Machine Learning for Economic Forecasting - An Application to China’s GDP Growth

Aug 2, 2026

Upload by Levi

Explores how machine learning can forecast China’s macroeconomic variables, focusing on quarterly real GDP growth. Multiple machine learning models are compared against traditional econometric models and expert forecasts, with error analysis across different periods. Results show machine learning typically achieves lower average forecast errors, especially under economic stability. At certain inflection points, expert forecasts may sometimes be more accurate, reflecting broader macro understanding. Interpretable methods are further used to identify key attributive variables behind model performance differences.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages40

Reading Time1 h 41 min

AccessFree to download

Introduction

GDP as a Core Indicator

Motivation and Research Approach

Paper Organization

Literature Review

Data and Data Processing

Forecasting Models and Evaluation

Model Comparison Across Periods

Interpretability Analysis

Robustness Check

Conclusion

arXiv :2407 .03595v1 [ econ .GN] 4 Jul 2024 Machine Learning for Economic Forecasting: An Application to China’s GDP Growth Yanqing Yang * 1,2 , Xingcheng Xu†1 , Jinfeng Ge‡1 , and Yan Xu §1 1 Shanghai Artificial Intelligence Laboratory 2Fudan University July 4, 2024 Abstract This paper aims to explore the application of machine learning in forecasting Chinese macroeconomic variables. Specifically, it employs various machine learning models to predict the quarterly real GDP growth of China, and analyzes the factors contributing to the performance differences among these models. Our findings indicate that the average forecast errors of machine learning models are generally lower than those of traditional econometric models or expert forecasts, particularly in periods of economic stability. However, during certain inflection points, although machine learning models still outperform traditional econometric models, expert forecasts may exhibit greater accuracy in some instances due to experts’ more comprehensive understanding of the macroeconomic environment and real-time economic variables. In addition to macroeconomic forecasting, this paper employs interpretable machine learning methods to identify the key attributive variables from different machine learning models, aiming to enhance the understanding and evaluation of their contributions to macroeconomic fluctuations. Keywords: Macroeconomic Forecasting; GDP Growth; Machine Learning; Interpretable Analysis. JEL Codes: C22, C24 *[yanqingyang@fudan.edu.cn](yanqingyang@fudan.edu.cn)[ ](yanqingyang@fudan.edu.cn)†[xingcheng.xu18@gmail.com](xingcheng.xu18@gmail.com)[ ](xingcheng.xu18@gmail.com)‡[gejinfeng@pjlab.org.cn](gejinfeng@pjlab.org.cn) §[xuyan@pjlab.org.cn](xuyan@pjlab.org.cn) 1 Introduction As China’s economy enters the “new normal”, changes in growth momentum and global political, economic, and financial environments have increased the difficulty of macroeconomic forecasting based on structural modeling. Exogenous shocks, such as epidemics and conflicts, have also contributed to a decline in forecasting accuracy. Concurrently, advancements in big data and artificial intelligence algorithms have introduced new tools and methods for macroeconomic forecasting and policy regulation. In this paper, we employ cutting-edge machine learning algorithms to forecast China’s macroeconomy and examine the interpretability of these forecasting methods. This paper focuses on forecasting China’s GDP. GDP is the core indicator of national economic accounting and the primary basis for assessing macroeconomic performance and formulating economic policies. In recent years, both international and domestic macroeconomic forecasting have advanced significantly. The continuous development of econometric models and improved data availability have facilitated research on forecasting China’s GDP growth (Tong, 2017 ; Chen et al., 2018 ; Zhang et al., 2018 ; Fei and Liu, 2019 ; Liang et al., 2021) . Meanwhile, the rapid development of artificial intelligence algorithms and the emergence of high-frequency data have positioned machine learning as a crucial emerging tool in economic forecasting (Belloni et al., 2014) . International research organizations and academics are increasingly applying machine learning models in macroeconomic forecasting. Existing international research indicates that machine learning models generally outperform traditional econometric models in predictive accuracy (Bajari et al., 2015 ; Richardson et al., 2021) . The application of machine learning models in macroeconomic forecasting in China remains atan early stage. To explore the use of machine learning models for forecasting China’s macroeconomy and to assess the effectiveness of different forecasting methods, this paper employs various machine learning models to forecast China’s GDP growth rate. These models include machine learning models, combined models of machine learning and econometric me

FAQs

What forecasting target does the paper focus on?

The paper focuses on forecasting China’s quarterly real GDP growth as a core macroeconomic variable for policy and performance assessment.

How do machine learning models compare with traditional econometric models and expert forecasts?

Machine learning models generally produce lower average forecast errors than traditional econometric models or expert forecasts, especially during economic stability, while experts may outperform at some inflection points.

How does the paper improve understanding of model performance?

It uses interpretable machine learning methods to identify key variables contributing to forecasting results from global and local perspectives, helping explain differences in model performance.

Aug 2, 2026

Upload by Levi