BANKING UNIVERSITY OF HO CHI MINH CITY NGUYEN THIEN NHAN EXAMINING HERDING BEHAVIOR IN VIETNAMESE STOCK MARKET GRADUATION THESIS MAJOR FINANCE – BANKING CODE 52340201 Ho Chi Minh City, April 2022 THE.
INTRODUCTION
Research background
The Vietnamese stock market has experienced 21 years from its foundation in
Since its inception in 1998, with the Ho Chi Minh City Stock Exchange (HSX) and the Hanoi Stock Exchange (HNX), the Vietnamese stock market has evolved from two listed companies in 2000 to a dynamic market marked by key milestones By September 30, 2021, it boasted 2,133 listed stocks and a market capitalization of over 8.3 quadrillion VND, equivalent to about 133.83% of GDP, according to the State Securities Commission of Vietnam.
In the recent two years, this emerging market has entered a new period of development with many impressive mileposts The value of the VN-Index has increased significantly, especially since March 2020; the Vietnamese stock market has broken many records in terms of liquidity as well as the number of new accounts At the end of 2021, there were 52 listed companies having capitalization reaching more than 1 billion USD, VN-Index increased from about 1120 points to 1498.28 points for gaining 378 points in a year, and the average daily trading volume was more than 26.560 trillion Dong in 2021 The first outbreak of the Covid-19 pandemic occurring in March 2020 has affected the market index adversely VN-Index dropped from 1000 to 650 in March 2020, equivalent to losing 35% points Afterwards, the market has recovered from the bottom and experienced am impressive growth until 2021 at 1498.28 points During the period from March
2020 to December 2021, investors also experienced many strong downward, unpredictable trading sessions and were difficult to explain Then, at each milestone where the VN-Index was about to break the new record in value, the market experienced strong declines from 100 points to 200 points So, is the market really efficient and does any mispricing or bubbles exists in our equity market? After a
2 financial crisises or a correction of the market, herding behavior is often considered to be a reason causing mispricing
Understanding what an asset bubble is helps explain the basis of herding phenomena in markets Hashimoto (2020) defines an asset bubble as a sustained deviation of an asset’s market value from its fundamental value, with prices often exceeding fundamentals prior to financial crises In standard finance theory, an efficient market is one where all publicly available information is quickly and accurately reflected in stock prices, enabling investors to use market prices as signals to guide investment capital allocation (Fama).
The Efficient Market Hypothesis (EMH) posits that asset prices fully reflect all available information and that investors act rationally, making unbiased forecasts of future prices If markets are truly efficient and participants rational, financial crises and asset bubbles should be rare or unexplained by mispricing alone Yet history shows episodes of economic downturns in Vietnam and around the world that have raised questions about EMH’s predictive accuracy Critics contend that irrational behavior, information asymmetry, and systemic shocks can drive prices away from fundamentals, creating bubbles and crises despite the claim of market efficiency This tension keeps the debate alive about when EMH holds and under what conditions markets may deviate from rational pricing.
Behavioral finance explains unusual market fluctuations by focusing on investor behavior and the psychological factors that shape stock prices, challenging the rationality assumed by efficient-market theory It combines psychology with conventional finance to understand how markets operate and how investors make decisions, arguing that people are not fully rational and markets are not perfectly efficient Statman (2014) identifies four main foundation blocks of standard finance and proposes four alternative foundation blocks for behavioral finance, underscoring a fundamental shift in how we view market dynamics Standard finance rests on the ideas that rational agents and efficient markets underlie price formation.
3) people should design portfolios by the rules of mean-variance portfolio theory
Behavioral Finance offers four foundational blocks: 1) people are normal, with cognitive biases that influence decision-making; 2) the market is not fully efficient; 3) investors design portfolios using the principles of Behavioral Portfolio Theory; and 4) expected investment returns are described by Behavioral Asset Pricing Theory, which frames returns as a function of risk.
The development of behavioral finance traces back to the 18th century when Adam Smith addressed the psychology of economic behavior in his landmark works Theory of Moral Sentiments (1759) and The Wealth of Nations (1776) These writings introduced ideas about how emotions, biases, and social influences shape financial decisions, laying an early foundation for the modern field of behavioral finance, which blends psychology and economics to explain investor behavior and market phenomena.
