The causal relationship between dividends and economic
growth: An analysis for Germany and The Netherlands
Bachelor thesis Daniela Rojas Morales Student number: 10167706 Faculty of Economics and Business
Finance and Organization Thesis supervisor: Philippe Versijp
February 2014
Abstract: The purpose of this paper was investigating if there is a causal relationship
between the dividend payout ratio of stock-‐listed firms and the economic growth for Germany and The Netherlands. For this purpose, stationarity-‐, co-‐integration-‐ and Granger-‐causality-‐ tests were applied. Due to restrictions with the stationarity of the data, it was not possible to form a conclusion for The Netherlands. For Germany, the results indicated that there is a uni-‐directional causality running from economic growth to dividends, without any feedback. This means that the dividend payout ratio of the stock-‐listed firms in Germany is affected by the economic growth in Germany. Contrarily, the economic growth of Germany is not affected by the dividend payout ratio of the stock-‐listed firms.
Table of Contents
1. Introduction 3
2. Literature review 4
2.1 Dividends and earnings 4
2.2 Earnings and economic growth 6
2.3 Dividends and economic growth 6
2.4 Variables analyzed 8
2.4.1 Economic growth 8
2.4.2 Dividend payout ratio 8
2.5 Hypotheses 9
2.5.1 Possible explanations hypotheses 9
3. Methodology 10
3.1 Stationarity 10
3.2 Co-‐integration 11
3.3 Granger-‐causality 11
3.4 Hsiao’s version of the Granger-‐causality test 13
4. Empirical results 13
4.1 Data 13
4.2 Results of stationarity test 15
4.3 Results of co-‐integration test 16
4.4 Results of Hsiao’s causality test 17
5. Discussions and Conclusion 17
References 19 Appendix 23
1. Introduction
Dividends are the most common method to distribute cash between firms and shareholders (Bodie, Kane and Marcus, 2005). The amount of dividend paid, varies per firm and year. Shareholders tend to prefer dividends stocks because these are generators of cash and are commonly known as a sign of financial strength. Yet, from a research by Eaton Vance (2006) it was found that only one in seven investors is aware that dividend-‐paying stocks in the S&P 500 Index have outperformed non-‐dividend payers by 8.21 percent over the past 10 years.
A lot of research has been done about the relationship between dividends and the rate of return, and dividends and the corporate earnings. From a different point of view, what this paper tries to find out is if there is a relationship between the amount of dividends paid and the economic growth of a country. Has the increase in dividend payout ratio in the last years something to do with the positive economic growth we have been experiencing with the crisis recovery? According to the literature, dividends tend to increase when earnings increase and earnings tend to follow the trend of economic growth. With this said, can a significant relationship running from economic growth to dividends be expected? Moreover, the relationship could also be expected running from dividends to economic growth as a consequence of dividend spending and consumer confidence.
Until today, research investigating this topic has been minimal. Greg IP, The Economist US economics editor, stated that a raise in dividends suggests a negative sign for the economy growth, that firms are paying the cash out as dividend due to pessimism expectations about the future. On the other hand, Mark Zandi, Moody’s analytics chief economist, argues that this raise in dividends is a benefit for the economy because about one third of this paid-‐out dividend is spent. Furthermore, Zandi argues that a raise in dividends indicates that companies are confident enough in their cash flows, think that it is safe to raise dividends and are performing better (“Dividends and Economic growth”, 2013).
Therefore, this study seeks to answer the following question: Is there is a causal relationship between the amount of dividends paid by the stock-‐listed
firms of a country and the economic growth of that country? It is focused on two European countries: Germany and the Netherlands.
The purpose of this paper is to contribute to the beginning of future research to completely disclosure the relationship between dividends and economic growth. Analyzing if the decision each firm makes regarding the amount of dividend paid influences the economic growth of the country may be relevant for the government in behalf of the investment decisions and economy stimulation. If a causal relationship running from dividends to economic growth is found, the government can boots the economy by investing or subsidizing in dividend-‐paying firms. On the other hand analyzing if the economic growth of a country influences the amount of dividend paid may be relevant for investors as one of the determinants for the investment decisions.
Using data of stock-‐listed firms and performing an ADF unit root test, an Engle-‐Granger co-‐integration test and Hsiao’s version of the Granger-‐causality test, this paper will attempt to answer this question. The paper is organized as follow: Section 2 highlights the review of related literature; Section 3 describes the used model and tests; Section 4 contains the data and the results and lastly section 5 contains the discussion, conclusion, limitations and suggestions for future research.
