Pertanika |. S<,<. Sci 8c Hum. 1(2): 171-177 (1993) ISSN: 0128-7702Universiti Pertanian Malaysia Press

The Dividend and Earnings Behaviour of Firms onthe Kuala Lumpur Stock Exchange

ANNUAR MD NASSIR and SHAMSHKR MOHAMADDepartment of Accounting and Finance, Faculty of Economics and Management,

Universiti Pertanian Malaysia,

43400, Serdang, Selangor, Darul Ehsan, Malaysia.

Keywords: Dividend per share, earnings per share, payout ratio

ABSTRAK

Terdapat banyak bukti berdokumen mengenai gelagaf dhidcn dan perolehan firma di negara-negara maju.l)i Malaysia, bukti tersiar mengenai isu-isu ini sangaf berkurangan. Kajian ini menyelidik mengenai gelagaldividen dan perolehan firma-firma yang tersenarai di Bursa Saham Kuala Lumpur (BSKL). Penemuan kajianini menunjukan: (i) keputusan dividen firma-firma ini sebahagiannya bergantung kepada perolehan lemasadan dividen niasa yang lepas; (ii) firma mempunvai dividen sasaran jangkamasa panjang yang dibentukberdasarkan kemampuan perolehan, dan (iii) perubahan perolehan firma berlaku secara rambang, inibermakna ada kemungkinan ramalan perolehan yang dibuat oleh para .malisis tidak berkesanan dari segiekonomi.

ABSTRACT

The dividend and earnings behaviour of firms in developed economies are well documented. In Malaysia,there is hardly an\ published evidence on these issues. This study investigates the dividend and earningsbehaviour of firms listed on the Kuala Lumpur Stock Exchange (KLSE). The findings suggest that: (i) thedividend da isions of these firms partially depend on their current earnings and past dividends; (ii) firms havelong-term target dividend which is condit ioned upon their earnings ability, and (iii) earnings changes o f firmsare random which implies that earnings forecasts by analysts might be of no economic significance.

INTRODUCTION

(a) Dividend Behaviour

There is a great deal of evidence in the financeliterature on the role of dividends in corporatepolicy, primarily concerned with providingexplanations on why firms actually pay dividends(Feldstein and Green 1983; Miller 1986).However, till today the role of dividends incorporate policy remains an unsettled issue.Dividend policy can be defined as the decision onhow much of the earnings should be paid as cashdividends to shareholders. There is also evidencethat firms in developed economies have a targetdividend payout ratio and only partially adjusttheir actual current dividend payout in anypayment period to a change in the basis for theirtarget payout ratio. Lintner (1956) suggested a

partial adjustment model of dividend

behaviour which views current dividend as afunction of past dividends and current earnings.This model or its modification has been appliedto U.S. data (Roy and Cheung 1985), I'.K. data(Ryan 1974) and Australian data (Shevlin 1982;Partington 1984) and the findings support theproposition that a target payout ratio and partialadjustment are reasonable representations of thedividend policy decision among firms indeveloped economies.

Lintner (1956) suggested the followingpartial adjustment relationship to explaindividend decisions:

where A Du : the change in dividend pershare of firm i from time t-1 to

Annuar Md. Nassir and Shamsher Mohamad

Dit* : the target dividend of firm i inperiod t

Dit, : the actual dividend of firm i inperiod t-1

c : the speed of adjustment individend to the differencebetween the target dividendand last period's dividend

a : intercept, anduu : a zero mean, constant

variance, non-autocorrelatederror term.

The target dividend, Di(* is assumed to berelated to a measure of earnings, E.t, such thatDit* = rEu, where r is firm is payout ratio. Hence,the above equation can be rewritten as:

A Dit = a, + c r Eit - c Dt_, + uit

Limner's model essentially suggests that changesin dividends of firm i is a function of earnings inthe current period t, and dividends in previousperiod.

