Target payouts and adjustment speeds of share
repurchases for frequent repurchasers in the European
Union.
Hugo Essink University of Groningen
Abstract:
I examine the target payout ratios and adjustment speeds of share repurchases for frequent repurchasers from 1998 to 2006 in the 15 EU nations that were a member of the EU before may 2004. I recognise the uncertainty on the measurement of share repurchases and therefore I identify seven share repurchase measuring methods. Using Lintner regressions, I find increasing target payout ratios, declining adjustment speeds and an increasing responsiveness to net income of share repurchases in the EU. This is in line with Skinner (2008) for the US and Von Eije and Megginson (2008) for the EU. I also present which of the seven identified share repurchase measuring methods can be best used to measure share repurchase patterns in the EU.
JEL Classification: G35
I. Introduction
In the past few decades extensive research has been done to get a better insight in the payout policies of companies. Much of this research builds on the cornerstone article of Lintner (1956) who concluded that managers, from 28 companies, determined their dividend policy by targeting a long-term payout ratio. Moreover, Lintner (1956) showed that managers were reluctant to adjust to that target payout ratio quickly. This implied that for cash dividends the speed of adjustment was relatively small, which in turn implies sticky cash dividends.
However, Lintner (1956) predominantly focused on cash dividends, and not on share repurchases. This can be explained by the fact that during the time that Lintner wrote his paper companies hardly repurchased their own shares (Brav et al., 2005).
This did not change for two decades but during the 1970’s firms slowly started to repurchase their own shares. From 1970 to 1976 New York Stock Exchange listed companies repurchased shares at levels around €2 billion per year. But after 1976 share repurchases started to increase year by year to almost 8 billion in 1980 (Shoven, 1986).
During the early 1980’s the emphasis of the research about payout policies shifted towards cash dividends and share repurchases because from that time on share repurchases became an economically significant phenomenon (Bagwell and Shoven, (1989); Grullon and Michaely, (2002)). During the 1980’s the aggregate value of share repurchases on the American Stock Exchange (ASE), NASDAQ and the New York Stock Exchange (NYSE) was about one third of the value distributed as cash dividends (Ikenberry et al. 1995). At the end of the 1980’s the total value of share repurchases had grown to about half the amount that companies distributed as cash dividends. Weston (2007) shows that over the 22 year period 1984 trough 2005 US share repurchases grew at a compounded rate of 11.2% whereas cash dividends only grew by 7.8%.
In this paper I combine the classical Lintner approach with share repurchases in order to measure both the target payout ratio of share repurchases as well as their adjustment speeds.
calculating these repurchases which makes it difficult to compare the results. For example some studies measure share repurchases by the reported amounts of share repurchases on the company’s cash flow statements (Von Eije and Megginson, 2008). And others use net repurchases by adjusting the gross repurchases to net out the effect of any offsetting financing transaction (Skinner, 2008). Moreover, Shoven (1986) calculates share repurchases by multiplying decreases in shares outstanding with the average share price of the month when the repurchase took place.
This paper recognizes this uncertainty on the measurement of share repurchase amounts and I will therefore present an analysis of the various ways in which share repurchases can be measured. In total I identify seven share repurchase measuring methods. For all these methods I present total amounts, number of repurchasing firms, series statistics, and correlations.
Then I measure the target payout ratios and speed of adjustments of all seven methods for companies situated in the European Union before May 2004, the so called EU15 countries. I study them over two periods. The first period ranges from 1998-2001 and the second from 2002-2006. I then analyse first, if the various share repurchase measuring methods give similar results for the target payout ratio and speed of adjustment and second, if for the seven methods these variables develop in the same way over the two studied periods.
Between the different methods the target payout ratios range between 10.7% and 27.6% in the period 1998 – 2001 and between 5.1% and 52.6% in the period 2002 – 2006. Moreover, six share repurchase measuring methods show an increase in the target payout ratio over time, and only one method did not.
In the period 1998 – 2001 the adjustment speeds of the seven methods range between 42.0% and 83.2% and they range between 27.0% and 91.9% in 2002 – 2006. Five out of seven methods show a decrease in adjustment speeds, where it only increased for two.
is not yet taken into account in this paper but it is recommended to calculate such a method in further research and analyse if the this adjusted method also has an increasing target payout ratio and a decreasing adjustment speed over time.
The structure of this paper is as follows. Section II explains why share repurchases have received so much attention in recent research and what the possible reasons are for
II. The rise of share repurchases
In this section I will distinguish between the quantitative effects (size and growth) and the theoretical reasons for the large growth in share repurchases.
Quantitative effects:
In the recent decades there has been a shift in the way by which companies reward shareholders. In theory, when perfect markets are assumed, companies can (in addition to cash dividends) distribute cash to investors through share repurchases. Companies can do so because of the following reason. When a company chooses to repurchase shares, the earnings are spread over less outstanding shares. This increase in the earnings per share causes the price of the remaining outstanding shares to rise. In perfect markets, this increase in price can compensate for the loss in cash required by the firm to repurchase the shares. So the companies distribute cash by buying the shares of shareholders who prefer a cash return, while they do not distribute to shareholders who prefer not to sell their shares in the repurchase transaction.
In the 1950’s and 1960’s companies used to predominately pay cash dividends and share repurchases were virtually nonexistent (Brav et al. 2005). But from the 1970’s onwards there is a sharp rise in cash distributions through share repurchases (Bagwell and Shoven, 1989).
Bagwell and Shoven (1989) show that between 1977 and 1987 real share repurchases in the US grew 824 percent. Over the period 1980-1998 share repurchase expenditures, relative to total earnings, grew at an average annual rate of 28.3 percent while cash dividends to earnings only grew at 7.5% over the same period (Grullon and Michaely, 2002). They also show that between 1984 and 1998 the dollar amount distributed to investors trough share repurchases relative to cash dividends was 51.5%. In 1998 US industrial firms for the first time in history spent more money on share repurchase programs than on dividends. This shows that share repurchases have become a very important tool of cash distribution to shareholders.
dividends paid will be. Skinner (2008) finds that cash dividends relative to total earnings are decreasing whereas share repurchases use up more and more earnings, suggesting a growing role of share repurchases in the payout ratio. For this reason I focuses on the payout through repurchases and on their relationship with net income in this paper.
All the articles mentioned above find an increase in share repurchase programs of US companies over the past decades. This increasing trend is also found in the EU15 countries during the period of my dataset. Figure 1. shows the yearly value of share repurchases in the EU15 from 1997 to 2006. Figure 1 shows a sharp increase of share repurchases over the 10-year period especially from 2003 onwards. This is in line with the trend of the recent decades in the US, and shows that share repurchasing is also increasingly used in the EU15.
