The Impact of Dividend Payout Policy on Performance of the Top 40 JSE Listed Firms

Introduction

Dividend payout policy remains a key topic in Corporate Finance, with Lintner contributing to the birthing principles in his 1956 paper along with Miller Modigliani’s 1961 dividend irrelevance hypothesis. Recent studies have aligned the significance of dividend payout policy, to tax, information and agency cost, in terms of determining financial performance of a firm. Researchers are still to reach an agreement on the optimal payout policy, that will enhance firm value, thus leading to an unsolved dividend riddle. A firm’s payout policy carries potential implications for changes in share prices and investors’ returns as it is an integral part of a firm’s larger financial ecosystem with important consideration on financing, investment and risk management (Farre-Mensa, Michaely, & Schmalz 2014) (Ross, 2002) (Foong, Zakaria, & Tan, 2007. Studies depict that payout policy affects firm performance as measured by the firm’s profitability and it must be a major determinant for a firm’s performance ((Amidu, 2007), (Brigham, 1995)). This study aims to investigate whether there is a direct relationship between payout policy and firm performance of the JSE top 40 listed companies.

This paper further extends on the methodology cited of our research proposal paper titled “THE IMPACT OF DIVIDEND PAYOUT POLICY ON PERFORMANCE OF THE TOP 40 JSE LISTED FIRMS” which aims to test if there is a relationship between dividend payout policy and firm’s performance. Our data looks at the period between 2009 to 2016, results of top 40 JSE listed entities.

Data Limitations

The top 40 listed firms vary year to year, which can distort results as the sampled entities can change annually. On the paper the sampled top 40 listed entities are as at 2019 and the study tracks their performance and payout policy from 2009 to 2016. The key factor and emphasise is to track and interpret top listed entities year on year performance returns versus their payout policy and not the entity’s name. On the final paper we will conduct a VLOOKUP year on year to calculate aggregate change in the overall entities listed annual see if the change is significant change in names.

Methodology

Packages Installed

The list below is packages installed on the R interface and included in the library to examine data. ‘data. table’which is a package used to manipulate data, subdivide rows,to enable to select and compute information in columns and calculate total by group. ‘dplyr’ which provides relatively easy to use functions when performing data analysis, allows for data to be transformed and tabulated. ‘ggplot2’ which enables interface to create graphics and map variables to aesthetics and provides more detail than primitive graphs. ‘tseries’ to convert a numerical vector into a time series which will unable to run normality test. ‘readxl’ efficient way to read rectangular data in excel format.

# install.packages("backports")
# install.packages("tidyr")
# install.packages("dplyr")
# install.packages("ggplot2")
# install.packages("moments")
# install.packages("forecast")
# install.packages("tseries")

library(moments)
library(ggplot2)
library(dplyr)
library(tidyr)
library(forecast)
library(tseries)
library(knitr)
library(readxl)

Importing Data

library(readxl) 
ROE <- read_excel("C:/Users/Tam/Desktop/Top 40 JSE LISTED CO'S.xlsx",
                  sheet = "ROE")
PE <- read_excel("C:/Users/Tam/Desktop/Top 40 JSE LISTED CO'S.xlsx",
                 sheet = "PE")
MBT <- read_excel("C:/Users/Tam/Desktop/Top 40 JSE LISTED CO'S.xlsx",
                  sheet = "MBT")
DP <- read_excel("C:/Users/Tam/Desktop/Top 40 JSE LISTED CO'S.xlsx",
                 sheet = "DP")
View(ROE)
View(PE)
View(MBT)
View(DP)

Pre-testing and Variable Creation

For us to analyse data, the first step involves examining the properties of the data.

Normality Testing

Test conducted to determine if the variables are symmetrical i.e. take the shape of a bell-shaped graph, are variables negatively or positively stewed?