Early work dating back to 1759 emphasized the role of sentiments in decision-making The 1960s and 1970s marked a new period in the development of behavioral finance Tversky and Kahneman, often regarded as the fathers of behavioral finance, articulated three key heuristics—availability, representativeness, and anchoring and adjustment—that shape judgments under uncertainty Today, behavioral finance has deepened our understanding of the financial markets, explaining phenomena that standard financial theory cannot fully explain and improving insights into investor behavior and market dynamics.
Herding behavior occurs when traders make decisions by imitating others' actions (Spyrou, 2013) According to Bikhchandani et al (1992), the non-transparency of the equity market is a key driver of herding A substantial body of research investigates herding across countries worldwide, and some studies find that herding is more pronounced in emerging markets (Voronkova and Bolh, 2015).
Motivated by the issues outlined earlier and the author's curiosity about herding behavior in the Vietnamese equity market, this study conducts an in-depth examination of herd behavior in the Vietnamese stock market The goal is to help investors gain a clearer understanding of how herding affects market dynamics in Vietnam’s financial landscape and to support the improvement of their investment decision-making processes.
Research gap
There have been numerous studies examining the presence of herding in both developed and emerging markets worldwide Tan et al (2008), Chiang et al (2010), Fu and Lin (2010), and Ju (2019) employ cross-sectional standard approaches to assess herding behavior across different market contexts.
4 deviation (hereafter CSSD) and cross-sectional absolute deviation (here after CSAD) model to examine herding behavior in Chinese stock markets Choi & Skiba
In 2015, the study employed the model proposed by Sias (2004) and Choi & Sias (2009) to investigate herding across 41 countries In 2020, Kumar et al applied the cross-sectional absolute deviation model to detect herding in Japan, Thailand, Taiwan, China, Indonesia, the United States, the United Kingdom, India, and Malaysia.
In Vietnam, Tran and Truong (2011) examine herding behavior by applying a GARCH model to confirm the estimated results after first using ordinary least squares (OLS) to detect the phenomenon Vo and Phan (2017) assess herding in the Vietnamese stock market over the 2005–2015 period by employing the CSAD and CSSD models, with findings indicating that herding exists in the Vietnamese equity market.
This study applies the Cross-Sectional Absolute Deviation (CSAD) model to test herding behavior in the Vietnamese stock market over 2016–2021, providing a more up-to-date dataset than previous work With liquidity surging and a wave of new investors entering Vietnam’s equity market in recent years, the research offers a current overview of stock market herding Building on the CSAD framework proposed by Chang et al (2000), the analysis examines whether price dispersion deviates in a way that signals herding and uses quantile regression to capture heterogeneity across return distributions The findings shed light on the presence and dynamics of herding under different market conditions, contributing to a clearer understanding of investor behavior in Vietnam’s increasingly active stock market.
Research question
This thesis is aimed to solve some issues:
Does herding behavior exist in the Vietnamese stock market and how it affects the stock market?
Observing herding behavior in various market conditions, such as up/down market and high/low trading volume market.
Research objectives
This study investigates the extent of herding behavior in the Vietnamese stock market during 2016–2021 by examining the relationship between equity return dispersion, measured by CSAD (cross-sectional absolute deviation), and the overall market return.
Besides that, this research aims to investigate herding behavior in different conditions market.
Research scope and methodology
Using an indirect approach, this study investigates the presence of herding behavior in the Vietnamese stock market by examining the relationship between market return and stock return dispersion The empirical framework relies on the Chang et al (2000) model, which is later refined by Chiang et al (2010) and widely used in research on herd behavior.
This study analyzes 1,500 daily observations from 270 companies listed on the Ho Chi Minh City Stock Exchange (HSX) over 2016–2021 To test the hypotheses, the authors employ traditional ordinary least squares (OLS) regression and quantile regression analysis, following Chiang et al This combination provides insights into average effects with OLS and distributional effects with quantile regression, enabling a robust examination of the relationships under study.
Research structure
The first chapter depicts the background of the research, researcher’s motivation, research objectives, general methodology and scope, research gap and research structure.
