Capital Structure Decisions and Its Impact on Dividend Payout Ratio during the Pre- and Post-period of Recession in Indian Scenario: An Empirical Study

Abstract

In this study, an attempt has been made to investigate the influence of capital structure decisions on dividend payout ratio for the companies belonging to BSE500 in India during the pre- and post-period of the recent global recession. The pre-recession period has been taken from 2001–2002 to 2006–2007 while the post-recession period from 2007–2008 to 2012–2013. The dependent variable taken into consideration is the Dividend Payout ratio and 10 independent variables which might have some impact on the dependent variable taken into consideration are Business Risk, Size (Log Sales), Size (Log assets), Growth Rate (Assets), Interest Coverage Ratio, Degree of Operating Leverage, Financial Leverage, Return on Assets, Tangibility and ‘Non-Debt Tax Shield. Logistic regression has been utilized in this study. It is found from the study that Growth Rrate (Assets) and Profitability (Return of Assets) are significant variables influencing the dividend payout ratio in the prerecession period, while Profitability (Return of Assets) and Financial Leverage are significant variables influencing dividend payout ratio in the post-recession period.

Keywords

Recession Capital Structure Dividend Payout Ratio Logistic Regression

Introduction

Dividend performance of a corporate entity largely depends on the dividend payouts, as influenced by the dividend policy of that particular entity. Determining the dividend policy has always increased the inquisitiveness of different researchers all over the globe because there is always a dilemma in the mind of the corporate management regarding the proportion to be paid as dividend to the shareholders and how much to retain as future provision.

Capital structure is a very important aspect of corporate finance. A number of literatures have dealt with various delicate issues of capital structure. This area is like a puzzle for many scholars all around the world. A number of past literatures of different scholars have tried to explore the variables influencing the capital structure decisions.

Before starting our research work, we scanned through the different past literatures dealing with the variables influencing the capital structure decisions. Overall we shortlisted 10 variables for our research work which could influence capital structure decisions. These variables were used as independent variables for our study. Dividend performance is measured through dividend payout ratio which was taken as dependent variable for the present study.

We have used logistic regression for our study because the dividend performance for a financial year can be ‘Good’ or ‘Poor’ for a company. A benchmark is set in this paper regarding the categorization of the dividend performance into ‘Good’ and ‘Poor’. We have used median to dichotomize the dividend performance rather than mean (average) as it is susceptible to the influence of outliers. Moreover, we have provided two examples of previous literary work which have also used median for classification purpose. One of the important advantages of using logistic regression is that it facilitates the dichotomization of the dependent variable in two categories, which is not provided in simple multiple regressions.

We have also observed from the past literature that the factors influencing dividend payout ratio changes from industry to industry. For example, the factors influencing dividend payout ratio in iron and steel industry would be differing from cement industry. It is one of the reasons why we have taken BSE 500 index for our research study as it comprises of 94 per cent of the market capitalization in Bombay Stock Exchange (BSE). Moreover, the BSE is oldest stock exchange in India. It covers all major 20 industries in the economy like cement, textile, iron and steel, banking, power generation and distribution, oil and gas, etc.

The entire period of our study has been segregated into pre-recession period (from 2001–2002 to 2006–2007) and post-recession period (from 2007–2008 to 2012–013). Recession have a negative impact on the entire economy of a country being characterized by decrease in overall production, employment as well as gross domestic product (GDP) of the country. India was not aloof from the impact of the global recession which basically originated from the sub-prime mortgage crisis in the US. In India, Index of Industrial production (IIP) growth reflected a drastic fall from 11.9 per cent in 2006–2007 to 8.7 per cent in 2007–2008. It was followed by again a steep fall in 2008–2009 to 3.2 per cent. The figures highlighted that the industrial production faltered in India during the period of recession. The growth rate of Indian economy also declined in terms of GDP at factor cost from 9.6 per cent in 2006–2007 to 6.8 per cent in 2008–2009. BSE Sensex, which is one of the barometers for testing the health of Indian economy, also witnessed a sharp decline from more than 20,000 points in December, 2007 to little more than 8,000 points in March, 2009. One of the greatest impacts on the Indian stock market was the withdrawal of the funds by foreign institutional investors to the extent of more than US$ 8000 million during the period of April to November, 2008. Apart from this, the recession also had an adverse impact on IT and IT enabled service companies like BPOs, KPOs, etc whose majority of the revenue was generated from the US market. Hence, we can observe that the adverse effect of the recession was felt in Indian scenario.

The above reasons motivated us to conduct the present study so as to understand the changes in the impact of capital structure decisions on dividend payout ratio during the ‘post-recession’ period from 2007–208 to 2012–2013 (i.e., 6 years) in comparison to the ‘pre-recession’ period from 2001–2002 to 2006–2007 (i.e., 6 years) in Indian scenario.

Literature Review

Some of the past works refereed by us while conducting the present research are given in brief as below:

Chen and Volpe (1998) have used median in their research paper to dichotomize the dependent variable in logistic regression to examine college student’s personal financial literacy.

Bauer (2004) has examined the factors determining the capital structure like profitability, tangibility, growth opportunity, volatility, etc. As per the empirical analysis, it was concluded that leverage is positively correlated to size and tax but is negatively correlated to profitability, tangibility and growth opportunity.

