AN EMPIRICAL STUDY OF DIVIDEND POLICY OF QUOTED COMPANIES IN NIGERIA

This study attempt to evaluate the observed dividend policy of a cross section of 27 Nigeria quoted companies using theories tested to explain dividend behavior of those firms. These theories which are several and varied; even contradict each other and considerable doubt exist as to which theory best represent the observed dividend behaviour of Nigerian firms; hence the need for this study. To carry out this study a more recent data for the period (1996 – 2006) were reviewed and a model with the necessary policy variables constructed. Factor upon which dividend decisions are based are identified and the magnitude of their effect estimated. Our estimation reveals that the traditional factors are significant in explaining and predicting their dividend decision within the period under review. The result provides strong support for the explanatory or predictive power of Lintner’s model. Also, factors which attempt to explain variations in share market prices were identified, and the magnitude of their effect estimated. The result confirms that share market price is a representation of market valuation of dividends.


INTRODUCTION
The finance profession has long struggled to develop a simple satisfactory model of dividend determination without much success.Modigliani and Miller (1961) show that in perfect capital market with no information asymmetry and predetermined investment decision, the value of the firm's is independent of the financing decisions.Hence, a firm's financing decision including dividends, have no effect on the value of the firm, nor the distribution of wealth between classes of security holders.
However, in an imperfect settings, dividend can influence shareholders wealth by providing information to investors or through wealth redistribution among claimants.
With information asymmetry, Bhattacharya (1979) demonstrates that dividends provide information about the firm's future cash flow and thus the dividend decision can change a firm value.Fama and Babiak (1968) and Jensen and Meckling (1976) demonstrated another potential real impact of financial decision transfer of wealth between classes of claimants can occur in the absence of imperfect priority rules.However, Kalay (1982) finds that firms generally are under these limitations.The payment of dividends conveys to shareholders that the company is profitable and financially strong.An increase in payment ratio signals to shareholders a permanent or long-term increase in firms expected earnings.Accordingly, the price of share may be affected by changes in dividend policy.
Dividend may offer tangible evidence of the firm's ability to generate cash, and as a result, the dividend policy of the firm affects the share price (Solomon, 1963).The market value of share is affected not because of the change in dividend but because of the information about changes in the future expected earnings conveyed through the payment (Pandey, 2000 pp.765).
It is contended that dividends are relevant because they have informational value.It is also believed that information content of dividend can go a long way to affect companies share market price by sending signals to prospective investors.Some of the pertinent problems are: why do companies pay dividend?
What actually informs the dividend policy?
What are the constraints of paying dividends and what should be the form of dividend?Do dividend matters?Of all the theories of dividend policy, which of one them best predict dividend policy behaviour in a specific case e.g.quoted firms in Nigeria?Can the magnitude of the factors that influence dividend policy be used in predicting market share prices of the firms under review?Answering these questions is absolutely not an easy task.Therefore, this study will seek to empirically analyse and evaluate, using conventional and non-conventional approach to investigate a number of factors related to these problems and seek how to evolve a long-term dividend policy and hence use the informational content of the dividend as declared by quoted firms.
The primary emphasis of this research work therefore is to identify the factors that influence the dividend policy of cross section of Nigerian quoted firms between 1996 and 2006 excluding 1998; to assess the stability of the result over time; and to test the relevance or applicability of dividend theories to share price behaviour in Nigeria.

