Key points are not available for this paper at this time.
Research background: The research employs the Cross-Sectional Absolute Deviation of returns (CSAD) model, augmented with modifications by Chiang and Zheng (2010) to address asymmetric investor behavior, facilitating the detection of herding behavior. Additionally, the study leverages Quantile Regression (QR), demonstrated by Barnes and Hughes (2002) to effectively capture extreme values in financial data with fat tails or skewed distributions. This approach is particularly relevant in the context of the volatile cryptocurrency market, allowing for the analysis of outliers and the assessment of the magnitude of return impacts using T-stat and Quantile Process Estimates. Purpose of the article: This study primarily centers its empirical analysis on identifying market-wide herding behavior (Henker et al., 2006) within the cryptocurrency market, spanning from January 1, 2016, to February 1, 2019, juxtaposed with the period from January 1, 2019, to January 7, 2022. The selected time frames were chosen to evaluate potential shifts in herding dynamics within this market, particularly during its phases of rapid expansion and subsequent stagnation. Methods: The Cross-Sectional Absolute Deviation (CSAD) methodology, as proposed by Chiang and Zheng (2010), was employed for herding detection, alongside the incorporation of dummy variables to discern the market conditions under which herding occurs. Herding behavior manifests when dispersion diminishes, or its increase is less than proportionate to market returns, indicating an inverse correlation between market returns and dispersion in the presence of herding. Additionally, CSAD estimation was conducted utilizing quantile regression to encompass a broader range of quantiles, facilitating the identification of herding tendencies across various return magnitudes. To delve further into investor behavior, Bitcoin was utilized as an illustrative example, elucidating investor reactions to market bubbles through the application of the Hodrick-Prescott (HP) Filter. Findings & value added: The findings reveal instances of herding behavior during downward market movements and at higher return levels preceding 2019. However, post-2019, herding is observed during upward market movements and at medium to higher return levels. This study presents compelling evidence of herding phenomena coinciding with the bursting of bubbles, particularly concerning Bitcoin. The findings provide a deeper understanding of how herding manifests differently across distinct market conditions and timeframes, offering actionable insights for investors and policymakers navigating the volatile cryptocurrency landscape. Additionally, by highlighting the correlation between herding behavior and market bubbles, particularly in the context of Bitcoin, this study contributes to the broader discourse on cryptocurrency market dynamics.
Gherghina et al. (Fri,) studied this question.