The influence of the credit crisis on ex-dividend stock price behaviour : an event study of the AEX

(1)

1

The influence of the credit crisis on ex-dividend

stock price behaviour. An event study of the AEX.

Economics and Business

Finance and Organization

Student R.J. Newton 6144047

Supervisor: M.A. Dijkstra

(2)

2 Abstract

The three main theories on ex-dividend stock price behaviour are the tax effect theory, the transaction cost theory and the market microstructure theory. Theoretically, the transaction cost theory and the market microstructure theory support the change of Dutch ex-dividend stock price behaviour during the credit crisis.

This study examines the ex-dividend price behaviour of stock listed on the AEX from 2005 to 2012. This period is split up into two time periods; one non-crisis period and one crisis period. The ex-dividend stock price behaviour is observed by analysing the abnormal returns that are computed using the CAPM. The abnormal returns of the ex-dividend day and the k-3 to k+3period are significantly larger than zero. None of the abnormal returns time intervals from the crisis period are significantly different from those of the non-crisis period. This empirically proves that the ex-dividend stock price behaviour did not change due to the credit crisis.

(3)

3 Table of contents

1 Introduction 4

2 Literature review 5

2.1 Stock price behaviour 5

2.2 Ex-dividend stock price behaviour 6

2.3 credit crisis 11 Summary 12 3 Empirical methodology 13 3.1 Abnormal returns 13 3.2 Regression 16 3.3 Data 18 4 Results 19 4.1 Abnormal returns 19

4.2 Impact of the credit crisis 20

5 Conclusion 24

References 26

(4)

4 1 Introduction

Stock dividend is a payment that a corporation makes to its shareholders (Berk and Demarzo, 2011). A corporation’s pay-out policy, concerning its cash flow to equity, is made out of two main options: Retain cash flow or Pay-out cash flow (Berk and Demarzo, 2011). For instance retained cash flow can be invested into new projects or it can be used to raise cash reserves. These options increase assets and when leaving debt unchanged increases equity. Pay-out of cash flow is done either by repurchasing shares or by paying dividend. By repurchasing shares from shareholders a corporation reduces its equity and assets. Cash dividend is a net outflow of cash to the corporation’s shareholders and is paid per share. Another type of dividend is stock dividend, where stockholders receive stock as dividend (Berk and Demarzo, 2011).

The four most important dividend-related dates are the declaration date, ex-dividend date, date of record and the payable date (Berk and Demarzo, 2011).. The Board of Directors decides on how much dividend and when the dividend is paid. The date on which the Board of Directors declares to pay dividend in the near future is called the declaration date. The shareholder who owns the share on the cum-dividend date will receive dividend. The ex-dividend date is the day after the cum-dividend date and it is the first day on which new shareholders, who did not own the stock on the cum-dividend date, do not own the right of receiving dividend on their stock. The payable date which is the date the dividend is paid to the shareholders.

The European stock market was affected by the credit crisis (Melvin and Taylor, 2009). The volatility of stock increased causing investor behaviour to change. The main question of this thesis is if the credit crisis affected the Dutch ex-dividend stock price behaviour.

The ex-dividend stock price behaviour is based on stocks listed on the Dutch AEX. There is no study on ex-dividend stock price behaviour of the Dutch stock market. An event study on the recent credit crisis further distinguishes this thesis from existing literature on ex-dividend stock price

(5)

5 The structure of this thesis is as follows. In paragraph 2 the related literature is discussed. In paragraph 2.1 stock price behaviour is described based on market efficiency. Paragraph 2.2 covers existing literature on dividend stock price behaviour. Paragraph 2.3 covers the possible change in ex-dividend stock price behaviour due to the credit crisis. Paragraph 3 presents the empirical

methodology. Paragraph 3.1 describes the methods that are used to compute the abnormal returns. Paragraph 3.2 handles the regression followed by paragraph 3.3 with a description of the data. Paragraph 4 presents the results. Paragraph 5 contains the conclusions.

2 Literature review 2.1 Stock price behaviour

Ex-dividend stock price behaviour is described by the efficiency of the capital markets. When a market is efficient all stock prices fully reflect all information (Fama, 1970; Berk and Demarzo, 2011). Therefore, stock prices are priced accordingly given all the available information. There are three different types of efficiency with different interpretations of what all information is: Strong form efficiency, semi-strong efficiency and weak form efficiency.

Strong form efficiency states that the price reflects all information, public and private information (Fama, 1970). Private information is not related to ex-dividend stock price behaviour.

Semi-strong form efficiency is the efficiency of which prices adjust to publicly available information (Fama, 1970). When information is made public, prices react promptly. For example, on ex-dividend day stock prices should drop immediately with the dividend amount (Miller and

Modigliani, 1961). When the market is not efficient the stock price takes longer to adjust.

Weak form efficiency states that all historical information is reflected in the current stock price (Fama, 1970). It implies that stock prices have a random movement. Future stock price fluctuations cannot be based on historical data because the current stock price already reflects all historical data.

(6)

6 2.2 Ex-dividend stock price behaviour

Miller and Modigliani (1961) state that in perfect capital markets, dividend policy does not affect stock prices. When capital markets are perfect, all corporations Net Present Value’s (NPV) are equal to zero. This means that all current and future profits are equal to zero due to competing corporations that drive out profit possibilities (Berk and Demarzo, 2011). Transaction costs and taxes do not exist in perfect capital markets (Miller, Modigliani, 1961). The other main assumptions Miller and Modigliani (1961) give are rational behaviour of investors and perfect certainty for investors. Investors display rational behaviour when they always choose for higher wealth over lower wealth (Miller, Modigliani, 1961). In an efficient market they are indifferent between receiving dividend or capital gains. Perfect certainty means there is no risk on future returns (Miller, Modigliani, 1961). The dividend policy does not affect the value of a corporation because the liability equity is reduced by the total dividend amount (Miller, Modigliani, 1961). This implies that the ex-dividend stock price drop should be equal to the dividend amount. The ex-dividend stock price drop is the reduction of equity compensating for the outflow of dividend. Though in practice, the ex-dividend price drop is largely not equal to the dividend amount.

