The Market Valuation of Share Repurchases in Europe

Transcription

1 The Market Valuation of Share Repurchases in Europe M. Ameziane LASFER * City University Business School, London, UK. Abstract This paper investigates the share price behaviour around the announcement of intention to repurchase ordinary shares by UK and other European firms over the period 985 to 998. Consistent with the US evidence, the results show that in the UK share repurchases are popular, the market reacts positively on the announcement date and share prices continue to increase 6 months after the announcement of the intention to repurchase. In contrast, in other European countries the repurchasing activity is recent but not common, the announcement date abnormal returns are insignificant and the post-announcement returns are negative. Information asymmetry and institutional factors explain a proportion of the differences in the abnormal performance. JEL classification: G; G5 Key words: Share repurchases; Signalling, information asymmetry, Corporate governance systems Correspondence to: City University Business School, Barbican Centre, London ECY 8HB, UK. Tel , Fax , m.a.lasfer@city.ac.uk. * I am grateful to Adri De Rider, Myron Slovin, Marie Sushka, Mike Vetsuypens and seminar participants at the 000 European Financial Management Association meeting in Athens, 999 Financial Management Association meeting in Orlando, Catholica University (Milan), Bologna University, Durham University, Tilburg University, Group HEC Paris and City University Business School for valuable insights and helpful comments. I would like to thank Michel Dubois, Mara Faccio, Edith Ginglinger and Chris Veld for sending me details on the treatment of share repurchases in Switzerland, Italy, France and the Netherlands, respectively. I have benefited from discussions with Murizio Murgia on the Italian system. I am grateful to Aslihan Ersoy-Bozcuk for providing me with part of the macro to undertake this research. All errors are my responsibility.

2 The Market Valuation of Share Repurchases in Europe Abstract This paper investigates the share price behaviour around the announcement of intention to repurchase ordinary shares by UK and other European firms over the period 985 to 998. Consistent with the US evidence, the results show that in the UK share repurchases are popular, the market reacts positively on the announcement date and share prices continue to increase 6 months after the announcement of the intention to repurchase. In contrast, in other European countries the repurchasing activity is recent but not common, the announcement date abnormal returns are insignificant and the post-announcement returns are negative. Information asymmetry and institutional factors explain a proportion of the differences in the abnormal performance. JEL classification: G; G5 Key words: Share repurchases; Signalling, information asymmetry; Corporate governance systems.

3 Share repurchases are increasingly becoming an important method for paying out cash to shareholders. In the US, for example, Jagannathan, Stephens and Weisbach (000) report that, over the last two decades, the number and value of repurchase programme announcements grew at a compounded rate of % per year from 5 ($5. billion) in 985 to 755 ($5 billion) in 996. Over the same period, cash dividends, although larger, have only doubled from $67.6 billion to.7 billion. About 80% of these announced programmes are executed in the open market and the actual repurchase increased from $8.8 billion in 985 to $6 billion in 996. In Europe, on the other hand, share repurchases are not so common mainly because of institutional and cultural constraints. For example, in Austria, the legal system allows companies to repurchase their shares only to fulfill their obligations under the stock option plan for employees. In Sweden, companies have to go through a lengthy process to get a court approval. In the Netherlands companies have to pay ⅓% withholding tax on the difference between the repurchasing price and the average paid up capital ( gemiddeld gestort kapitaal ) per share. In the UK, while they were made legal in 98, share repurchases are only becoming popular in recent years because of the ambiguous tax treatment and fear of signalling a lack of profitable investment opportunities. However, despite these constraints, a number of European companies have recently announced their intention to repurchase their shares. For example, in 997 repurchases proposals announced by firms in Europe amounted to $7.bn compared to $.bn in 996. This increase is, however, driven mainly by programmes made by such large firms as Diageo, BT, Reuters, ELF Aquitaine and TeleDenmark worth more than $bn each (Financial Times January 998). In addition, British companies are by far the biggest buyers of their own shares with nearly 80% of the total repurchases. A number of studies document the reasons for share repurchases. The primarily US-based empirical studies show that signalling is the most prevalent explanation for open market repurchases (e.g., Comment and Jarrell, 99; D Mello and Shroff, 000; Ikenberry, Lakonishok and Vermaelen, 995; Vermaelen, 98). These studies document that, on the announcement date of the intention to repurchase shares, share

