Tải bản đầy đủ
2 Large- cap firms in HOSE versus small-cap firms in HNX

2 Large- cap firms in HOSE versus small-cap firms in HNX

Tải bản đầy đủ

31

and Asness (2003). In more detail, both the small-cap firms in HNX and large – cap
firms in HOSE present this positive relationship between EEG and β (PR).
Firstly, comparing panel A with panel B, one important difference is the sample size;
23 versus 29. Another difference between the small-cap firms in HNX and the largecap firms in HOSE is the R2. In all models, the R2 for the firms on HOSE is higher
compared to the R2 of the HNX listed firms. This could mean that the impact of
dividend policy on future earnings growth is larger for listed firms on the HOSE
compared to listed firms on the HNX.
Table 4.3 panel data analyses subsamples

Panel A: HOSE

Mode I’ – EEG1YR

Model II’ – EEG2YR Model III’ – EEG4YR

Intercept (α)

-36.982

26.622

86.427

β(PR)

1.244**

1.162*

0.402*

T-test

3.880

3.106

0.645

R2

0.130

0.113

0.11

N

103

78

40

Panel B: HNX

Mode IV’ – EEG1YR

Model V’ – EEG2YR Model VI’– EEG4YR

Intercept (α)

-2.798

50.313

102.058

β(PR)

0.491*

0.675*

0.909*

T-test

2.401

2.672

2.464

R2

0.039

0.061

0.108

N

143

111

52

** *

, Significant at a 0.001; 0.05 level

The total sample is divided in two subsamples, respectively, the large-cap firms which have capital more than
350 billion vnd in HOSE and the mid-cap firms which have capital less than 150 billion vnd. The sample period
is 2006 - 2011. The dependent variable; The payout ratio which is calculated by dividing DPSi,t by EPSi,t. The
different levels of ‘annualized’ expected earnings growth are calculated by the change in EPS for t years, dividing
by t years. For example, EEG4YRi,t = (EPSi,t+4/EPSi,t -1)/4. The same method is used to calculate EEG1YRi,t and
EEG2YRi,t

32

4.3 Dividend yield and future profitability
To test hypothesis 2: The Dividend yield is positively correlated to the expected future
earnings growth of listed firms on the HOSE and HNX. Equation (3.2) is used.
EEGi,t+k = α + β(DY)i,t + ui,t+k(4.2)
Table 4.4 panel data analyses between EEG and DY

Mode I’’ – EEG1YR

Model II’’ – EEG2YR

Model III’’ – EEG4YR

Intercept (α)

49.488

119.206

192.685

β(DY)

-6.161**

-6.36*

-7.662*

T-test

-3.237

-2.624

-2.14

R2

0.052

0.043

0.057

N

192

154

78

Panel B: HNX

Mode IV’’ – EEG1YR

Model V’’ – EEG2YR

Model VI’’– EEG4YR

Intercept (α)

33.162

107.828

168.120

*

**

-5.233**

Panel A:
HOSE

β(DY)

-2.010

-3.386

T-test

-2.272

-3.129

-2.337

R2

0.027

0.063

0.076

N

187

147

68

** *

, Significant at a 0.001; 0.05 level

The total sample consists of firms on HOSE and HNX. The sample period is 2006 – 2011. The independent
variable; the dividend yield (DY) express the dividend per share as a percentage of the share price. The different
levels of “annualized” expected earnings growth are calculated by the change in EPS for t years, dividend by t
years. For example, EEG4YRi,t = (EPSi,t+4/EPSi,t -1)/4. The same method is used to calculate EEG1YRi,t and
EEG2YRi,t

The second hypothesis focuses on dividend yield. Is the dividend yield 7 positively
related to the expected future earnings growth of listed firms in the HOSE and HNX?
In table 4.4 the models are presented. In this model only the dividend yield is the
independent variable and the future earnings growth is the dependent variable. In all

33

models, the coefficients show that future profitability is negatively related to dividend
yield.

