Dividends Announcement and Semi Strong Form of Efficiency at The Nairobi Securities Exchange in Kenya

: The objective of this study was to establish the effects of dividend announcement to current market prices at the Nairobi Securities Exchange, with four specific objectives; to determine the information content of dividend announcements, to determine the extent to which prices converge to new values after dividend announcements on a sector by sector basis, to establish the market reaction to announced information and also to establish whether investors can secure excess returns by acting on announced information. A 66 days event timeline was employed from 2005 to 2015 on daily closing stock prices. A sample of 179 dividend announcements from 22 listed companies in 8 sectors were drawn and analysed using an OLS Market Model. Findings of the research conclude that; dividend announcements do have an impact on stock prices for the Agricultural, Banking, Commercial, Construction, Manufacturing and Telecomm Sectors and not for the Energy Sector, Insurance Sector and the Nairobi Securities Exchange Market. It takes more than five days for prices to adjust to their correct values and this makes it possible for market players to profit from the inefficiency by earning abnormal returns. We conclude that the Nairobi Securities Exchange is not semi-strong form efficient.


I. INTRODUCTION
The Nairobi Securities Exchange was formally started in 1954. The Government saw the exchange as a medium of raising finances while the brokers viewed it as source of raising funds for the expansion of the private enterprise (Muragu, 1990). In the recent years we have evidenced changes at the Nairobi Securities Exchange (NSE) The most notable are; settlement period moved from T+4 (Transaction date-T plus 4 Days) to T+3, live automated trading system (ATS) of securities, Dual listing within East African Security Exchange Associations, Review of the 20 Share Index in 2007, Introduction of the NSE All Share Index (NASI), demutualization of the NSE; the exchange became the second in Africa to demutualize after the Johannesburg Stock Exchange. (Nairobi Securities Exchange (NSE), 2015). The legal framework governing the operations of NSE has had its share of changes; in 1989 the Capital Markets Authority (CMA) was established to promote and maintain an effective and efficient market (Muragu, 1990). The CMA remains the regulatory authority charged with the responsibility of supervising, licensing and monitoring the activities of market intermediaries. A number of legislations have been passed to regulate the market which include; regulation of futures exchange, real estate investments, public offers, listing and disclosure, demutualization of the NSE, licensing requirements, foreign investors regulation, take over and mergers, collective investment schemes, Operational Rules, asset backed securities, venture capitalist among others. (CMA, 2014) The above changes are geared towards efficiency of the market, it is therefore imperative to establish whether NSE is efficient. The availability of information on the market is pertinent as it helps to determine the level of efficiency and sophistication attained by the market (Olweny, 2012).The efficient market hypothesis stipulates that whatever information you consider in making a decision has already been incorporated into prices and is useless for predicting future prices (Fama E. F., 1970). Fama goes ahead and classifies market efficiency into three categories. Weak Form Efficiency -asserts that current price fully incorporates all historical information of past prices movement only. The semi-strong form of market efficiency hypothesis suggests that the current price fully incorporates all publicly available information. Neither technical nor fundamental analysis can find a way for gaining excess returns. This form of market efficiency was researched. The strong form of market efficiency hypothesis states that the current price fully incorporates all existing information, both public and private (sometimes called insider information). Duncan Turere 1 , AFMJ Volume 5 Issue 10 October 2020 Despite the above initiatives and achievements the Market still faces several challenges, namely; inconsistent use of the capital markets by domestic and regional entities for capital mobilization; weaknesses in the conduct of business by intermediaries which have contributed to retail investor apathy and constraints on international investment flows thereby affecting liquidity in the market (Capital Markets Master Plan Steering Committee (CMMSC), 2014). A narrow retail investor base; 4% of the population currently participates in the local market (Oxford Business Group, 2016). Studies have documented contradictory evidence; (Muthama, 2013) indicates that the NSE appears to be an inefficient market, suggesting that there are opportunities to make excess returns which investors and market analysts can exploit. (Muragu, 1990)-indicates that small markets, such as the NSE, may provide empirical results consistent with weak-form efficiency. The NSE is not semi-strong form efficient as some investors can earn abnormal returns by having unequal access to public information (Olweny, 2012). Olweny, recommends further research on effects of dividend announcements on firm value to be conducted on a sector by sector basis, this was addressed in this research. This will help uncover the effect of dividend announcement on firm value on a sector by sector basis and what this implies on the efficiency of the NSE (Olweny, 2012).