4 prices increase significantly and the pre-announcement cumulative abnormal returns are negative suggesting that the repurchasing companies are undervalued. This signalling hypothesis is based on the premise that managers are better informed and they launch a repurchasing programme to signal to the market that their shares are undervalued. Survey studies also provide support for this hypothesis, as they show that managers repurchase their shares because they are undervalued and represent a good investment (Baker, Gallagher and Morgan, 98; Wansley, Lane and Sarkar, 989). This signalling hypothesis is, however, controversial because open market repurchases are not costly signals, as they do not commit the firm to actually buy back shares. Moreover, there is a distinction between signalling management s expectations of future increases in the firm s cash flow and earnings, and signalling disagreement with the current market valuation of the firm s performance. Grullon and Ikenberry (000) argue that while these two views are consistent with undervaluation, the former suggests that the firm is unable to communicate to the market its future prospects, while the later implies that market is inefficient because current prices do not reflect publicly available information. This later view is also supported by the findings that the markets under-react. For example, Ikenberry et al (995) analyse a large sample of US open market repurchase programme announcements in the 980s. They report significant positive abnormal returns of % in the four-year period following stock repurchases. This suggests that the markets fail to impound information in stock prices quickly. This abnormal performance is most prevalent in firms where under-valuation is likely to be a more predominant factor, i.e., high book-to-market firms (value stocks) which generate 5%, compared to.% for low book-to-market firms (glamour stocks). They suggest that the market is under-reacting because it treats repurchase announcements with skepticism leading prices to adjust slowly over time. Similar results are found in Canada where the three-year holding period cumulative abnormal returns amount to.%, with value stocks generating 7% and growth stocks % (Ikenberry, Lakonishok and Vermaelen, 000) and for the case of tender offer repurchases over shorter event windows (e.g., Dann, 98; Vermaelen, 98). 5

5 The purpose of this paper is to test the signalling hypothesis under different institutional settings and to highlight the extent to which institutional differences affect the short and the medium-term (up to six months) market reaction to share repurchases. Using data from all European countries, the paper contrasts the market reaction to the announcement of share repurchases in UK and in other European countries (referred to thereafter as Europe), and documents the impact of institutional factors on the short- and medium-term stock price performance. The sample is, first, split into market-based and bank-based systems following La Porta et al. (997, 998) classification. The UK is, like the US, a market-oriented country with common law that offers better protection to minority shareholders but characterized by liquid markets and unconcentrated company ownership, resulting in high information asymmetries. In contrast, other European countries are so-called bank-oriented and civil law countries where the pressure from shareholders to pay out cash is low as ownership is concentrated and markets are relatively illiquid, resulting in low information asymmetries. Therefore, if the market reaction to share repurchases is driven by signalling we would expect higher abnormal returns on the announcement date in the UK compared to other European countries. We also analyse the cross-section market reaction of firms within UK and other European countries. We expect firms with high information asymmetries to generate high abnormal returns on the announcement dates. Further research in this area is warranted for a number of reasons. First, the evidence from this analysis will highlight the extent to which European markets are homogeneous in their behaviour to facilitate the potential integration of European markets. Second, as Fama and French (998) argue, it is useful to investigate whether patterns in security returns documented in the United States hold in other countries to ensure that such patterns are not the outcome of data mining. Finally, the cross-country analysis will shed some light on the extent to which the behaviour of share prices when firms announce their intention to repurchase shares are affected by institutional factors. The analysis of the 6 announcements of share repurchases reported in Reuters Business Briefings and Extel Financial over the period reveals that the vast majority of these announcements (65 representing 7% of the total) are made in the 5

6 UK. In addition, about 59% of such announcements occurred in 997 and 998. In the pre-996 period, very few share-repurchasing programmes are announced outside the UK. The results also show that on the announcement date, share prices increase by.6% in UK and.06% in Europe. This abnormal performance is statistically significant but lower than the % documented in the US (e.g., Stephens and Weisbach, 998). In the UK, this abnormal performance extends 5 days after the announcement. For example, over the period + to +5 days after the announcement date, a buyand-hold strategy will result in an abnormal performance of.5% (t=.98), equivalent to 8.% per year. This abnormal performance is higher than the.9% annual return in the US (Ikenberry et al 995) and 7.08% in Canada (Ikenberry et al 000). In contrast, in other European countries, the event date abnormal returns are not strong and the post-announcement period abnormal performance is negative. Over the [+ to +5] period, share prices decrease by 5.9% (t = -.66). The analysis of individual European country reveals that in France and Italy, where share repurchases announcements are relatively more common, the announcement-period abnormal returns are not statistically significant. Over the postevent period [+ to +5] share prices decrease by.9% in France, though not significant, and by 6.5% in Italy (t = -.80) In other European countries, share prices decrease by.0% (t = -0.87). The same results are obtained when the announcements are split into common and civil law countries and into countries with high and low investor protection. In particular, in countries with high investor protection share prices increase by.80% in the post-announcement period while in countries with low investor protection share prices decrease by 6.0% (t-statistic of difference of.5 is significant at 0.0 level). The regression results indicate that this abnormal performance is also a function of the repurchasing method, a dummy for confounding announcements, the legal system of the country in which the announcing firm is operating and the forecast error which proxy for information asymmetry. The rest of the paper is organized as follows. Section I presents the data and the methodology. Section II discusses the results. Section III presents the conclusions. I. Data and Methodology 6

