Time-Varying Long Memory Property in the Cryptocurrency Markets

• 발행기관: 한국자료분석학회
• 저자명: Sang Hoon Kang
• 간행물 정보: 『Journal of The Korean Data Analysis Society (JKDAS)』Vol.23 No.1, 75~85쪽, 전체 11쪽
• 주제분류: 자연과학 > 통계학
• 파일형태: PDF
• 발행일자: 2021.02.28

영문 초록

This paper investigates the long memory property of four cryptocurrencies (Bitcoin, Dash, Ethereum, and Litecoin) using the Rescaled Range Hurst analysis. The presence of long memory test for the validity of efficient market hypothesis in the cryptocurrency markets. First, we use traditional long memory tests (Hurst-Mandelbrot R/S, GSP and GPH) to investigate the long memory property in the returns and volatilities of cryptocurrency markets. We find that the volatility shows strong long memory property. Second, we employs the rolling sample approach and calculate time-varying long memory propertty in the returns and volatilities of cryptocurrency markets. Emprical results show that both the volatility and returns of cryptocurrency markets possess the time-varying long memory property. The average Hurst exponents are well above 0.5, indicating the presence of long memory. The long memory property of volatility is stronger than that of returns. The time-varying Hurst exponent values for BTC are significant higher than those of other cryptocurrencies (DASH, ETH, and LTC). This finding indicates that BTC is less efficient than other cryptocurrency markets. Therefore, the presence of long memory is important to predict future cryptocurrency prices, for asset allocation, and for portfolio assessment.

목차

1. Introduction

2. Empirical methodology

3. Data

4. Empirical results

5. Conclusions

References

키워드

Cryptocurrency Time-Varying Hurst Exponent Market Efficiency Rolling Sample Approach