The Lunar New Year Effect on Stock Market Returns: Evidence from Ho Chi Minh Stock Exchange

Abstract

This study is devoted to investigating the Lunar New Year effect on market returns for the Ho Chi Minh Stock Exchange (HOSE). The data employed in this study include a daily series of the VN30-Index, which is a market capitalization weighted index of 30 large capitalization and high liquidity stocks traded on the HOSE, for the period from 6 February 2012 to 31 December 2024. The empirical findings derived from ordinary least squares (OLS), exponential-generalized autoregressive conditional heteroskedasticity [EGARCH(1,1)] regression models consistently confirm that the average return in the last two days and five days before the Lunar New Year are significantly higher than the average market returns on other days of the year. However, this study finds that the average return during the first two trading days and five trading days following the Lunar New Year are not significantly different from the average market returns on other days throughout the year.

1. Introduction

The Lunar New Year (LNY) is not only the biggest festival in the culture of several East Asian countries; it also has a profound impact on the psychology and behavior of investors in stock markets. Several studies have shown that due to optimistic sentiment, individual investors often engage in more trading during the early days of the LNY, leading to higher stock prices on those days (Bergsma & Jiang, 2016; Teng & Yang, 2018; Huang et al., 2022). Additionally, before the LNY, investors typically buy stocks to take advantage of prices increases at the beginning of the following year. As a result, stock prices tend to rise in the final days of the LNY. This increase in stock prices around the LNY is referred to as the LNY effect. This effect has been observed in the stock markets of China, Hong Kong, Taiwan, South Korea, Japan, Thailand, Malaysia, and Singapore (Yen et al., 2001; Wu, 2013; Yuan & Gupta, 2014; Chia et al., 2015; Bergsma & Jiang, 2016; Teng & Yang, 2018; Cui, 2024).

Like several countries in East Asia, the LNY (Tet Nguyen Dan in Vietnamese) plays a crucial role in Vietnamese culture, and it is the longest holiday of the year according to Vietnamese Labor Law. Therefore, the LNY effect can exist in the Vietnam stock market. Although the LNY effect has been documented in many stock markets in East Asian countries, very little has been known about this effect for the HOSE, a frontier market. Specifically, Khanh et al. (2020) examined the LNY effect on market returns for the Ho Chi Minh City Stock Exchange (HOSE) and found that the average return in the last 5 trading days before the LNY is higher than the average market return in the first 5 trading days following the LNY. However, this study does not provide statistical values to draw statistically significant conclusions, meaning that it cannot be concluded whether the LNY effect exists on the HOSE or not. In addition, Truong and Friday (2021b) investigated the effect of the LNY on the January anomaly for the HOSE and found that the January effect is present on the market for the entire studied period, but it disappeared in years when the LNY fell in January, suggesting that the LNY has a significant impact on the January anomaly in the Vietnam stock market. Moreover, Truong and Friday (2021a) examined the influence of the introduction of the VN30-Index futures contract on the daily returns anomaly for the HOSE. This study reported that the day-of-the-week effect is only present for the pre-index futures period, not for the post-index futures period. Therefore, this study is devoted to filling this gap in the literature by investigating the LNY effect for the HOSE. This study contributes to the existing literature in the following aspects. First, the HOSE offers fertile ground for exploring the LNY effect on market return due to the fact that it has been characterized by the predominance of individual investors who often rely on emotions rather than strict analysis, leading to decisions driven by fear, greed, or excitement (Truong et al., 2022). Therefore, it is expected that the LNY effect on stock returns will be more pronounced in the HOSE. By analyzing stock performance around the LNY, this study offers insights into how cultural events can influence market dynamics. Second, this study enhances our understanding of investor sentiment and behavior during festive periods, bridging gaps in the literature regarding seasonal anomalies in frontier markets. Finally, the findings of this study provide practical implications for investors in establishing a trading strategy in order to earn abnormal returns around the LNY.

The remainder of the paper is structured as follows. Section 2 summarizes the empirical findings of previous studies regarding the LNY effect. The data employed in the study and the research methodology are presented in Section 3. The empirical results of the study are reported in Section 4. Finally, conclusions of the study are presented in Section 5.

