Introduction
Purpose and contribution of research
This dissertation aims to evaluate the weak market efficiency hypothesis within the Vietnam Stock Exchange It focuses on objectives derived from market theory and statistical validation methods to investigate the level of information efficiency in Vietnam's stock market.
This article provides a comprehensive overview of the Vietnamese stock market, evaluating its effectiveness and highlighting key insights It also proposes strategic solutions aimed at fostering growth within Vietnam's stock market.
Limitation of the study
Although I have tried a lot, I still have some limitations as follows The thesis only stops at the weak market test and does not carry out effective semi-strong form and strong form Thesis focuses on researching data for the whole VN index, not for individual stocks.
Efficient Market Hypothesis Theory
Weak form efficiency
In a weak form efficient market, stock prices accurately reflect historical information, including past prices and trading volumes This concept posits that current prices incorporate all available past earnings and market data related to a stock Consequently, the hypothesis suggests that the returns on securities are uncorrelated with prior information, meaning investors cannot achieve abnormal returns by analyzing similar historical data Such efficient markets are commonly observed in developing or emerging economies.
Semi-strong form efficiency
Market theory operates effectively at the semi-strong form, positing that stock prices incorporate both past and current public information, such as stock prices, profitability, and trading volume Following the release of new information, stock prices adjust immediately, indicating that in a semi-strong market, price changes are unpredictable Consequently, investors acting on new information cannot achieve abnormal profits, as current prices already reflect all available public data This market behavior is typically observed in developed countries globally.
Strong form efficiency
The strong form of the efficient market hypothesis asserts that security prices fully reflect all relevant information, encompassing past, present, public, and even private data This theory integrates aspects of both weak and semi-strong market forms, ensuring that all investors have equal access to information, incur the same costs to obtain it, and face identical risks Consequently, no investor can leverage exclusive valuation-related insights to achieve excessive profits, reinforcing the notion of a truly efficient market.
Ngo Thi Minh Hoa, MSc in Finance, emphasizes that an effective strong-form market swiftly communicates information to investors, encompassing both public and internal data Consequently, for a market to be truly effective, it must assimilate all available information.
Testing models for EHM
EMH is an important financial theory in the foundation of modern financial theory Fama has identified three test models for EMH:
The Efficient Market Hypothesis (EMH) is rooted in The Fair Game theory, which serves as a framework for testing EMH in the stock market A simple illustration of this concept is a coin toss, where the probabilities of landing on heads or tails are equal In essence, The Fair Game represents a scenario where there is no systematic difference between actual and expected outcomes prior to the game Similarly, the stock market can be viewed as a fair game when there is no consistent discrepancy between actual and expected stock returns.
: Actual observation of stock at time t + 1
( ) : The expected price of stock i at time t + 1, in terms of market information aggregation is
: The difference between the actual price and the expected price of the stock
If the efficient market according to The Fair Game Theory, there is no difference between the actual stock price and the expected stock price:
On the other hand, if the market is efficient, the income of the stock will be similar to the stock price in the following equation:
Ngo Thi Minh Hoa_MSc in Finance | 7
: Actual observation of stock I at time t+1
: The difference between actual and expected returns of stock i
( ) : The expected return of stock I at time t + 1 in terms of market aggregation is
The new model, derived from the Fair Game Model, incorporates a slight adjustment to stock earnings, leading to a potential increase in income for securities investors, contrasting with the previous model where returns were zero This change can be mathematically expressed through a specific equation.
Despite a recent correction in future earnings, the model continues to illustrate the unpredictability of securities income While stock returns are generally expected to rise over time and dividends are likely to remain positive, the extent of these increases remains uncertain and difficult to forecast.
Price appreciation occurs due to market corrections triggered by the emergence of new information In an efficient market, information is released unpredictably, making it impossible to forecast when or what new data will arise This inherent unpredictability renders all stock price forecasting methods ineffective.
In the stock market, earnings per share represent the present value of the future income stream that investors can expect to receive.
