Is bitcoin a near stock? Linear and non-linear causal evidence from a price–volume relationship

Abstract

Purpose: The purpose of this paper is to examine the price–volume relationship in the bitcoin market to validate near-stock properties of bitcoin. Design/methodology/approach: Daily data of bitcoin returns, returns volatility and trading volume (TV) are utilized for the period August 17, 2010–April 16, 2017. Linear and non-linear causality tests are employed to examine price–volume relationship in the bitcoin market. Findings: The linear causality analysis indicates that the bitcoin TV cannot be used to predict return; however, the reverse causality is significant. In contrast, the non-linear causality analysis shows that there are non-linear feedbacks between the bitcoin TV and returns. The bitcoin TV, which represents new information, leads to price changes, and large positive price changes lead to increased trading activity. Similarly, in recent periods (post-break period), the results of the non-linear causality test show a unidirectional causality from TV to the volatility of returns. Research limitations/implications: This study uses the average index value of major bitcoin exchanges. But further research on this relationship using data from different bitcoin exchanges may provide further insights into the price–volume relationship of bitcoin and its near-stock properties. Practical implications: These findings from the non-linear causality analysis, therefore, suggest that investors cannot simply base their decisions on the linear dynamics of the bitcoin market. This is because new information in terms of the TV is neither linearly related to the price nor it is a one-to-one kind of relationship as most investors commonly understand it to be. Rather, investors’ decisions should be based on non-linear models, in general, and the best-fitting non-linear model, in particular. Originality/value: The study examines bitcoin’s near-stock properties in a price–volume relationship framework with the help of both linear and non-linear causality tests, which to the best of the authors’ knowledge remains unexplored.