Recently, the co-authored paper CPI Forecasting and Method Comparison Based on Macroeconomic Big Data by Lecturer Fan Xinyue from the Department of Econometrics, School of Economics, Shanghai University of Finance and Economics, has been formally published in Journal of Management Science in China, a domestic Class A authoritative economics journal.
Abstract
The advent of the big data era brings unprecedented opportunities and challenges to CPI forecasting. Fully tapping information from high-dimensional data and developing interpretable machine learning forecasting models are of great significance for both theoretical advancement and practical application. To this end, this study constructs a monthly macroeconomic database for China with 239 variables grouped into 9 categories, and compares the CPI forecasting performance of 13 models including traditional time-series models, regularized regressions, factor models and ensemble algorithms using large datasets. Furthermore, machine-learning derivative algorithms are developed based on the control-variable framework to interpret the estimation results and analyze underlying mechanisms. The empirical results show that random forest and XGBoost achieve satisfactory forecasting performance and exhibit prominent advantages especially in medium- and long-term forecasting. Further analysis attributes their strengths to nonlinear specifications and non-sparse variable handling: the former captures real-world variable relationships more faithfully, while the latter enables full exploitation of big-data information. Meanwhile, these two models identify autoregressive terms, prices, employment and other reasonably important variable categories for CPI prediction.
About the Author
Fan Xinyue holds a PhD in Quantitative Economics and is a Shanghai Chenguang Scholar. She is currently a Lecturer at the School of Economics, Shanghai University of Finance and Economics. Her research interests include macroeconometrics, textual analysis, macroeconomic monitoring and forecasting, as well as big-data modeling methods and applications. In recent years, she has published multiple papers in prominent journals such as International Journal of Forecasting, Journal of Forecasting, Economic Research Journal, The Journal of World Economy and Journal of Management Science in China. She presides over projects under the Shanghai Chenguang Program, the Shanghai Program for AI-Driven Research Paradigm Reform and Academic Advancement, and the Open Project of the Key Laboratory of Mathematical Economics of the Ministry of Education. As a core team member, she participates in major research projects including key base projects of the Ministry of Education for humanities and social sciences, major projects of the National Social Science Fund of China, General Projects of the National Natural Science Foundation of China, and key national statistical science research projects.
