The Netherlands: the ‘disappearing dividend puzzle’

The Netherlands: the ‘disappearing dividend puzzle’

Master of Science: Finance University of Groningen

Faculty of Economics and Business September, 2012

Author: M. Delleman Student number: 1705563 First supervisor: H. Gonenc Second supervisor: A. Plantinga

ABSTRACT

The percent of Dutch listed firms that pay out dividends declined from 88% in 1987 to 62% in 2010. This decline is partly explained by changes over time of the profitability, growth opportunities and the size of firms. This is consistent with the findings of Fama and French (2001). However, these traditional firm characteristics fail to explain the entire lower propensity to pay out dividends over time. However, when the variability of the net profit is added to the regression analysis, the lower propensity to pay over time disappears. This finding provides support for the signalling theory of dividends.

JEL classification: G35

Keywords: dividends, payout policy, signalling theory

I. Theoretical background...5

A. Dividend policy...5

B. Theories to explain dividend policy...5

B.1. Pecking order theory...5

B.2. Agency theory...6

B.3. Life-cycle theory...6

B.4. Signalling theory...7

B.5. Catering theory...7

B.6. Flexibility argument ...8

C. Influence of firm characteristics and general lower propensity to pay over time ..8

D. Differences between the U.S. and Europe...12

E. The Dutch situation...14

F. Discussion and criticism on the Fama and French (2001) study...15

II. Research method...17

A. Data selection and data description...17

B. Methodology...18

C. Hypotheses ...21

III. Empirical results ...23

A. Summary statistics ...23

B. Results ...29

IV. Conclusions...40

V. Appendix ...43

VI. Reference list...43

There has been done a lot of research on dividend policy. The fact that there are many firms that pay out dividends is interesting, because the tax rate on dividends is higher than on capital gains. Theories like the pecking order theory, agency costs theory, life- cycle theory, signalling theory, catering theory and the flexibility argument help to explain why firms decide to pay out dividends.

Recent studies find a decline in the number of firms that pay out dividends over time.

Fama and French (2001) examine if this decline can be explained by changing firm characteristics or a general lower propensity to pay over time. They find that profitability, investment opportunities and size explain a part of the decline in the number of non-financial and non-utility firms that pay out dividends in the U.S. from 1978 to 1999. These findings are confirmed by DeAngelo, DeAngelo and Stultz (2006) in the U.S., Fatemi and Bildik (2012) in a global context, Denis and Osobov (2008) in developed countries and Reddy and Rath (2005) in emerging markets. Fama and French (2001) call the part of the decline that they are not able to explain with firm characteristics: a general lower propensity to pay out dividends over time. Reddy and Rath (2005) call this phenomenon the disappearing dividend puzzle. Van Ees, Von Eije and Hooghiemstra (2008a) and Haan (1995) find that, in addition to the traditional Fama and French (2001) firm characteristics, the variability of the net profit has a highly significant influence on the decision to pay out dividends in the Netherlands.

This study explains the decline in the number of non-financial and non-utility firms that pay out dividends and are listed to the AEX, AMX, ASCX or the Dutch local market.

The number of Dutch firms that pay out dividends has declined after 1997. In 1987 88%

of the firms did pay out dividends and in 2010 the percent of paying firms has decreased to 62%. First, a test determines if the traditional firm characteristics of Fama and French (2001), profitability, investment opportunities and size, can explain the decline in dividend paying firms in the Netherlands or if there still remains a lower propensity to pay out dividends over time. It is hypothesized that firms that are more profitable, have less investment opportunities and are larger, are more likely to pay out dividends.

Second, calculations show if the unexplained general lower propensity to pay out

dividends can be explained by including the variability of the net profit in the regression

analysis. Two cross-sectional regressions show the difference between the regression

results with and without the variability of the net profit. In the first regression only the

proxies for profitability, investment opportunities and size are taken into account. The

variability of the net profit is included in the second regression. The regression results show the percent of the decline in the number of firms that pay out dividends that can be explained by firm characteristics and the part that is attributed to a general lower propensity to pay over time.

The variability of the net profit is included in the second regression, because this firm characteristic is found to have a significant negative influence on the number of firms that pay out dividends in the Netherlands. (Van Ees, Von Eije and Hooghiemstra, 2008a) The results of Haan (1995) are in agreement with the findings of Van Ees, Von Eije and Hooghiemstra (2008a). He claims that the negative influence of the variability of the net profit is proof of the signalling theory. Firms with a stable net profit want to pay dividends to signal that their profits are good. On the other hand, firms with a high variability of the net profit may prefer not to start paying dividends. Firms are afraid that a high variability of the net profit forces the firm to stop or decrease dividend payments.

The regression results for the Fama and French (2001) traditional firm characteristics confirm the hypothesis that firms that are more profitable, have less investment opportunities and are larger, are more likely to pay out dividends. The finding that more profitable firms are more likely to pay out dividends is in line with firms wanting to reduce agency costs (Easterbrook, 1984, Jensen, 1986) and the pecking order theory of financing. (Myers, 1984) The finding that larger firms are more likely to pay out dividends is in line with the life-cycle theory, because larger firms are older in general. (Grullon, Michaely and Swaminathan, 2002) According to the life-cycle theory, older firms have more internal funds and less investment opportunities. The finding that firms with less investment opportunities are more likely to pay out dividends is in line with the pecking order theory. However, the results show that investment opportunities are mostly insignificant for Dutch publicly listed firms. The results also show that the variability of the net profit has a significant negative influence on the number of firms that pay out dividends.

There remains a general lower propensity to pay out dividends over time when only

proxies for profitability, investment opportunities and size are taken into account. The

lower propensity to pay over time is 21.4% when both proxies for investment

opportunities, market to book ratio and growth rate of assets, are used and 13.6% when

only the growth rate of assets is used. However, when the variability of the net profit is

added to the traditional firm characteristics of Fama and French (2001), the lower propensity to pay out dividends over time disappears. With both proxies for investment opportunities there is a higher propensity to pay over time of 30.1%. When only the growth rate of assets is used, there is a higher propensity to pay over time of 28.0%.

Concluding, the variability of the net profit has a significant negative influence on the number of firms that pay out dividends in the Netherlands. Moreover, the variability of the net profit helps in explaining the remaining lower propensity to pay out dividends.

This result is in agreement with the findings of Van Ees, Von Eije and Hooghiemstra (2008a) and Haan (1995). This finding is also consistent with the signalling theory.

Firms are afraid to be punished by the market when they have to stop or decrease their dividend payment because of an increased variability of the net profit. (Black, 1976) The variability of the net profit has increased over time. Hence, fewer firms pay out dividends. Firms should try to stabilize the net profit as much as possible, because the results show that the variability of the net profit is an important variable in explaining the decrease in the number of firms paying dividends over time. When firms have a stable net profit they can avoid being forced to send a negative signal to the market by decreasing or cutting the dividend stream. Firms should keep this in mind when they decide on their dividend policy.