7 The sample is constructed by identifying all announcements of intention to repurchase ordinary shares reported in the Reuters Business Briefings (RBB) from January 985 through December 998. This database reports any news made by UK and European companies on share repurchases. These files are screened to isolate announcements of intention to repurchase ordinary shares only. This database is then complemented with Extel Financial, which reports news announcements made by companies. To be included, companies are required to have their share prices on Extel Financial or Datastream (an online financial database). Unlike previous US studies (e.g. Ikenberry et al, 995), the 987 period is not excluded as there was only one announcement. Table I, Panel A., reports the annual distribution of the announcements to repurchase shares. The last column shows that out off 6 announcements, 65 (7.%) are made by UK companies. France and Italy follow the UK with 5 (7.9%) and (6.8%), respectively. The relatively low level of share repurchasing activity in Europe is due to the stringent legislation or to the proposition that share repurchases do not fit the European corporate culture (Rau and Vermaelen, 00). The time-series distribution of the announcements is also disproportionate. The vast majority of repurchases (59.%) were announced in 997 and 998. Over the period, there were only 7 announcements, with in the Netherlands and in Italy in 989. The remaining 5 are in the UK. 6 In 990, UK firms made 9 announcements out of. The UK repurchasing programmes represent nearly twice the announcements made in the period. The announcements decreased to in 99 and 99 and to 9 in 99, but increased to 9 in 99. In 995/96 they stabilised at 65 but carried on increasing substantially in 997 and 998. UK programmes dominate the activity, increasing from 7 in 99 to 6 in 995 but decreased to 5 in 996. The large increases occurred in 997 and 998 with and 9 announced programmes. Table I, Panel B, reports the industry classification of the sample firms. The last column indicates that 7 (5%) announcements are made by companies from the Financial and General Manufacturing industries. These industries are followed by the services sector with about 8 (8%), utilities with 85 (%) and consumer goods with 7

8 7 (%) announcements. There are only (%) announcements made by companies from the Mining Extraction industry. This aggregate industrial distribution is similar to that observed in UK, Italy, and to a lesser extent Denmark. In other European countries, such as France, Finland and Sweden, most announcements are made by companies in the general manufacturing sector. [Insert Table I here] To assess the market reaction to the announcement of intention to repurchase shares, I compute the abnormal returns using the market model. The estimation period spans from day 90 to day relative to the announcement date. The coefficients of the market model are computed by running a regression of the each firm s raw returns against each firm s country main market index. The results are replicated using the market-adjusted returns, where a is set equal to zero and ß to one, Dimson (979) and Scholes and Williams (977) methodologies, and the size and book-to-market matched control firms. The results are not sensitive to the methodology used. The reported results are based on the straightforward market model. The use of event study methodology in this context is likely to be subject to three main biases. The first relates to inferences with event date uncertainty. Given that the data is collected from financial publications as reported by Reuters Business Briefings, it is difficult to identify the exact event date. To overcome this problem, I follow Lakonishok et al (995) and report results based on expanded event period [- to +]. 7 The second relates to the assumption that the returns are normal. Brown and Warner (985) show that the degree of non-normality in daily NYSE/ASE abnormal returns poses no serious problems for correct test statistic specification in such samples. However, Campbell and Wasley (99) document a substantial non-normality in the daily returns of NASDAQ securities. I report also the sign test and the median, which are free of specific assumptions concerning the distribution of abnormal returns. The third potential problem relates to inferences with clustering. The standard event study methodology assumes that the abnormal returns on individual securities are uncorrelated in the cross-section. This is a reasonable assumption if the event windows of the included securities are not overlapping in calendar time. In this case we can 8

9 calculate the variance of the aggregated sample CARs without concern about covariances between individual sample CARs, since they are zero. 8 However, given that, as reported in Table I, more than 59% of the events occur in 997 and 998, the clustering bias is likely to affect the results. To overcome this problem the t-statistics are computed as the sum of the standardized prediction errors divided by the square root of the number of sample firms, as suggested by Brown and Warner (985). The abnormal returns are computed over the 0 to +5 days relative to the announcement date. Data limitation constrained the post-announcement period and the use of unbiased methodologies such as these suggested by Barber and Lyon (997). Kothari and Warner (997) show that the misspecification bias induced by the use of the market model in testing long-term returns is less pronounced when the cumulation period is less than months. Nevertheless, the simulated results using the returns of size and book-to-market matched control firms as a benchmark are also reported. II. Empirical Results A. Announcement period abnormal returns Table II reports the distribution of the percentage cumulative abnormal returns over the event period and tests for their significance. Panel A. shows the aggregate results. Over the pre-event period [-0 to ] the cumulative abnormal returns are not significant and the distribution of positive and negative abnormal returns is even. The cumulative abnormal returns over the 0 to period (not reported) of 0. is negative but not statistically significant. These results do not provide support for the Ikenberry et al (995) findings in the US where the abnormal returns over the [-0 to ] period amount to.07% (t = -9.9), suggesting that companies time their announcements to repurchase shares. Over the 5 day-period (- to +), the cumulative abnormal returns amount to.8% for the whole sample,.6% in the UK and.06% for the rest of Europe. These abnormal returns are statistically significant but they are lower than the.5% found in the US market (e.g., Ikenberry et al, 995, Stephens and Weisbach, 998). The difference in means and in medians between the UK and Europe is not statistically significant. 9 9