This negative relationship between dividend yield and future profitability is not
consistent with the result found for the relationship between expected earnings growth
and the payout ratio. Therefore, the second hypothesis is not true because a negative
relationship was found. This negative relation means that a lower dividend yield results
in higher expected earnings growth in the future years. What is different between the
payout ratio and the yield dividend? These two variables have the same numerator but
a different denominator. Indeed, the payout ratio is scaled with earnings and the
dividend yield is scaled with the share price. For example, a company with a high
dividend has a negative earnings growth in the future. This could imply that investors
are skeptical about the firm’s future. Could the firm continue to pay the current
dividend payout ratio? It is possible that a negative relation exists because of this
skepticism by investors. It should be noted that further research is needed to find the
best explanation for this negative relation.

7

The dividend yield expresses the dividend per share as a percentage of the share price.

34

4.4 An expanded model to forecast future earnings growth
4.4.1 Variable description and model building when adding DY
In this section the research elaborates the simple model with two variables to check the
robustness of the earlier developed models. First of all, a Pearson correlation matrix of
all variables is made to analyze the correlation between the independent variables
within model. The whole correlation matrix is presented:

Table 4.5 Pearson’s correlations Matrix
PR1YRVN
Pearson Correlation
PR1YRVN

1

Sig. (2-tailed)
N
Pearson Correlation

DYHOSE

DYHOSE
**

.444

.000
206

206

**

1

.444

Sig. (2-tailed)

.000

N

206

231

**. Correlation is significant at the 0.01 level (2-tailed).

As shown in the Pearson correlation matrix, this expanded model makes use of the total
sample which listed on HOSE and HNX. All independent variables are included in
matrix, respectively PR, DY. In correlation matrix showed that DY and PR are mutual
highly correlated.

Afterwards, to test two hypotheses, equation (3.2) is used. Equation (3.2) presents:

EEGi,t+k = α + β1 (PR)i,t +β2(DY)i,t + ui,t+k(4.3)
EEGt+k

Earnings growth (k = 1, 2 or 4 years)

PR

Payout ratio (DPS/EPS)

35

DY

Dividend yield (Total dividends / (share price * number of share))

Ut+k

Error term

The dependent variable of interest in Equation (4.3) is future net profit at t+1, t+2 or
t+4, measured by the Expected earnings growth (EEG). Also other researchers have
used the future earnings growth as the dependent variable, like Arnott and Asness
(2003).

The most important independent variable is the payout ratio of the firm. The payout
ratio is calculated by dividing dividends per share by earnings per share. Furthermore,
the dividend yield is included. Gordon (1962) already developed the constant-growth
valuation model. He has used the dividend yield and the expected return to forecast
growth. Furthermore, the earnings yield was added. Like Gwilym et al. (2006) have
documented the earnings yield is most of the time negatively and significantly related
to the earnings growth.

In table 4.6 the average models are showed for the expected earnings growth on an
annualized 1, 2 and 4 years basis. In this table only the PR and the DY are variables in
the model. This research additionally tests the expected earnings growth as a function
of only payout ratio and dividend yield.

Table 4.6 An expanded model when adding DY

Model I’’’

Model II’’’

Model III’’’

HOSE

(EEG1YR)

(EEG2YR)

(EEG4YR)

Intercept (α)

-21.166

40.599

106.557

β 1 (PR)

1.776**

1.873**

2.374*

β 2 (DY)

-6.453*

-5.695**

-9.174**

36

R2

0.173

0.145

0.149

N

178

140

71

HNX

Model IV’’’

Model V’’’

Model VI’’’

Intercept (α)

13.165

80.026

133.120

β 1 (PR)

0.645**

0.847**

0.965**

β 2 (DY)

-3.345**

-4.857**

-6.541**

R2

0.090

0.142

0.172

N

187

147

68

** *

, Significant at a 0.001; 0.05 level

The sample includes the listed firms on HOSE and HNX. The sample period is 2006 -2011. This model has two
variables. The payout ratio which is calculated by dividing DPSi,t by EPSi,t. Dividend yield expresses the
dividend per share as a percentage of the share price. The expected earnings growth is a dependent variable in
these models. The different levels of ‘annualized’ expected earnings growth are calculated by the change in EPS
for t years, dividing by t years. For example, EEG4YRi,t = (EPSi,t+4/EPSi,t -1)/4. The same method is used to
calculate EEG1YRi,t and EEG2YRi,t.

To analyze the differences of adding DY to the models, compare table 4.6 with table
4.1. By adding the independent variable dividend yield, the variable payout remains
positive. Notice, the beta’s of payout ratio increase by adding this variable.
Furthermore, R2 increases, too. This means the variable DY helps to explain more of
the variable in the dependent variable ‘future profitability’. The dividend yield is
negatively related to the future earnings growth in all models.

