Multiple blockholders and rm valuation: Evidence from the Czech Republic

Transcription

1 Multiple blockholders and rm valuation: Evidence from the Czech Republic Ondrej Nezdara December 3, 2007 Abstract Using data for the Prague Stock Exchange in 996 to 2005, I investigate how presence and power of multiple shareholders aect rm valuation by studying stock price reaction to ownership changes. The presence of the second large shareholder (SLS) has about 4% positive eect on the stock market valuation of the rm with large controlling shareholder. The results are in line with the theory that the SLS may limit rent extraction of controlling shareholder. The increase in power of the SLS has also positive stock price eect. Stock price raises after a ownership change associated with a mandatory buyback oer. However, the positive eect is eliminated by a presence of the SLS. The presence of the SLS thus substitutes takeover code in limiting private benets of control. Introduction Number of empirical works demonstrates an inverted U-shaped relationship between ownership concentration and rm value. In low levels of ownership, the rm value is growing with concentration but as ownership gets beyond a certain point, the controlling blockholder starts to enjoy private benets on expenses of minority shareholders. As the ownership outside the USA and the UK is more concentrated, the focus has shifted from the traditional conict of interest between managers and dispersed shareholders towards a conict with large controlling shareholder on one side and minority shareholders on the other side. Existence of private benets raises a question whether there exist ways to block or at least limit the rent extraction. Since private benets are not directly observable, literature investigates whether the presence of another large shareholder may aect the distribution of cash ow among shareholders by studying rm dividend policy - Faccio et al. (200), Gugler and Yurtoglu (2003), and Bena and Hanousek (2005) - or the directly rm valuation (measured by market-to-book value) - Faccio and Lang (2003) and Maury and Pajuste (2005). However, the results are still mixed. This paper studies the eect of the presence and voting power of multiple blockholders on the rm valuation. Contrast to previous studies, I analyze an immediate reaction of the stock price to a publicly observed change in the ownership. This approach allows me to diagnostic ownership changes that have PhD Student in Economics, Swedish School of Economics and B.A., Arkadiankatu 7, 0000, Helsinki, Finland, ondrej.nezdara@hanken. For example, Shleifer and Vishny (986), Morck et al. (988), Wruck (989) and McConnell and Servaes (990)

2 an impact on the rm value but not on its dividend policy. As the rm unobserved characteristics may play a crucial role in the ownership-value relationship (Himmelberg et al. (999)), I do no either face the problem of using market-to-book ratio as measurement of the rm valuation. One of the commonly used mechanisms to protect minority shareholders is the mandatory buyback oer once the controlling shareholder reaches some level of ownership. This may be seen as alternative mechanism limiting private benets to the presence of a another blockholder. I also analyze whether the existence of such takeover code may have an impact on the stock price. To be able to study the link between ownership structure and stock price one needs to observe a suciently high number of ownership changes of publicly traded stocks. Using data from the developed economies like the USA or the UK cannot give us an opportunity to study such relationship since signicant ownership changes of publicly traded rms are rarely observed. Moreover, the presence of the second large shareholder may have smaller eect on rent extraction in countries with good investor protection. Transition economies, on the other side, went through dramatic changes from the socialistic to the capitalistic system and changes in the ownership structure are more frequent. However, in most of these countries, the stock market is not developed and only small number of rms has publicly traded stocks. The situation in the Czech Republic since mid 990s provides the best possible scenario to analysis this relationship. In the rst half of 990s, a signicant part of the Czech economy was privatized through a massive voucher privatization. The setting of the voucher privatization ensured a dispersed ownership structure of privatized rms. The Prague Stock Exchange (PSE) was established after the privatization and more than 600 stocks of privatized rms started to be traded in this stock market. The protection of the minority investors was on low level during 990s in the Czech Republic. Among others, La Porta et al. (998) argues that a relative weak protection of minority shareholders leads to the concentrated ownership structure and, indeed, the process of a concentration of ownership started immediately after the establishment of the PSE. In contrast to other countries, there have been a high number of changes in the ownership structure of publicly traded stocks. This gives a unique dataset to study the relationship between the ownership structure and stock price. The results show that the presence of the second large shareholder (SLS) has about 4% positive eect on the stock market valuation of the rm with large controlling shareholder. The price climbs after the formation of another blockholder and drops after the exit of the SLS. It is in line with the theory that the SLS may limit rent extraction of controlling shareholder. The eect is more signicant when the controlling shareholder holds majority stake. The increase in power of the SLS has also positive eect on the stock price. The existence of takeover code has a positive eect on the rm valuation as when the controlling shareholder crosses the threshold, the stock price signicantly climbs by about 8% in 2-month period around the ownership change. However, the price raises only in rms with a single blockholder. The positive eect is eliminated by a presence of the second large shareholder. The presence of the SLS thus substitutes takeover code in limiting private benets of control. The rest of the paper is organized as follows. In the second section, I bring a background of the Czech privatization and stock market. The overview of the related literature is in the Section 3. In Section 4, I describe the methodology. Section 5 contains information about data and descriptive 2