Figure 1. Total Value in real millions € for the period 1997 - 2006:
The share repurchase amounts used in Figure 1 are calculated following the method used by Von Eije and Megginson (2008).
Theoretical reasons:
influence payout policies. Personal tax rates are important since there is a difference between the effective tax rate on dividends and capital gains. This makes it more profitable for an investor to receive a return in the form of a capital gain than in the form of a cash dividend. Capital gains also offer another advantage over cash dividends by giving the investor the ability to time their payout and thus their tax liability. The tax liability created by cash dividends has to be paid when the dividend is received, while one can postpone paying these taxes by timing the realisation of capital gains through repurchases (Bagwell and Shoven, 1989).
But taxes cannot be the only explanation for the shift towards share repurchase programs because the U.S. Tax Reform Act of 1986 lowered the difference between the tax rates on dividends and the rates on capital gains for the individual investor. This should have created a shift back towards cash dividends but Bagwell and Shoven (1989) find no such trend in their data. Fama and French (2001) show that the propensity to pay cash dividends among NYSE, AMEX and NASDAQ non-financial non-utility firms drops from 66,5% in 1978 to 20.8% in 1999. Despite the decline in the cash dividend payout ratio the total payout ratio did not decrease between 1978-1999 (Grullon and Michaely, 2002). This means that U.S. firms have been increasingly distributing cash to shareholders in other ways than through cash dividends, to keep the total payout ratio at the same level.
Grullon and Michaely (2002) show that from the mid eighties onwards the majority of firms starting to payout cash to shareholders prefer to do that in the form of share repurchases rather than dividends. And that large established firms do not reduce their cash dividend payments. This can be explained by the fact that dividends can reveal information in a payout. Cash dividends can signal an increase in firm value to an investor. Consequently firms are very reluctant to cut their cash dividend payout ratio because this could send a negative signal to investors which will then lead to a drop in the firms share price. A decrease in share repurchases does not always lead to a drop in the firms share price. So if a firm wants to increase its payout ratio, it prefers to do so by using share repurchases instead of cash dividends because of the flexibility that share repurchases offer (Brav et al. 2005).
investment, which in turn leads to an increase in the share price due to an increase in demand for that stock in addition to the demand already created by the share repurchase.
Share repurchases can also be used to change the financial structure of a company. Weston and Siu (2003) demonstrate that a company’s debt to equity ratio (D/E ratio) can be changed using repurchases. In the D/E ratio, the debt component is calculated as debt net of excess cash and marketable securities. So when a company uses excess cash to repurchase some of its shares it not only decreases equity (if the shares are withdrawn) but also increases net debt. So share repurchases can bring the D/E ratio closer to its optimum. This can lead to a lower cost of capital which results in a higher share price.
Agency problems can also be reduced by the use of share repurchases. Jensen (1986) claims that share repurchases contribute to a better use of free cash flows. When managers use a company’s free cash flow to repurchase shares the managers cannot use the free cash flows to make unbeneficial investments. Share repurchases also mitigate the difference in interests between managers and owners (Weston and Siu, 2003). When shares are repurchased the non selling shareholders increase their percentage of ownership. As managers generally do not sell their shares they also increase their percentage of ownership over the firm. If these percentages are high enough managers are more bound to think as owners instead of just managers. This will lead to less agency costs.
III. Measuring share repurchases
A big problem regarding share repurchases is the way in which they are measured. Figure 1. Showed the yearly share repurchase amounts calculated by the method of Von Eije and Megginson (2008). This method results in so called gross amounts (which I will explain further in section IV). But in fact share repurchases can be calculated in many different ways.
One can measure share repurchases by looking at the announcements made by firms that want to repurchase shares. However, a firm does not have to announce its plans it can also repurchase shares without making an announcement. If a firm announces the share repurchase it, however, protects itself against the manipulation provision of the Securities and Exchange Act (Jagannathan et al., 2000). In the US these announcements are generally collected in the Security Data Company database (SDC). But the SDC often overstates the actual share repurchase numbers because the announcements give the companies the right but not the obligation to actually make all the share repurchases. In practise not many U.S. firms make all the repurchases that they announced and have the rights to do (Jagannathan et al., 2000). So using announcements to measure share repurchases amounts may not be the best measure.
Since 1984 U.S. firms have to report share repurchases on their cash flow statements but not all share repurchases can be treated as non-cash dividends (Fama and French 2001). According to Fama and French there are two cases where share repurchases are not substitutes for cash dividends namely: when repurchased stocks are used for employee stock ownership plans (the value of the stocks flow to the employee’s, and not to the investors), or when they are used to finance a merger1. The repurchased stocks are simply used to finance the investment (capital value in the form of shares flows to the shareholders of the acquired firms as payment). Acquired firms often prefer stock as payment for a merger because of the mentioned tax advantage.
Fama and French (2001) and Skinner (2008) therefore try to remove the effects of these two exceptions to derive so called net repurchase amounts. To do this Fama and French
(2001) and Skinner (2008) use Compustat2numbers and measure the increase in common treasury stock (Compustat #226). This calculation method is called the ‘treasury stock method’. However if firms report zero treasury stock in the current and past year Fama and French (2001) and Skinner (2008) conclude that a firm is using the ‘retirement method’ of repurchasing. This method measures share repurchases as the difference between stock purchases (#115) and stock issuances (#108). If both the ‘treasury stock’ and ‘retirement’ method give negative results, repurchases are set to zero. In practise the ‘treasury stock’ method is preferred over the ‘retirement’ method since the former nets out any issuances, including non-cash ones (Skinner, 2008). For example if a firm made net purchases (Computstat #115 - #108) of about €5 million but then issues treasury shares for the same amount for two purposes - to fund employee stock ownership plans or when they are used to finance a merger – the net increase in treasury stock is zero. But the difference between stock purchases and stock issuances would not change and still be €5 million. And thus the treasury stock method results in net repurchase amounts which can be treated as substitutes for cash dividends.
As mentioned earlier Von Eije and Megginson (2008) used gross share repurchase data from Worldscope3. Most share repurchase amounts after 1998 are available in euros except
those of companies situated in Great Britain, Denmark and Sweden. For those countries Von Eije and Megginson (2008) recalculate all amounts to euros using end-of-year exchange rates. And then they change all the nominal amounts to real values for the year 2000. Von Eije and Megginson also deflate the data of the period 1989 to 1998 using the consumers’ price index for all items in the individual countries. From 1999 onwards they deflate by using the consumers’ price index for all items in the EU15.
Shoven (1986) calculates share repurchases by multiplying decreases in common share outstanding with the average of the stock price at the end of the preceding month and the price at the end of the month in which the reduction took place. Shoven (1986) used the CRSP monthly stock return file to get the necessary information and adjusted the decreases in common shares outstanding for splits and reverse splits.
