Climate policy uncertainty and stock markets: Evidence from the United States : A thesis submitted in partial fulfilment of the requirements for the Degree of Doctor of Philosophy at Lincoln University

Abstract

Climate change is playing an increasingly pivotal role in financial markets, posing systemic risks to the financial sector. Concurrently, uncertainty surrounding climate policy presents significant challenges and has attracted growing academic attention. However, the existing literature has yet to comprehensively examine the relationships between climate policy uncertainty in the United States (U.S.) and its impact on financial markets. This study aims to bridge this research gap by systematically investigating the intricate relationships between the climate policy uncertainty and stock markets in the U.S. It identifies the best-fit copulas to represent the dependence structures between the climate policy uncertainty and stock markets and determines the most reliable machine learning model for forecasting U.S. stock markets based on climate policy uncertainty. To achieve this, the study uses the U.S. Climate Policy Uncertainty Index developed by Gavriilidis (2021), alongside U.S. stock market data covering major stock indexes and key industrial sectors, including transportation, mining, insurance, energy, waste and disposal services, and health care. The study uses Dynamic Conditional Correlation – Generalized Autoregressive Conditional Heteroscedasticity (DCC-GARCH) and copula models to investigate conditional correlations and dependence structures between climate policy uncertainty and stock markets, respectively. Various machine learning models are assessed to evaluate the predictive power of climate policy uncertainty in forecasting stock markets. The DCC-GARCH model findings highlight potential portfolio diversification opportunities in the transportation and insurance sectors because no information transmission is detected between these sectors and climate policy uncertainty using both symmetric and asymmetric models. Empirical results from dependence structure analysis suggest that U.S. stock markets are likely to rise alongside increases in climate policy uncertainty, because the Joe and Gumbel copulas are identified as the best-fit copulas for all examined pairs. Predictability analysis identifies Particle Swarm Optimization with eXtreme Gradient Boosting (PSO-XGBoost) as the most robust machine learning model for predicting U.S. stock markets based on climate policy uncertainty. The study also reveals that climate policy uncertainty exhibits the highest predictive power for the waste and disposal services sector and the lowest predictive power for the mining sector. The study’s findings provide critical insights into the complex relationships between U.S. climate policy uncertainty and stock markets, offering valuable implications for portfolio managers, policymakers and investors.