2. Literature review
2.1 Dividends and earnings
The first relationship to be considered is the relationship between dividend and earnings. Modigliani and Miller (1959) came up with a theorem, which stated that dividends are irrelevant in a perfect capital market. Specifically, a firm’s value is not affected by a firm’s choice of dividend policy. A perfect capital market is a market without taxes, transaction cost, information asymmetry and large buyers and sellers. However these perfect capital market assumptions are unrealistic and are inconsistent with the reality (Berk & DeMarzo, 2011).
The preference of investors for dividends certainly has consequences for the decisions a firm makes. The amount of dividends a company should pay out and how much they should keep for investment has been largely studied and has
contradicting results. On this, depends how many investors want to own the shares and at what price these will trade. Paying out too much dividend can lead to little cash left over for further investment or the risk that investors believe that there are not enough investment opportunities. On the other hand, keeping cash for investment and paying too little dividend or non at all can lead to a lack of investors and the risk of signaling a weak growth (Berk & DeMarzo, 2011).
Researchers have suggested many factors that influence a company’s dividend policy. Some focus on the aspects managers take in to account when determining the dividend. Others include market imperfections like taxes, agency costs, asymmetric information and behavioral explanations (Baker et al. 2001).
Information asymmetry arises when companies and managers have more or superior information than the investors. What information the dividends convey, has also been an intensely researched topic. Lintner (1956) was the first that suggested that companies only raise dividends when managers believe that earnings have permanently increased, which results in a different payout ratio than the firm’s targeted one. Subsequently, Miller and Modigliani (1961) suggested that managers use dividend policy to signal their expectations about future prospects of the firm, this hypothesis is known as the dividend-‐signaling hypothesis.
When a firm increases its dividends it sends the signal to investors that the manager believes the firm is able to afford higher dividends. Conversely when a firm decreases dividends, it sends the signal to investors that the firm needs to save cash and cannot afford the current dividend (Berk & DeMarzo, 2011). Several papers researching this topic support the signaling theory (see e.g. Bhattacharya 1979, Fama and French 2001, John and Williams 1985, and Miller and Rock 1985, Nissim and Ziv 2001), while others, contradicting the dividend-‐signaling hypothesis, fail to find significant evidence (see e.g. Watts 1973, DeAngelo et al. 1996, and Benartzi et al. 1997, Grullon et al. 2005). Therefore the conclusion about the validity of the dividend-‐signaling theory is indecisive.
On the other hand, dividends and earnings could also be negatively related because of the investment opportunities. When companies have more
investment opportunities the money available is used for these investments, thus less money is available to pay dividends. This implies that a decreasing dividend payout ratio is not related with decreasing earnings but with increasing future earnings (Berk & DeMarzo, 2011).
2.2 Earnings and economic growth
The second relationship considered is the relationship between earnings and economic growth. It is well known in the literature that corporate earnings are strongly pro-‐cyclical (see e.g. Blanchard and Perotti 2002, Longstaff and Piazzesi 2004). They have tended to move together with the economic growth over the decades, except for the Great Depression. Brown and Ball (1967) revealed that around 35-‐40 per cent of the variability of a firm’s earnings could be associated with the variability of the economy-‐wide earnings. Later, among other studies, Ball et al. (2009) confirmed that there exists a significant systematic influence to earnings.
2.3 Dividends and Economic growth
The last relationship that might apply and is studied in this paper is the relationship between dividends and economic growth. Although there is little literature about this, a relationship could be expected and reasoned as follows: The relationship might run from economic growth to dividends and rely on the previously explained dividend-‐signaling hypothesis and the relationship between earnings and economic growth. If dividends tend to increase when earnings increase and earnings follow economic growth, then a relationship between dividends and economic growth can be expected. Furthermore, earnings expectations increase.
On the other hand, the relationship might run from dividends to economic growth and could be explained by the fact that around one third of these dividends is spent (“Dividends and Economic growth”, 2013). In addition, dividends are known as a sign of financial strength, and thus a rise in dividends increases consumer confidence. This theory is linked with the self-‐fulfilling prophecy proposed by Robert K. Merton (1948) and the previously explained asymmetric information. If the investors believe that a rise in dividends is the
result of an increase in earnings and economic growth, consumer confidence will increase in addition to investment and consumption and in effect, the belief becomes reality. This is supported by many studies; among others, Acemoglu and Scott (1994) investigated the relationship between consumer confidence and the economy and found out that consumer confidence, measured with surveys, has a significant predictive power for growth in both consumption and labor income, even adding other economic indicators. Fig. 1 portrays the possible relationships between dividends and economic growth. Fig. 2 and 3 describe the trend of dividend payout ratio and real GDP growth, respectively, for Germany and the Netherlands over the period 2003–2012.
