The impact of dividends on the stock return in a period of crisis

The impact of dividends on the stock return

in a period of crisis

Master thesis

University of Groningen

Faculty of economics and Business

MSC Business Administration

Profile: Corporate financial management

By: Bart Snijders

Studentnumber: 1473336

The impact of dividends on the stock return

in a period of crisis

Abstract:

This study analyzes data from 348 Dutch and German firms around the crisis

period that started in 2007 to investigate the relationship between the

dividends paid out by firms and their stock return in times of crises.

While Fama and French (1998) and Pinkowitz and Williamson (2002) find

that in the long term there is a strong positive influence of dividends on firm

value due to dividends signaling, the results of this study show that in a

crisis period this strong influence is absent. Using a similar model, it is

found that in the crisis period, there is no significant relationship between

dividends and stock return. This result is explained by the high interest costs

and scarcity of cash within firms that occur in crisis periods, which make it

more attractive for firms to retain more of their earnings and pay a smaller

part out as dividends.

Key words: Dividends, dividends policy, dividends signalling, stock return and crisis

Table of Content

1. Introduction 4

2. The crisis period and theoretical background research question 7

2.1 The crisis of 2007 7

2.2 The impact on interest rates 7

2.3 The consequences for firms 9

3. Literature review 12

3.1 Relevancy dividend policy 12

3.2 The information content of dividends and dividends signalling 13 3.3 Dividend payout ratios and the traditional finance theory 14

3.4 Dividend policy vs. future earnings 15

3.5 Tax considerations 15

3.6 Dividend policy vs. stock performance 16

4. Data and methodology 18

4.1 Data and sample period 18

4.2 Methodology 19

4.2.1. Comparative analysis 20

4.2.2. OLS analysis 21

5. Results 29

5.1 Descriptive statistics 29

5.2 Results comparative analysis 33

5.3 Results OLS analysis 35

6. Discussion 41

7. Conclusion and limitations 43

7.1. Conclusion 43

7.2 Limitations of the paper and recommendations future research 44

1. Introduction

From the 1950’s, researchers look at the impact of a certain dividend policy on the stock performance of a firm. The dividend irrelevancy theory of Miller and Modigliani (1961) is seen as a good theoretical framework, but its practical appliance is limited since the assumption of a world without market imperfections is not realistic. A popular theory in explaining the effect of dividend policy on stock return is build around taxes. Dividends are seen as income for stockholders and these are more heavily taxed than capital gains. Therefore, the tax theory presumes that a higher dividend payout policy results in a lower stock price for the firm (Brennan, 1970). Few papers find evidence for the tax theory of dividends, for instance Black and Scholes (1974) and Miller and Scholes (1982) are not able to find support. Although few evidence is found for the tax theory of dividends, it remains prominent in the literature, possibly because of its intuitive sense.

Fama and French (1998) (F&F (1998) from now on) create a new model to test the effects of dividends on the market value of a firm. The model is mainly based on

regression equations that have the market value of the firm as the dependent variable and which have several static and dynamic independent variables that are shown or thought to have an influence on the market value of the firm. Dividends, changing dividends and changing dividend policies are taken as independent variables in this model and their influence on the stock return is tested. F&F (1998) hypothesize that due to the negative effects of taxes for dividends, high dividend paying firms should have lower market returns. However, they find the opposite result. With a good control for profitability, investment and debt, they find that dividends are positively related with firm value. They infer from their results that dividends convey important information about the

argue that there is an information content of dividends and that firms use cash dividend announcements to signal changes in the expectations of future prospects of the firm. In 2002, Pinkowitz and Williamson (P&W (2002) from now on) take an almost similar model as the one used in F&F (1998) and use it to investigate the market value of cash holdings for firms. As a by-product, they again find support for the positive affect of higher dividends on the market value of a firm, using a different sample period as F&F (1998).

Both F&F (1998) and P&W (2002) take a long sample period of 30 to 40 years to

investigate the impact of dividends on the value of a firm. By doing this, they do not take into account effects of the phase in the economical cycle or other economical

circumstances (that are specific for a certain period) in the relationship between dividends and firm value. In other words, they do not test whether the positive relation between dividends and the market value of the firm is present continuously, in both booming economical periods and in times of recession.

In this paper, I investigate whether the positive influence of high dividend payments on the stock return is also present in periods of (financial) crisis. As further explained in the next section, especially in financial crises, interest rates tend to be high. These high interest rates increase the costs of external financing for companies. Furthermore, during crises, the cash inflow for companies tends to be low. These developments can cause problems for a firm in the form of not being able to pay back their financial claims. Also other negative consequences can arise such as not being able to invest in profitable investment projects. Hence, when these developments occur, it could be better for a firm to pay out less dividends and retain a larger part of their earnings as cash within the company. Therefore, the impact of a certain dividend policy on the value of the firm could be different in a period of crisis.

were relatively high. The crisis starting in 2007 is an example of a (financial) crisis and the occurences of this year are used to demonstrate the developments which can occur during such crises.

2. The crisis period and theoretical background research question

2.1 The crisis of 2007

In 2007, a worldwide financial crisis began to originate, many financial institutions faced uncollectible loans and had to write off a large amount of assets. Europe’s biggest bank, HSBC, had to write down $20 billion for loan losses because there was trouble

recollecting mortgages from borrowers with poor credit ratings (Bloomberg1). BNP Paribas faced a 20% asset write off in three investment funds in only a couple of weeks time (Bloomberg2). US bank Merrill Lynch & Co reported in the third quarter of 2007 its biggest quarterly loss ever after having taken $8.4 billion of write offs on their assets due to bad loans (Bloomberg3). These are just a few examples of the causes of the financial crisis, which hit most banks worldwide. Most problems for the financial institutions were caused by the large amounts of bad loans that they had and the asset write offs that had to be done to account for the impairment losses on these bad loans. Faced with all these write offs, banks became hesitant and untrustworthy to take on more risk by providing new bank loans. This hesitance of banks to provide new loans is common to financial crises.

2.2 The impact on interest rates

This hesitance and lack of trust of banks lead to a change in the market interest rates, in two specific ways. Firstly, the Euribor rate in 2007 reached the highest level in 5 years (www.euribor-rates.eu4). The Euribor rate is the rate at which European banks borrow funds to each other, which is based on observations from a panel of 57 European banks. This interbank lending rate can be seen as an important determinant of the interest rates offered by banks on the market. In table 1, for instance, the monthly Euribor rates for 3

month loans for 2006 and 2007 are stated. Clearly, the Euribor rates are much higher in 2007 in comparison with 2006 and they are continuously rising.

5

Table 1. Euribor rates 3 months loans

In this table, the 3 months Euribor rates are of 2006 and 2007 are shown. This is the interest rate that European banks charge each other for loans with a 3 month maturity.

