- Email: [email protected]
Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America
Accepted Manuscript Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America Carlos Pombo, Rodrigo Taborda PII:
S1059-0560(17)30406-9
DOI:
10.1016/j.iref.2017.05.012
Reference:
REVECO 1432
To appear in:
International Review of Economics and Finance
Received Date: 21 November 2015 Revised Date:
10 May 2017
Accepted Date: 18 May 2017
Please cite this article as: Pombo C. & Taborda R., Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America, International Review of Economics and Finance (2017), doi: 10.1016/j.iref.2017.05.012. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Resubmitted version.R2
This version May 8 - 2017
SC
Abstract
RI PT
Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America
JEL Classification: G32, G34.
M AN U
This paper examines the relationship between firm value and blockholders in Latin America. Econometric results for a comprehensive data set of more than 550 firms listed in the six largest stock markets of the region support a positive effect of variables measuring the existence, contestability, dispersion and identity of blockholder on performance (Tobin’s Q) only for highly liquid stocks. The identity of the second largest blockholder (family, foreign, financial or the State) emerges as critical for such effects. The study supports that the voice and exit models are complementary rather than substitute mechanisms for firm corporate governance.
AC C
EP
TE D
Keywords: Multiple large blockholders, stock liquidity, firm value, corporate governance, Latin America
1
ACCEPTED MANUSCRIPT Resubmitted version.R2
1. Introduction Recent empirical literature on blockholder ownership has focused on the presence of multiple
RI PT
large shareholders and their interactions, and how these can affect firm management either through direct intervention—monitoring—or through share trading. Edmans and Manso (2011), for example, argue that the presence of multiple blockholders creates two conflicting governance effects. In the first, known as the voice mechanism or shareholder activism, multiple
SC
blockholders serve “as a commitment device to reward or punish the manager ex post for his actions.” By contesting control, blockholders can align interests to implement either profitable
M AN U
projects or simply to monitor managers. The second effect, known as the exit mechanism, occurs in firms with multiple blockholder structures and sees dispersed (and small) blockholders punish the largest shareholder, or management, by exiting the firm, i.e. by trading their shares, thus affecting firm value. This approach to corporate governance highlights that, by possessing better
TE D
information about firm value, smaller blockholders can use the exit strategy as a device to discipline the behavior of the largest blockholder. On the one hand, when blockholders contest control, a positive effect on monitoring appears, since the second or third blockholders are given
EP
more equal voting rights in circumstances where, due to other mechanisms, they cannot exert absolute control. These mechanisms may include the issuing of non-voting shares or multiple
AC C
share classes, pyramidal and cross-share ownership structures, or the existence of disproportionate board representation (Villalonga and Amit, 2010). On the other hand, a negative effect of blockholder ownership may result from the possibility that small shareholders might be subject to expropriation by large shareholders, turning the agency problem from one of controlling blocks to controlling minorities. In such circumstances firm management responds to the interests of the controlling owners, reducing firm value when
2
ACCEPTED MANUSCRIPT Resubmitted version.R2
large shareholders derive private benefits from related party transactions. In addition, an excess of monitoring by blockholders induces a loss of value, since management may refrain from certain profitable initiatives when a vigilante approach is followed by large shareholders.
RI PT
The paradigm of the benefits of a widespread ownership structure, mainly fostered after the prevalent, but not precisely right, view of diffuse ownership in the United States, led the academic agenda to study corporate governance in settings were large shareholders dominate the
SC
landscape.1 At the same time, this view defined a policy agenda towards developing countries in favor of increasing ownership dispersion as corporate governance mechanism. This perspective
M AN U
has evolved in line with the reality of blockholders as the dominant ownership structure in corporate finance. Moshirian et al. (2014) summarize how large ownership is also a prevalent corporate structure in developed countries. The growth of institutional investors worldwide during the last four decades as large shareholders explains in part the presence of blockholder
TE D
and investor activism within developed capital markets (Agrawal et al., 2011). Under this new reality, whereby widespread ownership has come to be used as a mechanism for increasing corporate governance and ultimately corporate value in developing countries, public
EP
policy is at a crossroads. Where to go? Should corporate governance be increased or ownership structures disseminated? The former approach has been widely accepted and implemented in
AC C
developing countries (McGee, 2009). Whether or not such positive initiatives exist in practice, beneficial effects on corporate value are not guaranteed. The latter approach, on the other hand, cannot be warranted, as a policy interventions aimed at reducing the concentration ownership are hardly an economic policy option. The highly concentrated ownership structure of public and private firms in Latin America provides a useful scenario for an empirical investigation of the 1
An example of the argument that ownership is widespread is provided by the work of Becht (2001), while Holderness (2009) has been pivotal in debunking this view.
3
ACCEPTED MANUSCRIPT Resubmitted version.R2
association between firm value and ownership in developing countries. Leaving aside the option of dispersed ownership or tighter corporate governance laws or guidelines, the remaining question is how intermediate ownership structures relate to firm value.
RI PT
Several papers have addressed some of these issues. The blockholder entry and market response was studied by Park et al. (2008), showing how the acquirer’s profile shapes abnormal returns differently. Ali et al. (2017) establish how better governance influence liquidity in
SC
Australian listed firms. And extending this work, Nadarajah et al. (2016) establish how more liquid firms have lower leverage once corporate governance mechanisms are accounted for.
M AN U
Closely related to Latin American experience with corporate governance, the work of Cueto (2013) examines corporate governance and ownership concentration and the effect upon firm value. Given that high ownership concentration is prevalent in the region, the author suggests that to minimize its negative effect, firms engage in corporate governance and institutional
TE D
investors engage in monitoring and restraint asset appropriation.
The paper contributes to the literature on multiple blockholders in empirically addressing two analytical alternatives that can shed light on approaches to increase firm value under a setting of
EP
high ownership concentration in emerging markets. First, we claim that blockholder direct monitoring or voice is an effective channel for reducing shareholders’ agency costs and
AC C
improving firm corporate governance but its effectiveness is conditional to stock liquidity. In other words, contestability among blockholders is only effective and shareholders’ exit strategy credible for high liquid stocks within financial markets where capital deepening is a structural issue. Second, we provide evidence that in firms with multiple blockholder ownership structure but without absolute control by the largest shareholder, the type of the second blockholder is central in explaining firm valuation.
4
ACCEPTED MANUSCRIPT Resubmitted version.R2
Ownership structures in Latin America led us to hypothesize that the second blockholder is crucial in this connection and that the situation described for the region might also elucidate the case of other emerging markets. Nearly 50% of the firms in this study have a controlling
RI PT
blockholder with equity rights above 50%, while fewer than 5% of firms within the sample have no kind of blockholder at all (i.e. all their shareholders have less than 10% of equity). The remaining 45% of firms in the sample exhibit multiple blockholders without direct control, but
SC
that are still of considerable size (more than 10% of equity), meaning that there is scope for forming controlling coalitions among the largest blockholders. The frequency distribution of
M AN U
ownership within this last group of firms shows that when the largest voting block colludes with the second largest, the coalition represents around 45% of direct votes, a figure that rises to 54% when the third top blockholder is included. High ownership concentration implies low separation ratios between ownership and control. Empirical country-level case studies on corporate
TE D
ownership in Latin America, based on real sector firms, show that separation ratios between ownership and control are close to 0.85 (Gutierrez and Pombo, 2009). This means that a controlling shareholder will need 42% of equity rights if they are to obtain control over a firm’s
EP
voting rights. Thus, focusing the empirical analysis on the role of blockholder coalitions and their effects on firm value, measured by direct votes, is fully justified by the circumstances on
AC C
the ground.2
Our work extends prior results of benchmark studies of multiple blockholders and firm value. Maury and Pajuste (2005) based on a sample of 136 Finnish firms for the 1993-2003 period, found that contestability and blockholder dispersion reduces firm value. They only report the regression coefficients and direct effect through Shapley coalitional value, the Herfindal 2
Indirect estimation of the wedge between ownership and control puts it at 0.75 - similar to the results of direct measurement. The indirect wedge indicator is defined as the Shapley value solution of a four-voting oceanic game to largest shareholder cash flow rights (Guedes and Loureiro, 2006).
5
ACCEPTED MANUSCRIPT Resubmitted version.R2
concentration and the differences in shareholder rights indices. There is no discussion on the role of stock liquidity. Similar results were found by Laeven and Levine (2008) for a sample of 1657 European firms but these results were limited to estimates for a cross section. The authors report
RI PT
a negative premium of cash flow rights dispersion between the largest and second largest blockholder on firms’ Tobin’s Q without the presence of a controlling owner with majority control rights (ownership share > 0.50). They did not consider any discussion on the role of
SC
market liquidity and blockholder trading as a blockholders’ disciplinary device. Konijn et al. (2011) study the influence of blockholder presence, size and dispersion on firms’ Tobin’s Q
M AN U
based on a sample of 3772 firm-year observations of US listed companies. They found a negative correlation between blockholder presence and size on firm value. They disentangle the blockholder effect produced by insiders and outsiders, where the former have a positive effect of firm value offset by the negative effect of outsiders. They control for stock liquidity in their
TE D
regression but they do not model blockholder contestability. Their results contradict what has been found for Continental Europe and Asia arguing that such results might depend on different regional and institutional features that need more research.
EP
The results of this study on the relationship between blockholders and firm value takes into account the dimensions of presence, contestability, control rights dispersion, and type of
AC C
blockholder. The empirical design presents a baseline regression that takes into account the full set of interactions among stock liquidity and blockholder ownership structure (refined in terms of presence, contestability, and dispersion). All results are presented in terms of the marginal effect of blockholder ownership on firm value conditioning for liquidity and blockholder type. In previous studies analyzing blockholder structure and corporate value, Latin America has not been studied in depth. The region is a classic example of emerging economies in terms of
6
ACCEPTED MANUSCRIPT Resubmitted version.R2
economic and financial development. Based on a comprehensive dataset of 562 Latin American corporations listed in the six largest stock markets of the region from 1997 to 2011, the econometric results support a positive effect from variables measuring contestability, dispersion
RI PT
and type of blockholder on Tobin’s Q only for high liquid stocks. In this sense, we report that the presence of multiple blockholders implies a premium of 4.5% on firms’ Tobin’s Q. Nonetheless, this premium increases by up to 13.5% after controlling for the interacting term of stock
SC
liquidity, and it rises to 17.1% if we expand interaction for the second blockholder type dummy variables (family, local, foreign, financial and state). Under this specification, we found that
M AN U
regardless of whether the second blockholder is a financial firm or foreign investor, there is a statistically and economically significant effect on firm value.
The effect of contestability and dispersion on firm value is in line with the presence effect when liquidity is considered. The overall marginal effect increases and becomes statistically robust
TE D
after including the interacting terms of stock liquidity and second blockholder type. Thus, the approach to address the blockholder effect goes beyond presence or size, and expands to shareholder dispersion and the ability to form coalitions and ultimately become a driving force
EP
behind firm value.
Specification tests for linear panel data models favored fixed effects within estimates in
AC C
contrast to the transformation of random effects. We include several control variables at firm level and we also control for country macroeconomic, financial, and institutional variables that are sensitive to local and foreign investors’ systematic risk perceptions. The remainder of the document is structured as follows: Section 2 presents the theoretical framework and development of the working hypotheses; section 3 analyzes the data and
7
ACCEPTED MANUSCRIPT Resubmitted version.R2
variables included in the empirical model; section 4 presents the econometric analysis; and section 5 concludes.