0 10 20 30 40 50 60 70 80 90 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 % Di vi d end p ayo u t Ra ti o Year Germany Netherlands Figure 1 Figure 2
Source: Author. Based on data from database DATASTREAM. Dividend payout ratio
2.4 Variables analyzed
2.4.1 Economic growth
Economic growth will be measured by the yearly growth of real GDP. Real GDP is the measurement for the economic output, adjusted for prices. The growth in percentage is calculated dividing the change of GDP by the value of real GDP in the past year and multiplying it by 100 (Boyes & Melvin, 2010).
2.4.2 Dividend Payout ratio
The dividends of the stock-‐listed firms of the country will be measured by the yearly average of the dividend payout ratio. The dividend payout ratio is calculated dividing the annual dividend paid by the net income. This ratio reveals how much of the net income is received by investors rather than what goes to the retained earnings account. A negative dividend pay ratio implies that the company is paying out more than the net income or that it is paying out dividend and incurred losses (Stock & Watson, 2012).
Figure 3 Real GDP Growth -‐6 -‐5 -‐4 -‐3 -‐2 -‐1 0 1 2 3 4 5 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 % R ea l G D P gr ow th Year Netherlands Germany
2.5 Hypotheses
The research question of this paper is: is there a causal relationship between dividends and the economic growth of a country? Based on this question the hypotheses can be formulated as follows:
H01: There is no uni-‐directional causality running from dividends to economic
growth.
H11: There is uni-‐directional causality running from dividends to economic
growth.
H02: There is no uni-‐directional causality running from economic growth to
dividends.
H12: There is uni-‐directional causality running from economic growth to
dividends.
H03: There is no bi-‐directional causality between economic growth and
dividends.
H13: There is bi-‐directional causality between economic growth and dividends
2.5.1 Possible explanations hypotheses
Rejecting the first null hypothesis means that there is uni-‐directional causality running from dividends to economic growth, which implies that increasing dividends leads to an increase in economic growth. Uni-‐directional causality running from dividends to economic growth could be explained by the increase in consumer confidence and the spending of the dividends.
On the other hand, rejecting the second null hypothesis means that there is uni-‐directional causality running from economic growth to dividends, which means that a higher economic growth leads to an increase in dividends. This supports the dividend-‐signaling hypothesis. If there is economic growth, firms are performing better, earnings and earnings expectations are higher and firms raise the dividends.
Finally rejecting the third null hypothesis, which occurs if the first and the second null hypothesis are rejected, means that there is a bi-‐directional causal relationship between dividends and economic growth. Economic growth causes more dividends whereas more dividends lead to economic growth.
3.Methodology
3.1 Stationarity
In the past, macroeconomic models were estimated by linear regressions without taking into account whether the variables were stationary or not. This caused spurious results, suggesting a significant relationship that was untrue (Granger & Newbold, 1973). This is why, according to Engle and Granger (1987), the first step to study if there is Granger-‐causality is to analyze if the variable set is stationary. Stationarity is when the joint distribution of the time series variable and its lagged value does not change over time, at least in a probabilistic sense. In other words, the variables have no clear tendency to return to a constant value or a linear trend. Stock and Watson define it as follow: “A time series Y is stationary if the joint distribution of (Ys+1, Ys+2,...., Ys+t) does not
depend on s, regarding of the value of T”. This is called being integrated of order zero and is usually notated as yt~I(0) (Granger, 1988). If the variables are
integrated of order zero the statistical results of a linear regression hold.
Using non-‐stationary variable sets can result in biased results; basically because it is difficult to predict the future if it is fundamentally different than the past (Stock & Watson, 2012). If the variables are stationary with the true values they are integrated of order zero. If instead, the variable is at first non-‐stationary, it can be examined if using differences results in a stationary variable set. A time series variable that is integrated of order d has been differentiated d times. If the first-‐differences are used the variables will be integrated of order one, notated as Δyt~I(1). If the second-‐differences are used the variables will be integrated of
order two, notated as Δyt~I(2) (Stock & Watson, 2012).
