Block trades and the benefits of control in Slovenia

Authors


Abstract

Ownership and control have been concentrating in most transition countries. The consolidation of control introduces changes in the power distribution within privatized firms and, most importantly, redirects the corporate governance problem to a conflict between large and small shareholders. In this study, we evaluate the ownership changes in Slovenian privatized firms through an analysis of stock price reactions to the entrance of a new blockholder (the shared benefits of control) and through an estimation of the premiums paid for large blocks (the private benefits of control). We provide evidence of and discuss the reasons for the failures of the privatization investment funds in implementing control over firm managers and in promoting the restructuring of firms in the first post-privatization years.

1. Introduction

Extensive privatization programmes supported the transition to market economies in Central and Eastern Europe. The withdrawal of the state, along with the changes in the institutional infrastructure, was meant to bring along substantial improvements in financing, decision-making and consequently, in the performance of the privatized firms. Performance was expected to improve also due to some redistribution of ownership in the post-privatization period, which should have directed shares into the hands of the most interested owners. Notwithstanding the differences in the privatization models, firm ownership and control has been in fact concentrating in all these countries (Berglof and Pajuste, 2003). Ownership concentration should in principle provide firms with owners that have the incentive to monitor managers and guide their behaviour towards firm-value maximization. However, it also makes it possible for the large owners to divert firm value to the cost of minority shareholders. The negative effects associated with the extraction of these so-called private benefits may even offset any value increases due to improved monitoring. The prevalence of one or the other effect seems to depend on the blockholders’ identity, their willingness, motivation and ability to monitor firm management (Koke and Renneboog, 2003), their interaction with other owners (Earle et al., 2003), as well as on the regulatory and institutional environment in which the firms operate.2

Most of the empirical studies so far evaluate the impact of privatization on the different measures of firm economic and financial performance, while more or less successfully addressing the problem of the endogeneity of the ownership structure. Our study partly avoids the latter problem by relying on alternative yet related analyses. It evaluates ownership redistribution, via the stock price reactions to the entry of a new blockholder and the premiums paid for large share blocks. Stock price reactions reflect the ‘shared benefits of control’, in other words how the market estimates the management and monitoring capabilities of the block purchasers and, consequently, their contribution to a firm's value. Block premiums, on the other hand, approximate the ‘private benefits of control’, that is the value that large investors attribute to holding control. Our study is one of very few3 that analyses the benefits of control and uses appropriate methods to tackle the specifics of the stock markets in transition countries (where the illiquid trading of shares is pervasive). The analysis is based on a set of share-blocks exchanged in the organized markets of the Ljubljana Stock Exchange in the first post-privatization year (2000 and 2001). Although limited in number, they represent a substantial part of all ‘officially disclosed’ acquisitions of qualified holdings in Slovenian public corporations in the period of our analysis. Given that we are able to distinguish the identity of the block buyers, we provide new and, in our opinion, more direct evidence on the role of privatization investment funds (PIFs) in privatized firms and discuss their incentives for creating rather than expropriating corporate value. The ability of PIFs to become involved in the supervision of companies and to help resolve the collective action problem was in fact among the main motives for mass privatization (Frydman and Rapaczynski, 1993).

Our analysis of stock price reactions to the acquisition of a share block by a PIF provides no support for the superiority of PIFs as corporate monitors. Any positive market reactions preceding the entrance of a new PIF as the largest shareholder are in fact reabsorbed within 20 days of the block trade announcement. More pronounced positive effects are observed only when the block buyer is an industry-related non-financial firm (industrial owner). However, in all the cases analyzed here the industrial firms acquiring the blocks also announced a takeover within six months of the block acquisition. Positive price reactions could thus be largely driven by the expectations of an imminent takeover rather than by the expected contribution of the industrial owner to firm performance. At the same time, non-majority share blocks trade at relatively high premiums in relation to the exchange price, indicating that even minority control is valuable. We find that the premiums increase with the percentage of a firm's shares transferred in a block and with the incontestability of the blockholder's control.

The main explanation of the observed results in our view lies in the characteristics of the Slovenian institutional framework, particularly in the interface between the newly emerging rules and the distribution of power between different interest groups at the time of privatization. Apart from preserving state influence, the Privatization Law (1992) introduced substantial employee ownership and thus firmed up the employee and management role in the privatized firms.4 At least initially, corporate decision-making in Slovenia could thus be described as a form of bargaining between the main constituencies from the pre-transition period5 and the new owners, namely PIFs and, later on, domestic non-financial firms. We believe that PIFs in particular lacked financial resources, accountability and proper incentives to confront firm managers and actively promote corporate restructuring. And, most importantly, they were left with an absence of opportunities for value extraction at both the firm and the fund level. In such an environment, we would expect that factors other than the owners themselves induced firm restructuring. These factors include product market competition and the characteristics of firm managers (their concern for reputation as well as their ability and motivation). In fact, there is a ‘leading’ group of Slovenian firms with dominant management that are export-oriented, promote internationalization and growth and have borne the largest share of the burden of the Slovenian transition (Domadenik, Prasnikar and Svejnar, 2006; Prasnikar and Gregoric, 2002).

The paper is structured as follows. Section 2 provides an overview of the Slovenian privatization and the dynamics of ownership in the Slovenian post-privatization period. The results of an event study of stock price reactions to trades of large blocks are presented in Section 3. An empirical model estimating the private benefits of control and the related results are discussed in Section 4. Section 5 concludes with final remarks.

2. Privatization design and post-privatization ownership changes in Slovenia

The privatization of Slovenian public companies consisted of a mandatory distribution of 40 percent of shares to the funds6 and the internal distribution of 20 percent of shares to employees, former employees and their relatives (inside owners). The companies (in the form of workers’ councils) furthermore could choose how to proceed with the privatization of the remaining 40 percent, whether through internal buyout or public sale of shares. With the internal buyout being the prevailing method of the Slovenian privatization, most of the companies ended up with substantial inside ownership.7 Managers however obtained only minor stakes (3.85 percent on average, 1.40 percent in listed firms).8 The participation of domestic non-financial firms was also quite limited (average stake: 1.86 percent in listed firms; 3.55 percent in non-listed firms’ capital) as was that of foreigners (0.33 percent).

To balance the prevailing role of insiders, the Privatization Law (1992) introduced two main groups of outside owners: state-controlled funds (the Pension Fund and the Restitution Fund) and PIFs. The state-controlled funds were expected to slowly transform themselves into minority shareholders or dissolve their holdings while settling the obligations to restitution claimants and the claimants of the social pension system. PIFs emerged on the other hand as a special form of investment fund with the purpose of collecting ownership certificates from the public and investing them in corporate shares. Apart from dispersing small investors’ risk, the funds were constituted with the aim of monitoring firm management. Consequently, it was expected that through active involvement in the secondary market transactions they would restructure their portfolios and increase their shareholdings in selected firms. All this was also reflected in the ‘selective’ provisions of the Slovenian Takeovers Act (1997), which initially allowed the PIFs to acquire ownership blocks of up to 40 percent without any obligation to make a public bid, but applied a 25 percent threshold to other buyers. Also, the law on Investment Funds and Management Companies (1999) limited a PIF's participation in shares to 10 percent of the PIF's total investments rather than to the percentage of equity held in a single firm.9 This law provided further support to the PIFs’ activism by allowing their formal transformation into financial holding companies.10 However, the delays and uncertainties related to the prompt closing of the so-called ‘privatization gap’11 and to the PIFs’ subsequent transformation into modern financial institutions made it difficult to predict when and how portfolio restructuring and greater involvement by PIFs in firm governance would actually take place.

As in other transition countries, the ownership of Slovenian firms has become more concentrated since privatization. By the year 2002, the number of shareholders in both listed and non-listed firms had in fact fallen by almost 50 percent,12 while the ownership share of main shareholders has been increasing, particularly in non-listed firms (Brezigar, Gregoric and Zajc, 2007; Simoneti and Gregoric, 2004). Descriptive statistics on the size of the largest ownership block in Slovenian listed firms are presented in Table 1. Although the concentration of ownership in these firms has grown at a slower pace, the percentage of shares owned by the largest shareholder increased on average by more than 10 percent over the 1999–2002 period (t-test = 5.26). Table 2 complements Table 1 and reports the percentage of the largest blocks and the average size of the blocks owned by different investor groups in the same period. Note that the mean values are calculated only for those firms in which a given group is the largest shareholder. Despite the initial expectations, we only observe a slow withdrawal of the state-controlled funds. PIFs on the other hand were stuck with their initial holdings or consolidated their financial investments in selected firms. In this regard, we often observe the representation of affiliated PIFs within the same firm.14 Two new groups of owners slowly emerged: domestic non-financial firms and foreign firms. Non-financial companies increased both in number and in the average size of the largest blocks held. The presence of foreigners grew both through ownership increases by the existing foreign owners and through new acquisition of controlling shares in non-listed firms.

