Evidences of herding behavior in vietnam stock market: A sectorial analysis

Nội dung text: Evidences of herding behavior in vietnam stock market: A sectorial analysis

1. EVIDENCES OF HERDING BEHAVIOR IN VIETNAM STOCK MARKET: A SECTORIAL ANALYSIS Mai Sinh Thoi*1 ABSTRACT: The study examines herding behavior in Vietnam stock market at sectorial level under different market conditions. The study employs the cross-sectional absolute deviation (CSAD) method developed by Chang, Cheng and Khorana(2000). The data contains daily 135 stock prices divided into 20 industries from 02nd Jan/2007 to 31st Dec/2017. Those stocks are listed on Hanoi stock exchange (HNX) or Ho Chi Minh stock exchange (HOSE). The estimated results demonstrate that herding is evident for the whole Vietnam stock market, under both up and down market. Herding is stronger in up market than down market. At sectorial level, all industries except for construction show the evidence of herding and 15 out of those 19 industries shows the stronger herding during up market than down market. However, the whole market and all industries do not show the evidence of herding under extreme up and down market conditions. During 2008 global financial crisis, the whole Vietnam stock market shows the marginally significant evidence of herding. Herding is weaker during the crisis period than normal period. At sectorial level, a half of 20 industries show the evidence of herding during the crisis and 8 out of those industries shows the weaker herding during the crisis period than normal period. Keywords: Herding behavior, sectorial level, Vietnam stock market. 1. INTRODUCTION: Herding behavior in stock market is defined as an imitative situation in which investors ignore their own analysis and follow the action of other investors that can be market consensus (Bikhchandani and Sharma, 2001).In behavioral perspective, herding behavior can be seen as irrational behavior. Herding behavior has adverse impact on stock market. It drives stock prices far away from fundamental value and raises volatility (Bikhchandani and Sharma, 2001). During booming time, herding behavior can make stock prices surge and cause bubbles. During shocking times, herding behavior can make stock prices fall and cause the overreaction of investors and crash for stock market, which often occurs in several financial crisis. Herding phenomenon often occurs when there is a limitation of information or knowledge which makes investors unconfident about their own analysis and choose to follow market consensus. Vietnam stock market as an emerging market is a good one to test for the existence of herding behavior. Due to the underdevelopment of information system, investors can face with insufficient information. Another reason is that the majority of investors in Vietnam stock exchange are individual investors with limited investment training. Hence, herding is likely to occur in Vietnam stock market. In the study, I plan to test for herding behavior in Vietnam stock market at sectorial level. The behavior is also examined during extreme market conditions and 2008 global financial crisis. The study is motivated by the following reasons: Firstly, my study makes contribution to literature. Most of studies investigating herding in Vietnam stock market concentrate on the entire market such as Bui et al. (2015), Vo and Phan (2016), Vo and Phan * University College Cork(UCC), Cork, Ireland, Phone number: (+84)349800533, Email: Mai.sinhthoi@umail.ucc.ie.
2. 472 HỘI THẢO KHOA HỌC QUỐC TẾ KHỞI NGHIỆP ĐỔI MỚI SÁNG TẠO QUỐC GIA (2017). To my best knowledge, Bui et al. (2017)is the only study testing herding at industry level. However, my study extends Bui et al. (2017) in a number of aspects. My study contains 5 more industries and herding is examined under more conditions like extreme market condition and crisis. The time period is extended from 2014 to 2017 to contain recent changes in Vietnam stock market. Secondly,Bui et al. (2015) and Vo and Phan (2017) shows inconsistent result about the herding direction between up and down market. My study helps to clarify the inconsistent result. Thirdly, my study result will be helpful to investors. Investors will understand further about herding behavior from each industry under different market conditions, their investment decision-making process. Based on that, they can create a more reasonable investment strategy. Fourthly, my study result will facilitate government authority to recognize the magnitude of herding behavior in each industry and propose solutions to enhance the efficiency ofV ietnam stock exchanges. The study employs the cross-sectional absolute deviation (CSAD) method developed by Chang et al. (2000), OLS regression method and secondary daily data to test for the evidence of herding behavior. The remainder of the paper is structured as follows: Section 2 - a literature review; Section 3 – methodology and data; Section 4- empirical results;Section 5 - conclusion. 