2. Literature Review

Although the calendar effect on stock returns has been extensively documented in the financial literature over the last several decades, the LNY effect has not been explored extensively. It is found that some studies investigated the LNY effect on stock market returns in East Asian countries. Specifically, Yen et al. (2001) tested the hypothesis of the LNY effect for six stock markets, including Hong Kong, Japan, South Korea, Malaysia, Singapore, and Taiwan for the period from 1991 to 2000. The researchers found that market returns tend to increase in the 15 days before and after the LNY across all markets. In other words, the LNY effect exists in all studied stock markets. Similarly, Yuan and Gupta (2014) explored the LNY effect for six stock markets in East Asia, namely China, Hong Kong, Japan, Malaysia, South Korea, and Taiwan. The results derived from the GARCH(1,1) model indicate that market returns in 3 days before the LNY are higher than on other days of the year in all studied stock markets. Moreover, this study found that market volatility in 3 days before the LNY is higher than other days of the year for the Chinese stock market, but there was no differences for the remaining stock markets studied. In addition, Chia et al. (2015) examined the existence of the LNY effect on the Hong Kong stock market from January 1988 to July 2012. Using GARCH-M, TGARCH-M, and EGARCH-M models, the researchers found that the average market returns in the two days before and one day after the LNY are higher than the average market returns on other days of the year. This study reported that the market returns’ volatility in the days following the LNY is higher than in the days before the LNY. According to the researchers, these findings can be explained by arguments derived from behavioral finance, where traditional culture and Chinese beliefs may shape investors’ risk attitudes and influence their decision making in stock trading. Moreover, Teng and Yang (2018) investigated the LNY effect for the Shanghai and Shenzhen stock exchanges during the period from 1993 to 2015 and found that market returns in the three sessions before and one session after the LNY are higher than on other days of the year. In a recent study, Cui (2024) re-examined the LNY effect on the Shanghai and Shenzhen stock exchanges. The results derived from GARCH-M and EGARCH-M models confirmed that market returns in the two trading days, three trading days before, and two trading days after the LNY are higher than the average returns on other days of the year for the Shanghai Stock Exchange. However, the LNY effect did not exist on the Shenzhen Stock Exchange. Using a GARCH model, Zhang et al. (2008) also found no evidence of the LNY effect on the Chinese stock market. In another aspect, Wu (2013) explored the effect of the LNY on stock returns of Chinese companies listed on the U.S. stock market during the period from 1993 to 2011. This study documented that the returns of stocks in the five days preceding the LNY are higher than on other days of the year by 0.226%, while the returns in the five days following the LNY were lower by 0.032%. Furthermore, Lai et al. (2024) investigated the Chinese National holiday effect, including LNY effect, on the Chinese stock market. The researchers found that market returns in the trading day before the national holiday are higher than the average returns on other days of the year.

Regarding the Taiwanese stock market, Huang et al. (2022) investigated the influence of culture on investors’ behavior by examining whether individual investors trade more aggressively during the early days of the LNY. Using a sample of 129,397 observations collected from 854 branches of 63 securities companies from January 2013 to December 2016, they found that individual investors trade stocks more than usual during the early days of the LNY due to feelings of happiness. Moreover, this study affirmed that individual investors incurred losses when buying and holding stocks for 1 to 5 days after the LNY. However, the findings of the study indicate that the LNY effect did not exist for the trading behavior of institutional investors. Specifically, Chien and Chen (2007) measured the impact of the LNY on the January effect for the Taiwan’s stock market from 1971 to 2004 and found that the January effect exists only when the LNY falls in February. Based on empirical evidence, the researchers concluded that culture plays a significant role in adjusting the seasonal behavior of investors in the Taiwan’s stock market.

In a broader context, Bergsma and Jiang (2016) investigated the New Year effect on 11 stock markets across six different cultures in which New Year holidays do not fall on 1 January. This study documented that the market returns around a cultural New Year are higher than the average returns on other days of the year. In addition, the researchers found that market returns in the first month of the New Year according to the calendar of the countries in the study are higher than the market returns in the remaining months of the year. They argued that the New Year effect is attributed to the optimistic sentiment of investors during New Year celebrations.