(b) Earnings Behaviour

Earnings forecast is an important input forthe valuation of securities and has importantimplications for asset-pricing theories. Indeveloped capital markets, researchers haveaddressed various issues concerning earnings offirms. For example, Ball and Watts (1972)address the computation of earnings forecastsusing naive models; and Deschamps and Mehta(1980) address the same issue using complexmodels. Neiderhoffer and Regan (1972)investigated the incidence of security pricechanges with earnings behaviour, that is whetheror not positive (negative) price changes areobserved when earnings changes are positive(negative). If such relationships exist, then thereis economic rationale for forecasting earnings.

There is also substantial evidence thatearnings behave as a random variable at best witha time trend. For example, Fama and Babiak's(1968) study on American firms, Finn andWhittred's (1982) study on Australian firms andA riff and Johnson's (1990) study on Singaporefirms support the proposition that growth inearnings follow a random walk and therefore it isunlikely that one could meaningfully predictearnings changes.

Ball and Watts (1972) tested whether a naiveearnings forecasting model is as good as acomplex model and concluded that a naivemodel is as good as any prediction model.Deschamps and Mehta (1980) reportedessentially the same findings.

There is hardly any published evidence onany of these issues in Malaysia, which in turnconstrains broader discussion about findingsfrom other related areas such as dividendbehaviour. This paper provides evidence on thedividend behaviour of firms listed on the KualaLumpur Stock Exchange (KLSE) over a fifteen-year period using both the Lintner's model and asimpler model of the analysis of distribution ofchanges in dividends conditional on the changesin earnings. It also presents evidence on therandomness of earnings changes of Malaysianfirms.

The remainder of this paper is organized asfollows: firstly, a description of the data andmethodology used. Next, a report on thepreliminary findings, followed by the highlightsof the main findings, and the conclusion.

DATA AND METHODOLOGYThe data used for analysis in this study consist ofannual earnings and dividends of sixty randomlychosen firms for the period 1975 to 1989.Dividends and earnings data were gathered andverified from a variety of sources such as theInvestors Digest, financial newspapers, annualCompanies Handbook and company annual reports.The earnings per share (EPS) is estimated bytaking the yearly earnings before tax and dividingit by the number of outstanding ordinary shares.Dividends per share (DPS) is estimated by takingthe total amount of yearly dividends and dividingit by the number of outstanding ordinary shares.A total number of 900 observations for EPS andDPS respectively were used in the empirical workreported in the next section. The payout ratiowas calculated by dividing the DPS by EPS. Thecoefficient of variation (CV) for earnings(dividends) is estimated by dividing the standarddeviation of EPS (DPS) by its mean.

To verify the relationship between dividendand earnings patterns of Malaysian listed firms,the simple model and the Lintner's model wereused. The simple model explains the change incurrent dividends conditioned upon changes incurrent and past earnings using direction of

172 Pertanika J. Soc. Sci. & Hum. Vol. 1. No. 2 1993

Dividend and Earnings Behaviour of Firms on KLSE

changes, whereas the Lintner's model explainsthe change in current dividends based on currentearnings and past dividends using both sign andmagnitude.

The randomness of earnings changes ismeasured by the difference between theexpected and actual percentage of earningschanges. Using historical frequencies of changesin earnings of each firm across the market, theconditional probabilities of observing thesignificance of runs of earnings of the sampledfirms were estimated. Ball and Watts (1972) andDeschamps and Mehta (1980) reported that theuse of the above (simple) model does as good ajob as any other complex model.

PRELIMINARY FINDINGSEvidence from Table 1 suggests that the averageDPS were much more stable over the period ofstudy than the average EPS. The relativevariability of EPS is more than twice that of DPS.The Mann-Whitney U test confirms that at a =0.001 the average DPS is more stable than theaverage EPS. This implies that firms listed on theKLSE only change dividends based on theperceived change in the ability to pay dividends

in the long term and not on transitory earningschanges. This is consistent with the theory thatmanagement will only change dividends whenthey are confident that the dividends can bemaintained in the future. The mean coefficientof variation for EPS over the fifteen-year periodwas lower (0.50) compared to the meancoefficient of variation for DPS (0.63). Theaverage payout ratio over the period was slightlyover 33 per cent.

The relative constancy of mean DPS over thestudy period does not imply that listed firms didnot change dividends, as about 55.5 per cent of alldividend announcements were dividend changesfrom the dividends declared in the previousperiod (see Table 2). Of the 55.5 per centdividend changes, 28.3 per cent were dividenddecreases while 27.2 per cent were dividendincreases.

