Determinants of Dividend Payout in Thailand's SET100

This paper examines the effects of risk (variability of profit), cash flow, investment opportunities (market to book value), firm size (market capitalisation), percentage of majority shareholders and financial leverage (debt ratio) on Thailand SET100 firm’s dividend policy. The results indicate that the stability of earnings (proxy risk) and financial leverage are significantly inverse related to firm’s payout ratio. It also implies that dividend policy is regardless of the firm’s cash flow, investment opportunities, firm size and agency cost. The implication of this study is if dividend policy does matter, the management cannot make the dividend decisions decision without taking consideration of the integration of business strategies including both financial and investment decisions.

1. Introduction

Dividend policy question has been a controversial issue since the introduction irrelevance of dividend policy theory by Miller and Modigliani (M&M) in the 1960’s when they believed in the world of efficient market, dividend policy does not affect the shareholder’s wealth. Basically, the principal hypotheses of dividend policy can be classified into signaling models, clientele effects, agency models, tax effects and free cash flow hypothesis (Frankfurter et al, 2004; Brav et al, 2005). There is an emerging consensus that there is no single explanation of dividend decision making (Abrutyn and Turner, 1990, Lease et al, 2000). Recent studies showed that the patterns of corporate dividend payout policies do not only differ across time periods (Pandey, 1995; Sarig, 2004) but also across countries (La Porta et al, 2000; Frankfurter, 2002) as well as between emerging and developed countries (Adaoglu, 2000; Aivazian and Booth, 2003).

An examination of corporate dividend policy practices in emerging countries are currently not well established in the literatures (Lease et al, 2000). Emerging markets differ from those in developed countries in terms of corporate governance (Mitton, 2004), taxation on dividends and capital gains (La Porta et al, 2000), and ownership structure (Lin, 2002). In addition, firms in emerging markets are subjected to more financial constraints than their counterparts in developed markets (Glen and Singh, 2004); they often have less information efficiency, more volatility, and are smaller market capitalization (Fuss, 2000; Bekaert and Harvey, 2003) which may have difference influence on their dividenpolicy. y.

As an example, in Adaoglu (2000) study, it showed that the emerging market firms followed unstable cash dividend policies and the main factor that determines the amount of cash dividends was the earnings of the corporation in that year. Aivazian and Booth (2003) also found out that companies in developing countries were shown to be less reluctant to change its dividends than their United States counterparts. These differences of the particular markets themselves raised the question about the extent to which the competing dividend policy theories could apply to such markets, in particular to Thailand.

This paper tries to address the determinants of dividend policy from a developing country perspective by focusing on SET100 firms in Thailand. We define dividend policy as dividend per share divided by earning per share before extraordinary item (Gul, 1998; Zeng, 2003; Amidu and Abor, 2006). We use variability of profit, cash flow, market to book value, market capitalisation, ownership of the majority shareholders, and debt ratio as proxies for risk, residual theory, investment opportunities, firm size, agency cost or clientele theory and financial leverage respectively. We adapted and modified the dividend policy model by D’Souza & Saxena (1999) in testing the dividend policy in international perspective. Their study findings indicated that the dividend payout ratio is significantly negatively related to institutional ownership of a firm’s shares and its risk but independent of investment decisions. However, they also suggested there were other factors determine dividend policy which we included in our study in order to find out to what extent the predictors which are mostly tested in developed countries applicable in developing countries.

The rest of the paper is organised as follows. Section two discusses the literature review on determinants of dividend policy. Section three discusses the empirical methods employed in this study. Section four describes the empirical analysis and Section five concludes the discussion.

2. Literature Review

Since Miller and Modigliani (1961) introduced the dividend irrelevance hypothesis and Black (1976) addressed “Dividend Puzzle” in their respective papers, there are numbers of researchers trying to solve the puzzle resulting development of theories in dividend policy. Lease et al (2000) argued by relaxing several of the assumptions of irrelevance dividend theory (taxes, agency costs and asymmetric information), the dividend policy may have impact on the share price.

Stability of Earning (Risk)

A firm that has relatively stable earnings is often able to predict its future earnings. Therefore, the firms with stable earnings are more likely to pay out dividends than the firms with fluctuated earnings. In Brav et al (2005), one of the main factors to determine the dividend decision is stability of future earnings and a sustainable change in earnings. Aivazian and Booth (2003) and Amidu and Abor (2006) study results show that dividend paa negative negative relationship with risk. Their study results also suggest that profitable firms with less variability in profit increase the ability of the firm to pay dividends. Meanwhile, in Nissim and Ziv (2002) study, they argued that under the signaling theory, dividend changes are related to firm’s future earnings changes not the past information leading to insignificant in relation.

