The effect of dividend policy on share price volatility in the Netherlands
Thesis by Rachella KordijkName: Student number: Field of study: Specialization: University: Thesis supervisor: Year of graduation: Rachella Kordijk 10016805
Business and Economics Finance and Organization University of Amsterdam Natalya Martynova 2014
Introduction
Whether to distribute profits to shareholders in the form of a dividend or to utilize them for investment opportunities should not only depend on the magnitude of the company’s income that is required for investment. Managers should also take into account the effect of dividend policy on stock price. On average, investors are risk-averse by nature so they prefer a less volatile stock, which is considered a safer and more stable investment. However, the relationship between dividend policy and stock price risk remains a source of controversy despite years of theoretical and empirical research. Over time, the topic has been examined based on listed companies from different countries in the world and results are contradictory. Most case studies are based on the investigation of U.S. companies by Baskin (1989) and the research conducted on Australian firms by Allen and Rachim (1996).
In addition to contradicting data from previous research, there is still no evidence based on listed companies in the Netherlands. The Netherlands is a civil law country with a
stakeholder-oriented governance system and strong ownership concentration. La Porta et al. (2000) found evidence that civil law countries pay lower dividends than common law countries like the United States on average, which results from the fact that common law countries usually have stronger minority
shareholder protection. This conclusion corresponds to the situation in the Netherlands since the Dutch corporate governance regime is such that most publicly listed companies impose severe restrictions on shareholder control. When domestic Dutch companies reach a certain size, they are required to adopt an institutional form named the structured regime, which aims to increase the rights of the works council at the expense of shareholders’ rights so that shareholders, among others, have less influence on determining dividend policy. In addition, more than 90% of companies in the Netherlands use one or more devices that restrict shareholder power: they are able to use poison pills and issue preference shares, priority shares and certificates. Renneboog and Szilagyi (2006) investigated the effect of this severe shareholder power restriction on dividend payout and they found that the payout ratio of Dutch companies is low and dividends are moderately smoothed. A remarkable result they reported is that the dividend decision of Dutch firms depends on operating cash flows instead of reported earnings, which is more common across the globe. Also, they concluded that dividend dynamics do not have a significant relationship with the intensity of agency problems, as proxied by firm size, leverage, and investment opportunities. They argue that their findings with respect to the low payout ratio could be extended to other countries with a stakeholder-oriented corporate governance system, but admitted that their argumentation for this is incomplete. However, they provided evidence that dividend policy in the Netherlands differs because of their stakeholder-oriented corporate governance regime with its special features. Hence, it could be the case that the effect of dividend policy on share price volatility in the Netherlands differs from this relationship in other countries, especially countries that adopt a shareholder-oriented governance approach such as the United States. This thesis aims to investigate the effect of dividend policy on share price volatility in the Netherlands.
To shed light on the effect of dividend policy on share price volatility in the Netherlands, dividend yield and payout ratio are compared to the volatility of the share price by estimating regression models. Since the two determinants of share price volatility are strongly correlated with each other, multicollinearity could be a problem. This is resolved by avoiding to include them in one and the same regression model at first. To control for certain factors that influence both share price volatility and dividend policy, control variables are included in the regression specifications. It is expected that payout policy has a negative relationship with volatility of share price. It has been found that payout ratio exhibits a significant negative relationship with share price volatility, after controlling for Size, Debt, Growth, Earnings volatility and Industry type, so this confirms the hypothesis. Against expectations, dividend yield appeared to have a significant positive relationship with share price volatility. Both Growth and Industry type appeared to have an insignificant effect on share price volatility and the R-squared of the regression model is relatively low. This could be the result of the small magnitude of the sample or poor definition of one or more of the variables used in the model.
First, the literature regarding the relationship between dividend policy and share price volatility is discussed. Thereafter, the hypothesis is formulated, the methodology of this thesis is explained, and the results are elaborated. Finally, the research question is answered and some suggestions for future research is given. The latter includes some drawbacks of this research.
Literature review
One of the most prominent investigations on the effect on dividend policy on share price volatility is done by Baskin (1989). He proposed four models which relate dividends to share price risk: the duration effect, the rate of return effect, the arbitrage realization effect and the informational effect.