Regarding the Vietnam stock market, Khanh et al. (2020) investigated the LNY effect on market returns for the HOSE. Using the VN-Index series for the period from 1 March 2002 to 31 December 2018, the researchers found that the average return in the last 5 trading days before the LNY is higher than the average market return in the first 5 trading days after the LNY. However, this study has the limitation of not providing statistical values to draw conclusions about the statistically significant differences. In addition, Truong and Friday (2021b) examined the effect of the LNY on the January anomaly for the HOSE during the period from 7 January 2009 to 26 December 2018. The empirical findings obtained from OLS and GARCH(1,1) revealed that the January effect is present on the market for the entire studied period. However, the January effect disappeared in years when the LNY fell in January. These findings suggest that the LNY has a significant impact on the January anomaly in the Vietnamese stock market. Based on the findings, the researchers concluded that the LNY influences the January effect on the Ho Chi Minh stock exchange.

In conclusion, the LNY effect has been found in all stock markets in countries with a tradition of celebrating the LNY. Specifically, market returns on the days before and after the LNY are higher than on other days of the year. The LNY effect can be explained by the optimism that alters investors’ attitudes toward risk and influences their decision making in stock trading during the days before and after the LNY. However, very little is known about this effect for the HOSE, a frontier market. Therefore, this study is devoted to filling the literature gap by exploring the LNY effect on the market returns for the HOSE.

3. Data and Research Methodology

The data employed in this study are primarily comprised of the daily VN30-Index series for the period from 6 February 2012 (the date the VN30-Index was officially launched) to 31 December 2024, obtained frominvesting.com (www.investing.com, accessed on 22 April 2025). Then, a natural logarithmic transformation is applied for the data to produce a time series of continuously compounded returns. Specifically, the market returns are computed using the following equation:

$$R_t=\log (I_t)-\log (I_{t-1})=\log (I_t/I_{t-1})$$

where

-

R_t is the market return of trading day t;

-

I_t is the VN30-Index at the closing day t;

-

I_t−1 is the VN30-Index at the closing day t − 1.

To test for the presence of the LNY effect on market returns for the HOSE, the ordinary least-squares (OLS) regression is first used in this study. Specifically, the model takes the following forms:

$$R_t={\alpha }_0+{\alpha }_{1}PRE{2}_t+{\alpha }_{2}POST{2}_t+{\beta }_1{D}_{1t}+{\beta }_2{D}_{2t}+{\beta }_3{D}_{3t}+{\beta }_4{D}_{4t}+{\epsilon }_t$$

(1)

$$R_t={\alpha }_0+{\varphi }_{1}PRE{5}_t+{\varphi }_{2}POST{5}_t+{\beta }_1{D}_{1t}+{\beta }_2{D}_{2t}+{\beta }_3{D}_{3t}+{\beta }_4{D}_{4t}+{\epsilon }_t$$

(2)

where

-

PRE2_t is a dummy variable, taking the value of 1 if observation t falls within the last 2 trading days before the LNY, and 0 otherwise.

-

POST2_t is a dummy variable, taking the value of 1 if observation t falls within the first 2 trading days following the LNY, and 0 otherwise.

-

PRE5_t is a dummy variable, taking the value of 1 if observation t falls within the last 5 trading days before the LNY, and 0 otherwise.

-

POST5_t is a dummy variable, taking the value of 1 if observation t falls within the first 5 trading days following the LNY, and 0 otherwise.

-

D_1t, D_2t, D_3t and D_4t are dummy variables for Monday, Tuesday, Thursday and Friday, respectively (i.e., D_1t takes the value of 1 if observation t occurs on Monday and 0 otherwise).

The selection of the PRE2 and POST2 variables to measure the LNY effect on the HOSE is based on Vietnam’s current stock trading regulation, which follows a T + 2 settlement system. This means that transactions (buying and selling) executed on day T are only completed after 2 business days, allowing the seller to receive payment and the buyer to receive stocks. Therefore, if a transaction occurs on the last two days of the lunar year, the money and stocks will only be available in investors’ accounts during the first two trading days after the LNY. Given this trading characteristic, market returns during the last two trading days before the LNY break may differ from those on other days. In addition, this study explores the LNY effect through the PRE5 and POST5 variables in order to broaden the scope of the study and enhance the robustness of the findings. The selection of these windows aligns with previous studies of Wu (2013) and Huang et al. (2022).