Month Rate Month Rate Month Rate Month Rate Jan-06 2.488% Jul-06 3.055% Jan-07 3.725% Jul-07 4.176% Feb-06 2.554% Aug-06 3.170% Feb-07 3.785% Aug-07 4.264% Mar-06 2.666% Sep-06 3.267% Mar-07 3.856% Sep-07 4.741% Apr-06 2.818% Oct-06 3.424% Apr-07 3.927% Oct-07 4.791% May-06 2.860% Nov-06 3.536% May-07 4.017% Nov-07 4.598% Jun-06 2.944% Dec-06 3.638% Jun-07 4.124% Dec-07 4.839% Source: http://www.euribor-rates.eu

Figure 1 compares the Euribor rates of loans with a 3 months maturity of 10 different years. In this figure it is visualized that the Euribor was at a clearly high rate in 2007, the highest rate of the past 5 years at that time. This high interbank rate is charged, along with a mark up, to the customers which are taking a loan from these banks.

6

Figure 1. Euribor 3 months interest rate (%).

Years. Over 10 year period

Secondly, the risk premium that banks asked over the loans they issued started to rise in 2007, as is seen in figure 2. In this graph, the difference in spread between an A2/P2 paper and a nonfinancial commercial paper is displayed, which shows the rate mark up

5

Sources: rates.eu/euribor-2007.asp?i1=6&i2=1 and http://www.euribor-rates.eu/euribor-2006.asp?i1=6&i2=1

6

that is used when commercial paper bears more risk. Clearly, in 2007 the difference in rates between the more risky thirty day A2/P2 loan and the less risky AA nonfinancial commercial paper is larger than in the 5 years before. This indicates a higher risk premium in 2007. A higher risk premium means that banks are asking for a higher interest rate when there is a specific default risk involved in a certain loan a customer is requesting for. Since all commercial parties bear risk to some extent, the mark ups on the rates of their requested loans increase.

Thus, both the increased Euribor rates and the higher risk premium cause the market interest rates to rise. This has consequences for companies in the real sector.

7

Figure 2. Discount rate spread.

2.3 The consequences for firms

The higher market interest rates and lack of trust by banks at periods of financial crises make it more difficult and more expensive for firms to get loans from these banks. The decreasing opportunities for attractive financing do not come at a good time for firms. Due to the crisis, consumers and firms have less faith in the economic future and as a consequence they decrease or delay their spending. This leads to a lower turnover, operating income and thus to lower incoming cash flows for companies. Hence, firms can

more easily get into liquidity or solvability problems. Not being able to get affordable financing can bring firms into problems because they need to pay back their claims to stay financially healthy as a firm, while their incoming cash flows are lower in a crisis. It is essential for most firms to meet their claims to be able to keep their business operations running and still generate incoming cash flows, therefore enough cash must be available.

The more likely liquidity problems and the changing market interest rates can be battled by companies by decreasing the amount of dividends that they pay out. According to Sawicki (2009), some firms decreased their dividends drastically after the onset of the crisis as the need to reserve cash became eminent to deal with the lack of capital and liquidity. Myers and Majluf (1984) recognize that restricting dividends is a good way for firms to build up financial slack and avoid negative consequences. When firms decrease their dividends, more cash will be available and this cash can be used to pay back the claims of creditors in order to stay financially healthy. Staying financially healthy, a firm can prevent damage to its reputation, business operations and contact with business partners.

Furthermore, next to being able to pay back the necessary claims to creditors, firms can avoid also the following negative consequences by restricting their dividends:

1. Not being able to make use of all available profitable investment opportunities. When a firm has a small cash position and borrowing is difficult, it might have to forego on profitable investment opportunities that arise (Myers and Majluf, 1984). These investment opportunities would increase the net value of the firm and therefore not being able to invest in these investment opportunities can be seen as a value reducing development. By decreasing the dividends paid out, more cash can be retained and this cash can be used for investment in these projects.

unattractive. A reduction in the dividend would be a very inexpensive way of providing necessary capital for the firm (Black and Scholes, 1974).

Hence, default on financial claims and several other negative consequences can be avoided when firms pay out less dividends and thus retain a larger part of their earnings within the company as cash holdings. Avoiding these negative consequences reduces the costs of the firm or increases its revenues, both of which are beneficial for firm value. Koller et al. (2005) argue that stock prices reflect real fundamental firm values, at least in the long term. Therefore, I investigate whether the positive signaling affect of dividends on the firm value is also present in a period of crisis. In other words, the goal of this research is to see whether the positive relationship between dividends and stock return, as was found by for instance F&F (1998) and P&W (2002), is also visible in a crisis period. I expect that the positive ‘signaling’ influence of dividends on the stock return is thwarted by the advantages of retaining more cash in the company in the crisis period, resulting in the absence of a clear positive relationship between dividends and stock return. Another scenario would be that I do find a strong positive relationship between dividends and stock return, this would lead to the conclusion that the positive signaling impact of dividends on stock returns is also present in a period of crisis. Additionally, the argumentation could be made that firms that decrease their dividends are often firms with risky future cash flows and since investors do not like this insecurity, these firms are punished with a lower firm value and stock return than firms that do not decrease their dividends.

To investigate whether the ‘dividend signalling’ positive impact of dividends on the stock return is visible in a crisis period with high interest rates, a nearly similar model as in F&F (1998) and P&W (2002) is used with data collected from the financial crisis that started in 2007.

3. Literature review

In this section I will describe the theory underlying my research hypothesis and the papers that have already been written on the topic are discussed. In section 3.1 elaborates upon the relevancy of dividend policy and specifically it is dealt with why firms pay attention to their dividend policy. After that, in section 3.2, I will explain the information asymmetries that are present between managers and investors and why this can lead to dividend signaling. In section 3.3 discusses the relevancy of the dividend payout ratio and it explains the traditional finance theory about the effect of dividends on stock return. Section 3.4 is devoted to describing the literature on how dividends affect future earnings. This research topic is related since when stock prices reflect real fundamental values (Koller et al. 2005), expected higher future earnings should be rewarded with a higher stock price. Section 3.5 dicusses the influence of tax and especially how tax affects the attractiveness of dividends. Finally, section 3.6 describes the papers with a comparable research topic as in this paper and which in particular look at the relationship between paying out dividends and the stock return or firm value.

3.1 Relevancy dividend policy

3.2 The information content of dividends and dividends signaling

There is an information asymmetry between the managers and the shareholders of a firm. For instance, Miller and Rock (1985) find evidence that there are clear information asymmetries between the investing public and the firm’s decision makers. Investors are not as fully aware of the future prospects of the firm as the managers of a firm is. Although, there is regular reporting from the firm, a firm will not report as detailed about projects as its managers have knowledge. Furthermore, firms will not report all information about the future of the firm that could benefit competitors to a large extent or firms might not report information on future opportunities which are relatively insecure or yet unknown to the public.