Campbell and Barenek (1955) were one of the first to conduct research on stock price behaviour on ex-dividend date. With data from the New York Stock Exchange, dating from October 1949 to April 1950 plus the last quarter of 1950, they found an ex-dividend stock price drop of roughly 90 percent of the dividend amount. This is against the irrelevance theory of Miller and Modigliani (1961).

The three main theories that help understand ex-dividend stock price drops that differ from dividends are the tax effect theory, transaction costs theory and the market microstructure theory (Isaksson, 2013).

The tax effect theory states that the relationship between the ex-dividend price drop and dividend is influenced by a tax differential between the income taxes and the capital gain tax (Elton and Gruber, 1970). Investors prefer capital gains when the taxes on dividend are higher than the tax on capital gains unless the ex-dividend price drop is less than the dividend, because in this situation the

(7)

7 shareholder return is higher when there is no dividend. An ex-dividend price drop lower than the dividend can generate a shareholder the same net return as when he does not receive dividend. Selling cum-dividend means the investor buys the share before cum-dividend for price (

P

o), sells the share for the cum-dividend price (

P

c) and pays capital gain taxes (

t

t

d

= t

g). The model states that the higher positive tax differential between dividend tax and capital gain tax, the lower the ex-dividend stock price drop relative to dividend will be. A higher dividend tax causes the ex-dividend price drop (

P

c

– P

e) to be lower than the dividend (D).

Barclay (1987) states that investors discount the value of dividend when the dividend tax is higher than the capital gain tax. He compared ex-dividend stock price drops on the NYSE during a period without tax differentials, 1900 to 1910, against a period with tax differentials, 1962 to 1985. The two main findings were: The ex-dividend stock price drop during the period 1900 to 1910 was equal to the dividend amount and the ex-dividend stock price drop during the period 1962 to 1985 was significantly lower than the dividend amount. The two main conclusions Barklay (1987) drew from these findings were: When there are no income or dividend taxes, investors are indifferent between receiving dividend or capital gains. A higher dividend tax relative to capital gain tax causes investors to discount dividend to capital gains.

(8)

8 Kalay (1982) states transaction costs are the main cause for ex-dividend stock price behaviour of short-term traders who have no tax differential. Transaction costs are the costs that investors pay for buying and selling stock. Kalay (1982) used execution costs as a measure for transaction costs. The data ranged from 1 April 1966 to 31 March 1976. At the time there was no tax differential between dividend taxes and capital gain taxes in the US. To earn a positive return investors trade short-term by buying stock cum-dividend and selling ex-dividend. The short-term trading returns are reduced by transaction costs. These lower returns reduce the ex-dividend stock price drop because short-term trading gives a negative return when the ex-dividend price drop is higher than the dividend amount minus transaction costs. Therefore Kalay (1982) states the reason for the limited stock price drop is due to transaction costs. The ex-dividend price drop should be equal to the costs of buying plus selling the stock.

Elton, Gruber and Rentzler (1984) criticized the data Kalay (1982) used for transaction costs, because it does not include all trading costs. Transaction costs are a part of the actual trading costs. Elton, Gruber and Rentzler (1984) state that short-term trading has no effect on ex-dividend stock price behaviour because Kalay (1982) failed to include transfers taxes, registration fees, clearance costs and the bid-ask spread. These costs are all made when buying and selling stock. Kalay (1982) used minimum execution costs as a measure for transaction costs. Actual transaction costs are therefore higher than execution costs. The transaction costs theory states that the higher actual transaction costs would lead to a lower ex-dividend stock price drop than is actually the case.

Lakonishok and Vermaelen (1986) examine the trading volume of stock around the ex-dividend date. Trading volume is higher than average before and after ex-ex-dividend day. To test for abnormally high trading volumes around ex-dividend date, data is taken from the NYSE and the AMX from the period 1970 to 1981. During the ten day time frame surrounding ex-dividend days trading volume is higher. The traded volume of high dividend and high liquidity stock increased 65 percent on the three days before ex-dividend days. A higher volume supports Kalay’s (1982) and Elton and Gruber (1970) statements on the influence of taxes and transactions costs on ex-dividend stock price behaviour (Lakonishok and Vermaelen, 1986).

(9)

9 The third theory concerning ex-dividend stock price behaviour is the market microstructure theory. The market microstructure theory relates ex-dividend stock price behaviour with market

characteristics. The market microstructure theory includes three elements that relate to stock prices. -Price formation and price discovery: static and dynamic issues like trading cost factors and the relation between certain information and stock prices. This element focuses on the relation between demand and prices & volumes.

-Market structure and design: The relation between trading protocols and stock price formation. This element is about the relation of market specific rules on stock price behaviour.

-Information and disclosure: The transparency of a market. This element focuses on the information that is available to investors (Madhaven, 2000)..

These elements differ between markets. Frank and Jagannathan (1998) conducted research on the market microstructure of Hong Kong. Data are from 1980 to 1993. At the time there were no taxes on income or capital gains. This implies there was no tax effect. The ex-dividend stock price drop was approximately 50 percent of the dividend amount. Frank and Jagannathan (1998) state this moderate ex-dividend stock price drop is due to the Hong Kong bid-ask spread. Investors that long stock pay the bid price. Investors that short stock receive the ask price. Rational sellers short stock cum-dividend and rational buyers long stock ex-dividend, causing trades to occur at the higher bid price cum-dividend and at the lower ask price cum-dividend. This market microstructure effect increases ex-dividend stock prices with the bid-ask spread.