The duration effect predicts that the price of high dividend yield stocks is less volatile as compared to the price of low dividend yield stocks. This is explained by the fact that high dividend yields imply more near-term cash flow, which results in the stock being less sensitive to fluctuations in discount rates and therefore show lower price volatility, according to Baskin (1989). However, both Easley and O’Hara (1992) and Xu (2013) found evidence that a short duration results in higher price volatility due to the impact of duration on the arrival of new information and following Baskin’s reasoning, this should result in higher share price volatility when dividend yield is higher. It is therefore questionable whether the duration effect that Baskin (1989) proposed still holds in today’s world.
The rate of return effect explains the issue from another perspective. According to pecking order theory (Donaldson, 1961; Myers & Majluf, 1984), companies show a preference regarding sources of financing: asymmetric information affects the decision whether to fund their projects by means of internal financing, issuing equity or issuing debt. The theory states that managers issue equity only as last resort. They prefer funding projects with retained earnings, which do not involve asymmetric
information, followed by debt financing. Because of this prioritization of retained earnings as compared to issuing equity, it may be rational for companies with large investment requirements to pay low dividends and use the cash retentions to fund their projects. This reasoning suggests that dividend yields and payout ratios may serve as proxies for growth opportunities. If predictions of profits from growth opportunities are more subject to error than forecasts of earnings from assets already in place, which is plausible, prices of shares from companies with low payout and low dividend yields are more volatile as compared to prices of shares from companies with high payout and high dividend yields, according to Baskin (1989). However, dividend changes may convey an ambiguous signal about growth opportunities since increases in dividend payout could also suggest the firm is disinvesting (Easterbrook, 1994). Whether the rate of return effect holds in practice, therefore, depends on the degree in which the market can distinguish between growing firms and disinvesting firms.
The third effect that Baskin (1989) proposed, arbitrage realization effect, is based on inefficiency of the financial market. Due to this inefficiency, it is possible to realize profit from mispricing and the higher the dividend yield, the greater the scope for such an arbitrage profit. This is demonstrated analytically:
Let P* be the present value of future dividends, A the discount from the intrinsic value of the stock, and P the actual value of the stock. By definition, P equals (1-A)P*. An investor who is fully informed and knows the magnitude of underpricing and the value of the expected dividend in each period, considers that the other investors are not aware of his superior information so that the stock continues to sell at a discount of 1-A at the end of the investment period. The appropriate discount rate (Ke) equals D/P* + g, where D is the expected dividend over the period and g is the expected capital gain. The informed investor buys at price P and his expected return (Ka) equals the expected dividend yield (D/P) and expected capital gain (g). So the investor’s return equals:
Ka = D/P + g = Ke + A(D/P)
As the formula specifies, the informed investor’s excess return over the period equals the product of the price discount and the dividend yield. Hence, the higher the dividend yield, the bigger the scope for realizing an arbitrage profit and the more volatile is the share price. However, arbitrageurs face a trad-off between arbitrage and inference. That is, they want to buy shares if their price temporary declines (the arbitrage effect) but they want to sell if their price permanently declines (the inference effect). According to Adrian (2004), arbitrage increases share price volatility when the inference effect dominates the arbitrage effect. He argues that what effect dominates depends on the amount of
uncertainty about the dividend policy that arbitrageurs face. If they face little uncertainty, the arbitrage effect dominates and therefore share price volatility is lower. If uncertainty about dividend policy is
high, the arbitrage effect is weaker than the inference effect and share price volatility is higher. Hence, whether the arbitrage realization effect that Baskin (1989) proposed holds such that higher dividend yield results in more arbitrage and lower share price volatility, depends on the degree of uncertainty about dividend policy.
As Miller and Rock (1985) suggest, a firm’s information can be transmitted to the public by dividend payments. By paying dividends while at the same time providing the earnings announcement, investors will have more confidence in the firm’s stability. This is named the informational effect and it indicates that it managers may be able to influence the share price volatility using dividend policy. For instance, they could increase the payout ratio to reduce volatility. However, some recently conducted studies found evidence that does not support the signaling function of dividends that is suggested by the informational effect (Chen et al., 2002; Abeyratne & Power, 2002; Vieira & Raposo, 2007; Hobbs & Schneller, 2012), so it could be the case that the informational effect does not hold anymore. This could be caused by the fact that there is many more information available nowadays, such that the additional informational content of dividends is negligible when investors’ form their beliefs about stability of the firm. According to Kuo and Lee (2013), magnitude of dividend tax burden on investors affects the signaling effect of dividends, so even if the informational effect still would appear to hold in today’s world in most cases, a certain tax system could result in absence of the effect in a particular country.