Kanwal and Kapoor (2008) have tried to identify the different variables that influence the dividend payout ratio of IT sector in Indian scenario. Statistical techniques of correlation and regression have been used to explore the relationship between key variables.

Gupta and Banga (2010) have tried to investigate into the various factors that determine the dividend decisions of a firm. The paper has utilized factor analysis to determine the different significant factors and then multiple regression analysis is performed on such factors. It is observed from the study that leverage and liquidity are considered to be significant factors determining the dividend policy decisions of Indian companies.

Lily, Venkatesh and Sukserm (2010) have examined the influence of a number of variables on the dividend payout ratio and concluded that financial leverage is significantly influencing the firm’s dividend payout ratio.

Ramachandran and Packkirisamy (2010) have investigated into the association of corporate leverage and dividend policy on a panel sample in the Indian scenario utilizing regression technique. It was concluded from the study that the dividend payout has an influence on the level of debt level in capital structure.

Rafique (2012) has concluded in his study that ‘Corporate Tax’ and ‘Firm’s Size’ are significant variables influencing the dividend payout.

Dutta, Bandopadhyay and Sengupta (2012) have made an attempt to understand the variables that determine the performance of the stock exchange in Indian scenario by utilization of logistic regression.

Acharya, Biswasroy and Mahapatra (2012) have made an attempt to investigate the determinants of the dividend behaviour for the companies belonging to BSE Sensex. The study suggests that earning per share and dividend per share are major variables having a significant influence on the dividend policy of the companies taken into consideration for the study.

Ranti (2013) has investigated into the determinants of dividends policy in the Nigerian stock exchange market for the year 2006–2011. The paper aimed to investigate the effect of financial performance, firm size, financial leverage and board independence on dividend payout decisions in Nigerian scenario by utilization of the regression analysis. A significant relationship has been revealed between firm’s financial performance and dividend payout decisions.

Yang, Zhang, Chai, Chen and Zhang (2014) in their study have utilized denture plaque and staining scores as dependent variable, which was dichotomized using median as statistical tool.

Objectives

The major objectives of the present research paper can be categorized as:

To develop a model through logistic regression which will help to predict the dividend performance during the pre-period as well as the post-period of recession.

To understand the impact of capital structure decisions on the dividend performance during the pre- and post-recession periods.

To validate the model developed during the pre- and post-recession periods (for the period of the study).

To understand the significant variables influencing the dividend performance during the pre- and post-recession periods and rank them accordingly as per their Wald statistics derived from logistic regression.

Dividend payout ratio has been used as the dependent variable and 10 independent variables related to capital structure decisions, namely, Interest Coverage Ratio, ‘Tangibility, Financial Leverage, Non-debt Tax Shield, Profitability (Return of Assets), Growth Rate (Assets), Degree of Operating Leverage, Business Risk, Size of Firm (Log Sales) and Size of Firm (Log Assets), is taken into consideration for the study.

Research Methodology

Source of Data

This study has been conducted based on secondary data of the selected companies of BSE 500 Index. BSE 500 is a major index in India comprising of 500 companies as per their market capitalization. Using Prowess database software, the list of all the companies belonging to BSE 500 Index in India has been determined. Using the same database software the data of required variables during the following periods were collected and used for this study.

2001–2002 to 2006–07 (i.e., 6 years as pre-recession).

2007–2008 to 2012–2013 (i.e., 6 years as post-recession).

Sampling and Population

There were certain companies among these 500 companies which either did not start its business or whose data for the period of the study were unavailable. The companies which did not start its business during any of this 12 years period (pre and post) were excluded. Those companies whose average dividend to average profit after tax is zero have also been excluded from the study because their dividend performance is certainly ‘Poor’. Hence, these automatic poor dividend performances are liable to be eliminated from the model because dividend performances (‘Poor’ or ‘Good’) have been determined on the basis of some criteria as discussed later (sub-section Statistical Methods). With the motive to remove outliers those companies which have paid excessive dividend (i.e., when average dividend amount is greater than the average profit after tax, i.e., DIV/PAT>1) are also excluded from the study. Hence, total of 419 companies were included for the study.

Statistical Methods

Binary logistic regression is used as statistical tool to understand the impact of capital structure decisions on dividend payout ratio. One of the important advantages of using binary logistic regression over simple regression is that the former facilitates the dichotomization of the dependent variable into two categories. The dividend performance of a corporate entity can be ‘Good’ or ‘Poor’ depending on the dividend payout ratio of the firm. Dividend performance is measured in this paper by dividend payout ratio.

In this paper, median is used for ascertaining the initial dichotomous classification on the available cases (i.e., 419 cases). Those companies whose average dividend performances (i.e., dividend payout ratio) for the period of study are greater than or equal to median are grouped as ‘Good’, otherwise are grouped as ‘Poor’.

There are a number of literatures which do support the utilization of ‘median’ for dichotomous classification of the dependent variables. Two of these literatures are given in the literature review (refer section two).

‘Good’ is coded as 1 and ‘Poor’ is coded as 0 for analysis purpose in the SPSS software.

It has been an endeavour in this research paper to develop a model which will help to predict the dividend payout ratio during the pre- and post-period of recession.

Logistic regression though itself is non-linear in nature but reflects function Z (which is linear) and involves the independent variables is taken into consideration for the study.

