Informational efficiency of London and Vietnam stock markets

Secondary markets are regulated by regulatory authority. (William, J. O'niel, June 8, 2009,How to make money in stocks, McGraw-Hill Education. ) Security prices are of great importance to investors hence its forecast enables better performance of those financial markets than the others. Since the significant study by Fam (1965), where he verifiably presented that mutual funds’ investments for stocks of the Dow-Jones Industrial Average do not outperform randomly selected portfolio with no dominance by the fund managers over the average investors. 06 trillion, in December 2014, London Stock Exchange, therefore, shifted to the third position in the world stock exchange. London Stock Exchange promotes equity markets deals such as professional securities, specialist fund market and Alternative Investment market that was launched on 19th June 1995. Notably, there has been an upgrade in the trading platform which is Linux- based named “Millennium Exchange” Ho Chi Minh City Stock Exchange has also experienced changing times after it was opened in, 2000.

As from 2006, this stock market has been receiving hot inflows of money and an increase in trading days. In 2007 there was instability of Vietnam’s transitional economy which caused a lot of losses to investors. Market efficiency is dynamic and efficiency is measured by taking into consideration stock indices of portfolios in Ho Chi Minh City Stock Exchange and London Stock Exchange on a daily basis. Specific values are assigned to share volumes based on the market capitalization. Market efficiency can be studied using a stochastic modeling approach that captures the time-dependent structure embedded in time series market efficiency data. This paper, therefore, seeks to fit random walk process to the daily share prices and exchange rates of Ho Chi Minh City Stock Exchange and London Stock Exchange.

Random Walk process is used because of its generality, it can handle many series regardless of whether they are stationary or not, with seasonal or without seasonal elements. He then described three types of market efficiencies. The strong form of market efficiency which refers to markets where information that is public or private reflects in prices of stock. There is believed that investors with access to insider information do not have an advantage over others. Weak market efficiency states that future prices are random and are not affected by past events. This, therefore, means that past data cannot be used to predict future prices. They discovered efficient market hypothesis from the non-stationary time series. However, their discovery was contrary to that of Shiller (1981).

Brown and Easton(1989) studied LSE,s daily data, they made effort to find out its efficiency in historical periods. They made use tools of run tests, x2, and serial correlation, they confirmed LSE’s efficiency in the historical time period as proven in contemporary markets. They discovered independence of price movements and serial correlation coefficient that was close to 0. On the other hand, Colombia, Indonesia, South Africa and Turkey showed weak form efficiency in their exchange rates. Empirical studies show a weak efficiency in Vietnam as the test is conducted on the basis of the interrelationship between past and current stock price (Fawson et al, 1996). Several tools are used including run test or uni1997), Abraham et al. (2002) Karemera et al (1999) all conducted run tests while.

Buguk and Brosen(2003), Groen et al (2003) and Seddighi and Nian (2004) conducted unit root test. 1 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 8. 915 res[-1] -0. 719 Residual standard error: 745. 4 on 96 degrees of freedom Multiple R-squared: 0. 2 Autocorrelation tests A process Xt is said to be an autocorrelation process of order P i. e AR (P) if it is a weighted sum of the past p values plus a random shock. It is like a regression where X is regressed on previous X's. The present depends on the past plus some error. (Seddighi,1996) Durbin-Watson test data: cpt ~ pt_1 DW = 2. 05 so do not reject the null hypothesis (there is no autocorrelation) 3. 3 Model selection criteria In this paper, the Akaike's, (1973) Information Criterion is used for model selection.

The criterion says to select the model that minimizes: AIC = -2log (maximum likelihood) + 2k Where k = p + q + 1 if the model contains an intercept or a constant term and k = p + q otherwise. The addition of the 2(p + q + 1) or 2(p + q) serves as a "penalty function” thus ensuring the selection of a parsimonious model. k is the number of parameters in the model. The mean function is constant over time 2. Variance is constant 3. The autocovariance doesn't depend on time In this paper, the Augmented Dickey-Fuller test has been used to test for stationarity. Testing unit root Consider the following model H0: Non- stationary H1: Stationary Augmented Dickey-Fuller Test data: pt_1_datatimeseries Dickey-Fuller = -3. 7858, Lag order = 4, p-value = 0. 05 hence reject the null => evidence of no unit root CHAPTER FOUR 4.

1 DATA DESCRIPTION The data employed in this study comprised of one hundred observations of the stock markets of London and Vietnam obtained from HSBC and Vietnam bank. In these two stock markets, social media plays a vital role in determining the prices of stocks. This is because any information available always has an impact on prices of shares. This data covers a span of one hundred observations, this was necessary to ensure a good coverage of the revision of stock prices for this period. Figure 4. 1: The time plot of the stock prices indicates a time series of the stock prices with an increasing trend over the time period. Further tests carried out i. e. plots of the histogram fig 4. The stock price Correlogram graph indicates non – zero autocorrelation of the lags with most of the lags being insignificant, this proves the non – stationarity of the time series.