II. MAIN OBJECTIVE
The aim of the study was to establish the effects of dividend announcements to current market prices at the Nairobi Securities Exchange. A. Specific objectives i) To determine the information content of dividend announcements to market prices ii) To determine the extent to which prices converge to new values after dividend announcements on a sector by sector basis iii) To establish the market reaction to dividends announcement iv) To establish whether investors can secure excess returns by acting on announced information

A. Random Walk theory
The random walk theory asserts that price movements will not follow any patterns or trends and that past price movements cannot be used to predict future price movements.
In an efficient market at any point in time the actual price of a security is a good estimate of its intrinsic value. (Fama E. F., 1965(Fama E. F., -1974. Fama, states that, in uncertain world the intrinsic value can never be ascertained exactly. Thus there is always room for disagreement among market participants concerning what they deem is an intrinsic value of a security.
In an efficient market such discrepancies between actual prices and intrinsic values should cause the actual price of the security to wander randomly about its intrinsic value.
If the discrepancies between actual prices and intrinsic values are systematic rather than random in nature, then the knowledge of this should help market participants to better predict the prices towards intrinsic values. When many traders try to take advantage of this they will tend to neutralize such systematic behavior in price series. (Fama E. F., 1965(Fama E. F., -1974. In an efficient market on average competition will cause the full effect of new information on intrinsic value to reflect "instantaneously" in actual price. Instantaneous adjustment has two implications; first, it might lead to overadjustment or under-adjustment to the intrinsic value. Second, the lag in the complete adjustment of actual prices to successive new intrinsic values will itself be an independent random variable. This means that instantaneous adjustment property of an efficient market implies that successive price changes in individual securities will be independent. (Fama E. F., 1965(Fama E. F., -1974 A market where successive price changes in individual securities are independent is by definition, a random walk market. The theory of random walk implies that a series of stock price changes has no memory-the past history of the series cannot be used to predict the future. (Fama E. F., 1965(Fama E. F., -1974, Serial correlation coefficients and analysis of runs of consecutive price changes of the same sign can be used to test the rand walk model. If the statistical tests tend to support the assumption of independence, then one can conclude that the historical prices cannot be used to predict market prices. (Fama E. F., 1965(Fama E. F., -1974. If the random walk theory is valid and security exchanges are efficient markets, then stock prices at any point will represent estimates of intrinsic or fundamental analysis, additional fundamental analysis will be of value if there is additional information which is not fully considered in forming current market prices.

B. Efficient market hypothesis
Efficient market hypothesis holds that any new information about a firm is incorporated into share prices rapidly and rationally with respect to the direction and magnitude of the share price movement. Fama presented the complete efficient market hypothesis in his paper: "Efficient capital markets: A review of theory and empirical work" in1970. He defined the efficient market as the one where "…security prices at any time "fully reflect" all available information (Fama E. F., 1970). He further distinguished three forms of market efficiency: weak, semi-strong and strong. Weak form of market efficiency implies that the future stock price developments are completely random, and no technical analysis can consistently help bit the market. To prove this form of efficiency to be present one should find no serial dependence among stock prices, no stock price movement patterns can be consistent over time.
Semi-strong form implies that investors cannot earn abnormal returns by trading on the new publicly available information as the information is immediately incorporated into stock prices. Neither technical nor fundamental analysis can find a way for beating the market. The speed of stock price adjustment is the determinant of this form of efficiency which can be tested with the highest certainty using the event study's methodology (Fama E. F., 1991). If no miss-reaction or delayed reaction of investors is discovered after the news announcement, the market is semi-strong efficient. This form of market efficiency was researched. Strong form of the market efficiency claims that at any point in time the share prices have fully incorporated all available information (both public and private information) and nobody can gain excess returns. Usually to test this form of efficiency it is looked whether someone can consistently earn higher than market index returns for a long period of time and currently there is no clear answer to this issue. (Jansen, 1968) Claims that "on average the funds apparently were not quite successful enough in their trading activities to recoup even their brokerage expenses", which proves the prevalence of the strong form of EMH. However, in 1991 Fama  reviews mutual funds performance and finds that some mutual funds actually out-performed the market by gaining tiny excess returns. Nevertheless, it does not prove anything as when there are thousands of investors, according to the normal distribution it is probable that some will actually beat the market. From the event study point of view, strong form of efficiency can be rejected if there are pre-event abnormal returns that signal trading based on insider information.