2. LITERATURE REVIEW Herding behavior attracts a big number of studies. My study relates to at least three strands of literature. The first strand is studies relating to herding behavior in Vietnam stock market. Le and Truong (2014), Phan and Zhou (2014), Bui et al. (2015), Vo and Phan (2016), Vo and Phan (2017) and Bui et al. (2017) shows the evidence of herding behavior in Vietnam stock market, but the magnitude of herding under different market conditions is inconclusive. Le and Truong (2014) develops a new method which enables to detect herding behavior in a given trading day. Phan and Zhou (2014) interviews 20 individual investors in Vietnam stock market. Applying the method from Chang et al. (2000) to analyses 30 blue chips, Bui et al. (2015) shows that herding in Vietnam stock market only exist during up market. Vo and Phan (2016) applies the method developed by Chang et al. (2000) and quantile regression technique. The study finds that herding is stronger after 2008 global financial crisis. Before the crisis, herding is stronger in down market than up market. The opposite is true after the crisis.Vo and Phan (2017) applies 2 methods developed by Christie and Huang (1995) and Chang et al. (2000) to analysis daily, weekly and monthly stock return. The research shows that herding is short-lived as it exists only in daily and weekly data and herding in down market is stronger than up market, which is controversial with Bui et al. (2015). Herding exists in both low and high trading volume, but be stronger in low volume case. Herding does not exist during 2008 global financial crisis. Bui et al. (2017) employing the method from Chang et al. (2000) with some extensions study herding at 16 sectors from 1st Jan/2007 to 17st Oct/2014. The study finds evidence of herding existing in 14 sectors and investors herd more in up market than down market. The study also observes that U.S. stock market affects herd behavior in the Vietnamese stock market. The difference between my study and those studies is that my study extends their studies to consider herding not only in the whole Vietnam stock market but also in each industry. The second strand is studies relating to herding behavior in stock market from sectorial perspectives. Gebka and Wohar (2013) investigates the existence ofherding in the global equity market, shows that herding exists in some industries such as especially in basic materials, consumer services, and oil and gas.Litimi et al. (2016) examines herding phenomenon and its effect on volatility in US stock market at sectorial level by employing both methods from Christie and Huang (1995) and Chang et al. (2000). The research finds that consumer non-durables, tublic utilities, and Transportation industries herd toward
3. INTERNATIONAL CONFERENCE STARTUP AND INNOVATION NATION 473 market return.Shah et al. (2017) tests for the herding behavior in Pakistan stock exchange from different perspectives such as herding of firms towards market, herding of firms towards industry portfolios and herding of industry portfolios towards market by using the method of Christie and Huang (1995). The study shows strong evidence of firms in several industries herding towards their industry portfolios, but weak evidence of industry portfolios herding towards the market. The third strand relates to studies about the empirical presence of herding behavior in international stock markets. Herding phenomenon is tested in various stock markets. Bui et al. (2015), Economou et al. (2016),Tan et al. (2008) report the presence of herding in Indonesia, the Philippines, Malaysia, Greece and China stock markets. Chang et al. (2000) developed a novel method based on the study by Christie and Huang (1995) to examine herding in different international markets. This study reports no evidence of herding in the most developed countries such as US, Hong Kong, Japan but a strong evidence in emerging markets such as South Korea and Taiwan. Hwang and Salmon (2001) report the similar results to Chang et al. (2000). In contrast, some studies show the evidence of herding in developed markets. For example, Chiang and Zheng (2010) shows the evidence of herding in some developed stock markets when applying market index data to compute herding propensities in country-specific level. Litimi et al. (2016) indicates the presence of herding in America and herding is a vital factor for bubbles in the market. The distinction between my study and those studies in the second and third strands above is that my study tests for herding in Vietnam stock market. 