4.4.2 Variable description and model building when adding DY, EIBT, TA and
ROE.
In this section the research elaborates the simple model with some other variables to
check the robustness of the earlier developed models. Is the payout ratio still
significant within the model if some other variables are inserted in the model? First of

37

all, a Pearson correlation matrix of all variables is made to analyze the correlation
between the independent variables within the model. The whole correlation matrix is
presented in Appendix III. As shown in the Pearson correlation matrix, this expanded
model makes use of the total sample which includes the firms listed on the HOSE and
HNX. All independent variables are included in the matrix, respectively PR, DY,
EIBT, TA and ROE.
We have an equation:
EEGi,t+k = α + β1 (PR)i,t +β2(DY)i,t + β3(EIBT)i,t + β4(TA)i,t + β5(ROE)i,t + ui,t+k (4.4)
In Table 4.6, 4.7 the average models are showed for the expected earnings growth on
an annualized 1, 2, and 4 years basis.
Table 4.7 An expanded model for firms in HOSE when adding DY, EIBT, TA and ROE

Model I^

Model II^

Model III^

(EEG1YR)

(EEG2YR)

(EEG4YR)

Intercept (α)

-27.952

74.398

134.684

β1 (PR)

1.886***

1.512**

1.983*

t-test value of PR

5.274

3.199

1.965

β2 (DY)

-5.835***

-4.822**

-8.050**

t-test value of DY

-3.662

-2.318

-2.009

β3 (EIBT)

-1.100 E-8

-1.618 E-7

-7.915 E-9

t-test value of EIBT

0.134

-1.302

-2.185

β4 (TA)

-9.465 E-9

4.636 E-9

5.85 E-8

t-test value of TA

-1.244

0.372

1.642

β5 (ROE)

0.525

-0.378

0.36

t-test value of ROE

0.834

-0.471

0.249

R2

0.226

0.193

0.238

HOSE

*** ** *

, , Significant at a 0.001; 0.05 and 0.1 level

38

The total sample consists of the firms in the HOSE and HNX. The sample period is 2006-2011. This model has six
independent variables. The payout ratio is calculated by dividing DPSt by EPSt. Dividend yield expresses the
dividend per share as a percentage of the share price. ROE is added, this variable measures the rate of return on
the ownership interest (shareholders' equity) of the common stock owners. The sixth variable that is included in
the expanded model is ROA. ROA Calculated by dividing a company's annual earnings by its total assets. The
expected earnings growth is the dependent variable in these models. The different levels of ‘annualized’
expected earnings growth are calculated by the change in EPS for t years, dividing by t years. For example,
EEG4YRt= (EPSt+4/EPSt -1)/4. The same method is used to calculate EEG1YR and EEG4YR.

Table 4.8 An expanded model for firms in HNX when adding DY, EIBT, TA and ROE.

Model IV^

Model V^

Model VI^

(EEG1YR)

(EEG2YR)

(EEG4YR)

Intercept (α)

2.070

76.047

110.416

β1 (PR)

0.717**

0.852**

1.103**

t-test value of PR

3.543

3.299

2.628

β2 (DY)

-3.441**

-4.844**

-6.524**

t-test value of DY

-3.619

-4.280

-2.922

β3 (EIBT)

-1.020 E-7

-2.199 E-7

-3.730 E-7

t-test value of EIBT

-0.079

-1.112

-0.586

β4 (TA)

4.647 E-9

2.854 E-8

8.628 E-9

t-test value of TA

0.259

1.077

0.135

β5 (ROE)

0.391

0.076

0.852

t-test value of ROE

1.292

0.201

0.787

R2

0.099

0.151

0.182

HNX

** *

, Significant at a 0.001; 0.05 level

The total sample consists of the firms in the HOSE and HNX. The sample period is 2006-2011. This model has six
independent variables. The payout ratio is calculated by dividing DPSt by EPSt. Dividend yield expresses the
dividend per share as a percentage of the share price. ROE is added, this variable measures the rate of return on
the ownership interest (shareholders' equity) of the common stock owners. The sixth variable that is included in
the expanded model is ROA. ROA Calculated by dividing a company's annual earnings by its total assets. The
expected earnings growth is the dependent variable in these models. The different levels of ‘annualized’

39

expected earnings growth are calculated by the change in EPS for t years, dividing by t years. For example,
EEG4YRt= (EPSt+4/EPSt -1)/4. The same method is used to calculate EEG1YR and EEG4YR.