Z = M_{0} + M_{1} X_{1} + M_{2} X_{2} ........ M_{n} X_{n}

(1)

Here,_{M₀is the constant of the regression, M₁, M₂,_{…,_{M_nare the regression coefficients and X₁,_{X₂, …, X_nare the independent variables.}}}}

Hypothesis Framed

In order to study the impact of various variables related to capital structure decisions on dividend payout ratio, the following two null hypotheses are framed which needs to be tested in the research study:

H1: Capital structure decisions do not have any impact on dividend payout ratio for the companies belonging to BSE 500 during the pre and post-period of recession.

H2: There is no significant difference between the variables related to capital structure decisions having an impact on dividend payout ratio for the companies belonging to BSE 500 during the pre and post-period of recession.

Variables Used in the Study

Capital structure does have a great influence on the dividend payout ratio. Sometimes when the market is on a high due to favourable circumstances, the company uses capital structure as a leverage to increase the dividend payout ratio. In favourable circumstances, the company may increase the debt component in the capital structure to reap the benefits of tax shield advantage. This may have an impact on the profitability. Hence, the company can increase the dividend payout ratio. In case of unfavourable circumstances (example in recessionary times), the company may decide to lower the dividend payout ratio and prefer to retain more in such circumstances. In such cases, retained earnings may be a preferred source of finance rather than equity capital for the company.

Various variables taken for the paper along with their codes, details and formulas are given in Table 1.

Table 1.

List of Dependent and Independent Variables

Category	Variable Code	Variable_Name	Details
Dependent	DPR	Dividend payout ratio	Average Dividend/Average PAT (Profit after Tax)
Independent	1. ICR	Interest coverage ratio	Average PBIT/Average Interest Paid
	2. FL	Financial leverage	Average Total Debt/Average Total Asset
	3. BR	Business risk	Standard Deviation of PBIT/Average PBIT
	4. GR	Growth rate (Assets)	Compound growth rate of total assets
	5. DOL	Degree of operating leverage	% change in EBIT to % change in Sales
	6. SIZE_1	Size of the firm (1)	Log (Average Total Sales)
	7. SIZE_2	Size of the firm (2)	Log (Average Total Assets)
	8. TAN	Tangibility	Average Net fixed Assets/Average Total Assets
	9. NDTS	Non-debt tax shield	Average Depreciation/Average Total Assets
	10. Profitability	Return on asset	Average Earnings before Interest and Tax (EBIT)/Average Total Assets

Source: Compiled by the authors.

Dependent Variable

The dependent variable which has been dichotomized in this paper is the dividend payout ratio. This paper tries to investigate the various independent variables related to capital structure decisions which might have an impact on the dividend payout of the companies belonging to BSE 500.

Independent Variables

In the present study, we have selected those independent variables related to capital structure decisions (derived after a thorough literature review) which may have an impact on dividend payout ratio.

Profitability Ratio

Profitability is an important variable influencing the dividend payout as per the previous research conducted. The profitability has been taken in this study as Return on Assets.

Previous studies indicate that a positive significant relationship exists between Return on Assets and dividend payout. If the profitability is high, then the firm will be more stable and the firm’s net earnings will allow them to afford large free cash flow which will in turn result in payment of larger dividend to the shareholders.

Growth Rate

Growth rate has been defined in this study as the compound growth in terms of total assets. It is seen from previous studies that it does have an impact on the dividend payout ratio of the firms.

Growth Rate (GR): Compound growth rate of total assets

G R = {{(T A_{n} / T A_{0})}^{\land} (1 / n) 1}

Where

TA_n = total assets at the end of the observed period, that is, 2012

TA_o = total assets at the beginning of observed period, that is, 2001

N = number of observed period.

Financial Leverage

The amount of debt component in the capital structure does affect the dividend policy of the firms. If more debt component is introduced in the capital structure, the firm may reap the benefits of tax shield; hence, its profitability will also increase. With increased profitability, the firm may declare more dividends to the shareholders to keep them happy.

Size of the Firm

Size of the firm is considered as an important variable influencing dividend payout of the firms. It may be assumed that older and established firms will pay a larger amount of dividend to the shareholders while the newer and small firms will pay lesser amount of dividend to the shareholders and will retain more for future growth and expansion.

In this study, two proxies have been taken for size of the firm

Size_1 = Log (Average Total Sales)

Size_2 = Log (Average Total Assets)

For the financial and banking companies, where sales figure is not available, total income figure has been utilized as a substitute.

Business Risk

Business risk is defined in this paper as variability in profit before interest and tax. If the business risk in the firm increases, the firm may prefer to have lesser debt in the capital structure. Hence the debt–equity ratio will be affected. The change in the debt–equity mix will in turn influence the dividend payout ratio of the company.

Degree of Operating Leverage

It has been defined in this study as the variability in future earnings for the firm. An increase of operating leverage may have an adverse impact on dividend payout ratio.

\begin{matrix} D O L = A V G {(E B I T_{t} - E B I T_{t}^{}_{- 1}^{}) / E B I T_{t - 1}} / \\ A V G {(S A L E S_{t} - S A L E S_{t - 1}) / S A L E S_{t - 1}} \end{matrix}

Where

EBIT_t = earnings before interest and tax of tth year

EBIT_t-1 = earnings before interest and tax of t − 1 year

Sales_t = net sale of t^th year

Sales_t-1 = net sale of t − 1 year

Example:

t = fiscal year 2002

t − 1 = fiscal year 2001

Interest Coverage Ratio

With the increase of interest coverage ratio, the profitability of the firms will increase to bear the interest burden on the debt component of the capital structure. Therefore, more funds will be available for the shareholders to be distributed as dividends. The firms may pay more dividends and retain less with the increase of interest coverage ratio.