These information asymmetries between the managers and shareholders are the cornerstone of dividend signaling models (Pal and Goyal, 2007). Dividend signaling models imply that dividends have an information content (Aharony and Swarv, 1980) which indicates that managers use cash dividends to signal changes in their expectations about the future prospects of a firm. Miller and Rock (1985) recognize that dividend policy is used by corporate managers to transmit private information to the market. For the long term, Lintner (1956) finds that managers try to maintain a policy with stable dividends in the foreseeable future. This is confirmed in the survey done by Baker et al (2002) in which the NASDAQ managers stress the importance of dividend continuity for the firm. A firm will therefore only increase its dividends if it thinks that the earnings can be permanently increased (Benartzi et al., 1997) and not when earnings increase incidentally. Von Eije et al (2008) find that instead of increasing dividends, firms use temporary increases in earnings or excess cash mostly for share repurchases as they provide more flexibility than cash dividends.

3.3 Dividend payout ratios and the traditional finance theory

Although it is interesting to see whether the extra information in dividend amounts leads to a lower or higher stock price for the firm, it is just as interesting to look at the influence of dividend payout ratios of firms on their stock price. Current research focuses mainly on the impact of higher dividends on the stock price, while the impact of a higher dividend pay out ratio on the stock price is largely neglected (Gwilym et al, 2006). For example, a higher dividend/earnings pay out ratio would mean that firms would pay a larger part of their earnings out as dividends, showing off a sign of confidence that dividends in the future are expected to be high as well and according to the dividends signaling model this would lead to a higher stock price.

3.4 Dividend policy vs. future earnings

Related to the question of whether high dividends or a high dividend payout ratio will lead to a positive or negative stock price development is the question what will happen to the future earnings of a company if that company has high dividends. The traditional finance theory states that as the dividend payout ratio is lower, future earnings will grow (Gwilym et al. 2006). This theory is again based on the view that a lower payout ratio and thus a larger retention of retained earnings of the company leads to a larger cash holding for the firm which it can use to reinvest in the business. These reinvestments would generate larger returns for the company in the future. Although, this reasoning sounds plausible, several other papers find a contradictive result.

Arnott and Asness (2003) investigate the US market and find historical evidence that firms with higher dividend payout ratios experience higher future earnings growth. Their evidence is strengthened by several robustness checks. Gwilym et al (2006) repeat this investigation using a sample of 11 countries, to see whether this result holds for different institutional, tax and legal environments. They get the same results which is that higher dividend pay out ratios lead to higher future earnings. The results of these papers are in line with the signaling theory (Arnott and Asness, 2003) in which companies with high potential signal this information to the investors by means of high dividend payout ratios. Their result would indicate that given that share prices reflect real fundamental company values, a higher dividend pay out ratio would lead to higher future stock prices. While, traditional finance theory comes to the conclusion that due to the lower retention of earnings, future earnings would be lower and this would lead to lower future stock prices.

3.5 Tax considerations

dividends states that a firm that pays out high dividends has a lower return on the stock in comparison with firms that pay out few dividends.

Besides the finding that investors have to pay more taxes on their dividend gains than on capital gains, these taxes also increase the cost of capital of the firm. F&F (1998) elaborate on this by using the assumptions that the capital gains on common stock are tax-free and that investors have to pay a 50% tax on their personal income out of dividends. In this case, the cost of capital of an all-equity firm that does not pay dividends is about half the value of an equivalent all-equity firm that pays out all its returns as dividends. This larger cost of capital that a high dividend paying firm is facing reduces the value of the firm for the investor. In other words, there is predicted to be a negative tax effect on the value of dividends for investors (Pinkowitz et al, 2006). Although, the theories about a tax effect on the pricing of dividends sound plausible, most papers testing the tax theory find no significant tax effect. For instance in the papers by Black and Scholes (1974), Miller and Scholes (1982), F&F (1998) and P&W (2002) no clear support is found for the tax effect on dividends.

3.6 Dividend policy vs. stock performance

remote relationship between dividends and stock return and this rejects the popular opinion in that time that high dividend paying firms have smaller stock returns because dividends are more heavily taxed.

4. Data & Methodology

4.1 Data & sample period

The data for this research is collected from Thomson Datastream. Initially the sample consisted of 144 firms, which stocks are listed at Euronext Amsterdam in the Netherlands and 1272 firms listed at the Frankfurt stock market in Germany. However, due to the rather large amount of data needed for this research (several financial variables and mostly for several years) a large fraction of these firms proves not to be applicable. Of these firms, there is at least for one variable for one year no data available. This selection results in a total sample of 82 Dutch firms and 350 German firms. These 432 firms are classified in 10 different industry groups. As I will explain in the OLS analysis section, firms from the financial or utility industry are not included in the sample data. There are 10 utility firms and 74 firms from the financial sector that are removed from the sample, leading to the ultimate sample of 348 firms. How these firms are distributed (in percentages) among the 8 industries is displayed in figure 3.

Figure 3. Distribution of firm s in industry categories

Oil and Gas, 4%

Health care, 8% Industrials, 32% Consumer goods, 16% Health care, 8% Consumer services, 13% Telecommuni-cations, 1% Technology, 18%

However, due to the rather large amount of missing R&D numbers, in this study that would lead to a bias in the results and therefore the R&D variables are skipped in the analysis.

The data is slightly adjusted by trimming the outliers. Clear outliers in the data can significantly downgrade the quality of the results of the analysis (Comrey, 1985). F&F(1998) and P&W (2002) deal with outliers by deleting the 0.5% and 1% most extreme outliers in both tails of each explanatory variable. The disadvantage of this method is that when using many variables, many data points get lost for the analysis. Therefore, I trim the data for this analysis in another way. The top and bottom 1% most extreme values of each variable are set equal to the values corresponding to the 99% and 1% respectively of that variable. By doing this, extreme outliers are trimmed, but no extra data points are lost.

The sample period of this research is from 01-01-2007 untill 01-01-2008 (year t), this is the first year that the credit crisis was present in the worldwide economy and in this year there were high interest rates on the capital markets. Since high interest rates make cash a scarcer good, it is particularly interesting to investigate the return consequences of a certain dividend policy in this period. It would not be possible to choose the sample year 2008, because the model that is used requires data of the following year and data of the year 2009 is not yet available at the time of this research. The reason that a period of one year is chosen is that it is consistent with the studies of F&F (1998), P&W (2002) and Black and Scholes (1974).

4.2 Methodology

describes the OLS analysis that is done. The advantage of the OLS analysis in comparison with the comparative analysis is that it is possible in the OLS analysis to control for other determinants of stock return, in order to investigate solely the influence of dividends on the stock return.