This study adds value to the existing literature for four reasons. First, the European, and especially the Dutch setting, has received very little attention. The existing literature mainly focuses on the U.S. and several larger countries in Europe. The research that has been done in the Netherlands did not use the Fama and French (2001) methodology to make a distinction between firm characteristics and a general lower propensity to pay over time. Second, the research period includes the years until 2010 and therefore it shows if the traditional variables are still important in recent years. Third, the variability of the net profit is added to the traditional firm characteristics. Fourth, the results do not only confirm the Fama and French (2001) results in a Dutch setting, but the results also add value because they show that the variability of the net profit can explain the remaining lower general propensity to pay out dividends over time.

Section I provides the theoretical background for the study. Section II explains the

data selection and the methodology. Section III presents the summary statistics and the

results. In section IV conclusions are drawn and the limitations of the study are

presented. Section V shows the appendix and section VI is the reference list.

I. Theoretical background

A. Dividend policy

Dividend policy is an important area of debate. Modigliani and Miller (1961) claim that dividend pay out is irrelevant under perfect market conditions. They claim that investors can create their own dividends by selling or borrowing from their own portfolio. When the perfect market conditions hold, there are no additional costs to one of those methods. However, in reality the three perfect market conditions, perfect capital markets, rational investor behaviour and perfect certainty, do not hold.

According to Haan (1995), the decision to pay or not to pay out dividends is influenced by all of the market imperfections. Therefore, it can be more profitable to pay out dividends. There are many theories that try to explain why firms decide to pay out dividends. DeAngelo and DeAngelo (2006) claim that, in an idealised world, stockholders profit the most when managers select maximum net present value projects and pay out the full net present value of the future cash flows of the project over the life of the firm.

B. Theories to explain dividend policy

B.1. Pecking order theory

The pecking order theory relies on the assumption that the capital structure of a firm is driven by preferences of a firm on how to finance projects. The pecking order theory predicts that firms will prefer to use internally generated funds before using external funds. (Myers, 1984) The pecking order theory can be explained from a transaction costs perspective. Fama and French (2001) and Van Ees, Von Eije and Hooghiemstra (2008b) start their papers with the remark that since the tax rate for dividends is higher than for capital gains, dividend gains should be presumed to be less valuable than capital gains.

However, paying dividends and repurchasing shares does decrease the firms’ internal

funds. In a perfect market this would not be a problem, but attracting external funds in an inefficient market results in paying transaction costs. (Van Ees, Von Eije and Hooghiemstra, 2008b, Haan, 1995)

B.2. Agency theory

Agency costs are the result of the separation of managers and shareholders. Agency costs are higher when managers have more opportunities to achieve their personal goals at the expense of the shareholders. When firms pay out dividends they have fewer opportunities to do this. (Jensen, 1986) Easterbrook (1984) provides two agency cost explanations of paying dividends. First, he claims that dividends may keep firms in the capital market if the monitoring of managers can be done at low cost. Second, dividends can be useful to adjust the risk that managers and investors take. In this way agency costs can be reduced. This reasoning helps in explaining the fact that firms pay out dividends and raise new funds in the capital market at the same time. La Porta et al.

(2000) find that the agency approach is indeed relevant in explaining dividend policy all over the world. They find support for an outcome model of the agency theory. This model predicts that minority shareholders pressure corporate insiders to pay out dividends to the shareholders. Therefore, firms in countries that have better minority shareholder protection tend to pay out more dividends. According to this theory, highly protected shareholders are expected to be more open to postponing dividend payment when there are investment opportunities. On the other hand, when shareholder protection is low, shareholders want to receive the dividends as fast as possible.

B.3. Life-cycle theory

The life-cycle view predicts that relatively new firms pay fewer dividends, because

their investment opportunities exceed their internally generated capital. In later years,

their internal funds exceed their investment opportunities and therefore firms optimally

pay out the excess funds to reduce the possibility that free cash flows are wasted. This

indicates that the life-cycle theory is based on a trade-off between the costs and

advantages of dividend payment. The costs are the agency costs of free cash flows and the advantages are the flotation costs savings, which are savings on costs paid by the firm when new securities are issued. This trade-off changes over time, because older firms tend to be more profitable and have less investment opportunities. Therefore, paying out dividends becomes more attractive over time. (DeAngelo, DeAngelo and Stultz, 2006) Grullon, Michaely and Swaminathan (2002) refer to this as the maturity hypothesis. A mature firm generates the most cash flows and therefore, in accordance with the agency theory (Jensen, 1986), it is more likely to pay out these free cash flows by paying out dividends or repurchasing stock.

B.4. Signalling theory

According to the signalling theory, the decisions of the management convey information to the outside of the firm, because managers have more information about the firm than outsiders. In this way shareholders can interpret the dividend policy decision as a signal about the future possibilities of the firm. Dividend changes, or the fact that the dividend does not change, signals information to the investors about what the managers really think. Therefore, if the forecasts are reliable, the dividend policy conveys information. (Black, 1976) According to Bhattacharya (1979), shareholders can be seen as relative outsiders and therefore they can obtain information about the growth opportunities of the firm by looking at the dividend payments of the firm.

However, Van Ees, Von Eije and Hooghiemstra (2008b) claim that such a signal should be interpreted with care, because a higher dividend payment could mean either that the firm expects lower growth opportunities in the future or that the firm expects a higher profit.

B.5. Catering theory

The catering theory hypothesizes that firms pay out dividends to meet investor

demand. This may indicate that the declining propensity to pay may be a result from a

change in investor demand away from dividends and towards capital gains. Catering

incentives vary over time and can be seen, to a certain extent, as separate from firm characteristics. (Baker and Wurgler, 2004a)

B.6. Flexibility argument

Chowdhry and Nanda (1994) provide another explanation why firms choose to pay out dividends. According to them, the optimal pay out policy is to pay out some cash in the form of dividends and carry the rest forwards to future periods. However, when the firm is undervalued, the firm is more likely to repurchase shares instead. Moreover, in periods with negative information, a firm is more likely to repurchase shares instead of paying dividends. This explains why firms repurchase shares infrequently. A multinomial analysis by Benltaifa (2011) shows that firms that use dividend payments and repurchases are larger, have less investment opportunities and less debt. The main motivation of firms to use share repurchases is to distribute cash in order to reduce agency costs, and stay flexible at the same time. Jagannathan, Stephens and Weisbach (2000) confirm that dividends are paid out of sustainable cash flows. On the other hand, share repurchases are paid when cash flows are much more volatile. They argue that repurchases do not replace dividends, but serve as a complementary mechanism to pay out short-term cash flows.

C. Influence of firm characteristics and general lower propensity to pay over time

Recent studies indicate a worldwide decline in the number of listed firms that pay out dividends. However, there are some contradicting results. Researchers claim that there might be several causes for the decline in the number of firms paying dividend.