It is important to note that the assumption of constant variance of residuals over time in the OLS model could not hold for time-series data in finance. According to Brooks (2002), if this assumption is violated and the OLS model is still used, the standard errors may be incorrect, leading to potentially biased conclusions. To address this issue, Engle (1982) developed the ARCH (Autoregressive Conditional Heteroscedasticity) model that allows the variance of residuals to change over time as a function of past residuals. Then, Bollerslev (1986) generalized the ARCH model into the GARCH (Generalized Autoregressive Conditional Heteroscedasticity) model, which permits conditional variance to depend on its own previous lags. Therefore, if the heteroscedasticity exists in the OLS model, the GARCH model is considered more appropriate than the OLS model.

Although the standard GARCH model addresses the issue of heteroscedasticity in regression models, it fails to consider the asymmetry in shocks that frequently arise in financial time-series data. To address this issue, Nelson (1991) introduced the EGARCH (exponential GARCH) model to measure the asymmetric volatility of a financial asset in response to positive and negative shocks (leverage effects). In addition, unlike TGARCH (threshold GARCH) and GJR-GARCH (Glosten–Jagannathan–Runkle GARCH), EGARCH does not require parameters to be constrained to ensure the non-negativity of the variance, which simplifies estimation and interpretation. Furthermore, EGARCH can accommodate persistent volatility clustering, making it suitable for financial time-series data. Therefore, this study employs the EGARCH(1,1) model to address the issue of heteroskedasticity if it exists in the model and the asymmetric effect of shocks on the market return volatility. Specifically, the EGARCH(1,1) takes the following forms:

$$\begin{gathered} R_t={\alpha }_0+{\alpha }_{1}PRE{2}_t+{\alpha }_{2}POST{2}_t+{\beta }_1{D}_{1t}+{\beta }_2{D}_{2t}+{\beta }_3{D}_{3t}+{\beta }_4{D}_{4t}+{\epsilon }_t \\ \ln ({\sigma }_t^2)=\omega +\delta \ln ({\sigma }_{t-1}^2)+\gamma \left[{\displaystyle \frac{\left|{\epsilon }_{t-1}\right|}{\sqrt{{\sigma }_{t-1}^2}}}-\sqrt{{\displaystyle \frac{2}{\pi }}}\right]+\phi {\displaystyle \frac{{\epsilon }_{t-1}}{\sqrt{{\sigma }_{t-1}^2}}} \end{gathered}$$

(3)

$$\begin{gathered} R_t={\alpha }_0+{\varphi }_{1}PRE{5}_t+{\varphi }_{2}POST{5}_t+{\beta }_1{D}_{1t}+{\beta }_2{D}_{2t}+{\beta }_3{D}_{3t}+{\beta }_4{D}_{4t}+{\epsilon }_t \\ \ln ({\sigma }_t^2)=\omega +\delta \ln ({\sigma }_{t-1}^2)+\gamma \left[{\displaystyle \frac{\left|{\epsilon }_{t-1}\right|}{\sqrt{{\sigma }_{t-1}^2}}}-\sqrt{{\displaystyle \frac{2}{\pi }}}\right]+\phi {\displaystyle \frac{{\epsilon }_{t-1}}{\sqrt{{\sigma }_{t-1}^2}}} \end{gathered}$$

(4)

where

-

$\gamma$ indicates the impact of past errors on the current volatility (ARCH effect).

-

$\delta$ represents the effect of past volatility on current volatility (GARCH effect).

-

$\phi$ indicates the asymmetric effects of positive and negative shocks on volatility (leverage effect).

4. Empirical Results

4.1. Descriptive Statistics of the Sample

Based on the collected data, descriptive statistics of market returns for the entire research period, 2 and 5 days before and after the LNY, are computed and presented in

Table 1. Summary statistics of the market returns.

Variables	Observation	Minimum	Mean	Maximum	Standard Deviation
R (entire)	3205	−0.0302	0.0001	0.0225	0.0052
PRE2	24	−0.0162	0.0031	0.0171	0.0064
POST2	24	−0.0137	0.0006	0.0104	0.0050
PRE5	60	−0.0167	0.0023	0.0171	0.0062
POST5	60	−0.0170	0.0002	0.0104	0.0052

. Table 1 shows that the average daily return on the HOSE during the studied period is 0.01 percent, ranging from −3.02 percent to 2.25 percent. In addition, the statistics shown in Table 1 indicate that the average return in the last two days (PRE2) and five days (PRE5) before the LNYare 0.31 percent and 0.23 percent, respectively, significantly higher than the average market returns for the entire studied period. However, statistical analyses reveal that there is no significant difference between the average market return for the entire studied period and the average returns for the 2-day and 5-day periods following the LNY.