In summary, all four effects predict a negative relationship between both dividend yield and payout ratio, and share price volatility. The duration effect and the rate of return effect treat dividends as a proxy for the timing of the underlying cash flow of the company. The arbitrage realization effect and the informational effect suggest that managers actively influence stock market risk. Baskin (1989) reported that a 1% increase in dividend yield causes a decrease of approximately 2.5% in the annual standard deviation of share price movements. However, it is questionable whether all of the four effects hold for all countries and even whether they still hold in today’s world in general.
After the investigation of Baskin (1989) and his proposition of the four models that relate dividend policy to volatility of the share price, the topic has been examined over time by different researchers, based on listed companies from different countries. Suleman et al. (2011) and Yasir et al. (2012) studied the relationship between dividend policy and share price volatility based on companies listed on the Karachi Stock exchange and found a positive relationship between dividend yield and share price volatility. This result is inconsistent with Baskin’s (1989) conclusion and could be evidence for the absence of one or more of the effects that the latter proposed. Pakistan has a unique tax system and following the reasoning of Koa and Lee (2013), this could cause the informational effect not to hold. The way in which the arbitrage realization effect applies could also be different, since it is likely that uncertainty about dividend policy in the emerging economy of Pakistan is bigger than in the developed economy of the United States. However, besides the reasoning that the effects might or might not hold, the contradicting results could also be caused by the fact that Pakistan
employs a stakeholder-oriented corporate governance system. Renneboog and Szilagyi (2006) argued that they expect a low payout ratio for countries with a stakeholder-oriented corporate governance system and this expectation is in line with the evidence Ho (2002) found in his comparative study of dividend policy in Australia and Japan. Opposed to Anglo-Saxon countries like the United States and Australia, Japan pursues a corporate governance system that focuses on all stakeholders instead of just shareholders and Ho concluded that Japan has a significant lower payout ratio than Australia. He reported that the environment has a significant influence on dividend policy and therefore findings about the relationship between dividend policy and share price volatility in countries that employ different corporate governance systems could differ from each other.
Allen and Rachim (1996) found that dividend yield is not correlated with share price volatility. They suggest that payout ratio, size of the firm, level of debt and earnings volatility are the main determinants of share price volatility instead. These results imply that there is no duration or arbitrage effect, since these effects suggest that dividend yield instead of payout ratio is the relevant
determinant. They neither support the rate of return effect as this effect suggest that both dividend yield and payout ratio determine share price volatility. In addition, the effect is likely to be correlated with the growth rate of the firm so when controlling for growth rate, the coefficients on dividend yield and payout ratio should change if the rate of return effect exists. Allen and Rachim (1996) did not observe such a change. They did find evidence supporting the informational effect, however. The absence of evidence for three of the four effects could be possibly explained by the fact that dividend policy in Australia is different from that in the United States due to differences in tax system between the two countries. Australia employs an imputation tax system instead of a classical tax system as is implemented in the United States, which induces higher payout of dividends.
Hussainey et al. (2011) studied the effect of dividend policy on share price volatility in the U.K. and found evidence in agreement with Baskin’s (1989) results. They showed a negative relation between share price volatility and both payout ratio and dividend yield. These corresponding results could possibly be explained by the fact that the United States and the United Kingdom are both Anglo-Saxon countries that employ a shareholder-oriented corporate governance approach, which result in similar dividend policies in the two countries even though they apply different tax systems: the United Kingdom employs an imputation tax system. Hashemijoo et al. (2012) found similar results from their study based on the Malaysian stock market. They argued that the duration effect, the rate of return effect, the arbitrage realization effect and the information effect apply to Malaysia. The consistency of findings could be explained by the fact that dividend policy in Malaysia is similar to dividend policy in the United States due to a similar tax system: both countries employ a classical tax system. Differences between the United States and Malaysia appear in for example the corporate governance regime: Malaysia pursues a stakeholder-oriented approach, while the United States employ a
shareholder-oriented approach to corporate governance. This implies that corporate governance system is not the predominant factor that causes differences in results regarding the relationship
between dividend policy and share price volatility. Differences in tax systems and more importantly the way and magnitude in which the duration effect, the rate of return effect, the arbitrage realization effect and the informational effect hold, are more important.