Akhmedov and Jacob (2010) looked at the Danish ex-dividend stock price behaviour before the credit crisis. Their data are from the Danish stock exchange and ranges from 1995 to 2005. The three main theories are estimated: tax effect theory, transaction cost theory and the market microstructure theory. The tax effect is calculated using Danish taxes. The transaction cost theory is not calculated but theoretically the ex-dividend price drop should lie in the estimated tax-effect window. The market microstructure is calculated using the Danish thick sizes. A thick size is the minimum change of a stock. The estimated ex-dividend day price drop ranges from 57% to 126% of the dividend amount. Two time intervals are used to test the theories: the cum-dividend closing price to the ex-dividend

(10)

10 closing price change and the cum-dividend close price to the ex-dividend opening price change. Opening prices are the prices of stock when the exchange is opened. Closing prices are the prices of stock when the exchange is closed. In the USA both these measures are 70% to 80% the dividend amount. The Danish cum open-to- ex open price drop is 32% and the cum open-to- ex close price drop is 18%. Akhmedov and Jacob (2010) note the smaller Danish ex-dividend stock price drop is due to the absence of a limit order adjustment mechanism together with low liquidity in Denmark. The limit order adjustment mechanism is a market microstructure that automatically reduces orders with the dividend amount (Dubofsky’s, 1992). The limit order buy value is reduced by the dividend amount before the start of the ex-dividend day. The limit order adjustment mechanism adjusts orders for the reduced value of stock because of stock-dividend. The trading volume is low when there is low liquidity, causing stock prices to react very slowly to stock going ex-dividend. Without a limit order mechanism and with low liquidity stock prices react less to stock going ex-dividend.

Isaksson (2013) conducted research on the ex-dividend stock price behaviour of blue-chip stock. Stock of a blue-chip corporations performs well, has low risk and is stable is times of crisis in comparison to other stock listed on the same exchange. Dutch examples of blue chip stock are Unilever and royal Dutch Shell (Hupperets and Menkveld, 2002). The data of Isaksson (2013) research cover the years 2005 to 2009 and are retrieved from the New York Stock exchange, Tokyo Stock exchange, Shanghai stock exchange and the London stock exchange. The raw price ratio, market-adjusted price ratio, raw price drop, market-adjusted price drop and the market-adjusted abnormal returns are calculated using the sample. Isaksson (2013) concludes that the New York stock exchange and Shanghai stock exchange showed no significant difference between the ex-dividend price drop and dividend. On the Tokyo stock exchange the ex-dividend price drop is significantly smaller than the dividend amount. As a result there are positive abnormal returns with a stock buying investment strategy (Isaksson, 2013). The ex-dividend price drop of the London stock Exchange was significantly higher than the dividend amount, resulting in abnormal returns. Although Isaksson (2013) does not study the cause, he states that the negative abnormal returns are a result of falling prices of London stock during the credit

(11)

11 crisis in 2008 and 2009 (Isaksson, 2013). The ex-dividend stock price drop is higher than the dividend amount due to dividend and the crisis.

2.3 The credit crisis

In 2007 a credit crisis erupted in the US that was mainly due to the US mortgage market. (Stiglitz, 2010). House prices dropped causing financial problems for homeowners with high mortgages. Especially sub-prime loans caused many problems (Stiglitz, 2010). A sub-prime loan is a loan that has low ascription standards. Sub-prime loans bare more risk which is why the rent on sub-prime loans is higher for then the rent on prime loans. The sub-prime crisis started a credit crisis in the US (Stiglitz, 2010). This credit crisis spilled over to Europe. The event that coincides with this spillover is the bankruptcy of Lehman Brothers on September 15, 2008 (Stiglitz, 2010). Lehman Brothers was an American bank that was not helped out of bankruptcy by the Federal Reserve and the U.S. treasury (Melvin and Taylor, 2009). It was considered not to be ‘too big to fail’. The bankruptcy of the Lehman brothers further increased credit uncertainty. Financial institution became more reticent, causing the credit supply to drop (Stiglitz, 2010).

The tax effect theory states that a change in the tax differential will result in a change in ex-dividend stock price behaviour (Elton and Gruber, 1970). The dividend tax changed from the beginning of 2007 (Staatsblad, 2006). The dividend pre-tax was reduced from 25 percent to 15 percent (Wet DB, 1965). This is a pre-tax paid by the corporation that pays out the dividend. Dividend from ownership of over 5 percent is taxed with 25 percent (Wet IB 2001). The remaining 10 percent is taxed after the dividend pre-tax (Wet DB 1965). If dividends are received from stock that is taxed in box 3: Savings and capital, the full 15 percent can be reclaimed by the receiver of the dividend (Wet DB 1965). Despite the change in structure of the dividend tax, the received dividends of shareholders almost remain the same (Wet DB 1965). Without looking at the effect of receiving the reclaimed dividend on a later date, there are no changes in taxes that matters to change the ex-dividend stock price behaviour from 2007 forward in the Netherlands.

(12)

12 The transaction cost theory states that changes in transaction cost will change ex-dividend stock price behaviour (Kalay, 1982). Elton, Gruber and Rentzler (1984) argue that the bid-ask spread should also be included because investors buy at the ask price and sell at the bid price.

The market microstructure theory predicts the increased bid ask spread reduces the

ex-dividend stock price drop (Frank and Jagannathan, 1998). Melvin and Taylor (2009) find an increased bid-ask spread during the credit crisis. Because financial institutions face greater risk because of high stock volatility, the bid-ask price increases to cover the greater risk. This reduces the ex-dividend stock price drop. Therefore the transaction cost theory and the market microstructure theory give leave room for a possible change of ex-dividend stock price behaviour during the credit crisis.

Summary: In an efficient market stock prices fully reflect all information. Miller and Modigliani (1961) state in a

efficient market, the ex-dividend stock price drop should be equal to the dividend amount. Campbell and Barenek (1955) provide empirical proof that the market is not fully efficient and stock price adjustments are lower than the dividend amount. The three main theories that explain ex-dividend stock price drops differ from dividends are the tax effect, transaction costs effect and the market microstructure. Empirically, taxes have an effect on ex-dividend stock price behaviour when there is a tax differential between the dividend tax and the capital gain tax (Elton and Gruber,1970). Transaction costs influence short-term trading around ex-dividend days (Kalay, 1982). The market microstructure studies the market characteristics that influence ex-dividend stock price behaviour. (Frank and Jagannathan, 1998). Akhmedov and Jacob (2010) found very high ex-dividend abnormal returns on the Danish stock market that are attributed by the Danish lack of a limit order adjustment mechanism. Isaksson (2013) found negative ex-dividend day abnormal returns in London during the credit crisis. This is attributed by the fall of the London stock market. The tax effect theory does not predict a change in ex-dividend stock price behaviour caused by the credit crisis in The Netherlands. The transaction cost theory and the market microstructure account for changes in the ex-dividend stock price behaviour because of changes in transaction costs and the bid-ask spread caused by the credit crisis.