Two tests to investigate for stationarity are the ADF and Phillips-‐Perron tests (PP-‐test). The regression for the ADF test for the variable X is estimated in the form:
ΔXt=βXt-‐1+ !!!!(δjΔXt-‐j)+εt
Using OLS the following hypothesis are tested: H0= β=0 Xt~I(1) or Xt~I(2) (unit root)
The problem with the ADF is that it doesn’t take autocorrelation and heteroskedasticities into account. However using the HAC variance estimator suggested by Newey and West (1987) in the co-‐integration test, these problems can be confronted. Phillips and Perron (1988) developed a unit-‐root test based on the ADF test, robust to serial correlation and time-‐dependent heteroskedasticities (Cheng & Lai, 1997). The reason the ADF test is preferred in this paper over the Phillips and Perron test is that is works better with small
samples (Schwert, 1989).
3.2 Co-‐integration
Some of the approaches to examine co-‐integration that have been proposed are the Engle-‐Granger test by Engle and Granger (1987), the autoregressive Distributed Lag (ARDL) approach suggested by Pesaran et al. (2001) and the maximum likelihood-‐based approach proposed by Johansen and Juselius (1990). In this study, following Engle and Granger (1987), the Engle-‐Granger test is employed because it is intuitive and easy to perform.
Co-‐integration can be described as the co-‐movement among economic variables over the long run, when two or more variables share a common stochastic trend. According to Engle and Granger (1987) one can expect that if x and y are non-‐stationary any linear combination would also be a random walk. However if x and y are non-‐stationary and any linear combination stationary, performing a Granger causality test would lead to biased results and hence an error-‐correction model should be applied. Therefore it is necessary to test for co-‐ integration. To test for co-‐integration the variables should be integrated of the same order (Dakurah et al. 2001). A I(1) and a I(0) variable for example can not be tested for co-‐integration or causality. Here the null hypothesis is that the regressors are not significant different from zero, and thus are not co-‐integrated.
3.3 Granger-‐causality If both variable sets are integrated of order zero or integrated of the order one, but co-‐integrated, causality and direction of the causality can be tested. The method that will be used to answer the question of causality is a Granger-‐
causality test. This test proved to be better over alternative tests in a Monte Carlo experiment by Geweke et al. (1983), especially for small samples.
Wiener wrote the first literature about Granger-‐causality in 1956. He
stated that if including a time series improves the prediction of another time series, then we could speak of causality. Later, Granger (1969) ratified this idea with a linear regression model. Granger defines Granger causality as “...X causes Y if we are able to predict Y using all available information than if the information apart from X had been used”. If the minimum error variance of Y using values of X is less than the minimum error variance of Y using only values
of Y, X Granger-‐causes Y:
σ2 (X|X, Y)< σ2 (X|X)
If this holds then X is a useful predictor of Y. Similarly, bi-‐directional causality occurs when X is causing Y and Y is causing X.
In the standard Granger-‐causality two regression models and two
autoregressive distributed lag (ADL) models are specified. The regression models (the restricted form) relate the time series with its past value and the ADL models contain lags of the dependent variable and of a single additional predictor (the unrestricted form). The models can be specified as follows:
Yt=α11+ !!!!!!β11Yt-‐i+u11t (1)
Yt=α12+ !!!!!!β11Yt-‐i+ !!"!!!β12Xt-‐j+u12t (2)
Xt=α21+ !!!!!!β21Xt-‐i+u21t (3)
Xt=α22+ !!"!!!β21Xt-‐i+ !!!!!!β22Yt-‐j+u22t (4)
Where X is natural logarithm of the dividend payout ratio, Y the change of real GDP, L the number of lags and α and β are parameters. Using a F-‐test, the restricted form with the unrestricted form are compared and tested for significance of the coefficients β12 and β22 for the lagged values. The null
hypotheses of no uni-‐directional causality are rejected if one the coefficients, is different from zero.
3.4 Hsiao version of the Granger-‐causality test
Hsiao (1981) developed a causality test based on Granger’s causality test combined with Akaike’s Final Prediction Error (1969). This method is used in this paper because it is easy to implement and simplifies the choice of the optimal lag length. The first step is to compute the sum of squared errors of Eq. 1 with the different lag orders. Subsequently, the FPE can be calculated as follows:
FPE (L)=!!!!!!!!!! SSE (L)/T
Where T is the sample size, L the lag size and SSE is the sum of squared errors. The smallest FPE indicates the optimal lag L1*. Similarly, the sum of squared
errors of Eq. 2 with the different lag order and the optimal lag L1* has to be
calculated. Lastly, the FPE has to be computed as follows:
FPE (L1*, L2)=!!!!∗!!!!!!!!!∗!!!!! SSE (L1*, L2)/T
The smallest FPE indicates the optimal lags L1*, L2*. If FPE (L1*, L2*)<FPE (L1*), X
Granger-‐causes Y. Following the same steps for Eq. 3 & 4 it can determine if Y causes X.