Methodology and hypothesis
Dividend policy in the Netherlands differs from that in the United States because of its stakeholder-oriented corporate governance regime with its special features (Renneboog and Szilagyi, 2006). That is, the Dutch corporate governance system causes publicly listed companies to impose severe
restrictions on shareholder control which results in a lower payout ratio on average, according to Renneboog and Szilagyi (2006). Their finding is in accordance with the evidence La Porta et al. (2000) found. They showed that common law countries, like the United States, pay higher dividends than civil law countries, like the Netherlands, on average. This is explained by the fact that common law countries usually have stronger minority shareholder protection.
It is expected that the duration effect in the way Baskin (1989) proposed does not hold for the Netherlands for the period 2008-2013. The reasoning that higher dividend yield results in higher share price volatility because of the effect of duration on the arrival of new information appears to be more applicable to today’s world. The rate of return effect is expected to apply since it is expected that the market can distinguish between growing firms and disinvesting firms due to the many information sources available in developed countries nowadays. The arbitrage realization effect is also expected to hold, because dividends in the Netherlands are moderately smoothed so uncertainty about dividends is relatively low. Therefore, the arbitrage effect dominates the inference effect and share price volatility is lower when dividend payout is higher. Finally, the informational effect is expected to hold for the Netherlands for the time period 2008-2013 too. Since both the United States and the Netherlands apply an imputation tax system, dividend tax burden on investors shows similarities and therefore the
signaling function and hence the informational effect is expected to hold for the Netherlands too. This study aims to investigate the effect of dividend policy on share price volatility in the Netherlands. Since it is expected that three of the four effects hold for the Netherlands for the time period studied, it is hypothesized that there is a negative relationship between dividend policy and share price volatility. That is, the hypothesis is that both dividend yield and payout ratio are negatively related to share price volatility. Because it is expected that the duration effect works in a way such that dividend policy and share price volatility are positively related, it is expected that the effect of
dividend policy on share price volatility is weaker than Baskin (1989) concluded. It is also expected that the fact that payout ratio in the Netherlands is lower than in the U.S. further mitigates the magnitude of the effect.
The sample that is used for investigation consists of the 24 companies that have been listed on the AEX since its start and that DataStream has information about for the time period studied. All of the panel data that is utilized in the analyses is obtained from DataStream, if not stated otherwise.
To answer the research question, a couple of least squares regression analyses are applied. The univariate specifications relate share price volatility and the two main measurements of dividend policy, dividend yield and payout ratio, for a period of five years from January 1st, 2008 to January 1st, 2013. The multivariate specifications relate share price volatility to the basic explanatory variable and a couple or all of the control variables. That is, the following equation is estimated:
Share price volatility = β0+ β1 Dividend yield + β2 Payout ratio + β3 Size + β4 Growth + β5 Debt + B6 Earnings volatility + β7 Dummy + ε
Share price volatility is calculated by using the extreme value method for estimating the variance of the rate of return that Parkinson (1980) derived. It is used by Baskin (1989) and all of the more recent studies on the topic also adopted this method (Allen & Rachim (1996), Hussainey et al. (2011), Habib et al. (2012), Hashemijoo et al. (2012), Zakaria (2012), Ramadan (2013)). Parkinson showed that this method is superior to the traditional one of simply using the standard deviation as proxy for share price volatility, not only because it is easy to apply in practice and a smaller time interval is needed to get reliable results, but also because of the fact that the calculation provides a better estimate of the diffusion constant than the traditional calculation does. The latter makes sense since the true variance of the rate of return of a share is equal to the diffusion constant of the underlying random walk, which is the value that characterizes the random walk of share prices. For each year, therefore, the variance of a company’s share price is calculated by dividing the annual range of share prices by the average of the highest and the lowest share price and then raised to the second power. The square root of the average of the variances for all years is taken to transform them into a standard deviation, which represents share price volatility of the particular company.