LITERATURE REVIEW
Theoretical literature review
Herding behavior can be defined in several ways Spyrou (2013) describes it as a process where economic agents imitate others or base their decisions on the actions of others Banerjee (1992) defines it as everyone doing what everyone else is doing, even when private information would suggest a different course Caparrlli et al (2004) argue that the fear of making a mistake drives investors to follow the crowd, with the belief that shared errors save face Nofsinger & Sias (1999) characterize herding as a pattern of investors trading in the same direction over a period Clements et al (2016) contend that herding in financial markets reflects similarity in decision making.
Bikhchandani and Sharma (2001) separate herding into two types: spurious herding and intentional herding Spurious herding occurs when a group of investors faces the same information sets, leading to similar investment decisions and potentially making the market more efficient as prices reflect the shared information For example, Investor A receives positive information about Apple's stock and buys Apple; subsequently, Investor B receives a positive signal and also buys, creating correlated demand across periods In this case, the phenomenon is considered unintentional herding because both investors are acting on the same information rather than imitating others.
8 investors just simply follow the same signal and do not actually follow each other’s trades
Intentional herding occurs when investors imitate the decisions of others and ignore their own information, which can generate systematic risk, excess volatility, and push a stock's market value away from its intrinsic value Two influential models used to interpret this behavior are the informational cascade frameworks introduced by Banerjee (1992) and refined by Avery and Zemsky (1998), with foundational contributions by Bikhchandani et al.
In 1992, rational traders copy the investment activity of other market participants, believing those investors may possess information the rest do not A second explanation for herding, proposed by Scharfstein and Stein (1990), is reputation-based: institutions and professional investors face reputational risk if they act against the crowd These two forms of herding—information-driven and reputation-driven—can produce opposite outcomes, and distinguishing between them remains challenging.
Several explanations have been offered for this phenomenon Bikchandani et al (2001) identify informational externalities as the first reason for herding: when later investors ignore their own information and rely on the investment decisions of earlier investors, an information cascade arises (Bikhchandani et al., 1992; Banerjee, 1992) They also note that when information is not common knowledge among investors, informationally inefficient herding may occur and can lead to price bubbles and mispricing Banerjee (1992) presents a decision model showing that decision-makers tend to mimic the behavior of the initial group since that group may hold important information.
Reputation-based factors can trigger herding in investment management, as managers may imitate the decisions of their peers to safeguard or enhance their standing The theories proposed by Scharfstein and Stein (1990), Trueman (1994), and Graham (1999) formalize how reputational concerns of fund managers or analysts can drive herding behavior.
Reputational herding arises when doubts about a manager's ability lead investment professionals to imitate the decisions of others rather than act on their private information In effect, reputation-based herding occurs as managers ignore their own private, substantive signals and mimic peers to avoid underperforming Rajan (2006) suggests that this herding provides a form of insurance that a manager's investment will not lag behind peers As a result, this protective dynamic can reinforce herd behavior, especially when other investment professionals face similar concerns about performance, creating a self-reinforcing cycle that can influence investment choices and market outcomes.
Thirdly, compensation-based is one of the reasons can lead to herding Trueman
A 2004 study shows that analysts tend to favor earnings forecasts that are closer to prior earnings expectations, a bias that can boost their compensation by shaping clients’ assessments of forecasting ability This convergence with previous results is linked to enhanced perceived credibility among investors, which can influence demand for the analyst’s services Bikhchandani et al discuss this dynamic and its implications for compensation structures in the analysis of sell-side forecasts.
A 2001 study shows that when an investment manager's compensation is evaluated by comparing her performance to that of similar professionals, incentives become distorted and she may end up with an inefficient portfolio Such peer benchmarking can also foster herd behavior among managers, leading to synchronized investment choices and reduced portfolio diversification.
Herding in financial markets does not necessarily undermine efficiency; when it is unintentional, it can actually improve stock market efficiency by causing investors to react in unison to the same fundamental information, thereby speeding the adjustment of prices to new fundamentals (Lakonishok et al., 1992) However, herding that is not grounded in fundamentals can impair efficiency by fostering mispricing and market instability Intentional or deliberate herding can destabilize markets and contribute to price bubbles and crashes (Scharfstein and Stein, 1990; Bikchandani et al., 2001).