In an efficient market, a company's share price accurately reflects all available information, meaning any price change directly impacts the stock's earnings stream Stock prices are influenced by new, often unpredictable market information, leading to unpredictable changes in income While tomorrow's market conditions will likely differ from today’s, these changes occur in a random manner This concept aligns with the Efficient Market Hypothesis (EMH), which underscores the inherent unpredictability of stock returns.
Ngo Thi Minh Hoa_MSc in Finance | 8
The random walk model, as defined by equation (*), illustrates the price and income fluctuations of stocks Professor Fama highlighted that this model expands upon The Fair Game Model, which focuses on the equilibrium of earnings per share Unlike The Fair Game Model, the random walk model provides stronger support for the Efficient Market Hypothesis (EMH), emphasizing the unpredictability of stock movements.
Vietnam Securities Market Overview
Fama's Efficient Market Hypothesis (EMH) is a foundational concept in modern finance that addresses investor concerns by explaining stock market dynamics The EMH posits that stock prices in an efficient market fully incorporate all available information, leading to random price changes that align with market indices As a result, predicting future stock prices is impossible since tomorrow's information remains unknown Consequently, in an efficient market, stock prices accurately reflect their true value, and any attempts to exploit inefficiencies are unlikely to yield abnormal profits for investors.
Despite the absence of a functioning real-world stock market, various factors contribute to this situation However, the primary objective remains consistent across all stock markets: to establish an ideal framework for countries to develop their own trading systems.
Testing the Efficient Market Hypothesis (EMH) in Vietnam's stock market is crucial for gaining a comprehensive understanding of market dynamics It enables us to assess the market's performance and development, allowing for timely and strategic measures to enhance its effectiveness and promote sustainable growth.
The stock market serves as a crucial mechanism for capital mobilization, significantly contributing to the development of both the financial market and the broader economy Established on November 28, 1996, the State Securities Commission (SSC) marked the inception of the Vietnam Securities Market.
Vietnam's stock market officially began operations in July 2000, making it relatively young compared to developed countries In its inaugural trading session on July 28, 2000, the market featured only two listed companies By 2005, this number had grown to 27 Over the past two decades, the market has significantly expanded, with the latest statistics from the State Securities Commission (SSC) reporting 2,090 listed companies and a registered volume of 125.92 billion shares as of May 2018 Additionally, the average trading value reached VND 14,203 billion in June 2018.
The Vietnamese stock market, despite its notable achievements, faces significant issues such as price manipulation and inadequate information disclosure Many companies report losses while their stock prices inexplicably rise, leading to consecutive sessions of extreme price fluctuations without any supporting data This raises concerns about the sensitivity of securities prices to market information and questions the overall efficiency of Vietnam's stock market To address these issues and analyze current phenomena, a study was conducted to test the Weak-Form Market Efficiency Hypothesis for the Vietnamese stock market from 2013 to 2017.
Literature review
In recent decades, numerous empirical studies have explored the implications of the Random Walk Hypothesis (RWH) and Weak Form Efficiency (WFE) in both emerging and developed stock markets, yielding varied results Early research on WFE theory predominantly relied on the RWH random step theory Several studies have assessed the effectiveness of weak markets through technical analysis returns Notably, Alexander (1961, 1964) introduced the "Filter Rule," a flexible trading strategy that advises investors to purchase stocks when prices rise by a certain percentage and to hold them until they fall by the same percentage, at which point they should consider short selling.
Ngo Thi Minh Hoa, MSc in Finance, highlights the findings of Fama and Blume (1966) regarding trading strategies in the stock market They compared a flexible business strategy, known as the Filter rule, with a buy-and-hold strategy over a fixed period If the flexible strategy yielded higher profits, it indicated market inefficiency; conversely, lower profits suggested weak market performance, as noted by Olowe (1999) The aim of a flexible trading strategy is to capitalize on systematic price trends in securities Alexander (1961, 1964) and Fama and Blume (1966) affirmed the existence of effective weak signals in the market By the late twentieth century, Fama and Blume's research was regarded as a seminal study in business rules Additionally, another research team analyzed the correlation of returns over time, with the random walk theory predicting a correlation coefficient of zero, as discussed by Cowles and Jones.