3 statistics. Section 6 brings the results and conclusions are in Section 7. 2 Voucher privatization and the Czech stock market The formal Czechoslovak economy started to transform to market oriented at the beginning of 990s. One of the biggest most challenges of the economic transformation was the privatization of the state property. The Czech mass privatization program was implemented through a voucher scheme in 992 and in The Prague Stock Exchange (PSE), an ocial stock market, was established after the rst wave in 993 and the stocks of privatized rms started to be traded in the PSE. However, the protection of investors was on low level at the beginning of the PSE. And even though authorities tried to improve the law several times, the protection of minority shareholders has stayed on a low level for several years. The private benets of control were then relatively high. Dyck and Zingales (2004) report the block premium, used as a measurement of private benets, to be 58% of rm value in the Czech Republic in second half of 990s, the second largest among the group of 39 developed and developing countries. Among others La Porta et al. (998) argues that such environment leads to the concentration of ownership since holding a minority stake is not optimal. The setting of the voucher privatization ensured a dispersed ownership structure of rms after their were privatized and thus many owners did not hold the optimal stake in the rms. The process of a concentration of ownership thus started immediately after the establishment of the stock market. Thus, in contrast to other countries, there have been a high number of changes in the ownership of publicly traded rms. The fact that there is a such high number of rms in the stock market raised the question whether it is optimal from the stock market authority and investors' point of view. Most of privatized rms would never go public in the standard conditions. World Bank (999) [page 8] also pointed on this problem: To improve securities market, the broad objectives of the Commission for the security market should be to restrict the number of publicly trade-able companies,... These actions would ensure the achievement of price convergence, increase market liquidity, restore condence, and open more possibilities for new equity issues. Stock market authority realized that this may hurt the whole capital market and started to clean up the market and massively delist beginning in The result was that there was only about 80 publicly traded stocks at the end of year Besides the PSE, there was established an OTC market (RM-system) in the middle 990s. RMsystem has been focused mostly on the small investors - usually those who earned they stocks in the voucher privatization and do not trade as frequently as a typical agent in the PSE. Almost all shares have been traded on both markets concurrently. Due to the existence of those two parallel stock markets, we observed situation that particular stocks that had been delisted from the PSE were still traded in the RM-system. RM-system. However, the process of massive delisting was also implemented in the The situation in the Czech Republic during the second half of 990s was very specic compared those in both developed and transition economies. We observe a high number of changes in the ownership structure of public rms, the liquidity of majority of traded stocks was uctuating and the main stock market indices have gone through out the dramatic development in term of their levels. 2 In 992, when the rst wave of privatization took place, the Czech Republic was part of Czechoslovakia. More detail description of the privatization scheme can be found, for example, in Claessens (997). 3

4 The last but the least we have seen authorities trying to improve a corporate law and regulation of the stock market but with not very signicant success. 3 Literature review Literature studying a relationship between ownership structure and rm value is extensive. However, the evidence is still mixed. Number of studies (e.g. Morck et al. (988), Wruck (989), McConnell and Servaes (990)) nd an inverted-u relationship between insider ownership and rm value. As insider ownership increases, the incentives of the managers are, indeed, in line with the other shareholders and the rm value increases. But as the ownership concentration reaches some level, the insiders start enjoy private benets of control on expenses of the other owners and the rm value decreases. Thus, there are both costs of dispersed ownership and concentrated ownership and the relationship between ownership and rm value is nonlinear. Others do not nd any relationship. Demsetz and Lehn (985) argue that the changes in the ownership structure are driven by a prot-maximization. Firms are the always in equilibrium or at least close to it. Any change in the ownership structure should increases the rm value as the ownership structure moves to the equilibrium. Using data from the USA for period 976 to 980, they do not nd any link between ownership concentration and rm performance. Himmelberg et al. (999) extend cross-sectional results of Demsetz and Lehn (985) and argue that much of ownership variation may be explained by unobserved rm heterogeneity. Similarly, Demsetz and Villalonga (200) nd no relation between ownership and performance after controlling for the endogeneity of ownership. Assuming that there is a inverted-u relation, few studies have tried to answer a question whether another blockholder may monitor the controlling shareholder to treat minorities fairly. However, the second large shareholders may also collude with the controlling shareholder and rent extraction may become even more severe. Studies dier in a way how they try to discover rent expropriation since the private benets are not observable. First stream of literature has looked at the dividend pay-out policy. As dividends are distributed to all shareholders, the rm dividend policy may signal the size of rent extraction by the controlling shareholder. Faccio et al. (200) for Western European rms, Gugler and Yurtoglu (2003) for German rms and Bena and Hanousek (2005) for Czech rms nd out that the rms with multiple large shareholder have a higher pay-out ratio. This evidence support the hypothesis that presence of a second large shareholder decreases the amount of rent extraction in the rms. On the other side, Faccio et al. (200) nd the opposite eect for rms from South East Asia - sign that large shareholders collude. Second group of papers studies the eect on the rm value and rm performance. Expecting some level of market eciency, the value of the rms with higher probability of rent expropriation should be smaller as minorities would recognize such rms and ask for higher expected return. Maury and Pajuste (2005) using a sample of Finnish listed rms over period from 993 to 2000 argue that more equal distribution of the votes among large blockholders has a positive eect on rm value measured by Tobin's Q. Faccio and Lang (2003), on the other side, do not nd any eect of the presence of the second large shareholders on excess value of western European rms. Gugler and Yurtoglu (2003) look at the stock price reaction when the dividend policy change (taking change in the dividend policy as 4