(13)

13 3 Empirical Method

mt

– R

ft

) +

it

OLS conditions that must hold are: The expected error is zero and the variance of the error is fixed.

E(

it

) = 0 , VAR(

it

) = σ²

i

(Berk and Demarzo, 2011)

Rit =Stock return. Rft = Risk free rate. Rmt = Market return. it = the error term for day t. βi = Systematic risk and αi = Constant term. i=Stock. t=Time

(14)

14 Ex-dividend day stock returns are based on the cum-dividend day closing price relative to the ex-dividend closing stock price and ex-dividend.

Returns on all days other than ex-dividend days are based on the closing stock price relative to the closing stock price of the day before.

The yearly risk free rate is converted to a daily risk free rate using the amount of trading days.

R

fk

= (1 + R

fy

)

1/250

- 1

Ri0=Ex-dividend day return. Pio= Ex-dividend day closing price. Di= Dividend. Pic=Cum-dividend closing price. Ri(k) = Day k≠0 return. Pi(k)= Day k≠0 closing price. Pi(k)-1= Day before Pi(k) closing price. Rfk= Risk free rate. Rfy= Yearly return risk free date.

The CAMP is estimated in the period ranging van Januari 1st 2000 to December 31st 2004. This time interval is before the event windows (Mackinlay, 1997). The CAPM parameters from outside the model are used to compute the expected returns in the event window. The event windows are

estimated in a sample ranging from Januari 1st 2005 to December 31st 2012. The ex-dividend days lie in the middle of the event window and are called K=0. K specifies days in the event in respect to the ex-dividend day. Cum-dividend days are called k=-1 , the day before that k=-2 and so on. The event window starts five days before and ends five days after the ex-dividend date and is called k=-5, K=+5. Eades, Hess, and Kim (1984) used the same eleven day event window. They found an average

abnormal return of 14 percent on ex-dividend date and an average abnormal return of 33 per cent in their k-5 to k+5 event window. They state some short term traders buy days before the ex-dividend day and some sell days after the ex-dividend day. This implies the importance of an eleven day event window because a shorter event window may cause relevant data to be omitted, because the total impact of dividend on the stock price is not included.

(15)

15 The expected returns in the event window are computed with the estimated alphas and beta. The Amsterdam Exchange Index (AEX) is used as a proxy for the market return.

E(R

ik

) = α

i

+ β

i

H1: AR≠0

The abnormal returns are equal to zero. When abnormal returns are not significantly zero the ex-dividend stock price drop is not equal to the ex-dividend amount.

β

8

VOL

ik

+

i

ARik=Abnormal return. αi=Constant term. CRISISik= Dummy variable for crisis period.

DYik=Dividend yield. TCik=Transaction Costs. MEik=Market Equity. BMik= Book-to-market

.

LEVik= Leverage. AVVOlik= Average trading volume. AVik= Trading volume.

The dummy variable (CRISIS) controls for two time periods. The variable CRISIS is zero for events from 2005 until the third quarter of 2008. The variable CRISIS is one for events from fourth quarter of 2008 until the end of 2012.

Dividend yield (DY) for every ex-dividend day is the second variable. Elton and Gruber (1970) state a higher dividend yield reduces the ex-dividend stock price drop. A lower price drop means higher abnormal returns. Lakonishok and Vermaelen (1986) state that a higher dividend yield increases short term trading, which signals the existence of abnormal returns. Elton, Gruber and Rentzler (1983) state that dividend yield is related to abnormal returns that are computed with the CAPM. Dividend yield is calculated by dividing dividend by the cum-dividend closing price (D/Pc).

Transaction costs (TC) influence short term trading behaviour (Kalay, 1982) (Lakonishok and Vermaelen 1986). Short trading can be viable when there are abnormal returns. The proxy for transaction costs is the inverse of the cum-dividend stock price (1/Pc) (Karpoff and Walkling, 1988). When transaction costs are constant stock size provide a relative measure for transaction costs. Fama and French (1992) state that Market Value (MV), price-to book and Leverage (LEV) are positively related with average stock returns. MV is the total value of a corporation’s outstanding stock in millions of euro. The Price-to-book (PB) ratio is the stock price divided by the book value of stock. It gives a measure for how a stock is priced relative to a stock’s actual price. Leverage (LEV) is the level of total debt relative to total equity. It reflects the financial composition of a corporation. Total debt divided by total assets is used as a measure for leverage.

(17)

17 The daily average volume for a month (AVVOL) is a proxy for liquidity. It is the average volume of stock traded per day. Liquidity is positively related with short- term trading behaviour and abnormal returns because more stock trading induces more liquidity (Kato and Loewenstein, 1995).

Transaction costs influence short-term trading behaviour (Lakonishok and Vermaelen 1986). Short-term trading behaviour is observed by the abnormal trading volume. The abnormal trading volume (VOL) is the daily average trading volume of the event window minus the daily average trading volume of the ex-dividend month (AVVOL).

To answer the question if the credit crisis changed ex-dividend stock price behavior the next hypothesis is tested:

The effect of the credit crisis on ex-dividend stock price behavior is computed with a regression. The abnormal returns in the period 2005 to the third quarter of 2008(Dummy CRISIS=0) are the same as the abnormal returns in the period from the fourth quarter of 2008 to 2012(Dumy CRISIS=1). H0:

β

1=0

H0:

β

ik

+

β

8

VOL

ik

+ β

9

Unilever

ik

+

β

10

Heineken + β

11

ING ………

β

28

Imtech +

i

The added variables Unilever to Imtechare dummy variables are one for abnormal returns that belong to that particular corporation.The only corporation for which there is no dummy variable is Shell, because the results of other companies will be relative to the results of Shell. There are 20 corporation dummy’s (Appendix 2)

(18)

18 3.3 Data

All the data are retreived from Thomson Datastream. The data are imported in to Excel and Stata 12. The dividend-paying stocks are companies listed on the AEX. The AEX is an index of the largest 25 stocks, based on market cap, traded on the Euronext Amsterdam. Aperam, D.E. Master Blenders, TNT and Post NL are deleted from the sample, beacause they where not listed on the Euronext Amsterdam in 2000 to 2004. 21 Corporations remain (appendix 2). The sample contains 256 events. The non-crisis period counts 127 events and the crisis period contains 129 events. Ex-dividend dates and dividend amounts taken from Datastream have been randomly checked using annual reports. Also missing data on ex-dividend dates have been added using annual reports. Non-trading days are deleted. This prevents the zero returns of non-trading days to cause bias.

