Predicting Stock Market Performance Using Individual Spending Data

This paper discusses the effectiveness of individual spending data in predicting the equity risk premium and greenium in the Australian stock market. We use adaptive LASSO, random forest and XGBoost models and the results show that some spending features, such as the spending ratio on Monday and the proportion of entertainment consumption frequency, have significant predictive ability for the two types of premium, respectively. The results remain stable even after the inclusion of traditional economic and financial variables and sentiment indicators. individual spending at a shows timeliness and foresight in capturing changes in market behavior. This study provides new evidence for understanding the relationship between consumption behavior and asset pricing, and expands the data dimension for the study of behavioral finance and sustainable finance.<p></p>