Table 1. Largest voting block: July 1999, July 2000, May 2001 and April 2002
 July 1999July 2000May 2001April 20022002–1999
Mean (SD)25.14 (14.94)27.51 (17.49)31.66 (19.82)35.46 (23.16)10.32 (20.76)
Median21.2521.9724.9927.043.56
Minimum5.007.226.825.11−44.51
Maximum91.1593.8796.6210091.97
P1011.9812.8213.7215.43−2.1
P2516.0617.1018.4819.740
P7529.6533.5539.6640.6912.93
P9039.8048.1655.5879.128.10
N firms112112112112112
Majority blockholder (N firms)911141910
At least 75% blockholder (N firms)2471210
Table 2. Frequency distribution of 112 listed companies by identity of the largest blockholder and the average voting stake held by the largest blockholder
 1999200020012002
% Firms% Share% Firms% Share% Firms% Share% Firms% Share
Individuals2.6817.344.4620.984.4617.694.4617.27
Foreigners6.2538.256.2538.156.2537.066.2552.25
Republic of Slovenia3.5723.473.5728.372.6822.881.7927.16
Non-financial companies12.5035.0619.6434.0919.6446.6725.8948.32
Banks2.6836.941.7928.612.6836.441.7942.63
Pension fund17.8617.6612.5017.6512.5017.2910.7117.95
Restitution fund8.0412.657.1412.846.2512.055.3612.19
PIFs43.7527.8442.8624.5543.7531.9141.0734.13
Workers’ association130.8939.070.8939.100.8939.042.6840.12
Development fund1.7964.030.8948.250.8964.120 0

What then is the impact of the observed ownership changes on the performance of Slovenian firms? Are the new owners actively involving in monitoring the management and consequently improving firm performance? Or is the observed consolidation merely driven by the owners’ desire to expropriate some pecuniary benefits that will not be shared with other owners? We provide some answers to these questions by analysing the specifics of block trades, namely the market reactions to the entrance of new blockholders (shared benefits) and the size and determinants of the premiums paid for share blocks (private benefits). These block trades suggest the characteristics of the reallocation of ownership in the post-privatization period in Slovenia. If it is efficient, this reallocation should be reflected in some way in the corresponding dynamics of share prices.

3. Block trades and the shared benefits of control in Slovenia

3.1 Theoretical framework and main hypotheses

Empirical studies of Western markets show that trades of share blocks of at least 5 percent are associated with significantly abnormal price returns. When positive, these abnormal price movements show that small shareholders anticipate benefiting from the change in control: they expect the new blockholder to bring in more efficient managerial or monitoring skills, to provide synergies in research, development and production as well as new incentives to increase the firm value. In the literature, the positive effects associated with the entrance of a new (better) owner that accrue to all shareholders are referred to as the ‘shared benefits of control’. Hence, assuming that the redistribution of ownership in the post-privatization period allocates the ‘exogenously’ assigned shares at the time of privatization into the hands of more interested and presumably better blockholders, the market should positively react to these transfers. In this regard, we are primarily interested in the impact of the two main groups of owners that have been the most active in the acquisition of these blocks in Slovenia: the PIFs and domestic non-financial firms.

In particular, there has been wide scepticism about the success of the PIFs as active owners of the privatized firms in Slovenia. We can think of several reasons for this. First, the privatization funds mostly lacked the capital and proper expertise to undertake the necessary restructuring, at least initially. Second, without a proper institutional and legal environment (as is the case in a transition country), economic incentives to monitor may be outweighed by other, stronger incentives for the expropriation of firm value. For instance, the funds may become involved in asset stripping and firm looting (Coffee, 1998), they may be providing business to their founders (banks) and so on. This potential for value extraction is particularly appealing during transition since the creation of markets for land and buildings makes these kinds of assets valuable after a long period in which they were ignored or undervalued in accounting statements (Wright, Buck and Filatotchev, 2005: 427). Some of the funds’ directors also seem to wrongly perceive their role on the boards as being that of an information agent who provides valuable information to investment companies for trading purposes (Coffee, 1998: 87, 103). In Slovenia, the redirection away from funds’ active involvement in governance was further fuelled by the lacunas in funds’ regulation and the uncertainty surrounding their formal transformation and subsequent listing on the Stock Exchange.15 All these inefficiencies enabled the funds’ managers (banks or private managing companies) to profit by charging relatively high managing fees, concentrating their ownership share in their funds at a very low price and exploiting the small shareholders’ passivity to promote their own agendas. As evidence of the latter, we can give an example of a PIF's formal transformation. For the purpose of transformation, a part of the PIF was organized as a financial holding company, while part of it was transformed into an investment company. The shareholders’ assembly approved the transformation with a mere 10.82 percent of votes cast. Slightly higher percentages (around 16.5 percent) are reported for other PIFs’ assemblies (Gregoric, 2003).16 Several factors thus increased the opportunity costs of restructuring and prevented the ‘active’ involvement of PIFs in corporate governance: on the one hand, costly monitoring and opposition from the state and inside owners, and, on the other hand, the lucrative trading profits from having access to inside information and the ‘post-privatization’ opportunities due to the (general) undervaluation of firm shares at firm and fund level.

Hence,

Hypothesis 1 (H1): Given the institutional settings (legal provisions, lack of the funds’ accountability, multiple blockholder structure with significant state-involvement and employee representation in firm ownership and corporate boards), we expect to observe limited (if any) positive abnormal stock returns following the acquisitions of blocks by a PIF.

On the other hand, assuming that certain preconditions are satisfied (including mutual trust and market competition) the presence of a significant owner with some industrial relationship brings to the controlling company a strong influence on both the governance and the operations of the controlled company, which should result in improved performance of the latter (Wright, Buck and Filatotchev, 2005). However, non-financial firms (particularly in the case of an industrial partner) can expropriate value from affiliated companies through, among other things, transfer-pricing and share dilution (Johnson et al., 2000). We assume that the former effect prevails so that,

Hypothesis 2 (H2): The market reacts positively to the acquisition of a share block by a non-financial firm announcing a take-over.

3.2 Data and the event-study methodology

We ascertain the influence of block trades on non-selling shareholders’ returns in Slovenia by applying the event-study methodology.17 The initial database is represented by all exchanges of stakes carrying between 5 and 25 percent of voting rights taking place on the Ljubljana Stock Exchange in 2000 and 2001 for which we could identify the parties involved in the stock transaction.18 There are three main reasons behind our choice of the block size. First, most empirical studies focus on trades of blocks of at least 5 percent as they are believed to provide their owners with enough power to actively influence the conduct of the firm's affairs (Barclay and Holderness, 1991). In Slovenia, 5-percent-ownership (voting) stakes in fact ensure a seat on a firm's Supervisory Board (Pahor, 2003). On the other hand, any acquisition of shares that, together with other shares, provides the buyer with 25 percent of the voting rights of a Slovenian public company is generally subject to a takeover bid. This determines the upper size of the blocks in our analysis. Information on the size and date of block trades was downloaded from the trading archive of the business review Gospodarski vestnik (http://www.gvin.com). We checked the accuracy of these figures by comparing them with those reported by the Ljubljana Stock Exchange. Stock prices and stock index values are those reported by the business daily Finance (http://www.finance-on.net). The parties involved in block trades were identified on the basis of articles from Finance and the Shareholders’ Register of the Central Clearing Securities Corporation. Except for three cases, in which the acquirer of a block was an industry-related non-financial company, the blocks were bought either by the PIFs or banks related to them. The data on takeover bids in 1999–2002 come from the Securities Market Agency. Information regarding the listing of companies, the number of shares outstanding and the constituents of stock indexes was downloaded from the web pages of the Ljubljana Stock Exchange.

The equilibrium models chosen for calculating normal stock returns are the market model and the market-adjusted model. The market index used for stocks traded on the official market is the ‘SBI20’ index; for shares traded on the free market it is the ‘IPT’ index. Both are value-weighted indexes and include the main and most liquid listed firms. Returns are daily returns calculated on the basis of the closing price of stock.19 We refer to the first trading day following the announcement of a block trade as the event day,20 while the announcement day is the effective announcement of the block trade. The abnormal performance for stock i on day t is calculated using the following equation:

image

where the Abnormal Return for stock i is obtained as a difference between its observed returns at time t and the ‘normal’ return estimated on the basis of the market index.