The correlations between the explanatory varibales of the regression without the corporation dummy’s are presented in table 1.

Table 1:

Crisis DY TC MV PB LEV AVVOL VOL(EX)

Crisis 1 DY 0,12 1 TC 0,054 0,38 1 MV -0,045 -0,39 -0,16 1 PB -0,015 0,044 0,39 -0,17 1 LEV 0,091 0,40 0,22 -0,34 0,0005 1 AVVOL -0,14 -0,049 0,52 0,57 0,069 -0,013 1 EX-VOL 0,11 0,19 0,27 -0,031 0,11 0,15 0,16 1 Correlations between explanatory variables.

The highest correlations of 0.52 and 0,57 are between the average volume and both transaction costs and market value. The values are below the 0,70 point on which a correlation is high.

(19)

19 4 Results

4.1 Abnormal returns

The CAPM model gives the measures for the stock systematic risk (Appendix2). The betas range from 0,1095 for Corio to 1,763 for ASML. The average of all the betas is 0,71 . Corio’s low beta states is it not effected much by market movements. While the high beta of ASML states that is it is very responsive to market movements.

In the period 2005 to 2012, the average dividend yield is 2,14 percent and the median is 1,7 percent (table 2). The highest dividend yield paid is 10,82 percent. The ex-dividend day abnormal returns have an average of 0,40 percent and a median of 0,17 percent (appendix3). The k-1 to k+1 CAR have a lower average of 0,16 percent. The k-1 to k+1 CAR has a higher median of 0,27 percent. The drop in average abnormal retuns is caused by negative abnormal returns on cum-dividend day and the day after dividend day. Stock prices do not react promtly on dividend day, therefore the day after ex-dividend day has a lower abnormal return. The k-1 to k+1 CAR median is higher than the median of ex-dividend day abnormal returns so large negative outliers could have caused the lower average of k-1 to k+k-1 CAR. The k-3 to k+3 cumulative abnormal returns average is 0,57 percent and the median is 0,68 percent. The k-5 to k+5 cumulative returns average is 0,54 percent and the median of 0,67 percent which are lower than the values for the k-3 to k+3 event window. Thefore the abnormal returns of the k-3 to k+3 event window are expected to be higher than those of the k-5 to k+5 event window.

Table 2

Variables Obs. Mean Median Std. Dev. Min Max

DY 256 .021 .017 .016 .0033 .11

EX-AR 256 .0040 .0017 .020 -.057 .086

CAR 1 256 .0016 .0028 .033 -.12 .12

CAR 3 256 .0057 .0070 .046 -.15 .18

CAR 5 256 .0054 .0067 .058 -.19 .18

The P-value for dividend abnormal returns score is 0,0013 (table 3). This states that the

ex-dividend abnormal return is significantly larger than zero with a confidence level of 99 percent. The P-value for the k-1 to k+1 event window is 0,4472 and is therefore not significantly different from zero

(20)

20 with a confidence level of 90 percent or higher. The P-value for the k-3 to k+3 event window is 0,0518 and is therefore significantly different form zero with a confidence level of 90 percent. This suggest that in the days k-3, k-2, k+2 and k+3 stock prices rise. The P-value for the k-5 to k+5 event window is 0,1382 and is therefore not significantly different from 0 with a confidence level of 90 percent or higher.

Table 3:

Variables Obs Mean Std. Error P>|t| Std Dev 95% Conf. interval EX-AR 256 .0040 .0012309 .0013*** .0196936 .0015844 .0064323 CAR 1 256 .0016 .0020748 .4472 .0331967 -.0025063 .0056655 CAR 3 256 .0057 .0028931 .0518* .0462902 -.0000044 .011351 CAR 5 256 .0054 .0036265 .1382 .0580247 -.001749 .0125346 ***=significant at 1 percent. * =significant at 10 percent

The ex-dividend day abnormal returns and the k-3 to k+3 cumulated abnormal returns are significantly higher than zero. This means that in these time intervals the ex-dividend stock price drop is

significantly lower than the dividend amount. It is notable that the k-1 to k+1 abnormal returns are not significantly larger than zero and the k-3 to k+3 abnormal returns are significantly larger than zero. The k-3 to k+3 time intervals include the k-1 to k+1 time intervals. The first possible cause a positive bias in the sample that is not related to stock going ex-dividend. A postive bias can have a larger effect on the k-3 to k+3 and k-5 to k+5 event windows because they contain more days than the k-1 to k+1 event window. The second possible cause is that stock prices drop the dividend amount with a 1 day delay. After the price drop stock prices increase within three days after the stock goes ex-dividend. In addition, stock price could increase during the days before the cum-dividend day. In particulair days k-3 and k-2. Confirming Eades, Hess, and Kim (1984) statements that investors acquire stock before cum-dividend day.

4.2 Impact of the credit crisis

The relation of the credit crisis with the event abnormal returns is tested by regressing abnormal returns on the dummy variable CRISIS. The two dependent variables ex-dividend abnormal return (EX-AR) and k-3 to k+3 abnormal return (CAR3) are tested first. The abnormal returns from

(21)

ex-21 dividend, k-1 to k+1, k-3 to k+3 and k-5 to k+5 joint together is the last dependent variable (CAR-ALL).

The regression on the ex-dividend day abnormal returns, excluding the corporation dummy’s, gives a coefficient for CRISIS of -0,0016 (table 4), meaning abnormal returns are 0,16 percent lower during the crisis. The regression on the ex-dividend abnormal returns does not give a significant measure for the dummy variable CRISIS. This implies the ex-dividend day abnormal returns during the credit crisis are not significantly different from the ex-dividend day abnormal returns during the non-crisis period. Dividend Yield (DY) and Market Value (MV) are significant explanatory variables for ex-dividend abnormal returns. Dividend yield has a significant postive coefficient of 0,27 with a confidence level of 99 percent. This means for every postive percentage change in dividend yield, ex-dividend abnormal return increases 0,27 percent. This is in line with the statements made by Elton and Gruber (1970) on the positive relation between abnormal returns and the dividend yield. Market value has a negative coefficient of -,066which is significant with a 95 percent confidence level. The smaller corporations have higher ex-dividend abnormal returns. This contradicts the statement of Fama and French (1992) who state larger companies have higher abnormal returns.