The decline may be caused by changing firm characteristics over time. However, several studies find a general lower propensity to pay out dividends, even when controlling for the changing firm characteristics. (Fama and French, 2001, Fatemi and Bildik, 2012)

An interesting finding is that although the percent of paying firms has decreased

over time, the real amount of dividend paid increased during this time period. DeAngelo,

DeAngelo and Skinner (2004) point out that this is because the reduction in payers

mainly appears in firms that did not pay out large amounts of dividends in the first place. Moreover, increasing dividends amongst the large firms appears to be compensating easily for the loss of the dividend reduction of many small dividend paying firms. This will not be discussed in further detail, because the real amount of dividend paid is not important in determining the factors that explain why firms decide to pay out dividends.

One of the most important articles about dividend payment since Lintner’s (1956) behavioural model for dividend payment is the paper of Fama and French (2001). They conclude that both, the changing firm characteristics and a general lower propensity to pay out dividends over time, explain a part of the decline in the number of firms that pay out dividend in their U.S. study on the NYSE, AMEX and NASDAQ from 1978 to 1999. The proportion of firms that paid dividend declined from 66.5% in 1978 to 21.3% in 1998.

Fama and French (2001) find that three firm characteristics are significant in explaining dividend payment. Profitability has a positive influence, growth opportunities have a negative influence and size has a positive influence on the number of non-financial and non-utility firms paying dividend. This result is confirmed by Banerjee, Gatchev and Spindt (2002) in a similar U.S. study. Even when they controlled for these important firm characteristics, there still remained a lower propensity to pay over time. Ferris, Sen and Yu (2006) confirm these findings in the UK. However, their findings are different in two ways. First, the decrease in percentage of dividend payers is smaller in the UK. Second, the decline in propensity to pay out dividends is smaller and has a shorter duration.

Fama and French (2001) also find that share repurchases are not a major factor in explaining the lower propensity to pay out dividends. Amihud and Li (2006) try to explain the lower propensity to pay by claiming that the signalling value of a dividend announcement has decreased over time. They find that the reason for the decrease in the informational value is the increasing number of institutional investors. Institutional investors are better informed and more sophisticated. As a result, firms use fewer dividend payments, because it is a costly signal to the firm.

Reddy and Rath (2005) document characteristics of payers and non-payers of

dividend in the emerging market of India. However, they do not look at the propensity to

pay out dividends. They find that the percent of firms paying dividends decreases from

57% in 1991 to 30% in 2001. Their regression results are in agreement with the Fama

and French (2001) results. They confirm that dividend paying firms are more likely to be

profitable and large. However, Reddy and Rath (2005) find no significant relationship between the firm’s growth opportunities and dividend payments, which contradicts the Fama and French (2001) results.

Denis and Osobov (2008) are also in agreement with Fama and French (2001). They find that profitability, growth opportunities and size are significant firm characteristics with the same signs as predicted by Fama and French. In addition, they find that the earned to contributed equity mix significantly impacts the dividend payout. They extend the literature by examining cross-sectional and time-series evidence on the propensity to pay out dividends in several developed financial markets in the period from 1989 to 2002. The propensity declines in the Denis and Osobov (2008) sample are smaller than the ones reported by Fama and French (2001). Moreover, the evidence is not as powerful in all countries. Especially the UK findings show very little evidence of a propensity decline when controlling for firm characteristics. This is in line with the findings of Ferris, Sen and Yu (2006). They attribute the declines in the propensity to pay out dividends to newly listed firms that fail to initiate dividend payments when they are expected to. Therefore, they are not able to reject the possibility of no significant change in corporate dividend policies. Denis and Osobov (2008) find no significant relationship between the lower propensity to pay out dividends and the repurchasing of shares, which is in agreement with the results of Fama and French (2001). Denis and Osobov (2008) argue that their contribution to the existing literature is that they narrow down the scope of the disappearing dividends to the dividend decisions of newly listed firms. However, they point out that the small amount of newly listed firms that pay out dividends in their sample of developed countries allows for the possibility of an overestimation of the expected number of dividend paying firms. They argue that this could be the reason for the lower propensity to pay out dividends instead of a systematic change in dividend policies. The benchmark model may not incorporate all characteristics of newly listed firms that influence them to make them less likely to pay out dividends.

Fatemi and Bildik (2012) conduct their research in an international context of 33

countries. They find a substantial variation in the propensity to pay out dividends at a

global level. Overall, they identify a common trend of a declining tendency to pay out

dividends. They find that firms with a higher profitability, firms with less growth

opportunities and larger firms, are more likely to pay out dividends, which is in line with

the Fama and French (2001) results. They find that publicly traded firms have become less profitable, have more investment opportunities and have become smaller.

Therefore, these firm characteristics have contributed to the decline in the propensity to pay out dividends over time. However, even when controlling for these firm characteristics, there still remains an unexplained decline in dividend pay out.

Therefore, they agree with Fama and French (2001) that dividends are disappearing and that controlling for changing firm characteristics fails in explaining the entire lower propensity to pay out dividends over time.

The conclusions of the aforementioned papers are summarized by Vieira and Raposo (2007) as papers that find that the traditional firm characteristics, profitability, growth opportunities and size, are significant. The authors conclude that firms that have a higher profitability, less growth opportunities and are larger, are more likely to pay out dividends. More profitable firms are more likely to pay out dividends, which is consistent with the agency theory (Easterbrook, 1984, Jensen, 1986) and the pecking order theory of financing (Myers, 1984). It is likely that they pay out the excess profits, to reduce agency costs and retain enough internal funds to finance investment projects.

The finding that firms with more growth opportunities tend to pay out fewer dividends is in line with the pecking order theory, because they prefer to invest with internal funds instead of external funds. The finding that larger firms are more likely to pay out dividends is consistent with the life-cycle theory, because larger firms have, on average, more internal funds than growth opportunities. (Grullon, Michaely and Swaminathan, 2002) Larger firms are older in general. Besides those firm characteristics, the authors find that there still remains a lower propensity to pay out dividends over time. Vieira and Raposo (2007) suggest that this can be explained by a declining perceived benefit of dividends through time, because of the fiscal disadvantage in relation to capital gains.

This is in line with the transaction costs theory.

One of the studies that is closely related to the Fama and French (2001) study is the study of DeAngelo, DeAngelo and Stultz (2006). Their sample also consists of non- financial and non-utility firms listed on the NYSE, AMEX and NASDAQ. Their time span lasts from the mid-1970s to 2002. Their main contribution is that they added an additional firm characteristic, the ratio of earned to contributed capital. They find that this ratio has more explanatory power than profitability and investment opportunities.

When controlling for the ratio of earned to contributed capital, they find that firms with

negative retained earnings show virtually no change in their dividend payments.