The market model coefficients α and β are estimated over a 200-day period starting 280 days and ending 80 days prior to an event. Abnormal returns are measured as prediction errors over the ‘event window’, the period around the event over which stock returns are examined. To take into account the slow incorporation of the announcement into stock prices (Banerjee, Leleux and Vermaelen, 1997: 30), we extend the event window from 20 days prior to 20 days after the event. Our main problem in computing abnormal returns was the low liquidity of Slovenian stocks since most shares do not trade every day. Hence, in the estimation period if a stock is not traded on a certain day that day is passed over for the stock and the market return. Stock is included in the analysis if it has at least 40 non-missing returns in the 200-day estimation period. In the event window, any non-trading day of individual stock is converted to its next trading day. The abnormal returns are then adjusted to take the multi-day character of the returns21 into account. In order to correct for the differences in stock returns’ variance and to release the strong assumption of cross-sectional homoscedasticity (De Jong, 1996: 7), we use standardized abnormal returns.

The abnormal performance measures used are: mean standardized abnormal return at date t (MSARt), that is, the average standardized abnormal returns across securities; this measure is a weighted average of the abnormal returns of individual stocks, with the weights inversely related to the estimated time-series standard deviation of the corresponding abnormal returns; mean standardized cumulative abnormal return (inline image), that is, the average across firms of the standardized cumulative abnormal returns over (t1, t2), where the standardized cumulative abnormal returns are defined as the mean of the abnormal returns of stock i over the event window (t1, t2) corrected by an estimate of the variance used in a standardized abnormal return test (or ‘Patell test’) below. The correction affects only multiple day windows and accounts for the eventual correlation of the abnormal returns within the window.

The statistical significance of the abnormal returns is assessed by applying three different tests:22

  • 1A standardized abnormal return test, which is asymptotically normally distributed and assumes that the ARs are cross-sectionally independent. On the assumption of cross-sectional independence,23 this statistic follows the standard normal distributions under the null hypothesis;
  • 2A t-test using a cross-sectional variance estimator, which depends on the number of firms in the sample and is robust to an increase in the variance of the ARs around the event dates,24
  • 3The rank test25: this non-parametric test can be used for event studies with small cross-sections. It further solves the problem related to the non-normality of abnormal returns as well as the thin trading of stock. Here we adopt the rank test proposed by Corrado (1989). This test takes into account the magnitude of abnormal returns, as the t-test does, but without the distributional assumptions which are necessary to implement the parametric t-test. The null hypothesis is that the shift in the distribution of event date excess returns is zero, that is, it should be uniformly distributed under the null that event periods are not different from non-event periods. The rank procedure assigns a rank to each daily return for each firm where rank 1 indicates the smallest abnormal return.

3.3 Empirical results

Figures 1 and 2 show the MSCAR for a clean sample composed of a total of 15 block trades, transferring on average 9.7 percent of the related voting rights (median value 7.5 percent) and referring to 15 listed non-financial companies. The reduction of the sample from the initial total of 40 blocks is mostly due to the fact that many of the stocks involved in block trades over the two years considered do not have the required minimum number of 40 non-missing daily returns over the estimation period (10 observations).26 We also excluded the trades in the shares of the same company that took place too close to be successfully distinguished one from one other (six observations). We further excluded block transactions that were actually acquisitions of shares by the firm itself (three observations) and block exchanges between PIFs and their management companies since these do not really involve a transfer of control (six observations). These generally represent facts or situations that add noise to the abnormal returns which interest us. While there may be concerns about the small sample size, Brown and Warner (1985), through a series of simulations with daily data, reinforce conclusions of previous works on event studies: ‘methodologies based on the OLS market model and using standard parametric tests are well-specified under a variety of conditions’ and even for ‘samples of either five or 20 securities, [ . . . ]. The goodness-of-fit tests do not indicate misspecification’. Thus, while non-normality and biases in estimating the market model appear to be unimportant in testing abnormal returns, the choice of the variance estimator is of some concern, affecting both the specification and power of the tests.

Figure 1 and 2.

Mean standardized cumulative abnormal returns (MSCAR) for the entire sample (15 companies): market model and market-adjusted model

In both plots, prices start to increase about 10 trading days before the announcement of the event (AD27 = 0); the market somehow anticipates the block trade. An additional upward movement in stock returns is observed four trading days before the event date. From approximately 10 days after the announcement, we observe different behaviour in the abnormal returns obtained in the market model compared to those estimated in the market-adjusted model. These differences might be due to the fact that the restrictions imposed in the market-adjusted model are not completely appropriate for some of the securities in our sample.28 Within 20 trading days of the trade, stock prices seem to settle close to the initial level, even if this downward turn is more pronounced in Figure 1.29Table A1 in Appendix A, reports the daily MSARt around the event date AD and the corresponding statistical tests. Given the possibility of a misspecification in the market-adjusted model, we mostly refer to the market model when analysing our results. The null hypothesis of a zero abnormal return is rejected on days (AD-9), (AD-2), (AD-1); abnormal returns on these days are positive at a 1- and 5-percent level of significance. Significantly negative abnormal returns are instead observed 10–11 days after AD. As the figures show, in the 20 days surrounding a trade (–10, +10) the stocks involved in the block trade experience a positive average cumulative abnormal return; most of this increase is concentrated in the 10 days preceding AD. In fact, according to Table A2 the highest and most significant values are observed over the windows (–2, 0) and (–9, 0).30 The stock performance observed before the announcement of the trade is similar to that reported for Poland (Trojanowski, 2002) and might indicate a leakage of information.

In order to account for the identity of the block acquirers and the potential for subsequent takeovers, we replicated the event study estimation separately for the companies that were acquired by non-financial firms (3) and those acquired by PIFs or their founders (banks). The mean cumulative standardized abnormal returns for these companies (Table A3) are positive and significant at 1 percent (st-test, adj-test) and 10 percent (r-test) over the windows (–2, 0), (–1, 0), (–9, 0), but not over (0, +20). Figure 3 shows that the first group of firms experienced positive abnormal returns over the period (–10, +20). Looking at Table A5, the highest and most significant increases are observed over the two days preceding the trade.

Figure 3 and 4.

Mean standardized cumulative abnormal returns for the three companies taken over within six months and for the companies that remained independent (market model)

MSCAR in Table A4 is positive over the windows (–10, +10) and (–2, 0), significant at a 10 percent level according to the ct-test and st-test statistics, respectively. The negative values observed over (0, +20) are not significant. For the sample of firms acquired by PIFs (Figure 4), the positive effect of a block trade announcement is lower but still statistically significant; it starts again about 10 days preceding the announcement of the block trade but is completely reabsorbed within 20 days of the AD. In Table A6, MSAR(–13), MSAR(–9) and MSAR(–2) (from the market model) are positive at the 1-percent level of significance. These results seem to support our first hypothesis and suggest that the entrance of a new blockholder, when a PIF, only temporarily affects the value of the firm stock.31

Although the post-trade abnormal returns associated with the three acquisitions by non-financial firms are not statistically significant, the prevalently positive values and superior pre-trade abnormal returns in comparison with other firms could lead us to accept the second hypothesis (H2). However, rather than reflecting the superiority of non-financial firms as company blockholders, we believe that the observed price reactions could be largely driven by investors’ expectations concerning the imminent takeover. All the three companies were in fact subject to a takeover by the block acquirer within six months of the block trade. The increases in the stock price might hence be a mere reflection of the increased likelihood of the investors’ vote being pivotal in a control contest and, consequently, of the expected additional payment that they will fetch in such a contest. In fact, when an event alters the expectations of a contested acquisition, a fraction of the private benefits of control may be reflected in the stock price and affect its variability in a substantial way (Zingales, 1995).32 This could also be one of the factors explaining the observed differences in the abnormal returns surrounding the block acquisitions by PIFs and industrial firms. Considering the less stringent takeover provisions (see Section 2) and the general scepticism on the PIFs’ active role in firm restructuring, it is indeed less likely that an acquisition of a block by a PIF would be actually perceived as a signal of an imminent takeover.

4. Block trades and the private benefits of control

4.1 Theoretical framework and hypotheses

Besides providing them with the incentive to actively monitor a firm's management, large share blocks confer on their owners some benefits that are not shared with other owners. Private benefits may take the form of the excessive compensation of those in control, large perquisites at the cost of minority shareholders, freeze-out mergers or the diversion of firm value, for example through acquiring inputs from other companies owned by large shareholders (managers) (Hart, 1995: 192). These are so-called pecuniary private benefits and have been widely emphasized in the literature. However, controlling owners may also benefit from synergies in production or individual prestige (non-pecuniary private benefits).33 At any rate, the expectations of some benefits in addition to the fraction of expected dividends that block buyers anticipate receiving should be reflected in the price premiums attached to share blocks. That is, investors are willing to pay more when the acquired blocks confer on them some degree of influence in the firm. However, blocks could also trade at a discount since monitoring is costly and requires effort. Moreover, when acquiring a large block, investors moreover give up some of the benefits of diversification and liquidity.