In 1937, a study utilized the Runs test to analyze the frequency of numerical strips, testing the random walk theory This research built upon Kendall's earlier findings, which suggested that stock price changes are nearly independent based on correlation coefficient measurements Subsequent studies by Dimson and Massavian (1998), Osborne (1959), and Fama (1965), along with Fama and Blume, further explored these concepts.
Research has shown that changes in stock prices are largely independent of investor decisions, as noted by Osborne (1959), and are primarily driven by the emergence of new information, according to Fama (1970) This randomness in information leads to random price movements, supporting the Weak Efficient Market Hypothesis (WEF) For instance, Kendall's (1953) analysis of stock prices for 22 US consumer securities and UK industry shares from 1883 to 1934 revealed that price fluctuations occurred randomly Similarly, Fama's (1965) study of the Dow Jones Industrial Average from 1957 to 1962, employing various statistical techniques, indicated a very low correlation coefficient, suggesting that extraordinary profits from trading strategies are unattainable and reinforcing the notion of market inefficiency.
(1973) used 234 stock models of eight major European stock markets from 1966 to
Ngo Thi Minh Hoa_MSc in Finance | 11
In 1971, the author examined the correlation coefficients for mice, weeks, and months regarding changes in profitability, discovering minimal signs of randomness that investors cannot leverage for price predictions Furthermore, Hagerman and Richmond (1973) conducted a study on 253 randomly selected stocks from the US OTC market between 1963 and 1967, concluding that the market aligns with the Random Walk theory Additional tests supporting the random walk market include findings from Cowles (1960), Cootner (1962), Mandelbrot (1966), Fama and Blume (1966), Sharp (1966), and Fisher.
Recent empirical studies in developed capital markets challenge the weak-form theory of the Efficient Market Hypothesis (EMH), suggesting that stock prices can be predicted in certain areas For instance, Lo and Mackinlay (1988) utilized the Variance ratio test on weekly returns of US securities from 1962 to 1985, which completely rejected the Random Walk theory Similar results were observed in other developed markets, including Japan, Hong Kong, Australia, and Singapore, indicating low market performance Furthermore, contemporary research highlights ongoing debates regarding weak market efficiency in emerging markets, attributed to irregular trading and insufficient supporting tools that hinder stock market effectiveness Havey (1995) posits that a higher potential profit margin correlates with a more efficient stock market; however, this relationship is not consistently observed, as recent studies indicate the presence of random walk behavior in some emerging markets, such as the correlation coefficients and chain tests conducted on the Kuala Lumpur Stock Exchange (KLSE) by Barnes (1986).
Ngo Thi Minh Hoa_MSc in Finance | 12
Table1 presents 8 studies on the weak form of the Vietnam stock market from 2000 to
Table 1 : Studies on the weak form of the Vietnam stock market over time
Source: National Institute for Finance
“x” is negative for effective market existence, “o” is to confirm the existence of effective markets in Vietnam for the period 2000-2013
On the Vietnam stock market, there have been some studies on the weak performance of the stock market However, the results are quite different Research of Thai Long
(2004) on VN-Index on HCM Stock Exchange from 2001-2005 The author uses the
The study employs the ARIMA method to analyze the efficiency of the Vietnam Stock Market, concluding that it exhibits weak market efficiency Research by Ho Viet Tien (2006) indicated ineffective market performance through correlation coefficients and profit distribution tests on the Vietnam Stock Exchange Similarly, chain testing of listed shares in Hanoi supported the notion of low market efficiency In contrast, Le Dat Chi (2006) found that certain stocks adhere to the Efficient Market Hypothesis (EMH) in its weak form Given these conflicting results regarding the Vietnam stock market, the research focuses on “Testing the Weak-Form Market Efficiency Hypothesis for Vietnam Stock Market from 2013 to 2017.”
Ngo Thi Minh Hoa_MSc in Finance | 13
Data
The VN Index serves as a key indicator of price fluctuations for all stocks listed on the HOSE, reflecting the overall market trends It compares the current market capitalization to that of July 28, 2000, marking the official launch of the stock market.
In order to test form efficient market hypothesis for the Vietnamese stock market, I use the market return variable of time series defined by the formula:
Daily market returns (Rt) are calculated from the daily index price and serve as a key time series variable for various testing methodologies These everyday showcase log returns play a crucial role in financial analysis and modeling.