4. Empirical results
4.1 Data
The sample of this study is a time-‐series for the time period 2003-‐2012. The decision concerning the time period is due to availability restrictions of the data. Furthermore, it is five years before the financial crisis and four years after, so this provides the possibility to study it in the different market conditions.
Using Compustat and Datastream, the annual dividend and net income of 30 firms that trade in the most important stock market indexes of each country were collected (see appendix 5 & 6 for list of the companies). The reason of using these stock-‐listed firms in the study is because these firms are almost always the largest and most influential in the country. Therefore a relationship is more
expected. To have enough data points companies of more than one index per country were used:
• The AEX and AMX for the Netherlands • The DAX and MDAX for Germany
To remain in the sample companies must have fulfilled the following requirements:
• It must have paid, at least once, dividend in the time period
• The firm’s income and annual total dividend paid for the time period must be available on Compustat or the Annual Report
The variables used in the models are: LD, natural logarithm of dividend payout ratio; and the real GDP. Because the dividend payout ratios were all positive, transforming it to natural logarithm can help if the variable is positive skewed. Including additional variables may mislead the principal objective and distract us when choosing the optimal lags.
Working with time series there are a couple of challenges like time lags, autocorrelation and heteroskedasticity of the errors. To deal with this challenges, as stated before, the heteroskedasticity-‐ and autocorrelation-‐ consistent (HAC) estimator of the variance proposed by Newey and West (1987) is used in the co-‐integration test. Additionally, using this estimator of the variance, Newey and West also proposed a formula to define the optimal lags: L*=0.75T1/3.
4.2 Results of stationarity test
As said before, because of the small sample, The ADF test is used as unit-‐root tests. First of all, plotting the variable could indicate if a drift, a trend or nothing was needed to add to perform the ADF test. Adding a drift is adding an intercept, and this has to be done when the variable has an approximate zero mean. When an ADF-‐test is performed on differenced variables a drift should not be included because differencing makes the constant part zero. When the plot of the variable shows an upward trend over time, a trend has to be added (See appendix 7 for an overview when drift or trend was added for each variable). The results for the unit-‐root test for the variables LD and Real GDP for Netherlands and Germany are reported above in table 2. A critical value of 0.10 is chosen, which is a proper level of significance for small samples (Yoo, 2005).
For Germany, the null hypothesis of no stationarity for LD is rejected at a significance level of 10% and for real GDP at al level of 5%. This means that for Germany both the dividend payout ratio and the real GDP are integrated of order zero. For the Netherlands, the null hypothesis of no integration of order zero is not rejected for the dividend payout ratio and rejected for the real GDP at a significance level of 10%.
The next step for the Netherlands after determining that the logarithm of dividend payout ratio is not integrated of order zero and thus the variables are integrated of different orders, is adding lags to the regression. Adding more lags resulted as showed above, in a higher p-‐value, which means that there is less statistical evidence to reject the null hypotheses of no stationarity. The next step is investigating if by differentiating the variable it results in a stationary variable. With the null hypothesis of integration of order one and the alternative hypothesis of integration of order two, the Dickey fuller test is performed. This results in a p-‐value of 0.0002, which indicates that the null hypothesis of integration of order one is rejected. Concluding, the variables dividend payout ratio and real GDP are integrated of different order for the Netherlands and integrated of order zero for Germany.
4.3 Results of co-‐integration test
Since it has been determined that the real GDP and the logarithm of the dividend payout ratio of Germany are integrated of the same order, the co-‐integration test can be performed. As said before, testing for co-‐integration is examining if there is a long-‐run linear relationship between the variables. The optimal lag length is defined with the Newey-‐West formula to confront heteroskedasticity-‐ and autocorrelation.
The null hypothesis is defined as no presence of co-‐integration and is rejected if the coefficient is not significant different from zero. The results are reported above in table 3. As shown in the table for both tests the p-‐value is higher than 0.10, which means that the null hypothesis of no co-‐integration is not rejected. That is, there exists no long-‐term relationship between dividend payout ratio and real GDP in the Germany. As previously mentioned, when the variable set is integrated of order zero, co-‐integration is not a requirement for testing for causality.
4.4 Results of Hsiao's version of the Granger-‐causality test
The results for Hsiao’s version of the Granger-‐ causality test for Germany are reported above in table 4. The F-‐values were computed with the restricted and the unrestricted formulas 1,2,3,4 using the optimal lag lengths L11*, L12*, L21*, L22*. As indicated, if FPE(L11*,L12)< FPE(L11*) then dividend Granger-‐causes economic growth and if FPE(L21*,L22)< FPE(L21*) economic growth Granger-‐ causes dividends.
