Bitcoin, Uncertainty, and Internet Searches

This thesis examines the predictive power of the Macroeconomic Uncertainty Index (MUI) with a horizon of one month and the volume of uncertainty-related online searches measured by Economic Uncertainty Related Queries (EURQ) on the Bitcoin returns. The sample includes 118 observations with monthly frequency and ranges from September 2010 to June 2020. In addition to the conventional methods, due to the presence of outliers and the departure from Gaussian assumptions; the quantile analysis framework was used to study the persistency of the shocks, the long-run relationships, and Granger causality among the variables. For this purpose, the quantile unit root test of Galvao (2009), the quantile cointegration test of Xiao (2009), and the Granger causality test in quantiles (Troster, 2016) were applied. The empirical findings highlight several vital points that lead to important policy implications. First, the variables display asymmetric behavior in response to the shocks along different conditional quantiles. Second, the Bitcoin-EURQ indicates different adjustment mechanisms towards long-run equilibrium in different quantiles. Third, the long-run and causal relationships between the series might be significantly different throughout the conditional distributions of the variables. Fourth, of particular interest, the findings provide evidence for the significant predictive power of the fluctuations in MUI and EURQ on Bitcoin returns, mostly in low and high quantiles. Implications of these findings for empirical researchers and Bitcoin investors are discussed in the conclusion section in detail. Keywords: Cryptocurrency, Bitcoin, Macroeconomic uncertainty, Internet search volumes, Economic uncertainty-related queries, Granger causality in quantiles.