Tangibility

Tangibility has been defined in this paper as the ratio between the net fixed assets and total assets. It is observed from the previous studies that with the increase of tangibility, the company prefers to have high dividend payout ratio.

Non-debt Tax Shield

It is defined in this study as the tax shield enjoyed by the firm in terms of depreciation. If the non-debt tax shield increases, it will have an impact on the profitability of the organization. This in turn will influence the dividend payout ratio of the company.

We can say that it entirely depends upon the dividend policy of the different firms regarding the proportion to pay-off as dividend to the shareholders and the proportion to be retained in the business for growth and expansion so as to maximize the return and minimize the cost associated with it.

Durbin Watson Test

Before proceeding with our study, we have to check if the data have time series influence or is a stationary one. Durbin Watson test has been conducted using SPSS to check the nature of the data during both the pre- as well post-recession periods. The test has been conducted through two methods:

Method 1: Computation of average Durbin Watson taking the dependent variable (Dividend Payout Ratio) and all the independent variables together.

The result observed from Table 2 reflects that average Durbin Watson test result is 1.899 during the pre-recession period and 1.900 during the post-recession period. The test result is out of the range of –1.5 to +1.5 (closer to 2), which proves that the data are not a time series data and are stationary. Durbin Watson table also shows that d_u < d < 4 − d_u._{Thus, there is no auto correlation between the dependent and independent variables during both the pre- as well as post-recession periods. The result of Table 2 is produced as follows:}

Table 2.

Average Durbin Watson Test Result

	Pre-recession	Post-recession
Average Durbin Watson Test	1.899	1.900

Source: Computed by the authors.

Method 2: Computation of Durbin Watson test results taking the dependent variable and each and every independent variable individually.

It is observed from Table 3 that Durbin Watson test results for the dependent variable and all the independent variables individually is out of the range −1.5 to +1.5. For example, the test result for dividend payout ratio and financial leverage is 1.816 for pre-period of recession and 1.894 for post-period of recession. The result of Table 3 is produced as follows:

Table 3.

Durbin Watson Test Result (dependent variable and independent variable taken individually)

Dependent Variable	Independent Variable	Durbin Watson Test (Pre-recession)	Durbin Watson Test (Post-recession)
Dividend payout ratio	Financial leverage	1.816	1.894
Dividend payout ratio	Interest coverage ratio	1.822	1.883
Dividend payout ratio	Tangibility	1.818	1.926
Dividend payout ratio	Non-debt tax shield	1.816	1.889
Dividend payout ratio	Return on assets	1.829	1.898
Dividend payout ratio	SIZE_1(Log sales)	1.821	1.889
Dividend payout ratio	SIZE_2 (Log assets)	1.823	1.897
Dividend payout ratio	Business risk	1.818	1.888
Dividend payout ratio	Growth rate	1.826	1.887
Dividend payout ratio	Degree of operating leverage	1.869	1.875

Source: Computed by the authors.

We can conclude from the above analysis that the data do not have time series influence and are stationary. Hence, we can utilize binary logistic regression for the present study.

Empirical Analysis

Pre-recession Analysis

Correlation Study Results

Multicollinearity is the undesirable situation where the correlations among the independent variables are strong. Hence, if multicollinearity problem exists among the independent variables then the regression results will not provide correct results.

What is a multicollinearity problem?

Lewis-Beck and Michael in their book Applied Regression: An Introduction have stated that if the correlation among the independent variables is greater than or equal to 0.80 then multicollinearity problem is assumed to exist. The same logic has been applied in this paper to define high correlation among the independent variables to give rise to multicollinearity problem.

The multicollinearity problem is checked through correlation matrix.

Correlation matrix is developed through SPSS between ‘Dividend payout ratio’ and 10 independent variables. It is observed from Table 4 (Correlation Matrix) that none of the independent variables are having correlation greater that 0.8 hence we can safely deduce that multicollinearity does not exist among the independent variables.

Table 4.

Correlation Matrix (Pre-recession Period)

		FL	ROA	ICR	TAN	NTDS	Size_1	Size_2	BR	GR	DOL	DPR
Correlation	FL	1.000	−.095	−.084	.647	.104	−.061	−.124	.006	.101	−.002	.033
	ROA	−.095	1.000	.211	−.057	.099	.165	−.059	.114	.013	.031	.198
	ICR	−.084	.211	1.000	−.063	.004	−.016	−.046	.009	.002	−.006	−.004
	TAN	.647	−.057	−.063	1.000	.234	−.048	−.099	.048	.099	−.017	−.034
	NDTS	.104	.099	.004	.234	1.000	.112	−.182	.009	−.043	−.007	.032
	Size_1	−.061	.165	−.016	−.048	.112	1.000	.732	.025	−.042	−.039	.009
	Size_2	−.124	−.059	−.046	−.099	−.182	.732	1.000	−.001	−.029	−.031	−.074
	BR	.006	.114	.009	.048	.009	.025	−.001	1.000	.034	−.004	.043
	GR	.101	.013	.002	.099	−.043	−.042	−.029	.034	1.000	.011	−.141
	DOL	−.002	.031	−.006	−.017	−.007	−.039	−.031	−.004	.011	1.000	−.097
	DPR	.033	.198	−.004	−.034	.032	.009	−.074	.043	−.141	−.097	1.000

Source: Computed by the authors.

