Effect of market sentiment on stock returns

Nội dung text: Effect of market sentiment on stock returns

1. EFFECT OF MARKET SENTIMENT ON STOCK RETURNS Lan Phuong To, PhD9., Cao Thi Thuy Trang - University of Economics and Business – Vietnam National University Abstract The stock market is affected by many factors. No single factor can affect the stock market. Instead, many factors come together to form an overall investment environment and impact the stock market. In the market, the psychology of investors always greatly affects the stability of the market, especially for the stock market, the impact of psychological factors is always very complicated, making the stock market always has a very high potential for instability, which is a key factor causing market panics, making the implementation of macroeconomic policies always face many difficulties. The results are contrary to the expected goals. For Vietnam, the impact of psychological factors on the market in recent years has been very complicated, causing bad consequences, threatening macroeconomic instability, and financial security is not guaranteed. This fact requires serious studies on the influence of investors' psychology on the stock market in Vietnam and investors must understand all the factors that can cause influence the stock market to make effective investment decisions. Keywords: Market sentiment, investor sentiment, behavioral finance, stock returns, Vietnam stock market INTRODUCTION The mainstay of traditional finance is the Efficient Market Hypothesis (EMH), which states that all information to a business is immediately reflected in stock prices, and all investors are smart and capable the same information and cannot pass the market. However, the assumptions of traditional finance have failed to explain the unusual, not "reasonable" phenomena that occur in reality in the market such as the January Effect, the day of the week effect, or the bubbles. Asset bubbles leads to stock market crash, etc Behavioral finance was created to fill the fundamental gaps of traditional finance with the assumption that financial asset prices are not always constant motivated by reasonable expectations of future profits, by the fact that people in general, and market participants in particular, are emotional beings, not just rational ones. Their investment decisions depend not only on qualitative factors according to the generalized model, but also on personal emotions or market movements. These factors have led to different decisions in the market causing stock prices to continuously rise and fall. In addition, the Capital Asset Pricing Model (CAPM) believes that the only factor affecting the expected return of a stock investment is the market risk premium and the sensitivity of that stock to 9Email: phuongtl@vnu.edu.vn 119
2. the general market movements is beta. The CAPM model has been criticized by scholars in the traditional and behavioral finance schools. In fact, psychological factors have an impact on investment returns, however, these psychological factors are difficult to quantify to research because psychology is a qualitative factor. So, the urgent question is, is there a way to quantify the psychological index? And how to fully assess the psychological factors affecting investment returns in the most comparable and general way? Do the effects of psychology have the same level of impact over a long period of time in the history of Vietnam's stock market, or do they differ through each development cycle? From the above urgent questions, the research focuses on quantifying and building a load factor representing the psychological factor by the method of principal component analysis (PCA) and then equity the impact of the above factor on the economic benefits return on investments through the capital asset pricing model CAPM. LITERATURE REVIEW Currently, there is a lot of research on behavioral finance, including psychological factors. The research mainly focuses on analyzing and evaluating the impact of psychological factors on the returns of stocks on the stock market. Specifically: Market psychology, also known as "investor psychology", is not always based on fundamentals, it is a branch of behavioral finance that uses results from models financial models and investor sentiment to measure and assess the impact of sentiment on price movements and expected returns of financial assets. ➢ Indicators measuring psychological factors in the stock market Huang et al (2013) propose to construct a new market sentiment index aligned to predict stock market aggregates, based on the six variables used by Baker and Wurgler (2006) and by using The Pool Least Square (PLS) method of Kelly and Pruitt (2013). The results of the research prove that investor sentiment has greater predictive power than macro variables for the stock market in general. In economic terms, market sentiment's ability to predict returns comes from investors' biased beliefs about future cash flows rather than discount rates. Therefore, the use of market sentiment is necessary in financial markets and the research has contributed in part to the position of behavioral finance. In the Taiwan stock market, the participants are mostly individual investors and only 5% of the participants are institutional investors (Huang and Yang, 2001) so it is really important to pay attention to the question of whether market sentiment affects stock returns. Lee's research (2018) answered this question by looking for indicators and measures that predict investor sentiment based on two approaches: One is an indicator of investor behavior measured by using proxy variables (such as short-term, long-term rates of return; average profit ratios, sales ratios, and price-earnings ratios) and other measured investor sentiment using proxy variables (investor sentiment index BW, consumer confidence index and market volatility index). The results show that the factors (BW sentiment index, consumer confidence index and VIX) are significantly related to price and have significant explanatory significance for average stock returns in the market. These factors are 120
