Cross-Market Signals: Economic Spillovers Across Markets

Abstract

This thesis explores the growing interconnectedness of global markets, with a focus on the explanatory power U.S. and China macroeconomic factors have on each other's stock market returns. Using the JKP Global Factors Dataset with 13 themes and 153 factors, we model this relationship using linear regressions, sparse additive models, and kernel regressions. The analyses were conducted using the Wilshire 5000 and Shanghai Composite returns. Across all models, U.S. economic data consistently improved the predictions of Chinese market returns, but Chinese economic data rarely added improved the predictions of U.S. market returns. Linear regressions revealed decent R^2 values using domestic data, with 0.70 R^2 0.80 for the U.S. market and 0.55 R^2 0.60 for the Chinese market. Sparse additive models and kernel regressions achieved higher R^2 values in the data. They were more likely to overfit to the data, with the partial dependence plots sometimes not following economic intuition. Future directions include accounting for interaction effects, attempting rolling-window methods, and exploring sector-level analyses to obtain more granular insights into market spillovers.