Note: ROA = Profitability (Return on Asset).

Logistics Regression Results (Pre-recession Period)

Logistic regression was run in SPSS 20.0 taking into consideration the dependent variable namely ‘Dividend payout ratio’ and 10 independent variables.

Classification Accuracy

The median of the dividend payout ratio is 30 per cent. Therefore, if in a year, the dividend payout ratio of the firms is more than 30 per cent, it is grouped as ‘Good’ and given a internal coding of ‘1’ by SPSS software otherwise as ‘Poor’ (internal coding is ‘0’).

Example: If X ltd company of BSE 500 has a average dividend payout ratio for the period of study (i.e., 2001–2002 to 2006–2007) of 31.66 per cent it is classified as ‘Good’, as its average dividend payout ratio is greater that median dividend payout ratio (30.0 per cent). Thus, it is a company having good dividend performance.

SPSS software has extracted the initial classification table (Table 5).

If we compare the initial and final classification table (i.e., after logistic regression) (Table 6), it can be observed that the accuracy of prediction in the latter table (60.1 per cent) is better than the former table (50.1 per cent).

It is observed from Hosmer and Lemeshow test (refer Table 7) that Chi-square value is 0.497 which is greater than 0.05; thus, we can say that there is not much deviation between the observed and predicted values.

Table 5.

Initial Classification Table (Pre-recession Period)

	Observed	Predicted
		DPR		Percentage Correct
		.00	1.00
Step 0 DPR	.00(Poor)	0	209	.0
	1.00(Good)	0	210	100.0
	Overall Percentage			50.1

Source: Computed by the authors.

Notes: a. Constant is included in the model.

b. The cut value is .500.

Table 6.

Classification Table (After Logistic Regression) (Pre-recession Period)

	Observed	Predicted
		DPR		Percentage Correct
		.00	1.00
Step 1 DPR	.00(Poor)	117	92	56.0
	1.00(Good)	75	135	64.3
	Overall Percentage			60.1

Source: Computed by the authors.

Note: a. The cut value is .500.

Table 7.

Hosmer and Lemeshow Test Statistic (Pre-recession Period)

Step	Chi-square	df	Sig.
1	7.376	8	.497

Source: Computed by the authors.

Developing the Model and Identifying the Variables Influencing the Dividend Payout Ratio during the Pre-recession Period

The variables along with their coefficients are produced in Table 8. The result of the Wald statistics reveals that Growth Rate (Assets) has the maximum contribution in influencing the dependent variable followed by Profitability (Return of Assets). It may be noted here that the significance level of other independent variables are more than 0.05, which means that these variables may be removed from the model.

Establishing the final model after removing the insignificant variables:

Those variables in the equation which are statistically insignificant variables are removed and again logistic regression is performed. Variables in the equation after the removal of statistically insignificant variables are given in Table 9.

Z function has been computed using the constant and coefficient value of the regression obtained from Table 9.

Z = - .033 + 3.294 * F L - 1.695 * G R

(2)

MODEL TO PREDICT THE DIVIDEND PERFORMANCE (Pre-recession period)

Hence, it can be observed from Table 9 that ‘Growth Rate (Assets)’ and ‘Profitability (Return on Assets)’ are the variables influencing dividend payout ratio for the firms belonging to BSE 500 during the pre-recession period.

Table 8.

Variables in the Equation along with Their β-coefficients and Significance Level (Pre-recession Period)

Variables in the Equation
		B	S.E.	Wald	df	Sig.	Exp(B)
Step 1a	FL	1.183	.668	3.137	1	.077	3.263
	Profitability	3.572	1.436	6.186	1	.013	35.590
	ICR	.000	.000	.840	1	.359	1.000
	TAN	−.849	.626	1.835	1	.176	.428
	NDTS	3.190	5.527	.333	1	.564	24.287
	SIZE_1	−.038	.192	.039	1	.844	.963
	SIZE_2	−.113	.201	.317	1	.573	.893
	BR	.036	.042	.731	1	.392	1.037
	GR	−1.894	.616	9.464	1	.002	.150
	DOL	−.004	.004	.678	1	.410	.996
	Constant	.278	.661	.177	1	.674	1.321

Source: Computed by the authors.

Table 9.

Variables in the Equation along with their β-coefficients and Significance Level (after the Removal of the Statistically Insignificant Variables, that is, Pre-recession Period)

		B	S.E.	Wald	df	Sig.	Exp(B)
Step 1a	Profitability	3.294	1.307	6.350	1	.012	26.945
	GR	−1.695	.568	8.898	1	.003	.184
	Constant	−.033	.208	.025	1	.875	.968

Source: Computed by the authors.

Post-recession Analysis

Similar to pre-recession period, the same process of analysis is repeated in post-recession period.