3. important determinants in predicting stock returns in the Taiwan stock market. ➢ The impact of psychological factors (psychological index) on stock returns Obviously, the impact of investor psychology factors in different approaches have a certain influence on stock returns. Apergis and Rehman (2018) by collecting data on all S&P 500 companies according to daily statistics, from 1995-2015 and 1-year US Treasury bond yields in the same period. , and use the investor sentiment setting variables from Baker and Wurgler (2006) in the residuals from the CAPM model. The results from the empirical test show that investor sentiment has a significant influence on stock returns according to the traditional CAPM model. The implication of this finding is that seeing the role of investment psychology in classical financial theory can lead to an imperfect picture of asset valuations or of having a single factor influencing stock profits. Also derived from the psychological factors of investors, by quantitative method, Nguyen Ngoc Tu Van (2018) uses the weighted regression model EGLS to analyze the impact of investor psychology on the price index. stocks through panel datasets of 5 countries in Southeast Asia for the period 2009-2017 including: (1) VN-Index of Vietnam, (2) JCI of Indonesia, (3) FTSE KLCI of Malaysia, (4) PSEI of the Philippines and SET of Thailand. In which, the consumer confidence factor represents investor sentiment, an indicator designed to measure the level of optimism (Sum, 2012) that consumers feel about the overall state of their products economy and their personal financial situation. The research results show that the consumer confidence index is a representative factor for investor sentiment and has an impact and function to predict the stock market price index in the future. Researching and building a set of market sentiment indexes to measure the reaction of the stock market, Phan Thi Nha Truc (2019) uses the component analysis (PCA) method based on the variables that make up the psychology of the stock market. investor BW (2006); Next, the least squares method is used to determine the correlation between the market sentiment index variable and the component variables to perform hypothesis testing. Factors affecting investor sentiment include: the number of transactions as measured by revenue, dividend premium, transactions of closed-end fund certificates, the number of shares issued for the first time, the number of shares new. The research shows the relationship between investor sentiment and information related to new issuance, dividend information, investment fund information and trading volume, market liquidity. The author has introduced the BW market sentiment index calculation model into Vietnam, however, the author only stops at building the model without assessing the impact of the market sentiment index on stock price fluctuations shares in the stock market. Psychology in finance and behavioral finance - cognitive influence on investor behavior ➢ Overconfidence, sensation seeking and overtrading The poor investment performance of investors is partly due to poor choices in buying and selling this stock and when [to buy and sell] (Hvidkjaer, 2008). Another reason mentioned is overtrading. Instead of holding onto the shares they own, investors tend to sell and buy too often, which incurs unnecessary costs in the form of commissions, fees, and taxes. 121
4. ➢ Underreaction, Overreaction and Price Change Rate Index When a company announces its pre-tax profit last quarter, the stock price rises if the news exceeds expectations, or the stock price falls if pre-tax profits fall. While this is consistent with the efficient market hypothesis, actual observations report the opposite, that stock prices in the case of an increase in corporate earnings before tax tend to continue to rise and in case [pre-tax profit] continues to decline in the following weeks. This anomaly is known as the post-earnings drop in stock prices (Bernald & Thomas, 1989). Barberis, Shleifer & Vishny (1998) argue that, because investors are conservatively biased (Slovic & Lichtenstein, 1971), they have considered new information relative to their previous beliefs and thus initial response was low when the pre-tax profit announcement was received. In the weeks following the announcement, the effects of the news were gradually recognized among investors buying (or selling) the stock, causing the stock price to "drift" slowly toward its underlying present value. share. However, if a pre-tax earnings announcement is in the same direction as previous announcements, investors may over-interpret this as a trend and thus over-react to the news. De Bondt & Thaler, 1985). Overreactions are more commonly observed within the first 15 minutes following the announcement of pre-tax profits, although the results of sharp price changes tend to return partially within hours (Patell & Wolfson, 1984). ➢ Diversification is not enough To avoid unnecessary risks, investors are advised not to "put all their eggs in one basket". The reason why the risk distribution recommendation should be related to the variability of returns: it is more likely that companies in the same industry (or country) perform worse at the same time than companies in different industries (countries) (Markowits, 1952). However, investors often do not diversify enough. Some investors may mistakenly believe that any multi-asset portfolio will be diversified (Goetxmann & Kumar, 2008) and thus use the primitive diversification heuristic that includes many stocks (or stock funds) in their portfolio without considering a variant partnership. This was shown in a research by Hedesstrom, Svedsater and Garling (2009), in which the majority of risk-averse test participants chose a portfolio of specialized funds over a single fund good variety which is actually the least risky option. ➢ Emotional impact Emotions arising from the evaluation of specific investment options can be used as decision signals. “Experience effect” predicts having a good (bad) feeling about something that causes people to over-associate with positive (or negative) attributes and thus choose stocks (or not stocks at all). Furthermore, only emotional forecasting can influence investment behavior. Summers and Duxbury (2012) suggest that investors' reluctance to sell stocks that have declined in value, as shown in this decision, is due to their desire to avoid the feeling of regret associated with creating the loss. These are consistent with the “ostrich effect,” which refers to the observation that investors during their time in the stock market seem to have a tendency to “bury their heads in the sand.” ➢ Social influence 122