Log returns are justified both theoretically and empirically, as they offer greater analytical simplicity and effectively link returns across different time periods Additionally, log returns exhibit a distribution pattern that aligns more closely with typical statistical methods, enhancing their applicability in financial analysis.
The daily closing prices of the Vn-Index were sourced from the reputable Vietnamese website stockbiz.vn, covering the period from January 1, 2013, to December 31, 2017 Prior to conducting any time series analysis using various methods, I established key statistical information for each time series through four distinct approaches.
Ngo Thi Minh Hoa_MSc in Finance | 14
In statistics, a skewness of 0 and kurtosis of 3 typically indicate a normal distribution of data However, the provided data shows a skewness of -16.82951 and a kurtosis of 422.6740, which suggests that this data string significantly deviates from a standard normal distribution.
Asymmetry quantifies the extent of imbalance in an experimental distribution, highlighting the uneven distribution of components relative to the center of gravity, or arithmetic mean.
When skewness is greater than 0, the distribution exhibits a right-deviated tail, indicating that values are scattered above the arithmetic mean Conversely, a skewness less than 0 signifies a left-sided distribution, where values are predominantly below the arithmetic mean In this case, with a skewness of -16.82951, the distribution demonstrates a pronounced left-shifted form.
The Kurtosis coefficient of 422.6740, significantly greater than 3, suggests a high dispersion in the market rate of return, resulting in a pointed graph Additionally, the Jarque-Bera p-value indicates that the market's rate of return does not conform to a standard distribution at any significance level of 10%, 5%, or 1% This information will enhance the accuracy of the test results in subsequent sections.
Ngo Thi Minh Hoa_MSc in Finance | 15
Methodologies for Weak Form market Efficiency Testing
Unit root test
First, we check whether the sequence lnP (t) is stationary or not (corresponding to A 1) according to the unit root method This method looks at the equation:
The equation is equivalent to:
Dickey Fuller (1979) suggested that the tau estimation of the coefficient a would follow the probability distribution tau (tau = the estimated p value/error of a) This test is estimated in three forms:
Ngo Thi Minh Hoa_MSc in Finance | 16
T is the trend variable and X (t) is the exogenous variable that includes α or α is the trend variable
Selecting the appropriate model based on test results hinges on the tau value of the coefficient in the Y variable (t-1) Since α must always be less than or equal to zero, any test yielding a coefficient of tau greater than zero lacks statistical significance Nonetheless, a correlation between u(t) variables may arise from omitted variables, prompting the use of the extended DF test, which incorporates the dependent variable ΔY(t).
In addition, in the case of suspicion of chain correlation between u (t), Philips Perron
In 1988, a method was proposed for estimating equation (1) by adjusting the tau value to ensure that the correlation with u(t) does not influence the tau distribution Additionally, the KPSS test introduced in 1992 is employed to examine the stationarity of Y(t), where the null hypothesis, H_0, asserts that Y(t) is stationary, represented by the corresponding equation.
Variance testing
In this analysis, we assume that the variance property, denoted as u(t), remains constant over time If this assumption holds true, the variance over a period of q will be equivalent to q multiplied by the variance over a single period.
Lo and Mackinlay (1988) proposed this test based on the ideal They use the variance ratio:
The hypothesis posits that U(t) maintains a constant variance over time q However, a key limitation of this test lies in the subjective nature of both the time period q and its selection If the variance var u(t) remains unchanged over time, then the condition VR(q) = 1 holds true for every q.
Ngo Thi Minh Hoa_MSc in Finance | 17
= 1, 2, Therefore, Chow and Denning (1993) proposed the following test:
For every q and H_1: there is at least 1 VR (q) 1 The test value here is
√ The calculated p-value corresponds to the above block so that the MV value is less than or equal to the upper block (2)
While the Lo and Mackinlay tests are commonly utilized, they exhibit notable limitations Firstly, these tests presume that the VR distribution follows a regular pattern, which may not hold true when the number of relations \( u(t) \) is limited Secondly, the VR test with \( q = 1,2, \) can erroneously suggest that the null hypothesis \( H_0 \) is false due to correlations among individual VRs caused by observations in VR(q) To address these issues, the Chow and Denning tests employ a bootstrap pattern replication method, yielding more reliable results The p-value is calculated by the ratio of the number of repetitions exceeding the upper limit, determined by (2) divided by the total number of reconstructions, with \( q \) set to 2, 4, 6, 8, 16, or 32 for date-based data, building upon the concepts established by Lo and Mackinlay.