4.2. Regression Results

To ensure the reliability of the estimated results, the augmented Dickey–Fuller (ADF) test is performed to examine the stationary of the studied variables. The results of the ADF test presented in

Table 2. Results of the ADF test.

Variables	Without trend	With trend
R (market returns)
Level	−54.77 ***	−54.77 ***
PRE2
Level	−13.80 ***	−13.80 ***
POST2
Level	−13.80 ***	−13.80 ***
PRE5
Level	−13.34 ***	−13.34 ***
POST5
Level	−13.34 ***	−13.34 ***

*** represents statistical significance at 1% level.

confirm that all variables are stationary at the level (the order of integration is zero).

Using Eviews software, version 10, the results of the OLS model summarized in Table 1 indicate that the average return in the last two days (PRE2) and five days (PRE5) before the LNY are significantly higher than the average market returns on other days of the year. However, the findings shown in

Table 3. Results of the OLS model.

Variable	Model 1		Model 2
	Coefficient	t-Statistic	Coefficient	t-Statistic
${\alpha }_0$ (constant)	0.000717	3.53 ***	0.00069	3.38 ***
${\alpha }_1$ (PRE2)	0.00227	2.15 **	-	-
${\alpha }_2$ (POST2)	0.00060	0.61	-	-
${\varphi }_1$ (PRE5)	-	-	0.00204	3.04 ***
${\varphi }_2$ (POST5)	-	-	0.00046	0.69
${\beta }_1$ (Monday)	−0.00134	−4.66 ***	−0.00135	−4.67 ***
${\beta }_2$ (Tuesday)	−0.00045	−1.56	−0.00044	−1.54
${\beta }_3$ (Thursday)	−0.00089	−3.12 ***	−0.00088	−3.09 ***
${\beta }_4$ (Friday)	−0.00027	−0.95	−0.00027	−0.92
ARCH-LM test (1 lag)	109.07 ***		108.10 ***

*** and ** represent statistical significance at the 1% and 5% levels, respectively.

confirm that the average return in the first two trading days (POST2) and five trading days (POST5) after the LNY are not significantly higher than the average market returns on other days of the year. On the basis of these findings, it can be concluded that the pre-LNY effect exists in the HOSE. However, this conclusion does not take into account the ARCH effect that is suspected to be present in the model. To test for the existence of ARCH effects, the Lagrange Multiplier approach, developed by Engle (1982), is employed. The results of ARCH-LM test presented in Table 3 confirm that ARCH effects are present in the OLS model. Due to the ARCH effect in the OLS models, the EGARCH models are employed in this study. In order to choose the appropriate EGARCH(p,q) model, some EGARCH(p,q) models (p = 1, 2, 3; q = 1, 2)were conducted. Based on the AIC, the EGARCH(1,1) is the most appropriate model for our study because its AIC value is lowest.

The results obtained from the EGARCH(1,1) models are summarized in

Table 4. Results of the EGARCH(1,1) model for the entire sample.

Variable	Model 3		Model 4
	Coefficient	z-Statistic	Coefficient	z-Statistic
Conditional mean equation
${\alpha }_0$ (constant)	0.00056	3.43 ***	0.00053	3.29 ***
${\alpha }_1$ (PRE2)	0.00256	2.37 **	-	-
${\alpha }_2$ (POST2)	0.00006	0.10	-	-
${\varphi }_1$ (PRE5)	-	-	0.00219	4.29 ***
${\varphi }_2$ (POST5)	-	-	0.00018	0.38
${\beta }_1$ (Monday)	−0.00084	−3.91 ***	−0.00082	−3.85 ***
${\beta }_2$ (Tuesday)	−0.00040	−1.78	−0.00038	−1.72 *
${\beta }_3$ (Thursday)	−0.00055	−2.54 **	−0.00055	−2.53 **
${\beta }_4$ (Friday)	−0.00024	−1.12	−0.00021	−0.95
Conditional variance equation
$\omega$	−0.51557	10.80 ***	−0.51940	−10.83 ***
$\gamma$ (ARCH effect)	0.20485	15.42 ***	0.20393	15.39 ***
$\delta$ (GARCH effect)	0.96619	231.96 ***	0.96580	231.13 ***
$\phi$ (Leverage effect)	−0.04729	−7.14 ***	−0.04809	−7.25 ***
ARCH-LM test (1 lag)	0.78		0.84