5 a signal of change in rent expropriation) and nd out that CARs around the announcement of the change in the dividend policy are positively eected by the presence of the second large shareholder. Other researchers try to discover how big premium the investors are willing to pay for controlling stakes when there is another blockholder present. Dyck and Zingales (2004) among others estimate the size of private benets by the premium paid for the share in the block. However, they do not nd any signicant eect of the presence of the second large shareholder with at least 20% of shares on the block premium. 4 Event study method I use the classical event study method with few adjustments. Many Czech stocks were suering from thin trading (see more details in chapter 5.2.). This may cause several problems with estimating beta coecients when using a market-return model for estimating normal returns around the events - see e.g. Dimson and Marsh (983). As several studies showed that usage of the constant-mean-return model does not usually lead to much higher standard error than in case of market-return model, I use a constant-mean-return model. Following MacKinlay (997), a constant-mean-return model is dened as: R i, t = µ i + ε i, t E (ε i, t ) = 0 var (ε i, t ) = σ 2 ε i where R i t is a return of share i in time t and µ i is a trend of share i. The estimation of constant return and standard error are then calculated as following: ˆµ i = T L t=t 0+ R i, t σˆ ε 2 i = L T t=t 0+ (R i, t ˆµ i ) 2 where L is a length of the pre-window estimating period starting in time T 0 and ending in time T. 5

6 I set the estimation window to be from T-270 to T-2, where T is a time of the ownership change. The abnormal return is dened as AR i, t = R i, t ˆµ i The sample abnormal return is AR i, t N ( 0, σ 2 (AR i, t ) ) where for constant-mean-return model the variance of abnormal return σ 2 (AR i, t ) = σ 2 ε i The individual cumulative abnormal return is CAR i (τ, τ 2 ) = τ 2 t=τ + AR i, t N ( 0, σ 2 ε i (τ 2 τ + ) ) the average abnormal return N AR t = N i= AR i, t the variance of abnormal return var ( AR t ) = N 2 N i= σ 2 ε i cumulative average abnormal return CAR t (τ, τ 2 ) = τ 2 t=τ AR t 6

7 and the variance of the cumulative average abnormal return var ( CAR t (τ, τ 2 ) ) = τ 2 t=τ var ( AR t ) I set the event period to be T-20 to T+20. I assume that such period should capture both possible stock price movements before the change due to the expectation of the change in ownership and slow adjustment to news after the event which may be the case in such undeveloped stock market as the Czech one. However, I also analyze the stock price movements in shorter periods before, after and around the event. The results for shorter periods should discover the stock price reaction to less expected changes in the ownership. 5 Data 5. Ownership structure The ownership data of publicly traded companies are provided by the Czech security register, Stredisko Cennych Papiru (SCP). The SCP discloses the information about ownership structure of all Czech rms with at least one issue of publicly traded stocks. The SCP provides information about all owners with more than 0% of shares. If the company has parallel issues, the security register inform about all shareholders that own at least 0% of any issue of rm stocks (even the ones that are not publicly traded). This law was implemented in August, 996. My sample contains information from period between August 996 and December Sample selection There have been a relatively small number of rms with parallel issues of stocks which I exclude due to a problem of identication of the voting power of the blockholders. A particular blockholder may hold over 0 per cent of one issue but less than 0 per cent of the other issue. Even though I know the voting power of both issues, I am not able to identify the blockholder's total voting power in the company. By excluding rms with parallel issue, I also get the sample of shares that fulll the condition one share - one vote. Further, I exclude stocks of privatization funds because of problematic comparison with the other rms. This leaves 55 rms in the sample. Even though, Hanousek et al. (2005) among others argue that most changes in ownership structure are observed between 993 and 996 which is outside of my sample period, the number of ownership changes is still high in my sample Descriptive statistics Changes in the ownership structure During the period between 996 and 2005, the total annual number of ownership changes has been permanently decreasing (see Table ). It is mainly due to the fact that the number of publicly traded stocks has been decreasing as the process of delisting has been 7