3. METHODOLOGY AND DATA Chang et al. (2000) and Christie and Huang (1995) developed 2 methods to investigate the presence of herding behavior which are applied by numerous methods such as Bui et al. (2015), Vo and Phan (2016), Vo and Phan (2017), Litimi et al. (2016), Shah et al. (2017). However, the latter is criticized because of a number of drawbacks. Firstly, the method suggested by Christie and Huang (1995) only recognizes herding under the condition of extreme return. However, herd behavior may be present at the entire return distribution and become stronger during period of market stresses (Chiang et al., 2010). Secondly, the definition of extreme return is arbitrary. Thirdly, the method is considerably affected by the existence of outliers(Vassilakopoulos, 2014).Hence, my study employs the method from Chang et al. (2000). The model to test herding for each industry is as follows: 2 CSADj,t = g0 + g1|Rmt| + g2R mt+ ut (1) Where,|Rmt| is the absolute value of market return at time t. CSADjtis a cross-sectional absolute deviationof the industry j which is calculated as follows: 1 N = − CSADjt ∑1 Rit R mt N Where, Ri,t is the daily return of stock i in the industry j at time t, which is determined as follows: P R=100* Lnit, , P and P : Closing daily stock prices at t and t-1. it P i,t i,t-1 it,1− The cross-sectional absolute deviation of the whole market (CSADt) is based on stocks from all industries. Rm,t isthe market return which can be calculated by weighted average market return or equally weighted average market return. In the study, I apply the later method as it is similar to the original paper of Chang et al. (2000). Rm,t is calculated as the cross sectional average of the N returns at time t and it is the same for all industries. The formula isas follows:
4. 474 HỘI THẢO KHOA HỌC QUỐC TẾ KHỞI NGHIỆP ĐỔI MỚI SÁNG TẠO QUỐC GIA 1 N = RRmt ∑1 it N Chang et al. (2000) demonstrates that return dispersion measured as CSAD is an increasing function of the market return and the relation is linear. However, in the case of herding behavior, investors follow market consensus and ignore their own analysis of stocks, the CSAD will go down, the increasingly linear 2 relationship no longer holds; instead, it is more likely to be non-linear increasing or even decreasing. R mt is added in equation 1 to capture the non- linear relationship. Herding is present if coefficient γ2 in equation (1) is significantly negative. Model 2 and 3below is used to test for herding in up (Rm>0) and down (Rm 0 (2) 2 CSADj,t = g0 + g1|Rmt| +γ4R mt+ ut, if Rm < 0 (3) If herding is present in up and down market, the coefficients g3 and γ4 are expectedto be significantly negative. Those coefficients g3 and γ4indicate the magnitude of herding in up and down marketrespectively and the difference between g3 and γ4 shows the herding asymmetry in the two states. If herding is stronger in up market than down market, g3 is expected to higher than γ4 in absolute value. Herding can be stronger during extreme market conditions(Christie and Huang (1995)). Because the definition of extreme market conditions is arbitrary, the study applies 5% up or down tail of market return distribution. Model 4 and 5 below are used to test for herding during extreme up and down market conditions as follows: extreme up 2extreme up CSADj,t = g0 + g1|R mt| + γ5R mt+ ut, Rmt in 5% up tail of distribution (4) extreme down 2extreme down CSADj,t = g0 + g1|R mt| + γ6R mt+ ut, Rmt in 5% down tail of distribution (5) If herding is evident, the coefficients 5γ and γ6 are expectedto be significantly negative. They can be used to compare with g3 and γ4 to check if herding is stronger during extreme market conditions. Vo and Phan (2017) shows that herding does not exist in Vietnam stock market during the global financial crisis. However,Shah et al. (2017) shows the opposite result in Pakistan stock market. The study will examine the question again. The model 6 below is used to test for herding during 2008 global financial crisis. 2 2 CSADj,t = g0 + g1D1* |Rmt| +g2(1 –D1)* |Rmt| + γ7D1*R mt+ γ8(1 –D1)*R mt+ ut (6) st st D2 if a dummy variable, is equal 1 if the time is from 1 Jan/2008 and 31 Dec/2008 and 0 otherwise. The coefficient 7γ represents herding during the global financial crisis. It is expected to be significantly negative if herding exists during the crisis. The coefficient 8γ represents herding during normal times and it can be compared with γ7 for the asymmetric effect of herding between the crisis period and normal period. All the equations above are estimated by OLS method and employ Newey-West (1987)consistent standard errors to counter heteroskedasticity and autocorrelation problems. The data includes daily 135 stock prices divided into 20 industries, from 02nd Jan/2007 to 31st Dec/2017. The sample includes all stocks listed before 01st Jan/2007. The total observations is 373,270. All stocks are listed onHanoi stock exchange (HNX) or Ho Chi Minh stock exchange (HOSE). The data is taken from Datastream, Thomson Reuters. Table 1 below shows descriptive statistics. Due to the unavailability of data, each industry‘s CSAD has number of observations from 2660 to 2741. No variables are stationary since the null hypothesis of no unit root is rejected for all variables. The market return has mean of around 0 but big standard deviation. This is because Vietnam stock market experiences fluctuating times like 2008 global financial crisis.