Cash Flows

Residual dividend policy theory is an approach that suggests that a firm pay dividend dividends if all the acceptable investment opportunities for those funds are currently unavailable (Lease et al, 2000). Therefore, it implies that firms with higher cash flow tend to have higher dividend payout. Zeng (2003), Deshmukh (2005), and Amidu & Abor (2006) study results showed that, firms with high cash flow have the probability to pay high dividend dividends to their shareholders. However, Baker and Smith (2006) argued that most firms nowadays practice “modified” residual policy where the firms carefully managed their payout ratio and dithe investment eam after investment decision was made. Whexperiencems may consistently experienced low free cash flows, the dividegoal. licy was not necessarily a corporate goal.

Investment Opportunities

Both residual theory and agency cost theory have the different explanation towards growth opportunitithe residual r residual theory, companies with high growth opportunities tend to pay lower dividends because they may use the available funds to finance ththe positive ments with positive net t value. This implies implies that, given investment opportunities, a firm with higher cash flow or earnings tends to pay higher dividends (Deshmukh, 2005). Collins et al (1996), Gul (1999), Zeng (2003) and Amidu and Abor (2006) study results indicate that significant negative relationship between firm growth and dividend payout. Gul (1999) and Desha significant study findings also show significant negative relationship between growth opportunities and dividend yields meaningin comparison to gh firms. firms have low dividend yields compared to low growth firms.

Under signaling perspective, high investment opportunities may be associated with high dividends as high quality firms basically may pay dividends to signal their quality to the market (Easterbrook, 1984; Zeng, 2003). Meanwhile, under agency cost theory, high growth firms may pay dividends to restrict managerial discretion (Zeng, 2003). However, D’Souza and Saxena (1999) study results that in the context of international firms, it seems that dividend are paid irrespective of the firm’s investment opportunities. They indicatthese findings this findings support the Miller and Modigliani (1961) argument that investment decisions are independent of dividend policy.

Ownership Concentration

Ownership concentration has mixed explanation. Under agency cost theory, insider ownership and institutional ownership are inversely related to agency costs as the shareholders can monitor the management more effectively (Alli et al, 1993; Collins et al 1996; Han et al, 1999; Ang et al, 2000). However, under the tax-based theory, institutional ownership is positively related to dividend payout because of tax differential and clientele effect (Short et al, 2002) because institutions prefer dividends thagains. tal gains.

Firm Size

Collins et al (1996), Lee S.R (1997), Zeng (2003), Mitton (2004) and Deshmukh (2005) study findings also indicate that firm size has a relationship with the dividend payout. Collins et al (1996) argued that larger firms have more generous payout resulting positive relationship with dividend payout. Lee S. R (1997) study results show that large companies are indeed the ones that are more likely to pay dividends explaining the decision of whether to pay dividends or not. Zeng (2003) argued that if the firm size is positively related to diversification and decentralisation, the large the firm size, the less observable the actions of management and the higher agency costs may be incurred. Therefore, paying high dividends may reduce the agency cost. Mitton (2004) and Deshmukh (2005) indicated that the firm size proxies for symmetric information where the larger firms have lesinformation. Therefore, c information therefore pay higher dividends.

Financial Leverage

Zeng (2003) indicated that if financial leverage is used as one indicator of the future default and positively related to the cost of financial costs, paying dividends may increase the financial distress for a firm a high high leverage ratio is high. His study results show that the firm leverage (short term plus long term debt/total assets) is inversely related to dividend payout. Fenn and Liang (2001) results study also indicate that firm financial leverage (debt to assets ratio) is inversely related to firm’s payout ratio. Nash et al (2003) ssupportedso support the argument due to the inclusion of debt covenants to minimize dividend payments by the bondholders.

3. Data and Empirical Methods

The aim of this study is to find out the determinants of dividend payout for public companies in Thailand Stock Exchange by taking SET100 listed companies as a benchmark. The dividend policy of the firm is indicated as firm’s dividend payout (dividend per share divided by earning per share before extraordinary item) as used in some studies (Gul, 1998; Zeng, 2003; Amidu and Abor, 2006). We chose earning per share before extraordinary because diversion of resources may occur before earnings are reported, in which case overestimatesverestimate the share of true earnings that is paid out as dividends (La Prota et al, 2000) and also to measure the true earnings obusiness. pany ordinary business.