***, **, and * represent statistical significance at 1%, 5% and 10% levels, respectively.

. These findings consistently confirm that the pre-LNY effect exists in the HOSE. Specifically, it is observed from Table 4 that the average return in the last two days (PRE2) and five days (PRE5) before the LNY are higher than the average market returns on other days of the year by 0.26 percent and 0.22 percent, respectively. The differences in market returns are statistically significant at the five percent level. However, the results presented in Table 4 indicate that the average return during the first two trading days (POST2) and five trading days (POST5) following the LNY are not significantly different from the average market returns on other days throughout the year. These findings are consistent with previous findings of Yen et al. (2001), Yuan and Gupta (2014), Chia et al. (2015), Bergsma and Jiang (2016), Teng and Yang (2018), Lai et al. (2024), and Cui (2024). However, this evidence is partly different from the findings of Zhang et al. (2008), Huang et al. (2022), and Cui (2024). Specifically, Zhang et al. (2008) and Cui (2024) found no evidence of the LNY effect in the Chinese stock market.

The LNY effect on market returns in the HOSE can be attributed to several factors. First, it is important to note that the HOSE has been characterized by the predominance of individual investors. Economists and psychologists have demonstrated that individuals engage in bounded rationality when making decisions. The LNY is a major cultural event in Vietnam, often associated with renewal and prosperity. Therefore, the LNY can boost investors optimism and sentiment. Heightened optimism and positive sentiment can lead to increased buying activity, driving stock prices up. In addition, many individual investors prefer to start the new year with profitable positions, which can drive up stock prices as they trade more aggressively in the days leading up to the holiday in anticipation of positive market movements. Moreover, in the Vietnamese culture, many companies issue bonuses before the LNY holiday, providing investors with additional capital to invest in stocks. This influx of cash can drive up the demand and price of stocks. Finally, traders often engage in speculative buying in anticipation of price increases after the LNY, further contributing to rising stock prices.

Furthermore, to check the robustness of the results, the data are divided into two subsamples based on the time the VN30-Index futures contract officially started trading in Vietnam (10 August 2017). Specifically, the first subgroup is the pre-index futures period, from 6 February2002 to 9 August2017, while the other subgroup is the post-index futures period, from 10 August2017 to 31 December2024. The estimated results of the EGARCH(1,1) model for the subsamples shown in

Table 5. Results of the EGARCH(1,1) model for the pre-index future subsample.

Variable	Model 3		Model 4
	Coefficient	z-Statistic	Coefficient	z-Statistic
Conditional mean equation
${\alpha }_0$ (constant)	0.00050	2.44 **	0.00047	2.28 **
${\alpha }_1$ (PRE2)	0.00285	2.70 ***	-	-
${\alpha }_2$ (POST2)	−0.00055	−0.89	-	-
${\varphi }_1$ (PRE5)	-	-	0.00207	3.33 ***
${\varphi }_2$ (POST5)	-	-	0.00003	0.05
${\beta }_1$ (Monday)	−0.00078	−2.91 ***	−0.00073	−2.74 ***
${\beta }_2$ (Tuesday)	−0.00045	−1.64	−0.00044	−1.64
${\beta }_3$ (Thursday)	−0.00053	−1.94 *	−0.00052	−1.90 *
${\beta }_4$ (Friday)	−0.00006	−0.21	0.00001	0.02
Conditional variance equation
$\omega$	−0.53080	−7.95 ***	−0.53217	−7.98 ***
$\gamma$ (ARCH effect)	0.20134	11.26 ***	0.19897	11.31 ***
$\delta$ (GARCH effect)	0.96514	169.61 ***	0.96487	169.75 ***
$\phi$ (Leverage effect)	−0.04601	−5.45 ***	−0.04668	−5.53 ***

***, **, and * represent statistical significance at 1%, 5%, and 10% levels, respectively.

and

Table 6. Results of the EGARCH(1,1) model for the post-index future subsample.