Table 4: EX-AR EX-AR EX-AR EX-AR EX-AR EX-AR EX-AR

CRISIS -,0005 _(,0025) -,0016 _(,0023) -,00067 _(,0024) -,0016 _(,0023) -,0015 _(,0023) -,0016 _(,0023) -,0016 _(,0027) DY ,27*** _(,092) ,27*** _(,094) ,27** _(,12) ,27** _(,12) ,27** _(,12) MVmln -,066** (,030) -,0044 (,035) -,0056 (,035) -,0032 (,062) P/B ,00012 (,0068) ,00010 (,00068) ,00010 (,00070) TC -,016 (,040) -,017 (,038) -,016 (,058) LEVERAGE ,000024 (,000090) ,000021 (,000089) ,000022 (,000096) EX-VOLmln ,11 (0,43) ,14 (0,39) ,15 (0,40) AVVOLmln -,020 (0,41) Constant ,0043 (,0014) -,00095 (,0023) ,0061*** (,0017) -,00093 (,0023) -,00094 (,0040) -,00073 (,0040) -,00076 (,0041) N 256 256 256 256 256 256 256 R2 ,0002 ,048 ,0088 ,0491 ,0498 ,0502 ,0502

(22)

22 Notes: mln is per million. ***, ** and * reflect significance at 1 percent, 5 percent and 10 percent. All the regressions are robust because there is suspicion of heteroscedasticity and outliers. The numbers under EX-AR without parenthesis are coefficients. The numbers with parenthesis are the standard errors of the coefficients.

The regression on the k-3 to k+3 abnormal returns, excluding the corporation dummy’s, gives a coefficient for CRISIS of -0,0025 (table 5). There are no significant coefficients for the dummy variable CRISIS. Implying the abnormal returns af the k-3 to K+3 during the crisis are not significantly different from those of the non-crisis period. Leverage is a significant explanatory variable with a coefficient of 0,00037, under a 90 percent confidence level. Implying every percentage increase in leverage increases the k-3 to k+3 abnormal returns with 0,037 percent. Confirming the statement of Fama and French (1992) of a postive relation between abnormal returns and leverage.

Table 5: CAR3 CAR3 CAR3 CAR3 CAR3 CAR3 CAR

CRISIS -,0019 (,0058) -,0015 (,0058) -,0034 (,0058) -,0034 (,0086) -,0034 (,0057) -,0029 (,0057) -,0025 (,0066) LEVERAGE ,00037* (,00021) ,00027 (,00022) ,00031 (,00022) ,00029 (,00022) ,00031 (,00022) ,00030 (,00023) DY ,21 (,21) ,26 (,22) ,23 (,22) ,27 (,23) ,27 (,23) MVmln ,093 (,090) ,089 (,089) ,10 (,092) ,073 (,17) P/B ,00097 (,0016) ,0010 (,0016) TC -,044 (,090) -,066 (,14) THREE-VOLmln 1,6 (1,3) 1,8 (1,3) 1,8 (1,3) AVVOLmln ,23 (1,14) Constant ,0066 (,0042) ,0050 (,0061) -,0035 (,0064) -,0079 (,0086) -,0073 (,0085) -,0091 (,011) -,0087 (,011) N 256 256 256 256 256 256 256 R2 ,0004 ,0117 ,0163 ,0189 ,0231 ,0247 ,0249

Notes: mln is per million, ***, ** and * reflect significance at 1 percent, 5 percent and 10 percent, All the regressions are robust because there is suspicion of heteroscedasticity and outliers, The numbers under CAR3 without parenthesis are coefficients, The numbers with parenthesis are the standard errors of the coefficients,

(23)

23 The regression on all abnormal returns joint together, excluding the corporation dummy’s, gives a non-significant coefficient for the dummy variable CRISIS of -0,0021 (Table 6). This implies that the abnormal returns during crisis are not different from the non-crisis abnormal returns. The control variables dividend yield, leverage and the Price-to-Book value have significant coefficient values of (0,30), (0,0012) and (0,0022). Dividend yield is significant with a confidence interval of 99 percent, leverage with a confidence interval of 95 percent and the price to book ratio with a confidence interval of 90 percent. These results affirm the statements of Elton and gruber (1970) and Fama and French (1992) that dividend yield, leverage and the price-to-book ratio are positively related with abnormal returns.

Table 6: CAR-ALL CAR-ALL CAR-ALL CAR-ALL CAR-ALL CAR-ALL CAR-ALL

CRISIS -,0021 (,0026) -,0033 (,0026) -,0026 (,0026) -,0028 (,0025) -,0028 (,0056) -,0028 (,0026) -,0021 (,0028) DY ,30 (,097) ,30*** (,096) ,23** (,099) ,27** (,11) ,26** (,11) ,29*** (,098) P/B ,0012* (,00066) ,0012* (,00065) ,00131* (,00068) ,0013* (,00068) ,0016 (,00078) LEVERAGE ,00019* (,000097) ,00021** (,00010) ,00021** (,000010) ,00022** (,00011) MVmln ,063 (,043) ,062 (,043) ,050 (,081) THREE-VOLmln ,56 (,61) ,96 (,73) AVVOLmln ,11 (,51) TC -,044 (,061) Constant ,0052 (-,0021) -,00066 (,0026) -,0038 (,0034) -,0068575* (,0037) -,010** (,0050) -,010** (,0050) -,01** (,0046) N 1024 1024 1024 1024 1024 1024 1024 R2 ,0006 ,0143 ,0171 ,0201 ,0215 ,0221 ,0227

Notes: mln in millions, ***, ** and * reflect significance at 1 percent, 5 percent and 10 percent, All the regressions are robust because there is suspicion of heteroscedasticity and outliers, The numbers under CAR-ALL without parenthesis are coefficients, The numbers with parenthesis are the standard errors of the coefficients,

(24)

24 The ex-dividend day abnormal return of ASML is the highest of the 21corporations(appendix 3). ASML has a significant coefficient of ,029 with a confidence level of 95 percent. Ahold has the smallest ex-dividend day abnormal return. Ahold has a significant coefficent of -,013 with a 90 percent confidence level. The regression on ex-dividend day abnormal returns, including all variables, gives a extra significant coefficient for ASML of ,030 with a 90 percent confidence level (Appendix 4). The regression on the k-3 to k+3 abnormal returns, including al variables also generates one extra significant coefficient. Reed has a significant coeffient of ,069 with a confidence level of 90 percent (appendix 4). Including the corporation dummy’s in the regressions did not give a significant value for the dummy variable CRISIS (Appendix 4).