At any rate, several empirical studies (for instance, Barclay and Holderness, 1989) show that the ‘additional’ benefits of holding large blocks compensate for the associated costs. Share blocks trade at a premium in relation to the stock-exchange price. Even more, blocks are also valuable when not conferring majority control (Dyck and Zingales, 2004; Franks and Mayer, 2000; Mikkelson and Regassa, 1991; Nicodano and Sembelli, 2004; Trojanowski, 2002). There are several reasons why a minority share block may be valuable. First, the extraction of private benefits may in fact take place also in firms with many large owners (Zwiebel, 1995). In this regard, holding a block makes it more likely that the owner will participate in the controlling coalition and enjoy rents. Second, a blockholder may be courted by the raider in a takeover attempt, collude with management in exchange for a favour and enjoy control amenities from sitting on corporate boards. Third, through monitoring the management, blockholders acquire private information that puts them in a superior position vis-à-vis competitive investors (Tirole, 2006: 404). These benefits could be particularly prominent in firms whose minority blockholders are not facing any controlling (majority) owner, when monitoring capital is scarce and in an environment (as in transition) characterized by low transparency of corporate actions and, consequently, high information asymmetries. Hence,

Hypothesis 3 (H3): Given the relatively dispersed ownership structure and insufficient disclosure of corporate actions, minority blocks should provide Slovenian blockholders with some benefits that are not enjoyed by small shareholders. Hence, we expect the share blocks to trade at a premium relative to the exchange price.

Assuming that the extraction of private benefits is costly and that a constant degree of control is obtained above a certain threshold, the incentives to extract private benefits and the corresponding premium should on average decrease with the ownership stake. However, in the case of minority blocks we would expect that a higher ownership share assigns the block buyer a stronger position against the management and other owners and, consequently, a greater probability of enjoying private benefits. Moreover, low investor protection, insufficient transparency, no real sanctioning of related-party transactions and the general lack of knowledge of good governance practice considerably decrease the costs of private benefits expropriation. All these factors could reduce the ‘alignment effect’ related to increases in the ownership of the largest blockholders.34 Consequently,

Hypothesis 4 (H4): While controlling for firm-specific characteristics, block premiums on average increase with the percentage of shares traded in the block and with the relative voting power gained by the block buyer.

4.2 Data and methodology

Our database consists of 31 blocks of 5–25 percent of firm shares traded on the Ljubljana Stock Exchange in 2000/2001. Information on the number, size of shares traded and the total value of the trades was downloaded from the Ljubljana Stock Exchange and the trading archive of the business review Gospodarski vestnik. Unfortunately, these sources provide no information on the identity of the parties involved in a block trade. However, we were able to identify the buyers and sellers by relying on two different sources: a) articles referring to block trades from the business daily Finance; and b) the Shareholders’ Register of the Central Clearing Securities Corporation. As already stated (see Section 3.3), we exclude blocks exchanged between two associated PIFs and between a PIF and its management company (6) and three repurchases of shares by the firms (3) since these transactions do not represent an actual change of control.

In estimating the private benefits of control, we rely on Barclay and Holderness (1989) who define the private benefits of control as the relative difference between the price paid for a share within a negotiated block trade and its post-transaction exchange price.35 As evidenced in Table 3, Slovenian shareholders on average exchanged blocks that transferred approximately 10 percent of voting rights. Despite the relatively small block size, these blocks were acquired at an average premium of 46.7 percent, representing around 5.8 percent of a firm's equity. Considering the size of the blocks traded and the fact that these trades did not always represent the acquisition of the largest block,36 we can conclude that in Slovenia block ownership allows relevant private benefits of control (Hypothesis 3). What then are the main factors driving the excess price paid for a block? In order to answer this question and confirm that the evaluated block premiums actually reflect the private benefits of control, we perform a cross-sectional regression analysis highlighting the determinants of block pricing in Slovenia. Descriptive statistics, with the respective P-values for mean and median, and definitions of the variables used in the regression models are reported in Table 3.

Table 3. Descriptive statistics for variables used in the regression analysis of block premiums
 NMean (t)SDMedian (z)MinimumMaximum
  1. Note: Pre-trade premiums (in percent) are calculated as ((Pbi – Pmi)/Pmi)*100, where Pbi is the price paid per share in the block and Pmi is the closing price three days prior to the announcement of a block trade. Post-trade premium has the same definition as Pre-trade premium but refers to the closing price two days after the announcement of the block trade instead of the pre-trade closing price. The missing values in four observations were replaced by the closing price one week after the announcement of the trade. The Standardized pre- and post-trade premiums are simply pre-trade premiums (post-trade premiums) multiplied by the fraction of shares in the block. These definitions follow those of Barclay and Holderness (1989), Mikkelson and Regassa (1991) and Trojanowski (2002). Leverage is defined as the percentage of the book value of debt in the book value of the capital calculated at the end of the year preceding the block trade. Firm size is measured as total assets (Firm size) and as the market value of equity at the end of the year preceding the block trade (Firm size1). Both variables are in SIT and enter the regression models in logarithms. Roa is the operating profit divided by the value of assets (excluding cash and marketable securities). Roe is the net profit per unit of equity, while Roe Adj. is the ratio between net profits and the value of equity adjusted for revalorization. The specification of the Market to book value follows the definition of the Ljubljana Stock Exchange. The latter defines the book value per share as the book value of capital of revalorization divided by the number of a firm's shares. The buyer's power ratio is the ratio between the Shapley value and the ownership share of the buyer of the block. Where the buyer of the block was already among the firm blockholders prior to acquisition of the block, the buyer's power ratio equals the increase in the buyer's Shapley value due to acquisition of the block divided by the percentage of shares transferred in the block. The ocean's power ratio refers to the power ratio of small shareholders, namely of the shareholders whose voting rights do not exceed 5 percent. The ocean's power ratio is thus calculated simply as the Shapley value of the ocean divided by the percentage of shares not tied up in the blocks. The latter definition follows that used by Zingales (1995). The difference in the power ratio is then the difference between the power ratio of the block buyer and the power ratio of the ocean. With regard to this, also see Rydqvist (1998).

  2. (*) Significant at the 10-percent level; (**) significant at the 5-percent level; (***) significant at the 1-percent level.

Pre-trade premium (%)3146.77*** (3.30)78.8415.38*** (3.43)−28.57280.12
Post-trade premium (%)3146.69*** (3.55)74.7420.77*** (3.92)−16.66258.4
Standardized pre-trade premium (%)315.82*** (2.99)10.851.55*** (3.59)−1.5738.77
Standardized post-trade premium (%)315.96*** (3.08)10.741.49*** (3.94)−0.9238.78
Block size (%)319.82*** (11.23)4.878.12*** (4.86)5.0021.24
Firm size (total assets)315.87e + 09*** (200.14)5.11e + 094.37e + 09 (4.86)***2.09e + 092.75e + 10
Firm size1 (market value)312.39e + 09*** (3.83)3.48e + 091.24e + 09*** (4.86)1.81e + 081.96e + 10
Leverage (%)3159.89*** (7.39)45.1658.06*** (4.78)3.92149.7
Roe (%)312.90** (2.48)6.493.44*** (2.65)−17.6214.70
Roe Adj. (%)314.36* (1.83)13.216.33** (2.39)−43.5325.53
Roa (%)311.43** (2.59)3.071.32** (2.33)−3.247.16
Market to book value 310.89*** (14.72)0.330.79*** (4.86)0.461.82
Buyer's power ratio201.42*** (6.978)0.911.20*** (3.92)0.554.22
Ocean's power ratio200.68*** (4.95)0.610.65*** (3.92)0.022.90
Difference in the power ratio200.74*** (3.09)1.130.61*** (3.02)−1.813.77

4.3 Empirical results

The results of the regression analysis are presented in Tables 4 and 5. We tested several regressions’ specifications, using different dependent variables and adding other potentially relevant explanatory variables. For the sake of brevity, we only present the most significant results. Table 4 refers to standardized post-trade block premiums calculated in relation to the closing exchange prices after the announcement of a trade. The dependent variables in Table 5 are Pre and post-trade premiums.

Table 4. Determinants of block premiums in Slovenia. Dependent variable: standardized post-trade block premium in percent37 (ordinary least-squares regression with robust standard errors)
 Regression 1 Coefficient (t-test)Regression 2 Coefficient (t-test)Regression 3 Coefficient (t-test)Regression 4 Coefficient (t-test)
  1. Note: For a definition of the variables reported, see the notes to Table 3.