THEORETICAL ISSUES AND LITERATURE REVIEW
Financing, investment and dividend decisions are the basic components of corporate policy.Financing decision requires an appropriate selection and combination of capital from available sources, investment decisions are concerned with the efficient deployment of capital funds while, dividend decision involves the periodic determination of proportion of a firms total distributable earnings that is payable to its shareholders.The larger the dividend paid, the less funds are retained for investment and the more the company will have to rely on other sources of long term funds (such as additional issues of equity and or debt capital) to finance projects.
In developed countries, the decision between paying dividend and retaining earnings has been taking seriously by both investors and management, and has been the subject of considerable research by economists in the last four decades (Lintner, 1956;Britain, 1964;Modigliani and Miller, 1961;Petit, 1976;Black and Scholes, 1974;Michael, Thaler and Womack, 1995;Dhillon and Johnson, 1994;Amibud and Murgin, 1997;Chariton and Vafeas, 1998) as cited in Adelegan(2001).
In Nigeria, until 1972/73 when the first indigenization decree was promulgated most quoted firms were foreign owned?
With the promulgation of the First and Second indigenization decree, foreign participation was restricted to forty per cent of the share capital.However, presently a major percentage of the sample firms in this study have foreign affiliation or investors.There is a disagreement over what type of investor is most interested in dividend.The question is whether individual investors, local investors or foreign investors are more interested in dividend than each other.
The argument centered on whether investors are expecting growth or cash flow.
According to Glen et al 1995, in many countries, management believes that local individuals and institutional investors are more interested in growth and re-investment of earnings than foreign investors who are more interested in dividend.
Multinational companies pay out proportionately more dividend than wholly domestic companies (Adelegan, 2001).
Dividend decision involves a trade-off between the retained earnings and issuing new shares.Over the years, the relationship between dividend policy and the value of the firm have been advanced by two school of thoughts of dividend theories.Those that claimed that dividends do not matter and those that claim they do.In summary, these theories can be grouped into two categories viz: -Theories which consider dividend decision to be irrelevant and Theories, which consider dividend decisions to be an active variable influencing the value of the firm.
The proponents of the dividend relevance school called the traditionalist or bird-in-hand propositions or rightists offered the first explanation for the relevance of dividend payment.Graham and Dodd (1934) founded the school.

Empirical Literature review
The earliest major attempt to explain dividend behaviour of companies has been credited to John Lintner (1956) who conducted his study on American Companies in the middle of 1950s.Since then there has been an ongoing debate on dividend policy in the developed 86 W. A. ADESOLA AND A. E. OKWONG markets resulting in mixed, controversial and inconclusive results.
This issues did not receive any serious attention among academic scholars in Nigeria until 1974 when Uzoaga and Alezienwa attempted to highlight the pattern of dividend policy pursued by Nigerian firms particularly since and during the period of indigenization and participation programme defined in the decree.Their study covered 52 company-years of dividend action (13 Companies for four years).They claimed that they "checked but found very little evidence" to support the classical influence that determine dividend policies in Nigeria during these period.They concluded that fear and resentment seem to have taken over from the classical forces.
However, Inanga (1977) and Soyode (1975) commented on the work of Uzoaga and Alozienwa.Inanga concluded that the problem arising from the change in dividend policy can be attributed to the share pricing policy of the Capital Issue Commission (CIC) which seemed to have ignored the classical factors that should govern the pricing of equity shares issues.This in turn made companies abandon "all the classical forces that determine dividend policy".
Soyode criticized Uzoaga and Alozienwa's work on the ground that it glossed over some important determinants of optimal dividend policy and questioned certain conclusions made in the study because they are inadequate or a mistaken evaluation.
Furthermore, Oyejide (1976) empirically tested for company dividend policy in Nigeria using Lintner's model as modified by Brittain.He disagreed with previous studies and concluded that "the available evidence provides a strong and unequivocal support for the conventional devices for explaining the dividend behaviour of Nigerian limited liability business organization.".Nyong (1990) conducted a study on dividend policy of quoted companies in Nigeria using the behavioural approach between 1983 -1987, to determine the factors that influence dividend policy of cross section of Nigeria quoted companies and also to assess the magnitude of these factors in predicting the observed share prices of the companies, he observed among others that the conventional Lintner's model performs creditably well.Adelegan (2001) in a more recent study of the impact of growth prospect, leverage and firm size on dividend behaviour of corporate firms in Nigeria between 1984 -1997; observed that the conventional Lintner's model does not perform quite creditably in explaining the dividend behaviour of corporate firms for the period under review.
Supports that factors that mainly influenced the dividend policy quoted firms are after tax earnings, economic policy changes (due to the partial liberation of the indigenization decree in 1989 and the subsequent simultaneous abolition of the indigenization decree of 1995), firm growth potentials and long term debts.
However, Adesola (2004) in his study of dividend policy behaviour in Nigeria using Lintner's model as modified by Brittan between 1996 -2000 appears to agree with Oyejide and Nyong's view that there is substantial and unequivocal support for the Lintner's model.