In our model, we presented six explanatory variables that may influence the probability of paying dividends namely risk (RISK), cash flows (CASH), investment opportunity (MTBV), percentage of majority shareholders (OWN), firm size (SIZE) and financial leverage (FIN). Our objective of choosing the proxies is to capture the management view as they are the one who made the dividend policy decision.

The dividend payout and the predictor variables used in this study are a five year average for the period 2001 to 2005 from SET100 firms. As the companies under SET100 index will be revised every six months, we selected the listed companies for January to June 2006 SET100 index calculation. The company financial data was collected through the online database at SETSMART [1] maintained by Thailand Stock Exchange. The data was annualithe calendar year. alendar year.

Multiple regression analyses are run to explain the relationship between firm’s dividend policy and the predictors. The study used the regression model used in D’ Souza (1999) study to explain the determinants of dividend policy. Any non-available data will be considered as a missing value. Before running the mregression’sressions testing, we deleted any extreme outliers and test for normality to make sure the variables are normal distribution to ensure our model is applicable. Table 1 shows the detail of predictor variables.

Our dividend payout model would be as follows:

PAYOUT = ¢0+¢1RISK+¢2CASH + ¢3MTBV+¢4OWN+¢5SIZE +¢6FIN+ ε

The following hypothesized relationships are predicted for each variable with respect to the dividend payout ratio:

H1a : MTBV, CASH, OWN and SIZE are expected to be positively related to PAYOUT;

H2a : RISK, OWN, FIN, MTBV are expected to inversely related to PAYOUT.

4. Empirical Results

Summary Statistics

Table 2 presents the descriptive statistics for all the regression variables. It shows the average indicators of variables computed from the SET100 company financial statements. The average (median) dividend payout ratio (measured as dividend per share/ earnings per share before extra ordinary item) is 37.42 percent [2] (36.39 percent). This means, on the average, SET100 firms pay about 37 percent of their profits as dividends and the average return on assets stands at 9 percent.

Average (median) risk (standard deviation of ROA) is 3.5 percent (3 percent) that can be considered lower. Cash flow, determined as the EBIT/ total assets had a mean (median) of 0.07 (0.11). This shows that in average EBIT represents 7 percent of the total assets for SET100 companies. Market-to-book value of the firms on average (median) is 1.36 (1.25). Average (median) percentage of majority shareholders is 56.29 percent (56.72 percent), suggesting that 56.29 percent of the companies shares are own by majority shareholders. Firm size determined as the natural logarithm of market capitalization has a mean (median) of 7.04 (7.04). The average (median) debt to total assets is 58 percent (57 percenimplyinging that in average SET100 firm’s total assets are financed through debt.

Correlation Matrix amongst Variables

Table 3 shows the correlation amongst predictors is between –0.416 (MTBV and FIN) to 0.404 (CASH and MTBV). Only two correlations have absolute value larger than 0.4 suggesting that multi-collinearity is not a major problem.

Regression Results

Table 4 shows that F-statistic is significant (F=3.834, p<0.05) at 95 percent confidence interval suggesting the model is useful to determine the variation in the criterion dividend payout.

As predicted, the results of this study show that risk and financial leverage are statistically significant negative relationship with dividend payout ratios, suggesting that, high-risk and highly financial leveraged firms pay lower dividends to their shareholders. Firms that are experiencing earning volatility find it difficult to pay the dividend. Therefore, therefore that such firms will have high probability to pay less or no dividend. On the other hand, firms with relatively stable earnings are often able to predict approximately what its future etherefore,ill be and therefore are more likely to pay out a higher pdividends. of its earnings as dividend. This result is consistent with Collins et al (1996), D’Souza & Saxena (1999), Aivazian and Booth (2003) and Amidu & Abor (2006) results study in relaticompany. vidend payout and risk of the company.

Meanwhile, highly levered firms usually have the obligation to pay interest to the bondholders at the first place. The bondholders may include bad covenants to the firms restricting the fidividendsthe certain dividend at certain level only. Paying dividends also may increase the firm fa high cial distress for firm with high leverage ratio leading to highly levered firms study end to have lower payout ratio. This result consistent with Fenn and Liang (2001), Zeng (2003) and Nash et al (2003) study findings.

The results also indicate cash flow is statistically positive relationship as predicted but insignificant with dividend payout. Here, the results are only partially reflected what we would have hoped where cash flow is significantly positive relationship to dividend payout (Deshmukh, 2005 and Amidu & Abor, 2006). If our model is correct, it implies that firms maintain its historical dividend payment without too much depends on its current cash as indicated in Baker and Smith (2006) study. It might be also consistent withistent to the signaling theory in Nissim and Ziv, (2001) study where dividend changes are related to firm’s future earnings changes. As the information for the cash is based on historical data, it might not reflect the future earnings resulting insignificant in relationship.