2Standard & Poor's Compustat is a database of financial, statistical and market information on active and inactive
companies throughout the world.
IV. Data and Methodology Data:
In this paper I examine different methods that can be used to measure share repurchase amounts. I use data from 15 EU countries over two five year periods, 1997-2001 and 2002-2006. The 15 countries studied are; Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, The Netherlands, Portugal, Spain, Sweden and the United Kingdom. The data are provided and described by Von Eije and Megginson (2009).
Since the dataset used in this paper is similar to the dataset used in the article by Von Eije and Megginson (2009) there is an opportunity to compare the results of their method of calculating share repurchases with various other methods described in this paper. And to see whether there are economic differences and if so what these differences tell us.
The dataset uses companies with usable ISIN codes and eliminates the companies with identical ISIN codes or names. Since it is common to use industrial firms in empirical studies they also exclude utility and financial firms. The dataset is also corrected for companies that are listed on multiple exchanges by only selecting those companies who are headquartered in the same country as where they are listed.
These selection criteria’s result in 5654 companies studied over a 10 year period resulting in 56540 studied firm years. Out of the 5654 companies 2080 (36.8%) are headquartered in the United Kingdom, 947 (16.8%) in France, 860 (15.2%) in Germany, 289 (5.1%) in the Benelux (Belgium, The Netherlands and Luxembourg), 747 (13.2%) from Southern Europe (Spain, Greece, Italy and Portugal) and 731 (12.9%) from other countries (Austria, Denmark, Finland, Ireland and Sweden).
All the collected data are corrected for exchange rates and consumer price index rates. Methodology:
repurchases based on POC amounts as PREP Repurchases. Von Eije and Megginson (2008) also include negative purchases of common stock amounts in their analysis. For comparison reasons I have recalculated their share repurchase amount in the same way (including negative POC amounts). I Also calculated PREP excluding these negative amounts in order to analyse the effects that the negative numbers have. I call the share repurchase amount that follow from this adjustment PREPA repurchases.
Fama and French (2001) and Skinner (2008) give two methods for calculating share repurchases namely the ‘treasury stock’ method and the ‘retirement’ method.
The treasury stock method is simply the change in common treasury stock over a time period, TSt+1 – TSt = ∆TS, where TStis the amount of common treasury stock at time t and
TSt+1 is the amount at time t+1. The treasury stock numbers are taken directly from the
balance sheet. If ∆TS is smaller than or equal to zero Skinner (2008) suggests using the retirement method.
The retirement method is calculated as the difference between purchases of common stock (POC) and the sales of common stock (SOC). These numbers are taken directly from a company’s cash flow statement4. If the treasury stock method and the retirement method are both negative then net share repurchases are set to zero. Skinner (2008) also states that it is preferable to use the treasury stock method over the retirement method since the first method nets out cash and non-cash issuances of stock. For the rest of this paper I will refer to share repurchases amounts calculated by using the treasury stock method (and if not available the retirement method) as the Skinner Method (SREP) repurchases. I will also analyse the retirement method and the treasury stock method separately. I call these share repurchases amounts respectively RREP and TREP repurchases.
The retirement method can also be adjusted for missing values of POC and SOC. This leads to a different repurchase variable which I call RREPA5. If one or both values is missing I exclude the POC and SOC variable from the share repurchase amount.
4 I treat negative sales of common stock as purchases of common stock. When my dataset has no value for POC I automatically give no value for the RREP amount. If there is a value for POC but no value for SOC I do include it in the RREP variable.
A problem with using treasury stock numbers is that in Europe firms can report these numbers at par value or at market values. While in the US firms are only allowed to report these numbers at market value. This makes it easier to compare US firms with each other since all the reported amounts have the same underlying valuation base. For the dataset used in this paper it is difficult to make a distinction between firms that report their treasury stock amounts at par value and those that report them at market value. I will give methods for identifying firms that report at par in the appendix.
Another method in which share repurchase numbers can be calculated is to take the change in common shares outstanding (CSO) over a time period t and, if this amount is negative, multiply this by the average stock price (ASP) during time period t (corrected for exchange rate and the consumer price index). This can be written as CSOt+1 – CSOt =
∆CSO. When ∆CSO < 0, multiply ∆CSO with ASP. This method makes it fairly easy to calculate share repurchase amounts since one only needs to know how many shares are outstanding during a year and the average share price. The number of common shares outstanding are taken directly from the end-year balance sheet. For the rest of this paper I call these share repurchase amounts Change CSO Repurchases (CREP). Shoven (1986) used a similar method but instead of year average stock prices Shoven (1986) used monthly average prices of the months in which the change in CSO took place.
I will also calculate the responsiveness to net income, the adjustment speeds and the target payout ratios of the different share repurchase measuring methods. To do so I use the following Lintner regression model:
∆Dt=
α
0 +α
1.NIt -α
2.Dt-1 +ε
t ……….(1)where:
∆Dt = the change in share repurchases
NIt= Net Income in constant prices.
Dt-1= the lagged variable of share repurchases (LPREP, LPREPA, LRREP, LRREPA,
LTREP, LSREP and LCREP)
ε
t= the error term of the regression equation.α1= the responsiveness to net income.
α2= the speed of adjustment.
The target payout ratio can be calculated by dividing α1 with α2 i.e. dividing the
V. Analysis and Results
I will start my analysis by looking at the different methods of calculating share repurchases, this is a good way to get a feel for the different methods. The seven methods I identified in Section IV are: POC Repurchases (PREP), adjusted POC repurchases PREPA, Retirement Method Repurchases (RREP), adjusted Retirement method repurchases RREPA, Treasury Stock Method Repurchases (TREP), Skinner Method Repurchases (SREP) and Change CSO Repurchases (CREP). All these different methods result in different share repurchases amounts, but how much do they differ? For an analysis of this issue I will present the differences in total values, number of repurchasing firms, series statistics, correlation and the results from the Lintner regression. In addition I will also try to answer the question “ which measuring method most accurately measures share repurchase amounts as a form of non-cash dividends? “ in appendix A.
Total values:
This table reports the total value in millions € per year for each share repurchase measuring method for the period 1997 to 2006. The measuring methods in this table correspond with the methods described in section IV.
When we look at Figure 2 one can see that PREP, PREPA, RREP, RREPA and (to a lesser extent) SREP share repurchases give comparable results throughout the 10-year period. The five methods follow each other closely, especially from 1997 to 2003. From 2004 onwards SREP values are lower than those of PREP, PREPA, RREP and RREPA repurchases but it continues to follow the movements of these others methods.