RI PT
2. Theoretical Framework and Hypotheses 2.1 Blockholders contest for control
Theoretical models and empirical studies in corporate governance have addressed settings with more than one large blockholder, circumstances that allow some level of contestability to emerge
SC
on the basis of coalitions that may be formed. Two competing views have been proposed. The first refers to the positive effect of blockholder presence, due to their monitoring role and the
M AN U
second to the benefits of control when large groups of shareholders form coalitions in order to conduct tunneling strategies that enable them to divert a firm’s cash flow. Regarding the first view, Bloch and Hedge (2001) analyzed a firm’s governance framework in which two large blockholders and dispersed shareholders interact in the shareholders’ general
TE D
meeting. These authors show that control contestability between two blockholders that are seeking to attract the votes of small shareholders tends to reduce rent extraction. Bennedsen and Wolfenson (2000) developed a model of firm control where firm founders dilute their (initial)
EP
control by distributing proportional equity holdings to several large shareholders, allowing them to form a coalition, through which, in turn, they can obtain control and curb private benefit
AC C
extraction. Gomes and Novaes (2006) model corporate governance according to two scenarios. The first occurs when insiders, or the initial founders of a firm, face investment opportunities that are difficult to evaluate to the extent that sharing control creates bargaining problems. As a result, ownership structures in which there is
only one large shareholder monitoring
management represent the most efficient system for protecting minority shareholders. However, they also find that in weak legal environments, the optimal ownership structure involves shared
8
ACCEPTED MANUSCRIPT Resubmitted version.R2
control between multiple blockholders. In these circumstances, the presence of multiple blockholders is positively associated with firm value. The second view concerns the benefits of control that accrue when large groups of shareholders
RI PT
form coalitions in order to conduct tunneling strategies that enable them to divert cash flow. Zwiebel (1995) argues that investors may choose to hold a significant block equity in a firm because it confers them partial benefits of control, especially in those cases where the next
SC
blockholder is insufficiently large and they feel the need to form a coalition.
The theoretical model of control contestability had its origins in La Porta et al. (2002), who
M AN U
modeled managerial rent extraction, and in Maury and Pajuste (2005), who extended the baseline model to reveal contestability behavior more explicitly among multiple large shareholders. The model has two main assumptions: that the existence of multiple large shareholders can reduce profit diversion (i.e., that control contestability is value-enhancing), and that diversion of profits
hypotheses of this paper:
TE D
by the controlling coalition is costly. The theoretical work of those studies supports the first two
EP
H1: Firm value (performance) increases with blockholder presence and higher contestability of power.
AC C
H2: If the probability of forming a controlling coalition between the largest shareholder and the next largest blockholders increases, rent diversion grows to the benefit of the coalition block, thus lowering firm value (performance).
2.2 The role of the second large blockholder Empirical research on corporate governance suggests that the type of controlling shareholder present in a given company is essential to understanding the link between ownership structure, agency costs and firm value. Claessens et al. (2002) find that the expropriation of minority shareholders is more likely when the largest controlling shareholder is a family or the state. The
9
ACCEPTED MANUSCRIPT Resubmitted version.R2
underlying assumption is that different types of blockholders have distinct incentives and abilities to monitor the controlling shareholder. Empirical results from regional or country case studies on the marginal effects of blockholder identity and firm value confirm that the kind of
RI PT
second blockholder is vitally important when it comes to contesting the agency costs of controlling owners. For instance, Jara-Bertin et al. (2008), based on sample of 1,200 European corporations, find that in firms where the two largest shareholders are family blocks decrease
SC
firm valuation.. Similarly, of Attig et al. (2009) who studied a sample of 1,252 East Asian public companies find that the presence of either a family or the state as second largest shareholder
M AN U
represents a value premium of between 0.23 and 0.96 on firms Tobin’s Q. In contrast, there is no significant effect if the second blockholder is a widely held firm.
Recent research on family firms has focused on analyzing the effects of the presence of multiple large shareholders on firm value. In two related works on family listed firms in Spain,
TE D
Sacristan et al. (2012, 2015) find that there is a premium of 3.5% on a firm´s return on assets (ROA) when a it has a second blockholder.
Research on the role of institutional investors, mainly from developed capital markets, shows a
EP
consistent positive effect of the presence of this type of investors on firm governance, efficiency and performance. Moreover, it is expected that independent institutional owners –e.g. investment
AC C
firms or banks- are more prone to behave as investors with high levels of control and to act as watchdogs of management because of their low monitoring costs and their capacity to collect information and implement risk management protocols. These investors have less natural potential for developing business relations with the corporations (Almazan et al. 2005). International evidence shows that the premium for the presence of institutional ownership is 0.12 on the Tobin’s Q ratio of a firm (Ferreira and Matos, 2008). Recent studies for Latin America
10
ACCEPTED MANUSCRIPT Resubmitted version.R2
note the premium there as 0.08 (Pombo and De-la-Hoz, 2015). These conceptual elements and findings support the third hypothesis of this study: H3: The presence of an independent second blockholder enhances firm performance.
RI PT
2.3 The role of market liquidity and blockholder voice/trade mechanisms
A liquid stock market provides the opportunity for non-shareholders to become shareholders, it encourages management compensation, reduces managerial opportunism, and promotes
SC
investment. Blockholders’ direct intervention and management monitoring—voice models— might be effective whether market liquidity reduces the cost of monitoring and discloses more
M AN U
private information on assets value or not. The main argument by which blockholder contest can act as a powerful internal firm governance mechanism is refined once market liquidity is studied. This argument can be traced back to Hirschman (1970) who investigated under what conditions the voice option prevails over the exit option and vice versa in shaping the optimal behavior of
TE D
business firms, organizations, and economic institutions. Theoretical models in the corporate finance literature predict different outcomes depending on whether stock liquidity enhances blockholder intervention or not. Maug (1988) develops an extensive theoretical game where a
EP
large shareholder (LSH) buys a firm’s shares and then decides to monitor or trade. He argues that liquidity can enhance monitoring because it allows the LSH to buy additional stakes at a price
AC C
that is lower than the expected equity price after blockholder intervention. In the same vein, Kahn and Winton (1998) within the context of institutional investors as blockholders, develop a two stage dynamic game where stock liquidity for some cases will help investors to monitor and, in other cases, to follow the Wall Street rule and sell their stakes. Liquidity on traded stocks might foster intervention when informed blockholders can exploit the ability to speculate in a given period. In period + 1, minorities are willing to sell to
11
ACCEPTED MANUSCRIPT Resubmitted version.R2
blockholders under adverse liquidity shocks and expected blockholder losses. These models highlight the existing tradeoff between voice and exit models. Dasgupta and Piascentino (2015) developed a moral hazard model built on the
RI PT
complementary hypothesis between blockholder intervention and trading suggested by Hirschman (1970). Their model is based on the fact that many blockholders are asset managers—i.e., mutual funds, hedge funds, pension funds—and therefore have different
SC
preferences in terms of long- and short-term portfolios. In their model, they consider active monitoring a feasible strategy to be used by asset managers when they are constrained by
M AN U
investors’ cash flows. Their model predicts that active voice is an equilibrium strategy when the threat of exit is credible and therefore the firm management’s best response is to exert effort to improve firm performance and better corporate governance. The results of this model provide an analytical explanation to the empirical behavior that mutual funds are less active than hedge
TE D
funds.
On the empirical side, Falkenstein (1996) suggested that investors (in particular mutual funds) prefer to invest in stocks with high visibility and low transaction costs. One of the arguments
EP
used to explain this behavior is that traders will choose stocks with high levels of visibility in the press and that have been listed for a long time. There is a well-grounded theoretical approach
AC C
supporting the association between market liquidity and firm performance. A liquid stock market provides an opportunity for non-shareholders to become shareholders, encourages management compensation, reduces managerial opportunism and promotes investment. The most important result of Fang et al. (2009) investigation of stock liquidity and firm performance is that firms with liquid stocks perform better in the market. The mechanism by which liquidity increases performance involves improved information on market prices and managerial compensation,
12
ACCEPTED MANUSCRIPT Resubmitted version.R2
This is particularly powerful for industries with high levels of business uncertainty. Chung et al. (2010) find that firms with better corporate governance typically display better financial performance. However, causality runs from corporate governance to higher liquidity. This result
RI PT
suggests that firms alleviate trading issues (liquidity) by improving their corporate governance record. The mechanism provided for such a dynamic is the improvement of financial and operational transparency, which decreases information asymmetries between current and outside
SC
investors.
Roosenboom et al. (2014), based on takeover data, show a negative relationship between lower
M AN U
stock liquidity and acquirers’ returns, particularly on those firms with a single blockholder. For the cases of multiple blockholders the correlation turns positive because block trades and exit become feasible. Their empirical results suggest that lower liquidity is related with withdrawal from takeover bids that would trigger negative announcements, higher CEO turnover, and lower
TE D
premium. Edmans et al. (2013) also address the issue of liquidity and governance finding a positive effect of stock liquidity on blockholder governance. Liquidity and block formation are linked, since the latter is favored by high liquidity, while the former increases the likelihood of
EP
blockholders governing through voice and exit. Hence, firm performance increases3. Latin American business environment, the issue of liquidity is of particular interest because
AC C
stock turnover is structurally low. For instance, the stock turnover median of all companies that belong to the S&P Latin America index was less than 20 basis points between 2000 and 2012. Representative companies of S&P 500 in the US, for the same time span, have a stock turnover
3
For more details, see Edmans (2014) for a comprehensive critical review on voice and exit models from theoretical and empirical standpoints.
13
ACCEPTED MANUSCRIPT Resubmitted version.R2
of above 150 basis points.4 The above theoretical elements and the regional financial markets structure statistics in the region lead us to state the following hypothesis:
RI PT
H4. Multiple blockholder contestability and voice intervention becomes effective in firms with high liquid stocks. Hence, this hypothesis implies that blockholders’ voice and exit disciplinary mechanisms are complements rather than substitutes; especially in markets such as the Latin American one where
SC
the presence of multiple blockholders is a common structure and market liquidity is structurally low.
3.1. Sources and data characteristics
M AN U
3. Data and method
This study uses an unbalanced panel of more than 550 non-financial firms and 6,785 firm-year data points from six Latin American countries (Argentina, Brazil, Chile, Colombia, Mexico and
TE D
Peru) from 1997 to 2011 (15 years). The main source of the dataset was Thomson’s Datastream platform of financial information; Worldscope platform was used to gather information on shareholding records. Other primary sources comprised databases of financial regulators in each
EP
country sampled, as well as the annual reports of each firm. In particular, the shareholder signatures for Chile, Colombia Peru and Brazil were obtained from the Economatica database
AC C
and the respective local regulatory agencies.5 Ownership information for Mexico and Argentina was also compiled using: i) U.S. Securities and Exchange Commission form 20F for cross-listed companies and, ii) annual reports compiled by the Mexican stock exchange. Table 1 (Panel A) summarizes the construction of the sample and its representativeness. The data extraction began with 4,809 firms. After removing non-active firms and those for which no 4
These estimates are based on data collected from Thomson’s Datastream. Superintendencia de Valores y Seguros (Chile), Superintendencia del Mercado de Valores (Perú), Comissao de Valores Mobiliarios (Brazil), Superintendencia Financiera (Colombia). Comisión Nacional Bancaria y de Valores (México), and Comisión Nacional de Valores (Argentina).