The average payout ratio for the period 1975-1989 is 33 per cent and has shown only a slightincrease from 31 per cent in 1975 to 34 per centin 1989, which is consistent with the assumptionof relatively constant payout ratio in the valuationliterature (Famaand Babiak 1968).

TABLE 1EPS, DPS and payout ratio of Malaysian listed firms by year

Year

197519761977197819791980198119821983198419851986198719881989

Average1975-1989

Mean

22.1926.0930.5732.8235.6634.3328.8624.6333.9817.3611.2411.05

1.1811.159.47

22.04

EPS in centsS.Dev.

27.1045.8855.9468.4251.5337.8041.2440.5779.7627.3427.7148.8034.5223.2224.85

11.05

C.V.

1.22

]c

]

]

1.761.832.081.45L101.43i.652.351.572.474.40

29.252.082.62

0.50

DPSMean

6.885.648.42

11.2412.5415.3711.518.54

14.553.922.822.341.223.653.19

7.46

in centsS.Dev.

18.9412.0415.5627.9130.5549.5729.8317.7442.545.687.815.547.088.705.96

4.68

C.V.

2.752.211.852.452.443.232.592.082.931.452.772.375.802.381.87

0.63

Payout ratioMean

0.310.210.280.340.350.450.390.350.420.230.250.211.030.330.34

0.33

PertanikaJ. Soc. Sci. & Hum Vol. 1. No. 2, 1993 173

Annual \1<1. Nasui and Shamsher Mohamad

TABLE 2Distribution In sign of A D , conditional on A E , A E , AE

Panel AE AE t AE ot t-1 t-2

No.

AD

%ofrow

total

No. %ofrowtotal

No. %ofrowtotal

Total Total

18160

5.7850.16

133148

42.7746.39

16011

51.453.15

311319

49.3650.64

Total 17S 28.25 281 44.60 171 27.15 630 100.00

7118377

5.079.82

56i32.91

344748152

24.64

41.96

32.88

64.96

9754155

70.29

48.2210.27

2.13

138112146234

21.90

17.78

23.17

37.14

Total 178 28.25 281 44.60 171 27.15 630 100.00

4444012394332

8.33

5.41

5.48

51.28

13.19

44.32

51.81

33.68

1222272953413463

25.00

29.73

36.99

37.18

58.24

46.59

40.96

66.32

324842926860

66.67

64.86

57.53

11.54

28.57

9.09

7.23

0

4874737891888395

7.62

11.75

11.59

12.38

14.44

13.97

13.17

15.08

Total 178 28.25 281 44.60 171 27.15 630 100.00

FINDINGS

(a) Dividends and Lagged Earnings - Simple ModelThe literature on dividends assumes that thecurrent dividend payments for a firm dependUpon both current and past profits. Thisassumption is verified in the Malaysian context byexamining the relationship between the changein dividend (A Djt or change in dividend for firm iin period t-1 to period t) and the change inearnings (A EJ in both current and past periods.The summary of the results of such a relationshipis presented in Table 2. This table shows thedistribution by the sign A D(, conditioned on thesigns of the per share earnings changes over thecurrent and two prior periods, that is A Eit, A Ei(1

and AEit_2.In Panel A of Table 2, when A Eit is > 0, 51.5

per cent of the cases have A Dj( > 0, whereas only5.8 per cent have A Dn < 0, with the rest

maintaining the dividend payments. In Panel B,when both A Eit and A Eit , are positive, theproportion of positive dividend changes is 70.3per cent, whereas when A Ei( is positive and A Eit,is negative, there is 48.2 per cent dividendincrease. When there are three consecutiveincreases in annual profits, the proportion ofpositive dividend changes is 66.7 per cent. Whenthere are two successive earnings increasepreceded by a decrease, 64.9 per cent of the caseshave an increased current dividend. Thesefindings provide evidence for a laggedrelationship between current and past earningschange and dividend changes. Table 2 also showsthat the effect of a change in earnings ondividends declines over time, which providesfurther evidence for a lagged relationship. Forexample, if two or three profit changes arenegative, the proportion of negative dividendchanges for the sequence ( — +)> is higher than

174 Pertanika |. Soc. Sci. & Hum. Vol. 1. No. 2 L993

Dividend and Earnings Behaviour of Finns on KI.SE

for the sequence (- + -) which in turn is higherthan for the sequence (+ — ) .