Barberis and Schleifer (2003) and Scharfstein and Stein (1990) argue that price movements should reverse after herding drives stock prices away from fundamentals There is some empirical evidence supporting this view, but findings are not uniform Brown et al (2010) and Puckett and Yan (2008) provide evidence of herding-related return reversals using weekly data, while, on the other side, other studies report mixed or inconclusive results, suggesting that the relationship between investor herding and return reversals may depend on data frequency and methodology.
Lakonishok et al (1992), Wermers (1999), and Sias (2004) find no evidence of return reversals following herds.
Herding in different conditions
Apart from detecting herding in various countries worldwide, many studies investigate herding in different aspects, such as the size of firms, trading volume and in up and down markets
Lakonishok, Shleifer, and Vishny (1992) examine herding in a sample of US equity funds and find that intentional herding is more prevalent in small-cap stocks because these firms have less public information, meaning small stocks have lower quantity and quality of public information than large-cap stocks This information gap increases uncertainty and makes managers more prone to imitate peers, aligning with Scharfstein's view on how information frictions shape herd-like trading in financial markets.
Stein (1990) interprets investment managers as tending to sell small-cap stocks that others are selling to avoid embarrassment, while continuing to hold large-cap stocks like IBM when others sell Choi and Sias (2009) and Venezia et al (2011) also document a greater degree of herding in small stocks, reinforcing the view that small-cap trading exhibits stronger herding behavior than large-cap stocks.
Trading volume is commonly used to gauge the relationship between market herding and liquidity, with intentional herding theories predicting stronger herding in markets or days characterized by low liquidity A large body of literature links information quality, market liquidity, and information asymmetry to herding behavior For instance, Vo and Phan (2017) show that herding occurs on both high- and low-volume days in the Vietnamese stock market, but is stronger on low-volume days Diamond and Verrecchia (1991) argue that information asymmetry is higher in less liquid markets, while Suominen (2001) finds that higher trading volume indicates better information quality.
A substantial body of empirical research has examined herding in both rising and falling markets Chiang et al (2010) analyze herding across 18 countries spanning Asia, Europe, the Americas, and Australia Their findings indicate that, with the exception of the United States and Latin American markets, herding is present in both upward and downward market movements.
11 markets, although herding asymmetry is more prevalent in Asian markets during rising markets
Tan et al (2008) analyze herding behavior in dual-listed Chinese A-share and B-share stocks and find evidence of herding in both the Shanghai and Shenzhen A-share markets, present in both rising and falling market conditions Herding among A-share investors in the Shanghai market is more pronounced during rising markets, periods of high trading volume, and high volatility, while no asymmetry is observed in the B-share market The authors explain this by noting that Shanghai-listed stocks are predominantly larger firms, many of which were formerly state-owned, which leads investors to be more optimistic and confident of government support in upswings.
Vo and Phan (2017) document herding behavior in the Vietnamese stock market across several market conditions, showing that herding is stronger in down markets than in up markets due to a flight-to-safety mentality during adverse times.
Empirical literature review
Overall, many empirical studies have been carried out since the 1990s concentrating on detecting the existence of herding behavior in different market conditions, perspectives and how it impacts the financial market Although some of them found no evidence of herd in the financial market, they have contributed a better understanding of this phenomenon
Christie & Huang (1995) collect data from 1925 to 1988 and utilized cross- sectional standard deviation (CSSD) to investigate the herd in the US stock market Nevertheless, there was no evidence of herd found Afterthat, Chang et al (2000) present a more powerful model named cross-sectional absolute deviation (CSAD) utilized to test the herd which was developed on the basis of the CSSD model The research detected herding behavior in multiple markets, such as the US, Hongkong, Japan, South Korea and Taiwan The result found evidence of herding in two
12 emerging markets - South Korea and Taiwan, partial evidence of herding in Japan and documented no evidence of herding in the US and Hong Kong
Chiang et al (2010) employ the CSSD model proposed by Chisties & Huang
Using a 1995 framework together with the cross-sectional absolute deviation (CSAD) model introduced by Chang et al (2000), the study examines herding behavior in Chinese stock markets The dataset covers stocks listed on the Shanghai and Shenzhen Stock Exchanges from January 1, 1996, to April 30, 2007 The findings show clear evidence of herding in both up and down markets within the Shanghai and Shenzhen A-share markets, while no herding signals are observed in B-share markets.