4.2.1. Comparative analysis

To perform the comparative analysis, I form a group with high return companies and a group with low return companies. To create these two groups, the following steps are taken. The first step is to calculate the individual returns of the sample stocks in the sample period. The stock returns consist of the sum of the capital gains/loss on the share and the value of the dividends paid out. After having calculated the returns on all shares in the sample, I rank them in an order of return that they granted during the sample period. The highest return stock is on top of the list and the lowest return stock is last on the list. Then I divide the companies into two groups of equal size, one half of the companies on the list with the highest return compose the ‘high return group’ and the other half of the companies on the list with the lowest return compose the ‘low return group’.

The variables that are compared in the analysis are listed in table 2. Furthermore, table 2 also states how these independent variables are constructed. These variables are also included in the regression analysis and a thorough explanation of these variables and why they are used can be found in the OLS analysis section.

which is appropriate for deviations of normality and different variances between the two groups is the Welch F-test (Eviews guide) and therefore I use this test for comparing the means of variables of the high and low return group. This test, like all statistical tests done for the purpose of this paper, is performed in the statistical application Eviews.

Besides comparing the means of the variables of the low and high return group, I also consider the median value of the variables of both groups. Due to the non-normality of the data, also a non-parametric test is needed to compare the medians of variable values of both groups. Non-parametric tests do not take an assumption about the shape of the distribution of returns (Brown and Warner, 1980). For the analysis, I use a Wilcoxon Signed ranks test, which can be used to assess whether two samples have a similar median.

4.2.2. OLS analysis

After having done the comparative analysis, I perform a regression analysis. The dependent variable in the regression analysis is the stock return in the sample period, which consists of the capital gain/loss during the period and the value of the dividends paid out. This dependent variable is slightly different than the dependent variable used by F&F (1998) and P&W (2002), who use the ratio of firm value over assets and who look at the difference that occurs in this ratio. The main difference between the stock return and the difference in the value/assets ratio is that value looks at the combined value of the debt and equity of the firm, while the stock return considers only the value of the equity of the firm. In this paper, I use the stock return because I consider it to be most relevant for equity investors to know whether they receive generally a higher return in high dividend paying firms.

Table 2 presents the non industry independent variables that are used in the regression analysis and the comparative analysis.

To test whether the dividends of the firm significantly influences the return on the stock, I use five independent variables. The first variable is the dividend/assets (D/A) ratio of year t, this variable is meant to explain the effect of the current D/A ratio on the stock performance in year t. Then, two variables are used to indicate the change in the level of dividends. dDt/At measures the change in dividends in year t with respect to year t-1, divided by total assets and dDt+1/At measures the change in dividends in year t+1 with respect to year t, divided by total assets. These two variables are used to measure the effect of a change in the level of dividends on the stock return of the firm.

Table 2. List of non industry independent variables

In this table, all non industry independent variables are shown. Dt represents the dividends in year t At represents total assets in year t, It represents interest costs in year t and Et represents earnings at year t The change in the level of a financial figure is always taken w.r.t. the level of the figure in the previous year. At the right hand side of the table it is shown how the variable is constructed.

Variable The variable is constructed by

Dt/At Dividing the level of dividends in year t by total assets in year t dDt/At Dividing the change in assets in year t by total assets in year t dDt+1/At Dividing the change in assets in year t+1 by total assets in year t

d(Dt/At) Dividing the dividend/assets ratio of year t by the dividend assets ratio of year t-1 d(Dt+1/At+1) Dividing the dividend/assets ratio of year t+1 by the dividend assets ratio of year t Et/At Dividing the level of earnings in year t by total assets in year t

dEt/At Dividing the change is earnings in year t by total assets in year t dEt+1/At Dividing the change in earnings in year t+1 by total assets in year t It/At Dividing the interest costs in year t by total assets in year t

dIt/At Dividing the change in interest costs in year t by total assets in year t dIt+1/At Dividing the change in interest costs in year t+1 by total assets in year t

d(It/At) Dividind the interest/assets ratio of year t by the dividend assets ratio of year t-1 d(It+1/At+1) Dividing the interest/assets ratio of year t+1 by the dividend assets ratio of year t dAt/At Dividing the change in total assets in year t by the total assets in year t

Hence, for the dividend payout ratio I use the ratio between dividends and assets (D/A), just as F&F (1998) do, instead of the ratio between dividends and earnings (D/E). Lintner (1956) and McManus et al. (2004) state that using the dividend earnings ratio in research has found support in several papers. McManus et al. (2004) suggests that the D/E payout ratio signals information beyond the level of dividends. However, there is an important limitation in using the D/E ratio in research, which is put forward by F&F (1998). The D/E ratio can be easily influenced by temporary fluctuations in earnings, which generates many extreme values and meaningless ratios. For example, the D/E ratio can explode when earnings are close to 0. When the earnings are negative, the D/E ratio will also become negative and this ratio will be meaningless. Both low earnings and negative earnings are frequent occurrences in times of crisis and therefore the D/E ratio is not used as the dividend payout ratio. By using the D/A ratio, I do take account of the relative size of the dividends with respect to the size of the firm.

The other non industry independent factors that I use in this analysis are earnings, growth and interest costs. Except for R&D costs, of which there is unfortunately not enough data available, these are the same factors as are included in the model of F&F (1998). F&F (1998) and P&W (2002) find that a higher level of profitability leads to a higher firm value and therefore earnings variables are included in the model.

The current earnings are represented by the variable Et/At and the change in earnings is represented by a variable that consists of the change in the earnings of the current year w.r.t. the past year (dEt/At) and a variable that consists of the change in the (expected) earnings of next year w.r.t the earnings in this year (dEt+1/At).

this year in comparison with last year (dAt/At) and a variable is formed to measure the (expected) growth of the assets in the next year in comparison with the current year (dAt+1/At).

Leverage can have two effects on the return for shareholders. A highly levered firm can have a relatively high operational income but a low return for their equityholders due to the high fixed interest costs. However, a highly levered firm has more debt financing than equity financing and since the required return by debt holders is normally lower than the required return by equityholders, more remaining return can go to the shareholders, increasing their return on equity. For these two reasons, the ratio of Interest costs/Assets and the change in interest costs can alter the equity’s return on the stock of a firm. Therefore, I use a variable in the analysis for the current Interest costs/Assets ratio (It/At), a variable to take into account the change in interest costs in the current year w.r.t last year (dIt/At) and a variable to take into account the (expected) change in the interest costs of next year (dIt+1/At). However, next to the possible effect of the level of interest costs on the stock return, the ratio between interest costs and assets is another possible determinant of the stock return (F&F, 1998). This ratio can be used to represent the leverage policy of the firm and changes in this ratio can be seen as changes in the company's leverage policy (F&F, 1998). Therefore, I add a variable to represent the change in the I/A ratio in this year with respect to the I/A ratio of last year (d(It/At)) and a variable to represent the (expected) change in the I/A ratio of next year (d(It+1/At+1).