(b) Dividends and Lagged Earnings - Lintner's ModelAn empirical test of Lintner's model using dataon Malaysian listed firms from 1975 to 1989 wascarried out and the findings are reported inTable 3. The speed of adjustment, c , isapproximately 0.08. The implied target payoutratio, r, is 0.34, which is below that of developedeconomies reported in Ariff and Johnson (1990).This could be due to several possible reasons, oneof which is that the EPS used in the analysis is nota good measure of firm's ability to pay dividends.Possibly, the ability to pay dividends is bettermeasured by cashflow estimation.

Another reason for the low estimated targetpayout ratio may be due to certain characteristicsof the Malaysian equity market, which encourage

TABLE 3Lintner's partial adjustment model using

Malaysian data

Time Period cr

1975-1989 -0.002 0.0276 -0.08 0.22

firms to adopt a low target payout ratio. Forexample, large shareholdings by institutions maybe for strategic reasons rather than dividendincome. In inter-company shareholdings or cross-holdings, the parent companies are likely toretain the earnings of their subsidiaries forexpansion and therefore favour a low dividendpayout ratio. The differences in the tax systems ofMalaysia and of developed economies mayaccount for the different payout ratios. In theU.S., for example, dividend income is subjectedto double taxation. The dividends are paid toshareholders at net amount and are taxed at thecorporate tax rate. Net dividend income receivedby shareholders is further taxed at their ownpersonal income tax rate. In Malaysia, whtMishareholders receive cash dividends, the\ benefitfrom a tax credit equivalent in the limit to theamount of tax paid by firms under the Malaysiantax imputation system. Furthermore, to ecouragethe public to invest, various tax incentives arcoffered to investors in the form of tax exemptionsand minimum level of taxable dividend income.Hence, though Malaysian firms adopt a lowerdividend payout ratio, actual after-tax dividendpayout ratios may not be much lower than forlii ins in the developed economies. This suggeststhat Malaysian firms need not adopt as high a

TABLE 4( ak ulated percentages of earnings sequences

Panel A E, A E> RelativeFrequencies

Calculated Actual

XP(+).P(+)P(+).PRP(-).P(+)P(-).P(-)

24.3624.9924.9925.64

21.9017.7cS

23.1737.14

Total 1()().()() 100.00

12.0312.3412.3412.3412.6612.6612.6612.99

7.6211.7511.5912.3814.4413.9713.1715.08

Total 1 ()().()() 100.00

The relative frequencies of the eai [lings sign are given as: P(+) = 0.4936, P(-) = 0.5064

Pertanika J. Soc. Sci. 8c Hum Vol. 1. No. 2, 1993 175

Annuar Md. Nassir and Shamsher Mohamad

dividend payout ratio as firms in developedeconomies which are subject to different taxsystems.

The estimated equation in Table 3 suggeststhat about 22 per cent of change in currentdividends are explained by changes in currentearnings and the last period's dividends.Therefore the Lintner's model described partiallythe dividend behaviour of Malaysian firms.

(c) Randomness of earnings changes

Table 4 shows the expected and actualpercentage of earnings changes. These wereestimated using the historical frequencies of thechanges of earnings of each firm across themarket and estimating the conditionalprobabilities of the sequences of runs in earnings.If the actual earnings sequences are close toexpected sequences, this suggests that earningsoccurrences or changes follow the normalsequence expected in a run. The overall observedrelative frequencies in terms of the signs (+indicating earnings increases and - indicatingearnings decreases) are 0.49 and 0.51.