5
14
ACCEPTED MANUSCRIPT Resubmitted version.R2
financial information or equity instruments were available, the number of firms fell to 558. The sample is representative because the selected firms imply more than 40% of total corporate
… Table 1 here…
RI PT
market capitalization reported by the World Bank for each country (Panel B).
Firms are gathered into three broad ownership brackets. Firms whose largest blockholder owns more than 50% of its stocks are named “Firms with absolute control”, while firms whose largest
SC
blockholder holds fewer than 10% of stocks are termed “Widely held firms”. Those in between, whose largest shareholder owns between 50% and 10% of shares are named “Firms with
M AN U
multiple blockholders”. This breakdown corresponds well to the purpose of the study, reflecting the highly concentrated ownership among first shareholders previously reported in Chong and Lopez-de-Silanes (2005). Table 2 summarizes the sample distribution according to the structure of blockholders. On average, firms with multiple blockholders -that is, whose largest shareholder
TE D
holds more than 10% and less than 50% of equity rights- make up about half the sample. In most of the remaining firms, at least 50% of company equity is controlled by a single shareholder. Around 13 companies, or 3% of the sample, are widely held. Colombia is the only country in
EP
which widely held firms account for more than 5% of the sample. This ownership distribution is similar to that reported by Leaven and Levine (2008) in their study of 1,657 European firms,
AC C
confirming that multiple blockholders without absolute control and high ownership concentration are common across large capital markets. … Table 2 here …
Table 3 shows the distribution within the sample of the first and second largest shareholders by firm ownership structure and blockholder identity. Panel A depicts the number of firm-year observations. Two aspects should be mentioned in relation to these figures. First, local firms –
15
ACCEPTED MANUSCRIPT Resubmitted version.R2
whether limited liability or incorporated are the main source of large shareholders, followed by firms owned by families or by financial institutions. The state is the final source of blockholders. Second, this observation might seem to contradict the findings of previous studies on ownership
RI PT
and control that families are the first source of large and/or controlling shareholders (La Porta et al., 1998; Claessens et al., 2002; Faccio and Lang, 2002). This phenomenon may be explained by cases where families use (country specific) alternative legal structures that are not observable in
SC
the information collected, resulting in an underestimation of the importance of family ownership. For instance, families might constitute a trust fund managed by a financial institution. Whenever
M AN U
this is the case it is the trust fund that is recorded as the shareholder.
Panel B depicts the average ownership of the first and second blockholder by identity and firm ownership structure. There are a few features worth mentioning. First, ownership participation of second blockholders is remarkably similar regardless of the nature of the blockholders in
TE D
question and of the firm’s ownership structure. For instance, for the sample of firms with multiple blockholders the largest shareholder has around 29% of equity rights regardless of whether it belongs to a family, a local company, a foreign firm or a financial institution. The
EP
same is the case for the second largest shareholder, whose equity rights range from 14.5% to 18.5%. Second, in firms with multiple blockholders there is scope for the second voting block to
AC C
contest the largest shareholder or to form a coalition in order to obtain absolute control. This prediction is independent of whether the second blockholder is a family, foreign or a financial institution. The ownership sum of both shareholder blocks is close to 50% in all cases. Eventually, all second blockholders will become pivotal in the formation of a coalition. … Table 3 here …
16
ACCEPTED MANUSCRIPT Resubmitted version.R2
The ownership structure of the firms with multiple blockholders is of particular interest. In any attempt to characterize the ability of shareholders to form coalitions in cases where there is no controlling shareholder, ownership distribution is highly relevant. Table 4 shows how, on
RI PT
average, the first two shareholders account for 45% of ownership, and 54% if they engage with the third blockholder. This holds true for all firms and for all types of blockholder (family, local,
shareholder wishes to engage in a coalition attempt. … Table 4 here…
SC
foreign, financial institution or state) regardless of the blockholder with which the largest
M AN U
3.2 Blockholder variables: presence, contestability, and dispersion
This study defines a blockholder as an investor with equity rights (direct votes) representing less than 50% and more than 10% of ownership in a listed company. Thus, blockholder presence is a dummy variable for firms fulfilling this shareholder characteristic. Three measures for
TE D
blockholder contestability are explored: i) the Shapley value of the top three shareholders; ii) the ratio of the second shareholder to the first and iii) the ratio of the second and third shareholders to the first. The Shapley value, as a solution to a cooperative game, measures the probability for
EP
the largest shareholder to form a coalition with either of the next two largest shareholders.6 By construction, the Shapley value is one for all firms where the largest shareholder has more than
AC C
50% of the firm’s equity rights, because it implies absolute shareholder control. The second measure is the contestability index 2nd to 1st shareholder defined as the ratio of the shares of the second blockholder to the shares of the largest one, that is: 1 =
6
ℎ 2 ℎ ℎ (1) ℎ 1 ℎ ℎ
We follow Leech’s (2002) construction of oceanic finite games to compute power indices.
17
ACCEPTED MANUSCRIPT Resubmitted version.R2
The index approaches one when the direct votes of the top two blockholders are similar. The closer to one means the second blockholder can contest the power of the first, therefore the optimal strategy is to engage in a coalition. The farthest from one, means that the first
RI PT
shareholder can advance in decision making without the second largest. This indicator grows in relevance for firms without absolute control and for multiple blockholders where the largest block is able to engage in a coalition with the second shareholder, achieving a controlling
SC
position. The third measure, contestability index 2nd and 3rd to 1st shareholder, proxies the
shareholder. It is defined as: 2 =
M AN U
ability of the second plus the third largest shareholders to challenge and monitor the largest
ℎ 2 + 3 ℎ ℎ (2) ℎ 1 ℎ ℎ
Dispersion estimates are defined using two Herfindahl-Hirschman Index (HHI) type measures. First, the standard HHI is constructed using the first to fourth shareholder ownership:
(3)
TE D
=
!
where is the share of ith shareholder´s shares.
EP
A higher HHI concentration indicates the presence of a large shareholder, and his inherited control; this is an obvious result for the segment of firms with a controlling shareholder, about
AC C
50% of the sample described above. For the 45% of the sample where non-controlling large shareholders have been identified the HHI concentration will always be lower than for firms with controlling shareholders.
The second dispersion estimate highlights the difference between subsequent shareholders in order to measure their dispersion: ""# = 18
( − % )
!
ACCEPTED MANUSCRIPT Resubmitted version.R2
where is the share of ith shareholder´s shares, and % other shareholder different from i. The HHI difference is a measure of shareholder distribution. The higher the index is the larger is the distance (in shares) between shareholders. In conceptual terms the HHI differences reflect the
RI PT
inequality in ownership. 3.3 Firm performance and control variables
In line with existing empirical research on firm value and multiple blockholder ownership
SC
(Claessens et al., 2002; Faccio and Lang, 2002; Mishra, 2011; Attig et al., 2009), several variables were considered to measure firm performance and financial characteristics. We relied
M AN U
in this study on Tobin’s Q as the primary proxy for firm performance, which follows the definition proposed by Black et al. (2006) who define the variable as the ratio between market value and the book value of assets. Market value is the sum of the market value of common stock and preferred stocks (if any), plus the book value of long-term liability and minority interest.7
TE D
Several financial variables are included in the econometric estimation, with the purpose of capturing the behavior and characteristics of firms appropriately, and particularly those related to blockholder structure. The natural logarithm of assets (deflated using the Consumer Price Index
EP
(CPI)) is used as a proxy of firm size. It is expected that a negative coefficient will be observed in the estimation results and that large and long-established firms will demonstrate slower market
AC C
dynamics than small and young companies. This variable can also capture a moral hazard effect easily observable in small firms, which are harder for external investors to monitor and where, in the absence of a watchdog, reckless management can divert resources. Wedge is measured as the ratio between Shapley value and the cash flow rights of the largest shareholder − ! −. Following Guedes and Loureiro (2006) this measure is used as an approach 7
We did alternative estimations using ROA instead of firms Tobin´s Q as performance variable. The econometric results are nor qualitatively different from the reported ones in this section. We used two definitions of ROA: i) EBITDA to book value of assets and ii) Operating Income to book value of assets.
19
ACCEPTED MANUSCRIPT Resubmitted version.R2
to identify shareholders voting rights in cases where data on direct ownership is available. Sales growth is measured as the annual growth rate of operating income. This variable is appropriate for measuring investment opportunities. Firms with better growth opportunities are expected to
RI PT
grow faster. Consequently, a positive relationship is expected between this variable and the value and performance measures.
Leverage is measured as the ratio of the book value of total liabilities to total assets. A high
SC
leverage indicates a scenario of financial stress affecting the firm, and challenges to its managerial decisions. A low value indicates financial leeway. Existing empirical evidence
M AN U
suggests that leverage may have either a positive disciplinary effect on the free use of cash flow by management, or a negative one if it increases the probability of bankruptcy or of a firm’s aggregate financial risk.
Stock beta is the standard measure of systemic risk for a firm’s stock with respect to the
TE D
market; it measures shares that have been actively traded in the stock market for more than 180 days in a given year. The Free Cash Flow (FCF) to equity ratio is defined as EBITDA minus expenditure on tax and interest to total firm equity. This is an indicator of a firm’s short term
EP
liquidity. Tangibility is measured as the ratio between plant, property and equipment, and total assets. Lower asset tangibility signals that a firm’s cash flow is presumed to be generated by
AC C
intangibles such as know-how or branding, which implies high firm market value. The expected relationship with the dependent variable is negative. Finally, country macroeconomic and institutional variables are included in the empirical baseline equation. Latin American countries share the same legal origin (French), and are therefore classified within the empirical literature of corporate governance as having similar investor protection standards. The institutional variables mainly deal with a country’s existing
20
ACCEPTED MANUSCRIPT Resubmitted version.R2
ability to undertake business initiatives, which ultimately have an effect on firms’ market value. Most of the variables fulfill the purpose of representing the financial market power to promote private enterprise. The included macro and country-institutional variables are the Emerging
RI PT
Markets Bond Index (EMBI), Legal rights, business freedom, financial freedom, per-capita GDP growth, inflation rate, and the corporate market capitalization to GDP ratio.8
Table 5 displays the summary of statistics of the above discussed firm performance and control
SC
variables that are included in the econometric baseline estimating equations.
4. Econometric analysis
M AN U
….Insert Table 5 here …
We estimate three baseline regression specifications in order to identify the predicted associations of blockholder interactions on firm market value. The three types of equations are consistent with the previous treatment of blockholder presence, contestability and dispersion. In
TE D
all three, the effect of liquidity and of the identity of the second blockholder is included as independent variables and interactions (level and slope). The equation to estimate the
EP
blockholder presence effect is:
Tobin Qit = β 0 + β1 Pr esenceit + β 2 Liquidityit + β3 Identityl + β 4 Pr esenceit × Liquidityit + β5 Pr esenceit × Identityl + β 6 Liqudityit × Identityl
AC C
+ β 7 Pr esenceit × Liquidity × Identity + β5 X1it + β 6 X 2 j + ε it
(5)
The equation used to estimate the contestability effect is:
8
Definitions of all variables included in the empirical model are summarized in the appendix (Table A.1).