As illustrated in the table, it appears that dividend payout ratio does not Granger-‐cause economic growth, given that 12,9692>11,0659. This is also evident from the F-‐value (1,62) and the p-‐value, that is higher than 0.1. Hence, there is not enough evidence to reject the first null hypothesis of no uni-‐ directional causality running from dividends to economic with a significance level of 10%. On the other hand given that 0,0346652<0,035989 it can be concluded that economic growth Granger-‐causes dividends. This is also supported by the F-‐value (3,661) and the p-‐value, that is smaller than 0.1. Therefore, there is enough evidence to reject the second null hypothesis of uni-‐ directional causality running from economic growth to dividends with a significance level of 10%.
5. Discussion and Conclusion
The purpose of this paper was to investigate if there is a relationship between dividends and economic growth. Previous literature only offered research on the relationship between dividends and return and dividends and earnings. To this end, the yearly average dividend payout ratio of 30 stock-‐listed companies and the economic growth data of the Germany and the Netherlands were collected. Subsequently, the stationarity of the data was analyzed with an ADF unit-‐root test and the co-‐movement of the variables with an Engle-‐Granger co-‐integration test. Lastly, Hsiao’s version of the Granger-‐causality test was performed. Due to the restrictions with the stationarity of the data it was not possible to form a conclusion for The Netherlands.
The main conclusion that emerges from this empirical research is that for Germany, as expected, there is a uni-‐directional causal relationship running from
economic growth to dividends. With this, the second null hypothesis of this study is rejected. This implies that the economic growth of a country has an impact on the amount of dividends paid by the stock-‐listed firms of the country. In other words, economic growth stimulates more dividends. As mentioned previously, this supports the dividend-‐signaling hypothesis of Miller and Modigliani (1961). When the economy grows, measured by an increase in real GDP, firms perform better and earnings and earnings expectations increase. Managers signal this increase by raising the dividend payout ratio. In the last years with the economic crisis there has been an economic decline in Germany. This decline could be the cause of the fact that German firms are paying fewer dividends out.
The absence of a uni-‐directional relationship from dividends to economic growth might indicate that the spending of dividends and the increase in consumption because of consumer confidence is not large enough. Thus, with a significance level of 10%, the first and third null hypotheses of this study are not rejected. Any increase in dividends will not significantly affect real GDP.
The results of this paper implicate the decisions investors make regarding investing in one stock or another because of the dividend paid. Investors can take the economic growth and earnings as determinants for the investment decisions.
The research question can now be answered. According to the empirical
evidence of this study, there is causal relationship between economic growth and dividends. This relationship is not bi-‐directional but runs from economic growth to dividends.
There are limitations to this research. One of the biggest limitations of this study is that the research sample is small. Only 30 companies were used for each country and the availability of the data was restricted to 10 years. Even though the tests applied were chosen for working better with small data, these test have a higher statistical power with large samples. Future research can take in to account more countries and it is recommended to use a larger sample of companies and time period. In the future research using a larger period might conclude in a bi-‐directional relation and a conclusion for The Netherlands. This paper serves as the beginning of further investigation to disclosure the causal relationship between economic growth and dividends completely.
References
Acemoglu, D., & Scott, A. (1994). Consumer confidence and rational expectations:
are agent’ beliefs consistent with the theory?. The Economic Journal, 104,
1-‐19.
Ali, A., & Urcan, O. (2012). Dividends increases and future earnings. Journal of Accounting & Economics, 19:1, 12-‐25.
Akaike, H. (1969). Fitting autoregressions for prediction. Annals of the Institute of
Statistical Mathematics, 21, 243–247.
Baker, H. K., Veit, E. T., & Powell, G. E. (2001). Factors influencing dividend policy
decisions of nasdaq firms. Financial Review, 38:3, 19-‐38.
Ball, R., & Brown, P. (1967). Some preliminary findings on the association between the earnings of a firm, its industry, and the economy. Journal of
Accounting Research, 5, 55-‐77.
Ball, R., Sadka, G., Sadka, R. (2009). Aggregate earnings and asset prices. Journal of Accounting Research, 47, 1097-‐1133.
Benartzi, S., Michaely, R., & Thaler, R. H. (1997). Do changes in dividends signal the future or the past?. Journal of Finance, 52, 1007–1034.
Berk, J., & DeMarzo, P. (2011). Corporate Finance (2nd ed.). Harlow: Pearson
Education Limited.