Correlation Study Results

It is observed from the correlation matrix that none of the independent variables are having correlation more than 0.80; hence, it can be assumed that multicollinearity problem does not exist among the independent variables.

The result of the correlation matrix during the post-recession period is given in Table 10.

Table 10.

Correlation Matrix (Post-recession Period)

		FL	ROA	ICR	TAN	NTDS	DPR	Size_1	Size_2	BR	GR	DOL
Correlation	FL	1.000	−.003	−.133	.645	.232	−.010	−.156	.005	.025	−316	.001
	ROA	−.003	1.000	.038	−.025	.284	.177	−.250	.087	−.053	−.261	−032
	ICR	−.133	.038	1.000	−.119	−.064	.062	−.068	−.081	.001	.060	.007
	TAN	.645	−.025	−.119	1.000	.309	−076	−.092	.018	−.020	−304	.024
	NDTS	.232	.284	–.064	.309	1.000	.013	−.269	.176	−.029	−.247	.006
	DPR	−.010	.177	.062	−.076	.013	1.000	−067	.003	−.070	−033	.002
	Size_1	−.156	−.250	−.068	−.092	−.269	−.067	1.000	.470	−.008	−.005	.004
	Size_2	.005	.087	–.081	.018	.176	.003	.470	1.000	−.004	−297	.003
	BR	.025	−.053	.001	−.020	−029	−070	−.008	−.004	1.000	−.040	−.010
	GR	−.316	−.261	.060	−.304	−.247	−.033	−.005	−.297	−.040	1.000	−024
	DOL	.001	−.032	.007	.024	.006	.002	.004	.003	−010	−.024	1.000

Source: Computed by the authors.

Note: ROA = Profitability (Return on Asset).

Logistic Regression Results (Post-recession Period)

Classification Accuracy

Median of the ‘Dividend payout ratio’ is 24.01 per cent. Therefore, for a particular year if the dividend payout ratio of the firm is greater than 24.01 per cent, it is classified as ‘Good’ or else is classified as ‘Poor’. The initial classification table is given in Table 11.

It is observed that the prediction of accuracy after conducting logistic regression (refer Table 12) is 60.7 per cent which is greater than the initial classification, that is, 53.5 per cent.

Hosmer and Lemeshow test (refer Table 13) reflects a significance level of 0.547 which is greater than 0.05; hence, the goodness of fit for the model can be established with accuracy.

Table 11.

Initial Classification Table (Post-recession Period)

	Observed	Predicted
		DPR		Percentage Correct
		.00	1.00
Step 0 DPR	.00(Poor)	0	195	.0
	1.00(Good)	0	224	100.0
	Overall Percentage			53.5

Source: Computed by the authors.

Notes: a. Constant is included in the model.

b. The cut value is .500.

Table 12.

Classification Table (After Logistic Regression) (Post-recession Period)

	Observed	Predicted
		DPR		Percentage Correct
		.00	1.00
Step 1 DPR	.00	86	109	44.1
	1.00	69	155	69.2
	Overall Percentage			57.5

Source: Computed by the authors.

Note: a. The cut value is .500.

Table 13.

Hosmer and Lemeshow Test Statistic (Post-recession Period)

Step	Chi-square	df	Sig.
1	6.909	8	.547

Source: Computed by the authors.

Developing the Model and Identifying the Variables Influencing the Dividend Payout Ratio during the Post-recession Period

The variables along with their β-coefficients and significance level in the equation are provided in Table 14. Wald statistics reveals that Profitability (Return of Assets) has the maximum contribution in influencing the dependent variable followed by Financial Leverage. It may be noted here that the other independent variables except profitability and financial leverage are having a significance level more than 0.05, which means they are not significantly influencing the dividend payout ratio of the companies belonging to BSE 500.

Table 14.

Variables along with Their β-coefficients and Significance Level (Post-recession Period)

		B	S.E.	Wald	df	Sig.	Exp(B)
Step 1a	FL	1.302	.620	4.408	1	.036	3.675
	Profitability	3.138	1.406	4.980	1	.026	23.054
	ICR	.000	.000	.507	1	.477	1.000
	TAN	−.518	.599	.748	1	.387	.596
	NDTS	−7.654	6.425	1.419	1	.234	.000
	SIZE_1	−.134	.181	.548	1	.459	.875
	SIZE_2	.029	.153	.037	1	.847	1.030
	BR	−.120	.157	.584	1	.445	.887
	GR	.023	.210	.012	1	.913	1.023
	DOL	−.002	.002	.447	1	.504	.998
	Constant	.085	.864	.010	1	.922	1.088

Source: Computed by the authors.

Note: a. Variable(s) entered on step 1: FL, Profitability, ICR, TAN, NDTS, SIZE_1, SIZE_2, BR, GR, DOL.

Establishing the final model after removing the insignificant variables:

Those variables which are statistically insignificant variables are removed, and again logistic regression is performed. Variables after the removal of statistically insignificant variables are given in Table 15.

With the help of the result from Table 15, Z function is calculated. The model developed is given as follows:

Z = - .527 + 2.991 * R O A + .746 * F L

(3)

MODEL TO PREDICT THE DIVIDEND PERFORMANCE (Post-recession period)

Hence it can be observed from Table 15 that ‘Profitability (Return on Assets)’ and ‘Financial Leverage’ are the variables influencing dividend payout ratio for the firms belonging to BSE 500 during the post-recession period.