METHODOLOGY:
Data are derived from secondary sources.Pool of data were extracted from publication of the Nigerian stock exchange(NSE) factbook 2001, 2005, and 2007 editions, Best shares selection guide various issues published by Flarmark and company, SEC annual reports.The sample data used contains all the one hundred and forty-five companies quoted on the Nigerian stock exchange as at 2007.However, only annual reports of 27 companies have all the data that is required for this study.Samples cover 15 sectors of NSE.

Model Specification
The following models were built for the study: (a) Dividend Payment Equations 1. DIV

AN EMPIRICAL STUDY OF DIVIDEND POLICY OF QUOTED COMPANIES IN NIGERIA
Where: GRT = Growth = Growth proxied by market value of equity divided by book value of Assets, SZ = Size = Size proxied by natural logarithm of total asset i.e.In (total Asset), CS= Capital structure = Debt divided by market value of equity b 0, b 1 and b 2 are regression parameters, DIV t = Dividend payment in year t, DIV t-1 = Dividend payment in year t -1, EARN t = Total earnings in year t, GRT t = Firm growth in year t, SZ t = Size of firm in year t, CS t = Capital Structure in year t MDIV = Mean value of Dividend Payment in (1996 -2006 1997, 1999, 2000, 2001, 2002, 2003, 2004, 2005, and 2006

DATA ANALYSIS
In this section therefore, we carry out the analysis of the estimated results.The analysis is on equation basis, starting with the dividend payment equation as below: Equation 4 (DIV 2000).Testing the economic a priori, the constant term has a negative sign instead of the expected positive sign this implies that the autonomous leverage decrease when the explanatory variables are fixed.All the other parameters estimates are correctly signed in support of the priori expectation.The parameter estimate of previous year dividend payment i.e.DIV2000 is statistically significant at one percent level of significance this means that pervious year dividend payment exerts significant influence on current dividend payment and hence, this provide a strong support for the explanatory or predictory power of lintner's model.The coefficient of multiple determinations (R 2 ) of 0.93 or 93% indicates that about 93 percent variations in the observed behaviour in the dependent variables is explained by the model.The remaining 7% may better be accounted for by other error term.The high R 2 indicated that the model fits the data well and is statistically robust; there is a tight fit of the model.
The F-statistic 127.755 is significant at the 1% level considering the table F-statistic (F 0.01 , (2, 7) = 9.55).The calculated F-statistic is greater than the table F-statistic (i.e.‫ح557.721‬9.55), therefore, it is significant at 1% level.This buttress the fact that the high R 2 is better than would have occurred by chance.Another essential test is the second-order or econometric criteria; the DW-Statistic is 1.504, the table DW at 5% level indicates the following, given K 1 = 2 (excluding the constant term) and sample size (n) equals 10; then dL = 0.697, du = 1.64), 4-du= 2.359 and 4 -dL = 3.303.Based on the decision rule, the calculated DW of 1.504 lies between the lower dL (0.697) and upper du (1.641).There is inconclusive evidence regarding the presence or absence of positive first order serial correlation.
In equation 4.2, We regress DIV payment in the year 2005 on EARN for the year 2005 and DIV payment lagged one year that is 2004.The quantitative result shows that all parameter estimates including the constant term are correctly signed.Also, in this equation just like equation 4.1, the parameter estimate of previous year dividend payment i.e.DIV2004 is the only parameter that is statistically significant and is significant at 1% level of significance.Specifically, the DIV2004 comes out with an estimated coefficient of 0.961.This means that an increase of one percent in DIV2004 will increase the dividend payment in 2005 by 0.961.And hence, a strong support for the explanatory or predictive power of Lintner's model.The coefficient of multiple determination (R 2 ) of 87% shows that the proportion of dividend payment explained by the regression equation is quite high; the implication of this result is that the factor that enter into the decision calculus for dividend payment in the year 2005 is previous year dividend payment i.e.DIV2004.The Adjusted R 2 is equally significant at one percent level based on the result of F-statistic of 54.330 which is greater than the table F-statistic of 9.55 at 1% level of significance.Also based on DW-Statistics test, there is no serial correlation.
In equation 4.3, We regress DIV2006 on EARN26 and DIV2005.Testing the economic a priori, the constant term and the earning for year 2006 (EARN26) are negatively signed instead of the expected positive sign.Only dividend payment lagged one year that is DIV2005 is correctly signed.All parameter estimates in the quantitative result are statistically significant except the intercept.EARN26 is statistically significant at 10% level of significance while DIV2005 is statistically significant at 1% level of significance, specifically the EARN26 comes out with an estimated coefficient of -0.019 while DIV2005 come out with an estimated coefficient of 1.219.This result also provides a strong support for the explanatory power of Lintner model.The proportion of dividend payment explained by the regression equation is quite high, being approximately 93.9 percent.The implication of this result is that the factors that enter into the decision calculus for dividend payment in year 2006 in order of importance include past year dividend followed by current earning.The F-statistic of 85.261 is significant at the one percent level of significance considering the table F-statistic of 9.55.i.e. (85.261 > 9.55) this supports the fact that the high R 2 of 93.9 percent did not occur by chance.The DW test result of 1.244 lies between the lower dL =0.697 and the upper du = 1.641 region which suggest based on decision rule that there is inconclusive evidence regarding the presence or absence of positive firstorder serial correlation.