The adjustments made to PREP and RREP to calculate PREPA and RREPA hardly have any influence on the share repurchase values. Figure 2 shows that PREP and RREP almost overlap respectively PREPA and RREPA. During the 10-year period PREP and PREPA values are very similar to each other and even the same in the years 1998 and 2006.
Table 1 shows that the Treasury Stock method of repurchasing (TREP) results in the lowest value in all years from 1998-2006 (in 1997 CREP repurchases were lower). In general TREP repurchases follow the same path as PREP, PREPA, RREP, RREPA and SREP repurchases but at lower values.
CREP repurchases on the other hand follow a different path. Where the companies repurchased shares for a value of €4.699 billion in 1997 (which is the lowest value of all methods in that year) they increased their repurchases in the three years that followed, to a value of €47.068 billion in 2000. After that CREP repurchases started to gradually decrease to a level of 23.961 billion Euro’s in 2004. In the last two years of the 10-year period CREP share repurchases sharply increased again to €60.635 billion and €47.680 billion in 2005 and 2006 respectively.
Table 1. Total values in millions €, per measuring method, from 1997 – 2006:
The volatile path that CREP repurchases follows can either be the result of a large decrease in common shares outstanding (CSO) or an increase in the price that is paid to buy back the outstanding shares, because those variables are used in the calculation of CREP repurchases6. Figure 3 shows the number of outstanding shares that were bought per year during the 1998 – 2006 period. It shows that the first spike in CREP repurchases in the year 2000, as shown in Figure 1, is the result of a spike in the number of repurchased shares in that year. Figure 3 shows another big spike in the number of repurchased share in 2003 however this spike does not result in a large increase in the CREP repurchase amount. This means that the prices of the repurchased shares were lower in 2003 than in prior years. In 2005 there is another spike in the number of repurchased shares. This spike in 2005 and the decrease in 2006 can also be seen in Figure 1 which means that the differences in share repurchase amounts can, in those two years, be explained by the number of repurchased shares.
Figure 3. Number of repurchased shares (used in CREP method calculations) from 1998 – 2006:
The total values of the different repurchase methods can also be divided into two subsequent periods. The first period ranges from 1997 – 2001 and the second from 2002 –
2006. The total value in real millions of euro’s for the two subsequent periods can be found in Table 2.
Table 2 reports the total value of all identified measuring methods for three different periods. The first period ranges from 1997 to 2001, the second from 2002 - 2006 and the third ranges from 1997 - 2006. Each method shows and increase between the first and the second period.
When we look at the first 5-year period in Table 2, 1997 – 2001, we see that during these five years CREP repurchases had the highest total value, €104.942 billion. This is mostly due to the enormous increase in CREP repurchases in the years 1999 and 2000. In 1998 total CREP repurchases were €5.182 billion. This grew to €15.702 billion in 1999, an increase of 203%, and to 47.068 billion Euro’s in 2000 which resembles an increase of another 200%. This increase is caused by the abolishment of government restrictions on share repurchasing throughout Europe in 1998. This led to an enormous increase in share repurchases especially in England, France and Germany in the years that follow.
In the second 5-year period, 2002 – 2006, share repurchases became even more popular. All share repurchase methods show an increase when compared to the prior five years. PREP and RREP (as well as PREPA and RREPA) repurchases even doubled and CREP repurchases showed an increase of 91% between the two periods. TREP repurchases shows the smallest increase of all the different methods. Between the two periods TREP share repurchases only increased by 37.5%. All methods result in increasing share repurchase amounts. Which is in line with the trend of the preceding decades.
Over the 10-year period share repurchases are the highest when the CREP method is used. From 1997 to 2006 CREP repurchases amounted to a total value of €305.568 billion. PREP, SREP and RREP repurchases had total values of respectively €290.325, €238.461
and €265.742 billion. Figure 1 and Table 2 show that in the first five years (1997 – 2001) PREP(A), SREP and RREP(A) repurchases result in similar values and remain to do so until 2004. The differences between the total values of these five methods can almost completely be explained by the values found in 2004, 2005 and 2006. In these years the difference between PREP and SREP amounted to €48.160 billion. Whereas the difference between PREP and SREP is about 51.864 billion Euro’s over whole the 10-year period. This means that the last three years, of a ten year period, were responsibly for 92.86% of the difference between the two methods.
The same is found when we look at the difference between PREP and RREP share repurchases. In the last three years the difference between the two methods was about €15 billion whereas the total ten year difference is €25 billion. The last three years thus account for 60% of the total difference in value.
Number of repurchasing firms:
Figure 4 shows the number of firms that repurchase shares in a given year for each measuring method. Table 3 gives the exact number of firms responsible for the total amount of repurchases per year for each method (as shown in Table 1). Table 3 shows that the PREP and PREPA method have identical numbers of firms generating the share repurchase amounts (resulting in a complete overlap of the PREP and PREPA lines in Figure 4).
Table 3. Number of firms responsible for the yearly repurchase amounts, per method, per year: Year Measuring Method 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 PREP 94 175 197 319 342 355 380 389 535 631 PREPA 94 175 197 319 342 355 380 389 535 631 RREP 77 164 192 289 336 329 369 320 424 495 RREPA 76 162 188 270 321 318 348 304 387 473 TREP 191 159 240 373 409 380 375 298 341 309 SREP 244 277 357 535 597 562 600 480 555 551 CREP 111 231 342 476 592 604 557 422 393 412
The numbers, per year, in this table represent the number of firms that are responsible for the share repurchase amounts as shown in Table 1. The number of firms is given for all seven share repurchase measuring method.
The PREP(A) share repurchase method is the only method that results in a constant increase in the number of repurchasing firms year after year. This increase is especially large in the last two years of the 10-year period, 2005 and 2006. Which also explains the increase in total PREP values in those years.
RREP and RREPA follow each other very closely since the yearly number of repurchasing firms are very close to each other. The lines in Figure 4 for the two methods are therefore also very similar. When one compares PREP(A) and RREP(A) it shows that especially up until 2003 they give comparable results but in the three years that follow they start to differ. In 2004 PREP(A) repurchases show a small increase of 9 firms contributing to its total share repurchase amount whereas RREP and RREPA show a decrease in firms of respectively 49 and 44 firms. In 2005 and 2006 the firms that repurchase start to increase rapidly for the PREP(A) and RREP(A) methods. The lines of PREP(A) and RREP(A) repurchases in Figure 4 follow the path of the total amounts of these methods in Figure 1. This indicates that during the 10-year period an increase (decrease) in repurchasing firms often leads to an increase (decrease) in PREP(A) and RREP(A) total repurchase amounts.
amounts in the years 2002 and 2003 whereas the number of repurchasing firms grew during those years.