Table 15.

Variables in the Equation along with Their β-coefficients and Significance Level (After the Removal of the Statistically insignificant variables, that is, Post-recession Period)

		B	S.E.	Wald	df	Sig.	Exp(B)
Step 1a	FL	.746	.391	3.636	1	.047	2.108
	Profitability	2.991	1.234	5.877	1	.015	19.907
	Constant	–.527	.243	4.690	1	.030	.590

Source: Computed by the authors.

Validity of the Model

The model that has been developed using logistic regression during the pre- and post-period of recession needs to be validated. The methodology adopted to validate the model developed has been explained with the help of examples.

Step 1: Calculation of the Z value for each company using the above model. The average values of the independent variables for the entire period of the study (i.e., 6 years for pre-recession period and 6 years for post-recession period) are substituted in the equation. For example, for X company, let the Z value be 0.88 during pre-recession period and 0.45 for post-recession period.

Step 2: Use the binary coding of dependent variable (Dividend payout ratio) for each company by using the same classification, that is, if the median of the dividend payout ratio for the period of pre-recession of a company is more than 30 per cent, it is classified as ‘1’ or else it is classified as ‘0’. For post-recession period if the median of the dividend payout ratio for the period is more than 24.01 per cent, it is classified as ‘1’ or else it is classified as ‘0’. Taking the same example as above: For X company, let the dividend payout ratio be coded as ‘1’, that is, good performance company for pre-recession period, and ‘0’ for post-recession period as poor performing company.

Step 3: Now using an excel sheet, if either of the two conditions given is satisfied then the model is be ‘True’. The two conditions are as follows:

Value as per Step 2 = 0 and Z value as per Step 1 < 0 is satisfied.

Value as per Step 2 = 1 and Z value as per Step 1 > 0 is satisfied.

Similarly, if neither of the two conditions is satisfied, the model is considered to be ‘False’.

The example for X company given in Step 1 and Step 2 can be concluded as below for pre- and post-period of recession.

Pre-recession period:

Z value as per Step 1 is ‘0.88’ and binary coding of the dependent variable as per Step 2 is ‘1’. Hence as value as per Step 2 = 1 and Step 1 > 0 is satisfied thus the model is ‘True’ for pre-recession period.

Post-recession period:

Z value as per Step 1 is ‘0.45’ and binary coding of the dependent variable as per Step 2 is ‘0’. Hence as value as per Step 2 = 0 and Step1 > 0 is satisfied, the model is ‘False’ for post-recession period.

Step 4: The total number of ‘True’ and ‘False’ are counted from the excel sheet. The main purpose of such an exercise is to understand whether the model is valid or false for each company taken into consideration. Then the number of cases having ‘True’ is divided by the total number of cases (i.e., 419) and the percentage of validation is calculated.

Following the above steps, the result of the validation is given as follows:

Pre-recession

‘True’ cases: 256

‘False’ cases: 163

Therefore, the percentage of validation of the model is 256/419 × 100 = 61.09 per cent.

Post-recession:

‘True’ cases: 239

‘False’ cases: 180

Therefore, the percentage of validation of the model is 239/419 × 100 = 57.04 per cent.

Hence, we can conclude from this section that the model developed through logistic regression has been validated to the extent of 61.09 per cent in the pre-recession period and 57.04 percent in the post-recession period.

Hence from the ‘Empirical Results’ section, two of the four basic objectives of our research paper have been fulfilled, namely, to (i) develop a model through logistic regression to predict the dividend performance during the pre- and post-period of recession and (b) to validate the model developed during the pre- and post-recession period during the period of the study.

Hypothesis Tested

One of the objectives of our research paper was to study the impact of capital structure decisions on the dividend performance during the pre- and post-period of recession.

The model developed during the pre- and post-recession period reflects the various significant variables related to capital structure decisions influencing the dividend performance of the companies belonging to BSE500. The models developed have been refined after removing the insignificant variables having p values greater than 0.05 (refer Tables 9 and 15).

The null hypothesis framed earlier has been accepted or rejected as below:

H1 is rejected as there is an impact of capital structure decisions on dividend payout ratio for the companies belonging to BSE 500 during both the pre- and post-period of recession.

H2 is rejected as there is significant difference between the variables related to capital structure decisions having an impact on dividend payout ratio for the companies belonging to BSE 500 during the pre- and post-period of recession.

Variables having an impact on dividend payout ratio are:

Pre-recession period: Growth Rate (Assets) and Profitability (Return on Assets).

Post-recession period: Profitability (Return on Assets) and Financial Leverage.

Conclusion

This empirical study has attempted to explore the impact of capital structure decisions on dividend payout ratio of companies belonging to BSE 500 during the pre-recession period (2001–2002 to 2006–2007) and post-recession period (2007–2008 to 2012–2013).

An attempt has been made in this research article to rank the significant variables influencing the dividend payout ratio of the companies belonging to BSE 500 as per their Wald statistics in Table 16.

We can observe from Table 16 that the ‘Growth Rate (Assets)’ and ‘Profitability (Return on Assets)’ are having an impact on the dividend payout ratio in pre-recession period while ‘Profitability (Return on Assets)’ and ‘Financial Leverage’ are having an impact on dividend payout ratio in the post-recession period.