5. In many areas of social life, people often follow others when making decisions. Social influence is either normative or informative (Deutsch & Gerard, 1955). In the case, people conform to external social pressures or internal social norms; in other cases, the motive is to obtain and use the information of others. When a group of investors follow each other to buy (or sell) the same stock, this is called a herd. Information leadership, or an “information layer,” arises when investors ignore their “private” information and instead imitate the choices of other investors because they believe that the Others will have better information. Psychological factors (shown by the market sentiment index) Market Sentiment - This term can be understood in a broad sense and it is used in very different contexts by academic researchers, financial analysts and even the media (Barberis et al., 1998; Kent et al., 1998; Welch and Qiu, 2004; Brown and Cliff, 2004; Baker and Wurgler, 2007). Academic articles by academics and from corporations operating in the financial sector (such as Merrill Lynch), financial newspapers (CNN) or the American Association of Individual Investors (American Association of Individual Investors). Investors (AAII), Investor Sentiment (Investor Sentiment) can also be called by different names as Market Sentiment or Investor Attention. In general, it means the common attitude of investors when predicting the price development of the market. This attitude is the accumulation of a variety of fundamental and technical factors, including price history, economic reports, seasonal factors, economic events, or is derived from the investor's own personality. On the other hand, some authors may consider investor sentiment as a tendency to measure investors' reactions by signals from noisy information in the market instead of official information; on the other hand, the same term is used to refer to investors' optimism or pessimism. Another definition would be as emotional, which is defined as investors' aversion to risk (Brad and Terrance, 2013). ➢ Methods of measuring market sentiment Market psychology is a topic that connects outcomes from behavioral finance, assessing the impact of investor sentiment on financial markets, and the fundamentals of asset pricing (Barberis et al. associates, 1998; Barberis and Thaler, 2003; Barker and Wurger, 2007). Scholars argue that investors' behavioral patterns have a significant impact on stock market returns. Up to the present time, there are five main methods of measuring market sentiment known in the scientific literature: (i) estimation based on fundamental market indicators, (ii) index survey-based sentiment, (iii) market sentiment data from specialized online sources (news analysis), (iv) Internet search behavior, and (v) non-economic factors. RESEARCH HYPOTHESIS Hypothesis: Psychological factors (SENT) have an impact on stock returns (Ri) of listed companies on the Vietnamese stock market in each stock market cycle. RESEARCH DATA 122 companies are listed on the Ho Chi Minh City Stock Exchange (HOSE) and the Hanoi Stock Exchange (HNX). These companies have been listed and traded since January 2009 or earlier and fully traded and not delisted since during the research period. 123
6. Research period The research period is the Vietnamese stock market from 2009 to 2019, going through 3 cycles and 2019 is the beginning of the sixth cycle. This research selects 3 research periods, which are three stock cycles (cycles II, IV, V), in which phase 3 is the extension of the fifth cycle until the end of 2019. #INSERT TABLE 1 RESEARCH RESULTS The State Securities Commission was established on November 28, 1996 and on July 28, 2000, Vietnam's stock market officially came into operation, with the representative index being VN Index with base value is 100 points. So far, after nearly 20 years of operation, the investment channel in the stock market has proven to be the most effective investment channel, achieving an average return of 12.5%/year. # INSERT FIGURE 1 Vietnam's stock market has grown strongly in size, continuously improved its structure, contributed to perfecting the market economy institution and promoted international integration, becoming an important capital channel importance of the economy. Vietnam's stock market currently has two stock exchanges (HOSE): HOSE and HNX. Derivatives stock market, born in 2017, is an effective investment and risk prevention channel. Thus, the structure of the stock market including: stock market, bond market and derivatives market has been realized in Vietnam and is constantly developing. Thus, it can be said that after nearly 20 years, Vietnam's stock market has not only developed rapidly in terms of "quantity", but also has markedly improved in terms of "quality". Vietnam's stock market is increasingly affirming its role and position in the economy, being an important capital mobilization channel for investment in socio-economic development. Results of the analysis of psychological load factor (SENT) # INSERT TABLE 2 In each research period, the data constituting the psychological index has changed sharply, in which period 3 is the outstanding growth, which is consistent with the reality of Vietnam's stock market. In terms of scale, it is the growth of the number of initial public offering (NIPO) shares that grow steadily over each period as shown by the number of issues. Basically, the change of the above factors has proven to be significant in part to reflect market sentiment and market sentiment will change over time for each period (stock market cycle, stock market cycle, etc.). Correlation between variables ➢ Period 1 #INSERT TABLE 3 The variable number of shares traded (TURN) has a negative relationship with the number of 124