Rs = a0 + a1*Dt/At + a2*dDt/At + a3*dDt+1/At + a4*Et/At + a5*dEt/At

+ a6*dEt+1/At + a7*It/At + a8*dIt/At + a09*dIt+1/At + a10*dAt/At +

a11*At+1/At + e

(1)

And

Rs = a0 + a1*Dt/At + a2*d(Dt/At) + a3*d(Dt+1/At+1) + a4*Et/At +

a5*dEt/At + a6*dEt+1/At + a7*It/At + a8*d(It/At) + a09*d(It+1/At+1) +

a10*dAt/At + a11*At+1/At + e

(2)

OLS analysis with industry variables

The independent variables in the first two regression equations are all also included in the papers of F&F (1998) and/or P&W (2002). However, next to the elements used in these papers that help to explain the stock return of a firm, Beck et al. (2008) and Tutticci et al. (2007) argue that the industry or sector where a company operates in is also an important determinant of the stock price development of a company. A reason for this industry effect on stock return is the different growth patterns of industries, during certain periods particular industries perform generally better in comparison with other industries (Beck et al. 2008). In addition, the outlook of industries is very different, which has an important impact on the valuation of the company and thus also on the stock return. Furthermore, Tutticci et al. (2007) argue that controlling for industry is also a good proxy for R&D costs, because R&D costs tend to be largely industry specific. Since, due to lack of data, R&D costs are not included in this analysis, taking into account the industry can also be seen as a substitute for R&D costs variables. Moreover, controlling for industries makes it possible to incorporate industry specific factors in the model.

the FTSE indexes8. The ICB system is used to segregate different industries or sectors within the macro economy, mainly the firms are categorized based on their source of revenue or where it constitutes the majority of their revenue9. The ICB recognizes 10 different industries in which all firms globally can be partitioned, these industries have a different code which can be assigned to companies belonging in that industry. An overview of the 10 different industries and their corresponding codes can be found in table 3.

Of the 10 industries that are classified in the ICB system, 2 industries are excluded from the analysis. Following P&W (2002), the Utilities industry and the Financials industry (in red assigned in table 3) are excluded. The firms with one of these industry classifications are eliminated from the sample because they are intrinsically different than the firms from the other industries. P&W (2002) state that financial firms are different because the unique role that cash plays for these companies, this can affect their dividend policy. They argue that utility firms are different because there is a small differential for these companies between the costs of internal and the costs of external funds, also affecting the consequences of paying dividends for these firms.

Table 3. List of industry categories of the ICB In this table the industry classification and the codes of the Industry Classification Benchmark are shown. Also the variable name we gave to the industry in the analysis is listed.

ICB code Industry Variable name

1 Oil & Gas O1

The firms belonging to the other 8 industries are included in the analysis. Following Beck et al. (2008) and Tutticci et al. (2007), I use dummy variables to include industries in the analysis. This is done by including for 7 out of 8 industries in the analysis a dummy variable, only the technology industry does not have a dummy variable. Every industry variable has the value of 1 if the company belongs to that industry and has the value of 0 if the firm does not belong to that industry. Hence, if the firm belongs to one of the 7 industries of which the variable is included in the analysis, one of those industry variables has the value of 1. If the firm belongs to the technology industry, all industry variables have the value of 0. In table 3, the names of the dummy variables that belong to each industry are displayed. Because it is not preferred to use only numerical codes such as the ICB codes of the industries, these variable names are introduced in the analysis.

In regression equation 3, the industry dummy variables are added to the original regression equation 1, which results in the following:

Rs = a0 + a1*Dt/At + a2*dDt/At + a3*dDt+1/At + a4*Et/At + a5*dEt/At

+ a6*dEt+1/At + a7*It/At + a8*dIt/At + a09*dIt+1/At + a10*dAt/At +

a11*At+1/At + a12*O1+ a13*BM + a14*ID + a15*CG + a16*H1+

a17*CS + a18*T1 + e

(3)

In regression equation 4, the industry dummy variables are added to the original regression equation 2, which results in the following:

Rs = a0 + a1*Dt/At + a2*d(Dt/At) + a3*d(Dt+1/At+1) + a4*Et/At +

a5*dEt/At + a6*dEt+1/At + a7*It/At + a8*d(It/At) + a09*d(It+1/At+1) +

a10*dAt/At + a11*At+1/At + a12*O1+ a13*BM + a14*ID + a15*CG +

5. Results

This section presents the results of the analyses. Subsection 5.1 provides some summary statistics for the data, 5.2 presents the results of the comparative analysis. Finally, subsection 5.3 describes the results of the regression analysis.

5.1 Descriptive statistics

Table 4. Mean variable table per industry

In this table the mean values of each variable per industry are given. D represents dividends, A is assets, E is earnings and I is interest costs. Both the dividend level and the interest costs level as the dividend policy and leverage policy variables are included in the table. The construction of the variables is explained in table 2.

Variable

Industry Dt/At (d)Dt/At (d)Dt+1/At Et/At (d)Et/At (d)Et+1/At It/At (d)It/At

(1)Oil & Gas 0.010 0.002 0.001 0.032 -0.003 0.003 0.013 0.005

(1000)Basic Materials 0.032 0.003 0.001 0.064 0.006 0.000 0.014 0.000 (2000)Industrials 0.027 0.007 -0.006 0.061 0.018 -0.017 0.014 0.001 (3000)Consumer Goods 0.024 -0.001 0.000 0.046 -0.003 -0.019 0.016 0.001 (4000)Health Care 0.009 0.003 -0.001 -0.092 0.005 -0.044 0.013 0.001 (5000)Consumer Services 0.021 0.004 -0.006 0.043 0.008 -0.046 0.017 0.001 (6000)Telecommunications 0.024 -0.005 0.001 0.062 0.014 -0.198 0.023 0.005 (9000)Technology 0.020 0.002 0.000 0.046 0.002 -0.036 0.010 0.001 Average 8 categories 0.021 0.002 -0.001 0.033 0.006 -0.045 0.015 0.002 Variable

Industry (d)It+1/At (d)At/At (d)At+1/At Return d(Dt/At) d(Dt+1/At+1) d(It/At) d(It+1/At+1)

(1)Oil & Gas 0.005 0.288 0.103 0.277 0.000 -0.001 0.001 0.004

Table 5. Median variable table per industry

In this table the median values of each variable per industry are given. D represents dividends, A is assets, E is earnings and I is interest costs. Both the dividend level and the interest costs level as the dividend policy and leverage policy variables are included in the table.The construction of the variables is explained in table 2