There are no abnormal returns measured during the credit crisis that are significantly different from the abnormal returns measured during the non-crisis period. The crisis ex-dividend stock price behavior is the same as the non-crisis ex-dividend stock price behavior. As discussed in paragraph 2,3 the

transaction cost theory and the market microstructure theoratically state that ex-dividend stock price behavior of Dutch stock could be affected by the credit crisis because of changes in investor behaviour concerning transaction costs and changes in the bid-ask spread. In theory the stable ex-dividend stock price behavior could main the effect of the credit crisis is too small or multiple effects counteract each other. For example, the increase in the bid-ask spread reduces the ex-dividend stock price behaviour and lower broker fees increase the ex-dividend stock price drop causing ex-dividend stock price behaviour to remain the same.

5 Conclusion

From 2005 to 2012 the ex-dividend day abnormal returns and k-3 to k+3 abnormal returns of stock listed on the AEX are significantly different from zero. In these time intervals the stock price drop is not equal to the dividend amount. The tax effect theory, transaction cost theory and the market microstructure theory account for price drops that are not equal to the dividend amount (Isaksson, 2013). The credit crisis influenced transaction costs and the bid-ask spread. The transaction cost theory en the market microstructure theory imply that as a result the ex-dividend stock price behaviour should

(25)

25 change. The ex-dividend day abnormal returns and the k-3 to k+3 abnormal returns are significantly larger than zero. The regressions on the ex-dividend abnormal returns, k-3 to k3 abnormal returns and all abnormal returns do not give any significant coefficients for the dummy variable crisis. This implies that the ex-dividend stock price behaviour of stock listed on the AEX has not changed due to the credit crisis. Empirically proving the changes in transaction costs and market microstructure due to the credit crisis did not have a significant effect on the ex-dividend stock price behaviour.

(26)

26 References

Akhmedov, U, and Jakob, K, (2010) The Ex-dividend Day: Action On and Off the Danish exchange, The Financial Review, V45, pp 83-103,

Bali, R, and Hite, G,L, (1998) Ex-dividend day stock price behavior: Discreteness or tax-induced clienteles?, Journal of Financial Economics, V47, pp, 127-159,

Barclay, M, (1987) Dividends, taxes and common stock prices: The ex-dividend day behavior of common stock prices before the income tax, Journal of Financial Economics, V19, pp, 31-44, Beranek, W,, & Campbell, J, A, (1955), Stock price behavior on ex-dividend dates, The Journal of

Finance, V10,4, pp, 425-429,

Berk, J, and Demarzo, P, (2011) Corporate Finance, USA: Pearson Education,

Bloomberg (2013) Consulted 07-11-2013 (http://www,bloomberg,com/quote/GNTH10YR:IND/chart) Elton, E,J, and Gruber, M,J, (1970) Marginal shareholder tax rates and the clientele effect, Review of

Economics and Statistics, V52, pp, 68-74,

Elton, E,J, Gruber, M,J, Rentzler, J, (1983) A simple examination of the empirical relationship

between dividend yields and deviations from the CAPM, Journal of Banking and Finance, V7, pp, 135-146,

Elton, E,J, Gruber, M,J, Rentzler, J, (1984) The ex-dividend day behavior of stock prices; a re-examination of the clientele effect: A comment, The Journal of Finance, V39,2, pp, 551-556, Frank, M, and Jagannathan, R, (1998) Why do stock prices drop by less than the value of the

dividend? Evidence from a country without taxes, Journal of Financial Economics, V47, pp, 161-188,

Fama, F, E, (May 1970), Efficient Capital Markets: A Review of Theory and Empirical Work, The Journal of Finance, V25, pp, 383-417,

Hupperetsa, E,C,J, and Menkveld, A,j, (2002) Intraday analysis of market integration: Dutch blue-chips traded in Amsterdam and New York, Journal of Financial Markets, V5, pp, 57–82, Isaksson, A, (2013) The Ex-Dividend-Day Price Behaviour of Blue-Chip Stocks: International

(27)

27 Kalay, A, (1982) The ex-dividend day behavior of stock prices: A re-examination of the

clientele effect, Journal of Finance, v37, pp, 1059-1070,

Karpoff, J,M, and Walkling, R,L, (1988) Short-term trading around ex-dividend days: Additional evidence, Journal of Financial Economics, V21, pp, 291-298,

Lakonishok, J, and Vermaelen, T, (1986) Tax induced trading around ex-dividend days, Journal of Financial Economics, V16, pp, 287-320,

Madhavan, A, (2000) Market microstructure: A survey, Journal of Financial Markets, V3, pp,205-258,

Melvin, M, and Taylor, M,P, (2009) The crisis in the foreign exchange market, Journal of International Money and Finance, V28, pp, 1317–1330,

Miller, M,H, and Modigliani, F, (1961) Dividend policy, growth, and the valuation of shares, Journal of Business, V34, pp, 411-433,

Staatsblad van het koninkrijk der Nederlanden (V631) (2006) Stiglitz, J, (2010) Vrije Val, W,W, Norton & Company, Houten Wet Inkomstenbelasting (2001)

(28)

28 Appendix 1

Dutch government 10 year bonds,

Rate per year %

(29)

