Analysis of the Dar Es Salaam Stock Exchange
Disclaimer: This work has been submitted by a student. This is not an example of the work written by our professional academic writers. You can view samples of our professional work here.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.
Published: Thu, 01 Mar 2018
3.0 Chapter Three
3.1 Research Methodology
This chapter aims at explaining the methodology which has been adopted in this study. Research approaches or style have been categorized into mainly two groups, the phenomenological approach and positivist approach. The phenomenological approach studies the phenomenon through observation, no theory at outset while the positivist approach use an existing theory or develop a new theory and test its validity.
Since this study has used the existing theory on market efficiency therefore positivist approach have been adopted with this study, the rationale behind the choice of this approach is due to the nature of the study.
3.2 Research Design
Research design can be broadly classified as exploratory research and Conclusive research .This study is conclusive research design because it involves the testing of specific hypothesis and examination of relationships as well as the data analysis is quantitative and research process is formal.
3.3 Data types and sources
Two types of data that has been used in this study, the daily closing stock of market index(Dar es salaam Stock Exchange IndexDSEI) and the weekly share prices for a sample of five listed companies from Dar es salaam Stock Exchange. The daily closing stock for market index has covered the period from July 2007 to August 2008 making total number of observation to be 280, excluding public holidays and non trading days. The daily data prior to July 2007 were not found therefore the study had to use the available data .
The second type of data that has been used in this study are weekly share prices of the five companies/securities included in the study .The weekly data runs from Jan 2002 to August 2008, which makes the total number of observation to be between 90 and 266. The final date is the same for the all companies but the initial date differs depend on when the company joined the stock market.
The weekly data refers to the Wednesdays closing stock price, however if Wednesdays data were not available then Thursday closing price were used, in absence of Thursday data , Tuesday was taken instead, but when both Tuesday and Thursday were not available as well , the data for that week was regarded as a missing data. The use of weekly data is appropriate for this kind of studies as Humphrey and Lont (2005) asserted that ‘weekly data helped to mitigate any nontrading effects and also reduced the effects of noise trading’. Even though the stock price was collected for the purpose of performing statistical tests, the actual test was conducted using natural logarithmic of the relative price. The stock return (denoted by R) was calculated by natural logarithmic difference of the weekly stock price given by the following equation
= [ )] (1)
Where:
Rt = Return at time t
P = Price at time t
The reasons why change in log price was used instead of the normal change in price, has been explain by Fama (1965), he mentioned that logarithms neutralize price level effects as well as producing a series of continuously compounded returns.
The daily closing stock for market index (DSEI) was used in performing the parametric serial correlation test and the weekly share prices for five listed companies was used to perform the non parametric runs test .In additional to the primary data (Stock prices) collected from Dar es salaam Stock Exchange, the study has also used the secondary source of data. The secondary data includes academic books, journals and other publications.
3.3.1 Thin Trading
As discussed earlier in literature review section, infrequency trading or thin trading is big problem in most of emerging stock market and failure to take into account can results into serious biasness of statistical results. In this study this problem has been taken into account and the weekly data were collected from infrequency trading.
The first step taken in controlling the infrequency trading was to eliminate the mostly thin traded securities / companies as Shanken (1987) depicted that some of researchers controls the thin trading problem by eliminating some of thin traded stocks. Initially the study was meant to include all ten companies listed in Dar es salaam Stock Exchange, however five companies were found to be very much affected by infrequency trading therefore were eliminated from the study.
The actual correction of weekly data from thin trading for the five companies included in this study was based on approach by Atchison et al (1987) as adjusted by Milambo et al (2003;cited in Mabhunu 2004). They suggested correcting thin trading problem by adjusting approach by Atchison at el (1987) who used uniform process which allocates returns equally over the days in multi days interval where security not traded.
According to Mlambo et al (2003 ;cited in Mabhunu 2004), if a stock is not traded for example after 14 days of non trading, then a single entry given by the following equation
(2)
should be used as an oppose to 15 entries of equal value.
Where:
= Length of time between a trade in a period t and previous successive trade
= Price of stock at time t
= Dividend at time t
Therefore the infrequency problem in this study have been controlled by applying equation 2 without taking into account the dividend adjustment as it has been suggested that adjustments of dividend does not have much effects.Also the use of weekly data instead of daily data for individual companies has helped to control this problem.
3.3.1 Test of goodness offit
One of the hypothesis in which the random walk has based on is about price changes to conform to some probability distribution. Therefore in testing the efficiency of stock market it is essential to identify the pattern and determine which known statistical distribution the pattern follows. In this study the JarqueBera test has beeen employed to test the normality of the stock return. This techniques has been used in several studies, includes the recent study of Market Return and Weak Form Efficiency: The case of Ghana by Frimpong and Oteng (2007). The JarqueBera test statistic is given by
JB = T(+ ) (3)
Where:
JB = Test Statistic
T = Number of observation
S = Sample Skewness
K = Sample Kurtosis.