Testing for Normality using Histogram on the ROE variables

library(ggplot2)
p1 <- qplot(ROE$`40178`, geom="histogram")
p2 <- qplot(ROE$`40543`, geom="histogram")
p3 <- qplot(ROE$`40908`, geom="histogram")
p4 <- qplot(ROE$`41274`, geom="histogram")
p5 <- qplot(ROE$`41639`, geom="histogram")
p6 <- qplot(ROE$`42004`, geom="histogram")
p7 <- qplot(ROE$`42369`, geom="histogram")
p8 <- qplot(ROE$`42735`, geom="histogram")

library(gridExtra)
grid.arrange(p1, p2, p3, p4, p5, p6, p7, p8, ncol = 4)

Testing for Normality using Jarque-Bera Test on the ROE variables

library(broom)
library(tseries)
library(knitr)
library(tidyr)
kable(tidy(jarque.bera.test(ROE$`40178`)))

statistic	p.value	parameter	method
69.2137	0	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`40543`)))

statistic	p.value	parameter	method
496.8081	0	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`40908`)))

statistic	p.value	parameter	method
97.17495	0	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`41274`)))

statistic	p.value	parameter	method
12.63585	0.0018037	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`41639`)))

statistic	p.value	parameter	method
18.44264	9.89e-05	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`42004`)))

statistic	p.value	parameter	method
9.077716	0.0106856	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`42369`)))

statistic	p.value	parameter	method
17.64217	0.0001476	2	Jarque Bera Test

kable(tidy(jarque.bera.test(ROE$`42735`)))

statistic	p.value	parameter	method
1335.76	0	2	Jarque Bera Test

kable(tidy(skewness(ROE$`40178`)))

x
1.906946

kable(tidy(kurtosis(ROE$`40178`)))

x
8.194463

Normality Test on MBV

kable(tidy(skewness(MBT$`40178`)))

x
NA

kable(tidy(kurtosis(MBT$`40178`)))

x
NA

Histogram

library(ggplot2)
p11 <- qplot(MBT$`40178`, geom="histogram")
p12 <- qplot(MBT$`40543`, geom="histogram")
p13 <- qplot(MBT$`40908`, geom="histogram")
p14 <- qplot(MBT$`41274`, geom="histogram")
p15 <- qplot(MBT$`41639`, geom="histogram")
p16 <- qplot(MBT$`42004`, geom="histogram")
p17 <- qplot(MBT$`42369`, geom="histogram")
p18 <- qplot(MBT$`42735`, geom="histogram")

library(gridExtra)
grid.arrange(p11, p12, p13, p14, p15, p16, p17, p18, ncol = 4)

Normality Test on PE

kable(tidy(skewness(PE$`40178`)))

x
NA

kable(tidy(kurtosis(PE$`40178`)))

x
NA

Histogram

library(ggplot2)
p21 <- qplot(PE$`40178`, geom="histogram")
p22 <- qplot(PE$`40543`, geom="histogram")
p23 <- qplot(PE$`40908`, geom="histogram")
p24 <- qplot(PE$`41274`, geom="histogram")
p25 <- qplot(PE$`41639`, geom="histogram")
p26 <- qplot(PE$`42004`, geom="histogram")
p27 <- qplot(PE$`42369`, geom="histogram")
p28 <- qplot(PE$`42735`, geom="histogram")

library(gridExtra)
grid.arrange(p21, p22, p23, p24, p25, p26, p27, p28, ncol = 4)

Normality Test on DP

kable(tidy(skewness(DP$`40178`)))

x
NA

kable(tidy(kurtosis(DP$`40178`)))

x
NA

Histogram

library(ggplot2)
p31 <- qplot(DP$`40178`, geom="histogram")
p32 <- qplot(DP$`40543`, geom="histogram")
p33 <- qplot(DP$`40908`, geom="histogram")
p34 <- qplot(DP$`41274`, geom="histogram")
p35 <- qplot(DP$`41639`, geom="histogram")
p36 <- qplot(DP$`42004`, geom="histogram")
p37 <- qplot(DP$`42369`, geom="histogram")
p38 <- qplot(DP$`42735`, geom="histogram")

library(gridExtra)
grid.arrange(p31, p32, p33, p34, p35, p36, p37, p38, ncol = 4)

Stationarity Test

Three stationarity tests were run per segment.