C. Dividend Signalling Theory
Market players are likely to interpret a change in dividend as a change in management's views of future profit prospects for the firm thus triggering share returns. Ross states that if managers possess insider information, then the choice of a managerial incentive and of the financial structure signals information to the market (Ross, 1977). In a competitive market the inferences drawn from the signals will be validated (Ross, 1977). Investors observe the firm's sources of funds and its utilization of funds including the overall dividend and then collectively price the firm (John & Williams, 1985). In a perfectly competitive market, both managers who are insiders and investors who are outsiders regard the pricing function as fixed by market forces. At equilibrium, managers of firms with valuable future cash inflows distribute larger dividends and receive higher prices for their stock whenever the demand for cash by both their firm and its current stockholders exceeds its internal supply of cash (John & Williams, 1985). Fuller and Thakor postulate that firms can be classified into three groups based on their history of performance; firms with the poorest prior performance pay the lowest dividends because they are correctly identified and therefore nothing to signal and they do face a Low Free-Cash-Flow (FCF) problem. Firms with the best performance are correctly identified in the market, they have no signalling concerns and pay higher dividends to cope with the High Free-Cash-Flow (FCF) problem. However, firms with intermediate prior performance pay dividend to resolve Free-Cash-Flow (FCF) problem and also to signal the market in order to separate them from prior poor performance (Fuller & Thakor, 2010

IV. EMPIRICAL REVIEW A. Dividend Announcements
The initial task of conducting an event study is to define the event of interest . Dividends announcement signal the market players on the performance of a company listed in the market. Dharmarathne (2013) examines stock price reaction to subsequent dividend announcements and information efficiency in Sri Lankan Share Market with a sample of 61 major companies from those listed on the Colombo Stock Exchange (CSE), which had made 137 dividends announcements for the period 1998 to 2008. The Market Model Methodology was applied in generating the Abnormal Returns surrounding the event day. The significance of the CAAR for the event period was tested using the t-statistic. The author states that despite the pre and post announcements effect the market reaction did not yield considerable figures and therefore concludes that the market is semi strong form efficient subsequent to the dividend announcements, dividends contain considerable information and the market reacted in an efficient manner to new dividend information.
Research conducted at the NSE indicated that Dividend policy influence the share prices and therefore companies There is no significant difference in short-term risk before and after the announcement date of dividends. 0 : There is no significant difference in long-term risk before the announcement date and after the ex-date of cash dividends. The authors conclude that; in short run, an increase in variance was reported which was roughly two times greater than the non-information period. The probable reason for an increase in event-period returns volatility in short-run may be due to 'low information environment'. In long run, the stocks seem to be less volatile due to the less intensity at which information arrives and is incorporated into security prices when stock exchanges are open. There seems to be a shift in investors' focus in long-run on information content of dividend announcements leading to a marginal decrease in 60-day variance as compared to 120-day period. These results justify the signaling ability of cash dividends announcements (Anwar, Singh, & Jain, 2015). The implications of the study were that managers may employ dividend policies to influence stock prices. For short term investment, investors can earn capital gains by investing in stocks that prices are highly volatile, and this is during the dividend announcements window. Kabiru From the results of the study the summation of the two coefficients from the ARCH and GARCH models was greater than 0.5 which supported the weak form efficiency and does not support the hypothesis of strong form efficiency. The results of the study therefore conclude that investors cannot bit the market using historical data but there is a possibility of market players beating the market if they are in possession of private information (Johannes, 2015 Jarque-Bera test rejected the normality assumption for all the countries under the study (Alagidede, 2011). Return predictability was evident in all the stock markets under study, evidence obtained was inconsistent with weak-form efficiency. This could be attributed to, buy and hold strategy by investors, limited instruments, low turnover and weak legislation. Risk return trade off was analyzed, the risk premium is significant and positive for Kenya and Tunisia at the 5% level implying that investors are compensated for the risk premium absorbed. For the other countries the relationship between mean returns and variance or standard deviation is insignificant implying that investors consider other risks apart from market volatility (Alagidede, 2011). Alagidede tested the asymmetry in African stock returns by examining the asymmetry in volatility, a fall in the value of a firm's stock causes the debt-to-equity ratio to rise (Alagidede, 2011). The E-GARCH-M was used, the parameters were positive and significant for Kenya, Morocco, and Nigeria. The findings were significant and consistent with existence of leverage effect on these markets. The existence of long memory is inconsistent with weak-form market efficiency, the FIGARCH estimates indicated the presence of long-memory in all return series. After controlling the results for nonlinearity, conditional heteroscedasticity and time varying premier, the author concludes that returns are predictable both in the mean and variance. Long memory is a major feature of African markets. Alagidade goes further and states that, investors in Kenya and Tunisia may be compensated for the risk assumed, leverage effect and leptokurtosis are not unusual to developed countries, but also prevalent in African stock returns (Alagidede, 2011).