Kurtosis which denoted by (K) in equation (3) measures the sharpness / peakness or flatness of the distribution of a series and is given by the following equation
K = (4)
A normal distributed series has kurtosis of 3, therefore whenever kurtosis of a series exceeds 3, the distribution of that series is regarded as leptokurtic relative to normal and if the kurtosis is less than 3, then the distribution is regarded as platykurtic(flat) relative to normal.
The skewness which denoted by (S) and computed by the following equation
S = (5)
Measures the asymmetric distribution of the series from its mean. A normal distributed series has skewness of zero, therefore if the skewness of the series is positive then the series is concluded to have a heavier right tail and if the skewness is negative the distribution is regarded as having a heavier left tail relative to normal.
The results of the JarqueBera test together with the skewness and kurtosis of the return series employed in this study have been reported in table 1.
3.4 Methods employed
As mentioned earlier this study aimed at achieving three main objectives, first to find empirical evidence of weak form efficiency hypothesis for Dar es Salaam Stock Exchange, secondly to identify the main barriers for the development of the Dar es salaam Stock Exchange i.e. the factors that hampers the growth of DSE and lastly to identify the quality of information available to investors at Dar es salaam Stock Exchange. Therefore in this chapter the methods used in achieving each objective have been explained in detail, starting with the first objective.
3.4.1 Objective 1: Empirical evidence for weak form efficiency hypothesis.
In achieving the first objective, the study intended to answer the following two specific questions
Is the Dar es Salaam Stock Exchange weakform efficient?
Do the stock prices in Dar es Salaam Stock Exchange follow the random walk?
The study was guided by the following hypothesis
Dar es salaam Stock Exchange is weak form market efficient.
Stock price follows a random walk
Various techniques have been used so far in testing for weak form hypothesis by different researcher as depicted in literature review section, the techniques includes the statistical test of independence and trading rules.
In determining whether a stock market is a weak form or not using statistical test, the correlation / relationship between stock price and return over the successive time interval is identified. If no significant correlation found then the market is regarded as weak form market as past return can not be used to determine future return. The market will be regarded as weak form inefficient if significant correlation will be found.
In testing our first null hypothesis, one statistical test of independence have been employed, the serial correlation test . The non parametric runs test was employed to test our second null hypothesis, the random walk hypothesis. The following is the explanation of each statistical test employed in this study.
Serial correlation test
It is among the widely used test of independence .The serial correlation test measures the correlation of a variable over consecutive time interval e.g. at time t and time t1.The reasons why this approach have been chosen to be used in this study is because of its familiarity in this kind of study. Several studies have employed this technique for example Vaidyanathan (1994) in the study of efficiency of the Indian capital market employed this approach. Similarly, Baral and Shrestha (2006) studying the daily stock behavior of commercial banks in Nepal, used the same approach.
In testing the weak form efficiency of the stock market using this approach, the correlation of log price/return series is determined, if autocorrelation is found the assumptions will be that the series does not follow the random walk, meaning that the stock price are not independent, past return can be used to determine the future return and hence the market is weak form inefficient.
The test statistic for the serial correlation coefficient for lag p can be express as
p = (5)
Similarly written as
P = (6)
In determining the autocorrelation of the return in this study, the LjungBox test was used. This is a portmanteau test which measures the autocorrelation of the variable. The LjungBox test statistic is given by
=T(T+2) (7)
Where by:
= Test Statistic
T = Number of observations
= Is the jth autocorrelation or autocorrelation coefficient (for lag j)
K = Number of coefficients to test autocorrelation, in other words the number of lag to be Tested.
Given the value of obtained from the test, the conclusion on the randomness of the log price/return can be reached if > , K at significance level α, where by , K means the αquantile of the Chisquare distribution with K degrees of freedom. Alternatively if the pvalue obtained from the statistical test is less than 0.05,then the test is significant at 95% level of confidence and therefore the null hypothesis of zero auto correction can be rejected. The results for this test have been reported in figure 1in the next chapter.
Runs test
This is the second test that had been employed in this study to test for the second null hypothesis. Unlike parametric tests such as serial correlation, a runs test is a nonparametric test which means that it does not require the normal distribution of the series. This is one of the advantage of using this approach and it is also the reason why this technique has been adopted in our study. A run can be define as a ‘set of identical (or related) symbols contained between two different symbols or no symbol (such as at the beginning or end of the sequence)’Spiegel et al (2000.p366).