8 Table : Number of changes in ownership structure and number of rms Year Number Average Number of rms number Number of rms in with at of of changes the least one changes sample change in per rm ownership 996 a) 8, , , , , , , a) Information available only since August 28, 996 Sources: SCP and author's calculations implemented. As only ownership changes of publicly traded stocks have been observable, the total number of observed changes decreased. The average annual number of ownership changes per rm has also decreased from the about in 997 to about 4 in 2000 but then has stayed relatively stable until the end of studying period (see Table ). Number of minority and majority blockholders in the rm In Table 2, I report the average number of owners having stake of specic size in the company. It shows that the privatized rms went through the process of concentration of ownership. At the end of 996, only 34.3 per cent publicly traded rms with one issues of stocks had a majority owner. In 2000, it was already 5.8 per cent of stocks. And at the end of year 200, almost two thirds of publicly traded rms with one issue of stocks had the majority owner. On the other side, the number of minority blockholders has decreased signicantly during the time. The average number of legal minority shareholders has decreased by almost 50 per cent from.93 to Similarly, the average number of listing minority has decreased. The average number of blocking minority shareholders in rm was relatively stable. The average number of all blockholders in a rm has decreased from in 996 to.683 in Stock market data Stock market data, i.e. stock prices and trade volume both in number of stocks and CZK, is provided by the private company Aliaweb. I have information on daily basis about all traded stocks in the 8

9 Table 2: Average number of owners per rm having specic stake at the end of years Year Average number of blockholders Delisting majority Simple majority Majority Blocking minority Listing minority Legal minority Total number of blockholders 996 a) a) Information available only since August 28, 996 Sources: SCP (security register) and author's calculations Prague Stock Exchange and RM-system for the period between March 993 (the time of establishment of the PSE) and June The trading mechanism in the PSE and RMS was set that most of the stocks have been traded for the xed price within one day. The price was set up after the authorities received the bid and ask orders in the pre-trading period. To be able to do analysis with all stocks, I set the price in days with non-zero trading to be the average price, i.e. trade volume in CZK divided by number of traded stocks Infrequently traded stocks As many stocks were illiquid, the days without any trade with a particular issue of stocks were not extraordinary. The price in days with no trade is set to be equal to price in nearest previous day with some trades. The stock price is then P i, t = trade volumei, t traded stocks i, t if traded stocks i, t > 0 = P i, t otherwise Using such price adjustment, we observe many zero-return days and few days with extreme changes in the price. The standard error estimated for such stocks will be possibly higher than actual. The result thus may be more likely insignicant. 9

10 5.2.2 Parallel stock markets and price arbitrage As mentioned before, there are two parallel markets where stocks of privatized rms are traded. Existence of parallel markets may lead to possible arbitrage. Indeed, the dierence in prices was sometimes signicant, especially in early years of stock markets and for illiquid stocks - see e.g. Hanousek and Nemecek (2002). There have been several cases when the price dierence was high enough to make a prot even with high transaction cost that investor may have faced. For example, stocks of Vysocina Vyklatice, were traded in April 995 on price between 65 to 83 CZK (about 2-3 euros per share) in the Prague Stock Exchange and on price between 260 to 325 CZK in the RM-system (approximately 4 times higher than in the PSE). There have been several trades on both markets during April 995. The dierences were less frequent in later year but there were still some cases even in the last year of the studying period. The existence of the arbitrage possibility may have caused several problems when using event study approach especially for illiquid stocks. Assume that the stock is traded for signicantly dierent prices in two parallel markets and with low frequency in both markets. When setting the stock price as average price from all trades in both markets, we may observe several jumps. The price for one day may be calculated from the trades from the rst market (as there was no trade in the second market) and the following day only from the trades from the second market (no trade in the rst market). Thus, I look at the stock price reaction in the market where a particular stock is traded more in the month of observed change in the ownership structure. 6 Results I investigate a short-term reaction of the stock price to a ownership change. I assume that stock price measures the rm valuation from atomistic shareholders' perspective since the blockholders only rarely traded in the Czech stock market and rather privately negotiated the trades. Thus, change in ownership structure may lead to a change in private benets even the actual performance of rm remains same. However, the disclosed performance changes. As the atomistic shareholders trade on knowledge and expectations of publicly known performance, the price should react to the ownership change if it leads to change in rent extraction. The role of the second large shareholder (SLS) in limiting the rent extraction of the controlling shareholder has to be studied in a rm with a large controlling shareholder that may enjoy private benets on the expenses of the other shareholders. Since I study the reaction of publicly traded stock, I dene the presence of a blockholder by holding a stake that is publicly observable (0% of shares in the Czech Republic). I analyze three aspects of the ownership: () the presence of the SLS; (2) the presence of the SLS with respect to the power of the largest shareholder; and (3) the power of the SLS. To be able to answer the question whether the presence (power) of the SLS has an eect on the stock price or not, the changes in the presence (power) of the SLS have to be observed. As the existence of takeover code may have an eect on the stock valuation of rms, I investigate the ownership changes when the controlling shareholder reaches the levels of ownership connected to the mandatory buyback to understand the eect of such legislation. As many ownership changes in presence (power) of the SLS 0