  2. (*) Significant at the 10-percent level; (**) significant at the 5-percent level; (***) significant at the 1-percent level.

Intercept45.258 (1.09)74.025 (2.10)30.03 (1.02)66.97 (2.17)
Block size1.037** (2.16)0.743* (1.80)0.585* (1.70)0.661* (1.70)
Leverage0.04 (0.87)0.021 (0.53)0.049 (1.29)0.025 (0.68)
Firm size−2.328 (−1.25)   
Firm size1 −3.637** (−2.19)−1.485 (−1.05)−3.463** (−2.33)
Roe Adj.  −0.337*** (−3.85) 
PIF buying   5.75** (2.32)
NF buying   5.44* (1.70)*
N31313131
R20.340.400.510.45
F2.22.45.52.7
Table 5. Determinants of block premiums in Slovenia. Dependent variable: pre-trade premium (regressions 1 and 2), post-trade premium (regressions 3 and 4) in percent (ordinary least squares regression with robust standard errors)
 Regression 142 (t-test)Regression 2 (t-test)Regression 3 (t-test)Regression 4 (t-test)
  • Note: For a definition of the variables reported, see the notes to Table 3.

  • (*)

    Significant at the 10-percent level;

  • (**)

    (**) significant at the 5-percent level;

  • (***)

    (***) significant at the 1-percent level.

Intercept327.18 (1.32)338.736 (1.18)276.566 (1.04)355.176 (0.699)
Difference in the power ratio17.025** (2.53) 8.833 (1.55) 
Firm size1−14.894 (−1.28)−15.341 (−1.17)−12.14 (−0.98)−15.184 (−1.09)
Buyer's power ratio 16.244* (1.70) 3.527 (0.39)
Ocean's power ratio −18.567*** (−2.85) −19.322*** (−3.95)
N20202020
R20.2730.27330.1160.144
F4.00**4.22**1.903.67**

The largest number of blocks in our sample was acquired by domestic firms (38.71 percent) and PIFs (32.26 percent), while banks appear as block buyers in eight cases (25.81 percent). Foreigners only acquired one block in our sample. The regression results mostly confirm a positive correlation between the percentages of shares transferred in the block and the size of the block premiums (Hypothesis 4). We find that PIFs and non-financial firms on average pay higher premiums than other owners (see regression 4 and the dummy variable PIF buying and NF buying).

We find that investors are on average willing to pay less for blocks in larger firms. This could be somehow expected considering the specifics and, consequently, the cash constraints of the block acquirers in Slovenia. A lower value is on average also attributed to the control of better performing companies.38 As suggested by Trojanowski (2002: 18), other factors besides the possibility of extracting private benefits (the expected dividends) may in fact be driving the acquisition of large blocks in these firms.39 Similarly to other empirical studies,40 the effect of firm leverage is not significant.

A positive relationship between the percentage of voting rights acquired in the blocks and the premium paid is also confirmed when applying an alternative measure of shareholders’ voting power – the Shapley Shubik power ratio41 (Table 5). Although limited by the small number of observations, these results are consistent with the findings of other empirical studies (Nicodano and Sembelli, 2004; Rydqvist, 1998; Trojanowski, 2002). In line with Hypothesis 4, investors pay more for blocks that carry greater probabilities of the buyer participating in the controlling coalition. The voting premium is positively related to the buyer's power ratio and significantly increases with the difference between the power ratio of the block buyer and the power ratio of the ocean (Difference in the power ratio). Further, the ocean's power ratio (regressions 2 and 4) has a negative and statistically significant influence on both post-trade and pre-trade block premiums. In fact, when control is contestable a certain fraction of private benefits and consequently the voting premium, is already incorporated in the stock exchange price, reflecting the expectation that voting rights attached to minority shares will become valuable in the case of a battle for control. The greater is the probability of minority investors being pivotal for a controlling coalition the higher is the probability of a contested acquisition. Consequently, the lower will be the price the buyer is willing to offer for a share block.

5. Concluding remarks

The introduction of private ownership constituted an important part of the socio-economic reforms in transition countries. Privatization should have provided the economic agents with new incentives and consequently constructed a basis for a better resource allocation and economic growth. However, the transfer of shares into private hands was carried out in a poor institutional environment, which opened up many opportunities for the dilution of firm value by the newly emerging owners. Slovenia is no exception here. For instance, the government failure to promptly address some of the privatization ‘mistakes’ (that is, the privatization gap) provided the privatization funds with further bargaining power and options to influence the final privatization outcome. In addition, the funds’ shares were held by a large number of small investors, whose interests were not fully protected and accounted for in the existing regulations. There are indeed several examples of institutional provisions regulating PIFs’ operations that were shaped under the influence of the PIFs (Giacomelli, Gregoric and Prasnikar, 2006) and which enabled the funds to capture substantial financial gains at the expense of minority shareholders and the development of the capital market (Mramor and Jasovic, 2004; Simoneti, 1999). These opportunities for rent extraction, combined with the potential opposition from inside owners or the state funds, certainly reduced the incentives of the PIFs to actively engage in the governance of the privatized firms and misdirected their incentives away from value maximization of the firms they owned. Indeed, the downturn in the prices following the announcement of the block acquisitions by PIFs indicates that these institutional owners somehow failed to fulfil expectations regarding their role as active players in the Slovenian corporate governance. This conclusion finds further support in other empirical studies. Slovenian firms dominantly owned by PIFs on average perform worse (Brezigar, Gregoric and Zajc, 2007; Knezevic, 2006). PIFs also do not outperform other owners with regard to replacing inefficient managers (Knezevic, 2006).

At the same time, substantial premiums for share blocks show that the investors are willing to pay substantial premiums, even when acquiring only minority control. We find that the premiums increase with the potential of the block buyer to form part of the controlling coalition. The latter in fact provides the blockholder with more ‘voice’ but also with the possibility to collude for extracting firm value and there are several other benefits that minority control provided to PIFs. The characteristics of Slovenian privatization, namely the division of firm capital in many blocks, gave them a board representation with relatively low ownership stakes (Prasnikar, Ferligoj and Pahor, 2004).43 This consequently provided them with access to firm inside information that they could rely on to reduce information asymmetry about firm prospects but also to expropriate in trading or to promote the interests of their founders (namely, banks). This information and participation in firm governance could analogously be of value to other minority blockholders (that is, non-financial firms), considering the poor investor protection and the limited disclosure of corporate actions which characterized the first years of the post-privatization period in Slovenia.

Footnotes

  • 1

    This is a revised and extended version of the paper that was awarded the 2003 Joseph de la Vega Prize for Emerging Markets. Special thanks go to Marco Becht and Janez Prasnikar for all their advice and support. We also thank Koen Schoors, the participants at the Global Corporate Governance Forum (Budapest, 3–5 July, 2003), the editor and the anonymous referee for their very useful comments. Financial support from the European Commission that funded the Research Training Network on ‘Understanding Financial Architecture’ <http://www.ufanet.org> is gratefully acknowledged.

  • 2

    For an excellent overview of privatization effects see for example Wright, Buck and Filatotchev (2005); Hanousek, Kocenda and Svejnar (2004).

  • 3

    See for example Trojanowski (2002) for Poland; Atanasov (2000) for Bulgaria.

  • 4

    In the absence of outside blockholder control and other forms of governance (i.e. developed capital markets, strong investor protection), and unless managers in a particular enterprise are market-oriented and employees are correspondingly compliant, managers and employees may form a coalition to preserve their entrenched interests and resist reforms (Wright, Buck and Filatotchev, 2005: 424). Thus, they may provide a strong countervailing power to other owners (i.e. PIFs).

  • 5

    These constituencies were the employees (due to self-management), operative leadership (managers), socio-political organizations and socio-political communities (Prasnikar and Svejnar, 1998).

  • 6

    Article 22 of the Privatization Law (also Law on the Ownership Transformation of Slovenian Companies, 1992) required the firms to transfer 10 percent of their shares to the Slovenian Pension Fund and 10 percent to the Slovenian Restitution Fund. Both funds are state controlled. The companies were furthermore obliged to transfer 20 percent of their capital to the Slovenian Development Fund for further sale to the PIFs for cash or in exchange for ownership certificates.

  • 7

    The term inside ownership refers to the ownership of employees (managers included), former employees and their relatives.

  • 8

    Simoneti and Gregoric (2004).

  • 9

    The only exception was the 20 percent limitation on the PIFs’ participation in the equity of any company that had a business relationship with a 10 percent or larger owner of the corresponding PIF. Otherwise, the PIF could have a 100 percent ownership stake in an individual company.