Bhattacharya, S. (1979). Imperfect information, dividend policy, and the “bird in the hand” fallacy. Bell Journal of Economics, 10, 259–270.
Blanchard, O., & Perotti, R. (2002). An Empirical Characterization Of The
Dynamic Effects Of Changes In Government Spending And Taxes On
Output. Quarterly Journal of Economics, 107, 1329-‐1368.
Bodie, Z., Kane, A., & Marcus, A. J. (2005). Investments (6th ed.). New York:
McGraw Hill.
Boyes, W., & Melvin, M. (2010). Economics (8th ed.). Boston: South-‐Western
College Pub.
Cheng, B. S., & Lai, T. W. (1997). An investigation of co-‐integration and causality
between consumption and economic activity in Taiwan. Energy
Economics, 4, 435-‐444.
Cassel, A. (2013, November 22). Dividends and Economic growth. CNBC
Retrieved January 5, 2014 from
https://www.economy.com/dismal/blog/blog.asp?cid=244181
Dakurah, H., Davies, S., & Sampath, R. (2001). Defense spending and economic
growth in developing countries, a causality analysis. Journal of Policy
Modelling, 23, 651–658.
DeAngelo, H., DeAngelo, L., & Skinner, D. (1996). Reversal of fortune: dividend
signaling and the disappearance of sustained earnings growth. Journal of
Financial Economics, 40, 341–371.
Eaton, V. C., (2006, January 18). First-‐Ever Eaton Vance Dividend Study Surveys
Top Finance Executives at Dividend-‐Paying US Corporations. Retrieved
December 20, 2013 from
www.eatonvance.com/alexandria/pressreleases/200601/Dividend
%20Surve%20Release.pdf
Engle, R. F., & Granger, W. J. (1987). Co-‐integration and error correction:
representation, estimation and testing. Econometrica, 55, 251-‐276.
Fama, E. & French, K. (2000). Forecasting profitability and earnings. Journal of
Business, 73, 161–175.
Fama, E. & French, K. (2001). Disappearing dividends: Changing firm
characteristics or lower propensity to pay. Journal of Financial Economics,
60, 3–44.
Geweke, J., Meese, R., & Dent, W. (1983) Comparing alternative tests of causality in temporal systems. Journal of Econometrics, 21, 161-‐194.
Granger, C. W. J. (1969). Investigating causal relations by econometric models
and cross-‐spectral methods. Econometrica, 37:3, 424-‐43.
Granger, C. W. J., & Newbold, P. (1973). Spurious regressions in econometrics.
Journal of Econometrics, 2, 111-‐120.
Granger, C.W.J. (1988). Some recent developments in a concept of causality,
Journal of Econometrics, 39, 199-‐211.
Grullon, G., Michaely, R., Bernartzi, S., & Thaler, R. H. (2005). Dividend changes
do not signal changes in future profitability. Journal of Business, 78, 1659
1682.
Healy, P. & Palepu, K. (1988). Earnings information conveyed by dividend
initiations and omissions. Journal of Financial Economics, 21, 149–176.
Hsiao, C. (1981). Autoregressive modeling and money-‐income causality detection. Journal of Monetary Economics, 7, 85–106.
Johansen, S., & Juselius, K. (1990). Maximum Likelihood Estimation and Inference
on Co-‐integration – with Applications to the Demand for Money. Oxford
Bulletin of Economics and Statistics 52, 169-‐210.
John, K. & Williams, J. (1985). Dividends, dilution, and taxes: A signaling
equilibrium. Journal of Finance, 40, 1053–1070.
Lintner, J. (1956). Distribution of incomes of corporations among dividends,
retained earnings, and taxes. American Economic Review, 46, 97–113.
Longstaff, F. A., & Piazzesi, M. (2004). Corporate earnings and the equity
premium. Journal of Financial Economics, 74, 401-‐421.
Melvin, M., & Boyes, W. (2010). Economics (8th ed). South Western: Cengage Learning
Merton, R. K., (1948). The self-‐fulfilling prophecy. The Antioch Review, 8, 193
210.
Michaely, R., Thaler, R., & Womack, K. (1995). Price reactions to dividend
initiations and omissions: overreaction or drift?. Journal of Finance, 50,
573–608.
Miller, M. & Modigliani, F. (1961). Dividend policy, growth, and the valuation of
shares. Journal of Business, 34, 411–433.
Miller, M. & Rock, K. (1985). Dividend policy under asymmetric information.
Journal of finance, 40, 1031–1051.