29 Appendix 2

Name α β Name α β

Dutch Shell -,0020879 ,6820587 DSM -,0030656 ,4930505

Unilever -,0034173 ,4744673 Reed Elsevier -,0011913 ,7708993

Heineken -,0046266 ,3245951 KPN ,00019 1,083,843

ING ,0033824 1,390,804 Randstad -,0022135 ,636549

ASML ,0060631 1,763,328 Wolters Kluwer -,0029949 ,5818122

Philips ,0048204 1,591,393 Fugro -,0045441 ,2486356

Unibail Rodamco -,0050467 ,1382418 Air France-KLM -,0005602 ,8481418

Arcelormittal -,0033316 ,3207477 Corio -,0057213 ,1094847

Ahold ,0007243 1,107,057 SBM Offshore -,0043321 ,3005647

Aegon ,0030877 1,468,563 Imtech -,0042918 ,3158722

Akzo Nobel -,0024576 ,6338674

R

it

– R

ft

= α

i

+ β

i

(R

mt

– R

ft

) +

it α Is the CAPM estimated constant term, β Is the CAPM estimated systematic risk,

(30)

30 Appendix3

Variables EXDAY CAR1 CAR3 CAR5

Coef Std. error Coef Std. error Coef Std. error Coef Std. error

Unilever ,0027805 ,0025523 ,008804* ,0047739 ,004481 ,0073777 ,0081486 ,0097316 Heineken ,0056889 ,0043781 ,0144151* ,007556 ,0085025 ,0116277 ,009666 ,015753 ING ,0010122 ,0023493 ,0038355 ,0034459 -,0109798 ,0086108 -,0127039 ,012603 ASML ,0289524** ,0136291 ,0040436 ,0158045 -,0244409 ,0218325 -,0253502 ,026005 Philips ,0063453 ,0065665 -,003164 ,0065458 -,03545*** ,011536 -,03865*** ,0113504 Unibail -,000101 ,0078867 -,0024447 ,0156426 ,0045923 ,014831 -,0221983 ,0191481 Arcelor ,0039216 ,0048175 -,0080037 ,0093869 -,0167632 ,0150145 -,0277889* ,0167213 Ahold -,0132264* ,0068493 -,0147573 ,0201906 -,0254525 ,0179174 -,0263956 ,0189203 Aegon ,0068048 ,0068299 ,0165407 ,0113768 ,0130136 ,0142026 ,0161893 ,0123504 Akzo ,0033256 ,004391 ,0009122 ,0061388 -,0057803 ,0095557 -,0061902 ,0162359 DSM -,0029479 ,0039987 -,0010349 ,0067014 -,0096447 ,0104553 -,004887 ,0172659 Reed ,0013786 ,0033008 ,0086905** ,0041192 ,0021295 ,0081496 ,0039382 ,0119121 KPN ,0094205 ,0057421 ,0150904 ,009381 ,0050119 ,0158987 ,0135569 ,0176719 Randstad ,0015998 ,0045756 -,0101083 ,0122156 -,044841** ,0209669 -,0425056* ,0240975 Wolters ,0041455 ,0041754 ,0088948 ,0092204 -,0019729 ,0090082 ,0082656 ,013882 Fugro ,0170045 ,0116427 ,0127039 ,0160173 ,0110657 ,0297524 ,0144396 ,0299851 Airfrance ,0193456** ,0092144 ,051075* ,0260735 ,0598808* ,0318389 ,0304699 ,0381647 Corio ,0077087 ,010991 ,0219441 ,0148562 -,0000451 ,0200106 ,0108312 ,017036 SBL ,0251792** ,0108005 ,0362825** ,0160983 ,0204582 ,0205235 -,0067944 ,0390885 Imtech ,0060132 ,0054737 ,0165646** ,0079057 ,0274468** ,0106066 ,026653 ,0194468 Constant -,0005282 ,0012762 -,0041506 ,0032539 ,007317 ,0062233 ,0090001 ,0067008 N 256 256 256 256 R2 ,1242 ,1249 ,1219 ,0919

Notes: ***, ** and * reflect significance at 1 percent, 5 percent and 10 percent. All the regressions are robust because there is suspicion of heteroscedasticity and outliers. The constant term is the estimate for Shell.

(31)

31 Appendix 4

EXDAY THREECAR

Coef Std, error Coef Std, error

CRISIS -,0023494 ,0031239 ,0024152 ,0077876 DY ,5228429*** ,1566187 ,5342715 ,3392598 EXVOLmln _,00674 _,415 _2,07 _1,27 TC -,0587543 ,1126183 -,3366387 ,2542027 MVmln _,0377 _,147 _,226 _,437 PB -,0007172 ,0012194 -,0001321 ,0021392 LEVERAGE -,0001982 ,0002682 ,0013472* ,0007062 AVVOLmln -,220 ,553 2,92* 1,57 Unilever ,0093093 ,009384 ,0065054 ,0254472 Heineken ,0124075 ,0123327 ,0160633 ,0336461 ING -,0015159 ,0054655 -,0273888 ,0169959 ASML ,0302183* ,0166791 -,0007761 ,0357687 Philips -,0007125 ,0111789 -,0188404 ,0304337 Unibail -,0051231 ,0162696 ,0007006 ,0410494 Arcelor ,0102973 ,0081975 -,0075531 ,0257496 Ahold -,0122031 ,0124556 ,006465 ,0356859 Aegon ,0027941 ,0115131 ,0450403 ,0318142 Akzo ,0032568 ,0134023 ,0145639 ,0360164 DSM -,0039605 ,0136423 ,016337 ,0376375 Reed ,0038925 ,0146679 ,0685657* ,0351948 KPN ,0145924 ,0184956 -,0313066 ,0468174 Randstad -,0038584 ,0143136 -,0235444 ,0401688 Wolters -,0005275 ,0144607 -,0012897 ,0380762 Fugro ,0123032 ,0185838 ,0191088 ,0481122 Airfrance ,0185452 ,0162241 ,074642 ,0456532 Corio -,0133751 ,0203637 -,0193603 ,0468284 SBM ,0232875 ,0165803 ,0333832 ,0411656 Imtech ,0005053 ,0142567 ,0621721 ,0387932 Constant -,0007774 ,0165989 -,050938 ,046999 N 256 256 R2 _,1758 _,1227

Notes: mln is per million. ***, ** and * reflect significance at 1 percent, 5 percent and 10 percent. All the regressions are robust because there is suspicion of heteroscedasticity and outliers. These regressions are an expansion of Table 4 and Table 5.