Market-to-book values are used as proxies for the firm’s future prospects and investment opportunities was found to have statistically positive associations but insignificant with dividend payout ratios. The results are inconsistent with the findings of previous studies (see Collins et al., 1996; Gul. 1999; Zeng, 2003 and Amidu & Abor, 2006) implies that emerging market has different characteristics that lead to difference findings in the developed markets. This may support the Miller and Modigliani (1961) theory as well as study of De’Souza and Saxena (1999) that investment decisions are independent of dividend policy. It implies that the SET100 companies may have stable dividends for the past 5 years.

Dividend payout also has a positive relationship but insignificant with firm size and ownership concentration. The positive relationship between firm and dividend payout is consistent with Collins et al (1996), Lee. S.R (1997) and Zeng (2003) results study that reflects agency costs. However, in our model, it seems that firm size is not significant indicating that dividend-paying company is regardless to its size. It implies that regardless the firm size is, the firms will cmanagey managed their dividend payout in order to maintain the stability of dividend payout. However, we must carethese findings since terpret this findings due to the fact that SET100 firm is maybe relatively low in Thailand. ation as they are top 100 public firms in Thailand.

If we considered institutional investors as the majority shareholders, tax-based theory has the better argument to explain the positive sign between dividend payout and ownership concentration because of dividend preference (Short et al, 2002). However, based on the insignificant regression results, the percentage of majority shareholders is not the main concern of the firms dividends. idends.

Based on these results, we can conclude that our Hypothesis H1a is not supported due to the insignificant relationship to dividend policy with 95 percent confidence interval. Hypothesis H2a is partially supported by which risk and financial leverage were negatively related to dividend payout with 95 percent confidence interval.

5. Conclusion and Recommendations

Conclusion

From the study, we may conclude that stability of earnings and financial policy of the firms are important variables that may influence companies to pay the low or hdividends.end. Their positive relationship with the dividend policy shows the willingness of the the higher dividends pay higher dividend when fewer obligations stable profit and less obligation to the lenders. It implies that the probability of the firmsa tight ay dividends is low if the company has tight obligation to pay interest to the bondholders due to financial distress or ithe longer on of debt covenants. Unstable earnings for longer period also may affect the ability of the company to predict its future earning resulting low in dividend payout.

It also suggested that, cash flows, investment opportunities, ownership concentration and firm size are regardless to dividend policy wherever the companies are due to their insignificant related to dividend policy. Besides, it seems that none of the main dividend theories (signalling, agency cost, clientele effect and cash flows) can really explain the behaviour of dividend payout in Thailand. However, the study suggested that the firm financial performance (earning) and financial policy (capital structure) may signal the information about the ability of the firms to pay dividends.

In overall, the results suggest that dividend does matter. If this is correct, the firms cannot decide dividend policy without taking consideration the integral part of business strategies including both financial and investment decision. Dynamic and characteristic change in the firm may require the firm to change its dividend policy if the firms want to maxithe value value for shareholders. An implication of this finding is that managers generally perceive that firms today set dividend payment in a manner consistent with that described by Lintner more than four decades back and may be different across the country because of the characteristics of the different market, objectives and strategies. Besides, investors who want to select the paying dividend firms might have to look into the two mentioned factors before selecting the companies.

Limitation

Caution should be taken into consideration to the limitation of this study. As our study concentrated on SET100 companiesa possibility possibility of non-response bias in our model for the remaining companies. The low R2 values for the regressions indicate other additional or appropriate explanatory variables are responsible, but the search for this problem is left for future research. However, it does not mean that our model is not applicable but getting more information may confirm our analysis as the low correlation amongst the predictors indicates that multi-colliproblem. is not a major problem.

Recommendations

Based on these findings, the following directions for future research should be considered: What determines dividend payout ratios of unquoted firms in Thailand? Why firms pay no dividends in Thailand? What determines the decision to pay or not to pay dividends in overall listed firms in Thailand?

The authors acknowledge Dr. Roy Kouwenberg, Ms Tin Moe Ni, Mr. Mori Kogid, Associate Prof. Dr. Fumitaka Furuoka, Mr. Charlie Albert and Ms. Jenny Ignatius for their valuable comments and assistance. The authors also acknowledge Asian Institute of Technology for the research grant.