In 2000, researchers suggested utilizing rank and rank scores of tssl as an alternative to the traditional tssl measure, demonstrating that this approach yields greater accuracy, particularly when tssl deviates from standard distribution patterns The study examined values of q at 2.5, 10, and 30.
Autocorrelation testing
Autocorrelation refers to the correlation among elements in a sequence of observations organized either chronologically (time series data) or spatially (cross-sectional data) The primary goal of autocorrelation analysis is to assess the likelihood of autocorrelation within the observed dataset A sequence is deemed correlated when the variance exhibits predictable cycles, indicating a discernible trend, which suggests that the sequence is not random.
The autocorrelation coefficient in this study indicates the linear relationship between stock return observations over time A non-zero autocorrelation coefficient signifies a dependency relationship among the observations Specifically, a positive coefficient suggests that past stock returns are positively correlated with future returns, indicating potential trends in the stock's performance.
Ngo Thi Minh Hoa, MSc in Finance, explains that the future return of a stock will generally follow the same trend as its current price If the stock price is currently rising, it is likely to continue increasing compared to previous levels Conversely, a negative correlation coefficient indicates that if the stock price is declining, it may experience a rise afterward A correlation coefficient different from zero suggests a non-random relationship among observations To analyze this correlation effectively, it's essential to identify its cycle, including the frequency of increases and the timing of declines, introducing the concept of latency The correlation coefficient can be calculated using a specific formula.
: Correlation coefficient of lag with lag time k k: latency
: Rate of return at time t
: Rate of return at time t+k r: Average return rate
This self-correlation test evaluates the market's rate of return with a specific latency, k For the rate of return to be deemed uncorrelated and the sequence to be classified as random, the correlation coefficient must equal 0.
The hypothesis in this case is:
The correlation coefficient in the market yield curve at each defined lag indicates the relationship between market yields at different latencies It is essential to assess whether these yields are correlated; if they are, the randomness of time series observations may be disrupted, even if they show no correlation at each individual lag Therefore, in addition to analyzing the correlation at specific delays, we will also investigate the correlation between yields across various latencies.
Ngo Thi Minh Hoa_MSc in Finance | 19
The hypothesis to be tested is:
To test for this hypothesis, we examine each individual p_k However, the statistical value Q is defined by the Ljung -Box as follows:
: Market Return Rate Correlation Ratio j
In a random data series, the correlation coefficients should ideally be zero; any non-zero correlation indicates an interrelationship among the observations, which contradicts the null hypothesis This article focuses on a two-tailed test using a Q test value, which follows a distribution with degrees of freedom denoted as k (the latency of observation) By applying a significance level α and the degrees of freedom k, we can determine the query value If the calculated value exceeds the threshold, we reject the null hypothesis and conclude that the sequence of observations exhibits autocorrelation.
Runs test
For a stock market to function effectively, stock prices and market indices must exhibit randomness, meaning past and future prices, as well as weekday and weekend prices, are independent This article employs the Runs test to evaluate the randomness hypothesis of the sample Chain testers are utilized to assess whether a continuous sequence of observations aligns with expected outcomes If the results fall within the anticipated range, we can accept the null hypothesis (H0) that the sequence consists of random numbers Conversely, if the number of runs exceeds the expected threshold, the hypothesis is rejected.
Ngo Thi Minh Hoa_MSc in Finance | 20
In a random process, the expectation of sequences is a critical consideration A limited number of strings may not adequately demonstrate the randomness or non-randomness of the process To assess this, we utilize a statistical tool that provides the formula for expected sequences, helping us understand the relationship between randomness and the number of observed outcomes.