7. shares first issued to the market (NIPO) and the number of additional shares issued (S). In addition, the correlation between the initial public offering (NIPO) variable and the initial public offering (RIPO) yield has the largest correlation, about 40.56%. However, in general, the linear relationship of 5 variables in period 1 is not strong (less than 60%), consistent with the condition of principal component analysis. ➢ Period 2 The variable number of shares traded (TURN) has a negative correlation with the variable dividend compensation (PDND), the remaining variables have a positive relationship with each other. In general, all variables have low correlation, consistent with the conditions of principal component analysis (PCA) and there is no autocorrelation between independent variables. However, the correlation coefficient between the variable NIPO - the number of shares issued for the first time and the variable RIPO - the yield of shares issued for the first time to the market is very high, almost equal to 1, which can be explained in the following reasons. In the second period from 2012 to 2014, the number of new shares issued for the first time to the market is very small, specifically, only April 2014 and September 2014 have new shares first issued, the rest of the month NIPO is equal to 0. When the number of shares initially issued to the market (NIPO) = 0 leads to the yield on the initial public offering (RIPO) also = 0. Therefore, the correlation coefficient of the NIPO variable and the RIPO variable in period 2 is so high (approximately 1). ➢ Period 3 In period 3, the correlation between variables is generally not significant. The correlation between the number of shares traded (TURN) and the number of initial public offerings (NIPO) is 24.53% and is also the highest. In addition, the pairs with negative correlation are dividend premium (PDND) and initial stock yield (RIPO); yield of shares issued for the first time and number of additional shares issued (S); yield of initial shares (RIPO) and number of shares traded (TURN). Therefore, the linear relationship of the variables is suitable for the condition of principal component analysis. Result of main component analysis 5 factors (PCA) ➢ Result of eigenvalue selection #INSERT TABLE 4 In all 3 periods, the main components 1 and 2 of all three research periods have eigenvalues greater than 1. Specifically, in phase 1, the eigenvalues of components 1, 2 and 3, respectively are 1,695,1,299 and 1,004, these three components account for the cumulative 80% of the variance. In period 2, the eigenvalues of components 1 and 2 are 2,180 and 1,107, respectively, which account for the cumulative 65.74% of the variance. The eigenvalues of components 1 and 2 in period 3 are 1,475 and 1,187, respectively, which account for the cumulative 53.25% of the variance. The higher the cumulative percentage of variance, the better it is that the new variable retains information from the original variables without losing its significance. However, in all 3 periods principal component 1 has a higher cumulative percentage of variance than the cumulative percentage of principal 125
8. component 2. This means principal component 1 (referred to as PC1) interpret the data results of the original variables better. Therefore, choosing the main component 1 (PC1) in all 3 periods is to represent the market sentiment index for each research period. ➢ Result of eigenvector #INSERT TABLE 5 Eigenvector of the components in each principal component (PC) reflects the variance vector value of each variable for that principal component. It can be understood that, each principal component is an axis in the space of the linear association between variables. Main component 1 (PC1) was chosen to represent the market sentiment index at each period. Therefore, for the main component 1 (PC1), the load vector value 1 is selected as the coefficients corresponding to the research variables that make up the market sentiment index. ➢ Inspection of the suitability of the model SENT 5 factors #INSERT TABLE 6 According to the analysis results, the KMO coefficients of all three research periods are less than 0.5, so it is not suitable for PCA analysis. Based on each KMO coefficient of each component, it is necessary to remove the variable with low KMO coefficient because it does not fit the data set. However, with some variables significant significance can be kept (KMO > 0.3). Therefore, the research removed the research variable with the lowest KMO coefficient in each period and then performed a new principal component analysis according to 4 factors. #INSERT TABLE 7 Result of main component analysis 4 factors (PCA) ➢ Result of eigenvalue selection #INSERT TABLE 8 With 4 variables/components, using PCA analysis on Stata, the results obtained are 4 main components (Principal Component) which are ranked in order of decreasing eigenvalues and have independent relationships with each other. For PC, eigenvalues >= 1 are considered suitable for representing variables. Principal component 1 in all 3 periods has the highest percentage of cumulative variance. This means that principal component 1 (referred to as PC1) explains the data results of the original variables better. Therefore, choosing the main component 1 (PC1) in all 3 periods is to represent the market sentiment index for each research period. ➢ Result of eigenvector #INSERT TABLE 9 In period 1, the RIPO variable has the strongest effect on PC1 while the variable S has a negligible effect (the S variable has the strongest effect on PC2). The other two variables TURN and PDND have a rather high correlation at PC1 (48.41% and 55.45%), respectively. 126