11 may be associated with mandatory buyback oer, I also limit my sample to changes when no level of mandatory buyback have been crossed to get more precise estimation of the role of the SLS in limiting the private benets. 6. Presence of the second large shareholder The eect of the presence of the SLS is measured by the stock price reaction to two pairs of events. The rst couple of events are moving in/out (or formation/exit) of the SLS. Formation of the SLS is dened as the ownership change where the initial ownership structure is characterized by a single large shareholder holding at least 0% of shares and controlling the rm and no other shareholder holds at least 0% of shares. After the change, the rm has two shareholders with at least 0% of shares. Even if the previous largest shareholder still controls the rm, his position has weakened as another blockholder stepped in. I assume that there is a positive probability that the SLS do not collude with the controlling shareholder and thus the SLS uses her power to monitor the controlling shareholder and possibly to stop or limit the rent extraction. The formation of the SLS should thus have a positive stock price eect. The moving out of the SLS is opposite ownership change, i.e. initially, there are two blockholders and only one after the change. Similarly, exit of the SLS gives the controlling shareholder more space for rent extraction and I expect a negative stock price reaction. The second pair of events is more specic than the rst one. The ownership changes when the SLS moves in/out and at the same time the stake of the largest shareholder remains same should measure the eect of presence of the SLS more precisely than the rst couple of events. First, no change of the largest shareholder's stake indicates that the two blockholders did not trade with each other. Stock market participants thus may expect that collusion of blockholders is less likely than in case when they trade with each other. Thus, I expect more signicant results than in case of studying all SLSmoving-in/out events. Second, when the SLS exits and sells her stake to the controlling shareholder, the controlling shareholder may reach one of the level associated with mandatory buyback. In such cases, the total stock price reaction may be positive due to the positive price eect of existence of such oer (see more in chapter 6.5) even the exit of the SLS has a negative price eect. When the controlling shareholder's stake remains same, there is no buyback oer. The results show that the price climbs before the actual formation of another blockholder by 2.53% in period T-20 to T- and.25% in period T-5 to T- (See Table 3 - Panel A). The positive return already before actual change may be explained by the fact that some blockholders do also trade in the market and thus there is short-term demand access which drives the price up. If they use third party to buy the shares for them, the buying from the market does show in the ownership structure until the shares are transferred to the blockholder's account. The alternative explanation is that at least some of changes in the ownership structure were anticipated - in cases of public tender or similar cases. The stock market thus priced the expectations about distribution of future prots already before the actual ownership change. After the change, the price does not react signicantly. The overall reaction is thus positive but not signicantly dierent from zero. Exit of another blockholder has a negative price eect before the actual change (again possibly similar explanation as in the moving-in case). However, we observe strongly positive eect after the change (rising from 2.0% for the period T to T+20 up to +7.95% for period T to T+20).

12 However, as mention above the result may be contaminated by a fact that part of the observations are the ownership changes associated with mandatory buyback oer. The results of less contaminated sample, i.e. moving in/out of the SLS and no change in the stake of the controlling shareholder, are in line with the expectations (Table 3 - Panel A). The positive CAR before formation of the SLS is even more signicant. The post-event CARs are not signicantly dierent from zero, but the CAR for whole window periods are signicantly positive in two out three event period (2.4% for T-0 to T+20 period and 6.39% for T-20 to T+20 period). The stock price reaction to the SLS-exit have changed even more signicantly. The total reactions become negative as expected. The CARs are still signicantly negative before the change (-3.86% for T-5 to T- period and -4.56% for T-0 to T- period) and insignicantly dierent from zero after the change. The total stock price reaction around the event is then signicantly negative (about -4% for both T-5 to T+5 period and T-0 to T+0 period). The presence of the SLS thus leads to approximately 4% increase in rm value in cases when the largest shareholders do not trade with each other. 6.2 Power of the controlling shareholder and the presence of the SLS The eect of monitoring the controlling shareholder by the SLS may depend on the voting power of the controlling shareholder. In cases when there is no other large shareholder, majority shareholder has perhaps more opportunities to extract the rent from the rm than minority controlling shareholder. Thus, the presence of the SLS may have a bigger impact on the stock price as the amount of private benets is higher in majority controlling cases. On the other side, the eectiveness of monitoring may dier in cases when the majority controls the rm and when minority controls the rm as the monitoring of the majority owner may be more dicult. The positive impact of the presence of the SLS may be also explained by the possibility of control contest (see for example Nenova (2003)). Thus, the magnitude of the stock price reaction may be higher for any of the cases. However, the sign of the stock price reaction is expected to be positive for moving-in of the SLS and negative for moving-out of the SLS. Similarly as before, I use rst all the changes of formation/exit of the SLS. Secondly, I use only changes when the stake of the largest shareholder does not change to see more precise estimate. The results are reported in the Table 3 in Panel B. First, I analyze changes in the presence of the SLS with largest shareholder holding majority stake before and after the ownership change (Lines and 2 in Table 3). For moving-in of the SLS, we observe the positive but insignicant CAR before the change and also after the change. The total stock price reaction around the event is thus positive (between.3% and 2.8%) but insignicantly dierent from zero. For moving-out of the SLS, we observe signicantly positive results after change (7.5% for the longest post-event period) and as the returns are insignicantly dierent from zero before the change. The overall reaction is positive and even signicantly positive for two longest around-event periods. Second, I analyze ownership changes when the controlling shareholder does not have majority stake either before or after the ownership change. The results are qualitatively similar. The results changes dramatically when we limit the sample to the cases when the stake of the largest shareholder remains same, specially for those when the controlling shareholder holds the majority. The 2