Stationarity Test on ROE

kable(tidy(adf.test(ROE$`40178`)))

statistic	p.value	parameter	method	alternative
-3.40756	0.0702073	3	Augmented Dickey-Fuller Test	stationary

kable(tidy(pp.test(ROE$`40178`)))

statistic	p.value	parameter	method	alternative
-39.38062	0.01	3	Phillips-Perron Unit Root Test	stationary

kable(tidy(kpss.test(ROE$`40178`)))

statistic	p.value	parameter	method
0.0751796	0.1	3	KPSS Test for Level Stationarity

Stationarity Test on MBV

kable(tidy(pp.test(MBT$`40178`)))

statistic	p.value	parameter	method	alternative
-33.78986	0.01	3	Phillips-Perron Unit Root Test	stationary

kable(tidy(kpss.test(MBT$`40178`)))

statistic	p.value	parameter	method
0.066735	0.1	3	KPSS Test for Level Stationarity
### Stationa	rity Test	on PE

kable(tidy(pp.test(PE$`40178`)))

statistic	p.value	parameter	method	alternative
-38.87409	0.01	3	Phillips-Perron Unit Root Test	stationary

kable(tidy(kpss.test(PE$`40178`)))

statistic	p.value	parameter	method
0.0969908	0.1	3	KPSS Test for Level Stationarity

Stationarity Test on DP

kable(tidy(kpss.test(DP$`40178`)))

statistic	p.value	parameter	method
0.2009999	0.1	3	KPSS Test for Level Stationarity

Linear Regression

The linear regression model developed for the study is as follows: \[DPit=β_0+β_1 ROE_it+β_2 MTB_it+β_3 PE_it+ε_it... (1)\]

reg <- lm(ROE$`40178` ~ MBT$`40178`)
reg1 <- lm(ROE$`40178` ~ PE$`40178`)
reg2 <- lm(MBT$`40178`~PE$`40178`)

library(broom)

kable(tidy(summary(reg)))

term	estimate	std.error	statistic	p.value
(Intercept)	14.6824560	4.348043	3.376796	0.0017364
MBT$40178	0.8411271	1.126661	0.746566	0.4600433

kable(tidy(summary(reg1)))

term	estimate	std.error	statistic	p.value
(Intercept)	16.4674653	3.5111733	4.6900179	0.0000366
PE$40178	0.0221877	0.0765439	0.2898688	0.7735349

kable(tidy(summary(reg2)))

term	estimate	std.error	statistic	p.value
(Intercept)	2.2198186	0.4854180	4.573004	0.0000523
PE$40178	0.0203509	0.0105822	1.923133	0.0621851

Normality Test Interpretation

Histogram

Looking at the visual inspection of the histogram diagrams we can deduce that the diagrams do not follow the normal bell curved shape thus implies that the distribution is non-normal. Further evidence is the “peakedness” of the histograms around the tails, variables are either negatively or positively skewed. We establish non-normality in the financial performance indicators and dividend payout measure, confirming that leptokurtosis is present in the return and divide payout distribution.

Jarque-Bera Test

A formal test for normality is the Jarque-Bera (JB) test, to examine skewness and kurtosis. The test on return reveals that p is zero, therefore the null of normality is rejected, that returns are normally distributed. The test agrees with the visual inspection which revealed leptokurtosis.

Linear Regression Interpretation

Examining the relationship between dividend payout ratio and financial performance indicators (ROE, MBT and PE), it is evident that the only variable that carries any statistical significant is the PE ratio, with a p-value of around 5%.