Variable

Industry Dt/At (d)Dt/At (d)Dt+1/At Et/At (d)Et/At (d)Et+1/At It/At (d)It/At

(1)Oil & Gas 0.000 0.000 0.000 0.049 0.008 0.003 0.012 0.004

(1000)Basic Materials 0.023 0.002 0.000 0.046 0.002 -0.007 0.013 0.000 (2000)Industrials 0.016 0.001 0.000 0.054 0.013 0.000 0.013 0.001 (3000)Consumer Goods 0.016 0.001 0.000 0.051 0.003 -0.005 0.014 0.000 (4000)Health Care 0.000 0.000 0.000 0.025 0.001 0.007 0.010 0.000 (5000)Consumer Services 0.007 0.000 0.000 0.037 0.002 -0.018 0.017 0.001 (6000)Telecommunications 0.030 0.000 0.000 0.064 0.019 -0.058 0.024 0.002 (9000)Technology 0.000 0.000 0.000 0.057 0.009 -0.005 0.005 0.000 Average 8 categories 0.012 0.001 0.000 0.048 0.007 -0.010 0.013 0.001 Variable

Industry (d)It+1/At (d)At/At (d)At+1/At Return d(Dt/At) d(Dt+1/At+1) d(It/At) d(It+1/At+1)

(1)Oil & Gas 0.003 0.310 0.092 0.117 0.000 0.000 0.000 0.000

The average return on the stock per industry is visualized in figure 4. These stock returns relate to the sample period from January 1st 2007 untill January 1st 2008. As can be seen in the graph, only the healthcare industry experienced on average a negative stock return in 2007. This value decrease was severe, as the return on the stock amounted up to a loss of 18,1%. The oil and gas industry was the biggest winner in 2007, having a positive return on average of 27,7%.

Figure 4. Average stock return per industry

-30.00% -20.00% -10.00% 0.00% 10.00% 20.00% 30.00% 40.00% 1 Industries R e tu rn ( % )

Oil & Gas Basic Materials Industrials Consumer Goods Health Care Consumer Services Telecommunications Technology

of the companies in the non paying group had a negative return on their stock during the year 2007.

Table 6. Summary statistics dividend payers and dividend non payers

In this table, some summary statistics are presentented of the group of companies that did not pay out any dividends in year t and of the group of companies that did pay out dividends in year t. Year t represents our sample year 2007.

Dt/At represents the dividend/assets ratio in year t, for example dDt/At represents the change in dividends and d(Dt/At) represents the change in the D/A ratio of year t. The construction of all variables can be read in table 2.

Number is the number of non payers or dividends payers in year t and Return is the average return of the stock of the non payer or payers group.

non payers in year t payers in year t Dt/At =0 Dt/At > 0

Variable Number Mean Median Number Mean Median Dt/At 123 0.000 0.000 225 0.035 0.023 dDt/At 123 -0.002 0.000 225 0.003 0.001 dDt+1/At 123 0.003 0.000 225 -0.001 0.000 d(Dt/At) 123 -0.002 0.000 225 0.002 0.001 d(Dt+1/At+1) 123 0.004 0.000 225 -0.002 -0.001 Return 123 0.019 -0.069 225 0.108 0.055

5.2 Results comparative analysis

Table 7 provides the results of the comparative analysis and it presents descriptive statistics for the whole sample group. The dividends paid out on average by all sample firms add up to 2,2% of the book value of their assets, while the average earnings are 4% of the book value of assets. The average stock return of the whole group is 7,6%. The average earnings, dividends and dividend payout ratio increased in the year 2007 and decreased in the year 2008. The high return group experienced a stock return of 37,1% on average and the low return group experienced a stock return of -21,9% on average.

Table 7. Results comparative analysis

In this table, firstly the mean and median values of all variables for the whole group are provided. Then the mean and median value of the high return group and of the low return group are given. The mean values are compared using the Welch mean test and the median values are compared using the Wilcoxon Signed rank test.

The significance is given by stars, where * represent significance at a 10% level, ** represents signifance at a 5% level and *** represents significance at a 1% level. The dividend amounts variables are given at the top and the extra dividend ratio variables are given below. The variables are all defined and clarified in table 2.

dividend amounts Whole Whole High return High return Low return Low return Mean Median

group group group group group group Welch test Wilcoxon

Variable Mean Median Mean Median Mean Median prob prob

As expected, the increase in earnings in year t and t+1 is significantly larger in the high return group than in the low return group. Also the high return group experienced significantly higher current Earnings/Assets ratios in comparison with its less successful peers. The interest variables generally not tend to be largely different between the two groups, only the increase in the interest costs/assets ratio in t+1 is significantly larger for the low return group. Furthermore, the result is found that the high return group had a significantly larger growth of the book value of assets than the low return group.

5.3 Results OLS analysis

This section presents the results of the regression analysis. However, before these regression results are presented, I consider the correlation figures between the independent variables of the regression equations. These correlation figures are shown in table 8 and 9. Table 8 presents all the correlations between the independent variables that are present in the regression equations 1 and/or 3. These correlations can be stated in one table, since all variables that are present in regression equation 1, are also present in equation 3. The same reasoning goes for the independent variables of regression equations 2 and 4, of which the correlations between the variables are presented in table 9. No correlation figures are presented between the industry dummy variables, because it is not possible for a firm to belong to more than one industry and therefore all companies will have only one score of 1 on an industry dummy variable (or 0 if it is a technology company) and all other dummy variables will have the score of 0. Hence, no correlation is possible between the industry dummy variables.

Table 8. Correlation matrix for regression equations 1 and 3

In this table, the correlations are given between the independent variables that are present in equations 1 and/or 3. There are no correlations given between two different industry dummy variables, since it is not possible for a company to belong to more than one industry category and therefore there will not be correlation between these dummy variables. The explanation of the variables is given in table 2 and table 3.

Variable Dt/At dDt/At dDt+1/At Et/At dEt/At dEt+1/At It/At dIt/At dIt+1/At dAt/At dAt+1/At

Table 9. Correlation matrix for regression equations 2 and/or 4

In this table, the correlations are given between the independent variables that are present in equations 2 and/or 4. There are no correlations given between two different industry dummy variables, since it is not possible for a company to belong to more than one industry category and therefore there will not be correlation between these dummy variables. The explanation of the variables is given in table 2 and table 3.

Variable Dt/At d(Dt/At) d(Dt+1/At+1) Et/At dEt/At dEt+1/At It/At d(It/At) d(It+1/At+1) dAt/At dAt+1/At

it can be concluded that there is no correlation which affects the validity of the regression analysis or which biases the inferences that can be made from the results of the analysis.