Dividend payers tend to have a high earned to contributed capital ratio and non-payers tend to have a low earned to contributed capital ratio, which is in line with the life-cycle theory of dividends.

Baker and Wurgler (2004b) examine if the catering theory helps to explain the lower propensity to pay out dividends. They create proxies for time-varying dividend premium, which is the catering incentive, and find that it helps to explain the rate of dividend initiation and omission. According to Baker and Wurgler (2004a) managers could try to cater to investor demand by paying dividends when a positive premium is put on firms that pay out dividends and by not paying dividends when the dividend premium is negative. Baker and Wurgler (2004b) show that catering incentives are able to explain, in an out of sample test, the actual disappearance of the post 1977 dividend of the Fama and French (2001) sample. According to them, it can be attributed to firms catering to sentiment-driven demand.

Hoberg and Prabhala (2009) examine the disappearing dividends with risk as an explanatory variable. They claim that risk can explain between one-third and one-half of the Fama and French (2001) disappearing dividends. They find little evidence of the catering effect. Hoberg and Prabhala (2009) claim that the dividend premium, the most important proxy of the catering effect, is only significant when risk is no control variable.

D. Differences between the U.S. and Europe

Bancel, Bhattacharyya and Mittoo (2005) investigate the differences in dividend payment in the U.S. and Europe. Their survey findings show small differences. Overall, the European managers value similar factors as important for pay out policies as their U.S. peers. Both European and U.S. managers do agree with Lintner (1956) that dividends are difficult to cut and are likely to be smoothed out. This confirms the findings of an earlier survey of Baker, Farrely and Edelman (1985). Bancel, Bhattacharyya and Mittoo (2005) find that the stability of earnings is considered to be very important. They find little support for the signalling theory and the agency theory.

There are differences between the U.S. and the European managers as well. European

managers do consider dividends to be equally important for the valuation today as it

was in the past. Moreover, in contrast to their U.S. peers, they do not agree with the statement that dividend changes lag behind changes in earnings.

Von Eije and Megginson (2008) perform their research in Europe. They find that dividend pay out and share repurchases in Europe are very similar to the U.S. situation.

They find that the fraction of European firms that pay out dividends has sharply declined in recent years. On the other hand, they find that the total amount of dividends paid has increased from 1989 to 2005. According to them, the fraction of retained earnings in a European firm’s total equity is not significantly correlated with the likelihood to pay out cash dividends. However, they find that the age of the firm is significant, which is consistent with the life-cycle view. Those results are contradicting to the international results of Denis and Osobov (2008) and the U.S. results of DeAngelo, DeAngelo and Stulz (2006). Von Eije and Megginson (2008) also investigate the catering effect. The results show that the catering effect is not relevant for European dividend policy. This is in agreement with the findings of Hoberg and Prabhala (2009).

Kirkulak and Kurt (2010) performed their research on a national level in Turkey. In

their research amongst firms on the Istanbul Stock Exchange from 1991 to 2006, they

conclude that firms prefer dividend omissions above dividend reductions. Moreover,

when firms start paying dividend, they put much effort in increasing the dividend. Firms

are reluctant to reduce dividend because they are afraid that it might signal to the

market that the prospects of the firm are bad. These conclusions are in line with the

signalling theory. (Black, 1976) Moreover, the results are in agreement with a survey of

Brav et al. (2005). They find that dividend policy is very conservative. The dividend

conservatism results in payers being afraid to cut dividends and makes non-paying firms

reluctant to start paying dividends. Moreover, paying firms wish that they did not pay

out dividends. According to Brav et al. (2005), the only firms that increase dividends

have stable and sustainable increases in earnings. However, even these firms would

prefer to pay out in the form of repurchases. They claim that the most important reason

for repurchases to be important is that they are considered more flexible than dividends,

because cuts are not always identified as a bad sign. They find that most managers

regret the current dividend level of the firm, because managers think that it prevents the

firm from having the desired level of flexibility in cash pay outs. These findings are in

line with the flexibility argument. According to Kirkulak and Kurt (2010) dividend

payments are also influenced by the current earnings of the firm and by a financial crisis.

E. The Dutch situation

The Dutch setting has received little attention. (De Jong and van Beusichem, 2008) Cools (1993) surveyed 50 chief executive officers of Dutch listed non-financial firms. His findings in the Netherlands do agree with the findings of Lintner (1956). Dutch firms aim at paying a target ratio and they try to stabilize the dividend payments. De Jong and Van Beusichem (2008) find that dividend payment requires additional financing when there are much investment opportunities. Moreover, they confirm that dividends can be perceived by shareholders as a signal of confidence in the future. De Jong and Van Beusichem (2008) mention several changes in the Netherlands during the time of the research. Since the dividend stop in 1974, there has been a continuing improvement of the regulations. Two codes of conduct were developed. The first code, by Peter’s Committee, was incorporated in 1997 and the second, by Tabaksblat’s Committee, was incorporated in 2003. As a consequence, the shareholders became much more important and involved in the firm by influencing the corporate policies. In the period following the codes, equity financing became more important instead of the debt financing. At the same time corporate governance becomes increasingly important in the Netherlands.

Van Ees, Von Eije and Hooghiemstra (2008b) find that in the Netherlands in 1998 approximately 17% of the publicly listed firms did not pay out dividends and in 2006 the percent has increased to approximately 28%. In another paper by Van Ees, Von Eije and Hooghiemstra (2008a), which is based on Dutch firms listed to the AEX, AMX or ASCX in 2006, they find a significant positive influence of the ratio of retained earnings to net assets and profitability. Moreover, they find a significant negative influence of the variability of the net profit and the market to book ratio on the propensity to pay out dividends. These results, except for the retained earnings to net assets variable, are consistent with the Von Eije and Megginson (2008) study. Moreover, the finding that there is a negative relation between the variability of the net profit and dividend payment is consistent with the findings of Jagannathan, Stephens and Weisback (2000).

They find that when the variability of the net profit increases firms are more likely to

prefer a share repurchase. Further, Van Ees, Von Eije and Hooghiemstra (2008a) find

that age, solvability and liquid resources have an insignificant influence. Moreover, they

find that the repurchasing of shares is positively related to dividend payment. Therefore,

they agree with the conclusions of Jagannathan, Stephens and Weisbach (2000) that share repurchases and dividend payment are complementary. However, the coefficients are not found to be significant and therefore the only conclusion that can be drawn is that there are no substitution effects between dividends and share repurchases. (Van Ees, Von Eije and Hooghiemstra, 2008a)

The research of Haan (1995) agrees with Van Ees, Von Eije and Hooghiemstra (2008a) that there is a negative influence of the variability of the net profit on the pay out of dividends of Dutch listed firms. According to Haan (1995) this is consistent with the signalling theory, because firms with a stable profit want to pay out higher dividends in order to signal to the investors that their profits are good. Haan also finds that Dutch firms with a higher Tobins Q measure, which is a measure for investment opportunities, pay out fewer dividends, because they need the internal funds to finance those investments. This is in line with the pecking order theory of financing. He also finds that there is a strong reliance of Dutch firms on internal financing. Therefore, investment opportunities are negatively related to the pay out of dividends. An interesting finding is that Dutch listed firms decrease their dividend payments faster than they do increase them. In this way they try to retain enough money for investment projects.