21
ACCEPTED MANUSCRIPT Resubmitted version.R2
Tobin Qit = β 0 + β1Contestabilityit + β 2 Liquidityit + β 3 Identityl + β 4Contestabilityit × Liquidityit + β5Contestabilityit × Identityl + β 6 Liquidityit × Identityl + β 7Contestabilityit × Liquidity × Identity + β5 X1it + β 6 X 2 j + ε it
(6)
Tobin Qit = β 0 + β1Contestabilityit + β 2 Liquidityit + β 3 Identityl
RI PT
The equation used to estimate the dispersion effect is:
+ β 4Contestabilityit × Liquidityit + β5Contestabilityit × Identityl + β 6 Liquidityit × Identityl + β 7Contestabilityit × Liquidity × Identity + β5 X1it + β 6 X 2 j + ε it
(7)
SC
where: &' , &( , &) &* are a vector of coefficients of length − 1 = 4 corresponding to the
− 5 identity values: family, local, foreign, financial institution, and state (family is used as the
M AN U
base group); and the interaction with Liquidity and the variables of interest. The vector - . is the set of firm-specific control variables and - / is the set of country institutional and macro control variables. Subscript i indexes firms while subscript t indexes time.
Stock liquidity is a dummy variable for those firms whose stock turnover belongs in the top
TE D
75th percentile. The set of firms belonging to the top quarter most traded firms in each country are also included in Latin American stock indices such as S&P Latin America 40, which comprises 70% of market capitalization of the region from Brazil, Chile, Colombia, Mexico and
EP
Peru. Therefore, if we use the 75th percentile as the cutoff point, we acknowledge that we are examining highly traded stocks in the region, which are also of international relevance. In
AC C
addition, we used alternative near cutoff points such as the 80th or 70th percentiles and the econometric results remained the same.9 Identity is a dummy variable for the identity of the second largest blockholder. The coefficients of interest are those associated to presence, contestability and dispersion, which shows the association between the different blockholder interests and firm market value. Furthermore, what
9
Thus, the usage of the top quartile is a standard measure for academics and financial analysts alike.
22
ACCEPTED MANUSCRIPT Resubmitted version.R2
matters here is the marginal effect associated with the corresponding interaction whenever they are “active”, in the case of liquidity: &! , & , &( , &* whenever Liquidity and Identity are not zero. The interactions are included in the equation in such a way that they modify the constant and
RI PT
slope in order to fully capture the effect of the firm’s liquidity and of the identity of the shareholder.
Given the longitudinal structure of the data, the econometric estimation falls within the fixed
SC
effects transformation for panel data structure, since the Hausman specification test rejected the null of Random Effects (RE) specification as the true model, and because of the need to control
M AN U
for unobserved fixed variables that are inherent in the estimating equation.
4.1 Econometric results: identification, liquidity and blockholder contestability marginal effects.
This section reports the main findings of this study. The first point worth to highlight is the
TE D
ability of the set of interactions gathering high liquid stocks and blockholder identity in the baseline equations become pivotal in the findings associated to the hypotheses of this study. The econometric results are the result of the identification of a simple equation where Tobin’s Q is
EP
regressed against the variable of interest, adding the firm and country control variables discussed above in batches. This process has proven useful when coefficients produce meaningful results
AC C
following the inclusion of control variables and when the estimation method is improved by using within−fixed effects (FE) rather than Ordinary Least Squares (OLS). This strategy helps overcome the two basic estimator biases suspected to be present in the econometric results: unobservable and omitted variables. Table A.2 in the appendix displays an example of the identification of the empirical baseline regression looking at the marginal effect of the Shapley value on firm’s Tobin’s Q. The reported marginal effects of the Shapley value display the
23
ACCEPTED MANUSCRIPT Resubmitted version.R2
expected sign (negative) and become statistically significant when the full set of controls and interactions are included. In addition, we observe that overall regression goodness of fit increases along with the specifications. Hence, adding the suggested control variables minimizes the
RI PT
omitted variable bias in our estimations.
Table 6a summarizes the marginal effects of blockholder presence, contestability and dispersion on firm Tobin’s Q without any of the interacting terms. These coefficients are related
SC
to the voice model. Table 6b displays the average marginal effects taking into account the interactions. These estimates capture the mechanism of how the blockholder variables respond
M AN U
when exit is observed through high liquidity. The full regression equations are displayed in the Appendix (Table A.3).
Several results are worth highlighting. First, across all estimates we observe that marginal effects for blockholder presence, contestability and dispersion on firm value become
TE D
economically and statistically significant when regressions include the liquidity interaction terms. This result validates the argument that voice and exit are complementary governance mechanisms within firms with multiple blockholders listed in low liquid markets (H4).
EP
Second, blockholder presence implies a premium on firm value of 0.045 units on Tobin´s Q [Eq.1 Table 6a]. This premium is three times larger when market liquidity is considered, rising to
AC C
0.135 units, and to 0.17 when the second blockholder identity interaction is included [Eq1. and Eq2 Table 6b]. Regression coefficients become significant at the 5% level. The positive effect suggests that the mere presence of blockholders increases value to shareholders. In value terms, a $10 million book asset firm and Tobin ratio close to 1 will be valued at $10.45m if there are blockholders with respect to absolute control firms. Now, if stock turnover is high (i.e., the firm belongs to the top quartile), this implies a value of $11.35m.
24
ACCEPTED MANUSCRIPT Resubmitted version.R2
The estimated size effect is greater than reported by Laeven and Levine (2008) for European firms, whose results show that there is a premium on Tobin’s Q of 0.037 units of the control rights of the largest shareholder for firms that only have multiple blockholders. The above results
RI PT
validate H1, which states that blockholder presence is value enhancing.
Third, the Shapley value marginal effects on firm value are all negative as expected. That is, as long as the probability to form a controlling coalition between the largest block with the
SC
subsequent blockholders increases, the firm value decreases. Marginal effect ranges from −0.14 to −0.34 when regressions include liquidity and second blockholder interactions. This finding is
M AN U
particularly important because if the probability of forming a controlling coalition raises 10 points (i.e. from 0.4 to 0.5) Tobin’s Q decreases by 0.034 units. In value terms, if a firm shows a Tobin’s Q ratio of 1 and corporate capitalization is 10 million dollars, rent diversion of a potential coalition will reduces asset value by 340k dollars. This result is in accordance with
TE D
Maury and Pajuste (2005) who report a −0.20 regression coefficient and by Gutierrez and Pombo (2009) who regression coefficients of −0.30 and −0.58 for medium and high liquid stocks for 140 Colombian firms. These results provide powerful validation for H2, which proposed a negative
AC C
coalition.
EP
effect on firm value due to potential tunneling and rent diversion by the blockholders’ controlling
Given that the Contestability Indices are inversely related to the Shapley value, the former display a positive association with firm value. They also capture the potential control exerted either by the second or the next two blockholders on the largest one. The marginal effect ranges from 0.11 to 0.13 for the voice model (Col.3 and Col.4, Table 6a). However, this effect becomes larger when regressions are controlled by stock liquidity and by the identity of the second blockholder. The marginal effects lie between 0.13 and 0.19 (Col.5 to Col.8, Table 6b). For the
25
ACCEPTED MANUSCRIPT Resubmitted version.R2
latter case, a change in one standard deviation in the contestability index (0.32) the Tobin’s Q will rise by 0.061 units. Fourth, regarding dispersion measures, the regression coefficients are displayed in the last two
RI PT
(four) columns in table 6a (6b). The HHI captures blockholders’ cash flow rights dispersion. The regression coefficients are negative related (as expected) with firm Tobin’s Q indicating that as dispersion increases there is more room for tunneling by the controlling shareholders. The
SC
marginal effect ranges from 0.39 and 0.49 and becomes statistically significant for the overall regressions with the interacting terms (Col.9 to Col.12, table 6b) suggesting that a change in one
M AN U
standard deviation on blockholders dispersion (0.26) will decrease Tobin’s Q by an average of 0.11 units. This result reinforces the one derived using the Shapley value. In fact the lower the control power of the second or third blockholder the greater the rent capture by the controlling voting-block.
TE D
The last set of results supports the third and fourth working hypotheses, in the sense that the second blockholder is a key element in improving the internal governance of firms, particularly in emerging markets such as Latin America. The consequence for firms with illiquid, seldom
EP
traded stocks, probably retained with the sole purpose of complying with stock exchange rules, is that the absence of such an internal corporate governance mechanism – i.e., blockholder
AC C
contestability− increases firm agency costs and deteriorates investor protection. …Insert Tables 6a and 6b here…
4.2 Identification results: second blockholder identity marginal effects The results presented in the previous section showed how the second blockholder is important to firm value and becomes a key player exerting direct monitoring on the largest shareholder or possibly resorting to the exit mechanism (voice) in order to discipline firm management for
26
ACCEPTED MANUSCRIPT Resubmitted version.R2
highly liquid stocks. It becomes crucial, therefore, to distinguish the identity of the second blockholder if firm value is to be maximized. This section is devoted to disentangling whether a particular kind of blockholder is more effective than others in enforcing managerial control.
RI PT
As described in the data section, the second blockholder is divided by their identity among family, local firm, foreign investor, financial institution or the state. In the second set of results the ownership effect by type of second blockholder is disaggregated. Figure 1 summarizes the
SC
marginal effect of each variable of interest, disentangled by the identity of the second largest shareholder. We have chosen to present the marginal effects in graphical form instead of using a
M AN U
table mainly for illustrative purposes. The coefficient display in a graph allows us to see the coefficient size, its confidence interval, and it allows us to contrast all of these elements with each of the indicators of blockholders’ presence, contestability, and dispersion measures included in the study. There are a total of 36 marginal effects that are displayed in the appendix
TE D
(Table A.4). For instance, the estimates of the baseline model of the overall marginal effect of the Shapley Value of firm Tobin’s Q is −0.345 (Table 6b) once regressions take into account the full set of interactions of stock liquidity and second blockholder type. Looking at Fig. 1, we can
EP
see that this negative effect is largest when the probability of conforming dominant coalition with the State as second blockholder is −0.75 within a confidence interval which lies between
AC C
−1.5 to −0.05. Hence, in this case, rent diversion is maximized.
….. Insert Figure 1 here…
The principal results may be summarized as follows. First, the positive and significant effect of the type second blockholder on firm value, for high liquid stock firms, is observed when they are financial firms, foreign investors or the state. The extent of the marginal effects of these types of blockholders is greater than that of family and local firms. Second, after separating by
27
ACCEPTED MANUSCRIPT Resubmitted version.R2
identity the findings show a substantial differential effect. The change between the mere fact of being a second large shareholder and it being of a value enhancing kind, such as a financial firm, is not trivial. For the case of blockholder presence, the corresponding marginal effect changes
RI PT
from 0.19 when he/she is family, to 0.41 or to 0.52 whether a financial firm or the state is the second blockholder.