Dividend yield is computed by dividing annual dividends per share by price per share. This ratio is averaged over all years studied. Payout ratio is expressed as the ratio of annual dividends per share to earnings per share and it is averaged over all years studied.
In addition, control variables are included in the regression model. They control for certain factors that influence both share price volatility and dividend policy. Size is one of those control variables since there are potential connections between size and share price volatility. Small
companies are likely to be less diversified in their activities and are less subject to investor scrutiny, which results in the stock market for smaller firms to be less informed and more illiquid and therefore suffering from greater price volatility (Allen & Rachim, 1996). In addition, Baskin (1989) suggests that companies with more dispersed shareholders, i.e. bigger firms, tend to more frequently use dividend as signaling device. Different studies use different measures as a proxy for firm size.
However, as discussed by among others Wald (1999), Mackay and Phillips (2002), and Dalbor et al. (2004), the natural logarithm of total assets is the best proxy for firm size and therefore this proxy is used in this thesis. The natural logarithms of total assets are averaged for the period studied. Royal Imtech and Post NL both have no assets at all on January 1st, 2013. The natural logarithm of zero should have been taken when calculating Size for these companies, but this is not possible. Therefore, the formula is not applied in calculating the value for Size for that year for both companies. Since they did not have any assets, zero is simply used as their value for Size.
Earnings volatility is also used as a control variable. Share price volatility is linked to the risks in the company’s product market and market risks could affect firm’s dividend policy (Allen & Rachim, 1996). Therefore, a control variable to account for the variability in the company’s earnings stream should be used. In accordance with Dichev and Tang (2009), earnings volatility is represented by the standard deviation of earnings before interest and taxes for the period studied. For Reed Elsevier, EBIT information is not found in DataStream for 2010 until 2013. Therefore, these figures are obtained from the company’s annual reports for those years. The same applies to Royal Wessanen for the year 2013.
Share price volatility is related to leverage because of operating risk. Since asymmetric information is involved, it is plausible to argue that borrowing is also related to dividend policy. To account for the leverage of the firm, debt is used as a control variable and is expressed as the ratio of total debt to total assets. The ratio is averaged for all years studied. For 2013, DataStream does not have information about the ratio of total debt to total assets from Royal Wessanen. Therefore, the figures for total debt and total assets for this year are obtained from the financial reports on the company’s website.
Growth is included as control variable too. The higher the company’s growth rate, the less dividends they pay to their investors since the earnings are retained to fund the company’s investment opportunities. Since the duration effect and the rate of return effect assume differences in timing in the company’s cash flows, it is likely to assign any relationship between dividend policy and share price volatility to the arbitrage effect or the informational effect, after there is controlled for growth in assets. Growth is expressed by averaging the ratio of change in total assets in a year to the level of total assets at the beginning of the year over all years studied.
Finally, it is possible that the relationship between dividend policy and share price volatility is caused by differences across different industries in addition to individual policies. Therefore, a dummy variable is included in the regression model. This dummy distinguishes between service companies and manufacturing companies. It equals zero for service companies and one for manufacturing companies.
Results
Table 1 reports some descriptive statistics of the variables included in one or more of the regression models. The average share price volatility in the Netherlands between 2008 and 2013 equals 46.49%, which is less than the average share price volatility that Baskin’s (1989) reported. He noted 61.42% for the sample of U.S. firms. If it is assumed that share prices are normally distributed and the small effect of corporations going ex-dividend is ignored, Parkinson (1980) argues that the standard deviation of stock market returns equivalent to share price volatility can be estimated by multiplying the reported average share price volatility by the constant 0.6008. This gives a standard deviation of 27.93%, which is again smaller than Baskin’s (1996) result of 36.9%. It is worth to remark that the standard deviation of payout ratio is approximately half of the magnitude of its mean. Also, the variation in price volatility is relatively high. This could be caused by the small amount of companies that could be included in the sample.
Mean Std. deviation Min. Max.