Caparrelli et al (2004) examine herd presence in the Italian stock market using a sample from September 1, 1988 to January 8, 2001, and find a non-linear relationship between dispersion and returns, supporting Christie and Huang’s conclusions that herding is present in extreme market conditions in Italy A subsequent study by Caporale et al (2008) tests for herding in the Athens Stock Market and indicates the existence of herd behavior for the years 1998-
2007 Applying the model suggested by Christie & Huang (1995) and Chang et al
In 2000, the authors subdivided the study period into semi-annual sub-periods They identify herding during the stock market crisis of 1999, and they observe that investor behavior has become more rational since 2002.
Choi and Skiba (2015) applied Sias’s model to detect herding behavior in international markets, examining 41 countries from Q4 1999 to Q1 2010 The study found statistically significant herding propensities across these target markets, particularly in countries with a substantial presence of institutional investors.
Hwang & Salmon (2001) propose a novel measure and test of herding based on the cross-sectional dispersion of asset factor sensitivities within a market They apply this approach to the US, UK, and South Korean stock markets, and their results reveal statistically significant evidence of herding.
”the market portfolio” during relatively quiet periods rather than when the market is under stress
Numerous studies investigate herd behavior in emerging markets For example, Hassan (2015) analyzes herding in the Pakistani stock market using the methodologies proposed by Christie and Huang (1995) and Chang et al (2000) The findings indicate an absence of herd behavior during the 2002–2007 period and show no support for the rational asset pricing model, suggesting that investor behavior was inefficient.
Tan et al (2008) study herding behavior in dual-listed Chinese A-share and B-share stocks by employing Chang et al.'s modified model, and they report evidence of herding in both the Shanghai and Shenzhen A-share markets under rising and falling market conditions Ju (2019) likewise detects herding in A- and B-share markets using the CSAD model, reporting a prevalent level of herding across both A- and B-share segments.
Bhaduri and Mahapatra (2013) examine herding in the Indian stock market by leveraging the symmetric properties of the cross-sectional return distribution instead of conventional test methodologies Their alternative approach reveals detectable herding in Indian equities over the sample period, with the effect appearing more pronounced during that time.
During the 2007 crash, the analysis shows that the rate of increase in security return dispersion was relatively lower on upmarket days than on downmarket days Fu and Lin (2010) analyze data from January 2004 to June 2009 to examine herding behavior and investors’ asymmetric reactions to good and bad news in China, and they report no overall evidence of herding in the China stock market; however, they find that low-turnover stocks are more prone to herd than high-turnover stocks, and that investors tend to herd in a downward market.
Kumar et al (2020) examine herding in commodity markets across major Asian economies, and their results reveal anti-herding behavior in India, Malaysia, and Taiwan, while Japan shows no evidence of herding.
Evidence of herding patterns—or anti-herding patterns—has been found in China and Indonesia In Singapore and Thailand, herding is detected in downward markets, while in the US and UK, herding does not appear to depend on the market’s development status Zhou et al (2013) propose a modified CSAD (cross-sectional absolute deviation) approach to study herding behavior and its asymmetry Their results show no overall herding in China's carbon market; regarding asymmetry, there is no herding in both up and down markets, nor in markets with high or low trading volumes or high or low volatility.
Vo and Phan (2017) study herd behavior in the Vietnamese stock market by applying Christie and Huang’s (1995) and Chang et al.’s (2000) models Using a dataset of 299 companies listed on the Ho Chi Minh City Stock Exchange (HOSE) from 2005 to 2015, the study finds persistent evidence of herding throughout the whole period and reveals an asymmetric effect that depends on market conditions and trading volume.
Tran and Truong (2011) examine the existence of herding in the Vietnamese stock market and whether it exhibits an asymmetric effect based on market direction Using daily price data for all securities listed on the Ho Chi Minh City Stock Trading Center from 2002 to 2007, they apply a GARCH(1,1) model to overcome limitations of ordinary least squares The results indicate the presence of herding in this emerging market, but there is no evidence of an asymmetric effect tied to the direction of market movements.