TREP repurchases had the lowest total value of share repurchases in almost every year (Table 1) but the number of firms that contributed towards this amount is close to the numbers of RREP(A) repurchases. So even though the number of firms of the two methods are quite similar their real total value of share repurchases differ a lot. This could be explained by the difference in reporting standards between the US and the EU. In the US firms are only allowed to report treasury stock at market values on their balance sheet whereas in the EU firms are able to choose between par values and market values. The fact that TREP share repurchase values are the lowest of all methods (while the number of repurchasing firms is not) suggests that, in the EU15, many firms report their TREP repurchases at par values.
The SREP and CREP share repurchases amounts, as shown in Figure 1, are quite different from 1997 – 2006, especially from 1999 to 2004, but interestingly the number of firms that are repurchasing during that period are similar. The years 2005 and 2006 show opposite results. In those two years the increase in the share repurchase amounts, of the SREP and CREP method, are similar but the number of repurchasing firms is not.
Series statistics:
Table 4 shows the standard deviations, mean value, minimum, maximum and medians for all methods for the period 1997 – 2006. As one can see, the standard deviations for all methods are high. This can be explained by the fact that many firms do not repurchase in most of the years in the dataset (every year consists of 5654 companies of which only the number of firms shown in Table 3 repurchase).
Table 4. Statistics per method measured for all firms in which observations were available during the period 1997 - 2006:
Measuring Method Standard deviation Mean Minimum Maximum Median
PREP 150.920 10.931 -488.174 10543.360 0.000 PREPA 150.876 10.979 0.000 10543.360 0.000 RREP 144.531 10.001 0.000 10543.360 0.000 RREPA 144.712 9.827 0.000 10543.360 0.000 TREP 85.628 3.455 0.000 8046.624 0.000 SREP 123.587 8.449 0.000 8046.624 0.000 CREP 184.692 9.453 0.000 20879.920 0.000
This table gives various statistics for all seven share repurchase measuring methods. The statistics are the standard deviation, the mean value, the series’ minimum and maximum and its median value.
All methods also have medians of zero which, in combination with the other values in Table 4, means that the deviations are all very skewed. The mean values of PREP(A), RREP(A) and CREP repurchases are very close to each other, while TREP repurchases measure the lowest mean value. This seems logical since the total values, from Table 2, are also close to each other, and TREP has the lowest total value.
The minimum values are zero for all methods except for PREP repurchases since Von Eije and Megginson (2008) included negative POC values in their calculations (as explained in section IV) whereas the other methods do not. The Maximum values for PREP(A) and RREP(A) are identical because those methods predominantly use purchases of common stock amounts in their calculations. The SREP measuring method has the same maximum value as the TREP method, even though the RREP measure (which is also used in SREP calculations) has a higher maximum, because SREP repurchases used TREP amounts even when higher RREP amounts are available.
Correlations:
Table 5. Correlation Matrix of the different Repurchase Methods measured for all firms in which observations were available during the period 1997 - 2006:
Measuring Method PREP PREPA RREP RREPA TREP SREP CREP
PREP 1.000 PREPA 0.999 1.000 RREP 0.997 0.997 1.000 RREPA 0.997 0.997 1.000 1.000 TREP 0.681 0.681 0.695 0.695 1.000 SREP 0.942 0.942 0.945 0.945 0.666 1.000 CREP 0.744 0.744 0.746 0.746 0.629 0.744 1.000
This table presents the correlation matrix calculated for the seven share repurchase measuring methods.
Table 5 shows the correlation matrix of the different share repurchase methods. This table again shows how PREP(A), SREP and RREP(A) are alike. The correlation between PREP and SREP is 0.942 which means that the relation between the two methods is positive and high, PREP and RREP are almost perfectly correlated with a correlation of 0.997. SREP and RREP also have a high correlation of 0.945. The adjustments made to PREP and RREP (indicated by PREPA and RREPA) hardly made any difference when we look at their correlations. The correlation coefficient between PREP and PREPA is 0.999 and RREP and RREPA are almost perfectly correlated with a rounded positive correlation of 1.000.
These high correlations are not that surprising since they can be explained by the way in which PREP, RREP and SREP are calculated. RREP is essentially PREP reduced with sales of common stock. And when there is no value for sales of common stock, RREP results in the same amount as PREP repurchases. SREP is a combination of TREP and RREP and when no value is found for TREP repurchases SREP is equal to RREP.
correlation with the other methods. Table 4 shows correlations of 0.629, 0.681, 0.666 and 0.695 with CREP, PREP, SREP and RREP respectively.
The results in this section show that PREP(A), SREP and RREP(A) have very similar results and follow each other’s movements over time. Whereas CREP and TREP do not. These two methods show quite different average and total results in yearly values and in their correlations with the other methods.
Which share repurchase measuring method can most accurately present share repurchase patterns in the EU?
In order to calculate accurate share repurchase amounts one needs to know how many share are repurchased and at what price. It does not matter where the repurchased shares are used for.
Since CREP repurchases is calculated as the decrease in common shares outstanding multiplied with the average market price of the year in which the repurchases took place it does not reflect true share repurchases. Since the average market price used is probably not the same as the actual price paid for the repurchased shares. So using CREP in share repurchase analysis can lead to under- or overstatements of repurchase amounts and that will distort repurchase patterns.
The TREP method, however, also has a big problem in the European Union. In the EU treasury stock amounts can be valued either at par or market values. This means that TREP share repurchase values can be understated when there are more companies that report at par (which is the case for my dataset), since it is improbable that the shares were bought at par values. This flaw in EU TREP repurchases also distorts SREP repurchases since the SREP method used TREP amounts. For a better measure of TREP and SREP repurchases one needs to make adjustments to TREP repurchase amounts. All par values used in TREP repurchases should be adjusted to reflect market prices. This “eight” method, however, goes beyond the scope of this paper.
PREP(A) and RREP(A) repurchase amounts, on the other hand, are taken directly from a company’s cash flow statement and are therefore valued at the exact prices paid for the repurchased shares. These four methods are thus more suitable for analyzing share repurchase patterns in the EU. But since PREP(A) amounts do not take into account sales of common shares in contrary to RREP(A) amounts I recommend to use RREP amounts in analyzing share repurchases in the EU. RREPA amounts are also suitable but the adjustment made to RREP to derive RREPA amounts hardly makes any difference in the studied variables.
VI. Lintner regressions
Besides the differences in levels, number of repurchasing firms, series statistics and correlations we may get a further understanding of the differences between the methods by using the Lintner regression model given in section IV. The model is used to calculate the sensitivity to net income, the speed of adjustment and the target payout ratios for each method. I measure the parameters of the Lintner regression still for all seven measuring methods used, though knowing that some methods may be more applicable than others.