Similar results are observed for the contestability indices. The capacity of the second
SC
blockholder to contest the largest blockholder is only significant for financial firms; the marginal effect of contestability highlights this role, increasing from 0.06 for family to 0.39 for financial
M AN U
firms. Third, controlling coalitions (Shapley value) will harm firm performance considerably. In this case, the marginal effect on Tobin’s Q decreases from −0.24 when the second voting block is a local firm to −0.59 and −0.75 when a financial firm or the state is involved in the coalition. Fourth, the role of the state as second blockholder deserves special attention. Many firms in the
TE D
sample were formerly state owned enterprises −some belonging to the power and energy sector − that had been privatized in the 1990s. In terms of market capitalization and size, privatized companies might be categorized as belonging to the subgroup of high stock turnover. In the
EP
shallow Latin American stock markets these firms have become major new players. The role of the state as second blockholder within other private firms is positive according to our findings.
AC C
However, this set of results should be accepted with care, given the small number of observations. For instance, in the case of Chile the development bank that is a state owned firm is an active shareholder in mining and domiciliary public services companies. The findings presented above highlight the importance of financial and foreign firms as second blockholders, reinforcing the positive effect on firm performance, in contrast to family and local ownership. These results are in line with the findings of the institutional investors and foreign
28
ACCEPTED MANUSCRIPT Resubmitted version.R2
investment literature. In general terms, it is expected that these two investor types will behave independently, providing better governance standards. This outcome supports H3, which predicts that the presence of an independent second blockholder enhances firm value.
RI PT
4.3 Endogeneity check
This section summarizes the endogeneity analysis conducted on the results that have just been discussed. The purpose is to determine whether the results hold to misspecifications tests due to a Empirical literature on
SC
potential endogeneity effect embedded in the estimation results.
corporate finance stresses the potential endogeneity between firm value and ownership structure
M AN U
(Renders and Gaeremynck, 2006; Bascle, 2008; Larcker and Rusticus, 2010; Roberts and Whited, 2013). To address these issues, we relied on instrumental variable estimations for the fixed effects regressions. The main endogeneity issue in the econometric estimation regards simultaneity; that is, large shareholders can pick stocks on high valued firms and, at the same
corporate value.
TE D
time, blockholder intervention and monitoring improves firm governance mechanisms and
Three variables were used as instrument for the blockholder ownership variables: (1) The
EP
blockholder definition −i.e. presence, contestability and dispersion− lagged one period; (2) The free cash flow standard error of the last three periods; and (3) The operating income standard
AC C
error of the last three periods. Each variable was used individually, as well as jointly in the IV estimation, using robust standard errors and serial correlation correction upon the estimates. The choice of lagged variables as an endogenous regressor (0 ) in a linear model = &1 + &! ! +. . +&0 0 + 3 as an instrument (4) is based on the assumption that satisfies the relevance
and the exclusion conditions, that is, #5(4, 0 ) ≠ 0 and #5(4, 8) = 0. Both conditions imply that in a linear regression the instrument z only has a relationship with the dependent variable
29
ACCEPTED MANUSCRIPT Resubmitted version.R2
through the relationship of the endogenous regressor 0 (Roberts and Whited, 2013). The relevance condition on the instrumented equation 0 = 91 + 9! ! +. . . +90%! 0%! + :; 4; + < (8)
RI PT
requires that : ≠ 0; The relevance and exclusion conditions ensure the correct identification of the original linear model. Thus, the lagged variable as an instrument variable is closely related to current value of the potential endogenous regressor and does not necessarily explain the
SC
dependent variable in the second stage estimation. Such requirements are met by the instrumental variables chosen, of which the corresponding statistical tests are reported at the bottom of the IV
M AN U
estimation results.
Following Himmelberg et al. (1999), we use the free cash flow volatility and operating income volatility as instruments. Unobserved firm heterogeneity comes from several sources such as managerial discretion or investor risk aversion. The former deals with intangible assets as a
TE D
source of discretional spending, as it is harder to monitor by shareholders and the blockholder ownership required to monitor rent diversion and managerial perk consumption would need to be greater.10 The second source concerns firms’ operating risk. All other things being equal, the
EP
higher a firm’s cash flow or operating income volatility, the less likely it is that multiple blockholder ownership will be present. These variables have the ability to isolate the potential
AC C
endogeneity between ownership structure and performance thanks to their ability to signal the estate of the company to potential investors. Such volatility informs the investors’ decision to increase his ownership stake or not. At the same time firm market value will have a lower response to such volatility.
10
We do not account for a good proxy that captures firm cash flow related with intangibles.
30
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 7 reports the IV regression results for joint instruments. The regression equations replicate the estimates of blockholder presence, contestability and dispersion with interactions. The focus is on the full estimation including high stock turnover firms and taking into account
RI PT
the identity of the second blockholder. In this case the table displays the full regressions, including the set of firm-specific financial controls, country institutional and macro variables. The principal result of this exploration of endogeneity is that the set of tests used to validate the
SC
IV analysis favor the original estimation and suggest that the IV estimation results should not be taken as definitive. To begin with, there is no statistical evidence of endogeneity. The tests listed
M AN U
at the bottom of the table accept the null hypothesis that the explanatory variables are not endogenous. The weakest case is the one corresponding to presence of blockholders, which rejects the null hypothesis at 10% significance; the remaining estimations reject the case of endogeneity at high levels of statistical significance. From this finding the remaining tests
TE D
comply with the standard procedure when estimating IV regressions. The weak instruments test is also rejected (high values of the statistic). Therefore, the set of instruments become a valid set of variables to uncover an endogenous process within the estimation. The under-identification
EP
test is also rejected, with zero probability on the null hypothesis of under-identification. Finally, the valid instrument test (Hansen J) is accepted, since we have one candidate variable for
AC C
endogeneity and three exogenous instruments (as tested above) … Insert table 7 here….
5. Conclusions
This article examined the relationship between multiple blockholder ownership (largest shareholders between 50% and 10%, as opposed to majority shareholders with more than 50%) and firm value. The results show that via different measures of presence, contestability and
31
ACCEPTED MANUSCRIPT Resubmitted version.R2
dispersion, blockholder presence and contestability are positively associated with market firm value within a representative sample of real sector listed firms across Latin America for the 1997 – 2011 period. The finding results show that voice models are limited if there is no further
RI PT
improvement on stock market liquidity. In this sense, blockholder direct monitoring and contestability among voting blocks, become effective internal corporate governance mechanisms within high traded stocks. That is, voice and trade models are complementary to governance
SC
mechanisms rather than substitutes for them. Further, the study evaluates whether the identity of the second largest blockholder enhances blockholder contestability when Tobin’s Q is already
M AN U
controlled by stock liquidity and finds that firms with financial or foreign second largest blockholders are those where the previous result strongly holds. This is not to say that family, local and state second largest blockholders are not important, but financial and foreign blockholders are those with the strongest effects on firm performance.
TE D
These findings extend previous literature on blockholder contestability considering the full set of interacting terms when firm value regressions control for stock liquidity and the second blockholder identity. The implications of these results from a standpoint of financial
EP
development highlight the role of financial deepening in emerging markets. On the one hand, by increasing market liquidity through greater firm listing and stock float ratios is dependent on
AC C
ownership concentration that might still present a natural restriction for further financial deepening. But on the other hand, the entry of large shareholders such as local or foreign institutional and other financial investors has a significant impact on firm performance by raising corporate governance standards and fostering the development of equity markets. These results coincide with studies on institutional investor heterogeneity and firm value.
Acknowledgements
32
ACCEPTED MANUSCRIPT Resubmitted version.R2
We would like to thank the editor (Carl R Chen) and one anonymous referee for their comments raised during the reviewing process that helped us to improve the paper’s theoretical framework and the empirical design. We want to thank to Luis Gutierrez, for helpful comments
RI PT
to previous versions of this paper and María Camila de la Hoz for her helpful research assistance, the comments by the discussant at the World Finance Conference-2016 at NY, and seminar
SC
participants at the research workshop at Universidad de los Andes School of Management.
References
M AN U
Ali S., Liu B., & Su J.J. (2017). Corporate governance and stock liquidity dimensions: panel evidence from pure order-driven Australian market. International Review of Economics and Finance. Forthcoming. Almazan A., Hartzell, J., & Starks, L. (2005). Active institutional shareholders and cost of monitoring: evidence from executive compensation. Financial Management, 34, 5-34. Attig, N., El Ghoul, S. & Guedhami, O. (2009). Do multiple large shareholders play a corporate governance role? Evidence from East Asia, The Journal of Financial Research, 32, 395-422. Bascle, G. (2008). Controlling for endogeneity with instrumental variables in strategic management research, Strategic Organization: 6(3), 285–327
TE D
Becht, M. (2001). Beneficial ownership in the United States. In F. Barca & M. Becht (Eds.), The control of corporate Europe: Chapter 11, pp. 285–298. Oxford University Press. Bennedsen, M. & D. Wolfenson (2000). The balance of power in closely held corporations. Journal of Financial Economics, 58: 113-139
EP
Bloch F. & Hege U. (2003). Multiple shareholders and control contests. MPRA Paper 42286, University Library of Munich, Germany.
AC C
Claessens S., Djankov S., Fan J. & Lang L. (2002). Disentangling the incentive and entrenchment effects of large shareholders. Journal of Finance, 57: 2741–2771 Chong, A. & Lopez-de-Silanes, F. (Eds.) (2005) Privatization in Latin America, myths and realities, Washington: Stanford University Press - World Bank. Chung K. H., Elder J. & Kim J. C., (2010) Corporate governance and liquidity. Journal of Financial and Quantitative Analysis, 45 (2): 265-291. Cueto, D. (2013). Substitutability and complementarity of corporate governance mechanisms in Latin America. International Review of Economics and Finance, 25: 310-325. Dasgupta A, & Piacentino G. (2015) The Wall Street Walk when blockholders compete for flows. Journal of Finance, 70(6): 2853-2896 De-la-Hoz M.C. & Pombo C. (2016) Institutional investor heterogeneity and firm valuation: Evidence from Latin America. Emerging Markets Review, 26, 197-221
33
ACCEPTED MANUSCRIPT Resubmitted version.R2
Edmans A., Fang V.W., & Zur E. (2013) The effect of liquidity on governance, Review of Financial Studies, 26(6), 1443–82 Edmans A., & Manso, G. (2011) Governance through trading and intervention: A theory of multiple blockholders. Review of Financial Studies, 24(7), 2395–428
RI PT
Edmans, A. (2014) Blockholders and corporate governance, Annual Review of Financial Economics, 6, 23-50 Faccio, M., & Lang, L. (2002)The ultimate ownership of western European corporations. Journal of Financial Economics, 65, 365–395 Falkensein, E.G. (1996) Preferences for stock characteristics as revealed by mutual fund portfolio holdings, Journal of Finance, 51(1): 111–135
M AN U
SC
Fang V.W., Noe T., Tice S. (2009). Stock market liquidity and firm value, Journal of Financial Economics, 94(1), 150–69 Ferreira, M., & Matos, P., (2008) The colors of investors’ money: The role of institutional investors around the world. Journal of Financial Economics, 88, 499–533 Gomes, A., & Novaes, W. (2006) Sharing of control versus monitoring as corporate governance mechanisms, Working paper, May 2006 Guedes J., & Loureiro, G. (2006) Estimating the expropriation of minority shareholders: Results from a new empirical approach, European Review of Finance, 12: 421-448 Gutiérrez, L. H., & Pombo, C. (2009). Corporate ownership and control contestability in emerging markets: The case of Colombia. Journal of Economics and Business, 61, 112-139
TE D
Heritage Foundation (2013). Index of economic freedom, Web site: http://www.heritage.org/index/. Accessed: December, 2014 Hirschman, A.O. (1970). Exit, voice, and loyalty: Responses to decline in firms, organizations, and states. Cambridge, MA: Harvard Univ. Press.