5. INTERNATIONAL CONFERENCE STARTUP AND INNOVATION NATION 475 5. EMPIRICAL RESULT Table 2 below demonstrates the estimated results of herding for the whole market, each industry at 2 states, up market and down market. For the whole market, herding is evident as the coefficient 2g is significantly negative, which is consistent with previous studies such as Le and Truong (2014), Phan and Zhou (2014), Bui et al. (2015), Vo and Phan (2016), Vo and Phan (2017). Herding is stronger in up market than down market since the coefficient 3g is higher than the coefficientγ4. The result is consistent with Bui et al. (2015) and Bui et al. (2017). All 20 industries, except for construction shows the evidence of herding because of having negative significant 2g , which is consistent with Bui et al. (2017). 15 out of those 19 industries shows that herding is more evident in up market than down market. Only 2 industries, including pharmacy, health, chemical and mineral shows the opposite direction. The remaining 2 industries, including electricity, oil, gas and banking, insurance industries do not show the herding asymmetry between up and down market. Table 3 below shows the empirical result of herding under extreme up and down market. Based on the sign and significance of γ5 and γ6, for the whole market, herding is not evident under both extreme up and down market. The result is not consistent with Vo and Phan (2017). For 20 industries, none of γ5 and γ6 is negatively significant, so no industries shows herding under extreme market conditions. Table 4 below shows herding during 2008 global financial crisis. For the whole market, herding is evident during the crisis because of the negative sign and marginal significance of γ7 at 10%. The result is inconsistent with Vo and Phan (2017) who shows the absence of herding during the crisis. The difference is probably because of the difference in the method of calculating Rmt. My study applies equally weighted average market return while Vo and Phan (2017) applies weighted average market return. Since γ8 is higher than γ7 in absolute value, herding during normal period is stronger than herding during crisis. Among 20 industries, 10 industries, including: Real estate; rubber; securities; pharmacy, health and chemical; education; mineral; electricity, oil and gas; steel; food; construction show the evidence of herding during the crisis because of the negative significance of γ7. 8 out of those 10 industries shows that herding is stronger in normal period than crisis period. Construction industry shows the opposite direction. Pharmacy, health and chemical industry shows the equality of herding between 2 periods. The weaker herding during the crisis implies that investors are less likely to follow market consensus during crisis. 6. CONCLUSION The study investigates herding behavior in Vietnam stock market at sectorial level. Herding behavior is also examined under up and down market, extreme market condition, 2008 global financial crisis period. The study employs the cross-sectional absolute deviation (CSAD) method developed by Chang et al. (2000). The data contains daily 135 stock prices from 20 industries, from 02nd Jan/2007 to 31st Dec/2017. Those stocks are listed on Hanoi stock exchange (HNX) or Ho Chi Minh stock exchange (HOSE). The estimated results demonstrate that herding is found evident for the whole Vietnam stock market, under both up and down market. Herding is stronger in up market than down