Table 10 presents the results of the analysis of the regression equations 1 and 2. The coefficients of the ‘change in level of dividends’ variables of equation 1 are all negative, indicating a negative relationship between an increase in dividends and the return on the stock. However, these coefficients are far from significant. The coefficients of the ‘change in D/A ratio’ variables are positive, indicating a positive relationship between an increase in the D/A ratio and the return of the stock. However, these coefficient are also far from significant. Hence, I find no clear relationship between changes in the level of dividends or D/A ratio and the return on the stock of a company.

Table 10. Results regression equations 1 and 2

In this table, the regression results of equation 1 and 2 are displayed. At the left side, under dividend amounts, the results of regression 1 are showed. At the right side, under dividend ratios, the results of regression 2 are showed. The explanation of all variables can be found in table 2. The significance of the coefficients of the variables is indicated by stars, where * represents significance at a 10% level, ** represents significance at a 5% level and *** represents significance at a 1% level.

Dividend amounts Dividend ratios

Variable Coefficient Prob. Variable Coefficient Prob.

c 0.106 0.0168** c 0.102 0.022** Dt/At -0.814 0.310 Dt/At -0.814 0.303 dDt/At -0.430 0.708 d(Dt/At) 0.415 0.697 dDt+1/At -0.314 0.816 d(Dt+1/At+1) 0.370 0.794 Et/At 0.445 0.0853* Et/At 0.517 0.046** dEt/At 0.693 0.0265** dEt/At 0.745 0.018** dEt+1/At 0.497 0.0287** dEt+1/At 0.526 0.022** It/At -3.302 0.121 It/At -3.243 0.126 dIt/At -2.355 0.516 d(It/At) -0.197 0.951 dIt+1/At -11.247 0.0002*** d(It+1/At+1) -8.937 0.0021*** dAt/At 0.399 0.002*** dAt/At 0.369 0.0029*** dAt+1/At 0.482 0*** dAt+1/At 0.290 0.0033***

Number of obs. 348 Number of obs. 348 R-squared 0.199 R-squared 0.188 Adj. R-squared 0.173 Adj. R-squared 0.161

of next year has a strongly significant negative impact on the stock return of year t, this result is visible in the results of both regresson equations.

Furthermore, I find a clear strong positive impact of the growth of the book value of assets on the stock return of companies. This result is present for both years and for both equations.

The R-squared and adjusted R-squared of equation 1 (0.199 and 0.173) are higher than the R-squared and adjusted R-squared of equation 2 (0.188 and 0.161), indicating that the explanatory power of the ‘change in dividend and interest costs level’ variables is larger than the ‘change in dividend policy and leverage policy’ variables.

Table 11 presents the results of the analysis of the regression equations 3 and 4. With the industry dummy variables added to the analysis, the results do not appear to change much. There is no significant relationship between a changing level of dividends or D/A ratio and the stock return. The ‘change in dividend level’ variables have a negative coefficient and the ‘change in dividend policy’ variables have a positive coefficient, but both are far from being significant. Just as appeared from the results of the analysis of equations 1 and 2, the results of analyzing equations 3 and 4 reveal that an increase in earnings has a strong positive impact on the stock return of the company.

Table 11. Results regression equations 3 and 4

In this table the regression results of equation 3 and 4 are displayed. At the left side, under dividend amounts, the results of regression 3 are showed. At the right side, under dividend ratios, the results of regression 4 are showed. The explanation of all variables can be found in table 2 and 3. The significance of the coefficients of the variables is indicated by stars, where * represents significance at at 10% level, ** represents significance at a 5% level and *** represents significance at a 1% level.

Dividends amounts Dividend ratios

Variable Coefficient Prob. Variable Coefficient Prob.

c 0.076 0.206 c 0.072 0.228 Dt/At -0.916 0.257 Dt/At -0.898 0.259 dDt/At -0.222 0.848 d(Dt/At) 0.558 0.602 dDt+1/At -0.378 0.783 d(Dt+1/At+1) 0.304 0.833 Et/At 0.328 0.232 Et/At 0.394 0.152 dEt/At 0.747 0.019** dEt/At 0.804 0.012** dEt+1/At 0.470 0.043** dEt+1/At 0.505 0.031** It/At -4.067 0.066* It/At -4.024 0.067* dIt/At -2.181 0.550 d(It/At) 0.063 0.984 dIt+1/At -11.066 0.000*** d(It+1/At+1) -8.597 0.003*** dAt/At 0.369 0.006** dAt/At 0.344 0.007*** dAt+1/At 0.471 0*** dAt+1/At 0.284 0.004*** O1 0.168 0.146 O1 0.159 0.169 BM 0.093 0.301 BM 0.095 0.296 ID 0.072 0.254 ID 0.070 0.274 CG 0.083 0.258 CG 0.083 0.262 H1 -0.064 0.502 H1 -0.066 0.493 CS 0.007 0.923 CS 0.011 0.888 T1 0.258 0.275 T1 0.270 0.257

6. Discussion

The results of this paper reveal that there is no clear relationship between the dividend amount or dividends/assets ratio and the stock return of the firm in the crisis year 2007. This result is found in both the comparative analysis and the regression analysis and is robust for both including other determinants of stock return and industry dummy variables and for not including these.

The results of this study are compared with the results of the study of F&F (1998) in table 12. The results of this study can be best compared with the results from the study by F&F (1998) because the non industry independent variables are almost identical in both papers.

Table 12. Comparison of research results

In this table, the results of this study are compared with the results of the study by Fama and French (1998). Neg indicates that the coefficient of the regression variable in the study has a negative sign and pos indicates that the coefficient of the regression variable has a positive sign. * means that the coefficient is significant at a 10% level, ** means that the coefficient is significant at a 5% level and *** means signifance at a 1% level. The explanations of the variables can be found in table 2.

Paper Our results Fama and French (1998)

Variable

Dt/At neg pos***

dDt/At neg pos***

dDt+1/At neg pos***

d(Dt/At) pos pos***

d(Dt+1/At+1) pos pos***

Et/At pos pos***

dEt/At pos** pos***

dEt+1/At pos** pos***

It/At neg* neg*

dIt/At neg neg***

dIt+1/At neg*** neg***

d(It/At) pos pos

d(It+1/At+1) neg*** neg**

dAt/At pos*** pos***

dAt+1/At pos*** pos***

7. Conclusion & limitations

7.1 Conclusion

This paper considers the impact of dividends on the stock return that a firm experiences during a crisis period. To perform the analysis, nearly the same model is used as the one in Fama and French (1998) and Pinkowitz and Williamson (2002). With a good control for profitability, leverage, growth and industry, the results show that there is no clear relationship between the dividends of a firm and its stock return in this crisis period. Both dividends amounts and dividend/asset payout ratios are included in the analysis.