Table 2.1 A summary of relevant empirical research about herding behavior
Authors Period Studied Models Countrie s Main Findings
Daily and monthly data from 7/1962 to 12/1988
CH US The results did not show the presence of herd for both daily and monthly returns
Hong Kong, Japan, South Korea, and Taiwan
RESEARCH METHODOLOGY
Data collection and sample description
This study uses the daily closing prices of all stocks listed on the Ho Chi Minh Stock Exchange (HSX) to examine the presence of herding in the Vietnamese stock market The daily dataset is secondary data covering the period from 01/04/2016 to 31/12/2021 and was collected from the HSX website and cophieu68.com.
409 stocks listed on HSX and after collecting and filtering, the final sample comprises 270 companies equivalent to 270 stocks which provided 1500 daily observations over the period studied
Selected stocks must satisfy the going-concern requirement during 2016–2021 and be listed on the Ho Chi Minh Stock Exchange (HSX) In addition, the dataset is divided into four subsets to reveal insights into the herding phenomenon under different market conditions.
Table 3.1 Summary of data observations used in the study
In market analysis, an up market day is defined as a day when the market return is greater than zero, while a down market day is defined as a day when the market return is less than zero Additionally, a high trading volume day occurs when the day's trading volume exceeds the previous 30-day moving average.
On the contrary, “low trading volume days have trading volume less than the previous 30-day moving average.
Regression model for testing the hypotheses
Two widely used models for detecting herding behavior in financial markets are the cross-sectional standard deviation (CSSD) method developed by Christie and Huang in 1995 and the cross-sectional absolute deviation (CSAD) approach introduced by Chang and colleagues in 2000 The CSSD method analyzes how dispersion in individual stock returns across firms at each point in time signals abnormal clustering of price moves, while the CSAD approach regresses the absolute deviation of stock returns from the market return to capture non-linear patterns indicative of herding Together, these tools quantify the intensity of herding and help researchers study its drivers, variation across market conditions, and implications for market efficiency.
Christie & Huang (1995) employ the cross-sectional standard deviation (CSSD) method to detect herding behavior, which is defined as:
Return dispersion at time t can be represented by two indicator variables: one equals one when the market return at time t lies in the extreme lower tail of the distribution and zero otherwise, while the other equals one when the market return at time t lies in the extreme upper tail and zero otherwise However, the Christie & Huang model has drawbacks Munkh-Ulzii et al (2018) argue that defining extreme returns is arbitrary, so traders may disagree on what constitutes an extreme move and the return distribution characteristics may change over time Additionally, Chiang et al (2010) point out that herd behavior can be present across the entire return distribution and become more dominant during periods of market stress.
Christie and Huang’s model identifies herding only under extreme returns, so the study adopts an alternative to the Christie and Huang (1995) approach by applying the CSAD method proposed by Chang et al (2000) The CSAD method is developed based on the Christie & Huang (1995) framework, offering a return-based measure of herd behavior.
CSAD i , t = γ 0+ γ 1 | | + γ 2 Where CSAD i,t is the cross-sectional absolute deviation of i company at time t
Let R_t^abs denote the absolute value of the equally weighted average stock return for the dual-listed portfolio of N companies in period t In the model, γ1 is the coefficient on this absolute average return, while γ2 is the coefficient on the squared market return, also known as the herding coefficient, which captures the degree of coordinated or herd-like movement in prices Together, these terms describe how the magnitude of the average stock performance and the squared market performance influence the portfolio’s return dynamics.
The model included independent and dependent variables which were described the calculation as follows:
Dependent variable: the dependent variable in this research was return dispersion measured by cross-sectional absolute deviation, which was expressed as:
CSAD t = ∑ | | (3) where N is the number of firms in the portfolio, is the return of market portfolio at time t and is the return of stock i at time t
Independent variables: the study uses market return as the key independent variable Based on the VN-Index closing prices from the original data, the daily market return is calculated as the daily percentage change in the VN-Index closing price, typically expressed as Return_t = (P_t − P_{t−1}) / P_{t−1} (or 100 × [P_t / P_{t−1} − 1]), reflecting day-to-day movements in the Vietnam stock market and serving as the primary metric for the analysis.
= 100 ( - (4) where: P t is closed price of VN-Index at time t and P t-1 is closed price of VN-Index at time t-1
Chiang et al (2010) argue that higher market returns can widen stock return dispersion (CSAD) because each asset differs in sensitivity to market movements and investors make independent decisions under normal market conditions, producing a typically linear CSAD–market return pattern Consequently, a negative coefficient γ2 (γ2 < 0) would indicate the presence of herd behavior in the market.