In the regression analysis the share repurchase variables are adjusted to only include the companies that pay cash dividends for at least for 5 years in the 10-year period and repurchase shares at least three times or more from 1997 to 2006. These criteria result in the following number of included firms for the different repurchase variables; 511 companies are included in the CREP repurchase amount, 397 in the PREP and PREPA repurchases, 594 in the SREP variable, respectively 338 and 323 firms in RREP and RREPA amounts and 394 companies in TREP share repurchases. We call these companies the “frequent repurchasers”.
Table 6 shows the results of the Lintner regression analysis for these frequent repurchasers. I performed the regression for two subsequent periods using pooled annual data. The first period is a four year period ranging from 1998 to 2001 and the second period ranges from 2002 to 2006. C represents the coefficient of the explanatory variable in the Lintner regression model. The probability of the t-statistic is represented by P. The independent variable for this regression is the change in the share repurchases amounts for the different methods. NI stands for Net Income and the lagged values of the share repurchase variables are denoted with an L. These are the independent variables of the model. N is the number of observations.
The first period is a four year period ranging from 1998 to 2001 and the second period ranges from 2002 to 2006. C represents the coefficient of the explanatory variable in the Lintner regression model. The probability of the t-statistic is represented by P. The independent variable for this regression is the change in the share repurchases amounts for the different methods. NI stands for Net Income and the lagged values of the share repurchase variables are denoted with an L. These are the independent variables of the model. N is the number of observations (note that one frequent repurchasing firm may occur more often in the regression analysis).
Table 6. Lintner Regression Results, for two periods measured only for firms that pay cash dividends for at least for 5 years in the 10-year period and repurchase shares at least three times or more from 1997 to 2006.:
C P C P
Panel A 1998 - 2001 2002-2006
Intercept 4.121 0.514 Intercept -8.227 0.174
NI 0.119 0.000 NI 0.142 0.000
LPREP -0.622 0.000 LPREP -0.271 0.000
Adj R-squared 0.302 Adj R-squared 0.220
N 1354 N 1742
Panel B
Intercept 4.166 0.509 Intercept -8.148 0.177
NI 0.119 0.000 NI 0.141 0.000
LPREPA -0.622 0.000 LPREPA -0.270 0.000
Adj R-squared 0.302 Adj R-squared 0.220
N 1354 N 1742
Panel C
Intercept 2.021 0.013 Intercept -8.731 0.216
NI 0.064 0.000 NI 0.137 0.000
LRREP -0.596 0.000 LRREP -0.279 0.000
Adj R-squared 0.184 Adj R-squared 0.288
N 1172 N 1492
Panel D
Intercept 1.963 0.019 Intercept -8.807 0.221
NI 0.075 0.000 NI 0.150 0.000
LRREPA -0.601 0.000 LRREPA -0.295 0.000
Adj R-squared 0.187 Adj R-squared 0.264
N 1131 N 1429
Panel E
Intercept -1.971 0.656 Intercept 7.607 0.003
NI 0.143 0.000 NI 0.047 0.000
LTREP -0.832 0.000 LTREP -0.919 0.000
Adj R-squared 0.344 Adj R-squared 0.682
N 1139 N 1608
Panel F
Intercept 3.059 0.479 Intercept -0.601 0.885
NI 0.144 0.000 NI 0.108 0.000
LSREP -0.805 0.000 LSREP -0.385 0.000
Adj R-squared 0.402 Adj R-squared 0.216
N 1816 N 2453
Panel G
Intercept -0.2856 0.953 Intercept -1.249 0.844
NI 0.116 0.000 NI 0.284 0.000
LCREP -0.420 0.000 LCREP -0.910 0.000
Adj R-squared 0.188 Adj R-squared 0.508
Table 6 shows that for PREP repurchases (Panel A) the effects of net income on repurchases increases between the two sub-periods. In the first period, 1998 – 2001, coefficient C for net income is 0.119 in the second period this has increased to 0.142. This means that PREP repurchases became more responsive to net income. The speed of adjustment of PREP repurchases to net income decreases over the two sub-periods. In the period of 1998 to 2001 the speed of adjustment was 0.622, this declined to 0.271 in the period 2002 – 2006. This suggests that the companies in our dataset are repurchasing shares not immediately after net income is earned and that it takes even longer for the firms to do so in the second period. For PREPA repurchases (Panel B) the Lintner regression model gives almost entirely the same results.
RREP repurchases (Panel C) also shows an increasing effect of net income on share repurchases. In Table 6 coefficient C for net income is 0.064 in the first 4-year period, which increases to 0.141 for the 2002 – 2006 period. So as with PREP repurchases RREP also gets more responsive to net income. The speed of adjustment, between the two sub-periods, decreases from 0.596 to 0.279. The adjusted RREP variable RREPA (Panel D) shows similar results. RREPA responsiveness to earnings also increased between the two periods (From 0.075 from 1998 – 2001 to 0.150 in 2002 – 2006) and it’s speed of adjustment decreased (0.601 in the first period to 0.295 in the second).
The Treasury Stock method (Panel E) results in a decreasing responsiveness to net income. In the period 1998 – 2001 the responsiveness was 0.143 which declined to 0.047 in the period 2002 – 2006. TREP repurchases thus became less responsive to net income, this is in contrast to the result found by using PREP(A) and RREP(A) repurchases. The speed of adjustment increased between the sub-periods. In the first period the speed of adjustment was 0.832 and in the second period it was 0.919. Because TREP has a declining responsiveness to net income whereas RREP has an increasing responsiveness there are implications for the SREP regression analysis.
The speed of adjustment of SREP was 0.805 in the first period and declined to 0.385 in the second.
The CREP share repurchase variable (Panel G) shows an increasing responsiveness coefficient, namely 0.116 in the first period and 0.284 in the second. The speed of adjustment is increasing for CREP repurchases (For the period 1998 – 2001 it was 0.420 and for the period 2002 – 2006 it was 0.910). This suggests that firms will repurchase shares quicker after net income is earned.
Table 7. Target Payout ratios per method for two periods:
Period Measuring Method 1998 - 2001 2002 - 2006 PREP Value 0.191 0.526 PREPA Value 0.191 0.522 RREP Value 0.107 0.491 RREPA Value 0.125 0.509 TREP Value 0.172 0.051 SREP Value 0.180 0.281 CREP Value 0.276 0.312
This table shows the target payout ratios of all seven share repurchase measuring methods for two periods. The first period is a four year period and ranges from 1998 – 2001, the second period, a five year period, ranges from 2002 – 2006.