9. In period 2, variables NIPO and RIPO have the strongest and positive impact on PC1 (equivalent to 66.01%) while variable S has the smallest effect (13.41%). This effect is reversed for PC2, while NIPO and RIPO have a negative effect on PC and the effect of S is the largest, followed by TURN. In period 3, there is a fairly uniform level of impact among the variables, showing the dominance of each variable on PC in order of NIPO, TURN, S and PDND, respectively. The vector relationship of variables for PC is shown through the loading plot chart for each period as follows: #INSERT FIGURE 2,3,4 Thus, looking at the graph, we can see a clear explanation of the relationship of each component variable with respect to principal components 1 and 2. In Figure 2, all variables are located on the positive axis of PC1 and the variables. RIPO TURN and PNDN are closer to the axis than S, the opposite is true for PC2. In Figure 3, the two variables NIPO and RIPO are closest to each other and close to the axis, showing that these two variables have the strongest impact on PC; while the variable S and TURN tend to be located quite far and further away from PC2. In Figure 4, the variable S is closest to the axis, but the value is not large compared to the two variables NIPO and TURN. While the PNDN variable is closer to the second axis and has the smallest impact on PC1. #INSERT FIGURE 5,6,7 Considering the three periods, the linear trend among the variables in the second period is strongest, with only a few outliers such as Sep-2013 or Oct-2013. The linear trend of the third period is less strong, and the last period 1 is quite discrete, but basically, they show a linear trend in all three research periods. ➢ Inspection of the suitability of the model SENT 4 factors INSERT TABLE 10 The KMO test results show that the average KMO coefficient of all three periods is greater than 0.5, the KMO coefficients of the component variables have relatively good values (greater than 0.3). Therefore, the dataset is suitable for PCA analysis. Results of the construction of the Market Sentiment Index Based on the PCA principal component analysis, it can be concluded that the market sentiment index is affected by 4 factors out of 5 factors according to the research of Baker and Wurgler (2006) when applied in the Vietnamese market. Male. Moreover, the impact of the factors in each period is different. From the PCA analysis, the first principal of all components has an eigenvalue greater than 1 and explains the high total variance extracted for the entire data set. Therefore, the market sentiment index in the Vietnamese stock market is expressed by the value of the first principal component when analyzing the PCA for each period. The market sentiment index had the strongest impact in mid-2014, when the market was 127
10. gradually recovering, and positive sentiment prevailed over the market. The most negative impact of market sentiment lies in the early years of 2009 and 2019, with the SENT index approximately negative 2 (-2). The results of the impact of psychological factors on stock returns through CAPM Correlation analysis among the variables in the model #INSERT TABLE 11 Through three research periods over a period of 11 years, it can be seen that the average values of the stock risk premium and market risk indicators tend to decrease, and at the same time the narrowing of the deviations benchmarks show that RIRF and RMRF tend to be less volatile over time. Besides, the minimum value of CAPM variables (RIRF and RMRF) increased, confirming a positive sign in the stock market when investing in this market showed signs of being safer. However, in the period 2, the maximum (maximum) values at RIRF and RMRF are the smallest, showing that this period the investment in the market is not as explosive as the previous two periods. The average value of SMB and HML fluctuates not much in all 3 periods. The standard deviation of SMB fluctuates strongly at period 1 which is a remarkable point in descriptive statistical analysis. DISCUSSION Comment on load factor SENT Based on the results of the PCA, it can be concluded that psychological factors exist in all three research periods and can be measured through research variables. In period 1, the SENT index is influenced by 4 factors: Yield of shares first issued to the market, Number of additional shares issued, Number of shares traded and Dividend premium. In period 2, the market sentiment index is measured by 4 variables: Number of shares issued for the first time to the market, Yield of shares issued for the first time to the market, Number of additional shares issued and Number of shares traded. In period 3, the market sentiment index is measured by 4 variables: Number of shares issued for the first time to the market, Number of additional shares issued, Number of shares traded and dividend premium. Comments on the impact of psychological factors on stock investment returns on the Vietnamese stock market Through the analysis of psychological factors and the impact model of SENT on investment returns of all three periods, it can be concluded that the market sentiment index is statistically significant for both models valuation is CAPM. Through the analysis results, it can be said that the results of the models are an extension of the traditional CAPM when it is proved that the psychological factor (SENT) has an impact on return on investment (Ri). Beta coefficient (which determines the sensitivity of stock returns to market volatility) tends to decrease in three periods. Thus, when looking at the results of model analysis by 128
11. panel data for 122 companies observed during the 11-year research period (2009-2019), it can be concluded: at period1, the coefficient The beta of 122 research companies is close to the market beta coefficient, proving that the coverage of the observed objects is close to covering the entire stock market. This analysis result is consistent with the data selection condition of the research. In the 2nd and 3rd periods, the beta coefficient is strongly reduced, proving that the observed objects gradually have a smaller and narrower impact compared to the stock market for the following years. This is a normal phenomenon and is consistent with the laws of market movement. More and more large-scale companies are listed, and liquidity will be concentrated in prominent and large-cap companies. Since the launch of the VN100 index, it can be seen that there has been a significant decrease in the number of 122 companies observed in the above index. Therefore, along with the ability to cover the market, the coefficient