24 exists in Vietnamese stock market, in contrast, if γ 2 is greater than 0 and significantly statistical, there is no presence of herd in Vietnamese stock market
3.2.2 Regression model for estimation the degree of herd in rising and falling market:
Beyond detecting herding in the stock market, the author also examines whether herding behavior varies with market conditions The research investigates whether there is a difference in the level of this phenomenon between up markets and down markets To assess this, the author uses models designed to test herding in both up and down market conditions.
Let R_m,t^up denote the market return at time t when the market is rising, and let R_m,t^down denote the market return at time t when the market is falling Equations (5) and (6) define these two returns as the baseline α plus terms that reflect the average absolute deviation of each stock’s return from the return of the equally-weighted market portfolio at time t, conditioned on rising or falling market returns The average absolute deviation measures how far each stock typically lies from the market portfolio’s return under the respective market condition The market is considered up when its return is greater than zero, and down when its return is less than zero.
3.2.3 Regression model for estimation the degree of herd in high and low volume trading market:
Research has documented varying levels of herding across different trading volumes Li et al (2009) find a positive correlation between average trading volumes and market returns realized by individual investors Building on this, the study investigates whether herding behavior differs between high- and low-trading-volume periods in the Vietnamese stock market The herding regression is estimated separately for high and low trading volumes using regression specifications designed to capture the tendency of investors to mimic others in different liquidity regimes.
Figure 8 defines market returns conditional on trading volume levels: R_t^H denotes market returns on days with high trading volume, and R_t^L denotes market returns on days with low trading volume, where high or low is determined by whether the day's volume exceeds or falls below its 30-day moving average The measure A_t^H (A_t^L) represents the average absolute value of each stock's return relative to the return of the equally-weighted market portfolio at time t, conditional on high-volume (low-volume) markets In practice, a day t is classified as having high trading volume if the volume exceeds the preceding 30-day moving average, and as having low trading volume if it is below that average.
EMPIRICAL RESULT
Descriptive statistics
Table 4.1 Summary statistics of cross-sectional absolute deviation and absolute market return
Var Obs Mean Max Min S.D
In high trading volume state
In low trading volume state
(Source: Synthesis by the author from Eviews 8)
Table 4.1 displays descriptive statistics for the absolute market return and the cross-sectional absolute deviation (CSAD) across the 2016–2021 period The table provides summary statistics for the entire sample and is subdivided into four smaller data sets, enabling comparison of these key measures—absolute market return and CSAD—across sub-samples.
1) Rising market 2) Declining market 3) High trading volume and 4) Low trading volume
Analyzing 1,500 observations from 2016–2021, the Vietnamese equity market shows an average daily return of about 0.07%, with extreme moves reaching a maximum of 4.9% and a minimum of −6.6% The sample’s overall fluctuation level over this period is 1.1% Return dispersion, measured by CSAD for the entire sample, ranges from 1.1% to 3.6%, with a volatility magnitude of 0.37%.
Across the Vietnamese equity market dataset, there are 866 observations when the market is up and 634 observations when it is down The greatest daily return in rising conditions is 4.9%, with a daily standard deviation of 0.6%, while the minimum daily return during declining sessions is -6.67% with a daily standard deviation of 0.98% The volatility magnitude is higher in declining markets than in rising markets, a pattern that aligns with prior studies on herding in the Vietnamese market, such as McQueen et al.
(1996) suggest that there is a delayed reaction to good news, whereas investors react more quickly to bad news Having the same conclusion, Bernartzi & Thaler (1995) state that market participants may fear the potential loss during a down market period of stress more than they enjoy the potential gain during an up market period of stress, and evaluate their positions relative to the market with more anxiety which is described as “myopic loss aversion”
In high trading volume market, the minimum value of the market return is -6.67% with a standard deviation of 1.2% In low trading volume market, the maximum and minimum value of the market return is about 3.9% and -4.9%, respectively, with a standard deviation of 0.9%
Testing for mean different of CSAD
Because there is a similarity of CSAD value among sub-sample which is about 0.018 The author decides to test for mean different of CASD among Overall data set and all the rest sub-sample T-test is conducted to test for this section and the results are shown in the below table
CASD Up Down High Low
P(T