  • 10

    The Privatization Funds were supposed to gradually transform themselves into financial holding companies, investment funds and mutual funds. By the end of 2003, most of the PIFs (46 out of 52) actually opted to transform the whole or part of the PIF into a financial holding company, while only a few adopted the form of mutual funds and investment companies (4 and 11, respectively).

  • 11

    The privatization gap refers to the discrepancy between the (estimated) capital available for privatization and the value of the ownership certificates that were distributed to the population. The government was expected to fill the gap by selling-off remaining State ownership.

  • 12

    Simoneti and Gregoric (2004).

  • 13

    Workers’ associations are special legal entities constituted by workers to carry out internal buy-outs in privatization.

  • 14

    In 2002, this was the case in 17 companies. In 89 percent of these cases we observed that, apart from the block held by the largest ownership PIF, at least one of the other blocks was held by one or more PIFs managed by the same management company as the largest PIF (affiliated PIFs). Since we expect these PIFs to vote together, we added the shares of the affiliated PIFs to the voting stake of the largest PIF in the affiliation when calculating the voting rights of the largest shareholders (Table 1 and Table 2).

  • 15

    The main requirement for funds’ listing on the stock exchange was that all vouchers collected were exchanged for company shares. Initially, most of the PIFs found it difficult to fulfil this criterion due to the huge discrepancy between the capital available for privatization and the ownership certificates distributed in privatization (privatization gap), which was reflected in the PIFs portfolios. The Slovenian government managed to fill this gap only by the beginning of 2002. All this caused substantial delays in funds’ formal transformation and listing on the stock exchange. In addition, once listed, the PIFs shares were traded at large discounts. This largely limited the market discipline (i.e. the investors’ willingness to exit). For more, see Mramor and Jasovic (2004).

  • 16

    For more on the expropriation of minority investors, see Simoneti (1999).

  • 17

    Regression equations and statistical tests used are in the Appendix. For more on event studies, see Campbell, Lo and MacKinlay (1997).

  • 18

    These trades represent 81 percent of all the large share blocks (between 5 and 25 percent) transferred on the Ljubljana Stock Exchange in the period of our analysis.

  • 19

    There is one exception. If the block trade was also the last deal of the day, the average daily price instead of the closing price (which in this case is the price of the block) was used in calculating the daily returns.

  • 20

    At the time of our analysis, every trade of securities of a value exceeding EUR 140,000 (block trade) had to be reported to the Ljubljana Stock Exchange on the same day if it was settled at least half an hour before the Stock Exchange closed, otherwise on the first day after the trade. The Stock Exchange published information on block trades on its website within 30 min of receipt of the notification. Further, information on block trades was provided by the business daily Finance and the daily newspaper Delo on the first day following the trade.

  • 21

    To adjust for missing returns we followed the ‘Eventus’ procedure. See Appendix B for details.

  • 22

    See Appendix B for details.

  • 23

    See Patell (1976) and ‘Eventus Technical Reference’, the technical reference of the registered trademark for the software used to do event study analyses, Eventus (see references).

  • 24

    See Asquith (1983) and ‘Eventus Technical Reference’.

  • 25

    Corrado and Zivney (1992) showed that the rank test dominates over the t-test and the sign test.

  • 26

    These companies are all relatively small and rarely trade. In fact, in most of the cases excluded, the price of the shares did not change substantially after the block exchange. Thus, excluding these cases from our analysis should not introduce substantial biases to our results. Moreover, due to thin trading the share prices of these companies are anything but efficient.

  • 27

    Announcement date.

  • 28

    See Appendix B for details.

  • 29

    We replicated the analysis on a reduced sample consisting only of the securities included in the official or in the free market index (respectively, four and two companies). These stocks are the most traded and have very few or no missing returns. The results from both the market model and the market-adjusted model confirm the conclusions drawn for the entire sample. Again, the abnormal stock returns are positive from approximately nine days prior to the event and there is a decline in stock returns starting from the first day after the announcement of the trade.

  • 30

    Results of the market-adjusted model are even stronger with reference to the magnitude and significance of the positive effect on returns over the windows (−2, 0), (−9, 0) and (−9, +20), according to the st-test, adj-test and cs-test. The positive trend in abnormal returns highlighted in Figure 2 over the 20 days after the announcement of a trade has no statistical support. See the right side of Table A1 and Table A2 in Appendix A.

  • 31

    MSCARs from the market-adjusted model confirm the results, with the difference in the window (0 + 20), where MSCAR is positive, even if not significant. However, this positive value might be due to a misspecified market-adjusted model for a sample that includes companies that are not part of the market indexes, as already explained.

  • 32

    Zingales argues that the value of the vote increases only when a change in control is contested. The probability of a contested acquisition should be higher when there are multiple large shareholders with similar stakes (Zingales, 1995: 1063). The latter holds good for each of the three companies in our analysis.

  • 33

    For more, see, for example, Barclay (1992).

  • 34

    For more on the ‘alignment effect’ see, for instance, Burkart, Gromb and Panunzi (2000).

  • 35

    Indeed, this estimation seems to measure private benefits (and not overpayment) and introduces smaller biases compared to alternative methods (Dyck and Zingales, 2004).

  • 36

    This is the case in 8 out of 20 cases for which we could identify the exact ownership structure after the block trade.

  • 37

    Firm size is measured as the logarithm of a firm's total assets (Firm size) or its market capitalization (Firm size1). All the other explanatory variables are in percent.

  • 38

    We measure performance as profit per unit of capital net of revalorization. In addition, we ran a regression with profit per firm share as a measure of a firm's performance and the expected dividend stream. The impact was still negative but not statistically significant (t-statistics = −1.35), while the relationship between the percentage of shares acquired and the premium turned out positive and significant (t-statistics = 2.16). The impact of the variables ‘leverage’ and ‘size’ remained insignificant and of the same sign.

  • 39

    We included other additional explanatory variables (for instance, the logarithm of the value of a block traded as a proxy for liquidity costs, the ratio of dispersed shares, the market-to-book value etc.). They turned out not to be significant and to have no influence on the overall conclusions of our regression analysis. We also ran the same regressions with reference to the price one week after the trade. The coefficient of the variable ‘Block size’ remained positive, but lost significance. For the sake of brevity, these results are not reported.

  • 40

    Barclay and Holderness (1989); Trojanowski (2002); Banerjee, Leleux and Vermaelen (1997); Nicodano and Sembelli (2004).

  • 41

    For more on the calculations of the power index, see Milnor and Shapley (1978).

  • 42

    Controlling for other variables, such as prior firm performance and leverage, does not improve the model (all variables are highly insignificant).

  • 43

    Board seats are mostly divided between the employees (36.5 percent), non-financial companies (15.82 percent) and PIFs (15.16 percent). Except for a few cases, board representation automatically follows the PIFs’ block holdings. For more see Gregoric (2003).

  • 44

    See Patell (1976).

  • 45

    Corrado and Zivney (1992) showed that the rank test dominates over the t-test and the sign test.

Appendix A

Table A1. Daily mean standardized abnormal returns and test statistics for the entire sample (15 companies) – market model and market-adjusted model
AD = 0Market modelMarket-adjusted model
MSARst-testcs-testr-testMSARst-testcs-testr-test
  1. Notes: The st-test is the standardized abnormal return test and is asymptotically normally distributed; the cs-test is based on a cross-sectional variance estimator and has a Student's-t distribution with (N–1) degrees of freedom; the r-test is the rank test, asymptotically normally distributed. Values significant at the 1 percent, 5 percent and 10 percent levels are identified by ***, **, *, respectively.

