Analyzing Structural Breaks in China's  Macroeconomy with High Dimensional Factor Model

Since the reform and opening-up,China has experienced rapid economic growth and become the largest economic entity in Asia and the second in the world. Gradually,it has grown into an important engine driving global economic growth. China's macroeconomy has also received more attention from the academic community and become a key research object. Studying China's macroeconomy plays a crucial role in optimizing the government's macro-control policies,addressing structural issues in the economic system,and maintaining healthy development of China's market economy. Today's world is in the midst of great changes that have not been seen in a century. Although peace and development remain the theme of the times,the risks and uncertainties of world economic operations have significantly increased. Against the backdrop of a downturn in the world economy,volatile international situation,and domestic reform pains,China's macroeconomic data are often subject to structural changes,wherein different economic sectors present strong interlinkages and co-movements. Understanding the phenomenon of structural breaks in China's macroeconomy plays an indispensable role in enabling the government to better perceive the risks and uncertainties of the world economy's operation,grasp China's macroeconomic structure and laws of development,and evaluate the effectiveness of regulatory policies. Thus,the structural change analysis of China's macroeconomy has become an urgent need for accurately predicting economic development trends,improving the macroeconomic control system,and promoting the overall macroeconomic stability and high-quality development.

In recent years,the analysis of the structural breaks in China's macroeconomic data has received increasing attention. However,there are two shortcomings in the existing literatures,which make it difficult to capture structural changes in the macroeconomy. First,the current studies mainly use the sequential testing procedure to detect the existence and locations of the structural changes,but the sequential estimation procedure would involve the multiple testing problems. As a result,it is more likely for the test to reject the null of no structural changes,and then the number of breaks will be overestimated,thereby affecting the subsequent analysis of structural changes. Therefore,to avoid the impact of sequential testing procedure on the structural breaks detection,some researchers reformulate the identification of multiple structural breaks as a variable selection problem in the high-dimensional regression with group-sparse coefficients. Then,they can apply the shrinkage estimation methods to estimate the numbers and locations of structural changes simultaneously. This type of method has the advantages of fast calculation and high accuracy for identifying structural changes. Second,the existing literatures generally ignore the inner relationships among various macroeconomic variables,only analyzing the structural changes in a certain field of national economy with a single macroeconomic variable. However,there exist close internal connections among these macroeconomic variables actually. Therefore,depicting and utilizing their coordinated changes is crucial for the government to analyze China's overall macroeconomic operating situation and regulate the macroeconomic system. In recent years,more and more researchers have tried to construct potential factors that can capture macroeconomic dynamics to analyze China's macroeconomy. However,they usually assume that the factor loadings do not change over time,ignoring the impact of technological innovation,policy reform and other uncertain factors on macroeconomic data. Consequently,the constructed factor models may not accurately capture the common factors behind the China's macroeconomic data. To capture structural changes in factor models,Baltagi et al. (2021) and Ma and Tu (2023a) recently proposed estimation methods based on the least squares and Group Fused Lasso,respectively. The theoretical contributions of Ma and Tu (2023a) include the consistency and limiting distribution of the break fraction estimators and consistent break date estimators,rather than just the consistency of the break fraction estimators (Baltagi et al.,2021). In practice,the procedure of Ma and Tu (2023a) is practically easy-to-implement with standard statistical packages,overcoming the drawbacks of the existing methods that they often involve multiple tuning parameters and are computationally demanding in dealing with multiple unknown breaks. Therefore,the method of Ma and Tu (2023a) is both theoretically accurate and practically appealing,and has the potential for analyzing structural breaks in China's macroeconomy.

In summary,China's macroeconomic data typically exhibit characteristics of structural breaks,and there exists a factor model structure,where macroeconomic variables in different sectors have strong co-movement features. However,few scholars have applied factor models with multiple structural breaks,together with an efficient and accurate estimation method,to study the structural break phenomenon in China's macroeconomy. This paper aims to fully utilize the factor characteristics in China's macroeconomic data and accurately capture the structural breaks in the macrosystem,providing a comprehensive analysis of China's macroeconomy. In particular,this paper proposes the factor model with multiple unknown structural breaks to model 23 China's macroeconomic time series,and further utilizes the estimation method of Ma and Tu (2023a) to identify the unknown numbers and locations of the change points,in contrast to Baltagi et al. (2021). The results show that via the estimation method of Ma and Tu (2023a),China's macroeconomic data have undergone six structural breaks from 1990 to 2022,and the estimated break dates are closely related to the important historic events,such as,Deng Xiaoping's South Tour Speeches and the 14th National Congress of the Communist Party of China,the SARS  in 2003,the  COVID-19  in 2020,etc. The method of Baltagi et al. (2021) detects four structural breaks,however it fails to detect the structural breaks in 2008 and 2020. This shows that the method of Ma and Tu (2023a) benefits from its more accurate change point estimation and fewer tuning parameter choices in practice,so that it can identify more structural breaks in the macroeconomic system and better capture the dynamic development of the macroeconomy. In addition,thanks to our factor model considering synergistic relationships in macroeconomic data,as well as a more accurate method for estimating change points,this paper can simultaneously detect the structural breaks in China's macroeconomy which were only separately discovered for each series in the existing literatures.