Newey, W. K., & West, K. D. (1987). A simple positive semi-‐definite,
heteroskedasticity and autocorrelation consistent covariance matrix.
Econometrica, 55, 703-‐708.
Nissim, D. & Ziv, A. (2001). Dividend changes and future profitability. Journal of
Finance, 56, 2111–2133.
Penman, S. H. (1983). The predictive content of earnings forecasts and dividends. Journal of Finance, 38, 1181–1199.
Pesaran, H., Shin, Y., & Smith, R. (2001). Bound testing approaches to the analysis of level relationships. Journal of Applied Econometrics.16, 289-‐326.
Schwert, G. W. (1989). Why does stock market volatility change over time. Journal of Finance, 44, 1115-‐1153.
Stock, J. H., & Watson, M. M. (2012). Introduction to econometrics (3d ed.). Harlow: Pearson Education Limited.
Watts, R. (1973). The information content of dividends. Journal of Business, 46,
191-‐211.
Wiener, N. (1956). The theory of prediction. In Beckenbach, E. (ed.): Modern
Mathematics for Engineers. New York: McGraw-‐Hill.
Yoo, S. H. (2005). Electricity consumption and economic growth: evidence from Korea. Energy Policy, 33:12 , 1627-‐1632.
Appendix
Appendix 1
Average dividend payout ratio for The Netherlands for the period 2003-‐2012
Year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Dividend Payout ratio 30,63 63,67 35,49 34,61 36,28 24,61 38,28 46,93 41,29 78,26 Appendix 2
Average dividend payout ratio for Germany for the period 2003-‐2012
Year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Dividend Payout ratio 27,02 38,40 42,57 34,50 42,96 55,54 74,27 45,24 60,53 49,81 Appendix 3
GDP growth rate for The Netherlands for the period 2003-‐2012
Year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 GDP growth rate 0,337 2,237 2,046 3,394 3,921 1,804 -‐3,668 1,528 0,945 -‐1,247 Appendix 4
GDP growth rate for Germany for the period 2003-‐2012
Year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 GDP growth rate -‐0,387 0,694 0,846 3,886 3,389 0,807 -‐5,085 3,857 3,399 0,896 Appendix 5
List of the 30 DAX-‐ and MDAX-‐ listed firms ALLIANZ SE
AURUBIS AG
AXEL SPRINGER VERLAG AG BEIERSDORF AG
BILFINGER SE
BMW-‐BAYER MOTOREN WERKE AG DAIMLER AG
DEUTSCHE BANK AG DEUTSCHE BOERSE AG DEUTSCHE EUROSHOP AG DEUTSCHE POST AG DEUTSCHE TELEKOM ELRINGKLINGER AG
FRESENIUS MEDICAL CARE AG&CO FUCHS PETROLUB SE
LEONI AG LINDE AG MERCK KGAA METRO AG
MTU AERO ENGINES AG MUNICH RE CO
SAP AG SIEMENS AG
THYSSENKRUPP AG VOLKSWAGEN AG
WEBER (GERRY) INTERNATNL AG BASF SE BAYER AG ADIDAS AG Appendix 6
List of the 30 The AEX-‐ and AMX-‐ listed firms
AEGON NV AIR FRANCE -‐ KLM AKZO NOBEL NV ASML HOLDING NV CORIO NV FUGRO NV HEINEKEN NV KONINKLIJKE AHOLD NV KONINKLIJKE KPN NV KONINKLIJKE PHILIPS NV POSTNL NV RANDSTAD HOLDINGS NV REED ELSEVIER NV ROYAL DSM NV
ROYAL DUTCH SHELL PLC ROYAL IMTECH NV SBM OFFSHORE NV UNILEVER NV WOLTERS KLUWER NV ARCADIS NV BINCKBANK NV BOSKALIS WESTMINSTER NV CSM NV HEIJMANS NV
KONINKLIJKE BAM GROEP NV KONINKLIJKE TEN CATE NV
Appendix 7
Overview when drift or trend was added for each variable
Log dividend payout
ratio Real GDP growth
Germany -‐ Drift Levels Netherlands -‐ Drift Levels Appendix 8
Variables used in research
Variable construction
Dividend payout ratio
Total dividend paid
Net Income
Real GDP growth For year n=( !"#$ !"# !" !"#$ ! !(!"#$ !"# !" !"#$ !!!))
(!"#$ !"# !" !"#$ !!!) *100
NIEUWE STEEN INVESTMENTS NV NUTRECO NV
VOPAK (KONINKLIJKE) NV UNIT 4 NV