Variable	Model 3		Model 4
	Coefficient	z-Statistic	Coefficient	z-Statistic
Conditional mean equation
${\alpha }_0$ (constant)	0.000599	2.67 ***	0.00057	2.57 ***
${\alpha }_1$ (PRE2)	0.00261	1.92 *	-	-
${\alpha }_2$ (POST2)	−0.00031	−0.41	-	-
${\varphi }_1$ (PRE5)	-	-	0.00310	4.39 ***
${\varphi }_2$ (POST5)	-	-	0.00011	0.15
${\beta }_1$ (Monday)	−0.00087	−2.97 ***	−0.00086	−2.95 ***
${\beta }_2$ (Tuesday)	−0.00042	−1.32	−0.00042	−1.35
${\beta }_3$ (Thursday)	−0.00069	−2.39 **	−0.00067	−2.38 **
${\beta }_4$ (Friday)	−0.00021	−0.68	−0.00017	−0.56
Conditional variance equation
$\omega$	−0.456697	−8.45 ***	−0.46221	−8.56 ***
$\gamma$ (ARCH effect)	0.193017	11.65 ***	0.19350	11.71 ***
$\delta$ (GARCH effect)	0.970159	199.79 ***	0.96972	199.76 ***
$\phi$ (Leverage effect)	−0.058865	−6.78 ***	−0.06177	−7.14 ***

***, **, and * represent statistical significance at 1%, 5% and 10% levels, respectively.

consistently confirm that the average return in the last two days (PRE2) and five days (PRE5) before the LNY are significantly higher than the average market returns on other days of the year.

Additionally, the results obtained from model 3 and model 4 consistently confirm that the day-of-the-week effect is present in the HOSE. Specifically, the average market returns for Monday and Thursday are negative and statistically significant at the five percent level. This evidence aligns with the previous findings of Truong and Friday (2021a) for the HOSE. Moreover, the results of the EGARCH(1,1) model confirm that the leverage effect on the market volatility exists in the HOSE. Specifically, the results presented in Table 4 show that the leverage effect coefficient is statistically negative at the one percent significance level, implying that negative market shocks lead to greater volatility than positive shocks of the same magnitude.

5. Conclusions

This study investigates the LNY effect on market returns in the HOSE. Using the daily VN30-Index data from 6 February 2012 to 31 December 2024, we find significant evidence of a pre-LNY effect, characterized by higher average returns in the final days leading up to the LNY. Specifically, market returns in the last two trading days and five trading days before the LNY are statistically higher than those on other days of the year. However, this effect does not extend to the days immediately following the holiday, where no significant differences in returns are observed. These findings align with previous studies conducted in various East Asian markets, reinforcing the notion that cultural factors significantly influence investor behavior and market dynamics. The optimism associated with the LNY appears to drive increased trading activity and speculative buying, contributing to price surges prior to the holiday. Additionally, this study findsthat the effect of shocks on the market return volatility is asymmetric for the HOSE. Specifically, negative market shocks lead to greater volatility than positive shocks of the same magnitude.

The contributions of this study are to enrich the literature on seasonal anomalies in frontier markets and highlight the importance of cultural events in shaping investor sentiment and market behavior. The insights gained from this study can aid investors in establishing a trading strategy in order to earn abnormal returns around the LNY. Specifically, investors can benefit from increased stock prices in the days leading up to the LNY. Understanding this seasonal trend can help investors make informed decisions about when to buy or sell stocks. Given that no significant price increases were observed in the days following the LNY, investors should be cautious about holding positions into this period. The potential for lower returns post-holiday suggests a strategy focused on short-term gains in the days leading up to the LNY.

Although this study has enriched the literature on seasonal anomalies in frontier markets, it still has several limitations that should be addressed in future empirical studies. First, this study does not take into account the factors that create the LNY effect on the HOSE, such as trading volume on the days before and after the LNY. In addition, this study only focuses on the LNY effect for the overall market, without examining this effect on specific groups of stocks. The LNY effect may differ between big and small stocks. These limitations could serve as interesting topics for future research.