In performing this test, each change in return/price is classified as positive (+), negative () and zero change (0). Alternatively change in return could be classified alphabetically for example A ,could be each return that equal or exceeds the mean value and B could be each return that are below mean value. The test can be executed to obtained the actual number of runs (denoted by V), and then the actual number of runs (V) can be compared with the expected number of runs () which is given by the following equation
= (8)
Where
= Expected number of runs
N = Total number of return observations
= Sample size of each category of price change
If actual number of runs will be greater than expected runs, it will be indications of negative serial correlation and if actual runs fall below expected return it will indicate the positive serial correlation of the return. Alternatively the pvalue obtained can be used to conclude on the results of this test, if pvalue is less than 0.05, then the test is significant at 95% level of confidence and therefore the null hypothesis of randomness can be rejected.
For a large sample i.e (N>30), the sampling distribution of V is approximately corresponds to a normal distribution and thus
Z = (9)
Where:
Z = ZTest Statistic
V = Actual return
= Expected return
= Standard deviation given by the following equation
= [{+ N (N+1)} – 2N – (10)
Therefore at appropriate level of significance, the Zstatistic can be used to test for independence of return series.
The reason why the randomness tests such as non parametric runs test are used to test for the efficiency of the stock market is because efficiency of the stock market is determined by the way information are incorporated in current stock price. For a well efficient market , new information is incorporated instantaneously and spontaneously and therefore no arbitrage opportunity can exist.
Since new information is incorporated instantaneously and spontaneously in current stock price then stock price/returns will be generated in random fashion i.e there will be no any pattern. In relation to the weak form efficient market all past information is expected to be incorporated in current stock price in such a way that a positive change in returns is not expected to be followed by positive change in return or negative to be followed by negative as the returns generated randomly. However, for the weak inefficient market all past information are not incorporated instantaneously and spontaneously as the results the change in returns is generated in a pattern which can lead to opportunity of making fortune.
Therefore testing of randomness helps to reveal the how new information is incorporated in current stock price and the way returns are generated, if its in a random fashion or with pattern. This helps in drawing conclusion regarding the efficiency of a stock market. The results for this non parametric runs test are shown in table 2 & 3 and discussed in the next chapter.
Objective 2:Factors affecting the growth/ development of Dar es salaam
Stock Exchange
Despite aim of finding empirical evidence of weak form hypothesis, also the second objective of this study was to identify and discuss major factors/ challenges that have been affecting the development and progress of Dar es Salaam Stock Exchange.
April 2008, the Dar es Salaam Stock Exchange celebrated its tenth years anniversary, however for the period of its ten years of operations; we have witness the slow growth / development of the stock market, only few companies have been listed so far. But what are the main causes of this slow growth? in terms of listing of companies?, what are the challenges faced by the stock market?.
Further more the numbers of individuals participating in the market as investors is not so impressive, in a speech by the minister of finance and economic affairs on 10^{th} anniversary of DSE , he said the market so far DSE has enable more than 116,651 Tanzanians to own shares. This is small figure to be as a minimum figure for the country with population of approximately 39.4 million people, we would expect a good number of individual to be aware of operations of the stock market and hence participating and a minimum figure could have been a million and above, however the situation is different then what is the real problem?, are there any efforts by the market authority to ensure the general public is aware of the stock market operations and hence increase the number of investors in the market?.
Despite of the barriers and challenges for its growth, what measures have been and will be taken to ensure the stock market is growing? .What are the future prospects of the market? . In achieving our second objective the above mentioned questions will be addressed and discussed. This was done through reviewing and studying of the existing literature and publications regarding Dar es Salaam Stock Exchange and African stock markets in general, since most of the emerging African stock markets share the same kind of the obstacle/ challenges. The findings and discussion of these issues have been presented the following chapter.
Objective 3: The quality of information to investor and other stakeholders at DSE.
The last objective was to determine the quality of information available to investors and other stakeholders at Dar es Salaam Stock Market. There various sources of information for investors in any stock market and one of the sources is financial statements. Even though financial statements are sometimes subjective to the manipulation of management and by the time financial statements are published some changes might have already happen, yet financial statements remains to be crucial source of information for investors and analysts.
Normally the existing investors as well as potential investors would like to know how the investment have been well managed as this will give them the overall picture on how safe investing in the company has been or will be. Using the published financial statements, investors and analysts can acquire valuable information which can help in their decision making.
However, investors will be deprived from using this type of source of information, if the information provided with the financial statements are not of good quality and required standard. According to Benston (2003), if the information provided by financial statements is not useful and accurate then its reception will not give investors the kind of insight they wanted and as the results investors will incur costs to find information somewhere else.
Therefore with this objective, the quality of information available to investors in DSE was determined and discussed. This was achieved through a comparison of financial statement of Tanzania Breweries Limited (TBL) a company listed in Dar es salaam Stock Exchange and Sanisbury PLC – a company listed in London Stock Exchange.