13 formation of the SLS with majority owner controlling the rm brings a positive CAR before and after the ownership change. The overall eect is signicantly positive with about 6.5% for T-20 to T+20 period. The exit of the SLS brings a signicantly negative CAR before and signicantly positive CAR after the change. The total price reaction is negative but with exception of the shortest period, it is not signicantly negative, as the price recovers after a month. The results also change for cases when the controlling shareholder does not hold majority, especially for exit of the SLS type of change. The reaction is no more signicantly positive and even sometimes signicantly negative. The explanation of such changes in the results may be again that positive eect of the exit of the SLS was aected by existence of mandatory buyback oer. The formation of the SLS has thus higher positive eect when the majority shareholder controls the rm. The SLS is thus able to monitor eectively the distribution of rent even in majority-controlled rms where private benets are expected to be bigger. The formation of the SLS in minority-controlled rm does not have positive price eect as the private benets may be smaller. The exit of the SLS from majority-controlled rm leads to a drop in the stock price before the ownership change, but the price recovers within one month after the change. The exit of the SLS from minority-controlled rm has limited price eect. 6.3 Power of the second large shareholder Similarly in the previous section, we may assume that more the SLS holds in the rm more he is willing to monitor the behavior of the controlling shareholder. Thus, increase in the stake of the SLS should have a positive stock price eect (in case that two largest shareholders do not collude) since the minority shareholders may expect the SLS be more active in the monitoring. And following the same logic, the decrease in the voting power of the second large shareholder should lead to a drop in the stock price. To ensure that the probability of collusion is small, I focus only on the ownership changes in the power of the SLS when at the same time the largest stake remaining same. As I study the changes in the power rather than changes in the presence in this section, I excluded the cases when the SLS exits. Thus, the SLS has to hold stake of at least 0% both before and after change to be publicly observable. The results (see Table 4 - Panel A) show that the increase in the power of the SLS brings a positive CAR of 2% (although not signicantly dierent from zero) in period T-20 to T-. After the increase of power of the SLS, the price increases by about 2.5% (signicantly positive). The total stock price reaction is signicantly positive of 4.5% for the longest period (but signicantly positive for all periods). For a decrease in the power of the SLS, the results are not signicant. The stock price reaction does not depend on the power of the controlling shareholder (see last four bottom lines of Table 4 - Panel A) although the results for increases in power of the SLS when controlling shareholder has majority stake are less signicant perhaps due to low number of observations. 6.4 Mandatory buyback oer One of the law implemented to the protect minorities is a mandatory buyback oer. When majority shareholder reaches certain levels of ownership, he has to oer minority shareholders to buy out their 3

14 stake for certain price that is set up by the law. By the Czech law, the price in the oer was set up as the maximum of three prices: () the weighted average price in last 6 months; (2) price set by the licensed adjuster; (3) 85% of the price paid for the share in the stake (5% control premium). The limits are set as: 50%, 66.6%, and 75% of shares. Security commission has to approve the price oered to minorities. Until the commission approves the oer, the usage of newly bought stocks by the majority shareholder is blocked. The Czech security commission has not rejected the oer very often even though the prices of shares were many times criticized to be too low. However, the majority shareholder could not oer too small price since he would risk the rejection and the possibility that atomistic shareholders will take control over the rm for some time. The stock price reaction is strongly signicantly positive for majority of events associated with mandatory buyback oer (see Table 4 - Panel B). First, I study all crossings of the mandatory buyback threshold without focusing on the second large shareholder's presence. Interestingly, the before the actual change of the ownership, the CARs are not signicantly dierent from zero - for crossing-66% case, there are even signicantly negative. Thus, it seems that most the ownership changes were not anticipated by the stock market. After the change, we observe price increase of 7-8% within month after the ownership change. Similar results can be seen when studying only change when new majority shareholder stepped in, i.e. the majority stake changes hands (middle part of Table 4 - Panel B). However, the price reaction is less positive for cases when the SLS is present (5.35% and 6.20% for T-20 to T+20 period) and more positive when is not (.08% for T-20 to T+20 period). Finally, when the majority stake just changes hands and the amount of shares is same for the new and previous controlling shareholder. It indicates that the new controlling shareholder was not holding any shares in the rm before and all shares of the previous controlling shareholder were sold. Again it helps to eliminate the cases when the third party was involved in the trade. The presence of the SLS play a crucial role, now. If the SLS is present (and does not change the number of shares holding, i.e. she was not involved in the trade), we do not observe any signicantly positive stock price reaction. If the SLS is not present, the stock price reaction is strongly positive (3.22% for period T-20 to T+20). It goes in line with the theory that the presence of the SLS already aected the distribution of rent among shareholders and mandatory buyback oer will not signicantly improves the positions of minority shareholders in the rm. Thus, presence of the SLS eliminates the positive eect associated with crossing the mandatory buyback threshold. In summary, the mandatory buyback oer has a positive price eect of about 8% for T-20 to T+20 period. However, the results are driven mostly the cases when there is only single large shareholder. The presence of the SLS signicantly eliminate the positive eect of the mandatory buyback oer and thus we may see these two mechanisms as substitutes in limiting the private benets. 7 Conclusion This paper studies a short term stock price reaction to changes in a presence and a power of multiple blockholders to understand how the ownership structure eects private benets of control. I use data for the Prague Stock Exchange in 996 to 2005 since there were many signicant changes in the 4