On the other hand, Renneboog and Szilagyi (2006) find that dividends pay out is low and moderately smoothed in general for Dutch firms. They find that the decision to pay out dividends is unresponsive to changes in earnings and they find little relationship with size, leverage and investment opportunities. This is contradicting to the findings of Haan (1995) and Van Ees, Von Eije and Hooghiemstra (2008a). The findings of Renneboog and Szilagyi (2006) show that dividends are not used to minify concerns about free cash flows, which is contradicting to the agency theory. They argue that this is because Dutch shareholders are too week to enforce optimal pay out policies.

F. Discussion and criticism on the Fama and French (2001) study

Julio and Ikenberg (2004) find that dividends are reappearing instead of

disappearing. They find that the decline in the number of dividend paying firms found by

Fama and French (2001) is sharply reversed in the new millennium. They find that the

Bush tax cut in the U.S. helps in explaining the reappearing dividend. However, this is

only part of the story. Julio and Ikenberg (2004) think that as long as investors respond positively to the announcement of a dividend increase there will be no disappearing dividends. They argue that the maturation of new firms will keep renewing the stream of dividend paying firms, which is consistent with the life-cycle theory. Besides those two reasons for reappearing dividends, they find that positive changes in the underlying economy, which increased the firm profitability, have contributed to reappearing dividends as well. This is in line with the agency theory, signalling theory and the pecking order theory. If this is a major reason for reappearing dividends it is very likely that dividends have disappeared again since the start of the financial crisis. Grullon et al.

(2011) also find contradicting findings to the Fama and French (2001) study. They find that the propensity to pay out dividends in the U.S. has remained constant over the last 30 years. According to them, firms are as likely to return cash in 2003 as in 1978.

Moreover, they find that the net cash payments of firms with low retained earnings have increased over time. Therefore, they find that the Fama and French (2001) results do not hold when considering the total firm’s net cash pay outs. Grullon et al. (2011) suggest that the reason might be the loosening of the legislation regarding repurchases for less mature firms in the United States. They conclude that the shift from cash dividend distributions to share repurchases shows that firms are moving towards an optimal dividend policy to minimize tax losses. This is consistent with the transaction costs theory and with firms wanting to be more flexible.

Chahyadi and Salas (2012) do criticize Fama and French (2001) and other studies,

because they use a method that looks at the difference between the actual percent and

the expected percent of dividend players. This method does not discriminate between

changes in firm characteristics and changes in the attitude of managers towards paying

dividends. They find, using a modified Oaxaca decomposition methodology for a binary

dependent variable (Chahyadi and Salas, 2012), that 76% of the decline of dividend

payers in the Fama and French (2001) sample is caused by changes in firm

characteristics. Chahyadi and Salas (2012) add several other firm characteristics to the

traditional Fama and French (2001) characteristics. They find that the most important

variables are the earned to contributed capital mix, profitability, investment

opportunities and liquidity. When they also incorporate the firms’ share repurchases,

they find a slightly higher propensity to pay out dividends over time. Therefore, they

find that share repurchases and firm characteristics explain 100% of the decline in the

number of dividend paying firms. Chahyadi and Salas (2012) conclude that the repurchasing of shares substitutes for dividend payment. This finding is contradicting to the findings of Jagannathan, Stephens and Weisbach (2000) who find that share repurchases and dividend payments are complementary rather than substitutes.

This study uses the method of Fama and French (2001) in spite of the criticism by Chahyadi and Salas (2012). Most studies confirm the Fama and French (2001) findings and use their methodology without encountering any problems. However, the criticism will be incorporated as a weakness of the research and future research could try to improve the method. In addition to the traditional Fama and French firm characteristics, the variability of the net profit is used as an additional firm characteristic. The next section explains the methodology in more detail.

II. Research method

A. Data selection and data description

The data is collected for Dutch firms that are publicly listed on the AEX, AMX, ASCX

and the local Dutch market. Similar to the Fama and French (2001) study, all financial

and utility firms are excluded, in order to avoid criticism about regulations that

influence the dividend payment of these firms. A newly listed firm is added to the

sample at the beginning of year T if it was added to the DataStream database between

January and December of year T-1. A firm that was delisted from the DataStream

database in year T is removed from the dataset in year T. Firms that became listed

before 1987 are all included in the sample at the start of 1987. The information on the

variables used, is collected from Thomson Reuters DataStream. When there is only few

data missing for a specific firm, the data is collected from Bureau van Dijk’s Orbis. The

data that is collected for each firm are the dividend yield (DY), earnings before interest

and taxes (WC18191), income taxes (WC01451), preferred stock (WC03451), total

stockholder’s equity (WC03995), market capitalization (WC08001), total assets

(WC02999) and the net income before preferred dividend (WC01651).

The research period lasts from 1987 to 2010. Unfortunately, not all variables were available for all firms. The firms, on which not all the data is available for every year that the firm was publicly listed, are not included in any year. As a result, the number of firms in the sample is the smallest in 1987, with 26 firms and slowly increases to 75 firms in 2001. In 2010 the number of firms has decreased again to 53 firms. The reason for the changing sample size, are listings and delistings of firms during the research period. The small number of firms at the start of the sample period can be explained by the data unavailability for many firms before 1990. Therefore, a lot of firms are excluded in the whole sample period, but especially in the early years. The decreasing sample size after 2001 can be explained by the data unavailability for new lists in more recent years.

Moreover, there were delistings during these years which also contributed to the decreasing sample size. Therefore, a percent of dividend paying firms is calculated in addition to the total number of firms that pay out dividends. As long as the regression results are reliable, the fluctuation in the sample size across years is not a problem.

Fama and French (2001) and Fatemi and Bildik (2012) have a fluctuating number of firms over time as well. However, their sample size is much larger.

B. Methodology

The aforementioned variables are needed to calculate the proxies for profitability, investment opportunities, size and the variability of the net profit. The proxy for profitability is the firm’s earnings before interest divided by the total assets. The proxies for investment opportunities are the market to book ratio, which is calculated by dividing the market value of the firm in year T by the total assets of the firm in year T and the firm’s growth rate of assets. The firm’s growth rate of assets is calculated by dividing the total assets in year T minus the total assets in year T-1 by the total assets in year T-1. The proxy for size is calculated by calculating the percent of firms that have the same or smaller market capitalization in year T. Market capitalization is defined as the market price at the end of the year times the number of common shares outstanding.