The various variables related to capital structure decisions having an impact on dividend decisions are being briefly described below (refer Table 16):

Table 16.

List of Significant Independent Variables and Their Relative Importance in the Study (during the Pre- and Post-period of Recession)

Period	Significant Variables	Wald	Impact on Profitability (based on beta values)	Rank
Pre-recession	1) GR (Growth Rate)	8.898	Negative	1
	2) Profitability (Return of Assets)	6.350	Positive	2
Post-recession	1) Profitability (Return of Assets)	5.877	Positive	1
	2) Financial Leverage	3.636	Positive	2

Source: Computed by the authors.

It is observed from the study that Profitability (Return on Assets) is having a positive influence (on the basis of beta coefficient) on the dividend payout ratio for the companies belonging to BSE 500 during the pre- and post-period of recession. It means that dividend decisions are directly influenced by the financial performance of the firms. If the firm is more profitable then it is obvious that it will have more internal financing which will allow them to declare more dividends. It also confirms with the pecking order theory that the firm prefers to finance its activities by using internal financing first and then resort to debt and equity as a last option.

It is also observed that growth rate is having a negative impact on dividend payout ratio during the pre-recession period. With the increase of growth rate, the firms prefer to retain more rather than to pay higher dividends. It may be due to the fact that increase of growth rate attracts incremental investment expenditure. Due to high cost of external financing as a result of this, the firm will require funds which can be financed through retained earnings. Hence, the firm will declare lesser dividend and retain more in the business. This also confirms the pecking order theory where the firms are relying more on internal financing to finance its investment cost and then rely on debt and equity as external financing options.

It is very interesting to observe that financial leverage is positively influencing the dividend payout ratio of the firm during the post-recession period. With the decrease of leverage in capital structure, the firm prefers to pay fewer dividends. Due to the recessionary situation, the firms are wary of increasing more debt component in their capital structure as it may increase their financial risk and may lead to bankruptcy. Decrease in leverage of the firms may have an impact on its profitability as high debt firm seem to be more profitable due to the advantage of tax shield. Due to the decrease in leverage, the firms prefer to pay lesser dividend and aim towards retaining more future growth and expansion. This may be the reason why due to decrease in the capital structure of the firms, they prefer to declare fewer dividends and retain more.

These facts conclude that ‘Growth Rate’ and ‘Profitability (Return of Assets)’ are significant variables influencing the dividend payout for the firms belonging to BSE500 during the pre-recession period while ‘Profitability (Return on Assets)’ and ‘Financial Leverage’ are significant variables influencing the dividend payout for the firms during the post-recession period. The competent authorities should attach considerable importance to these variables while framing the dividend policy for the companies.

References

Acharya

P.N.

, Biswasroy

P.K.

, & Mahapatra

R.P.

(2012). Determinants of corporate dividend policy: A study of Sensex included companies. Indian Journal of Finance, 6(4), 35–43.

Dutta

, Bandopadhyay

, & Sengupta

(2012). Prediction of stock performance in the Indian stock market using logistic regression. International Journal of Business and Information, 7(1), 105–136.

Gupta

, & Banga

(2010). The determinants of corporate dividend policy. Decision, 37(2), 63.

Kanwal

, & Kapoor

(2008). Determinant of dividend payout ratios: A study of Indian information technology sector. International Research Journal of Finance and Economics, 15 (15), 63–71.

Lewis-Beck

M.S.

(1980). Applied regression: An introduction. Newbury Park, CA: SAGE Publications.

Lily

, Venkatesh

, & Sukserm

(2010). Determinants of dividend payout in Thailand. UBU Journal, 11(3), 73–81.

Mehta

(2012). An empirical analysis of determinants of dividend policy—Evidence from the UAE companies. Global Review of Accounting and Finance, 3(1), 18–31.

Miller

M.H.

, & Modigliani

(1961). Dividend policy, growth and the valuation of shares. Journal of Business, 34(4), 411–433.

Okpara

G.C.

(2010). A diagnosis of the determinant of dividend pay-out policy in Nigeria: A factor analytical approach. American Journal of Scientific Research, 8(8), 57–67.

10.

Patrik

(2004). Determinants of capital structure: Empirical evidence of Czech Republic. Czech Journal of Economics and Finance, 54(1-2), 2–21.

11.

Rafique

(2012). Factors affecting dividend payout: Evidence from listed non-financial firms of Karachi Stock Exchange. Business Management Dynamics, 1(11), 76–92.

12.

Ramachandran

, & Packkirisamy

(2010). The impact of firm size on dividend behaviour: A study with reference to corporate firms across industries in India. Managing Global Transitions, 8(1), 49–78.

13.

Ranti

O.U.

(2013). Determinants of dividend policy: A study of selected listed firms in Nigeria. Change and Leadership, 17(17), 107–119.

14.

Volpe

R.P.

, & Chen

(1998). An analysis of personal financial literacy among college students. Financial Services Review, 7(2), 107–128.

15.

Yang

, Zhang

, Chai

, Chen

, & Zhang

(2014). Multiple logistic regression analysis of risk factors associated with denture plaque and staining in Chinese removable denture wearers over 40 years old in Xi’an—A cross-sectional study. PLoS ONE, 9(2), 1–8.