty and financial uncertainty on economic fluctuations is analyzed.Based on the idea of "symbiosis and co-prosperity" of economy and finance,this chapter examines the mechanisms of economic uncertainty and financial uncertainty on output and price fluctuations in the framework of their coexistence.Applying the TVAR model and the generalized impulse response function to examine the impact of economic and financial uncertainty on output and prices under different volatility regimes,we find that:(1)there are spillover effects between economic uncertainty and financial uncertainty and the spillover effect of financial uncertainty on economic uncertainty is stronger and the lag is weaker.(2)Compared with financial uncertainty,the shock effect of economic uncertainty on output and price fluctuations is more prominent,and the shock effect is further amplified in the high volatility regime of output and price.(3)Under the high growth rate of economic uncertainty scenario,the elevated financial uncertainty plays an amplifying role in output and price fluctuations,while the high growth rate of financial uncertainty does not affect the shock effect of economic uncertainty on output and price fluctuations.The main marginal contributions of this dissertation are summarized as follows.(1)Improving existing uncertainty measurement methods.This dissertation extends the existing economic uncertainty and financial uncertainty measurement methods from three aspects:indicator set selection,data preprocessing,and uncertainty measurement model.In terms of indicator set selection,there is no intersection between the new economic indicator set and the financial market indicator set to easily identify the source of volatility,and at the same time,the financial market indicator set covers the financial market as a whole and no longer focuses excessively on the stock market.In terms of data processing,this dissertation applies the NBS-SA method in use by the National Bureau of Statistics and combines the search method rather than the subjective setting method to determine the moving holiday variables to overcome the subjective setting bias;at the level of data dimensionality reduction,this dissertation applies the ROSPCA method instead of the principal component method to enhance the ability of the principal components to compress the information and to provide the non-noise information.In terms of model construction,this dissertation extends the SV model commonly used in the existing literature to the TVP-SVM model and combines the Bayesian model selection method to select the optimal model for different indicators,which improves the accuracy of individual uncertainty measurement and ensures the accuracy of the economic uncertainty and financial uncertainty measurement results in this dissertation.(2)Further expansion of the research questions.Existing literature pays less attention to the impact of economic uncertainty and financial uncertainty linkage on economic fluctuations,this dissertation puts them in a unified framework,and investigates the impacts of economic uncertainty and financial uncertainty on economic fluctuations and their mechanisms with the TVAR model.The research questions not only take into account the structural changes in economic development,but also consider the issue of the differences in the mechanisms of economic uncertainty and financial uncertainty under different economic fluctuation regimes.