Decoding Trade War Signals: Dynamic Hierarchical Factor Modelling of Financial Market Responses to Policy Uncertainty

Trade policy uncertainty has emerged as a critical force influencing global financial markets, particularly during escalating trade tensions between major economies. This thesis introduces an application of Dynamic Hierarchical Factor Models (DHFM) to examine the forecasting value of uncertainty measures for US financial markets, developing a targeted approach to measuring trade-related market risk through hierarchical uncertainty decomposition. Using monthly data from 1992 to 2019, this research leverages a comprehensive dataset combining 446 macroeconomic indicators from the FRED-MD database and specialized uncertainty measures from multiple sources: Economic Policy Uncertainty indices from Baker, Bloom, and Davis (2016); Geopolitical Risk measures from Caldara and Iacoviello (2020); bilateral trade statistics from the U.S. Census Bureau; and global supply chain pressure indices from the New York Federal Reserve. The analysis evaluates the predictive performance of Trade Policy Uncertainty (TPU), Economic Policy Uncertainty (EPU), Geopolitical Risk (GPR), and a novel Trade Economy-Wide (TEW) factor across seventeen model specifications for S&P 500 returns and VIX volatility. Results reveal systematic comparative advantages across market segments. Trade Policy Uncertainty emerges as the premier equity market predictor, achieving 11.50% forecasting improvement with statistical significance (p=0.0706). In contrast, the DHFM-extracted TEW factor demonstrates great effectiveness for volatility forecasting, achieving 23.68% improvement (p=0.0008), validating factor-based approaches over traditional media-based indices. The findings provide nuanced evidence on market efficiency: equity markets exhibit semi-strong form efficiency with selective forecasting opportunities, while volatility markets show systematic inefficiency. The differential effectiveness establishes uncertainty measures as complementary tools requiring multi-dimensional approaches for comprehensive financial market analysis, contributing to both econometric literature on factor models and practical applications in financial forecasting, risk management, and policy analysis.