15 ownership structure of publicly traded rms. I nd a support for the hypothesis that the presence of the second large shareholder (SLS) has a positive eect on the limitation of private benets of control. The presence of the SLS has about 4% positive eect on the valuation of a rm with large controlling shareholder, as the stock price raises after the formation of another blockholder and drops after the exit of the SLS. The eect of the presence of the SLS is even higher for cases when the controlling shareholder holds majority stake. The increase in the power of the second large shareholder have a positive price eect but the decrease in the power of the SLS has not a signicant price eect. The existence of takeover code has a positive eect on the stock market valuation of rm as when the controlling shareholder crosses the threshold, the stock price signicantly climbs by about 8% in 2-month period around the ownership change. However, the price raises only in rms with a single blockholder. Thus, the positive eect is eliminated by a presence of the second large shareholder. The mandatory buyback code and presence of the second large shareholder may work as substitutes to limit private benets of control. References Bank, W.: 999, `Country Report (Summary Report): Czech Republic, Towards EU Accession'. World Bank. Bena, J. and J. Hanousek: 2005, `Rent Extraction by Large Shareholders: Evidence Using Dividend Policy in the Czech republic'. CERGE - EI Working Paper. Claessens, S.: 997, `Corporate Governance and Equity Prices: Evidence from the Czech and SLovak Republic'. The Journal of Finance 52(4), Demsetz, H. and K. Lehn: 985, `The Structure of Corporate Ownership'. Journal of Political Economy 93, Demsetz, H. and B. Villalonga: 200, `Ownership structure and corporate performance'. Journal of Corporate Finance 7, Dimson, E. and P. Marsh: 983, `The Stability of UK Risk Measures and the Problem of Thin Trading'. The Journal of Finance 38, Dyck, A. and L. Zingales: 2004, `Private Benet of Control: An International Comparison'. Journal of Finance 59(2), Faccio, M., L. Lang, and L. Young: 200, `Dividends and expropriation'. American Economic Review 9, Faccio, M. and L. H. Lang: 2003, `The Separation of Ownership and Control: An Analysis of Ultimate Ownership in Western European Countries'. Journal of Financial Economics 65(3), Gugler, K. and B. B. Yurtoglu: 2003, `Corporate governance and dividend pay-out policy in Germany'. European Economic Review 47,

16 Hanousek, J., E. Kocenda, and J. Svejnar: 2005, `Origin and Concentration: Corporate Ownership, Control nad Performance'. CERGE-EI Working Paper 259. Hanousek, J. and L. Nemecek: 2002, `Market structure, liquidity, and information based trading at the Prague Stock Exchange'. Emerging Markets Review 3(3), Himmelberg, C., R. Hubbrad, and D. Palia: 999, `Understanding the determinants of managerial ownership and the link between ownership and performance'. Journal of Financial Economics 53, La Porta, R., F. Lopez-de Silanes, A. Shleifer, and R. W. Vishny: 998, `Law and Finance'. Journal of Political Economy 06, 355. MacKinlay, A. C.: 997, `Event Studies in Economics and Finance'. Journal of Economic Literature 35, 339. Maury, C. B. and A. Pajuste: 2005, `Multiple Large Shareholders and Firm Value'. Journal of Banking and Finance 29, McConnell, J. J. and H. Servaes: 990, `Additional evidence on equity ownership and corporate value'. Journal of Financial Economics 27, Morck, R., A. Shleifer, and R. W. Vishny: 988, `Management Ownership and Market Valuation: An Empirical Analysis'. Journal of Financial Economics 20, Nenova, T.: 2003, `The value of corporate voting rights and control: A cross-country analysis'. Journal of Financial Economics 68, Shleifer, A. and R. W. Vishny: 986, `Large Shareholders and Corporate Control'. Journal of Political Economy 94(3), Wruck, K. H.: 989, `Equity Ownership Concentration and Firm Value'. Journal of Financial Economics 23,