EP
Himmelberg, C., Hubbard, R., & Palia, D. (1999). Understanding the determinants of managerial ownership and the link between ownership and performance, Journal of Financial Economics, 53: 353–384 Holderness, C.G. (2009). The myth of diffuse ownership in the United States. Review of Financial Studies, 22(4), 1377–1408.
AC C
Konijn, S. J, Kräussl, R., & Lucas, A. (2011). Blockholder dispersion and firm value. Journal of Corporate Finance, 17, 1330–1339. Jara-Bertin, M., López, F. & López-de-Foronda O. (2008). The contest to the control in European family firms: How other shareholders affect firm value. Corporate Governance: An International Review, 16: 146-159. La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and Vishny, R.W. (1999). Corporate ownership around the world. Journal of Finance, 59, 471–517. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (2002). Investor protection and corporate valuation, Journal of Finance, 57, 1147-1170. Laeven, L., & Levine, R. (2008). Complex ownership structures and corporate valuations, Review of Financial Studies, 21 (2): 579-604
34
ACCEPTED MANUSCRIPT Resubmitted version.R2
Larcker, D. F. & Rusticus, T. O. (2010). On the use of instrumental variables in accounting research, Journal of Accounting and Economics, 49(3), 186–205. Leech, D. (2002). Designing the Voting System for the Council of the European Union, Public Choice, 113(3-4), 437–64.
RI PT
McGee, R. W. (2009). Corporate governance in developing economies. Springer. Maug, E. (1998). Large shareholders as monitors: Is there a trade-off between liquidity and control?, Journal of Finance, 53(1), 65–98. Maury, B., Pajuste A. (2005). Multiple Large Shareholders and Firm Value, Journal of Banking and Finance, 29, 1813-1834.
SC
Mishra, D.R. (2011). Multiple large shareholders and corporate risk taking: Evidence from East Asia, Corporate Governance: An International Review, 19, 507-528.
M AN U
Moshirian, F., Nguyen T. T. and Zhang, B. (2014). Large shareholders and firm value: An international analysis. SSRN Scholarly Paper ID 2407845, Social Science Research Network Nadarajah, S., Ali, S., Liu B., & Haung, A. 2016. Stock liquidity, corporate governance and leverage: new panel evidence. Pacific-Basin Finance Journal. Forthcoming. Park Y.W., Zekiye S., & Moon H.S. (2008). Large outside blockholders as monitors: evidence from partial acquisitions. International Review of Economics and Finance, 17, 529-545. Renders, A. & Gaeremynck, A. (2006). Corporate governance and performance: controlling for sample selection bias and endogeneity. Working Paper 0606, Katholieke Universitiet Leuven
TE D
Roosenboom, P., Schlingemann, F.P., & Vasconcelos, M. (2014). Does stock liquidity affect the incentives to monitor: Evidence from corporate takeovers. Review of Financial Studies, 27(8): 2392–433. Roberts, M. and Whited, T. (2013). Endogeneity in empirical corporate finance. In G. Constantinides, Harris, M. and Stulz, R. (Eds.), Handbook of the economics of finance, Volume 2A, Chapter 7, pp. 493–572. Elsevier.
EP
Sacristán-Navarro, M., Gómez-Ansón, S. & Cabeza-García, L. (2011). Family ownership and control, the presence of other large shareholders, and firm performance: Further evidence, Family Business Review, 24(1), 71-93. Sacristán-Navarro, M., Gómez-Ansón, S. & Cabeza-García, L. (2015). The Company you keep: The effect of other large shareholders in family firms, Corporate Governance: An International Review, 23 (3), 216-233.
AC C
Villalonga, B. & Amit, R. (2010). Family control of firms and industries. Financial Management 39(3), 863–904. World Bank (2008). World development indicators. Web site, http://data.worldbank.org/indicator. Accessed: December, 2014 Williams R., (2012). Using the margins command to estimate and interpret adjusted predictions and marginal effects. The Stata Journal, 12 (2), 308-331. Zwiebel, J. (1995). Block investment and partial benefits of corporate control, Review of Economic Studies, 62: 161-185
35
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 1 - Sample Construction and Representativeness Panel A. Sample Construction. Number of firms in the Sample Total Number of firms (Averages 2003-2011) Average Reported Datastream Mnemonics Removing non-active firms Romoving firms with non-equity instruments Removing banks and financial firms Removing firms with insufficient ownership information Total Sample
Panel B. Sample Representativeness (Averages 2000-2011)
Sample Mean Sample Median
0.300 0.143
SC
0.170 0.062 1.132 0.248 0.115 0.071
0.223 0.118
0.432 0.404
TE D
Argentina Brazil Chile Colombia México Peru
Corporate value - Corporate value sample /GDP sample /Mcap. (World Bank) 0.144 0.705 0.048 0.091 0.778 0.777 0.187 0.482 0.092 0.326 0.091 0.211
M AN U
Total Assets/GDP
RI PT
4809 1552 1154 937 604 562
Source: Own estimates based on Thomson-one and World Bank Development Indicators
Table2 Blockholder ownership structure by country (Firm-year observations) Latin America Obs Firms/1 3,383 226 3,299 220 193 13
EP
Blockholder Structure
AC C
Control 1st. shareholder Multiple blockholders Widely held Total
6,875
Dist. 0.49 0.48 0.03
Argentina Obs 335 114 0
Brazil Obs 853 1,206 57
449
2,116
458
Chile Colombia Mexico Obs Obs Obs 604 184 653 643 289 489 17 66 17 1,264
539
1,159
Peru Obs 754 558 36 1,348
Notes: 1/ average number of firms with ownership information for the 1997-2011 period. Sources: Own estimates based on Thomson's World-Scope , Economatica only ownership info; countries' security regulators; SEC-form 20-F for cross-listed companies, and companies' annual reports when applicable.
36
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 3 First and second blockholder type by firm ownership structure (Firm-year observations)
2nd
Control 1st. shareholder Multiple blockholders Widely held
478 539 95
664 836 40
1,021 1,140 28
RI PT
Panel A - Total Observations Shareholder Firm Ownership Type Family Local Foreign Financial State Control 1st. shareholder 585 1,188 702 605 142 1st Multiple blockholders 696 1,030 1,086 266 79 Widely held 92 47 34 8 0 296 317 16
119 50 1
2nd
Control 1st. shareholder Multiple blockholders Widely held
M AN U
SC
Panel B - Ownership share Shareholder Firm Ownership Type Family Local Foreign Financial State Control 1st. shareholder 0.694 0.697 0.678 0.715 0.721 1st Multiple blockholders 0.274 0.322 0.285 0.317 0.267 Widely held 0.076 0.082 0.083 0.085 0.111 0.164 0.146 0.162 0.060 0.070
0.105 0.154 0.066
0.217 0.147 0.185 0.185 0.075 0.092
TE D
Notes: The table shows the number of firms and ownership belonging to a specific blockholder identity and whether they are first or second largest shareholder, and if they determine the ownership structure among controlling 1st shareholder, multiple blockholders or widely held. For example, 585 firms have as first shareholder a family firm that makes the firm belong to one of controlling shareholder, correspondingly this group of firms has an average ownership of 69%. Sources: Own estimates based on Thomson's Worldscope , Economatica (only ownership info; countries' security regulators; SEC-form 20-F for cross-listed companies, and companies' annual reports when applicable.
All 0.298 0.156 0.086 0.055
All (Max) All (Min) 0.500 0.101 0.500 0.003 0.272 0.000 0.244 0.000
AC C
Shareholder 1 2 3 4
EP
Table 4 Ownership Structure - Blockholder firm All (SD) 0.118 0.091 0.053 0.041
Family 0.279 0.146 0.079 0.055
Local 0.295 0.162 0.090 0.059
Foreign Financial 0.300 0.316 0.154 0.185 0.086 0.102 0.057 0.058
State 0.341 0.185 0.066 0.042
Notes: The table shows the ownership structure from the first to the fourth largest shareholder for the sample and by ownership identity. Source: Authors’ estimation.
37
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 5 Performance and Control variables – Summary of Statistics min
max
mean
0.08 -1.36
7.03 2.27
1.21 0.12
6,682
0.000
1.000
0.494
6,679 6,681 6,681
0.000 0.000 0.000
1.000 1.000 2.000
0.703 0.381 0.581
6,678 6,678
0.013 0.000
1.000 1.000
0.348 0.234
M AN U
6,453 6,679 5,993 6,453 4,892 6,265 6,448
0.72 0.12
0.500
0.341 0.315 0.508
SC
6,916 7,176
sd
RI PT
N
0.256 0.264
-0.258 0.000 -24.32 0.001 0.000 -30.27 0.000
12.650 5.556 23.95 1.437 7.736 31.96 0.967
5.953 1.394 0.02 0.475 0.100 0.096 0.411
1.854 0.331 2.20 0.212 0.399 0.953 0.246
6,682 6,682 4,086 6,682 6,682
64.9 0.055 0.333 0.574 0.333
5,742.0 0.474 0.778 1.000 1.000
482.6 0.199 0.490 0.764 0.716
684.7 0.110 0.154 0.114 0.156
6,682 6,682 6,682
-1.17 9.57 -11.73
31.76 157.00 8.66
6.85 53.12 2.62
4.30 33.29 3.10
AC C
EP
TE D
VARIABLES Firm Performance Tobin's Q ROA Presence Blockholder presence dummy Contestability Shapley Contest. Index 1 Contest. Index 2 Dispersion Herfindal Concentration Herfindal Index Finance Size Wedge Sales (Growth) Leverage Beta Free cash flow Tangibility Institutional Country Finance EMBI Domestic credit (w.r.t U.S.) Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.) Macroeconomic Inflation (cpi) Market cap. / GDP GDP growth (Per-capita)
Notes: The table shows the summary statistics for the explanatory variables of interest: presence, contestability, dispersion; and the remaining control variables: country specific financial, institutional and macroeconomic variables. Domestic credit, legal rights index, business freedom, financial freedom and market capitalization are expressed with respect to (w.r.t) the same variable for the United States. All variable definitions are listed in Appendix – table A.1. Source: Authors’ estimation
38
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 6a – Firm value Fixed Effects Regression - Marginal Effects – Voice model Dependent Variable Tobin' Q
Col.1
Col.2
Col.3
Col.5
0.0455 (0.0362)
Contestability Shapley
0.129* (00552) 0.112** (0.0374)
Dispersion Herfindal Index Concentration
-0.11 (0.0794)
Herfindal Index Differentials 3,282
M AN U
Contest. index 2
Col.6
SC
-0.148 (0.0718)
Contest. index 1
Observations
Col.4
RI PT
VARIABLES Presence Blockholder presence
3,282
AC C
EP
TE D
Significance: *** p < 0.01, ** p < 0.05, * p < 0.1 Source: Authors’ estimation.