of determination of the model also tends to decrease. INSERT TABLE 18 Thus, through the results of analysis and assessment of the impact of psychological factors on stock investment returns through the valuation model, it can be summarized as follows: ✓ Firstly, the market has psychological factors and has the impact of psychology (expressed through psychological factors) on the return of stock investment. ✓ Second, psychological factors have different effects on stock valuation in three different periods, which is caused by the narrowing of the size and the strength of the market of the observed subjects. ✓ Third, psychological factors have a positive impact on large companies. This is shown through period 1 and period 2 when companies' ability to cover the market is high. Besides, market sentiment negatively affects small companies. This is reflected in the third phase when the ability to cover the market of the companies under the research shrinks with the number of companies on the VN100 list decreasing sharply over the years (since 2015 there are 54 companies in the VN100 list). By 2019 there are 35 companies on the VN100 list). CONCLUSION AND RECOMMENDATION Recommendations for the development of Vietnam's stock market through the results of research on psychological factors affecting the return of stock investment Through the results of quantitative analysis throughout the research, it can be confirmed that the Vietnamese stock market has psychological factors, and psychological factors have a clear impact on the return on investment in stocks on the stock exchange. deal. Therefore, this proves that investors in the Vietnamese stock market are irrational, just as the Vietnamese stock market is imperfect. This implies that the explanation of decisions in the investment process of investors cannot be based on standard financial theories but must be based on theories of behavioral finance. Thus, from the evidence that is clearly confirmed on the model and the theory from the development on the Vietnamese stock market, the research makes some recommendations to develop the Vietnamese stock market for the parties. related, including: (i) Individual investors and (ii) Policy 129
12. recommendations to develop a high-class and integrity stock market system. Recommendations for individual investors Choose the right valuation method The first prerequisite when participating is that individual investors must improve their understanding of the stock market to equip themselves with basic knowledge before investing. In addition, it is necessary to improve the ability to analyze, evaluate and build an investment philosophy to avoid the situation of investment following the crowd. Controlling emotional factors in investing All individual investors need to pay attention and identify their own psychological factors when participating in investment. Market sentiment is a factor that affects stock prices. The behavior of stock investors is influenced by emotional factors. Therefore, in order to be successful in the stock market, besides equipping with standard financial knowledge, investors need to have understanding of behavioral finance or emotional finance. to control emotions in their investment decision-making process, as well as being able to identify deviant emotional factors that are common in the market such as: ➢ Avoid “overconfidence”: Overconfidence in investment behavior is also the main cause of bubbles in the stock market and the risk incurred by investors will be high ➢ Avoiding the “allocation effect”: Investors have a psychological tendency to dislike losses, leading to often selling stocks that are rising rather than selling stocks that are falling. ➢ Avoiding “deviant dependence on reference price”: Investors should avoid being too dependent on a reference price or when forecasting market trends too much focus on historical movements will lead to mistakes in making investment decisions. Enhancing the financial literacy of individual investors Constantly improving knowledge from practical lessons and past experiences is the most important factor to help individual investors succeed in the stock market. Individual investors need to develop an investment philosophy to avoid the phenomenon of crowd investing. Policy recommendations The research clearly demonstrated the impact of psychological factors on the stock market. Therefore, it is absolutely necessary for the Government and competent parties to recognize and promptly implement reasonable policies to develop the stock market. Therefore: Completing the legal framework in the financial sector, ensuring the agreement between financial regulations and other relevant regulations. When a legal corridor is completed, many benefits will be achieved. Investors, especially individual investors, will increase their confidence in a healthy and developed market, reduce irrational decisions that come from lack of information, or fake news, and even current events. price 130
13. object of some objects in the market. The stock market develops and will positively affect all economic sectors in society. Establish an agency to protect individual investors and enhance transparency on the Vietnamese stock market ➢ Decentralizing supervision activities to the Stock Exchanges, the State Securities Commission only supervises trading members. ➢ Separation of direct supervision to monitor compliance with legal regulations and indirect supervision of unfair transactions. Renovating the policy of price fluctuation range and short selling policy In the long term, remove the price fluctuation range and perform a market circuit breaker if there is a strong fluctuation. This solution will prevent deviations in investors' behavior, forcing them to consider and analyze carefully before making decisions. Besides, according to each market movement and cycle, different circuit breaker amplitude can be applied for each market period. Continue to widely inform policies It is necessary to establish different communication channels so that the Government can capture investor sentiment at different times through surveys and surveys so that it can collect and promptly reflect changes in sentiment investors' management, thereby implementing appropriate policies to regulate the market. Strengthening market supervision to identify complex psychological trends that may cause bad developments in financial markets. It is necessary to improve the capacity of supervisory staff and modernize the technical and technological infrastructure, especially the application of market management software, thereby, contributing to further improving the quality of the company forecaster 131