Price volatility 0.4649 0.1729 0.1574 0.9571 Dividend yield 3.6965 11.9348 0.0000 8.2300 Payout ratio 37.2053 18.0668 0.0000 74.3017 Size 14.2532 1.5563 10.3429 16.5268 Growth 0.0900 0.1563 -0.1829 0.5222 Debt 25.5964 12.6616 0.5200 59.9367 Earnings volatility 401094.4000 359292.8000 41589.3500 1207285.0000
Table 1: descriptive statistics variables
Table 2 reports correlations between the variables. As expected, the correlations between share price volatility and both dividend yield and payout ratio are negative, a low correlation of -0.3818 and a correlation of -0.6320 respectively. Baskin (1986) reported correlations of -0.643 and -0.542
respectively. Dividend yield and payout ratio are significantly correlated with each other (r = 0.5531), which could result in multicollinearity that causes errors when regressing share price volatility on both of them simultaneously. Therefore, they are not included in one and the same regression model at first. First, share price volatility is regressed on dividend yield and the full list of control variables. Results of all regression specifications are shown in table 3. As expected, the coefficient on dividend yield is negative. However, it is not significant. The coefficients on all of the control variables are insignificant too. This may be the result of the small size of the sample or the large spread in the variables, especially share price volatility and payout ratio. To see whether results would become significant when another determinant of dividend policy is used, payout ratio is included in the regression model at the expense of dividend yield. It appears that payout ratio has a significant but small negative relationship with share price volatility: when payout ratio increases by 1%, share price volatility declines by 0.63%. The coefficients on all of the control variables remained insignificant after replacing dividend yield for payout ratio. This could imply that the control variables are
inappropriate to control for certain factors that influence both share price volatility and dividend policy in the Netherlands, even though they seemed to serve as appropriate control variables for other
countries.
Price vol. Div. yield PO ratio Size Growth Debt Earn. vol. Price vol. 1.0000 Div. yield -0.3818 1.0000 PO ratio -0.6320 * 0.5531 * 1.0000 Size -0.3036 0.0486 0.3356 1.0000 Growth 0.2395 -0.5349 * -0.2126 0.0430 1.0000 Debt -0.1448 0.4506 * 0.0933 -0.1132 -0.2845 1.0000 Earn. vol. 0.0120 -0.0445 -0.0490 0.5149 * -0.0905 -0.0367 1.0000 Table 2: correlations
* = significant at the five-percent level
(1) (2) (3) (4) (5) Dividend yield -0.0279 (0.0273) 0.0024 * (0.0261) Payout ratio -0.0063 * (0.0020) -0.0060 * (0.0021) -0.0060 * (0.0016) -0.0062 * (0.0025) Size -0.0439 (0.0288) -0.0091 (0.0280) -0.0101 (0.0221) -0.0090 * (0.0289) Growth 0.1285 (0.2802) 0.1395 (0.2126) 0.1366 (0.2713) 0.1294 (0.2454) Debt -0.0000 (0.0035) -0.0004 (0.0026) -0.0009 (0.0025) -0.0002 * (0.0030) Earnings volatility 0.0000 (0.0000) 0.0000 (0.0000) 0.0000 * (0.0000) Industry dummy -0.0235 (0.0875) -0.0832 (0.0750) -0.0463 (0.0721) -0.0852 (0.0806) R² 0.2691 0.4721 0.4416 0.3994 0.4723 Adjusted R² 0.0111 0.2857 0.2865 0.3721 0.2415 Number of obs. 120 120 120 120 120
Table 3: Regression specifications
* = significant at the five-percent level
As shown in table 2, earnings volatility is highly correlated with size. Its correlations with both dividend yield and payout ratio are relatively low. In addition, table 3 shows that the insignificant coefficient of earnings volatility equals 0.000 in both specification (1) and (2), which implies that it does not have a significant positive or negative effect on share price volatility. To check whether the use of earnings volatility as a control variable causes the insignificance of all control variables, the variable is dropped from regression (3). Surprisingly, the coefficients on all control variables remain insignificant. The R-square of the model becomes even lower: only 44.16% of the variance in share
price volatility is explained by all of the independent variables together. Of all control variables utilized in all of the regressions, only growth has an (insignificant) positive effect on share price volatility: if growth increases by 1%, share price volatility goes up by 13.66%. This is also
unexpected, since it is hypothesized that growth would have a negative effect on share price volatility. Since the coefficients of all of the control variables remain insignificant in all specifications, all of the control variables are excluded to estimate a single linear regression model, including only payout ratio as independent variable. The significant coefficient on payout ratio equals -0.0060, which is exactly the same as resulted from specification (3). This might suggest that the control variables are not important in estimating the effect of dividend policy on share price volatility in the Netherlands. To check whether this is the case, a restricted F-test is conducted. The F-statistic equals 0.3592 (df = 4,19). Using a 5%-significance level, this means that the single linear regression model does not explain the relationship between dividend policy and share price volatility better than the model from specification (4). This could have been anticipated by comparing the square of both models. The R-square of the regression from specification (5) is lower: the variance in payout ratio explains only 39.94% of the variance in share price volatility.