3,282
39
3,282
-0.104 (0.0718) 3,282 3,282
Resubmitted version.R2
ACCEPTED MANUSCRIPT
Table 6.b - Firm value Fixed Effects Regression – Average Marginal Effects Voice and Trade models Dependent Variable Tobin' Q
VARIABLES
x P75=1 x P75=1 x x P75=1 x P75=1 x x P75=1 x P75=1 x x P75=1 x P75=1 Type Type Type x Type Col.1 Col.2 Col.3 Col.4 Col.5 Col.6 Col.7 Col.8
x P75=1
x P75=1 x x P75=1 x P75=1 x Type Type Col.9 Col.10 Col.11 Col.12
Presence
Contest. index 1
0.191** 0.166* (0.0792) (0.0859)
0.147** 0.130** (0.0511) (0.0552)
SC
Contest. index 2 Dispersion Herfindal Index Concentration
866
720
866
M AN U
0.385*** 0.493*** (0.12) (0.14)
Herfindal Index Differentials Observations
RI PT
Blockholder presence 0.135** 0.171** (0.0561) (0.0597) Contestability Shapley 0.264** 0.345** (0.0909) (0.0983)
720
866
720
866
720
866
720
0.391*** 0.437*** (0.113) (0.146) 866 720
Notes: Column P75 = 1 reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) variable in a regression where only these two variables are interacted (two way interaction). Column p75=1 x Type reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) corresponding to a regression where ownership type is included in the interaction specification (three way interaction) as in the full base line regression equations 5 to 7.
AC C
EP
TE D
Average marginal effect estimates the predicted value of the estimated equation for each observed value of the variable of interest and the average effect is the average of all predicted values. For binary variables, it obtains the predicted values for the ones and zeroes and estimates the difference. In the two way interaction the average marginal effect can be calculated by examining the raw coefficients presented in table A.3 from which the single marginal effects of blockholder presence is 0.0245 and the single marginal effect of the interaction between blockholder presence and high liquidity is 0.11 (Table A.3 Eq.2). The combined marginal effect of blockholder presence is the derivative of the dependent variable with respect to presence: ∂Q/∂Presence = &! + & = 0.0245 + 0.11 = 0.135 which is the value in Col.1; Similarly, the single marginal effect of the interaction between the Shapley value and high liquidity is -0.148 (Table A3 Eq.5). Thus overall marginal effect is ∂Q/∂SV = &! + & = −0.116 − 0.148 = −0.264; which is the value reported in Col.3. The remainder marginal effects are obtained in the same way. Columns label as x P75=1 x Type report the marginal effect corresponding to the three way interaction between stock liquidity dummy, type of second blockholder and the variable of interest. In this case the average marginal effect implies a nonlinear effect after taking the average of all observed values. Williams (2012) details the procedure undertaken in the statistical analysis software Stata. Heteroscedasticity corrected standard errors in parentheses; Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation.
40
Resubmitted version.R2
ACCEPTED MANUSCRIPT
TE D
M AN U
SC
RI PT
Figure 1 - Firm type, marginal effect.
AC C
EP
Source: Own estimates. Notes: The figure shows the marginal effect of the variables of interest (blockholder presence, contestability and ownership concentration) whenever the high concentration and the identity blockholder are equal to one. These figures allow distinguishing the effect of the blockholder upon performance when identity is studied.
41
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 7 - Instrumental variables regressions full over-identified equations (m = 3) Dependent Variable Tobin' Q
Eq.1
Eq.2
...
...
...
...
...
...
...
...
...
-0.148
...
...
...
...
...
(0.161)
...
...
...
...
...
...
0.0713
...
...
...
...
...
(0.132)
...
...
...
...
...
...
...
...
Herfindal Index Differentials
... ...
...
...
(0.0955)
...
...
...
SC
...
...
...
-0.184
...
...
...
...
(0.211)
...
...
...
...
...
-0.199
...
...
...
...
(0.200)
-0.199***
-0.199***
(0.0560)
(0.0469)
(0.0470)
(0.0471)
(0.0475)
(0.0473)
0.647
0.0263
-0.0590
-0.0596
-0.0806*
-0.0932*
-0.205***
-0.204***
(0.421)
(0.0906)
(0.0523)
(0.0551)
(0.0476)
(0.0488)
1.42E-03
-3.55E-03
-3.47E-03
-3.32E-03
-3.58E-03
-3.57E-03
(5.36E-03) (2.72E-03)
(2.74E-03)
(2.73E-03)
(2.73E-03)
(2.76E-03)
0.235*
0.238**
0.243**
0.243**
TE D
AC C
Lag tangibility
...
-0.201***
0.240**
(0.175)
(0.120)
(0.120)
(0.120)
(0.120)
(0.120)
-0.0441
-0.0252
-0.0232
-0.0237
-0.0284
-0.0291
(0.0307)
(0.0199)
(0.0204)
(0.0204)
(0.0200)
(0.0204)
EP
Lag FCF ratio
...
0.027
-0.216***
0.375**
Beta
RI PT
...
Dispersion Herfindal Index Concentration
Leverage
Eq.6
...
Contest. index 2
Sales (Growth)
Eq.5
1.640*
Contest. index 1
Wedge
Eq.4
(0.914)
Contestability Shapley
Finance Size
Eq.3
M AN U
VARIABLES Presence Blockholder presence
0.105**
0.021
0.018
0.018
0.020
0.020
(0.0441)
(0.0126)
(0.0119)
(0.0116)
(0.0125)
(0.0127)
-0.0441
-0.110
-0.111
-0.114
-0.110
-0.110
(0.147)
(0.108)
(0.109)
(0.109)
(0.108)
(0.108)
Country Finance and Institutional EMBI -5.84e-05* -6.13e-05** -6.35e-05*** -6.27e-05*** -5.86e-05** -6.02e-05** (2.38e-05)
(2.38e-05)
(2.40e-05)
(2.40e-05)
Domestic credit (w.r.t U.S.)
3.059***
2.042***
1.951***
1.961***
2.069***
2.054***
(0.712)
(0.333)
(0.324)
(0.324)
(0.343)
(0.333)
Legal rights (w.r.t U.S.)
0.431**
0.331*
0.319*
0.325*
0.322*
0.310*
(0.207)
(0.173)
(0.173)
(0.173)
(0.175)
(0.174)
Buss. freedom (w.r.t U.S.)
-1.109**
-0.291*
-0.264*
-0.258*
-0.305**
-0.295**
Fin. freedom (w.r.t U.S.)
(3.40e-05) (2.38e-05)
(0.505)
(0.152)
(0.144)
(0.144)
(0.153)
(0.149)
0.0560
-0.165
-0.165
-0.167
-0.147
-0.143
(0.247)
(0.180)
(0.181)
(0.181)
(0.181)
(0.182)
42
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 7 (Cont.) - Instrumental variables regressions over-identified equations Dependent Variable Tobin' Q
Year dummy (2008) Observations Number of id_firm_n Sector dummy Year dummy Robust SE HAC Interaction R2 centered F-stat
Eq.5
Eq.6
3.66E-03
-4.43E-03
-4.84E-03
-4.70E-03
-4.16E-03
-4.31E-03
(7.03E-03) (3.77E-03)
(3.74E-03)
(3.74E-03)
(3.83E-03)
(3.81E-03)
0.0051*** 0.00451***
0.0044***
0.0044***
0.0046***
0.0045***
(8.78E-04) (6.50E-04)
(6.45E-04)
(6.46E-04)
(6.49E-04)
(6.45E-04)
0.0150*** 0.0159***
0.0159***
0.0158***
0.0158***
0.0159***
(3.83E-03) (2.60E-03)
(2.63E-03)
(2.63E-03)
(2.61E-03)
(2.61E-03)
0.203***
0.192***
0.192***
0.192***
0.194***
0.193***
(0.0383)
(0.0287)
(0.0288)
(0.0288)
(0.0289)
(0.0289)
2,689
2,689
2,689
2,689
2,689
2,689
426
426
426
426
426
426
No
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Slope
Slope
Slope
Slope
Slope
0.132
0.128
0.129
0.132
0.130
9.482
9.375
9.378
9.531
9.481
2.445
0.0364
0.000817
0.214
0.443
1.083
0.118
0.849
0.977
0.644
0.506
0.298
3.339
60.86
65.02
60.22
60.70
39.35
7.117
101.3
108.2
106.6
76.48
60.64
Yes Slope -0.475 6.140
TE D
EP
Underidentification. H_0: Underidentification
Eq.4
Yes
Endogeneity H_0: x_j is not endogenous Vs. x_j is endogenous No Endogeneity p-value Weak inst. H_0: Weak identification
Eq.3
RI PT
GDP growth (pc)
Eq.2
SC
Market cap. / GDP
Eq.1
M AN U
VARIABLES Macroeconomic Inflation (cpi)
0.0285
0
0
0
0
0
2.860
9.221
8.472
8.511
9.462
9.381
Valid inst Hansen J. p-value
0.0908
0.00995
0.0145
0.0142
0.00882
0.00918
AC C
Underidentification. p-value Valid inst Hansen J. H_0: Overidentification
Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 Notes: Estimation results for equations 1 to 6 using instrumental variables to check endogeneity of the dependent variable. HAC stands for heterosckedastic and autocorrelation consistent standard errors. Each regression uses 3 join instruments: i) the lag of blockholder presence, contestability and dispersion definition; ii) operating income volatility; iii) cash flow ratio volatility. Slope interactions refers to presence, contestability and dispersion variables times 75th percentile of stock liquidity dummy. Regressions are controlled by second blockholder type dummies. All variable definitions are listed in Appendix – Table A.1. Source: Authors’ estimation.
43
ACCEPTED MANUSCRIPT Resubmitted version.R2
APPENDIX Table A.1 Definition of Variables Ownership Blockholder Presence
Source
Definition
Variable
Shareholder 2
Equity share of the 2nd largest blockholder
Shareholder 3
Equity share of the 3rd largest blockholder
Shareholder 4
Equity share of the 4th largest blockholder
Wedge
Shapley Value to control rights largest shareholder ratio
ROA
Market value/book value: [stock market capitalization + long term liabilities + book value of prefered stocks (if any) + minority interest]/[book value of assets] The ratio of net profits after tax to total assets
Contestability Shapley Contestability Index 1 Contestability Index 2
Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters
M AN U
Tobin's Q
Anual Reports-Thomson Reuters
Own estimates
SC
Performance
RI PT
Shareholder 1
Dummy variable for firms with blockholder as an investor with Anual Reports-Thomson Reuters equity rights (direct votes) with less than 50% and more than 10% ownership in a listed company Equity share of the largest blockholder Anual Reports-Thomson Reuters
The Shapley Value solution for the largest shareholders in a three-voting game The share of the second to the largest blockholder ratio.
Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters
The share of the second and third blockholders to the largest blockholder ratio. Shapley Value to control rights largest shareholder s1
Anual Reports-Thomson Reuters
H.I Concentration
Herfindahl concentration index
Anual Reports-Thomson Reuters
H.I Differences
The sum of the squares of the differences between the first and the second H.I Con.