(b)
Determinant of share market price Two equations are formulated in this section to assist in explaining variations in share market prices.They go beyond the descriptive nature of Linter's model to provide motivations for the payment of dividend in the first place.The interpretations of the empirical result are as given below: In equation 4.4, We regress stock price for 2001 i.e. (SP2001) on dividend payment in the same year i.e. (DIV2001) and earnings per share in the year i.e.EPS2001.The quantitative result shows that: The constant term has the right sign (positive) and conforms to econometric a priori criteria.This means that when the independent variables or explanatory variables are zero, other factors not specified in equation 4.4 will still cause SP2001 to increase at the rate of 0.634 per cent.The quantitative result also shows that the sign of the coefficient of DIV2001 is in agreement with Gordon Model and also signaling theory of dividend policy.This implies that the realized positive relationship between stock price in 2001 and dividend payment in the same year is in line with theoretical expectation.Also, worthy of note is that the t-value of 2.864 is statistically significant at five percent two tailed test level of significance.Our estimated result of earnings per share in 2001 also indicate the relationship to be positive and statistically significant at one-percent level of significant and hence a strong support for the explanatory or predictive power of Gordon Behavioral model and Bhattacharya's signaling theory model.Equation 4.4 indicates that the best explanation for current share market price is current earnings per share (EPS2001) and current dividend (DIV2001).These two factors explain over 88 per cent of variations in share market price for the year under review.This is confirmed by the fact that the adjusted R 2 is statistically significant at one percent level based on the calculated fstatistic result of 89.544.The DW statistic (1.801) shows that there is no serial correlation in the residual of the model.Therefore, our estimates are reliable.
In 2 when tested with F-statistic is statistically significant at one per cent level of significance.The DW test for the incidence of serial correlation shows inconclusive evidence regarding the presence or absence of positive firstorder serial correlation.

Mean of Dividend Payment
For this section, we formulate one regression equation to determine the mean of dividend payment.In equation 4.6, we regress mean of dividend on mean of earnings per share (MEPS), mean of earnings (MEARN), mean of growth (MGRT) and mean of log of asset i.e. size (MSZ).The quantitative result shows that out of the four explanatory variables, it is only mean of earnings per share that is statistically significant, being significant at 10 per cent level of significance.The relevant implication of this result is that the influence of these three explanatory variables on the dependent variable is insignificant and could be considered as not accounting for variations in the dependent variable.The value of the adjusted R 2 of 3.8 percent implies that the equation does not give a good fit to the empirical sample data and the omitted variables might have performed better.The DW test shows inconclusive evidence regarding the presence or absence of positive first order serial correlation.

DISCUSSION OF FINDINGS
The main findings of this study are: 1.
The dividend policies of quoted companies in Nigeria are significantly influenced by their earnings and previous year dividend and that because of the reluctance to cut dividends, companies only partially adjust their dividends to changes in earnings.

2.
Average earning per share is the significant determinant of Average dividend payment, which confirms the fact that the most important decision calculus for payment of dividend is the current earning.

3.
Growth prospect and firm size has no impact on the dividend behavior of quoted firms in Nigeria for the period under review.