The variability of the net profit in year T will be calculated by calculating the standard

deviation of the values for net profit in year T and the two years before year T divided by

the firm’s total assets in year T.

The methodology used is similar to the Fama and French (2001) methodology.

Summary statistics are provided on the variables constructed. The summary statistics are useful to identify trends and differences between dividend payers and non-dividend payers. According to the literature, the number of firms that pay out dividends has declined over time and it is expected that the selected firm characteristics have a significant influence on this lower propensity to pay out dividends. However, it is hypothesized that the firm characteristics fail in explaining the entire lower propensity to pay out dividends over time.

The first step is to test if the independent variables, which are the firm characteristics, do have a statistically significant influence on the dependent variable. In this approach it is assumed that the proxies for profitability, investment opportunities, size and the variability of the net profit have a constant meaning over time. The dividend yield is the dependent variable in the logit regressions. The dividend yield is defined as the annual dividend per share divided by the price per share. The dividend yield is a dummy variable, because the actual value of the dividend yield is not important for the regression analysis. This is true, because the goal of the study is to identify explanatory factors that explain the trend of a declining number of firms that pay out dividends. If in year T dividend is paid out, the variable is assigned a value of 1.0 and 0.0 otherwise.

Cross-sectional regressions 1 and 2 are tested for each year separately. The

advantage of running cross-sectional regressions for each year is that it allows us to see

if firms change the way in which they determine dividend policy over time. (Chahyadi

and Salas, 2012) In addition to the traditional firm characteristics from the study of

Fama and French (2001), the variability of the net profit from the Van Ees, Von Eije and

Hooghiemstra (2008a) study is added to the second regression. The first regression

allows testing if the traditional Fama and French (2001) firm characteristics can explain

the entire decline in propensity to pay out dividends or if there still remains a general

lower propensity to pay out dividends over time. Moreover, when we add the variability

of the net profit in the second regression, it is possible to determine the difference

between the two regressions and to see if this additional variable is able to explain the

expected general lower propensity to pay out dividends over time.

The dependent variable in regression 1 and 2, , is the dividend yield at time T.

This variable is a dummy variable and will be assigned a value of 0.0 if there is no dividend yield or 1.0 if there is a dividend yield. is the constant and the other betas are the betas for the independent variables. is the profitability at time T, measured by the earnings before interest divided by total assets. are the investment opportunities at time T, measured by the market to book ratio and the growth rate of assets. is the size of the firm at time T, measured by the percent of firms that have a smaller or equal market capitalization. is the variability of the net profit in year T. This firm characteristic is only included in the second regression. is the error term at time T.

Every independent variable in regression 1 and 2 is winsorized at the 0.05 and the 0.95 level to correct for outliers that make the regression results less reliable.

Winsorization at the 0.01 and 0.99 level was not successful in eliminating all far outliers.

It is not necessary to winsorize the dependent variable, because it is a dummy variable.

The regressions are run with White heteroskedasticity-consistent standard errors and covariance, because the White test shows that there is heteroskedasticity in the regressions.

The tables presented in section III show the means across years of the regression intercepts, slopes and the t-statistics for the mean. Therefore, the results show if the firm characteristics have a negative or positive influence on the number of firms that pay out dividends and if this influence is found to be significant. The literature suggests that the changes of the selected firm characteristics over time explain at least a part of the decline in the propensity to pay out dividends over time.

The next and most important step is to check if there still remains a general change

in the propensity to pay. Controlling for changing firm characteristics over time is

needed to make a distinction between the influence of changing firm characteristics and

the influence of a general changing propensity to pay out dividends. The results of the

regression analysis are used to predict the expected percent of firms that pay out

dividends. Fama and French (2001) observed, in the U.S., a peak of paying firms in 1977.

Therefore, they selected the 1963 to 1977 years as a base period to predict the expected number of firms paying dividends in the years after 1978. In this paper the base period is the 1987 to 1997 period, because after 1997 there is a sudden decline of 10% in the number of firms that pay out dividends. The averages of the regression results for the base period, from 1987 to 1997, can be used to calculate the expected number of firms that pay out dividends in the years after 1997. The percent of expected payers is calculated by applying the average logit regression coefficients for the base period to the actual values of the firm characteristics for each firm in year T. These results will be summed over firms and divided by the number of firms and multiplied by 100. The actual percent of payers is calculated by multiplying the ratio of payers to the total number of firms by 100. The change in the expected percent of firms that pay out dividends measures the effect of changing firm characteristics over time. The expected percent of firms paying minus the actual percent of firms paying measures the effect of a change in the general propensity to pay. When another base period is used, the results should be approximately the same. This is the same method that Fama and French (2001) used to check for biases.

C. Hypotheses

Considering the literature, a decrease in the number of firms paying dividends is expected. This trend has been identified by a lot of researchers in the U.S. (Fama and French, 2001), worldwide (Fatemi and Bildik, 2012), in Europe (Von Eije and Megginson, 2008) and in the Netherlands (Van Ees, Von Eije and Hooghiemstra, 2008a).

These studies identify firm characteristics that could have a significant negative or

positive influence on the number of firms that pay out dividends. Most studies find that

when firm profitability increases, it is more likely to pay out dividends. Investment

opportunities provide a firm with another possibility to invest their money. Therefore,

firms with more investment opportunities are less likely to pay out dividends. Firm size

is expected to have a positive influence on the likelihood that a firm pays dividend,

because larger firms usually have a more constant cash flow. Therefore, they are

expected to be able to continue to pay dividends more easily. (Chahyadi and Salas, 2012,

Fama and French, 2001). These traditional firm characteristics results are tested and

confirmed in numerous studies. Therefore, the following hypotheses are developed for the Dutch setting.

1a) More profitable firms are more likely to pay out dividends.

1b) Firms with more investment opportunities are less likely to pay out dividends.

1c) Larger firms are more likely to pay out dividends.

Fama and French (2001) and numerous other studies find that the traditional firm characteristics; profitability, investment opportunities and size, fail to explain the entire general lower propensity to pay. This result is confirmed in different settings. This results in the second hypothesis.

2) The traditional firm characteristics fail to explain the entire general lower propensity to pay out dividends over time.

When an additional variable is added, which is the net variability of the net profit, it

is expected that a larger percent of the decline in the number of firms that pay out

dividends can be explained by firm characteristics. A lower variability of the net profit

means that it is easier for the firm to keep paying dividends. Van Ees, Von Eije and

Hooghiemstra (2008a) and Haan (1995) find that this variable is very significant in the

Netherlands. The theoretical background shows that it is very important in the decision

to start paying dividends or to raise the amount of the dividend payment. (Bancel,

Bhattacharyya and Mittoo, 2005, Lintner, 1956) Therefore, when the variability of the

net profit is high, the firm will not start paying dividends, because it fears the

punishment of the market when it decreases or stops paying dividends. The signalling

theory provides the best reasoning for these findings. (Black, 1976) It is expected that,

when the variability of the net profit is included, the explanatory power of the firm

characteristics increases and the difference between the expected and the actual percent

of payers will become smaller. Hence, the general lower propensity to pay over time is

expected to be smaller. These theoretical findings result in the third hypothesis.