10 Table II also reports the behaviour of share prices in the post-event periods. Over the [+ to +0] period share prices for the sample as a whole increased by 0.99% (t =.75). In the UK prices increase by.% (t =.). In contrast, in Europe the increase in prices by 0.6% is not significant (t = 0.70). However, the difference in means and medians between UK and Europe is not statistically significant. Over the [+ to +5] post-event period, share prices in UK continue to increase by.5% (t =.98) while those in Europe decrease by 5.9% (t = -.56). In terms of medians, share prices increased by.70% in the UK and decreased by.97% in Europe. The t-statistic of the difference in means between the two CARs of.8 is significant at 0.0 level. 0 The p-value of the Mann-Whitney statistics indicates also that the differences in medians between the two samples are significant. Consistent with the US evidence, these results suggest that, in the UK, the market is under-reacting to the announcement of share repurchases. Figure charts the cumulative abnormal returns in UK and other European countries over the [-0 to +5] period. [Insert Figure here] Table II, Panel B, reports the simulated results using the returns on size and book-to-market matched control firms as a benchmark. The results mimic those reported in Panel A. In the pre-event period the abnormal returns are negative but not significant. On the event period [-, +] the average abnormal returns of.% are significant at 0.0 level. This abnormal performance is mainly driven by UK companies which generate.7% (t = 6.0). The average abnormal returns of other European companies of 0.97% are not significant. The last row indicates that the mean and median abnormal returns of firms in UK are significant higher than those in continental Europe (t of differences in means of.9 and Mann Whitney p-value for differences in medians of 0.06). Over the [+, +0] period, only UK firms extent their positive abnormal performance by 0.8%. Finally, in the [+, +5] period, firms in the UK carry on increasing significantly by.0% while those in continental Europe decrease by.%. Thus, the results documented in Panel A. do not appear to be sensitive to the choice of event study methodology. In the remaining sections, the results are based on the market model. 0

11 Table II, Panel C, provides details of the abnormal returns of two individual European countries with sufficient data points. In France where 5 announcements are made, the abnormal performance on the announcement period, - to +, amounts to 0.78%, not statistically significant at any confidence level. Similarly, in Italy the abnormal returns of 0.6% are not statistically significant. The other remaining European countries generate abnormal returns of.5% (t =.9). In the post event period share prices in France continue to decline steadily. Over the [+ to +5] post-event period share prices decrease by.9% (t = -.7). In Italy, share prices decrease by 6.5 (t = -.80) despite the increase in the shares prices of.98% (t =.) over the [+ to +0] period. In these two countries the median abnormal returns are also negative and the proportion of positive abnormal returns amounts to only 7% and 9%. In the remaining European countries the postannouncement cumulative abnormal returns are negative but not significant. Panel D, Table II, tests for the impact of event period clustering in the UK by comparing the abnormal returns in different sample periods. Over the event period [- to +] the average abnormal returns are all positive and significant except in the subperiod where the abnormal returns of 0.8% are not significant. Over the [+ to +5] period, the cumulative abnormal returns are all positive and significant. Panel E, Table II, reports the results for other European countries. The abnormal returns, over the event period [- to +], are not significant. In the post-event period, the abnormal returns are negative but significant only in 998. The differences in means and median abnormal returns, as reported in Panel F, are not significant, except for 997 and 998. Thus, the impact of event date clustering is not likely to have a significant impact on the results. [Insert Table II here] B. Impact of changes in institutional settings Over the sample period, two countries, namely France and UK, have undertaken a set of institutional changes on how companies should repurchase shares and how selling shareholders should be taxed. In France, the Law of July 998 gave greater freedom to companies to repurchase their shares. In the UK, the tax and regulatory

12 environment relating to the treatment of the advanced corporation tax when firms repurchase their shares changed drastically. These two countries are analyzed to assess the extent to which institutional settings affect share-repurchasing activity. [Insert Table III here] B.. The introduction of the simplified buyback procedure in France in July 998 In France, traditionally share repurchases were prohibited on the grounds that they entail an undesirable confusion between the roles of debtors and creditors. Thus companies can buyback their shares only in the framework of a reduction of capital not motivated by losses and the shares acquired have to be cancelled. Over the years, however, various measures were taken which alternated between relaxation of the rules and their further strengthening. For example, the set of measures introduced to implement the Second EC Directive in 98 extended the prohibition to acquisition by subscriptions, acquisitions through third parties, and the acceptance of a firm s own shares as a guarantee or as a guarantee for the third party to acquire the firm s shares. The law has, however, introduced more generous measures such as buying back the shares to distribute to employees in the framework of employee participation schemes or to temporarily support the share price quotation. The Law of July 998 amended these possibilities and gave greater degrees of freedom. First, the law repealed the possibility of buying back shares to support the stock market quotation but introduced the simplified buyback procedure where a company is allowed to buyback and hold up to 0% of its shares without any specific purpose. Second, the law extended the possibility of buying back shares in the framework of reduction of capital not motivated by losses by revoking the prohibition for a company to proceed with buybacks if it had earlier issued exchangeable or convertible bonds or warrants. This law is expected to result in an increase in the number of firms repurchasing their shares but to a reduction in the event date abnormal returns as firms are allowed to buyback their shares for other reasons than signalling. Table III, Panel A, reports the results. Between 990 and -July 998, a total of 9 share repurchases announcements are made, an average of. announcements per year. In the 6 months after the law of July 998 share repurchases reached. Thus, as

13 expected, the Law of July 998 has lead more French companies to announce their intention to repurchase their shares. However, the t-statistics of the differences in means abnormal returns in the pre- and post July 998 are not significant suggesting that the law has not resulted in a change in the market perception of this activity. In the pre-998 period the abnormal returns over the cumulating periods are not significant at any confidence level. In the post-july 998 period, the abnormal returns over [ to +] days of 0.6%, although positive, are not significant (t = 0.). However, in the preevent period [ 0 to ], the cumulative abnormal returns of -.6 are significant (t = -.), suggesting that companies time the announcement of their intention to repurchase their shares. In contrast, in the post-event period + to +0, the abnormal returns decrease by.6% (t = -.66), suggesting that the market is mean reverting. This negative abnormal performance extends to the 5 days after the announcement date with share prices decreasing by 7.% (t = -7.89). These results suggest that, although the July 998 law increased the number of share repurchase announcements, the market does not appear to value this strategy favorably. B.. The UK tax and institutional changes The market perception of share repurchases depends on their tax treatment, i.e., on whether they are treated as dividends or capital gains. This, in turn, is a function of the repurchasing method adopted and on whether the Inland Revenue, the UK tax authority, allows companies to treat share repurchases as dividends. In general, onmarket repurchases are treated as capital gains and off-market repurchases where shareholders sell directly to the company through tender offers and private transactions, or through a broker who is acting as an agent for the company, are treated as dividends. In this case the firm is committed to repurchase the shares, shareholders know that they are selling to the company and the Inland Revenue allows them to claim the tax credit. In the UK this split between on- and off-market is not a clear cut. Before September 99, companies had to repurchase their shares in the open market. These repurchases are considered as on-market. However, exceptionally, companies get a permission from the Inland Revenue to treat part of the share repurchases as dividends. In this case, they pay the tax credit in the form of Advanced Corporation Tax (ACT)

14 which they can deduct from their corporation tax liability and shareholders can claim back from the Inland Revenue. For example, in 99, The Inland Revenue agreed that.6 per share bought back by Reuters at would be treated as capital repayment and therefore subject to capital gains tax. The balance of.8 per share was treated as a distribution i.e., equivalent to a dividend, thereby creating a tax credit of. (5% of.8). Similarly, Boots Plc announced its intention to repurchase shares on 5 November 99. The price paid in excess of 78p per share will constitute distribution for tax purposes and certain shareholders can reclaim part or all of the associated tax credit of 5% equal to 08.75p on basis of the previous day s closing price of 5p. On September 99 BZW invented the agency buyback with Northern Electric s PLC share repurchase programme. This method is viewed as an open-market share repurchase but through a broker-agent. Shareholders know that they are selling indirectly to the company and they are able to claim the tax credit. This feature lead taxexempt institutions, such a pension funds, to have a strong preference for share repurchases. This method was in operation until 7 October 996 when the Inland Revenue abolished this loophole and shareholders are not allowed to claim the tax credit. In July 997, the UK government announced changes in the taxation of dividends distributed to pension funds. As from July 997, pension funds are not allowed to claim the tax credit on dividends they receive, making these investors indifferent between dividends and share repurchases. Table III, Panel B shows the impact of tax changes on the behaviour of share prices around the announcement of the intention to repurchase shares. Over the period 986 to 0 September 99, when share repurchases are taxed as capital gains, there were only a total of 8 announcements, an average of 9 per year. Over this period the announcements were highly scattered. In the months just before the September 99, when BZW invented the agency buybacks method, there was no single announcement. In the following years to 7 October 996, a total of announcements were made, an average of 6 announcement per year. 5 In the following period which spans from 8 October 96 to July 98, the number of announcements decreased to. However, taking account of the time period, the average number of

15 announcements is In between July 997 and December 998, the number of announcements increased to 9, mainly because of the abolition of the advanced corporation tax which removed the penal tax charges on buybacks. This trend in share repurchases is consistent with Rau and Vermaelen (00) findings. In the first case, i.e., in the pre-september 99 period the abnormal returns of.8 over the event period to + are not statistically significant (t = 0.6). Similarly, in the days before and 0 days after the announcement date, the abnormal returns are not significant. However, over the [+ to +5] days, share prices increase significantly by 6.6%. In the second case when companies used agency buybacks, i.e., from September 99 to 7 October 996, share prices increase by.% on the announcement period to + and carried on increasing up to day 0 by.7%. Thus over the period to +0, the cumulative abnormal performance amounts to 5.79%. Thereafter, the abnormal performance of 0.6% is not significant at any confidence level. In both these cases the pre-announcement abnormal returns are not significant, suggesting that companies are not timing their repurchases announcements. In the third case when the tax authorities abolished the loophole (from 8 October 996 to July 997), the abnormal returns over the event period to + of 0.% are not significant. However, in the pre-announcement period [-0 to ] and the post announcement period [+ to +0] share prices decrease by.9% and.6% respectively, suggesting that companies are timing their announcement but that the market penalizes such decision. Over the period [+ to +5], share prices recover to result in an abnormal performance of.0% (t =.). Finally, share prices of companies that announce their intention to repurchase their shares after the decision to abolish the tax credit for pension funds increase by.% (t = 8.0) in the announcement period [- to +]. They carry on increasing by.90% (t = 7.0) over the period [+ to +0] and by 5.9% (t =.85) in the next 0 days to day +5. Thus, a buy-and-hold strategy from day to day +5 will result in a cumulative abnormal performance of.%. In the pre-event period the abnormal performance is not statistically significant. 5

16 Table III, Panel C, reports the t-statistics of the differences in means across the cases. Over the event period [- to +], the abnormal returns in the post-july 997 are statistically different from other cases. This suggests that the abolition of the tax credit has lead the market to value more share repurchases as companies do not incur the potential advanced corporation tax costs and they do not repurchase their shares for the sole purpose of offering tax credits to tax-exempt institutions. In the post-event period [+ to +5] the cumulative abnormal returns in the post-july 997 period are different from the second and third cases. This suggests that the market is under-reacting more to the announcement of repurchases after the abolition of the tax credit. C. Impact of Legal Settings Under the agency setting share repurchases may be used to transfer wealth from existing atomistic shareholders to large shareholders. In particular, companies in countries whether the investor protection is low can use more often privately negotiated buybacks whereby firms repurchase shares from a large shareholder at a negotiated price. This method can be used to consolidate control and to eliminate a troublesome shareholder. In order to shed more light on these issues, I split the sample of announcements into their respective level of investor protection. La Porta et al (998) show that among the countries in the world that offer strong protection for their minority shareholders are UK, Ireland and Spain. Other European countries are considered to offer low protection for small investors. Table IV reports the cross-legal system differences in returns. Panel A shows that in the pre-event period [-0 to ] the difference in cumulative abnormal returns across the two legal systems is not statistically significant. On the event date 0, the abnormal returns of 0.5% in low protection countries are statistically lower than the 0.95% in high protection countries (t =.67). Over the [+ to +5] period, share prices in low protection countries decrease by 6.0% while those in high protection countries increase by.80%. The difference in means between the two systems is statistically significant at the 0.0 level. Table IV, Panel B, compares the characteristics of companies repurchasing shares in both systems. The results show that, in low-protection countries, financial 6

17 companies account for.5% of announcements compared to 0.% in high-protection countries. The t-statistics of differences in means of.9 is significant at the 0.05 level, suggesting that in high protection countries, financial companies are more likely to repurchase shares. As shown in the industrial classification of the sample firms in Table I, in high protection countries (mainly UK) there are more investment trusts buying back their shares to reduce their discount, thus to signal their under-valuation. Table IV, Panel B. also shows that in the high protection countries the proportion of shares companies intend to repurchase of.6% is significantly higher than the 9.% sought by companies in low protection countries. In the next three rows of Table IV, Panel B. a comparison of the methods used by the sample firms to repurchase their shares is presented. Consistent with the US evidence (e.g., Stephens and Weisbach, 998), the vast majority of companies in both systems intend to use open market repurchasing method, although the proportion is significantly higher in the low protection countries. 7 In contrast, there are more companies using the tender offer method in high protection countries. 8 This is due to the UK legislation that made companies treat all their shareholders equally under the "preemption rights rule. The differences in means in the use of negotiated method are not significant. These results are not consistent with the expropriation hypothesis implied in LaPorta et al (000). Finally, the last row of Table IV, Panel B. shows that 5.6% of announcements in low protection countries are made at the same time as other announcements compared to.9% in high protection countries. The differences in means are significant suggesting that firms in low protection are more likely to announce their intention to buyback their shares when they announce other news such as earnings. Table IV, Panel C, provides summary statistics on the financial characteristics of companies operating under the two systems. The average book-to-market of companies in low protection countries amounts to 0.9, compared to 0.6 in high protection countries. The difference in means of.6 implies that, on average, firms in low protection countries are value stocks while those in high protection countries are growth stocks. If as shown by Ikenberry et al (995), value stocks generate higher returns than 7

18 growth stocks as these companies are undervalued, we would expect companies in low protection countries to generate higher returns than those in high protection countries. The results reported in above are not consistent with this hypothesis. It is however possible that the book-to-market ratios are not directly comparable across the European countries because of accounting differences. To overcome this problem, I use the average -year growth rate in R&D as a proxy for growth. The results show that companies in high protection countries are more likely to invest more in R&D (thus higher growth) than low protection companies, suggesting that the under-valuation hypothesis observed in the US is not likely to apply in Europe. The results also suggest that the negative post-event abnormal performance of companies in low protection countries cannot be attributed to the punishment by the market of companies paying out cash in the form of buyback instead of financing growth opportunities. Table IV, Panel C. also shows that the average payout ratios of companies in both systems are equal. However, consistent with La Porta et al (000), the average growth rate in dividends paid by companies in high protection countries is significantly higher, while the growth rates in earnings are relatively equal. The last two rows provide summary statistics of the level of information asymmetry. Following Krishnaswami and Subramaniam (999), I proxy the degree of information asymmetry by analysts forecast errors (AFE) of earnings computed as: AFE i, T, h = F i, T, h A A i, T i, T where FI,T,h is the analyst s forecast of firm i s earnings per share in period T made at horizon h. AI,T is firm s i s actual earnings per share in period T. The mean absolute forecast error is used as a measure of accuracy. The forecast error is measured in the last reported earnings before the announcement of share repurchases. The forecasted earnings are taken as those made one quarter before the announcement of earnings. The data is extracted from the Institutional Brokers Estimate System (IBES). Firms with more information asymmetries are expected to have higher forecast errors, and therefore higher abnormal returns on the announcement dates of share repurchases. The results reported in Table IV, Panel C. indicate that UK companies have significantly 8

19 lower forecast errors than European companies. This suggests that companies in high protection countries have lower information asymmetries, implying that their share repurchases announcements are not expected to provide more information to the market than other European companies. The results reported in Panel A. which show higher abnormal returns in high protection countries are not consistent with the information asymmetry hypothesis. [Insert Table IV here] Table V reports the results of the regressions where the impact of these factors on the announcement date and post-event abnormal returns are taken simultaneously into account. The results show that low protection dummy, negotiated dummy, other announcements dummy and size have negative impact on day 0 abnormal returns. In contrast, the analysts forecast errors and the analysts accuracy are positively related to the announcement date abnormal returns. However, the book-to-market does not appear to affect the announcement date abnormal returns. These results suggest that smaller companies that are subject to high information asymmetries, that operate in high investor protection countries, that do not announce other news with share repurchases and do not use negotiated method to repurchase shares generate significantly higher abnormal returns on the announcement of share repurchases. The low protection dummy, the legal system in which the firm is operating, and the degree of information asymmetry, as shown in Equations (6) to (8), affects the postevent cumulative abnormal returns. The results indicate that companies that operate in common-law countries and/or in high investor protection and those that are subject to high information asymmetries generate significantly higher post-event abnormal returns. This suggests that for these companies, the markets under-react to the news conveyed by the announcement of the intention to repurchase shares. [Insert Table V here]. Conclusions 9

20 The purpose of the paper is to document the market reaction to share repurchases in UK and continental Europe. The analysis of the institutional framework shows that UK companies are likely to be different from their European counterpart because of a number of reasons. First, unlike other European countries, UK is a marketbased system where agency costs and information asymmetry problems are higher. UK is also governed by common law, which offers better protection to small shareholders. Finally, in the UK share repurchases are cancelled while in Europe, companies can choose between canceling the shares they repurchased or keeping them as treasury stock. These institutional differences are expected to affect the market reaction to the announcement of share repurchases. In particular, it is hypothesized that in the UK the market will react more favorably to the announcement of the intention to repurchase the shares than in Europe. In addition, the under-reaction hypothesis documented in the US is also expected to be supported in the UK because of the similarities between the structures of the US and UK market systems. Using a comprehensive sample of 6 announcements, the paper shows that the vast majority of these announcements are made in the UK. This is consistent with the proposition that share repurchases are not in the culture of continental European companies and that the restrictive legislation and heavy tax burden prevents these companies from returning their excess cash to their shareholders through buybacks. The results also show that there are stringent differences between the market reaction to share repurchases in UK and continental Europe. On the announcement day the increase of share prices increase significantly more in the UK and in continental Europe the announcement date abnormal returns are barely significant. This increase in UK by.6% is however substantially lower than the.5% documented in the US. Over the post-event period, + to +5 days relative to the announcement date, share prices in the UK continue to increase to reach.5%, suggesting that, like in the US, the UK market is under-reacting to the announcement of the intention to repurchase shares. In contrast in Europe, share prices decrease by 5.9%. The paper also documents the impact of the legal system, repurchasing method, levels of information asymmetry, growth, and size on the abnormal performance. The 0

21 results indicate that the market reaction is negatively related to negotiated method, size, and other confounding announcements but positively related to information asymmetry variables. In addition, in common law countries and in countries where the rights of minority shareholders are protected the market reacts positively to the announcement of the intention to repurchase the shares. In the post-event period, share prices increase by an annual compounded rate of about 8%, compared to a decrease by 6.0% in low investor protection countries. These results suggest that institutional settings as well as firm specific factors explain the market valuation of share repurchases in Europe. The analysis is concentrated mainly on the announcement of the intention to repurchase shares. It did not consider whether companies have actually bought back their shares and whether this has resulted in the positive post-announcement abnormal returns in the UK and the negative performance in the other countries in Europe. The analysis has also not considered the long-term performance of repurchasing companies. In addition the study has not considered alternative measures of information asymmetry such as volatility in stock returns, standard deviation of forecasts, and number of analysts following companies. The extent to which these factors will provide support or alter the conclusions is the subject of further research.

22 References Bagwell L. S. and J. B. Shoven, 989, Cash distribution to shareholders. Journal of Economic Perspectives, 9-0. Baker, H. K., P. L. Gallagher and K. E. Morgan, 98, Management's view of stock repurchase programs. Journal of Financial Research, -7. Ball, C.A. and W.N. Torous, 988, Investigating security-price performance in the presence of event-date uncertainty. Journal of Financial Economics,, -5 Barber, B. M., and J. D. Lyon, 997, Detecting long run abnormal stock returns: The empirical power and specification of test-statistics. Journal of Financial Economics, -7 Bernard, V.L., 987, Cross-sectional dependence and problems in inference in marketbased accounting research. Journal of Accounting Research, 5, -8. Brown S. J. and J. B. Warner, 985, using daily stock returns; the case of event studies, Journal of Financial Economics, -. Campbell, C.J., and C.E. Wasley, 99, measuring security price performance using daily NASDAQ returns. Journal of Financial Economics,, 7-9. Comment R. and G. A. Jarrell, 99, The relative signaling power of Dutch-Auction and fixed-price self-tender offers and open-market share repurchases, Journal of Finance 6, -7. Dann L. Y., 98, Common stock repurchases; an analysis of returns to bondholders and shareholders, Journal of Financial Economics 9, -8. Dimson, E., 979, Risk measurement when shares are subject to infrequent trading. Journal of Financial Economics, 7, Dittmar A. K., 000, Why do firms repurchase stocks? Journal of Business 7, -56 D Mello, R., and P. Shroff, 000, Equity undervaluation and decisions to repurchase tender offers: An empirical investigation. Journal of Finance 55, 99- Fama, E. F., 998, Market efficiency, long-term returns, and behavioral finance. Journal of Financial Economics 9, 8-06

23 Fama, E. F., and K. R. French, 998, Value versus growth: The international evidence. Journal of Finance 5, Fenn, G.W. and N. Liang, 00, Corporate payout policy and managerial stock incentives. Journal of Financial Economics 60, 5-7 Grullon, G. and R. Michaely, 000, Dividends, share repurchases, and the substitution hypothesis. Unpublished manuscript, Cornell University, Ithaca, NY. Grullon, G. and D. Ikenberry, 000, What do we know about stock repurchases? Journal of Applied Corporate Finance, -5. Ikenberry D. L. and T. Vermaelen, 996, The option to repurchase stock. Financial Management 5, 9-. Ikenberry D. L., J. Lakonishok, and T. Vermaelen, 995, Market underreaction to open share repurchases. Journal of Financial Economics 9, Ikenberry, D., J. Lakonishok and T. Vermaelen, 000, Stock repurchases in Canada: Performance and strategic trading, Journal of Finance 55, 7-97 Jagannathan, M., C.P Stephens and M.S. Weisbach, 000, Financial flexibility and the choice between dividends and stock repurchases. Journal of Financial Economics 57, Kothari, S.P, and J.B. Warner, 997, Measuring long-horizon security price performance. Journal of Financial Economics,, 0-9 Krishnaswami, S. and V. Subramaniam, 999, Information asymmetry, valuation, and the corporate spin-off decision. Journal of Financial Economics 5, 7- La Porta, R., F. Lopez-De-Silanes, A. Shleifer and R. W. Vishny, 997, Legal determinants of external finance. Journal of Finance 5, -50. La Porta, R., F. Lopez-De-Silanes, A. Shleifer and R. W. Vishny, 998, Law and finance. Journal of Political Economy 06, -55. La Porta, R., F. Lopez-De-Silanes, A. Shleifer and R. W. Vishny, 000, Agency problems and dividend policies around the world. Journal of Finance 55, -. Lasfer M. A., 996, Taxes and dividends: The UK evidence. Journal of Banking and Finance Vol. 0, No., April 996, 55-7.