The finding in this paper is different than the finding in the papers of F&F (1998) and P&W (2002). They find that, using a large time frame, there is a positive relationship between the dividends paid out by a firm and the value of the firm. As an explanation of their finding they argue that dividends signal information about the future profitability of the firm. From this research it appears that in times of financial crisis with high interest costs, this signalling theory is not (dominantly) visible.

7.2 Limitations of the paper and recommendations future research

There are several limitations concerning this paper. A first limitation is that the results of this paper can be explained using other theoretical reasoning. It appears from the results that on balance there is not a clear relationship between the dividends and the stock return of a firm in the crisis period. In this paper it is inferred that the positive signaling effect of dividends on stock return, which is seen in the results of comparable papers with a long sample period, is counterbalanced by the benefits of retaining more earnings in the company in the crisis period. However, it could also be possible that the positive effect of dividend signalling is counterbalanced by the negative effects of dividend taxation on the stock price, as is put forward by Brennan (1970). Since dividends are often more heavily taxed than capital gains, this could put pressure on the stock price of high dividend paying firms (Miller and Scholes, 1978).

Another possible explanation for the results of this paper is that the positive effects of the dividend signaling on the stock return are counterbalanced by the costs of dividend signalling mentioned in the paper by Bhattacharya (1979). Both of these explanations do not seem most likely, since these effects should also be visible with a long sample period and they were not found by F&F (1998) or P&W (2002). Nevertheless, it would be possible that the absence of a clear positive relationship between dividends and earnings is due to another theoretical explanation than the one given in this paper. Therefore, future research should be devoted to repeating this research using another sample to find more support for the ‘crisis theory’ and to perform analyses for both crisis periods and flourishing economical periods to be able to directly compare the results of these analyses. Conducting more future research with other samples is also important because a second limitation of the paper is that only one crisis period is used. To be able to better generalize the finding in this paper it is important that the same result is reached in other papers using other samples.

References:

- Aharony, J. and Swarv, I. 1980. Quarterly Dividend and Earnings Announcements and Stockholders’ Returns: An Empirical Analysis. The journal of Finance, vol. XXXV, no.1

- Alltizer, R.L. and Hamill, J.R. 1999. The effects of tax rates and dividend payouts on common stock returns. American Business Review, vol. 17, issue 2, p24

- Arnott, R.D. and Asness, C.S. 2003. Surprise! Higher Dividends = Higher Earnings Growth. Financial Analysts Journal, vol. 59, issue 1, p70-87

- Baker, H.K., Powell, G.E., Velt, E.T. 2002. Revisiting Managerial Perspectives on Dividend Policy. Journal of Economics and Finance, vol. 26, no. 3

- Beck, T., Demirguc-Kunt, A., Laeven, L and Levine, R. 2008. Finance, Firm Size and Growth. Journal of Money, Credit & Banking (Blackwell), vol. 40, issue 7, p1379-1405 - Benartzi, S., Michaely, R. and Thaler, R. 1997. Do Changes in Dividends Signal the Future or the Past? The Journal of Finance, vol. LII, no. 3.

- Bhattacharya, S. 1979. Imperfect information, dividend policy, and the “bird in the hand fallacy,” Bell Journal of Economics and Management Science, vol. 10, p 259-270

- Black F. and Scholes, M. 1974. The effects of dividend yield and dividend policy on common stock prices and returns. Journal of Financial Economics, vol. 1, issue 1, p1-22 - Brennan, M.J. 1970. Taxes, market valuation and corporate financial policy. National Tax Journal, vol. 23, issue 4, p417-427

- Brook, Y., Charlton, W.T. and Hendershott, R.J. 1998. Do firms use dividends to signal large future cash flow increases? FM: The Journal of the Financial Management

Association, vol. 27, Issue 3, p46-58

- Brooks, C. 2008. Introductory econometrics for finance. Cambridge University Press, 2nd edition.

- Brown, S.J. and Warner, J.B. 1980. Meausuring Security Price Performance. Journal of Financial Economics, vol. 8, p205

- Comrey, A.L. 1985. A Method for Removing Outliers to Improve Factor Analytical Results. Multivariate Behavioral Research, vol. 20. Issue 3, p273

- Fama, E.F. and French, K.R., 1998. Taxes, Financing Decisions, and Firm Value. The Journal of Finance, vol. LIII, no. 3

- Gwilym, O., Seaton, J., Suddason, K. and Thomas, S. 2006. International Evidence on the Payout ratio, Earnings, Dividends, and Returns. Financial Analysts Journal, vol. 62, no. 1

- Keim, D.B. 1985. Dividend Yields and Stock Returns. Journal of Financial Economics, vol 14, p473-489

- Koller, T., Wessels, D. and Goedhart, M.(2005). Valuation: Measuring and Managing the Value of Companies. Wiley, New York

- Lintner, J. 1956. Distributions of Incomes of Corporations among Dividends, Retained Earnings, and Taxes. American Economic Review, vol. 46, p97-113

- McManus, I., Gwilym, O. and Thomas, S. 2004. The Role of Payout Ratio in the Relationship Between Stock Returns and Dividend Yield. Journal of Business Finance & Accounting, vol. 31

- Miller, M. and Modigliani, F. 1961. Dividend Policy, Growth, and the Valuation of Shares. Journal of Business, vol. 34 p411-433

- Miller, M.H. and Rock, K. 1985. Dividend Policy under Assymmetric Information. Journal of Finance, vol. 40, issue 4, p1031-1051

- Miller, M.H. and Scholes, M.S. 1978. Dividends and taxes. Journal of Financial Economics, vol. 6, p333-364

- Miller, M.H. and Scholes, M.S. 1982. Dividends and Taxes: Some Empirical Evidence. Journal of Political Economy, vol. 90, issue 6, p1118-1141

- Myers, S.C. and Majluf, N.S. 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, vol. 13, issue 2, p187-221

- Ogden, J.P. A dividend payment effect in stock returns. 1994. Financial Review, vol. 29, issue 3, p345

- Pal, K. and Goyal, P. 2007. Leading Determinants of Dividend Policy: A Case Study of the Indian Banking Industry. Journal of Risk Finance, vol. 8, issue 4, p349-363

- Pinkowitz, L. and Williamson, R. 2002. What is a dollar worth? The Market Value of Cash Holdings. Working paper. Mc Donough School of Business Washington,

Georgetown University

- Sawicki, J. 2009. Corporate governance and dividend policy in Southeast Asia pre- and post-crisis. European Journal of Finance, vol. 15, Issue 2, p211-230

- Tutticci, I., Krishnan, G. and Percy, M. 2007. The role of External Monitoring in Firm Valuation. Journal of International Accounting Research, vol. 6, issue 2, p83-107