14. REFERENCES Alan Kraus, Robert H Litzenberger ,(1976), “Skewness Preference and the Valuation of Risk Assets”, Journal of Finance, vol. 31, pp. 1085-1100. Apergis & Rehman, (2018), “Is CAPM a Behavioral Model? Estimating Sentiments from Rationalism”, Journal of Behavioral Finance, vol. 19, pp. 442- 449. Baker & Wugler, (2007), “Investor Sentiment in the Stock Market”, Journal Of Economic Perspectives, vol. 21, pp. 129-152. Baker, Wurgler & Yuan, (2009), “Global, Local, and Contagious Investor Sentiment”, Journal of Financial Economics, vol. 104, pp. 272-287. Barberis & Thailer, (2003), “A survey of behavioral finance”, Handbook of the Economics of Finance, pp.1-78. Basu, (1977), “Investment performance of common stocks in relation to their price‐earnings ratios: A test of the efficient market hypothesis”, Journal of Finance, vol. 32, pp. 663-682. Burcu Dinỗergửk, (2018), “Financing of working capital requirement and profitability: evidence from Borsa Istanbul chemical, petroleum, rubber, and plastic sector”, Financial Management from an Emerging Market Perspective, pp.175-187. Byrne & Brooks, (2008), “Behavioral Finance: Theories and Evidence”, Research Foundation Literature Reviews, Vol 3. Chaudhuri & Smiles, (2004), “ Stock market and aggregate economic activity: evidence from Australia”, Applied Financial Economics, vol. 14, pp. 121-129. Chen, Roll & Ross, (1986), “Economic forces and the stock market”, The Journal of Business, vol. 59, pp. 383-403. Dimitrios Tsoukalas, (2003), “Macroeconomic factors and stock prices in the emerging Cypriot equity market”, Managerial Finance, vol. 29, pp. 87-92. Fama, (1970), “Efficient Capital Markets: A Review of Theory and Empirical Work”, The Journal of Finance, vol. 25, pp. 383-417. Fama & French, (1992), “The cross-section of expected stock returns”, Journal of Finance, vol. 27, pp. 427-765. Hirshliefer & Teoh, (2003), “Herding Behavior in Futures Market: An Empirical Analysis from India”, Theoretical Economics Letters, Vol.7. Hirshliefer & Teoh, (2009), “Driven to Distraction: Extraneous Events and Underreaction to Earnings News”, Journal of Finance, vol. 64, pp. 2289-2325. Huang, Yang & Sheng, (2013), “An Empirical Study of the Effect of Investor Sentiment on Returns of Different Industries”, Mathematical Problems in Engineering, vol. 45, pp. 1-11. Ibrahim, (1999), “Macroeconomic Variables and Stock Prices in Malaysia: An Empirical Analysis”, Asian Economic Journal, vol. 13, pp. 219-231. Kelly & Pruitt, (2013), “Market Expectations in the Cross‐Section of Present Values”, 132
15. Journal of Finance, vol. 68, pp. 1721-1756. Mahmood & Dinniah, (2007), “Stock Returns and Macroeconomic Influences: Evidence from the Six Asian-Pacific Countries”, Financial Economics and Futures Market Research Paper, pp. 1-21. Maysami & Koh, (2000), “A Vector Error Correction Model of the Singapore Stock Market”, International Review of Economics and Finance, vol. 9, pp. 79- 96. Mian & Sankaraguruswamy, (2012), “Investor Sentiment and Stock Market Response to Earnings News”, The Accounting Review, vol. 87, pp. 1357-1384. Mukherjee & Naka, (1995), “ Dynamic Relations between Macroeconomic Variables and the Japanese Stock Market: An Application of a Vector Error Correction Model”, Journal of Financial Research, vol. 18, pp. 222-237. Nguyen Ngoc Tu Van (2018). Impact of investor sentiment on ASEAN stock price index. Review of Finance. Oprea & Brad, (2014), “Investor Sentiment and Stock Returns: Evidence from Romania”, International Journal of Academic Research in Accounting, Finance and Management Sciences, Vol. 4, pp. 23-29. Pandey & Sehgal, (2019), “Investor Sentiment and its Role in Asset Pricing: An Empirical Study for India”, IIMB Management Review, vol. 31, pp. 127-144. Phan Thi Nha Truc (2019). Building a sentiment index to measure factors affecting investor sentiment in Vietnam, Review of Finance. APPENDICES Table 1. Analysis periods Period Cycle Start End Trend Time Bottom Time Bottom point point 2009-2011 III 23/02/2009 244.02 1/2/2012 350 Increase 2012-2014 IV 1/2/2012 350 22/12/2014 537.54 Increase 2015-2019 V 22/12/2014 537.54 24/12/2018 908.56 Increase Source: Author Table 2. Descriptive statistics of variables representing psychological indicators by period Name NIPO RIPO S TURN PDND Obs Period 1 36 36 36 36 36 133
16. Period 2 36 36 36 36 36 Period 3 60 60 60 60 60 Mean Period 1 9.266 0.350 0.009 0.123 0.381 Period 2 3.083 0.003 0.004 0.070 0.215 Period 3 31.861 1.05 0.003 0.057 0.300 Std.Dev Period 1 16.453 0.784 0.012 0.097 0.338 Period 2 18.425 0.017 0.005 0.026 0.228 Period 3 31.098 4.291 0.004 0.017 0.334 Min Period 1 0.000 0.000 0.000 0.027 -0.526 Period 2 0.000 0.000 0.000 0.022 -0.144 Period 3 0.000 -1.000 0.000 0.028 -0.367 Max Period 1 72.993 3.594 0.048 0.363 1.040 Period 2 100.56 0.100 0.02 0.132 0.753 Period 3 465.73 29.000 0.025 0.096 1.112 Table 3. Correlation between the variables Period Variable NIPO RIPO S TURN PDND NIPO 1.000 RIPO 0.046 1.000 Period 1 S 0.070 0.043 1.000 (2009-2011) TURN -0.144 0.353 -0.209 1.000 1.000 PDND 0.076 0.396 0.180 0.096 NIPO 1.000 RIPO 1.000 1.000 Period 2 S 0.083 0.086 1.000 134
17. (2012-2014) TURN 0.277 0.277 0.137 1.000 1.000 PDND 0.111 0.112 0.149 -0.005 NIPO 1.000 RIPO 0.212 1.000 Period 3 S 0.245 -0.031 1.000 (2015-2019) TURN 0.245 -0.076 0.100 1.000 1.000 PDND 0.126 -0.127 0.045 0.145 Table 4: Results eigenvalues 5 factors each period Period Component Eigenvalue Difference Proportion Cumulative Comp1 1.695 0.396 0.339 0.339 Comp2 1.299 0.295 0.260 0.599 Comp3 1.004 0.332 0.201 0.800 Period 1 Comp4 0.671 0.341 0.134 0.934 Comp5 0.331 0.066 1.000 Comp1 2.180 1.073 0.436 0.436 Comp2 1.107 0.146 0.221 0.657 Period 2 Comp3 0.961 0.208 0.192 0.850 Comp4 0.753 0.753 0.151 1.000 Comp1 1.475 0.288 0.295 0.295 Comp2 1.187 0.261 0.238 0.533 Comp3 0.926 0.090 0.185 0.718 Period 3 Comp4 0.836 0.261 0.167 0.885 Comp5 0.575 0.115 1.000 135