N: the total number of observations of the sample
: the total number of strings in each string type
When the number of observations exceeds 30, the sample distribution of the expected value (m) approximates a normal distribution The standard deviation of this distribution can be calculated using a specific formula.
Here the hypothesis that we need to test is: the chain of observation of the daily profit of stocks and the VN index is random, ie using tail test:
The Z test statistic is utilized to evaluate the null hypothesis by comparing the calculated Z value to a critical value from the statistical table The Z value is determined using a specific formula, which plays a crucial role in the hypothesis testing process.
R: number of sequences of the observed range in reality
M: The number of sequences expected by chance
In 1956, scientists Wallis and Robert introduced a coefficient that adjusts based on the discrepancy between actual and expected string counts If the real count falls short of the expected value, the adjustment coefficient is incrementally increased by 0.5 Conversely, if the practical count of sequences is lower than the anticipated number of random sequences, the correction factor is progressively decreased by 0.5.
Ngo Thi Minh Hoa_MSc in Finance | 21
When the number of strings R deviates significantly from the expected range, the null hypothesis is rejected, indicating a dependency of the sequence of nonzero random variables on observations Consequently, the theory of an efficient market is not applicable to the Vietnam stock market.
Test results analysis
Unit root test
Null Hypothesis: LOG_RETURN has a unit root
Lag Length: 0 (Automatic - based on SIC, maxlag#) t-Statistic Prob.*
Augmented Dickey-Fuller test statistic -38.32323 0.0001
Augmented Dickey-Fuller Test Equation
Variable Coefficient Std Error t-Statistic Prob
S.E of regression 0.008469 Akaike info criterion -6.703406
Sum squared resid 0.106947 Schwarz criterion -6.696295
Log likelihood 5006.093 Hannan-Quinn criter -6.700757
Table 3 : Unit root test result
Ngo Thi Minh Hoa_MSc in Finance | 22
The Augmented Dickey-Fuller test reveals a test statistic of -38.32323, significantly lower than the MacKinnon critical values With a p-value less than α at the 10%, 5%, and 1% significance levels, we reject the null hypothesis, indicating that the market yield curve is not a stationary process and suggesting the absence of a random walk Furthermore, the t-statistic exceeds the critical values for all significance levels, reinforcing the rejection of the null hypothesis and indicating that the weak form of the market hypothesis does not hold.
Variance test
Null Hypothesis: Log LOG_RETURN is a random walk
Standard error estimates assume no heteroskedasticity
Joint Tests Value df Probability
Period Var Ratio Std Error z-Statistic Probability
*Probability approximation using studentized maximum modulus with parameter value 4 and infinite degrees of freedom
Period Variance Var Ratio Obs
Ngo Thi Minh Hoa_MSc in Finance | 23
Since I have defined more than one test period, there are two sets of test results The
“Joint Tests” are the tests of the joint null hypothesis for all times, whereas the
Individual Tests refer to variance ratio tests applied to specific periods The Chow-Denning maximum statistic of 11.97295 is associated with the individual test for period 5 With an estimated p-value of 0.0000, which is less than 0.05, the null hypothesis (H0) is strongly rejected, indicating significant results.
Autocorrelation test
Autocorrelation Partial Correlation AC PAC Q-Stat Prob
Ngo Thi Minh Hoa_MSc in Finance | 24
This test is used to test the relationship between observations in the series over time
If there is a correlation between the observations, the stock price does not change randomly but in a certain trend, this is a sign of an inefficient market
AC is the correlation coefficient, which represents the correlation of observations at each specific latency; the observed sequence is considered random when the correlation coefficients are zero
Q-Stat is a time-tested value, which also represents the correlation between the observations in the data series but the correlation between the lags Observations are only considered random when the Q-Stat is zero If only one of 36 correlations at 36 different latencies is 0 or not statistically significant then the series of data is considered to be correlated and results in rejecting the null hypothesis even though the
AC coefficient satisfies the condition Therefore Q-Stat is very important in making conclusions
Accreditation under this methodology is covered in the Methodologies section This method will provide additional evidence for a relationship between observations as a basis for conclusions supporting or rejecting the hypothesis
The analysis reveals 36 distinct latencies in Table 5, with all AC coefficients significantly different from zero, indicating that the VN Index's profitability is not random Furthermore, all Q-Stats across these latencies also differ from zero, confirming that the VN Index's profit between January 1, 2013, and December 31, 2017, exhibits a non-random pattern This self-correlation phenomenon suggests that investors can predict trends, with a notable finding that the negative correlation coefficient surpasses the positive one, implying a tendency for the VN Index to decline.