A. Research design
Empirical study was employed where tests were carried out for successive daily price changes of stocks trading at the NSE to determine the efficiency of the market. The design was adopted as it allows collection of large amounts of data from the target population. This design was useful in studying the randomness of stock price returns to test whether they reflect all available public information. A regression technique based on the event study methodology was used to determine existence of abnormal returns, parametric tests were performed to test the significance of the abnormal returns. This was supported by other authors who have

B. Target Population
The population of the study were all the companies quoted at the Nairobi Securities exchange. To ensure a well-represented sample size, it was important to have the entire population of listed companies and zero down to the companies that have data available for stock prices and dividend announcements. Secondary data was obtained from Bloomberg Professional Services over the period January 2009 to December 2015.

C. Sampling frame
In order to have a well-represented sample the sample consisted of companies that have continuously constituted the NSE 20 share index and/or traded for a period of three years from Jan 2009 to Dec 2015. A sample of each sector was considered in order to ensure that every sector was represented in drawing up the conclusions of the study. Apart from the security being part of the NSE 20 share index, the security ought to have paid dividend and traded during the event window and period. This will not only avoid issues of thin trading but also ensure we have data to check for abnormal returns.

D. Event Study Specifications
Three kinds of event windows were considered [-1; +1], with hypothetical event day being day 0, [-5; +5], and for every day from -5 to +5 separately. The aim of looking at the days from -5 to +5, was to look at the bigger view where due to inefficient markets the reaction to an event could continue for several days. Below is an event study timeline to illustrate; Estimation Period Event Window 55 days 5 days 5 days Where; R iτ is the return for the day τ for stock i, P iτ is the split-adjusted stock prices for the current trades, P iτ−1 is the split-adjusted stock prices for the preceding trades and D iτ is the dividend for the day τ, if available.

F. Normal Returns
To estimate the parameters of the normal return, the below regression should be performed; Where; Riτ are the day τ returns on the security i, Rmτ are the returns on the market portfolio in the estimation period, αi and βi are the coefficients to be estimated and 2 are the variance of the residual

G. Abnormal Return Calculation
The abnormal return is the actual return on a security during the event window minus an estimate of its normal return (expected return unconditional the event taking place) over a Where; , are abnormal returns for the day τ in an event window, are actual returns for the day τ in an event window, and ⟨ | ⟩ are expected normal returns for the day τ in an event window

H. Estimation of the Market Model
Under the general conditions; ordinary least squares (OLS) is a consistent estimation procedure of the market model parameters (Mackinlay, 1997). For the ℎ firm in the event time, the OLS estimator for the market model parameters for an estimation window of observation are; Where; is the return in event period τ for security , is the return in event period τ for market

I. Existence of Abnormal Returns
Given the market model parameters above, the abnormal returns can be derived; ̂= −̂−̂ (3.10) Therefore, it is true to say; ̂= − ⟨ | ⟩ = −̂−̂ (3.11) The abnormal Return is the disturbance term of the market model calculated on out of sample basis. Under the null hypothesis, conditional on the event window returns, the abnormal returns should be jointly normally distributed with a zero conditional mean and conditional variance (Mackinlay, 1997). Where; The below null hypothesis was tested; H0; Dividends announcements have no impact on the behavior of stock prices (mean or variance). Under H0 the distributional of the sample abnormal returns of a given observation in the event window is; ̂~( 0, 2 (̂)) (3.13)

J. Aggregation of abnormal returns to determine market reaction and price convergence;
Where ̂( 1 , 2 ) is the sample cumulative abnormal return (CAR) from 1 to 2 where; 1 < 1 ≤ 2 ≤ 2 . The CAR from 1 2 is the sum of the included abnormal returns; ̂( 1 , 2 ) = ∑2 = 1 (3.14) As 1 increases the variance of ̂ is; 2 ( 1 , 2 ) = ( 2 − 1 + 1) 1 2 (3.15) The distribution of the cumulative abnormal return under 0 is; ̂( 1 , 2 )~(0, 2 ( 1 , 2 )) (3.16) Given the null distributions of the abnormal return and the cumulative abnormal return, tests of the null hypothesis can be conducted. However, tests with one event observation are not likely to be useful so it is necessary to aggregate. (Mackinlay, 1997). It is assumed that there is no clustering i.e. there is no any overlap in the event windows of the included securities. The abnormal returns can be aggregated as follows; ̅̅̅̅ = 1 ∑= 1 (3.17) And for the Large 1 its variance is Using the above estimates the abnormal returns for any event period can be analysed. The average abnormal returns were aggregated over the event window using the same approach as that used in calculating the cumulative abnormal returns for each security for any interval in the event window; ̅̅̅̅̅̅ (