4.
That both current dividend and earnings per share or earnings are significant in explaining the observed differential share market prices of companies.The fact that the magnitude of the impact of earnings or earnings per share on share market prices is greater than that of dividend payment suggest that the main determinant of market share value for Nigeria firm is no longer dividend but earnings for recent data.This is inconsistent with findings by Graham, Dodd, and Cottle (1962), Nyong (1990), Adesola (2004) and this does not provide a strong empirical support for Gordon Models.

5.
The Nigerian market capitalizes the estimates of cash flows receivable by shareholders as dividend and hence that share market price is a representation of market valuation of dividends.6.
The empirical result of positive and significant effect of dividend payments on share market prices for the sample of Nigerian Companies indirectly cast some doubt on the empirical validity of Modigliani and Miller's preposition of Dividend irrelevance in the context of Nigerian business environment.

CONCLUDING REMARKS
In conclusion therefore, our empirical evidence indicates that the hypotheses of Lintner/Gordon as well as that of signaling theory of Bhattacharya performs remarkably well with respect to the dividend policy of quoted companies under review.this confirms previous result as cited in Nyong(1990), Adesola(2004) that Average earnings per share or average earnings is still the most significant determinant of average dividend payment, We also confirm that current dividend payment and earning per share are significant in explaining the observed differential share market prices of quoted firms in Nigeria.However, recent data reveal that the magnitude of the impact of earnings or earning per share is now greater than that of current dividend payment which used to be the most significant as reported in previous studies (see Nyong 1990, Adesola 2004).Furthermore, we also confirm that growth prospect and firm size has no significant impact on the dividend behavior of corporate firms; and this is inconsistent with the findings of Adelegan(2001).
Based on the findings from the study, we recommend as follows: (i) That government should assist in improving the quality and availability of secondary data bank available for research in Nigeria (ii) That the result of this study has at least one policy implication.
The fact that dividend is still an important determinant of share market prices means that companies may increase their share market price by increase in the rate of dividend paid.In order words, there is sufficient empirical evidence to believe that a liberal dividend policy will lead to a higher average market value of common stocks than will penurious dividend policies.
In effect we suggest that corporate management should follow a generous dividend policy which will maximize the long term benefits to its stockholders.

(iii)
Firms should try all their possible best in improving their total earnings from each transaction year, since recent study reveals that it now has greater impact than any other factor in determining the market share value for Nigerian firms from year 2001 till date.
) MEARN = Mean value of Earnings in (1996 -2006) MGRT = Mean value of Growth in (1996 -2006) MCS = Mean value of Firm's Capital Structure in (1996 -2006) SP t = Stock price for year t, PBT t = Profit Before tax in year t EPS t = Earnings per share in year t, ASSET t = Total Asset in year t DEBT t = Debt in year t, EQUITY t = Equity in year t PBT t-1 = Profit before tax in year t -1, EPS t-1 = Earnings per share in year t -1 ASSET t-1 = Total Asset in year t -1, DEBT t-1 = Debt in year t -1 EQUITY t-1 = Equity in t -1 ; and the 27 companies covered span 15 sectors namely: Automobile and Tyre (Dunlop Nigeria Plc), Banking (Access Bank Nigeria Plc), Breweries (Guinness Nigeria Plc and Nigerian Breweries), Building Materials (Ashaka Cement Plc, Cement Company of Northern Nigeria), Chemical and Paints (Berger Paints Nigeria Plc), Conglomerates (Chellarams Plc, CFAO Nigeria Plc, John Holt Plc, UAC of Nigeria Plc, UNILEVER equation 4.5, We postulate that stock price in 2006 is a function of Dividend payment in 2006 and earning per share (EPS2006) in 2006.The quantitative result shows that even though Dividend payment in 2006 indicate a positive relation, the t-value of 0.469 is not statistically significant.The implication of this is that DIV2006 has no significant influence on the determination of stock price value in 2006.Worthy of note is that our estimated result of earning per share in 2006 is statistically significant at one percent level of significance.This signifies that the most important factor in the decision calculus for stock price behavior in 2006 is earning per share.The measure of the explanatory power of the regression equation using adjusted R

Table F :
Standard Multiple Regression Result for equation 4.6