3) Including the variability of the net profit increases the explanatory power of the firm characteristics and reduces the remaining general lower propensity to pay out dividends over time

III. Empirical results

A. Summary statistics

In this subsection summary statistics are provided. It is important to know if there is a decline in the number of firms that pay out dividends over time in the Netherlands.

Without such a trend this study is not useful. Table I shows the number and percent of firms in the sample that pay out dividends and the number of firms that do not pay out dividends for each year in the 1987 to 2010 period. Table I shows that the number of firms in the sample is different in each year and therefore it is best to look at the percent of payers. The most important trend in table I is the decline in the percent of paying firms after 1997. Between 1987 and 1997 the percent of paying firms is approximately 90% and after 1997 it steadily decreases from 90% to 62% in 2010. However, in 2008 and 2009 the percent of paying firms increased again to approximately 70%. Figure 1 is a graphical representation of table I that makes the trend more visible.

Table IIa, IIb and IIc show summary statistics on all the firm characteristics for the

whole research period from 1987 to 2010. There are 1247 firm observations in the

sample. Table IIa shows the mean, 25

percentile, median, 75

percentile and the

standard deviation on all the firm characteristics for all firms. There are 961 dividend

paying firm observations in the sample. Table IIb shows the mean, 25

percentile,

median, 75

percentile and the standard deviation on all the firm characteristics of

dividend paying firms. There are 286 non-dividend paying firm observations in the

sample. Table IIc shows the mean, 25

percentile, median, 75

percentile and the

standard deviation on all firm characteristics of non-dividend paying firms.

Table I

Percent and number of firms, payers and non-payers

The table shows the number of firms that pay out dividends and the number of firms that do not pay out dividends and the percent of payers for each year from 1987 to 2010.

Payers (%) Firms Payers Non-Payers

1987 88% 26 23 3

1988 89% 28 25 3

1989 87% 30 26 4

1990 91% 34 31 3

1991 84% 37 31 6

1992 87% 38 33 5

1993 90% 39 35 4

1994 90% 40 36 4

1995 91% 43 39 4

1996 94% 48 45 3

1997 90% 52 47 5

1998 81% 59 48 11

1999 81% 67 54 13

2000 73% 75 55 20

2001 77% 75 58 17

2002 75% 73 55 18

2003 70% 70 49 21

2004 66% 64 42 22

2005 65% 63 41 22

2006 62% 63 39 24

2007 62% 61 38 23

2008 70% 56 39 17

2009 72% 54 39 15

2010 62% 53 33 20

Figure 1

The percent of firms that pay out dividends from 1987 to 2010

The figure is a graphical representation of table I. It shows the percent of firms that pay out dividends and

the percent of firms that do not pay out dividends from 1987 to 2010.

Table IIa

Summary statistics for all firms in the 1987-2010 period

Summary statistics for all firms, which includes dividend payers and non-payers, on the mean, 25

percentile, median, 75

percentile and the standard deviation of the earnings before interest divided by total assets (EBIt/At), market to book ratio (Vt/At), growth rate of assets (dAt/At), market capitalization (MDPt) and the variability of the net profit (SDPt/At) in the 1987-2010 period.

Observations Mean 25

percentile Median 75

percentile Standard Deviation

EBIt/At 1247 0.1018 0.0581 0.1136 0.1692 0.1257

Vt/At 1247 1.7242 1.0964 1.4137 1.9240 1.0216

dAt/At 1247 0.1048 -0.0240 0.0663 0.1788 0.2404

MDPt 1247 0.5000 0.2500 0.5000 0.7500 0.2895

SDPt/At 1247 0.0531 0.0076 0.0177 0.5573 0.1094

Table IIb

Summary statistics for dividend paying firms in the 1987-2010 period

Summary statistics for dividend paying firms on the mean, 25

percentile, median, 75

percentile and the standard deviation of the earnings before interest divided by total assets (EBIt/At), market to book ratio (Vt/At), growth rate of assets (dAt/At), market capitalization (MDPt) and the variability of the net profit (SDPt/At) in the 1987-2010 period.

Observations Mean 25th

percentile Median 75th

percentile Standard Deviation

EBIt/At 961 0.1251 0.0785 0.1222 0.1740 0.0914

Vt/At 961 1.6312 1.0916 1.3881 1.8235 0.8650

dAt/At 961 0.1031 -0.0054 0.0715 0.1707 0.2033

MDPt 961 0.5498 0.3214 0.5593 0.7895 0.2802

SDPt/At 961 0.0283 0.0066 0.0135 0.0325 0.0494

Table IIc

Summary statistics for the non-dividend paying firms in the 1987-2010 period

Summary statistics for firms that do not pay dividends on the mean, 25

percentile, median, 75

percentile and the standard deviation of the earnings before interest divided by total assets (EBIt/At), market to book ratio (Vt/At), growth rate of assets (dAt/At), market capitalization (MDPt) and the variability of the net profit (SDPt/At) in the 1987-2010 period.

Observations Mean 25th

percentile Median 75th

percentile Standard Deviation

EBIt/At 286 0.0234 -0.0636 0.0459 0.1276 0.1816

Vt/At 286 2.0366 1.1126 1.4839 2.3026 1.3840

dAt/At 286 0.1106 -0.1081 0.0444 0.2635 0.3366

MDPt 286 0.3722 0.1314 0.3017 0.5862 0.2785

SDPt/At 286 0.1366 0.0252 0.0611 0.1873 0.1871

Table III

Average values for different periods for all independent variables

The average values for profitability, investment opportunities, size and the variability of the net profit divided by total assets (SDPt/At). Profitability is measured by the earnings before interest divided by total assets (EBIt/At). Investment opportunities are measured by the market to book ratio (Vt/At) and the growth rate of assets (dAt/At). Size is measured by the market capitalization (MDPt). The average values are divided in different time periods and payers are separated from non-payers.