Thus, the test results show that self-correlation with the 95% confidence interval between the observations in the chain Positive correlations show the same change in
Ngo Thi Minh Hoa_MSc in Finance | 25 tssl between present and future, while negative correlations show the inverse mutation of tssl
The test results indicate a correlation in the daily tssl of the stock market, suggesting that the sequence of observations is consistent rather than random This leads to the rejection of the assumption, demonstrating that Vietnam's stock market is ineffective at the weak level.
Autocorrelation test (LBQ test) p-value 0.0086 t-statistics 38.0857 c-value 31.4104 h=1: rejects the null that the residuals of the returns are not autocorrelated
Another autocorrelation (LBQ test) shows the p-value is 0.0086 which is lower than 0.05, thus Hypothesis H0 is rejected; Vietnam stock market is not effective at the weak level
Ngo Thi Minh Hoa_MSc in Finance | 26
Runs Test ( median) log return Test Value a 0.0005208128108423
Runs Test 3 (mode) log return Test Value a 0128456578440902 b
Runs Test 2 (mean) log return Test Value a -0.000083268186200 Cases < Test Value 682
The Runs test is used to analyze data that adheres to a normal distribution, focusing on criteria such as median, mean, and mode According to J.L Sharma and R.E Kennedy (1977), the null hypothesis is rejected if |Z| > 1.96 In this study, the Z indices for median, mean, and mode were -0.518, -0.905, and 0.163, respectively, indicating that their absolute values do not exceed 1.96, suggesting a non-autocorrelation chain Additionally, the significance values for the median, mean, and mode were 0.605, 0.366, and 0.871, all significantly above the 5% level Consequently, the hypothesis is accepted, indicating that the stock market in Vietnam exhibits weak efficiency.
Conclusion
Testing techniques Weak form efficient market hypothesis in that sample period
Non- weak form efficient market hypothesis in that sample period
Table 7: Testing techniques results summary
The analysis presented in Table 7 indicates that, during the period from January 1, 2013, to December 31, 2017, the Vietnamese stock market exhibits characteristics of a weak form efficient market, as evidenced by the Runs test However, alternative testing methods, including unit tests, variance testing, and autocorrelation testing, suggest that the market does not fully align with weak form efficiency These findings reflect the prevailing market conditions in Vietnam, which are comparable to those observed in other developing markets.
Analysis of aggregate data reveals that the price and profit of the VN-Index do not fluctuate randomly Instead, the Vietnamese stock market exhibits three notable phenomena that influence these changes.
Over-the-counter (OTC) stock prices are typically lower prior to their official listing, as investors often rush to purchase these stocks in anticipation of their debut on the stock exchange Additionally, a trend known as "the price of the week" suggests that early week prices frequently reverse the fluctuations observed at the end of the previous week, with weekend gains often followed by declines at the start of the new week, or vice versa Furthermore, each new issuance or bonus can significantly impact market dynamics.
Ngo Thi Minh Hoa_MSc in Finance | 28 price will increase before the closing date of the listing Such market types are considered to be inefficient markets (Khanh, 2007)
The fluctuations in time series data can lead to varying increases or decreases over time, which limits the effectiveness of technical analysis in securities investment This phenomenon is evident when market enthusiasm surpasses analysts' expectations, and during downturns, the absence of a solid support level can exacerbate market declines.
When market inefficiencies arise, investors compete to exploit these anomalies for profit, ultimately driving the market back to equilibrium This process eliminates the inefficiencies, paving the way for new phenomena to emerge As these new dynamics develop, the market becomes stronger, often in ways that investors have yet to recognize.
For many professional investors, the stock market never has the past, meaning that the phenomenon never repeats the same, so it is difficult to predict
Ngo Thi Minh Hoa_MSc in Finance | 29