−200.170.640.87−0.730.311.211.41−0.12
−19−0.13−0.50−0.510.76−0.14−0.54−0.49−0.12
−180.000.010.011.290.060.230.270.71
−170.010.020.020.150.150.560.410.40
−16−0.30−1.15−1.630.89−0.16−0.62−0.780.42
−15−0.04−0.15−0.19−0.150.050.180.211.20
−140.180.710.51−1.290.552.13**0.88−1.69*
−13−0.62−2.38**−1.14−0.46−0.48−1.86*−0.910.09
−12−0.29−1.13−1.720.91−0.12−0.47−0.45−0.56
−11−0.17−0.67−1.33−1.01−0.03−0.14−0.27−0.16
−100.371.440.350.130.511.98**0.470.71
−90.722.77***1.051.69*1.827.04***1.11−0.80
−80.351.371.31−0.250.652.52**1.56−0.21
−7−0.03−0.11−0.140.73−0.01−0.02−0.03−0.05
−60.100.380.29−1.490.331.290.990.07
−5−0.50−1.95*−1.09−1.09−0.45−1.76*−0.94−1.18
−40.050.190.12−2.05**0.110.420.232.52**
−3−0.02−0.09−0.050.510.230.900.48−1.62
−20.963.70***2.07*1.291.003.89***1.81*0.63
−10.662.56**0.802.38**0.933.62***1.050.28
0−0.16−0.61−0.640.35−0.05−0.18−0.160.80
1−0.46−1.78*−1.720.05−0.40−1.55−1.021.39
20.220.840.830.280.371.421.260.96
30.100.370.360.430.090.340.28−0.75
40.150.580.571.67*0.250.960.76−0.14
5−0.01−0.05−0.030.15−0.04−0.17−0.09−1.69*
6−0.22−0.87−1.060.30−0.30−1.16−1.21−0.96
70.341.310.96−0.860.672.61***1.411.08
8−0.15−0.57−1.29−0.760.080.310.530.71
9−0.40−1.55−0.72−0.66−0.06−0.24−0.08−0.85
100.863.31***1.11−0.730.813.14***0.981.03
11−0.56−2.17**−1.23−1.09−0.52−2.03**−1.141.46
12−0.23−0.91−2.46**−0.08−0.19−0.75−1.76*−1.39
130.140.541.010.940.491.92*1.16−0.07
14−0.32−1.24−1.85*−0.81−0.42−1.63−1.34−1.15
150.431.67*1.05−0.330.762.93***1.030.00
16−0.13−0.51−0.89−1.01−0.11−0.44−0.50−0.31
17−0.40−1.55−0.821.59−0.20−0.78−0.381.39
180.210.810.330.560.210.810.31−0.09
190.060.230.21−0.71−0.07−0.27−0.170.24
20−0.27−1.06−1.50−1.49−0.14−0.52−0.77−2.16**
Table A2. Mean standardized cumulative abnormal returns and corresponding significance tests over different event windows for the 15 securities of the whole sample
 Market modelMarket-adjusted model
MSCARst-testadj-testcs-testr-testMSCARst-testadj-testcs-testr-test
AD-10, AD + 100.161.200.622.38**0.450.632.47**2.45**2.21**0.42
AD-2, AD0.873.23***3.36***1.96*2.32**1.074.17***4.15***2.33**0.99
AD-1, AD0.401.391.550.641.93*0.622.42**2.41**1.030.76
AD-9, AD0.582.57**2.26**2.17**0.341.425.55***5.52***3.22***0.29
AD-9, AD + 200.081.120.300.56−0.241.094.24***4.21***2.33**−0.06
AD, AD + 20−0.19−0.68−0.74−1.87*−0.480.251.000.980.72−0.11
Table A3. Mean standardized cumulative abnormal returns and corresponding significance tests for the companies taken over within six months of a block trade
 Market modelMarket-adjusted model
MSCARst-testadj-testcs-testr-testMSCARst-testadj-testcs-testr-test
AD-10, AD + 100.110.280.181.521.470.220.370.371.961.11
AD-2, AD2.553.90***4.42***1.681.632.484.29***4.30***1.852.27**
AD-1, AD2.123.00***3.67***1.051.65*1.973.40***3.40***1.081.36
AD-9, AD1.672.64***2.88***1.681.82*1.953.38***3.38***1.971.60
AD-9, AD + 200.782.25**1.341.940.821.743.00***3.01***5.84**0.23
AD,AD + 20−0.010.41−0.02−0.18−0.070.400.690.690.53−0.73
Table A4. Mean standardized cumulative abnormal returns and corresponding significance tests for the companies that remained independent in the six months following a block trade
 Market modelMarket-adjusted model
MSCARst-testadj-testcs-testr-testMSCARst-testadj-testcs-testr-test
  • Notes on Tables A2, A3, A4: the st-test is the standardized abnormal return test for the null MSCAR = 0 over the window and is asymptotically normally distributed; the adj-test is the test corrected for any correlation of the abnormal returns over the multiple day windows (technical reference of ‘Eventus’); the cs-test is based on a cross-sectional variance estimator and has a Student-t distribution with (N–1) degrees of freedom; r-test is the rank test, asymptotically normally distributed.

  • *

    Significant at 10 percent;

  • **

    ** Significant at 5 percent;

  • ***

    *** Significant at 1 percent.

AD-10, AD + 100.171.190.602.09*−0.240.742.57**2.55**2.09*−0.11
AD-2, AD0.451.66*1.551.251.610.722.52**2.49**1.61−0.10
AD-1, AD−0.030.04−0.10−0.051.200.291.010.990.470.12
AD-9, AD0.311.551.091.63−0.521.294.51***4.48***2.54**−0.49
AD-9, AD + 20−0.100.14−0.34−1.09−0.650.933.24***3.20***1.61−0.17
AD, AD + 20−0.23−0.97−0.81−1.90−0.460.220.770.750.520.24
Table A5. Daily mean standardized abnormal returns and test statistics for the companies taken over within six months of a block trade – market model and market-adjusted model
AD = 0Market modelMarket-adjusted model
MSARst-testcs-testr-testMSARst-testcs-testr-test
  • Note: see Notes on Table A1.

  • *

    Significant at 10 percent;

  • **

    ** Significant at 5 percent;

  • ***

    *** Significant at 1 percent.

−200.110.200.571.570.180.321.151.52
−190.040.080.05−0.920.010.010.01−0.93
−180.691.201.88−0.540.821.411.71−0.49
−170.811.410.88−1.081.232.13**1.01−0.20
−16−0.54−0.93−0.60−1.08−0.28−0.48−0.26−1.03
−15−0.68−1.18−4.76**−1.08−0.65−1.12−3.42*0.05
−14−0.55−0.96−2.68−0.54−0.43−0.75−1.85−0.10
−130.360.620.78−1.79*0.520.891.480.69
−12−1.09−1.88*−2.160.81−1.12−1.95*−2.12−0.79
−110.030.050.11−0.220.060.110.31−0.89
−100.470.811.480.490.631.093.50*−0.30
−90.290.500.642.28**0.250.440.551.87*
−8−0.01−0.02−0.041.300.110.190.610.34
−7−0.22−0.38−0.520.43−0.30−0.53−0.610.30
−6−0.15−0.26−0.49−1.080.150.260.29−0.69
−5−0.02−0.04−0.03−0.110.200.340.210.74
−40.150.260.510.270.140.240.371.08
−30.851.471.850.111.232.14**6.37**−0.84
−21.462.54**1.240.491.532.65***1.332.02**
−13.435.93***1.911.034.016.95***1.910.98
0−0.97−1.68*−1.081.30−1.23−2.13**−1.420.93
10.170.300.200.760.570.990.40−0.15
20.480.840.920.270.560.971.191.23
30.250.420.86−1.250.320.551.47−2.36**
4−0.52−0.90−0.920.11−0.54−0.93−0.88−0.15
50.250.440.280.000.470.810.380.20
60.160.280.241.520.080.130.11−1.48
71.853.20***1.14−0.272.484.30***1.071.87*
8−0.32−0.55−3.62*−1.08−0.40−0.69−2.14−1.23
9−0.27−0.47−0.420.38−0.53−0.93−0.62−0.54
101.091.88*0.80−0.221.081.87*0.751.23
11−0.90−1.56−1.23−1.08−0.94−1.62−1.420.30
12−0.25−0.43−2.761.14−0.25−0.42−1.70−1.03
13−0.12−0.20−0.860.38−0.20−0.35−0.94−0.10
140.090.150.33−2.39**0.120.200.42−0.25
150.140.240.291.030.050.080.10−1.08
160.070.120.77−1.030.120.220.67−0.34
170.761.311.410.380.761.311.53−0.93
18−0.03−0.05−0.09−1.080.00−0.01−0.02−0.93
19−0.12−0.20−4.34**0.54−0.07−0.12−0.671.23
20−0.72−1.25−0.720.27−0.59−1.02−0.650.25
Table A6. Mean standardized abnormal returns and test statistics for the companies that remained independent in the six months following a block trade market model and market-adjusted model
AD = 0Market modelMarket-adjusted model
MSARst-testCs-testr-testMSARst-testcs-testr-test
  • Note: see notes to Table A1.