The expected percentage of differentearnings sequences as the weighted average ofthe observed sequences are computed and shownin column 5: the probability of earnings increaseoccurring consecutively for three periods[P(+).P(+).P(+)] is 0.1203 or 12.03 per cent.Columns 6 and 7 show that the expectedpercentages for different combinations ofearnings sequences are close to the actualfrequencies observed. For both the panels X andY, a chi-square test at 5 per cent level confirmsthat the earnings changes are random. Thissuggests that earnings changes (or at least theirsigns) are independent over subsequent annualperiods. These findings are consistent with thoseof the developed capital markets (Ariff andJohnson 1990).

CONCLUSIONThis study attempts to assess the dividendsbehaviour of firms listed on the KLSE, specificallywhether the current dividend decision isdependent on the current earnings and pastperiod's dividends and whether firms have a long-term target dividend. The findings usingLintner's model provide evidence that thedividend decision of firms listed on the KLSEpartially depends on their current earnings andthe past period's dividends. The findings from

Table 2 (simple model) suggest that thesequence of earnings increases is associated withthe decision to increase dividends whiledividends decreases are likely to occur whensequence of earnings is showing a decliningtrend. These findings support the notion thatfirms have a long-term target dividend which isconditioned upon their earnings ability.

This study also provides evidence thatearnings changes (or at least their signs) ofMalaysian listed firms are random (independent)and there is an even chance for a firm to reportearnings increases or decreases in any particularaccounting period. This implies that earningsforecasts by analysts might be of no economicsignificance. However, these findings are basedon average values over a large sample of firmsand not targeted to any particular firm. Asuperior analyst may be able to reward himselffor his efforts to forecast earnings changes of aspecific individual company.

REFERENCESANNUAR, M.N., M. ARIFF and M. SHAMSHER. 1991.

Market's reaction to accounting information:The case of earnings and dividendannouncements on KLSE. Akauntan Nasional(June Issue) 28-32.

ARIFF, M. and L.W.JOHNSON. 1990. Securities Marketsand Stock Pricing: Evidence From a DevelopingCapital Market in Asia. Singapore: Longmans(Singapore) Publishers Ltd.

BALL, R. and R. WATTS. 1972. Some time seriesproperties of accounting income. / Finan.27(3): 663-668.

DESCHAMPS, B. and D.R. MEHTA. 1980. Predictiveability and descripitive validity of earningsforecastings models./ Finan. 35(4): 933-950.

FAMA, E.F. and H. BABIAK. 1968. Dividend policy: anempirical analysis./ Amer. Statist. Assoc. 63(4):1132-1161.

FELDSTEIN, M. and J. GREEN. 1983. Why docompanies pay dividends. Amer Econ. Review73(1): 17-30.

FINN, E.J. and G.P. WHITTRED. 1982. On theusefulness of naive expectation of earningsper share as experimental benchmarks.Economic Records June Issue: 169-173.

LiNTNER, J. 1956. Distributions of incomes ofcorporations among dividends, retainedearnings and taxes. Amer. Econ. Review 46(1):97-113.

Pertanika J. So . Si i. fc Hum. Vol. I. No. 2 1993

Dividend and Earnings Behaviour of Firms on KLSE

MILLER, M.H. 1986. Behavioural rationality infinance: the case of dividends. / Bus, 54(4):451-468.

\ i IDKRHOFFER, V. and P. REGAN. 1972. Earningschanges, analysts' forecasts and stock prices.Financial Analysts Journal 28(3): 65-71.

PARTINGTON, G.H. 1984. Dividend policy and targetpayout ratios. Accounting and Finance 24(2): 63-74.

ROY, S.P. and J.K. CHEUNG. 1985. Target payoutand the association between dividends andshare prices. Accounting and Finance 25(2):57-76.

RYAN, T.M. 1974. Dividend policy and marketvaluation in British industry. / Bus. Finan. andAccoun. 1(3): 415-428.

SHEVLIN, T. 1982. Australian corporate dividendpolicy: empirical evidence. Accounting andFinance 22(1): 1-22.

WATTS, R. 1973. The information content ofdividends./ Bus. 46: 191-211.

WOOLRIDGE, J.R. 1983. Dividend changes andsecurity prices./ Finan. 38: 1601-1615.

(Received 26 June 1992)

PertanikaJ. Soc. Sci. 8c Hum Vol. 1. No. 2, 1993 177