Even though dividend yield and payout ratio are correlated with each other and multicollinearity could appear when they are used as independent variables simultaneously, a
regression model that includes both of them together is estimated (6). The full list of control variables is also included in the specification. This is done to see whether the correlation between dividend yield and payout ratio is weakened by controlling for certain variables. If this is the case, significant results could be obtained despite the correlation of 0.5531 between the two determinants of dividend policy. As expected, payout ratio is negatively related to share price volatility: a 1% increase in payout ratio results in a 0.62% decrease in volatility of the share price. The coefficient is also significant.
Unexpectedly, dividend yield has a significant relationship with share price volatility: if dividend yield increases by 1%, share price volatility increases by 0.24%. The coefficients on Size, Debt and
Earnings volatility are also significant. If size increases by 1%, share price volatility decreases by 0.90%. This negative relationship is expected, since the stock market for smaller firms tend to be less informed and more illiquid and therefore suffers from greater price volatility. The results also show that if a firm leverages up, share price volatility declines. However, even though this effect is significant, it is only small: if debt goes up by 1%, share price volatility decreases by 0.02%.
Strikingly, the significant coefficient of Earnings volatility still equals 0.0000. The number is rounded to four decimal places, so that is why an effect that appears to equal 0.00% is still significant. The insignificant coefficient of growth implies that if growth increases by 1%, share price volatility increases by 12.94%. The insignificant coefficient of the industry dummy implies that the volatility of manufacturing firms’ share price is 8.52% lower than the volatility of service firms’ share price on average. The R-squared of this regression specification equals 47.23%, which means that the variance in Payout ratio, Dividend yield, Size, Debt, Growth, Earnings volatility and Industry type explains
47.23% of the variance in share price volatility. Even though it is still low, it is higher than the R-squared of all other specifications. To check whether this specification is best indeed, a restricted F-test is conducted. The F-statistic equals 0.3917 (df = 6,17), Using a 5%-significance level, this means that the single linear regression model does not explain the relationship between dividend policy and share price volatility better than the model from specification (5).
Conclusion and discussion
The aim of this study is to investigate the effect of dividend policy on share price volatility in the Netherlands. It is expected that both payout ratio and dividend yield exhibits a negative relationship with volatility of the share price. The sample that is used consists of 24 companies that are listed on the AEX index. Single and multiple linear regression analyses are applied, differing in the amount of (control) variables used.
The findings suggest that both dividend yield and payout ratio are determinants of dividend policy in the Netherlands indeed. As expected, there is a small but significant negative effect of payout ratio on share price volatility. Dividend yield appeared to have a significant positive relationship with share price volatility in the Netherlands. This is against the expectations and suggests that the duration effect such as Baskin (1989) proposed does not hold for the Netherlands for the period 2008-2013. Instead, the evidence found supports the belief that because of the impact of duration on the arrival of new information, higher dividend yield results in higher share price volatility in the Netherlands nowadays, which is exactly what both Easley and O’Hara (1992) and Xu (2013) stated. The
coefficients of the control variables Growth and Industry dummy appeared to be insignificant. The R-squared of the model is also relatively low. This could be explained by the small size of the sample. In addition, the sample size might be the reason why the variation in variables, especially payout ratio and share price volatility, is relatively high. The AEX is most commonly used as a proxy for the Dutch stock market, but it only includes 25 companies on average. It is tried to get the sample as extended as possible by including all companies that have been listed on the index since its start, but DataStream has information about 24 of them for the period studied only. Another explanation of insignificance of those control variables might be associated with the definition of them, especially regarding the distinguishing of industry type. The distinguishing between service companies and manufacturing companies might be too simple and possibly discriminating between more industry types would have benefited the results. Further research should be conducted regarding the optimal distinguishing between industry types.