The target payout ratios are also interesting to look at. Table 7 shows the target payout ratios for the two subsequent periods for all methods. CREP repurchases result in the highest target payout ratio in the first period. However in the second period PREP repurchases have the highest ratio. PREP, RREP and the adjusted versions of the two methods show the biggest increases. SREP and CREP show much smaller increases and the target payout ratio using TREP repurchases even declined. This means that when PREP and RREP repurchase methods are used in analysis the target payout ratios will be much higher than with the other methods. So a company that wants to signal an increase in its target payout ratio could just calculate their repurchase variable with a different method to get the wanted result and vice versa.
Table 8. The sign of the change in the Responsiveness to net income, Adjustment speed and Target payout ratio between periods:
Measuring Method Responsiveness to net
income Adjustment speed Target payout ratio
PREP Value + - + PREPA Value + - + RREP Value + - + RREPA Value + - + TREP Value - + -SREP Value - - + CREP Value + + +
This table presents the development over time between two periods. The first period ranges from 1998 – 2001 and the second from 2002 – 2006. A + resembles an increase of the variable between the two periods and a – resembles a decrease of the variable between the two periods.
Table 8 shows that for SREP and TREP repurchases the responsiveness to net income decreases over time whereas it increases for all other methods. The speed of adjustment decreases for almost every method except for TREP and CREP repurchases. This means that PREP, SREP and RREP repurchases are becoming more sticky over time, whereas TREP and CREP repurchases react quicker over time.
The share repurchase amounts used in Von Eije and Megginson (2008), the so called PREP repurchases, are responding better to net income, over time, than those used in Skinner (2008) (SREP repurchases). In the first period SREP responsiveness was higher than the responsiveness of PREP repurchases but this changes in the second period. In the second period, 2001 – 2006, the responsiveness to net income for PREP repurchases was higher (0.142 for PREP repurchases and 0.108 for the SREP method).
CREP repurchases is the only method that results in increasing values for all three variable’s in Table 8.
The most different method is the TREP measuring method as it is the only method that has a declining target payout ratio over time. For the TREP method this difference may be explained by the fact that in the EU treasury stock may be reported at par or at market values. This means that in the calculation of TREP repurchases some share repurchase amounts are valued at par and others at market values. This leads to distortions in the total TREP values and in underestimations of the repurchase amounts (and thus also in SREP values, since SREP repurchases use TREP amounts).
VII. Summary and conclusions:
I used data of 5654 listed companies in the EU 15 countries (the countries that were a member of the European Union before May 2004) during the period of 1997 – 2006 in order to analyse what the target payout ratios and adjustment speeds were for share repurchases in the EU. To do so I identify seven share repurchase measuring methods; the method introduced by von Eije and Megginson (2008) PREP repurchases, the adjusted PREP repurchases PREPA, the Retirement Method RREP, the adjusted RREP repurchases RREPA, the Treasury Stock Method TREP, the method as used in Skinner (2008) SREP repurchases and a method based on Shoven (1986) CREP repurchases. Then, I perform Linter regressions on those companies that both frequently pay cash dividends and frequently repurchase shares (frequent repurchasers) in order to analyze the responsiveness to net income, the adjustment speed and the target payout ratios for the different repurchasing methods.
Six out of the seven measuring methods show increasing target payout ratios and five out of the seven show decreasing adjustment speeds in the EU for the period 1997 – 2006. In the European Union I recommend using the retirement method RREP to measure share repurchase amounts and to analyze share repurchase patterns. This method concluded that for the period 1998 – 2006 the target payout ratio of companies increased (from 19.7% in the period 1998 – 2001 to 49.1% from 2002 – 2006) ). At the same time the speed of adjustment decreased from 59.6% in 1998 – 2001 to 27.9% in 2002 – 2006. The responsiveness to net income also increased implying a growing relation between net income and share repurchases. This is in line with the results found by Skinner (2008) for the US and with the findings of Von Eije and Megginson (2008) for the EU.
References
Bagwell, L. and J. Shoven, 1989. Cash distributions to shareholders. Journal of
Economic Perspectives 3, 129-149.
Bagwell, Laurie Simon, 1992, “Dutch Auction Repurchases: An Analysis of Shareholder Heterogeneity,” Journal of Finance, 47 (No. 1, March), 71-106.
Brav, Alon, John R. Graham, Campbell R. Harvey, and Roni Michaely. 2005. Payout policy in the 21st century. Journal of Financial Economics 77, 483-527.
DeAngelo, Harry and Linda DeAngelo. 2006a.The irrelevance of the MM dividend irrelevance theorem, Journal of Financial Economics 79, 1415-1431.
Fama, Eugene F. and Kenneth R. French. 2001. Disappearing dividends: Changing firm characteristics or lower propensity to pay? Journal of Financial Economics 60, 3-43.
Grullon, Gustavo, and Roni Michaely. 2002. Dividends, share repurchases, and the substitution hypothesis. Journal of Finance 57: 1649-1684.
Hackethal, Andreas and Alexandre Zdantchouk. 2006. Signaling power of open market share repurchases in Germany. Published online 24 May 2006.
Jagannathan, Murali, Clifford Stephens and Michael Weisbach. 2000. Financial flexibility and the choice between dividends and stock repurchases. Journal of Financial Economics 57, 355-384.
Jensen, Michael C., 1986, “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers,” American Economic Review, 76, 323-329.
Lintner, John. 1956. Distribution of incomes of corporations among dividends, retained earnings, and taxes, American Economic Review 46, 97-113.
Miller, M.,Modigliani, F., 1961. Dividend policy, growth, and the valuation of shares. Journal of
Business 34, 411–433.
Modigliani, F. and Miller, M. (1958). "The Cost of Capital, Corporation Finance and the Theory of Investment". American Economic Review 48 (3): 261–297
Shoven, John B., "The Tax Consequences of Share Repurchases and Other Non-Dividend Cash Payments to Equity Owners." In Summers, Lawrence, ed. Tax policy and the Economy Vol. I. Cambridge, MA: NBER and MIT Press, 1986, pp. 29-54.
Skinner, Douglas J. 2007. The evolving relation between earnings, dividends, and stock repurchases. Journal of Financial Economics 87, 582-609.
Von Eije, J. Henk and Megginson, William L., Flexibility of Dividend Policies and Shareholders' Returns in the European Union (2009). Available at SSRN: http://ssrn.com/abstract=1342671
Appendix A:
Which share repurchase measuring method results in an accurate measure of non-cash dividends in the European context?:
First I will look at the method used by Von Eije and Megginson (2008) the so called PREP repurchases and the adjusted PREP repurchase variable PREPA. These methods both take purchases of common share (POC) amounts from the company’s cash flow statement and use those amounts as substitutes for cash dividends. These methods however do not take into account any reissues or sales of common share. For example a company that repurchased shares could decide to sell all the repurchased share in the same year they were bought. The company’s sales of common stock would offset the purchases and lead to zero shares held to act as non-cash dividends. The PREP(A) method, however, does not recognize this sale and will then overstate yearly share repurchases. And it will thus also overstate the amount of share repurchases that are used as substitutes for cash dividends. This can be seen in Figure 1 where PREP(A) has the second highest share repurchase amounts from 1997 – 2005 and the highest total value in 2006. So in order to calculate true share repurchases used as non-cash dividends it is wise not to use one of these two methods.
