Econometrics and Machine Learning Approach on Correlation in Stock Return between US Firms and Chinese Suppliers

Abstract

This study investigates machine learning on influential factors on the correlation of the stock return movements between US firms and their Chinese suppliers, given that the overall correlation has already been proven in the previous study. The present investigation was conducted by performing econometrical analysis and applying a variety of regression models to US firms' stock returns (independent variable) and their Chinese suppliers' stock returns (dependent variable) with various factors involved. From the results, it was evident that industry differences are a factor that caused variations in the degree of stock return correlations. Furthermore, firm size and stock trading volume yielded positive impacts on the significance of the correlation. Subsequently, predicated on said findings, a prediction model was generated using the Random Forest machine learning approach. Using US customer firms' monthly stock return data, as well as the three aforementioned factors, the monthly stock return value of their Chinese supplier firms can be predicted.