Bitcoin's Energy Consumption, Price Volatility, and Environmental Pollution: Contagion and Causality Dynamics Under Heteroskedasticity and Nonlinearity

Abstract

Bitcoin's (BT) energy demand due to the proof of work mining algorithm and transaction methods has been criticized in recent literature because of BT's global energy consumption (EC) being equal to or even greater than that of some industrialized economies. This study explores the contagion and causality dynamics under nonlinearity and heteroskedasticity with Markovian-type regime switches among BT price volatility, EC from BT, and its effects on the global carbon dioxide (CO2) emissions with a daily sample of January 2, 2012-December 24, 2023. In the empirical methodology, following the preliminary tests that indicated nonlinearity and heteroskedasticity, this study employed novel Markov-switching-based nonlinear volatility copula and causality models to capture the marginal distributions of BT, its EC and CO2 emissions, and their joint distributions for contagion and tail dependence and determine nonlinear and asymmetric causality relations in distinct regimes of high and low volatility governed by Markov chains. The empirical results indicate significant and positive copula parameters with high magnitudes demonstrating asymmetric contagion and tail dependence occurring at the extreme levels of BT price volatility in the upper and lower tails during both regimes. Novel Markov-switching-based copula causality tests designated unidirectional causality from BT to CO2, as well as from BT's EC to CO2 emissions during each regime coupled with existent bidirectional causality confirming cyclical feedback effects between BT prices and its EC in both regimes. For robustness and comparison, single-regime copula models were employed, and their findings confirmed the results concerning significant and positive tail dependence and contagion. However, the single-regime approach led to a set of inconsistencies in causality results owing to omitting nonlinearity while capturing heteroskedasticity and confirming the nonrejection of causality between BT, EC, and CO2 emissions. Important policy recommendations include green alternatives to cryptocurrency-mining algorithms.