References
Allen, D.E. & Rachim, V.S. (1996). Dividend policy and stock price volatility: Australian evidence, Journal of Applied Economics, 6 (1), pp. 175-188.
Adrian, T. (2004). Inference, arbitrage, and asset price volatility, Journal of Financial Intermediation, 18 (1), pp. 49-64.
Baskin J. (1989). Dividend policy and the volatility of common stock, The Journal of Portfolio Management, 15 (1), pp. 19-25.
Dalbor, M. C., Kim, A., & Upneja, A. (2004). An initial investigation of firm size and debt use by small restraurant firms, Journal of Hospitality Financial Management, 12 (1), pp. 41-48. DeAngelo H., DeAngelo L., & Skinner D. J. (1996). Reversal of fortune: dividend policy and the
disappearance of sustained earnings growth, Journal of Financial Economics, 40 (3), pp. 341-371.
Dichev, I. & Tang, V. (2009). Earnings volatility and earnings predictability, Journal of Accounting and Economics, Journal of Accounting and Economics, 47 (1), pp. 160-181.
Donaldson, C. (1961), Corporate debt capacity: A study of corporate debt policy and the determination of corporate debt capacity. Harvard University Press, Cambridge, MA.
Easley, D. & O’Hara, M. (1992). Time and the process of security price adjustment, The Journal of Finance, 47 (2), pp. 577-605.
Easterbrook F. H. (1984). Two agency-cost explanations of dividends, The American Economic Review, 74 (1), pp. 650-659.
Habib, Y., Kiani, I. Z., & Kahn, M. A. (2012). Dividend policy and share price volatility: Evidence from Pakistan, Global Journal of Management and Business Research, 12 (5), pp. 79-84. Hussainey, K., Mgbame, C. O., & Chijoke-Mgbame, A. M. (2011). Dividend policy and share price
volatility: UK evidence, The Journal of Risk Finance, 12 (1), pp. 57-68.
Khan, T. (2006). Company dividends and ownership structure: Evidence from UK panel data, The Economic Journal, 510 (1), pp. C172-C189.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. W. (2000). Agency problems and dividend policies around the world, The Journal of Finance, 55(1), pp. 1-33.
MacKay, P., & Phillips, G. M. (2002). Is there an optimal industry financial structure? (Working Paper 9032). Retrieved from: https://webkuliah.unimedia.ac.id/ebook/files/mackay-p
phillip(2002)%20insutry%20finstru.pdf
Miller, M. H., & Rock, K. (1985). Dividend policy under asymmetric information, The Journal of Finance, pp. 1031-1051.
Moh'd, M. A., Perry, L. G., & Rimbey, J. N. (1995). An investigation of the dynamic relationship between agency theory and dividend policy, The Financial Review, pp. 367-385.
Myers, S.C. & Majluf, N. (1984). Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics, 13 (2), pp. 187-221. Nazir, M. S., Musarat, M., Waseem, N., & Ahmed, A. F. (2010). Determinants of stock price volatility
in Karachi stock exchange: The mediating role of corporate dividend policy, International Research Journal of Finance and Economics, 55 (1), pp. 100-107.
Parkinson, M. (1980). The extreme value method for estimating the variance of the rate of return, Journal of Business, 53(1), pp. 61-65.
Ramadan, I. Z. (2013), Dividend policy and price volatility: Empirical evidence from Jordan, International Journal of Academic Research in Accounting, Finance and Management Sciences, 3 (2), pp. 15–22.
Renneboog, L. D. R., & Szilagyi, P. G. (2006). How relevant is dividend policy under low shareholder protection?, Discussion Paper 2006-73, Center for Economic Research.
Wald, J. K. (1999). How firm characteristics affect capital structure: an international comparison, Journal of Financial Research, 22(2), pp. 161-187.
Zakaria, Z., Muhammad, J., & Zulkifli, H. A. (2012). The impact of dividend policy on the share price volatility: Malaysiian construction and material companies, International Journal of