RREP and RREPA repurchases on the other hand do correct for sales of common shares and could therefore be seen as a better measure of non-cash dividends. However according to Fama and French (2001) and Skinner (2008) RREP(A) repurchases do not correct for repurchased stocks that are used for employee stock ownership plans (ESOP’s) or repurchased stocks to finance a merger. And therefore do not measure non-cash dividends accurately.
actual paid price and the average market price is substantial. This might explain the volatile path CREP repurchases follows in Figure 1. The number of shares repurchased as shown in Figure 3 shows a large spike in 2003 but this spike isn’t not followed by a large increase in the total value of CREP share repurchases in that year, which might be caused by an underestimation of the price paid for the repurchased shares. The large increases in repurchased shares in 2000 and 2005 that resulted in high share repurchase amounts could in turn be caused by overestimations of the share prices paid.
Another reason why the CREP method may be flawed is the same as with RREP(A) repurchases and that is that CREP also does not correct for stocks used in ESOP’s or to finance a merger. This problem, however, can be taken away by using the method introduced by Fama and French (2001) and Skinner (2008) the SREP method.
The SREP method, as mentioned earlier, is a combination of the Treasury Stock Method (TREP) and the Retirement Method (RREP), using RREP amounts only when no TREP amounts are available. Skinner (2008) prefers TREP repurchases over RREP repurchases because balance sheet treasury stock amounts do net out stocks reissued for employee stock ownership plans (ESOP’s) and stocks used to finance a merger. This means that TREP repurchase amounts can be seen as a better measure of non-cash dividends then PREP(A), RREP(A) and CREP repurchases.
SREP will, therefore, be a more accurate measure of non-cash dividends when there are relatively more TREP values used in its calculations than RREP values. However, from 1997 – 2006 SREP repurchases is actually getting less accurate due to the fact that relatively more RREP values are used over time. Table 9 shows the number of RREP repurchasing firms that are used in SREP calculations as well as RREP firms as a percentage of total SREP firms from 1997 – 2006.
Table 9. Number of RREP firms used in SREP calculations from 1997 -2006: YEAR 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 SREP Firms 244 277 357 535 597 562 600 480 555 551 TREP Firms 191 159 240 373 409 380 375 298 341 309 RREP Firms 53 118 117 162 188 182 225 182 214 242 RREP/SREP 21.721% 42.599% 32.773% 30.280% 31.491% 32.384% 37.500% 37.917% 38.559% 43.920% This table presents the number of TREP and RREP firms used in SREP calculations for the period 1997 –
One can see that, over time, firms that repurchase using the retirement method (RREP) are becoming a relatively larger part of total SREP firms. This results in SREP repurchases becoming a less accurate measure of non-cash dividends, from 1997 to 2006, since TREP repurchase amounts are preferred over RREP repurchases.
Nonetheless SREP is still the most accurate measure of non-cash dividends due to the fact that, in comparison to the other methods, TREP is the only method that corrects for stocks used for ESOP’s and for stocks used to finance mergers and RREP corrects for sales of common stock.
However SREP repurchases are less accurate in the European context then in the US. In the US TREP and RREP amounts must be valued at market values. These share repurchase amounts therefore correspond with the exact amounts paid for the repurchased share. In the EU firms are allowed to report treasury stock on the balance sheet at par or at market values. This could lead to underestimations of TREP amounts in the EU (and also in SREP amounts) when not all amounts are reported at market values. As explained, this is not a problem in the US due to its accounting standards. For the EU, however, adjustments need to be made in order to value TREP repurchase amounts at market value7.
One way of adjusting TREP repurchases is to apply an adjustment ratio that increases the repurchase amount for individual firms to better reflect market prices. Since RREP share repurchase amounts are at market value one could multiply TREP repurchases by the RREP/TREP ratio when RREP/TREP > 1 (if <1, I assume that TREP amounts are already at market prices). When RREP repurchase amounts are higher than the TREP repurchases, TREP repurchases will be increased by the ratio. This ratio can only be calculated for firms who have values for both RREP and TREP amounts in a given year. Table 10 gives the number of firms that report RREP as well as TREP amounts during the 10-year period 1997 – 2006.
This table gives the number of firms that report repurchases for both the TREP and RREP share repurchase measuring methods for the period 1997 – 2006.
Table 10 shows that not many firms report values for both methods which limits the usability of the ratio, since it is impossible to determine an accurate adjustment ratio for individual firms with no TREP and RREP values. Therefore one could calculate the average RREP/TREP ratio, per year, for the firms that do report both values and multiply that ratio with total yearly TREP repurchase amounts. This however will also increase the amounts already valued at market prices, which will lead to overestimations of TREP share repurchase amounts. Another disadvantage of this adjustment ratio is that companies that do have values for both TREP and RREP repurchases are often larger, older and renown firms. For these firms the difference between the stocks’ par value and market value is more likely to be larger than for younger and smaller firms. This will lead to a relatively higher average adjustment ratios and, again, to overestimations of TREP share repurchase amounts.
It is clear that it is hard to determine correct adjustment ratios for the individual firms and for total yearly TREP repurchase amounts. Therefore adjusting TREP repurchases by the mentioned ratio’s will probably not make TREP amounts a more accurate measure of non-cash dividends in the EU.
A different way of adjusting the Treasury Stock Method (TREP) is to identify the firms that report their treasury stock at par using financial statements and, assuming that managers are not able to outperform the market, multiply the number of shares these firms repurchased with the average market price to adjust them to market prices. However, when there are relatively many firms that report at par values, which is probably the case for our dataset, it will be time consuming to make this adjustment.
An alternative is not to look for the companies that report at par but to look for the companies that do not report at market values. When treasury stock is valued at stock prices lower than the yearly average stock price one could increase the TREP repurchase amounts by valuing them at average stock prices. If the treasury stock is valued at stock
Table 10. Number of firms with values for both TREP and RREP repurchases from 1997 -2006:
YEAR 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
prices higher than the average share price then no correction is needed. This can be done without knowing if the treasury stock amounts are listed at par or at market prices. The resulting TREP repurchase amounts will then at least be valued at average stock prices and otherwise at higher prices. One problem with this adjustment is the fact that it also increases TREP repurchases valued at prices lower than average market prices but higher than par.