The aim of the comparison was to determine if an investor in DSE using financial statement will get the same quality of information similar to an investor in London Stock Exchange. In this comparison in additional of looking the contents and standards in which these annual report have been prepared also the study looked at the general accessibility of the annual report and other company’s information which might be helpful to investors between these company .Also the general overview of corporate governance between these two companies was analyzed and discussed. The results and discussions of this comparison have been presented in the following chapter.
3.5 Data Analysis
Since the study had involve the statistical tests, therefore data was analyzed with the help of statistical packages. The parametric serial correction test and parametric runs test was performed using SPSS (Statistical Package For Social Science) and the Jarque Bera test was performed using EVIEWS .Both quantitative and qualitative approach have been used in interpreting the results of analysis
4.0 Chapter Four: Data Analysis, Presentation & Discussion of Findings
4.1 Introduction
The aim of this chapter is to present the analysis and discuss the findings of the study. The chapter have been divided in three main part (A, B, C). The first part (A), reports the results and the discussions from statistical tests relating to the first objective of this study. Part B and C report the findings and discussions relating to the second and third objectives respectively .
4.2 Part A : Empirical evidence for weak form efficiency hypothesis
In finding the empirical evidence for weak form efficiency hypothesis in Dar es salaam Stock Exchange ,three main statistical tests were performed , firstly the normality test i.e the Jarque – Bera test, parametric serial correlation test and non parametric runs test, the findings of these test are presented in that order.
4.2.1 Test of goodness – of fit
Since it is essential to determine the probability distribution of the series when performing efficiency tests /statistical tests, therefore the returns for the stock market index(DSEI) as well as the returns for the five companies used in the study were firstly analyzed to determine if the return series follows the normal distribution. The result of nomality test are shown in table 1
Table 1: Jarque Bera Test
DAHACO 
DSEI 
SIMBA 
TBL 
TCC 
TWIGA 

Mean 
0.002166 
0.000227 
0.006390 
0.002727 
0.000724 
0.008349 
Median 
0.000000 
0.000000 
0.000000 
0.000000 
0.000000 
0.000000 
Maximum 
0.141970 
0.021668 
0.277632 
0.202941 
0.074108 
0.287682 
Minimum 
0.182322 
0.024520 
0.253781 
0.146093 
0.117783 
0.072759 
Std. Dev. 
0.028979 
0.002794 
0.045971 
0.027860 
0.021852 
0.040065 
Skewness 
0.462822 
0.623957 
0.863549 
2.133508 
0.934278 
4.314131 
Kurtosis 
15.81011 
39.73707 
17.61886 
23.35083 
11.19382 
29.39428 
JarqueBera 
1216.549 
15707.35 
1751.608 
4792.031 
532.6697 
2891.643 
Probability 
0.000000 
0.000000 
0.000000 
0.000000 
0.000000 
0.000000 
Sum 
0.383447 
0.063240 
1.239581 
0.725336 
0.130979 
0.751392 
Sum Sq. Dev. 
0.147799 
0.002170 
0.407877 
0.205694 
0.085951 
0.142861 
Observations 
177 
279 
194 
266 
181 
90 
Source : Analyzed data
As shown in the table 1, the pvalue of the jarque bera test for stock market index series(DSEI) and the pvalue for the five individual companies is below 0.05 i.e pvalue < 0.05 .This implies that the test is significant at 95% confidence level and therefore the null hypothesis of normality distribution of the returns series is rejected for the both market index and individual companies.
As depicted in the table 1 above, the return for the market index(DSEI), TCC Ltd and DAHACO/SWISSPORT Ltd are negatively skew(heavier left tail) as their skewness less than zero. I,e skewness < 0. The skewness of TBL Ltd, Simba Ltd and Twiga Ltd are greater than zero( skewness > 0) which means that their returns are positively skew( a heavier right tail) relative to normal. A perfectly symmetrical distribution such as normal distribution has skewness which equal to zero.
Regarding the kurtosis which describe the flatness or peaknedness of the distribution the results shows that the returns of both market index and individual companies have kurtosis greater than three i.e kurtosis > 3, which implies that the distribution of the returns are sharply peaked (leptokurtic) relative to normal. The pvalue from test statistic, kurtosis and skewness indicates the rejection of normality for the returns so the general conclusion which can be drawn from the test of goodnessoffit is that the returns employed in this study are not normally distributed and therefore non parametric statistical tests are more appropriate to be used than parametric statistical tests.
4.2.2 Results of Serial Correlations/ Autocorrelation Test
Though it has been suggested that when the series is not normally distributed then non parametric tests wo
Cite This Work
To export a reference to this article please select a referencing stye below: