Actual share repurchases, timing and corporate liquidity

Transcription

1 Actual share repurchases, timing and corporate liquidity Edith Ginglinger and Jacques Hamon a January 2005 Key words: open market share repurchases, timing, liquidity JEL classification: G14, G32, G35 Abstract Research into the impact of open market share repurchases has been hindered by the lack of data available on actual share repurchases in many countries, including the U.S. Using a previously unaccessed database containing detailed information on 36,848 repurchases made by 352 French firms, we show that corporate share repurchases have a significantly adverse effect on liquidity as measured by bid-ask spread or depth. Our results also indicate that managers' apparent timing ability is largely due to systematic contrarian trading. a Both authors are from Cereg, University Paris-Dauphine, place du Maréchal de Lattre, Paris cedex 16 edith.ginglinger@dauphine.fr, jacques.hamon@dauphine.fr. We are grateful to Gérard Rameix for providing the data about daily repurchases. We would also like to thank Bruno Biais, François Degeorge, Thierry Foucault, Bertrand Jacquillat, Meziane Lasfer, Myron Slovin, and Marie Sushka for their comments. 1

2 Open market repurchases have become increasingly popular in recent years in a large number of countries. Dittmar and Dittmar (2002) report that while repurchases in the United States accounted for 11.82% of total payouts in 1971, they represented 44.42% of payouts in An increase in repurchase activity is also documented for Canada and the U.K. Countries where stock buybacks were prohibited, such as Germany or Taiwan, now allow firms to repurchase shares on the open market. In France, open market stock repurchase programs have also increased dramatically since the July 2 nd 1998 law introduced provisions simplifying their implementation. Prior work focuses on valuation effects at announcements of open market repurchases. The choice between dividends and repurchases (Grullon and Michaely (2002)), the reasons why firms repurchase (Dittmar (2000)), the information content of buyback announcements (Grullon and Michaely (2004)), and the long-term performance of repurchasing firms (Ikenberry, Lakonishok and Vermaelen (1995)) have also been studied. There is less work about the impact of share repurchases on liquidity. Barclay and Smith (1988) find that repurchase announcements have a negative impact on liquidity. Miller and McConnell (1995) find no evidence of an increase in bid-ask spreads surrounding announcements of repurchase programs. Wiggins (1994) and Singh, Zaman and Krishnamurti (1994) report evidence that bid-ask spreads decline when buyback programs are announced. While companies may announce share repurchase programs, they are under no obligation to carry them out. The proportion of repurchases actually undertaken varies between countries and depends on the period 1, and knowledge of actual repurchases is required to gain a thorough insight into repurchase decisions. However, these factors are rarely studied. One reason is that data on actual repurchases, especially the dates of trades, are not available for most countries. The U.S. disclosure environment makes it difficult to study actual share buybacks. Firms can repurchase shares without making announcements, and can announce buyback programs without fulfilling them. There are no disclosure requirements for repurchasing, beyond a statement of the number of shares outstanding at quarter-end. This means that share repurchases in the U.S. can be neither observed at the time the transaction occurs, nor directly measured afterward. Previous studies have used two approaches to cope with this difficulty. One approach, used by Stephens and Weisbach (1998), relies on the quarterly decrease in the firm s shares outstanding as a proxy for the number of shares repurchased. However, the use of proxies generates inaccuracies because it understates the true amount of shares repurchased when a firm grants shares to 2

3 employees or stock options are exercised. This approach also provides little information about how firms implement open market repurchase programs, or the impact of the repurchases on market liquidity, since the true transaction dates are not known. Cook, Krigman and Leach (2004) use voluntarily disclosed data for 64 firms' repurchase programs over the period 1993 and An alternative approach is to look at stock exchanges where firms are required to disclose information about share repurchases. Brockman and Chung (2001) use data from the Hong Kong Stock Exchange to study the effect of managerial trading on liquidity, given that firms listed on that stock market are required to disclose repurchases the following business day. In this paper, we use data from Euronext Paris (the Paris Stock Exchange) to study the timing of repurchases and the impact of repurchase activities on liquidity. Firms listed in Paris are required to disclose data about repurchases for a given month at the beginning of the following month. We also obtain the precise trading dates for repurchases from the Autorité des Marchés Financiers (AMF) 2 database. We are therefore able to examine precise repurchase timing and analyse the impact of repurchases on different measures of corporate liquidity. We study the liquidity and information effects of actual repurchases on the Paris Stock Exchange for 36,848 firm-trading days over the period This sample size exceeds that used in prior studies of the impact of actual repurchases using time-stamped data. Our first objective is to assess whether buybacks are motivated by the opportunity to profit from private information or by price support objectives. Brockman and Chung (2001) find that managers are able to acquire shares at lower costs than a naïve accumulation strategy, but their findings could reflect opportune timing based on insider information, or straightforward price support. We find that companies repurchase their own shares following periods when the share price has been falling, evidence that companies purchase against market trends. There is no significant rise in share prices after buybacks. Thus, our results are consistent with the price support hypothesis, but provide little evidence for the theory that managers use private information to repurchase stock before prices rise. Our second objective is to determine the effect of repurchases on different measures of corporate liquidity. We examine two competing hypotheses. The first hypothesis is that liquidity is enhanced when management buys back shares by entering limit orders in the order book. The second hypothesis is that liquidity deteriorates on repurchase days. This deterioration may be due 3

4 to trading on private information or to contrarian trading. Prior studies that examine the liquidity effects of actual repurchases obtain conflicting results. Brockman and Chung (2001) find that liquidity deteriorates during repurchase periods at the Hong Kong Stock Exchange. Cook, Krigman and Leach (2004) show that repurchases contribute to U.S. market liquidity by narrowing bid-ask spreads and attenuating the price impact of order imbalances on days when repurchase trades are completed. Our study compares trading days when repurchase trades are executed with trading days surrounding non-repurchase days. We examine changes in the effective relative bid-ask spread, as well as in the depth, on the date of repurchase trading, and find a highly significant reduction in liquidity. This result is applicable to all market segments. Our results are broadly consistent with those of Brockman and Chung (2001). However, these authors argue that managers trading on private information explains their results. We find that managers mainly adopt contrarian trading. Contrarian repurchases reflect the existence of heterogeneity of expectations in the market. Such repurchases increase asymmetric information, which in turn increase the bid-ask spread. Buybacks are large enough to stabilize prices, but do not lead to higher share price. Our results are in contrast to the findings of Cook, Krigman and Leach (2004), who argue that differences in disclosure regulations between the U.S. and Hong Kong explain the divergence between their results and those of Brockman and Chung (2001). Despite differences between disclosure regulations in France and Hong Kong, we find that the two markets produce similar results. Our evidence suggests that it is the order flow that communicates the information that repurchasing is taking place. The remainder of the paper is organized as follows. In Section I, we develop hypotheses related to timing and liquidity effects of corporate repurchases. In Section II, we describe the structure and repurchase disclosure environment of the Paris Stock Exchange. In Section III, we describe the data. In Section IV, analyses of the empirical results are reported. Section V summarizes and provides conclusions. 4

5 I. Timing and liquidity hypotheses A. Timing or price support The explanation most frequently put forward for stock repurchases is the informationsignaling hypothesis. Share repurchases are a means by which management can signal its belief that the firm s stock is undervalued (Vermaelen (1981)). Stephens and Weisbach (1998) report negative abnormal returns during the period preceding the announcement of the buyback program, indicating that firms repurchase when managers perceive that the shares are undervalued. Ikenberry, Lakonishok and Vermaelen (1995) find negative price movement before a repurchase announcement, followed by a period of abnormally positive performance. However, these authors did not have access to the dates of the actual repurchases, and are thus unable to verify whether managers have timing ability. Brockman and Chung (2001) use a bootstrapping technique to assess managerial timing ability on the Hong Kong market. They construct empirical distributions of repurchase costs and compare them to the actual costs, finding that bootstrapped costs significantly exceed actual costs, results that are consistent with managerial timing. Cook, Krigman and Leach (2004) find that NYSE firms exhibit some timing ability, since acquisition costs are lower than various benchmarks, but Nasdaq firms do not on average beat their benchmarks. They document that firms repurchase shares following declines in price and that prices stabilize following repurchase trades, evidence that suggests firms repurchase to trade against the trend to support prices in a depressed market. Neither Brockman and Chung (2001), nor Cook, Krigman and Leach (2004) take into account the general trend of share prices over the period. Using data from the Paris stock exchange, we provide evidence as to whether timing or price support is the main motivation for repurchases. The market-timing hypothesis implies that share price for repurchase days should be lower than the price for subsequent non-repurchase days. The price support hypothesis implies that the share price for repurchase days should be lower than the price for prior non-repurchase days, but not significantly different from subsequent non-repurchase days. We discriminate between these hypotheses by examining 5

6 relative prices for repurchase days compared to non repurchase days. We conduct these comparisons for raw ratios and market adjusted ratios. B. Liquidity hypotheses The two prior studies that examine the liquidity effects of actual repurchases obtain conflicting results. Brockman and Chung (2001) find that liquidity deteriorates during repurchase periods at the Hong Kong Stock Exchange. Cook, Krigman and Leach (2004) show that repurchases contribute to U.S. market liquidity by narrowing bid-ask spreads and attenuating the price impact of order imbalances on days when repurchase trades are completed. We examine two competing hypotheses. The first hypothesis is that liquidity is enhanced when management buys back shares by entering limit orders in the order book. If this hypothesis is confirmed, we should observe a decrease of the bid-ask spread and an increase of the depth when the firm repurchases shares. The second hypothesis is that liquidity is reduced when the firm repurchases shares. This hypothesis is consistent with either market-timing hypothesis or price support hypothesis. According to the market-timing hypothesis, managers have an informational advantage over other investors and trade on it. The presence of a better-informed trader leads to a widening of the bid-ask spreads, increasing transaction costs for all investors and reducing liquidity. According to the price support hypothesis, managers trade against the trend in order to support prices in a depressed market. Thus, the bid ask spread widens due to the increase of asymmetric information resulting from the heterogeneity of expectations in the market. 6

7 II. The French disclosure environment and market structure A. Repurchase disclosure environment The legal and regulatory environment of share buybacks on the Paris Stock Exchange differs in some ways from that in the U.S. As established by the July 2 nd, 1998 law, open market stock repurchases are authorized up to the limit of 10% of capital and can extend for a maximum period of 18 months. The objectives and terms of the open market stock repurchase are defined at the annual general shareholders' meeting. For each 24-month period, subject to authorization by shareholders, shares representing up to 10% of existing capital can be cancelled. Firms can also use the acquired shares to grant shares or options to their employees, stabilize stock prices, exchange stocks as part of merger and acquisition transactions, reduce or eliminate dilution related to conversion of convertible bonds, or hold the shares as treasury stock, within the 10% limit. When a French company s repurchase program is approved, a registration statement is issued to obtain approval by the AMF. No subsequent announcement of the company s intention to repurchase shares is necessary before it proceeds with the actual buybacks. By comparison, in the U.S. there is no registration statement, but firms can voluntarily inform the market of their intentions just before the repurchase transaction. In the U.S., the only guide for executing open market repurchases is the SEC s "safe harbor" Rule 10b-18, which describes a code of conduct that, if followed, protects the corporation against share price manipulation charges based on the timing or price of repurchases. It is commonly believed by traders, the SEC, and the stock exchanges that the safe harbor rule is followed by most repurchasing firms (Cook, Krigman and Leach, 2003). Although French regulations are broadly similar to those in the U.S., there are some differences. The French rules are as follows: - purchases and bids must be executed through a single broker or dealer on any given day; - on the continuous market, buybacks are prohibited at the opening and closing call auctions; - non-block purchases for any given day must not exceed 25% of the standard trading volume specified for the relevant security. In the U.S., the reference period covers the four calendar weeks preceding the week of a Rule 10b-18 purchase or bid. In France, the 7

8 reference period covers the three trading days before the repurchase for the most liquid shares, listed on the deferred settlement segment. For cash only shares, the reference period is 15 trading days before the repurchase date; - the purchase price must fall within the lowest-highest price interval for the elapsed portion of the current trading day; - managers cannot repurchase shares during the two weeks preceding the disclosure of annual results, or when they are in possession of non-public information. There are two important differences between the U.S. and French regulations. The first difference concerns the possibility of reselling shares that are acquired through a repurchase program. In the U.S., these shares cannot be resold. American firms follow a strategy of pure accumulation, and the shares acquired are later cancelled or granted to employees when stock options are exercised. In France, however, repurchased shares can be resold, facilitating market timing. When the purpose of the repurchase program is price stabilization, companies are allowed to buy back and then may subsequently resell the shares. The second difference concerns disclosure. Every month, French firms are required to publicly report the total number of shares they repurchased or sold during the previous month. These French disclosure regulations are in contrast to those of the U.S. and Hong Kong. Until 2004, in the U.S. there were no disclosure requirements. In Hong Kong any shares repurchased on a given day must be reported before the start of trading on the following business day. France represents an intermediate disclosure environment that can provide further insight of the impact of disclosure. Examination of the French market also provides a framework for assessing the consequences of recent changes to U.S. regulations. In December 2003, the Securities and Exchange Commission adopted several amendments to Rule 10b-18. To enhance the transparency of issuer repurchases, the amendments require disclosure of all issuer repurchases of equity securities for the previous fiscal quarter, including the total number of shares purchased (reported on a monthly basis) and the average price paid per share. The rules also require footnote disclosure of the principal terms of publicly announced repurchase programs. Thus, in general, as of 2004, the disclosure of information concerning repurchases by U.S. companies is now similar to French 8

9 practices. Information is published quarterly, although data are reported monthly, whereas in France, data are disclosed and publicly reported monthly. Although the monthly disclosed data are useful, they are not sufficiently detailed for the purposes of this study. However, French firms must also report precise trading dates to the AMF. Although this detailed information is not public, we were granted access to it. The AMF collects a 0.15% tax assessed on a basis calculated by multiplying the number of shares repurchased by the weighted average price of the repurchases. Since sales are not subject to this tax, no salesrelated information is included in the database. Our work therefore focuses on share purchases only. Using published monthly sales data, we verify that, on average, resales comprise less than one fifth of the securities repurchased. For robustness, we also tested subsamples of companies that purchased shares, but sold no shares, over the period of our analysis. The results are similar to the full sample, and thus are not reported in the tables below. B. Euronext Paris market structure Euronext Paris is an electronic limit order market. There are two main market models: continuous trading, for the more actively traded stocks, and a double auctions market for the less liquid stocks. A description can be found in Biais, Hillion, and Spatt (1999). A call auction determines the opening and closing price in the continuous market. Trading takes place continuously from 9:00 a.m. to 5:25 p.m. Traders mainly use limit orders and market orders. Deferred settlement is possible for a subset of the more liquid shares: this market segment is called the service à réglement différé or SRD. For all other shares, only cash trading is possible. The distinction between SRD stocks and cash-only stocks is important with respect to the market model (all SRD stocks belong to the continuous market), but also to liquidity, that is, relative spread, turnover, and depth. The SRD market segment accounts for approximately 89% of the total market value of Euronext Paris, and 97% of the turnover. In several cases, we report results in three categories, according to the type of trading and market model: SRD, cash market continuous trading, and cash market - auction trading. 9

10 Euronext Paris offers substantial transparency. The five best bid/ask limits (price and quantity) in the order book are publicly released and members have access to all orders outstanding in the book, although the identity of the broker is not shown. Traders can also submit iceberg orders (also referred to as hidden-size orders). For these orders, only specified tranches are successively entered in the order book, and disclosed to the market, with the current time stamp after full execution of the preceding tranche. Over the period of study, transactions on the regulated block trades market were possible only for a small number of shares. Shares listed on the auction market, for example, are not eligible. 3 Trades of more than the normal block amount (NBA) defined by Euronext can either be executed through the Central Order Book or the block trade facility. For the period of our study, the NBA, which is established for each share and revised quarterly, corresponds approximately to the quantity of shares in the order book for the five best price limits. In both quantity and value, the NBA is highly variable from one share to another. An ordinary block could be traded within a price range extended to approximately the weighted average price of the five best price limits. Most transactions, even very large trades, are executed through the Central Order Book rather than block trading facilities. Off-hours transactions are also possible for all shares. Transactions made outside the trading hours are executed at a price within a range of 1% around the last traded price. III. Data We use the Euronext intraday database. Our dataset contains a time-stamped record of all transactions and orders submitted to the market from January 2000 to December 2002 for 918 French firms. These data include transaction prices, volumes, and the best limits of the order book (bid and ask prices, and bid and ask size), as well as market capitalization. All data are stamped to the nearest second. We also use data from the AMF, which provides the actual daily repurchases for all firms listed on the Paris Exchange from January 2000 to December The data include the name of the repurchasing company, the date, the number of shares repurchased, the share price, and the 10

11 daily total repurchase value. In some cases, a breakdown of all buyback transactions for a given day is reported, while in other cases only a weighted average price for the day is reported. On the whole, even when a detailed breakdown is available, it is difficult to locate the exact transaction in the database that corresponds to a particular buyback. On the SRD market, the average number of separate transactions in a day including repurchases is 970. There are two explanations for this relatively large number of trades on a given day on the Paris market. One, any order quantity is accepted, even a quantity of 1 share, with no penalty applicable. Two, it is an agency market, where orders are automatically transferred from the investor and the grouping of trades by brokers is limited. We therefore focus our work on the daily weighted average repurchase price. Figure 1 shows the daily value of repurchase trades on the Paris market, together with the CAC40 and SBF250 indexes, from January 2000 to December The repurchase volume is highly variable over the period. The end of 2000 and September 2001 in particular registered major waves of repurchases, but there was little buyback activity from February to May [Insert Figure 1 here] In Table I, the characteristics are reported for the 371 repurchasing firms (at least one trading day involving a repurchase over the three years) and the 547 non-repurchasing firms (no repurchase at all over the three years). Over the period, 40.41% of Paris-listed firms repurchased shares. [Insert Table I here] The repurchasing firms are significantly larger than non-repurchasing firms (average market capitalization is more than four times greater). The average daily rates of return are comparable for both categories of firms. The average effective relative spread for repurchasing firms, 2.31%, is significantly lower than for non-repurchasing firms, 4.32%. Turnover is higher for the repurchasing firms. Depth is not significantly different between the two groups and share volatility (whether measured daily, hourly or quarter-hourly) is lower for the repurchasing firms. Results on listing frequency confirm that repurchasing firms have higher liquidity than non- 11

12 repurchasing firms. Finally, the average number of daily transactions is 127 for firms that made no repurchases compared to 337 for firms that repurchased shares at least once. Proportionately, on the continuous market with deferred settlement facilities, 62.26% of firms repurchased shares at least once during the period studied. On the cash-only continuous market, 43.02% of firms repurchased equity, while on the auction market only 18.90% of firms repurchased equity. In general, the descriptive statistics indicate that the companies that repurchase shares are the largest and most liquid firms. IV. Empirical results Our results describe the main characteristics of share repurchases, managerial timing ability, and the impact of repurchases on liquidity. A. The repurchase decision Table II shows summary statistics for all repurchasing companies included in our sample. To be retained in the final sample, we require a precise reported date for the repurchase, that the reported repurchase price is higher than the minimum and lower than the maximum trading price that day, and that the quantity repurchased is lower than the total quantity traded on the exchange. Our final sample consists of 352 firms and 36,848 repurchase trading days over the 36- month period of study. [Insert Table II here] For our sample, we find that firms repurchase shares with values of between % and 14.15% of their market capitalization, with an average of 1.78%. The average number of shares repurchased in a single trading day was 12,495 for all shares, 40,532 for firms listed on the SRD market (continuous market with deferred settlement facilities) and 262 for firms listed on the auction market. Firms listed on the Paris Exchange repurchase million shares for a total of 33.9 billion. The most active firms trade on the SRD market, purchasing 33.3 billion shares, 12

13 or 98.11% of the total. The average repurchase accounts for 27.90% of the total trading value on the transaction date. Repurchases took place on 20.21% of firm-trading days. Repurchase frequency varies widely. Some firms spread their repurchases over several months, while others make fewer transactions and time the buying back of shares after a significant decline in price. The average number of repurchase days per firm is Firms listed on the SRD repurchase on 85.9 days, and firms on the auction market (cash-only continuous market) repurchase on 62.2 days (119.5 days). We next look at the distribution of firms in terms of number of repurchase trading days. There are 16 firms (4.54%) that repurchase shares on only one session over the three-year period. There are 35 (9.94%) that repurchase on 2 to 5 trading days, and 35 on 6 to15 trading days. There are 164 firms (46.59%) that repurchase on 16 to 100 trading days, and 16 firms (4.54%) intervene in their own equity market on more than 300 trading days. The distribution is similar for all three market type-based categories. We then divided the repurchase periods into sequences. A sequence is defined as a continuous series of trading sessions in which repurchase activity occurs, including a maximum break of two consecutive sessions with no repurchases. If the break is longer than two trading sessions, a new repurchase sequence is considered to have begun. A repurchase sequence is on average 7.69 trading days (8.49 on the SRD market and 5.64 for the auction market). The minimum sequence is 1 trading day and the maximum is 320 trading days. On average, there are repurchase sequences per firm. The average break between two consecutive sequences is 20.2 trading days. B. Managerial timing and private information We discriminate between the private information and the price support hypotheses, by examining the prices around repurchases. First, we utilize univariate tests (Table III). We compute the value-weighted average price (Vwap) as the average price of all transactions through the central order book, weighted according to the proportional value of the transaction in the trading day. We use several relative prices for our tests, both raw and market adjusted. We first compute the raw ratios for the Vwap for each trading day to the average Vwap over a two-, four- 13

14 and six-month period (beginning 1, 2,or 3 months before and ending 1, 2, or 3 months after the day concerned, excluding repurchase days). We also calculate market adjusted ratios, equal to the previous ratios divided by the same ratios over the same periods, calculated using the SBF250 index. According to the timing and price support hypotheses, both raw and market adjusted ratios should be lower for days with repurchases than for days with no repurchases. We also look at two other classes of ratios: the Vwap for each day relative to the average Vwap over the preceding one, two or three months or over the subsequent one, two or three months (excluding repurchase days). Both raw and market adjusted ratios are reported.. The market-timing hypothesis contends that managers buy stock because they are in possession of more favorable information than when they do not buy stock. We would thus expect the Vwap/(Vwap n monthsafter) ratio for repurchase days to be lower than the same ratio for non-repurchase days. The price support hypothesis contends that managers repurchase stock after a price drop, although they have no private information about future price developments. This means the Vwap/(Vwap n monthsbefore) ratio for repurchase days should be lower than the same ratio for non-repurchase days, while the Vwap/(Vwap n monthsafter) ratio should not differ significantly between the two. We find that the average Vwap/(Vwap 2n months) raw ratio is significantly lower on repurchase days (by 1.2%, 1.45% and 1.68% for respectively n= 1,2,3 and for the full sample). This effect is more pronounced for SRD shares and less marked for the cash only market. When the market adjustments are taken into account, the results are less pronounced but still very significant. These findings are consistent with both the timing and price support hypotheses. Taking the average Vwap/(Vwap n monthsbefore), we show that it is significantly lower for repurchase days. For instance, the average differential is 2.83% (3.12%) when repurchase days are compared to non-repurchase days for n=3 for the full sample (SRD market firms). For cash only market firms, the differential is 2.71%. The equivalent market adjusted ratios are also significantly lower on repurchase days: for three months, the differential is 1.33% for the full sample, 1.52% for SRD firms, and 1.29 for the cash only market. The raw Vwap/Vwap n monthsafter ratio, meanwhile, is also lower on repurchase days but the dip is small: only 0.3% on average for n equal to three, and the significance level of the differential is lower than for the preceding result. Further, the market adjusted Vwap/(Vwap n monthsafter) ratio is not significantly different on repurchase days, whatever the period and the market. While these findings are not sufficient to fully dismiss the existence of managerial timing, they are more consistent with the price support 14

15 hypothesis. 4 Over the medium term (one, two or three months), companies mainly repurchase their shares after the relative share price has fallen. The impact of private information is limited, and disappears when market adjusted ratios are considered. [Insert Table III here] We also observe the price trend during the repurchase day, and the trading days immediately before and after that day. If managerial actions are motivated by price support, repurchases would be expected to take place on trading days where prices are falling, and/or just after a fall in price. By comparison, if managers are exercising their timing skills, we would expect to see rising price trends in the trading days following the repurchase day. We observe the rates of return from two days before the repurchase to five days after. Rates of return are measured based on midpoints, so as to eliminate any effects due to microstructure (e.g. bid-ask bounce). The rate of return for a given trading day is calculated from that day's opening to the next day's opening. A more detailed analysis is used for the repurchase day and the preceding day. The rate of return for the day before a buyback transaction is separated into two components. One component is the rate observed from the first to the last transaction price. The other component is the rate observed off hours, that is from the last transaction price to the first transaction price of the repurchase day. Three rates of return are examined for the repurchase day. The first measures developments in price from the opening to the time of repurchase, the second covers the time between the repurchase and the close of trade, and the third from the close to the next day's opening. The repurchase price may come from a single transaction, or it can be a weighted average for several transactions. These results are compared with the rates of return when no repurchase takes place. For non-repurchase days, the rate of return breakdown is limited to the rate of return for trading hours (open to close), and the off hours rate of return (from the close to the next day s opening). The results are presented in Table IV for the full sample and for subsamples by market type. 5 [Insert Table IV here] 15

16 For the full sample, we find that repurchases take place after a fall in price over the previous two trading days. The midquote rate of return from preceding close to open is % for repurchase days, versus 0.121% for days with no repurchases for the full sample. These results are similar on all markets. The fall in price accelerates during the repurchase day. For the full sample, the midquote rate of return during trading hours is % (-0.022%) for repurchase (non-repurchase) trading days. Dividing the day into sections, the rate of return from open to repurchase is % for the full sample (-0.606% for SRD stocks and % for the others). After the repurchase, for the full sample, there is a 0.076% rise on average, while the rise for SRD market companies is 0.237%. Finally, in the five days after the repurchase, price trends are not significantly different from those when no repurchase had taken place, in any of the markets. Figure 2 illustrates the effect of price support trading, showing curves for the cumulative rates of return around repurchase days. [Insert Figure 2 here] These results suggest that firms act against market trends, timing their purchases to take advantage of falling prices. These results are also consistent with a price stabilization motive for repurchases. The results are particularly significant for stocks listed on the SRD market. However, companies do not appear to be able, in the short term, to repurchase stock just before a price rise. The price support hypothesis would thus appear to be confirmed. C. Corporate liquidity effects A repurchase program can affect the liquidity of a stock in two ways. One, the repurchase trading can increase liquidity on the bid side and make the firm s share more liquid. Two, the possibility of trading by an informed trader or a firm s contrarian trading strategy may reduce liquidity. We discriminate between these effects by using univariate and then multivariate evidence from the French market. 16

17 Table V reports summary statistics for a range of liquidity measures, divided into trade-based measures and order-based measures for trading days with and without repurchases. Trade-based measures include the number of trades and the turnover ratio, i.e., the ratio between the capital exchanged during the session and the market capitalization at opening. Order-based measures include effective relative bid-ask spreads and depths. Depth at the opening is measured by the natural logarithm of the sum of the value in Euros of shares posted at the best limits first observed in the order book that trading day. The bid-ask spread at opening (closing) is calculated based on the first price and the first spread available for the session (last price and last spread available). We also report the average relative effective spread, estimated using full day data records. In this case, the number of spreads over a trading day varies according to the number of transactions. The effective spread takes into account the possibility of a transaction within the spread, although in practice this is rare on the Paris exchange. 6 The relative effective spread relates the spread to the midpoint of the spread 7 : 2 Effective Relative Spread = Ask Price + Bid Price Midpoint = 2. Trade Price - Midpoint Midpoint, where We calculate means and medians for repurchase days and non-repurchase days, and also report t- tests. [Insert Table V here] For our sample of 352 firms that repurchase shares at least once over the period , most of the liquidity indicators deteriorate significantly on repurchase days. Results are presented for all shares (table V, panel A), for SRD shares (table V, panel B) and for shares on the cashonly market (table V, panel C). The turnover ratio falls from 0.137% on non-repurchase days to 0.12% for repurchase days for the full sample, from 0.259% to 0.234% for SRD shares, and from 0.076% to 0.073% for shares on the cash-only market (the last difference is not significant). The daily average number of transactions is (295.5) for days without repurchases (with repurchases) for the overall sample. For all subsets, there is a reduction in volatility, calculated over the 21 days preceding the day concerned, for repurchase days. When we look at the 17

18 volatility just prior to the repurchase (based on 36 hourly or quarter-hourly rates of return), however, the results show a sizeable increase in volatility. In annualized data, volatility, based on the 36 preceding quarter-hourly rates, rises for the full sample from 90%, when there is no repurchases, to 106% when repurchases occur. As for the number of rates of return available for the 36 quarter hours preceding the repurchase session, there is a reduction for days with repurchases from 17.8 to 14.5 for the full sample, and from 27.5 to 25.5 (7.3 to 6.3) for SRD shares (cash-only market shares). Depth and spread also deteriorate. Depth drops from 9.28 when there are no repurchases to 9.15 when there are repurchases for the full sample, and from to (8.83 to 8.76) for SRD shares (cash-only market shares). This result is confirmed by examination of the spreads. The relative effective spread at opening increases significantly, from 2.13% for days without repurchases to 2.40% for days with repurchases, for all shares. For SRD shares, the spread rises from 0.77% to 0.90%, while on the cash-only market, it rises from 2.85% to 3.05%. We also carry out an in-depth analysis of the bid-ask spread around repurchase days. The comparison of trading days with and without repurchases may introduce a bias related to changes in spreads over time. In addition, the least liquid firms may be overrepresented on repurchase days. Table VI compares the first available (daily average) spreads for repurchase days with the spreads for three benchmark days. A lag (lead) benchmark corresponds to the day before (after) the repurchase day when no repurchase took place. The last benchmark corresponds to a fourweek period before the repurchase, after eliminating any repurchase days in the period. The average first available (daily average) spread for the full sample on repurchase days is 2.4% (2.08%), which is 0.07% (0.06%) higher than the previous day's average spread, and 0.14% (0.11%) higher than the spread for the four weeks preceding the repurchase. For companies listed on the SRD market (cash only market), the first available spread 8 is 0.90% (3.05%) on repurchase days, compared to 0.85% (2.83%) the previous day and 0.90% (2.94%) the following day. The average spread over the four weeks before the repurchase is 0.85% (2.84%). The average spreads on repurchase days are significantly higher than the priorperiod benchmarks (lag day and pre-repurchase period) for both samples, whereas they are significantly different from the lead day benchmark only for the firms listed on the cash only market. Despite the more limited differentials between spreads, these results are consistent with our previous findings. 18

19 [Insert Table VI here] Overall, our results support the hypothesis that liquidity is reduced when firms undertake share buybacks. To complement the univariate analyses, we use several regression models. Table VII reports the estimation results from the following regression model. LIQUIDITY i = α + β0 REPURCH i + β1 REPURCH1 i + β2repurch2 i + γ1 SRD i + γ2 MM i + γ3 VOLAT i + ε i [Insert Table VII here] Liquidity i is the dependent variable represented by either relative spread or depth for the firm-day i. 9 We log-transform the variable depth because of skewness in its distribution. We introduce three control variables. The cost components of the bid-ask spread fall into three categories: order-processing costs, inventory-holding costs, and adverse selection costs. Orderprocessing costs are largely fixed. The bid-ask spread should therefore decrease with trading volume. A market maker incurs inventory-holding costs when providing liquidity. In an orderdriven market, limit order traders run the risk of non-execution, and inventory costs grow as volatility rises. The bid-ask spread should therefore increase with volatility. Adverse selection costs result from the risk of trading with individuals who possess private information. Several measures of adverse selection costs have been used in the literature. The bid-ask spread should increase with insider ownership (Glosten and Harris (1988)) and more generally with volatility (Copeland and Galai (1983)), while it should decrease with the size of the firm (market value of the shares) and the trading volume. 10 In order to take into consideration the combined effect of the control variables suggested by previous studies of the determinants of the bid-ask spread, we introduce two dichotomous variables corresponding to the settlement methods and the market models. Variable SRD takes the value of 1 if the share is continuously quoted and has deferred settlement facilities. Variable MM is equal to 1 if the share is quoted on the auction market. These variables correspond to 19

20 different market capitalizations and trading volumes. We also introduce a volatility variable (average annualized daily estimation over the last 21 trading days) 11, that several previous empirical studies have found is a determinant of the spread (Stoll (1978) or Harris (1994)). Repurchasing operations are taken into account by using three dummy variables. The first, REPURCH, takes the value of 1 for a given day if the firm repurchases shares during that day, and zero otherwise. As discussed above, on the French stock market repurchases are often spread over a period of time, in sequences of variable duration. Even though the information is only officially disclosed the following month, it is possible that the first sequence will alert the market to the presence of a large trader (the repurchasing firm). The first day of a sequence opens a repurchase period. Although no public announcement has been made, observation of order flows may lead investors to infer that the firm is likely to continue buying back shares. To take into consideration the impact of the market's being taken by surprise by the first intervention, and the subsequent disclosure of the information, we introduce two variables. REPURCH1 takes the value of 1 if the day of repurchase is the first day in a sequence, and zero otherwise. REPURCH2 takes the value of 1 if the trading day concerned is not the first of a sequence. Liquidity, by whatever measure, is higher for firms on the SRD market and lower for firms on the auction market. It decreases as volatility for the 21-day period preceding the trade increases. This is in keeping with previous findings on the determinants of the bid-ask spread. Our results show that relative spread increases and depth decreases on days when repurchases take place. These findings are particularly robust, being independent of any particular trading or settlement method. Liquidity is reduced as soon as the firm trades in its own shares, in any market. 12 This supports the hypothesis that information asymmetry is greater when an informed trader (in this case the firm itself) intervenes in the market. We also compare the impact of first repurchases in a sequence with the impact of subsequent repurchases. The results indicate that the adverse effect on liquidity, which starts as soon as the first repurchase takes place, is particularly noticeable for repurchases subsequent to the first in a sequence. Depth, both at opening and closing, deteriorates considerably more for the repurchase days that come after the first repurchase. Despite the lack of immediate disclosure, the evidence suggests that the order flow conveys the information that repurchase is taking place. We observe that at the opening of a session, effective relative spread deteriorates by on a first repurchase day in a sequence and by for a subsequent repurchase day. The impact on 20

21 liquidity increases even during the first repurchase day. The first repurchase day leads to a decline of in the relative spread at close of trading, and a decline of in subsequent sessions. When we consider the relative spread averaged over a trading day, we find a decline on repurchase days. Our results support the hypothesis that information asymmetry increases when the firm trades in its own shares. The repurchase orders are large enough to stabilize prices and reduce liquidity, but do not lead to a price increase. Our results show that the deterioration in liquidity on repurchase days in France is similar to the effect observed in Hong Kong, although the disclosure deadlines are different in the two markets. Thus, it is unlikely that the divergence between the results for the U.S. and Hong Kong can be attributed solely to differences in disclosure regulations and practices. Our findings contribute to a debate that began in the U.S. in December 2002 about the compulsory disclosure of information as to when repurchases actually take place. Since the new U.S. disclosure regulations, adopted in December 2003, closely resemble the French environment, it should eventually be possible to determine whether the differences in observed results are attributable to differences in disclosure requirements or to other factors such as differences in the way the markets operate. V. Conclusion Our study contributes to an understanding of the impact of open market share repurchases on the stock markets. Following the 1998 reform of regulations governing share repurchases on the French stock market, there has been a wave of large-scale corporate share repurchases. We are able to study a large sample of actual share repurchases in France by obtaining access to a previously unavailable database. The lack of comparable information disclosure makes this type of analysis impossible in the U.S. Thus, our results contribute to the literature on stock repurchases. French firms must disclose all repurchases in a given month at the start of the following month, a constraint that falls between the non-disclosure that was the case in the U.S. before 2004 and the next-day disclosure compulsory in Hong Kong. 21

22 Within this context, we study both the managerial timing ability hypothesis and the impact of repurchases on liquidity. Our sample of 352 firms over three years, with corresponding timestamped data, is the largest sample used to date. On average, French firms repurchase shares at a lower price than that paid by other investors. We find that this outcome is mainly due to one of the purposes of stock buybacks, that is, price stabilization. Shares are repurchased after an observable decline in share price. Our results provide little evidence to support the theory that managers use private information to repurchase stock before the share price rises. Our conclusions are unambiguous on the issue of the impact of repurchases on liquidity. Whatever the trading method, repurchases significantly reduce the liquidity of the market for relevant shares. This reduction in liquidity has also been reported for the Hong Kong market, but not for the U.S. This divergence from U.S. results may be due to the lack of data on actual repurchases for the U.S. prior to Differences in disclosure regulation may be another factor. The similarity of our results to those for Hong Kong indicates that the choice of requiring firms to disclose repurchases either one day or one month after execution does not affect the impact on liquidity. In the Paris market, the order flow communicates the information that repurchases are taking place so that the spread widens and depth falls. There is also a difference between the market's reaction on the first day of a repurchase sequence, and the reaction on subsequent days. Our results suggest that there are limited benefits from greater post-trade transparency of corporate share repurchases. 22

23 References Barclay, Michael J., and Clifford W. Smith, Jr., 1988, Corporate payout policy: cash dividends versus open-market repurchases, Journal of Financial Economics, 22 (1), Biais, Bruno, Pierre Hillion and Chester Spatt, 1999, Price discovery and learning during the preopening period of the Paris Bourse, Journal of Political Economy, 107 (6), Bollen, Nicholas P., Tom Smith and Robert E. Whaley, 2003, Modeling the bid/ask spread: measuring the inventory-holding premium, Journal of Financial Economics, 72 (1), Brockman, Paul and Dennis Y. Chung, 2001, Managerial timing and corporate liquidity: evidence from actual share repurchases, Journal of Financial Economics, 61 (3), Cook, Douglas O., Laurie Krigman and J. Chris Leach, 2003, An analysis of SEC guidelines for executing open market repurchases, Journal of Business, 76 (2), Cook, Douglas O., Laurie Krigman and J. Chris Leach, 2004, On the timing and execution of open market repurchases, Review of Financial Studies, 17 (2), Copeland, Thomas E. and Dan Galai, 1983, Information effects of the bid-ask spread, Journal of Finance, 38 (5), Dittmar, Amy K., 2000, Why do firms repurchase stock? Journal of Business, 73, Dittmar, Amy K. and Robert F. Dittmar, 2002, Stock Repurchase Waves: An Explanation of the Trends in Aggregate Corporate Payout Policy? Working paper, Indiana University Glosten, Lawrence R. and Lawrence E. Harris, 1988, Estimating the components of the bid-ask spread, Journal of Financial Economics, 21 (1), Grullon, Gustavo and Roni Michaely, 2002, Dividends, share repurchases and the substitution hypothesis, Journal of Finance, 57 (4), Grullon, Gustavo and Roni Michaely, 2004, The information content of share repurchase programs, Journal of Finance, 59 (2), Harris, Lawrence E., 1994, Minimum price variations, discrete bid-ask spreads, and quotation sizes, Review of Financial Studies, 7 (1), Ikenberry, David, Josef Lakonishok and Theo Vermaelen, 1995, Market underreaction to open market share repurchases, Journal of Financial Economics, 39 (2/3),