K. Diagnostic Tests for CAAR for information content of dividends Parametric Test -Ordinary Student t-distribution
To test the null hypothesis that the abnormal returns are zero. In practice because 1 2 is unknown an estimator must be used to calculate the variance of the abnormal returns. The usual sample variance measure of 1 2 from the market model regression in the estimation window is an appropriate choice (Mackinlay, 1997). Using this to calculate ( ̅̅̅̅ ) above 0 can be tested using; = ̅̅̅̅̅̅( 1 , 2 ) ̅̅̅̅̅̅( 1 , 2 ) 1/2~ (0,1) (3.23) The cumulative abnormal returns are assumed to follow a student-t distribution. The calculated t-value would be compared with the critical value of the t-statistic at N-1 degrees of freedom from the tables to determine the significance of the average cumulative abnormal returns. Parametric tests; which rely on the assumption that individual firm's abnormal returns are normally distributed and there is no clustering hence no cross-sectional correlation between the returns. Duncan Turere 1 , AFMJ Volume 5 Issue 10 October 2020 VI. RESULTS AND DISCUSSIONS Table I below summarises the number of firms that met the sample criteria per sector and the number of dividend announcements used as the event.    Table III below indicates the t-statistics (equation 3.23) for the Cumulative Abnormal Returns carried out for all the sectors in the study and the market during the event window. The Variance of CAAR (equation 3.21) was also calculated and the Standard deviation being the square root of the variance.  (Dharmarathne, 2013), (Bulla, 2016). However, for the Energy Sector, Insurance Sector and the NSE Market, the null Hypothesis, is accepted at 96% Confidence level, concluding that Dividend announcements have no impact on the behavior of stock prices for the two sectors and the market.

C. Price Convergence to new values
To test how quickly prices, converge to the new theoretically correct fundamental values after an announcement and whether the market reacts efficiently; we need to evaluate the stock prices movements during the event window; i.e. before and after the date of dividend announcements (refer to Fig. 3  below). During the event window we expect prices to react instantaneously to their true fundamental price after the announcement. Generally, from the data obtained post the event date the CAAR is not stable; an indication that it takes more than five days after the event for market prices to converge to the new theoretically correct fundamental values.

E. Abnormal Returns during the Event Window
If abnormal returns exist during the event window; market players can take advantage and beat the market buy either selling or buying securities. The descriptive statistics shown in Table II depict presence of abnormal returns across all sectors, with the Agriculture sector having the highest CAAR and the highest variance.

F. Summary of hypothesis testing
From the findings of the research dividend announcements do have an impact on the behaviour of stock prices for the Agricultural, Banking, Commercial, Construction, Manufacturing and Telecomm Sectors. However, for the Energy Sector, Insurance Sector and the NSE Market in general, dividend announcements have no impact on the behaviour of stock prices. In addition, there is empirical evidence of price movements before dividend announcements; an indication of information asymmetry. After dividend announcements there is a sluggish response of stock prices where it takes more than five days for prices to adjust to their correct values. This makes it possible for market players to profit from this inefficiency by either selling or buying stocks and earning abnormal returns hence the Nairobi Securities Exchange has not attained the semistrong form of market efficiency.

VII. CONCLUSION
From the findings of the study dividend announcements do have an impact on the behaviour of stock prices for the Agricultural, Banking, Commercial, Construction, Manufacturing and Telecomm Sectors. On the other hand, for the Energy Sector, Insurance Sector and the NSE Market in general, dividend announcements have no impact on the behaviour of stock prices. In addition, there is empirical evidence of price movements before dividend announcements; an indication of information asymmetry. Subsequently, after dividend announcements there is a sluggish response of stock prices where it takes more than five days for prices to adjust to their correct values. This makes it possible for market players to profit from this inefficiency by either selling or buying stocks and earning abnormal returns. Due to the above compelling evidence we confidently conclude that the Nairobi Securities Exchange is not semi-strong form efficient. Existence of Abnormal Returns before the event date draws questions on whether some information which is not available to the wider public may have licked. It is therefore interesting if a research can be conducted to test the possibility of insider trading at the NSE market. At the same time the speed at which stock adjust to information is of concern especially when no arbitrage profits are to be realized; we have evidenced a lot of changes on the market structure and a future research agenda would provide more insights to whether the changes have an impact on the speed of price formation.