From Tables 4.7 and 4.8 that the relationship remains positive and only the PR and DY
are most of the time significant variables within the models. For that reason, this
research additionally tests the expected earnings growth as a function of only payout
ratio and dividend yield. The R2 of Model I^ - Model VI^ are in the range [0.099;
0.238]. All coefficients of the variable ‘payout ratio’ remain positive and all
coefficients of variable ‘dividend yield’ remain negative. These support two
hypotheses- the payout ratio is positively correlated to the expected future earnings
growth and the dividend yield is negatively correlated to the expected future earnings
growth.
Model I^ and Model IV^ explain the influence of the payout ratio and dividend yield of
listed HOSE firms and listed HNX firms on the 1 year expected earnings growth.
Models II^, III^, V^ and VI^ explain the relationship between the payout ratio,
dividend yield and respectively the 2 years annualized EEG and 4 years annualized
EEG for the listed firms in HOSE or HNX.
These models create a more detailed insight in the relationship. For example, the
coefficient of 1.512 in Model II^ implies that if the payout ratio increase 10% the
annualized 2 years expected earnings growth is 15.12%. This research focuses on the
average t-test to conclude whether the payout ratio variable and dividend yield are
significant within these average models. The average t-statistics, measured over a
period of six years, are above the minimum threshold of 1.96. For all the average
models it is assumed that the average t-statistic can be used to analyze the significance
of the variable within the model.

40

Chapter 5: Conclusions and Recommendations

This research has focused on the influence of the payout ratio and dividend yield on
expected future earnings growth for listed firms on HOSE and HNX. The model
developed by Arnott and Asness (2003) was used this research. Some new interesting
results about the influence of dividend policy and dividend yield on the future
profitability of listed firms on HOSE and HNX.

5.1 Conclusions with respects to the first hypothesis
After researched, it showed a positive relation on the indices level between dividend
distribution and future earnings growth of listed firm on HOSE and HNX. This
relationship on the indices level is in accordance with the results found by Arnott and
Asness (2003) for the United States. Therefore, the first hypothesis- The payout ratio is
positively correlated to the expected future earnings growth for listed firms on HOSE
and HNX- is confirmed.

Another important result of the panel data analyses is that the small-cap index (HNX)
exhibited smaller PR coefficients compared to the coefficients of the large-cap index
(HOSE). One possible explanation is that the HNX includes smaller firms which tend
to grow faster. For this reason, the listed firms on HXN have a higher retention rate and
a pay less dividend. These firms need their money to finance future growth. It should
note that this supports the life cycle theory. If the firm ends up in a more mature stage
of the life cycle, the firm pays more dividends. Indeed, the firms on HOSE are in a
more mature stage of their life cycle compared to the firms on HNX.

41

Besides, the R2 for the large-cap firms on HOSE is higher compared to the R2 of the
HNX listed small-cap firms. This could mean that the impact of dividend policy on
future earnings growth is larger for listed firms on the HOSE compared to listed firms
on the HNX.

The last, to find the differences of adding DY to the models, we compare table 4.6 with
table 4.1. By adding the independent variable dividend yield, the variable payout
remains positive. Notice, the beta’s of payout ratio increase by adding this variable.
Furthermore, R2 increases, too. This means the variable DY helps to explain more of
the variable in the dependent variable ‘future profitability’. When we add DY, EIBT,
TA and ROE that the relationship remains positive and only the PR and DY are most
of the time significant variables within the models

5.2 Conclusions with respect to the second hypothesis
The second part of this research focuses on relationship between dividend yield and
future profitability of listed firms on HOSE and HNX. However, in all yearly
regressions which were run for the period 2006 – 2011, DY was negatively related to
the future profitability.

In the end, an average expanded model with the variables payout ratio, dividend yield
were used to conduct an analysis. With this expanded model one can analyze whether
the payout is still significant after adding dividend yield.

5.3 Recommendations
During this research the positive relationship between payout ratio and expected
earnings growth was proved for listed firms on HOSE and HNX, during the period
2006-2011. It would be interesting for further research to analyze the possible