TL;DR
This study investigates investor herding behavior in BRICS nations from 2007-2021, revealing that herding is a short-lived phenomenon, primarily driven by high volatility and not significantly by crises, except for China during the Global Financial Crisis. Interestingly, the U.S. "fear index" (VIX) has a limited, short-lived impact, primarily during COVID-19, with China remaining largely unaffected.
What Problem Does This Paper Address?
Herding behavior, where investors mimic others rather than relying on private information, is a well-documented phenomenon in financial markets. However, much of the existing research focuses on developed markets or single emerging economies. This paper addresses the gap by providing a comprehensive analysis of herding behavior across all BRICS countries (Brazil, Russia, India, China, and South Africa) simultaneously. It aims to understand the commonalities and differences in their investor behavior, particularly how factors like market volatility, economic crises, investor sentiment, and the influence of the U.S. market (specifically the VIX) contribute to or mitigate herding. The study also seeks to clarify the complex, often debated, relationship between volatility and herding, and whether sentiment truly drives herding.
Methods
The study employs the Chang, Cheng, and Khorana (CCK) model to measure herding behavior, using the cross-sectional absolute deviation of returns (CSAD) as the primary indicator. To analyze the determinants of herding, the researchers utilize several advanced econometric techniques:
- Principal Component Analysis (PCA): Used to construct a sentiment index from various economic indicators, aiming to capture a more nuanced view of investor mood.
- Multiple Volatility Measures: The study goes beyond single volatility measures by employing six different models (GARCH, EWMA, Parkinson, Garman-Klass, Moving Average, and Capitalization-Weighted) to assess volatility's impact, acknowledging that different measures might yield different results.
- Chow Tests: Used to identify structural breaks in the data, allowing for the analysis of distinct periods, including pre-crisis, during-crisis, and post-crisis phases.
- Granger Causality Tests: Applied to determine the direction of influence between variables, such as sentiment and herding, or the VIX and herding, to establish causal relationships rather than mere correlations.
- Quartile Regression: Used to examine how different levels of volatility (low, medium, high) affect herding behavior.
Datasets
The study utilizes daily, weekly, and monthly stock market data for the BRICS countries (Brazil, Russia, India, China, and South Africa) from July 2, 2007, to September 30, 2021. The data sources are DataStream and Bloomberg. The analysis is segmented into three main crisis periods:
- Global Financial Crisis: September 15, 2008 – March 31, 2009
- European Debt Crisis: April 1, 2010 – January 31, 2012
- COVID-19 Crisis: December 31, 2019 – September 30, 2021
Key Insights and Ideas
- Herding is a Short-Lived Phenomenon: The study strongly supports the idea that herding is a temporary behavior, most evident when using high-frequency (daily) data and disappearing with lower frequencies (weekly/monthly). This suggests that herding is a fleeting reaction rather than a persistent market characteristic.
- Volatility is a Key Driver, Not Crises: High market volatility, rather than the occurrence of crises themselves, is identified as a significant amplifier of herding behavior, particularly in China. Crises, in general, did not consistently trigger herding across all BRICS nations.
- China's Unique Position: China stands out as an exception, showing herding only during the Global Financial Crisis and remaining largely immune to the influence of the U.S. VIX, making it a potential "counterbalancing force" in international portfolios.
- Volatility Measure Matters: The impact of volatility on herding is not uniform across different measurement methods. This highlights the importance of considering various volatility metrics for a comprehensive understanding.
- Limited Role of Sentiment and U.S. Fear Index: While sentiment (derived from PCA) shows some limited Granger causality with herding, its role is not as significant as initially hypothesized and is often unrelated to crises. Similarly, the U.S. VIX has a short-lived and limited effect, primarily during the COVID-19 pandemic, and notably, not in China.
Key Results
- Herding is Present in China During GFC: Herding was observed only in China during the Global Financial Crisis, with a significant negative coefficient for Rmt² in daily data.
- High Volatility Amplifies Herding: In periods of high volatility, herding was found to be present in Brazil, China, and India, indicated by significant negative coefficients of Rmt². Russia and South Africa showed no significant herding, regardless of volatility levels.
- VIX Impact During COVID-19: The U.S. VIX showed a significant effect on herding (CSAD) in all BRICS countries except China, but only during the COVID-19 crisis period.
- Two-Way Causality Between VIX and CSAD: Granger causality tests revealed a two-way relationship between the VIX and CSAD in Brazil and Russia, independent of crises, challenging the notion that VIX solely drives herding.
Limitations and Future Work
- Volatility Measure Homogeneity Not Supported: The study found that different volatility measures have varying effects on herding, meaning the assumption of volatility measure homogeneity is not supported.
- Limited Evidence for Sentiment Causality: While PCA was used to construct a sentiment index, the evidence for sentiment Granger-causing herding was limited and period/market-specific.
- US Influence Varies: The U.S. market's influence (spillover effects) on BRICS was found to be short-lived and generally unrelated to crises, though it was more pronounced in certain periods and countries.
- Future Research Directions: The authors suggest exploring the impact of political/military crises, investigating how current events like the Russia-Ukraine conflict affect BRICS markets, and examining the role of high-frequency data for a deeper understanding of spillovers and causalities. They also recommend further research into the nuances of investor behavior in China and the potential for it to act as a counterbalancing force in international portfolios.
Summary updated