17 T-5 to T- before T-0 to T- T-20 to T- T to T+5 T to T+0 T to T+20 T-5 to T+5 total T-0 to T+0 The largest 2nd largest Number of shareholder shareholder observations moving IN %*.29% 2.53%* 0.06% -0.07% 0.0%.3%.23% 2.63% - moving OUT %* -.95%* -.56% 2.0%* 4.36%* 7.95%*.07% 2.4%* 6.39%* no change moving IN %* 3.34%* 2.99%* 0.4% 0.26% -0.45% 3.5%* 3.59%* 2.55% no change moving OUT %* -4.56%* -3.07% -0.50% 0.36% 2.59% -4.36%* -4.20%* -0.48% T-5 to T- before T-0 to T- T-20 to T- T to T+5 T to T+0 T to T+20 T-5 to T+5 total T-0 to T+0 The largest 2nd largest Number of shareholder shareholder observations MAJ->MAJ moving IN %.7% 0.53% 0.89%.3%.02%.34% 2.84%.55% MAJ->MAJ moving OUT % 0.29% 0.84% 0.30% 2.54%* 7.54%* -0.77% 2.83%* 8.38%* MIN->MIN moving IN %*.4% 2.76% 0.07% -.36% -.68% 2.07% 0.05%.07% MIN->MIN moving OUT % -0.92% -0.23% 2.54%* 5.06%* 7.04%* 3.2%* 4.3%* 6.8%* MAJ no change moving IN %* 4.36%* 2.07% 0.95% 3.24%* 4.4%* 3.7%* 7.6%* 6.48%* MAJ no change moving OUT %* -7.75%* -7.38%* 2.75%* 4.48%* 6.3%* -7.5%* -3.27% -.24% MIN no change moving IN %* 2.54% 3.7%* 0.00% -2.05% -4.2%* 3.35%* 0.49% -0.50% MIN no change moving OUT % -2.96%* -0.90% -2.3% -.7% 0.8% -2.95% -4.67%* -0.0% * 5% significance level TABLE 3 PANEL A: Presence of the second large shareholder PANEL B: Presence of the second large shareholder and power of the controlling shareholder after after T-20 to T+20 T-20 to T+20

18 T-5 to T- before T-0 to T- T-20 to T- T to T+5 T to T+0 T to T+20 T-5 to T+5 total T-0 to T+0 T-20 to T No change UP %.89%* 2.0%.39%*.7%* 2.56%* 2.42%* 3.60%* 4.57%* No change DOWN % 0.2% -.38% -0.52%.78% 2.39% -0.04%.9%.0% MAJ no change UP %.% 3.50% 2.06%.47%.60% 3.08%* 2.58% 5.0% MAJ no change DOWN % -0.95% -.0% -3.06% 0.50% 2.89% -2.78% -0.44%.80% MIN no change UP % 2.32%*.2%.03%.84% 3.09%* 2.07%* 4.6%* 4.30%* MIN no change DOWN % 0.54% -.49% 0.48% 2.29% 2.20%.04% 2.83% 0.70% T-5 to T- TABLE 4 PANEL A: Power of the second large shareholder PANEL B: Mandatory buyback offer before T-0 to T- T-20 to T- T to T+5 T to T+0 T to T+20 T-5 to T+5 total T-0 to T+0 The largest 2nd largest Number of shareholder shareholder observations Crossing one ALL % 0.07% 0.69%.44%* 3.92%* 7.23%*.68%* 3.99%* 7.9%* Crossing 50% ALL % 0.80% 0.8%.95%* 5.0%* 7.73%* 2.84%* 5.82%* 8.53%* Crossing 66% ALL %* -2.0%* -0.40%.96%* 4.89%* 0.9%* 0.8% 2.79%* 9.79%* Crossing 75% ALL % -0.82% -2.0%.% 3.8%* 7.39%* 0.68% 2.36% 5.29%* after after T-20 to T+20 New majority owner New majority owner and no change in stake ALL %.37%* 2.07%* 2.44%* 4.48%* 7.09%* 2.78%* 5.85%* 9.5%* above 0% % 0.% 0.73%.28% 2.56%* 4.62%* 0.8% 2.67% 5.35%* under 0% 34.2% 2.6%* 2.74%* 3.39%* 6.0%* 8.35%* 4.60%* 8.60%*.08%* no change (>0%) % 0.28% 0.59%.50% 2.52%* 5.60%* 0.74% 2.8% 6.20%* no change %.9%*.48%.92%* 4.4%* 6.78%* 2.33%* 6.3%* 8.26%* no change (>0%) %* -0.22% -.6%.73% 2.33% 2.55% -0.07% 2.0%.39% under 0% %* 3.43%* 3.39%* 2.06%* 5.92%* 9.83%* 4.07%* 9.35%* 3.22%* * 5% significance level