1987-2010 1987-1991 1992-1996 1997-2001 2002-2006 2007-2010 EBIt/At

All firms 0.1018 0.1375 0.1370 0.1370 0.0712 0.0546

Payers 0.1251 0.1418 0.1425 0.1410 0.1096 0.1031

Non-Payers 0.0234 0.1063 0.0852 0.1210 -0.0008 -0.0418

Vt/At

All firms 1.7242 1.3122 1.5860 2.1172 1.8848 1.5554

Payers 1.6312 1.3223 1.6161 1.9245 1.7146 1.5041

Non-Payers 2.0366 1.2394 1.3032 2.8823 2.2039 1.6572

dAt/At

All firms 0.1048 0.0840 0.1045 0.1939 0.1142 0.0667

Payers 0.1031 0.0865 0.1072 0.1821 0.0577 0.0925

Non-Payers 0.1106 0.0665 0.0793 0.2408 0.2201 0.0155

MDPt

All firms 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000

Payers 0.5498 0.5463 0.5196 0.5420 0.5650 0.5952

Non-Payers 0.3722 0.3001 0.4403 0.3673 0.3997 0.3337

SDPt/At

All firms 0.0531 0.0135 0.0190 0.0321 0.0978 0.0765

Payers 0.0283 0.0115 0.0165 0.0266 0.0378 0.0440

Non-Payers 0.1366 0.0282 0.0421 0.0542 0.2103 0.1412

Table III shows the average values for profitability, investment opportunities, size and the variability of the net profit (SDPt). The profitability is measured by the earnings before interest divided by total assets (EBIt/At). Investment opportunities are measured by the market to book ratio (Vt/At) and the growth rate of assets (dAt/At). Size is measured by the market capitalization (MDPt). The means shown are split up in different periods and payers are separated from non-payers. Profitability is higher for payers. Payers have a profitability ratio of 0.1251 and firms that do not pay have a profitability ratio of 0.0234. This is also true for all sub-groups and is in line with hypothesis 1a which claims that more profitable firms are more likely to pay out dividends. The mean of the market to book ratio is higher for non-paying firms (2.0366) than for paying firms (1.6312). This is in line with hypothesis 1b, which states that firms with more growth opportunities are less likely to pay out dividends. However, in the periods from 1987 to 1991 and from 1992 to 1996, the market to book ratio of paying firms is a bit larger. The growth rate of assets is a bit larger for non-payers (0.1106) than for dividend payers (0.1031). However, the sub-group results are mixed. In the period from 1997 to 2001 and from 2002 to 2006 the non-payers do have a higher growth rate of assets, but in the other three periods payers do have a higher growth rate of assets.

The hypothesis that firms that do pay out dividends have a lower growth rate of assets does hold for the whole sample but does not hold for several sub-groups. The average market capitalization is higher for firms that do pay out dividends (0.5498) than for firms that do not pay out dividends (0.3722). This finding holds for all sub-groups. This is consistent with the theory and hypothesis 1c, that larger firms are more likely to pay out dividends. The variability of the net profit is a lot higher for firms that do not pay dividends (0.1366) than for firms that do pay out dividends (0.0283). This holds for all sub-groups. The difference is especially large in the period from 2002 to 2006 with 0.2103 for non-payers against 0.0378 for payers. Those findings are in line with the hypothesis that the variability of the net profit has a negative influence on the decision to pay out dividends. Overall, the summary statistics are in line with the hypotheses.

However, the findings for the growth rate of assets proxy of investment opportunities are mixed.

Table III also shows that the average firms’ profitability has decreased over time.

This is true for payers and non-payer. Both investment opportunities, the market to

book ratio and the growth rate of assets, have increased until the period from 1997 to

2001 and afterwards they decreased again. Market capitalization is calculated against the other firms in the sample and therefore the average for all firms stays the same over time. However, payers have become larger and non-payers have become smaller. The variability of the net profit has increased over time, which is in line with the hypothesis that it will help in explaining a larger part of the decline in the number of firms that pay out dividends. These summary statistics show no proof of the significance of the trends over time. The next subsection provides the results of the significance tests.

B. Results

Tables IV and V, which are presented in this subsection, show the results of the cross-sectional regressions on the influence of the selected firm characteristics on the pay out of dividends over time. Tables VI and VII show the percentages of the influence of changing firm characteristics and a general lower or higher propensity to pay over time on the decision to pay out dividends.

There are two sets of results. One uses both measures for investment opportunities,

which are market to book ratio and the growth rate of assets, and the other one uses

only the growth rate of assets. There are two reasons to do this. First, it is consistent

with the Fama and French (2001) method. They find that the market to book ratio drifts

up over time. The second and more important reason is that in several years a high

correlation between the proxy for profitability and the market to book ratio proxy for

investment opportunities exists. This problem could not be solved when other proxies

or ratios are used instead. Therefore, it is useful to run the regressions without the

market to book variable in order to see if the results are subject to a multicollinearity

bias.

Table IV

Logit regression results on which firms do pay out dividends

The logit regression results are estimated for all DataStream firms with the required data items available for each year in the 1987-2010 period. The dependent variable is a dummy variable which is set to 1.0 in year t if a firm does pay out dividends and it is set to 0.0 if a firm does not pay out dividends in year t. The explanatory variables are profitability (EBIt/At), market to book ratio (Vt/At), growth rate of assets (dAt/At) and the percent of firms in the sample that have the same or lower market capitalization (MDPt).

Means (across years) are shown for the regression intercept (Int) and the slopes of the variables. T- statistics for the mean are shown, which are defined as the mean divided by its standard error.

Average coefficients Average T-statistic

Int. EBIt/At Vt/At dAt/At MDPt Int. EBIt/At Vt/At dAt/At MDPt 1987-2010 0.72 1.42 -0.15 0.01 0.30 4.56 2.36 -1.73 -0.08 1.72

1987-1997 0.75 1.38 -0.12 0.15 0.21 4.50 1.11 -0.72 0.44 1.06

1998-2010 0.70 1.46 -0.17 -0.11 0.38 4.62 3.41 -2.58 -0.52 2.29

1987-1991 0.74 1.39 -0.18 0.16 0.32 3.58 1.14 -0.87 0.42 1.50

1992-1996 0.74 1.64 -0.07 0.14 0.06 5.02 1.30 -0.49 0.30 0.36

1997-2001 0.75 1.01 -0.15 -0.12 0.48 5.89 2.44 -2.96 -0.58 3.27 2002-2006 0.75 1.45 -0.17 -0.21 0.25 4.75 3.05 -2.57 -0.89 1.38

2007-2010 0.61 1.67 -0.19 0.10 0.42 3.32 4.25 -1.76 0.44 2.22

1987-2010 0.60 0.91 -0.05 0.25 4.28 1.94 -0.28 1.34

1987-1997 0.71 0.70 0.08 0.17 4.23 0.86 0.27 0.78

1998-2010 0.51 1.09 -0.16 0.31 4.32 2.86 -0.74 1.81

1987-1991 0.65 0.56 0.06 0.30 3.26 0.78 0.23 1.10

1992-1996 0.73 1.20 0.09 0.02 4.67 1.40 0.18 0.22

1997-2001 0.68 0.04 -0.27 0.36 5.29 0.86 -1.21 2.45

2002-2006 0.51 1.36 -0.22 0.20 4.43 2.95 -1.03 1.12

2007-2010 0.37 1.53 0.11 0.37 3.62 4.18 0.62 1.91