market, which is consistent with Bui et al. (2015) and Bui et al. (2017). At sectorial level, 19 out of 20 industries shows the evidence of herding and 15 out of those 19 industries shows the stronger herding during up market than down market. The remaining 4 industries shows the opposite direction or equality. However, under extreme up and down market conditions, the whole market and all industries do not show the evidence of herding. During 2008 global financial crisis, the whole Vietnam stock market shows the marginally significant evidence of herding, which is not consistent with Vo and Phan (2017). Herding is found to be weaker during the crisis period than normal periods. At sectorial level, a half of 20 industries show the evidence of herding during
6. 476 HỘI THẢO KHOA HỌC QUỐC TẾ KHỞI NGHIỆP ĐỔI MỚI SÁNG TẠO QUỐC GIA the crisis and 8 out of those industries shows the weaker herding during the crisis period than normal period. One limitation of the study is that it overlooksstocks listing on Vietnam stock exchanges from 2007, so the sample may not represent for the whole population. I suggest that future study should include those stocks to examine for the presence of herding. Future study can also examine herding under different extreme market conditions from the study such as at 1% or 10% up and down tail of market return distribution. Table 1: Descriptive statistics Variable Obs Mean SD Min Max ADF test CSAD (the whole market) 2741 1.83 0.49 0.40 4.42 -10.104 CSADJ (Real estate) 2660 1.55 0.95 0.03 7.44 -12.829 CSADj (Rubber) 2681 1.47 1.12 0.00 7.54 -13.02 CSADj (Securities) 2621 1.51 1.17 0.00 8.35 -11.724 CSADj (Information technology) 2682 2.07 0.99 0.03 12.00 -14.249 CSADj (Oil and gas) 2682 1.65 1.13 0.02 14.32 -14.868 CSADj (Tourism) 2741 1.95 1.36 0.00 18.12 -12.916 CSADj (Pharmacy, health and Chemical) 2740 1.56 1.02 0.01 8.20 -13.971 CSADj (Education) 2741 2.10 1.05 0.01 8.49 -11.319 CSADj (Mineral) 2682 1.75 1.08 0.03 8.61 -12.372 CSADj (Electricity, oil and gas) 2682 1.34 0.66 0.11 5.33 -12.029 CSADj (Banking and insurance) 2741 1.57 0.94 0.00 10.75 -13.295 CSADj (Plastics and packing) 2741 1.95 0.79 0.15 5.71 -13.483 CSADj (Manufacturing and business) 2741 1.85 0.69 0.10 5.79 -12.05 CSADj (Steel) 2713 1.60 0.96 0.01 10.08 -13.618 CSADj (Food) 2741 1.88 0.71 0.21 6.04 -13.224 CSADj (Commerce) 2681 1.68 0.94 0.02 6.66 -13.928 CSADj (Aquiculture) 2741 1.81 0.91 0.02 10.30 -12.996 CSADj (Logistics) 2741 2.11 0.75 0.32 6.99 -11.631 CSADj (Construting material) 2741 2.26 0.79 0.22 10.62 -10.483 CSADj (Constrution) 2741 2.24 1.11 0.02 18.15 -14.223 Rm 2741 -0.02 1.47 -7.02 6.23 -16.983 |Rm| 2741 0.98 1.09 0.00 7.02 -8.660 R2m 2741 2.15 4.87 0.00 49.28 -10.013 : means that null hypothesisof stationary is rejected at 1%
7. INTERNATIONAL CONFERENCE STARTUP AND INNOVATION NATION 477 Table 2: Herding behavior of the whole, up and down market Both states Up market Down market g0 g1 g2 g0 g1 g3 g0 g1 γ4 The whole 1.61 0.37 -0.07 1.59 0.42 -0.08 1.63 0.34 -0.06 market (107.6) (12.36) (9) (76.83) (9.99) (7.39) (77.38) (8.28) (6.02) 1.36 0.41 -0.09 1.35 0.45 -0.11 1.36 0.37 -0.08 Real estate (42.40) (7.17) (6.29) (29.72) (5.15) (4.24) (29.35) (4.79) (4.43) 1.21 0.51 -0.11 1.21 0.51 -0.11 1.21 0.51 -0.10 Rubber (33.05) (9.32) (8.78) (23.27) (6.22) (5.53) (23.11) (6.77) (6.64) 1.03 0.81 -0.14 1.0 0.88 -0.17 1.0 0.76 -0.13 Securities (31.09) (12.88) (9.52) (22.65) (9.94) (7.54) (20.95) (8.36) (6.08) Information 1.93 0.31 -0.08 1.93 0.32 -0.08 1.93 0.31 -0.07 technology (56.8) (6.41) (7.15) (40.7) (4.51) (4.90) (38.96) (4.55) (5.14) 1.38 0.43 -0.07 1.33 0.55 -0.10 1.45 0.32 -0.05 Oil, gas (40.12) (7.46) (5.02) (27.92) (6.46) (4.73) (28.46) (4.03) (2.44) 1.68 0.43 -0.07 1.65 0.53 -0.11 1.7 0.38 -0.05 Tourism (38.39) (6.94) (5.12) (27.32) (5.69) (5.25) (26.74) (4.68) (2.89) Pharmacy, health, 1.30 0.38 -0.05 1.31 0.34 -0.04* 1.29 0.4 -0.05 Chemical (42.32) (8.28) (4.41) (29.73) (4.68) (1.85) (29.56) (6.58) (4.21) 1.68 0.61 -0.08 1.62 0.77 -0.13 1.73 0.49 -0.05 Education (51.04) (12.28) (7.13) (35.42) (10.4) (7.29) (36.26) (7.54) (3.51) 1.55 0.38 -0.08 1.54 0.35 -0.08 1.58 0.39 -0.09 Mineral (44.98) (7.30) (7.23) (33.57) (4.87) (4.49) (29.95) (5.23) (5.49) Electricity, oil, 1.16 0.32 -0.06 1.15 0.33 -0.06 1.18 0.31 -0.06 gas (53.80) (8.44) (6.32) (36.55) (5.45) (3.68) (38.47) (6.09) (5.06) Banking, 1.34 0.34 -0.05 1.34 0.32 -0.05 1.34 0.35 -0.05 insurance (46.35) (7.45) (4.14) (34) (4.98) (2.57) (30.96) (5.43) (3.2) Plastics, 1.77 0.35 -0.08 1.72 0.44 -0.11 1.81 0.3 -0.06 packing (68) (8.53) (7.83) (47.6) (7.31) (7.1) (48.42) (5.37) (4.78) Business 1.67 0.34 -0.07 1.7 0.36 -0.09 1.62 0.36 -0.07 Manufacturing (75.66) (9.17) (8.14) (55.42) (6.7) (6.27) (51.12) (6.88) (5.83) 1.40 0.4 -0.09 1.4 0.43 -0.11 1.4 0.38 -0.08 Steel (44.78) (7.91) (7.34) (31.51) (5.89) (5.93) (31.59) (5.81) (5.13) 1.61 0.43 -0.07 1.63 0.43 -0.08 1.58 0.45 -0.07 Food (69.01) (9.93) (6.28) (51.82) (7.36) (5) (47.47) (7.53) (4.53) 1.53 0.34 -0.09 1.53 0.39 -0.1 1.53 0.31 -0.08 Commerce (48.73) (7.53) (8.43) (34.48) (5.97) (6.56) (34.03) (4.83) (5.48) 1.66 0.32 -0.07 1.65 0.33 -0.08 1.66 0.3 -0.07 Aquiculture (56.4) (6.67) (6.46) (42.43) (5.3) (4.62) (37.97) (4.39) (4.4) 1.93 0.32 -0.06 1.93 0.33 -0.07 1.92 0.32 -0.06 Logistics (78.88) (7.42) (5.48) (59.62) (5.94) (4.55) (53.44) (5.19) (3.74) Constructing 2 0.40 -0.07 2.02 0.43 -0.08 2.02 0.37 -0.06 material (83.93) (9.91) (7.45) (58.04) (7.03) (5.23) (59.65) (7.16) (5.38) 2.26 -0.01 -0.01 2.24 -0.01 -0.01 2.29 -0.01 -0.01 Construction (59.31) (0.09) (0.66) (41.61) (0.07) (0.23) (41.84) (0.18) (0.56) Note: T-statistics in parenthesis. , : means that the coefficients are significant at 1% and 5%.Both states, up market and down market are empirical results of the models 1, 2 and 3below respectively. Newey-west (1987) standard errors is applied to counter heteroskedasticity and autocorrelation problems. 2 CSADj,t = g0 + g1|Rmt| + g2R mt + ut (1) 2 CSADj,t = g0 + g1 |Rmt| + g3R mt + ut , if Rm > 0 (2) 2 CSADj,t = g0 + g1 |Rmt| + γ4 R mt + ut , if Rm < 0 (3)
8. 478 HỘI THẢO KHOA HỌC QUỐC TẾ KHỞI NGHIỆP ĐỔI MỚI SÁNG TẠO QUỐC GIA Table 3: Herding behavior of extreme up and down market Extreme up market Extreme down market g0 g1 γ5 g0 g1 γ6 3.13 -0.52 0.04 4.47 -1.13 0.12 The whole market (3.46) (1.11) (0.78) (6.73) (3.55) (3.21) 5.17 -1.8 0.19 5.32 -1.8 0.19* Real estate (3.29) (2.03) (1.60) (3.05) (2) (1.69) 2.73* -0.42 0.02 4.88 -1.17 0.08 Rubber (1.76) (0.5) (0.15) (2.67) (1.3) (0.75) 4.40 -1.25* 0.13 5.69 -1.8 0.19 Securities (3.17) (1.8) (1.6) (2.62) (1.65) (1.5) 3.58 -0.75 0.07 4.29 -0.94 0.08 Information technology (2.87) (1.14) (0.84) (3.09) (1.37) (1.01) 3.78 -0.94 0.10 3.33 -0.62 0.07 Oil and gas (2.17) (1.06) (1.01) (2.05) (0.73) (0.63) 3.61 -0.57 0.03 2.17 0.19 -0.03 Tourism (2.19) (0.69) (0.34) (1.16) (0.21) (0.764) Pharmacy, health, -0.18 1.07 -0.12 0.19 1.03 -0.13 Chemical (0.1) (1.07) (0.99) (0.12) (1.26) (1.41) 4.36 -0.77 0.07 2.61 -0.04 0.02 Education (2.68) (0.93) (0.7) (2.16) (0.07) (0.27) 4.02 -1.05 0.10 3.73 -0.85 0.07 Mineral (3.02) (1.52) (1.25) (2.92) (1.33) (1) 3.92 -1.04* 0.09 4.15 -1.32* 0.14 Electricity, oil, gas (3.24) (1.67) (1.27) (3.34) (1.92) (1.56) 4.16 -1.22 0.15 4.55 -1.28 0.14 Banking, insurance (2.63) (1.45) (1.45) (3.55) (2.09) (2.1) 4.89 -1.37 0.13 4.98 -1.25 0.12 Plastics, packing (5.3) (2.87) (2.19) (5.02) (2.52) (2.11 4.11 -1.04 0.10 4.26 -0.92* 0.08 business Manufacturing (4.04) (2.02) (1.6) (4.08) (1.81) (1.34) 4.42 -1.34* 0.13 5.12 -1.55 0.15 Steel (3.45) (1.95) (1.5) (3.97) (2.5) (2.23) 2.98 -0.43 0.04 4.73 -1.13* 0.12 Food (2.02) (0.54) (0.38) (4) (1.85) (1.55) 4.234 -1.18 0.11* 4.04 -0.93 0.07 Commerce (3.79) (2.13) (1.66) (3.14) (1.41) (0.85) 2.7 -0.27 0.01 5.40 -1.69 0.17 Aquiculture (1.51) (0.3) (0.04) (4.38) (2.83) (2.42) 2.3* -0.06 -0.01 6.13 -1.86 0.2 Logistics (1.78) (0.09) (0.02) (6.85) (4.13) (3.8) 2.81 -0.18 0.02 4.54 -0.96 0.1 Constructing material (2.21) (0.27) (0.17) (5.45) (2.42) (2.26) 0.96 0.48 -0.04 3.65 -0.78 0.09 Construction (0.65) (0.62) (0.52) (3.01) (1.31) (1.28) Note: T-statistics in parenthesis. , : means that the coefficients are significant at 1% and 5%. Extreme up and down market demonstrate empirical results of model 4 and 5below respectively. Newey-west (1987) standard errors is applied to counter heteroskedasticity and autocorrelation problems.
9. INTERNATIONAL CONFERENCE STARTUP AND INNOVATION NATION 479 extreme up 2 extreme up CSADj,t = g0 + g1|R mt| + γ5R mt + ut, Rmt in 5% up tail of distribution (4) extreme down 2 extreme down CSADj,t = g0 + g1|R mt| + γ6R mt + ut, Rmt in 5% down tail of distribution (5) Table 4: Herding behavior during 2008 global financial crisis Extreme up market g0 g1 g2 γ7 γ8 1.6 0.11* 0.43 -0.02* -0.07 The whole market (99.64) (1.75) (12.61) (1.82) (7) 1.34 0.3 0.46 -0.07 -0.11 Real estate (36.57) (3.09) (5.66) (3.37) (4.03) 1.2 0.38 0.52 -0.09 -0.1 Rubber (31.72) (3.6) (8.14) 94.14) (5.62) 0.1 0.61 0.94 -0.09 -0.18 Securities (28.9) (5.91) (13.28) (4.12) (10.01) 1.9 -0.11 0.42 0.01 -0.09 Information technology (53.06) (1.48) (7.21) (0.03) (5.6) 1.36 0.25 0.51 -0.03 -0.09 Oil and gas (38.59) (2.66) (8.02) (1.4) (5.54) 1.66 0.09 0.51 -0.01 -0.08 Tourism (36.66) (1.15) (6.9) (0.73) (4.04) 1.29 0.27 0.37 -0.04 -0.04 Pharmacy, health, Chemical (41.46) (3.57) (7.26) (2.26) (2.58) 1.66 0.35 0.68 -0.03* -0.09 Education (48.53) (4.05) (12.01) (1.84) (6.04) 1.53 0.22 0.47 -0.05 -0.11 Mineral (43.03) (2.31) (7.89) (2.38) (6.99) 1.15 0.16 0.36 -0.03 -0.07 Electricity, oil, gas (46) (2.55) (6.55) (2.59) (3.7) 1.33 0.27 0.36 -0.03 -0.05 Banking, insurance (44.57) (3.29) (7.26) (1.59) (3.84) 1.74 -0.01 0.47 -0.01 -0.1 Plastics, packing (62.2) (0.08) (9.05) (0.35) (6.63) 1.64 -0.04 0.43 -0.01 -0.08 business Manufacturing (72.45) (0.59) (10.54) (0.36) (6.99) 1.38 0.17 0.47 -0.04 -0.11 Steel (40.98) (2.29) (7.59) (2.52) (6) 1.6 0.18 0.48 -0.03 -0.07 Food (62.76) (2.43) (8.77) (2.12) (4.14) 1.5 -0.03 0.45 -0.02 -0.1 Commerce (45.24) (0.38) (8.06) (1.21) (6.65) 1.63 -0.09 0.41 -0.01 -0.08 Aquiculture (52.84) (1.33) (7.08) (0.34) (4.77) 1.91 -0.02 0.38 -0.01 -0.06 Logistics (74.53) (0.26) (7.53) (0.46) (3.85) 1.2 0.06 0.5 -0.01 -0.08 Constructing material (77.69) (0.87) (10.28) (0.43) (6.21) 2.24 -0.33 0.08 0.05 -0.02 Construction (56.48) (4.49) (1.23) (3.4) (0.94) Note: T-statistics in parenthesis. , ,*: means that the coefficients are significant at 1%, 5% and 10%.
