Bitcoin in the Macroeconomic Landscape: Price Determinants, Regime Dynamics, and Volatility Modelling

Abstract

Bitcoin has grown from a cryptographic experiment into a trillion-dollar asset class with tightening links to traditional financial markets. This thesis studies the forces behind Bitcoin's price formation and volatility through four empirical chapters, using daily and monthly data from 2014 to 2024. A vector autoregression (VAR) with Granger-causality tests shows that investor attention, proxied by Google Trends, predicts Bitcoin returns more strongly than conventional safe-haven variables such as gold or the VIX. Five policy dummies for major regulatory events show no statistically significant effects on any variable, consistent with Bitcoin prices being driven primarily by internal market dynamics rather than discrete policy interventions. A mixed-frequency VAR pairs daily Bitcoin prices with monthly consumer-price indices for the United States, South Korea, and Japan. Bitcoin returns Granger-cause Japanese CPI but not Korean or US inflation, a result that suggests a jurisdiction-specific transmission channel concentrated in Japan's retail-heavy crypto market. US inflation in turn Granger-causes Bitcoin returns, consistent with a macro-drives-crypto channel, and cross-country CPI spillovers from the United States to East Asia are the dominant dynamic in the system. A two-state Markov-switching VAR identifies regime-dependent volatility dynamics in which conditional variance of Bitcoin returns in the high-volatility state is nearly three times that in the tranquil state. Interest rates do not significantly predict Bitcoin returns in either regime, so the model's main contribution is capturing variance-regime shifts rather than return predictability. EGARCH and DCC-GARCH specifications show that retail attention, proxied by Google Trends, is the strongest driver of conditional variance, while short-term interest rates amplify and long-term rates dampen Bitcoin volatility. Correlations with the MSCI World Index spike during global sell-offs, which weakens the case for Bitcoin as a portfolio diversifier. Bitcoin behaves as a sentiment-driven speculative asset in calm periods but becomes increasingly correlated with mainstream markets under liquidity stress. These findings inform how monetary authorities assess cryptocurrency spillovers, how portfolio managers budget for regime-conditional correlation risk, and how regulators design disclosure requirements for digital-asset exposures.

Date of Award	25 Mar 2026
Original language	English
Awarding Institution	University of Bath
Supervisor	Bruce Morley (Supervisor) & Asgerdur Petursdottir (Supervisor)