Uncertainty and bubbles in cryptocurrencies: Evidence from newly developed uncertainty indices

Abstract

In this paper, we examine whether newly developed crypto price and policy uncertainty indices based on news coverage (Lucey et al., 2022) are associated with the emergence of bubbles in cryptocurrencies. Using probit regressions, we show that these indices have a higher explanatory power than factors previously considered in the literature. Furthermore, using a random forest model, we show that these classifiers are associated with the largest information gain (reduction in the Gini impurity measure) of the model. While the COVID-19 pandemic has exacerbated the occurrence of bubbles, these crypto uncertainty indices remain the best predictors of bubbles both before and during the pandemic. These results are robust to alternative definitions of a bubble, variations in the time horizon, and the inclusion of various regressors known to be related to the price movements in crypto assets.