10. 480 HỘI THẢO KHOA HỌC QUỐC TẾ KHỞI NGHIỆP ĐỔI MỚI SÁNG TẠO QUỐC GIA The table demonstrates empirical results of model 6 below. Newey-west (1987) standard errors is applied to counter heteroskedasticity and autocorrelation problems. 2 2 CSADj,t = g0 + g1 D1* |Rmt| +g2(1 –D1)* |Rmt| + γ7D1*R mt + γ8(1 –D1)*R mt + ut (6) REFERENCES: Bikhchandani, S. and Sharma, S.(2000). Herd behavior in financial markets. IMF Staff Papers, pp. 279 –310. Bui, N. D., Nguyen, T. B. and Nguyen, T. T. (2015). Herd behavior in Southeast Asian stockmarkets – an empirical investigation.Acta Oeconomica, 65(3), pp. 413–429. Bui, N. D., Nguyen, T. B., Nguyen, T. T. and Titman, G. F. (2017). Herding in frontier stock markets: evidence from the Vietnamese stock market. Accounting and finance. Chang, C., Cheng, W. and Khorana, A. (2000). An examination of herd behavior in equity markets: An international perspective. Journal of banking and finance, 24, pp. 1651- 1679. Chiang and Zheng, D. (2010). An empirical analysis of herd behavior in global stock market. J. Bank. Finance 30, pp. 2471–2488. Christie, G. and Huang, D. (1995). Following the Pied Piper: Do Individual Returns Herd around the Market? Financial Analysts Journal, 51(4), pp. 31-37. Economou, F., Katsikas, E. and Vickers, G. (2016). Testing for herding in the Athens Stock Exchange during the crisis period. Finance Research Letters, 18, pp.334-341. Gebka, B. and Wohar, E. (2013). International herding: Does it differ across sectors? Journal of International Financial Markets, Institutions & Money, 23, pp.55-84. Hwang, S. and Salmon, M. (2001). A new measure of herding and empirical evidence. Working paper series WP01-02. Financial econometrics research center. Le, U. and Truong, H. (2014). An Exploratory Study of Herd Behavior in Vietnamese Stock Market: A New Method. Asian Journal of Finance & Accounting, 6(1). Litimi, H., BenSaùda, A. and Bouraoui, O. (2016). Herding and excessive risk in the American stock market: A sectoral analysis. Research in International Business and Finance, 38, pp.6-21. Phan, K and Zhou, J. (2014). Vietnamese Individual Investors’ Behavior in the Stock Market:An Exploratory Study. Research journal of social science and management, 3(2). Shah, M., Shah, A. and Khan, S. (2017). Herding behavior in the Pakistan stock exchange: Some new insights. Research in International Business and Finance, 42, pp.865-873. Tan, L., Chiang, T.C., Mason, J.R. and Nelling, E. (2008). Herding behavior in Chinese stock markets: An examination of A and B shares. Pac.-Basin Finance J. 16 (1–2), pp.61–77. Vassilakopoulos, P. (2014).Herding in financial markets:evidence from the Athens stock exchange. Journal of Computational Optimization in Economics and Finance, 6(3), pp. 279-309. Vo, V. and Phan, D. (2016). Herd Behavior in Emerging Equity Markets:Evidence from Vietnam. Asian Journal of Law and Economics, 7(3), pp.369–383. Vo, V. and Phan, D. (2016). Further evidence on the herd behavior in Vietnam stock market. Journal of Behavioral and Experimental Finance, 13, pp.33–41.