  • *

    Significant at 10 percent;

  • **

    ** Significant at 5 percent;

  • ***

    *** Significant at 1 percent.

−200.190.670.78−1.510.351.201.25−0.86
−19−0.17−0.60−0.681.22−0.17−0.60−0.580.33
−18−0.16−0.57−0.741.59−0.13−0.45−0.580.96
−17−0.19−0.68−0.670.68−0.13−0.43−0.390.50
−16−0.23−0.81−1.87*1.43−0.13−0.45−1.080.93
−150.120.430.530.360.220.760.881.19
−140.371.270.85−1.070.802.76***1.03−1.67*
−13−0.86−2.99***−1.320.39−0.73−2.52**−1.14−0.24
−12−0.09−0.32−0.710.550.130.450.48−0.19
−11−0.22−0.78−1.49−0.94−0.06−0.20−0.370.26
−100.341.190.25−0.100.481.67*0.350.86
−90.832.86***0.960.652.217.65***1.08−1.72*
−80.441.541.34−0.890.792.73***1.52−0.38
−70.020.080.090.550.070.240.24−0.19
−60.160.560.38−1.020.381.310.930.41
−5−0.62−2.15**−1.11−1.07−0.62−2.14**−1.09−1.55
−40.020.080.05−2.24**0.100.350.172.03**
−3−0.24−0.82−0.480.47−0.02−0.07−0.03−1.24
−20.832.89***1.611.090.873.02***1.34−0.33
−1−0.03−0.10−0.031.95*0.160.570.19−0.19
00.040.130.18−0.260.250.860.970.36
1−0.63−2.17**−2.33**−0.31−0.64−2.23**−1.86*1.48
20.150.510.460.160.321.100.910.38
30.070.240.211.040.030.110.080.38
40.321.121.121.67*0.441.541.20−0.07
5−0.08−0.26−0.170.16−0.17−0.59−0.30−1.82*
6−0.32−1.11−1.50−0.42−0.39−1.36−1.48−0.26
7−0.04−0.15−0.32−0.760.220.771.400.19
8−0.10−0.36−0.73−0.260.200.691.181.31
9−0.44−1.53−0.64−0.860.060.200.06−0.60
100.812.81***0.86−0.650.742.58***0.740.45
11−0.48−1.67*−0.87−0.60−0.42−1.45−0.761.34
12−0.23−0.80−1.94*−0.63−0.18−0.63−1.34−0.91
130.200.701.220.780.672.31**1.28−0.02
14−0.42−1.46−2.13*0.31−0.55−1.92*−1.45−1.05
150.511.76*1.00−0.830.933.23***1.020.53
16−0.18−0.63−0.99−0.55−0.17−0.60−0.62−0.14
17−0.69−2.39**−1.201.46−0.44−1.53−0.681.86*
180.270.930.331.090.260.910.310.36
190.110.380.30−0.99−0.07−0.25−0.14−0.36
20−0.16−0.56−3.50***−1.67*−0.02−0.07−0.29−2.32**

Appendix B

Formulas used for abnormal returns and statistical tests

In Section 3.1 we study the effects of block trades on stock returns by performing an event study analysis. In order to do this, the following main elements need to be specified: the event and timing of the event; the benchmark model for normal stock return behaviour and the abnormal return of the stock around the event dates (De Jong, 1996: 2). The events are the exchange of share blocks. The benchmark models used for estimating normal stock returns are the market and the market-adjusted models. The abnormal returns are measured as the prediction errors over the ‘event window’. That is, the period around the event over which stock returns are examined. The abnormal return for stock i on day t is calculated as follows:

image

The market-adjusted model can be viewed as a restricted market model. While in the latter the coefficients are estimated over the estimation period, in the former fixed values of coefficients α = 0 and β = 1 are imposed, namely the normal stock returns are approximated by the market return (measured by the market index). This restricted model is particularly appropriate in the analysis of events for which the limited data availability prevents an accurate estimation of the coefficients (for example, in the case of Initial Public Offerings). Given the presence of missing returns, this might also be the case in our study.

To adjust for missing returns, we followed the ‘Eventus’ procedure. If q is the number of non-trading days (the number of days with no closing price reported), the abnormal return for the first post-missing day is calculated as follows:

image

In order to correct for differences in the stock return variance and to release the strong assumption of cross-sectional homoscedasticity (De Jong, 1996: 7), we standardise the abnormal returns in the following way:

image

The variance inline image is then estimated by:

image

When the abnormal returns are calculated as residuals from the estimated market model, an unbiased estimate of inline image is given by:

image

where

image

Rmk is the return on market index observed on day k;

inline image is the mean market return over interval inline image through inline image used to estimate the parameters for i (the estimation period); and

Di is the number of non-missing trading day returns in the estimation period of i.

This unbiased estimate adjusts for the fact that the coefficients α and β are estimated from the market model and it is further corrected (following the ‘Eventus Technical Reference’) for the multi-period character of the returns in the event window.

Hence, we compute the following measures:

  • 1The cross-sectional average of abnormal returns at date t:
    image
  • 2The cross-sectional average of standardized abnormal returns at date t:
    image
    This measure is a weighted average of the abnormal returns of individual stocks, with the weights inversely related to the estimated time-series standard deviation of the corresponding abnormal returns.
  • 3Cumulative abnormal return, defined as the cumulative sum of the abnormal returns of stock i over the event window (t1, t2):
    image
  • 4Standardized cumulative abnormal return of stock i over (t1, t2):
    image
    where
    image
    and L = (t2 − t1 + 1) is the length of the window (t1, t2) over which we cumulate the abnormal returns.This variance estimate is used to construct a corrected version of the ‘standardized abnormal return test’ (or ‘Patell test’) below. The correction affects only multiple day windows and accounts for the eventual correlation of the abnormal returns within the window. The latter might occur due to the fact that the abnormal returns are all functions of the same market model intercept and slope estimators. The bias for uncorrected tests is more serious in longer event windows (‘Eventus’, technical reference: 81). For the market-adjusted model, abnormal returns are calculated as the difference between the realized return of the security over the event window and the return on the market index. In this case, there is no estimation of the mean and the expression for the variance is simply:
    image
  • 5Mean cumulative abnormal return defined as the average of the cumulative abnormal returns across the observations:
    image
  • 6Mean standardized cumulative abnormal return, the average of the standardized cumulative abnormal returns across the observations:
    image

The statistical significance of abnormal returns is assessed by applying three different tests:

  • (i) The ‘standardized abnormal return test’, which is asymptotically normally distributed and assumes that the ARs are cross-sectionally independent:
image

referred to as the st-test in the MSAR tables in Appendix A;

The corresponding test statistic for the null hypothesis that inline image is corrected for the correlation of abnormal returns over the window (t1, t2) (see definition 4):

image

referred to as the adj-test in the MSCAR tables in Appendix A;

We also employ the following test for the null hypothesis that inline image

image

referred to as the st-test in the MSCAR tables in Appendix A,

where

image

is a small sample correction term.

On the assumption of cross-sectional independence44 this statistic follows the standard normal distributions under the null.

  • (ii) A t-test that uses a cross-sectional variance estimator, which depends on the number of firms in the sample and is robust to an increase in the variance of the ARs around the event dates. On the assumption that abnormal returns are cross-sectionally independent and identically normally distributed, the test for day t in the event period is defined as:
image

referred to as the cs-test in the MSAR tables in Appendix A, where

image

By following the ‘Eventus’, we extend this method to multi-period windows to get the standardized cross-sectional test statistic for the null hypothesis that inline image:

image

referred to as the cs-test in the MSCAR tables in Appendix A, where

image

The expressions of the cross-variance estimates and t-tests for the non-standardized MARt and MCARt are analogous. According to Brown and Warner (1985), the cross-sectional test is well-specified for event–date variance increases but not very powerful. The standardized cross-sectional test is, on the other hand, well-specified and more powerful.

  • (iii) The rank test;45 this non-parametric test can be used for event studies with small cross-sections. It further solves the problem related to the non-normality of abnormal returns as well as the thin trading of stock. Here we adopt the rank test proposed by Corrado (1989). This test takes into account the magnitude of the abnormal returns, as the t-test does, but without the distributional assumptions which are necessary to implement the parametric t-test. The null is that the shift in the distribution of event date excess returns is zero, that is, it should be uniformly distributed under the null that event periods are not different from non-event periods. The rank procedure assigns a rank to each daily return for each firm where rank 1 indicates the smallest abnormal return. Hence, the expected rank over a window L2 = (t2 − t1) around t = 0 is (L2 + 1)/2. Letting Kit be the rank of the excess return ARit at event date t, the day 0 test statistic is:
image

referred to as the r-test in the MSAR tables, where

image

The test of the null is implemented by using the result that the asymptotic null distribution of this statistic is standard normal. Compared with the t-test, the rank test is expected to work better in small samples because it may converge faster to the normal distribution. In practice, non-parametric tests are used in conjunction with parametric tests to check the robustness of the conclusions based on the last ones.

When testing the significance of the cumulative abnormal returns over a multiple day window (t1, t2), we apply the following version of the rank test (assuming the independence of daily returns ranks within the window):

image

referred to as the r-test in the MSCAR tables, where inline image is the average rank across N observations, through days (t1, t2); inline image, and E is the number of non-missing returns in the event period.

Ancillary