Figure 4. 5 time series ACF plot. Figure 4. 6 time series PACF plot. 01 that leads us to reject the null hypothesis at 1% level of significance and conclude that the series is now stationary. AUTOCORRELATION TESTS Durbin-Watson test data: pt ~ pt_1 DW = 1. 9683, p-value = 0. 3977 Alternative hypothesis: true autocorrelation is greater than 0. in DW p has a value of 0. Error t value Pr(>|t|) (Intercept) -0. 946 res[-1] 0. 895 Residual standard error: 30. 05 on 96 degrees of freedom Multiple R-squared: 0. 0001839, Adjusted R-squared: -0. 01765 on 1 and 96 DF, p-value: 0. 8946 The null hypothesis is that there is no autocorrelation confirming their independence. The T value is -0. 068 thus do not reject the null hypothesis. 4 SELECTION OF THE MODEL The principle of parsimony (we assume that the model with the fewest parameters is best) is inadequate in deciding which model was the best.

The study has presented us with an opportunity to have an extensive understanding of the theory of time series analysis in the area of non-linear models and its application to real life situation. The stages in the model building (that is the identification, estimation, and checking) strategy has been explored and utilized. In coming up with the model we used historical stock prices from 100 data entries. After estimation of the parameters of selected models, a series of diagnostic and forecast accuracy test was performed. Having satisfied all the model assumptions, the model was judged to be the best model for forecasting. For them to make this technical analysis they require predictions on the future movement of prices. It is from this analysis they either decide to sell, buy or hold on the stock.

Market intermediaries: They include stock brokers and clearing houses, market efficiency is important for them since they buy and sell securities on behalf of investors. Suggestion for further studies 1. A further study using GARCH models should be done to incorporate variances in price. IEEE Transactions on Automatic Control 19 (6): 716–723. (accessed June,22nd,2018) Andreja Pufnik, Davor Kunovac, (2006). Short-Term Forecasting of Inflation in Croatia with seasonal ARIMA processes. Retrieved December, 2006, from Croatian National. Bank, Working Paper No 16:http://www. |(accessed 23rd June, 2018) Meyler, A. G. Kenny and T. Quinn, (1998). Forecasting Irish Inflation using ARIMA models, Central Bank of Ireland, Technical Paper. (accessed 22nd June, 2018) Chow, K. V. , and K. C. Denning, 1993, A simple multiple variance ratio test,Journal of Econometrics 58, pp. Koh and S.

Ouliaris, 1997, Joint variance-ratio tests of the martingale hypothesis for exchange rates, Journal of Business and EconomicStatistics15, pp. (accessed 21st June, 2018) Hoque, H. , J. H. S. Chung, 1996, Price limits and stock market efficiency, Journal of Business Finance and Accounting 23, pp. (accessed 21st June 24, 2018) Seddighi, H. R(2012), Introductory Econometrics: practical Approach. London and New York. Brooks, 2009, Price limits and stock market efficiency:Evidence from rolling bicorrelation test statistic, Chaos, Solitons and Fractals 40, pp. (accessed,22nd June, 2018-06-24) Lo, A. W. , and A. C. d)To match the same times interval as the five datasets. (accessed 22nd June,2018) Appendices Appendix 1:R-SOFTWARE CODES USED IN ANALYSIS • # Load TTR package before running ACF and PACF • acf(pt_1_datatimeseries) # Autocorrelation plot on original CPI data • pacf(pt_1_datatimeseries) # Partial Autocorrelation on original CPI data • # Load tseries package before testing stationarity • adf.

test(pt_1_datatimeseries, k=1) # Test for stationarity • pt_1_datatimeseriescomponents<- decompose(pt_datatimeseries) pt_datatimeseriescomponents # Decomposing Time series into constituent components • plot. ts(ptandpt_1) # Plot on the differenced data • adf. test(ptandpt_1, k=1) # Test of stationarity on the differenced data • acf(ts(ptandpt_1, freq=1, 30)) # Autocorrelation values on differenced data • acf(ptandpt_1,plot=FALSE) # Autocorrelation plot on the differenced data • pacf(ts(ptandpt_1, freq=1, 30)) # Partial Autocorrelation values on differenced data • pacf(ptandpt_1,plot=FALSE) # Partial Autocorrelation plot on differenced data • modelcpi<- rm(pt_datatimeseries, order=c(2,2,0)) • # Load forecast package before running auto random model • tsdisplay(residuals(modelpt)) # Residual test on the fitted RM model whether it indicates a white noise • mean(residuals(modelpt_1)) # Test on the mean of the residual whether its Zero or close to zero • plotForecastErrors<-function(forecasterrors) # set up residual forecast plot.