18. Table 5: Results eigenvector 5 factors each period Period Component Eigenvalue Difference Proportion Cumulative Comp1 1.695 0.396 0.339 0.339 Comp2 1.299 0.295 0.260 0.599 Period 1 Comp3 1.004 0.332 0.201 0.800 Comp4 0.671 0.341 0.134 0.934 Comp5 0.331 0.066 1.000 Comp1 2.180 1.073 0.436 0.436 Comp2 1.107 0.146 0.221 0.657 Period 2 Comp3 0.961 0.208 0.192 0.850 Comp4 0.753 0.753 0.151 1.000 Comp1 1.475 0.288 0.295 0.295 Comp2 1.187 0.261 0.238 0.533 Period 3 Comp3 0.926 0.090 0.185 0.718 Comp4 0.836 0.261 0.167 0.885 Comp5 0.575 0.115 1.000 Table 6: KMO results 5 factors each period KMO Variable Period 1 Period 2 Period 3 NIPO 0.343 0.497 0.477 RIPO 0.427 0.497 0.38 S 0.508 0.349 0.529 TURN 0.350 0.906 0.546 PDND 0.514 0.187 0.545 136
19. Chung 0.411 0.492 0.486 Table 7: Result of removing variable Period Variable Components for the 4-factor PCA Period 1 NIPO RIPO, S, TURN, PDND Period 2 PDND NIPO, RIPO, S, TURN Period n 3 RIPO NIPO, S, TURN, PDND Table 8: Results eigenvalues 4 factors each period Period Component Eigenvalue Difference Proportion Cumulative Comp1 1.581 0.348 0.395 0.395 Comp2 1.233 0.559 0.308 0.704 Period 1 Comp3 0.674 0.162 0.169 0.872 Comp4 0.512 0.128 1.000 Comp1 2.156 1.127 0.539 0.539 Comp2 1.029 0.214 0.257 0.796 Period 2 Comp3 0.815 0.815 0.204 1.000 Comp4 0.000 0.000 1.000 Comp1 1.473 0.492 0.368 0.368 Comp2 0.981 0.135 0.245 0.614 Period 3 Comp3 0.846 0.147 0.212 0.825 Comp4 0.700 0.175 1.000 Table 9: Results eigenvector 4 factors each period Period Variable Comp1 Comp2 Comp3 Comp4 137
20. RIPO 0.6752 0.0047 0.0599 -0.7352 S 0.0473 0.7633 0.6365 0.1002 Period 1 TURN 0.4841 -0.5202 0.5117 0.4829 PDND 0.5545 0.3832 -0.574 0.465 NIPO 0.6602 -0.1926 -0.1643 RIPO 0.6601 -0.1931 -0.1643 Period 2 S 0.1341 0.8673 -0.4795 TURN 0.3323 0.4165 0.8462 NIPO 0.6105 -0.1918 -0.1087 -0.7607 S 0.462 -0.619 0.4324 0.4652 Period 3 TURN 0.5271 0.2656 -0.669 0.4517 PDND 0.3687 0.7138 0.5947 0.0309 Table 10: KMO results 4 factors each period KMO Biến Period 1 Period 2 Period 3 NIPO . 0.516 0.562 RIPO 0.519 0.516 . S 0.473 0.45 0.575 TURN 0.503 0.913 0.594 PDND 0.54 . 0.638 Chung 0.515 0.532 0.581 Market sentiment index over time 138
21. Period 1: 푆 1 = 0.675푅 푃 + 0.047푆 + 0.484 푈푅 + 0.555푃 Period 2: 푆 2 = 0.660 푃 + 0.660푅 푃 + 0.134푆 + 0.332 푈푅 Period 3: 푆 3 = 0.611 푃 + 0.462푆 + 0.527 푈푅 + 0.369푃 Table 11. Matrix of correlation coefficients between variables in the research model RMRF SENT Period 1 RMRF 1 SENT 0.4836 1 ECON 0.0673 0.1814 Period 2 RMRF 1 SENT -0.1068 1 ECON 0.1045 -0.0903 Period 3 RMRF 1 SENT 0.0194 1 ECON 0.0908 -0.1595 The model of the impact of psychological factors on stock returns through CAPM Period 1: Ri = Rf + 1.028(Rm- Rf)+0.011SENT Period 2: Ri = Rf + 0.828(Rm-Rf) + 0.011SENT Period 3: Rf + 0.439(Rm-Rf) - 0.006SENT Table 12: Results of regression analysis method POLS 139
22. Period RIRF Coef. S.E t P>t R-Squared Period 1 RMRF 1.028 0.021 48.78 0.000 0.4575 SENT 0.011 0.002 6.83 0.000 Period 2 RMRF 0.828 0.033 25.43 0.000 0.1373 SENT 0.011 0.002 9.37 0.000 Period 3 RMRF 0.439 0.027 16.16 0.000 0.0356 SENT -0.006 0.001 -5.3 0.000 Table 13: Inspection of the suitability of the model by POLS Multicollinearity Heteroskedasticity Autocorrelation F Prob > SENT RMRF MEAN Chi2 P-value (1,121) F Period 1 1.34 1.31 1.23 186.31 0.000 0.093 0.762 Period 2 1.02 1.02 1.02 60.82 0.000 3.403 0.068 Period 3 1.03 1 1.02 2.000 0.9914 2.209 0.140 Table 14. FEM result Period RIRF Coef. S.E t P>|t| F(3,n-5) Prob >F RMRF 1.028 0.021 48.63 0.000 Period 1 SENT 0.011 0.002 6.81 0.000 1225.48 0.000 _CONS 0.002 0.003 0.89 0.374 RMRF 0.828 0.032 25.32 0.000 Period 2 SENT 0.011 0.001 9.33 0.000 140
23. _CONS -0.001 0.003 -0.47 0.638 230.84 0.000 RMRF 0.439 0.027 16.13 0.000 Period 3 SENT -0.005 0.001 -5.29 0.000 89.64 0.000 _CONS -0.03 0.002 -17.64 0.000 Table 15: REM result RIRF Coef. S.E Z P>|z| Wald chi Prob RMRF 1.028 0.021 48.78 0.000 Period 1 SENT 0.011 0.002 6.83 0.000 3699.99 0.000 _CONS 0.002 0.003 0.89 0.372 RMRF 0.828 0.326 25.43 0.000 Period 2 SENT 0.011 0.001 9.37 0.000 698.47 0.000 _CONS -0.001 0.003 -0.47 0.000 RMRF 0.439 0.027 16.16 0.000 Period 3 SENT -0.006 0.001 -5.30 0.000 269.95 0.000 _CONS -0.03 0.002 -17.67 0.000 Table 16. Hausman result (choose REM or FEM) Period Variable FEM REM S,E Chi2 Prob (V_b - (b) (B) V_B) Period 1 RMRF 1.028 1.028 0.002 0.000 1.000 SENT 0.011 0.011 0.000 Period 2 RMRF 0.828 0.828 0.003 0.000 1.000 SENT 0.011 0.011 0.000 141
24. Period 3 RMRF 0.439 0.439 0.002 0.000 1.000 SENT -0.006 -0.006 0.000 Table 17. REM test Kiểm định phương Kiểm định tự sai sai số thay đổi tương quan F Prob > Chi2 P-value (1,121) F Period 1 0.000 1.000 0.093 0.7615 Period 2 0.000 1.000 3.403 0.068 Period 3 0.000 1.000 2.209 0.140 Table 18: Number of companies in 122 companies researched in the VN100 index period 1 from 2015 to 2019 Year 2015 2016 2017 2018 2019 Number of companies in 54 49 40 38 35 VN100 Source: HNX&HOSE Figure 1: VN Index and trading volume of Vietnam's stock market since its establishment 2000- 2019 7,000 1,400 6,000 1,200 5,000 1,000 4,000 800 3,000 600 2,000 400 1,000 200 0 0 31-Jul-2009 31-Jul-2019 31-Jul-2000 31-Jul-2001 31-Jul-2002 31-Jul-2003 31-Jul-2004 31-Jul-2005 31-Jul-2006 31-Jul-2007 31-Jul-2008 31-Jul-2010 31-Jul-2011 31-Jul-2012 31-Jul-2013 31-Jul-2014 31-Jul-2015 31-Jul-2016 31-Jul-2017 31-Jul-2018 Volume Close 142
25. Figure 2: Relationship between eigenvector of variables for PC1 and PC2 period 1 Figure 3: Relationship between eigenvector of variables for PC1 and PC2 period 2 Figure 4: Relationship between eigenvector of variables for PC1 and PC2 period 3 143
26. Figure 5: Distribution of the data set period 1 Figure 6: Distribution of the data set period 2 Figure 7: Distribution of the data set period 3 144
27. 01-09 06-09 Figure 11-09 04-10 09-10 8 02-11 : Graph depicting movements of the SENT index theSENT of movements Graph depicting 07-11 12-11 05-12 10-12 145 03-13 08-13 01-14 06-14 11-14 04-15 09-15 02-16 07-16 12-16 05-17 10-17 03-18 08-18 01-19 06-19 11-19