ABSTRACT
This study tests for the weak-form market efficiency of 15 cryptocurrency prices. The conventional unit root tests and stationary test results reveal that most cryptocurrency markets are efficient markets. However, the non-linear quantile unit root test proposed by Li and Park (2018) rejects the unit root null hypothesis over the whole quantile level. To derive more informative ideas, we split the whole quantile interval to several sub-intervals and find asymmetric behaviour of the market efficiency across the lower and upper sub-intervals in several cryptocurrency markets. Moreover, non-linear quantile unit root tests for Chainlink, Bitcoin Cash, Binance Coin, EOS, Tron, and Stellar indicate that markets for these cryptocurrencies are efficient at the upper sub-intervals.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/13504851.2022.2096851.
Notes
1 Cryptocurrency is traded 24 hours a day. The closed price denotes the latest data in the range based on coordinated universal time (CUT). The time series of cryptocurrency are available at https://www.coinmarketcap.com.
2 QKS test results are quite similar to those of the QCM test, and the difference between QKS and QCM test results is that the total number of rejections in the QCM test is lower than those in the QKS. In this study, we cannot confirm what factors constitute the difference between QKS and QCM test results; however, according to Li and Park (Citation2018), QKS tests reject unit roots more often than QCM tests.
3 To gain better insight, we also consider the linear quantile test of Koenker and Xiao (Citation2004). However, to save space, we only report non-linear quantile test results. The results for linear quantile test are available in the online version of the supplementary materials.