Anual Reports-Thomson Reuters
Size
Natural Logarithm of total asssets
Anual Reports-Thomson Reuters
Sales (growth)
Percentage of sales growth compared to previous year.
Anual Reports-Thomson Reuters
Leverage
The ratio of total liabilities to total assets
Anual Reports-Thomson Reuters
Wedge
Standard measure of systemic risk for each firm’s stock with Anual Reports-Thomson Reuters respect to the market. It was measure for stocks which have been active in the stock market for more than 180 days in a given year. Free Cash Flow (FCF) to equity ratio Is defined as EBITDA minus Anual Reports-Thomson Reuters tax expenditure minus interest expenditures to total firm equity. This is an indicator for firm short term liquidity. Total property, plant and equipment divided by total assets Anual Reports-Thomson Reuters
AC C
Stock Beta
EP
Finance
TE D
Dispersion
Free Cash Flow
Tangibility
44
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.1 Definition of Variables (Continued) Source
Definition
Variable Institutional
Domestic Credit (w.r.t.US)
Emerging Markets Bonds Index. Measures financial risk in each Anual Reports-Thomson Reuters country. Expressed in basis points Domestic credit to private sector: refers to financial resources The World Bank — world provided to the private sector, such as through loans, development indicators purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment
RI PT
EMBI
Measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and thus facilitate lending. Business Freedom (w.r.t.US) Is an overall indicator of the efficiency of government regulation of business. The score for each country is a number between 0 and 100, with 100 equaling the freest business environment. The score is based on 10 factors, all weighted equally, using data from the World Bank’s Doing Business study. Financial Freedom (w.r.t.US) Is a measure of banking efficiency as well as a measure of independence from government control and interference in the financial sector. Index from 0 to 10. Macroeconomic
The World Bank — world development indicators
Inflation (CPI)
The World Bank — world development indicators
AC C
EP
GDP Growth (pc)
Inflation as measured by the consumer price index reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services Market capitalization (also known as market value) is the share price times the number of shares outstanding, divided by GDP Annual percentage growth rate of GDP per capita based on constant local currency.
TE D
Market Cap/ GDP
M AN U
SC
Legal Rights (w.r.t.US)
45
Index of economic freedom
Index of economic freedom
The World Bank — World development indicators The World Bank — world development indicators
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.2 Identification - empirical equation Tobin's Q and Shapley Value Fixed Effects Regressions VARIABLES
Coeff
+ Firm + Country + Turnover Marginal + Turnover Marginal specific and Macro interaction Effect and Type x Effect controls controls x P75 = 1 p75 x Type
Contestability Shapley
AC C
EP
TE D
M AN U
SC
RI PT
0.217*** 0.199* -0.148 -0.116 -0.264** -0.0114 -0.345** (0.0804) (0.120) (0.110) (0.112) (0.123) (0.152) (0.135) Firm financial and specific variabless Size 0.0780** -0.217*** -0.218*** -0.206*** (0.0391) (0.0577) (0.0576) (0.0585) Wedge -0.133 0.00756 0.0181 0.0321 (0.0940) (0.0796) (0.0798) (0.0811) Sales (Growth) 0.00535 -0.00150 -0.00150 -0.00369 (0.00417) (0.00340) (0.00339) (0.00287) Leverage 0.245* 0.277** 0.277** 0.211 (0.130) (0.135) (0.135) (0.152) Beta 0.0120 -0.0170 -0.0140 -0.0282 (0.0209) (0.0214) (0.0209) (0.0214) Lag free cash flow ratio 0.0160** 0.00158 0.00173 0.0153 (0.00754) (0.0170) (0.0168) (0.0123) Lag tangibility -0.559*** -0.235** -0.238** -0.122 (0.119) (0.101) (0.101) (0.119) Country institutional variables EMBI -5.68e-05** -5.38e-05** -6.49e-05** (2.23E-05) (2.23E-05) (2.74E-05) Domestic credit (w.r.t U.S.) 2.204*** 2.200*** 2.051*** (0.430) (0.430) (0.422) Legal rights (w.r.t U.S.) 0.407** 0.404** 0.330 (0.187) (0.188) (0.210) Buss. freedom (w.r.t U.S.) -0.293* -0.300* -0.275 (0.176) (0.175) (0.181) Fin. freedom (w.r.t U.S.) -0.165 -0.183 -0.195 (0.213) (0.212) (0.237) Macro economic variables Inflation (cpi) -5.24E-03 -5.10E-03 -5.85E-03 (3.65E-03) (3.66E-03) (3.62E-03) Market cap. / GDP 0.00468*** 0.00472*** 0.00456*** (6.99E-04) (6.99E-04) (7.41E-04) GDP growth (pc) 0.0156*** 0.0156*** 0.0160*** (2.60E-03) (2.59E-03) (2.47E-03) Year dummy (2008) 0.203*** 0.202*** 0.194*** (0.0313) (0.0311) (0.0306) Constant 1.068*** 0.939*** 2.062*** 2.060*** 1.873*** (0.0565) (0.242) (0.346) (0.346) (0.368) Observations 6,305 4,426 3,282 3,280 866 2,699 720 R-squared 0.005 0.031 0.118 0.120 0.137 Sector dummy No No No No No No No Year dummy No No No No No No No Robust SE Yes Yes Yes Yes Yes Yes Yes Level + Level + Level + Level + Level + Level + Level + Interaction slope slope slope slope slope slope slope Number of id_firm_n 548 475 474 474 435
Notes: The table reports the identification process of the Tobin’s Q empirical equation for the case of the Shapley Value. It begins with a simple regression and then regressions add the firm-specific, country controls, and the interacting terms of stock liquidity and second blockholder identity. Columns Marginal effects show the marginal effect of the immediate left column. Robust standard errors in parentheses. Significance: *** p < 0.01, ** p < 0.05, * p<0.1
46
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within) Shapley
Coeff.
Variables
x P75=1 x Type
x P75=1
Eq.1
Eq.2
Eq.3
Contest Index 1
Coeff.
x P75=1 x Type
x P75=1
Eq.4
Eq.5
Eq.6
Coeff.
Contest Index 2 x P75=1 x Type
x P75=1
Eq.7
Eq.8
Eq.9
x P75=1
Eq.10
Eq.11
Coeff.
HI diferrences x P75=1 x Type
x P75=1
Coeff.
x P75=1 x Type
x P75=1
Eq.12
Eq.13
Eq.14
Eq.15
Eq.16
Eq.17
Eq.18
0.0455
0.0245
-0.0144
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.0479)
(0.0482)
(0.0914)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
Contestability
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
Shapley
...
...
...
-0.148
-0.116
-0.0114
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.110)
(0.112)
(0.152)
...
...
...
...
...
...
...
...
...
...
...
...
Contest. index 1
...
...
...
...
...
...
0.129*
0.100
-0.121
...
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.0683)
(0.0732)
(0.125)
...
Contest. index 2
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
SC
Presence
Dispersion
...
...
...
...
...
...
...
H.I. Concentration
...
...
...
...
...
...
...
...
...
...
...
...
...
...
H.I. Differences
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.0992**
-0.0919
...
0.0548
0.266**
...
...
(0.0419)
(0.0969)
...
(0.0603)
(0.132)
...
Interaction stock liquidity dummy
0.110**
0.203*
...
...
...
(0.0558)
(0.123)
...
...
P75 x Shapley
...
...
...
...
...
...
...
p75 x contest_index 1
...
...
...
...
...
...
...
...
P75 x contest_index 2
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
P75 x hi_differences P75 x hi_concentration
...
...
...
...
...
...
...
0.0977*
-0.0464
...
...
...
...
...
...
(0.0502)
(0.0545)
(0.0815)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.110
-0.0546
-0.0827
...
...
...
...
...
...
...
...
(0.120)
(0.124)
(0.238)
...
...
...
...
...
...
...
...
...
...
...
-0.104
-0.0462
-0.0324
...
...
...
...
...
...
...
...
(0.105)
(0.106)
(0.234)
-0.0762
-0.0364
...
-0.0725
-0.0527
...
0.0569
0.173*
...
0.0281
0.124
(0.0488)
(0.115)
...
(0.0463)
(0.106)
...
(0.0474)
(0.102)
...
(0.0376)
(0.0779) ...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.148**
-0.352*
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.0751)
(0.180)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
0.0909
0.184
...
...
...
...
...
...
...
...
... ...
...
...
(0.0764)
(0.174)
...
...
...
...
...
...
...
...
EP
...
...
...
...
...
...
0.0495
0.153
...
...
...
...
...
...
...
...
...
...
...
...
(0.0487)
(0.101)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.344***
-0.506
...
...
...
...
...
...
...
...
...
...
...
...
(0.0992)
(0.311)
...
...
...
...
...
...
...
...
...
...
-0.330***
-0.536*
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.108)
(0.324)
...
...
...
AC C
P75 x blockholder presence
...
0.112**
...
TE D
P75 dummy
...
M AN U
Blockholder presence
Coeff.
HI Concentration x P75=1 x Type
RI PT
Presence
47
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.3 Tobin’s Q regressions (Cont.) Presence Coeff.
Variables
Shapley x P75=1
x P75=1 x Type
Coeff.
Contest Index 1 x P75=1
x P75=1 x Type
Coeff.
Contest Index 2
x P75=1
x P75=1 x Type
Coeff.
Leverage Beta L.fcf_r L.tangibility
Coeff.
x P75=1
x P75=1 x Type
Eq.5
Eq.6
Eq.7
Eq.8
Eq.9
Eq.10
Eq.11
Eq.12
Eq.13
Eq.14
Eq.15
Eq.16
Eq.17
Eq.18
-0.218***
-0.221***
-0.214***
-0.217***
-0.218***
-0.206***
-0.217***
-0.217***
-0.201***
-0.217***
-0.217***
-0.200***
-0.217***
-0.221***
-0.212***
-0.217***
-0.222***
-0.211***
(0.0582)
(0.0581)
(0.0598)
(0.0577)
(0.0576)
(0.0585)
(0.0578)
(0.0579)
(0.0572)
(0.0573)
(0.0573)
(0.0573)
(0.0580)
(0.0580)
(0.0583)
(0.0581)
(0.0579)
(0.0582)
-0.0480
-0.0426
-0.0522
0.00756
0.0181
0.0321
-0.0413
-0.0365
-0.0654
-0.0262
-0.0215
-0.0481
-0.0759
-0.0685
-0.0931
-0.0803
-0.0726
-0.0949*
RI PT
Eq.4
(0.0617)
(0.0617)
(0.0607)
(0.0796)
(0.0798)
(0.0811)
(0.0599)
(0.0600)
(0.0561)
(0.0599)
(0.0599)
(0.0564)
(0.0585)
(0.0580)
(0.0565)
(0.0579)
(0.0574)
(0.0562)
-0.00143
-0.00141
-0.00362
-0.00150
-0.00150
-0.00369
-0.00149
-0.00141
-0.00366
-0.00138
-0.00133
-0.00346
-0.00153
-0.00161
-0.00407
-0.00156
-0.00154
-0.00381
(0.00339)
(0.00338)
(0.00287)
(0.00340)
(0.00339)
(0.00287)
(0.00339)
(0.00339)
(0.00284)
(0.00284)
0.282**
0.285**
0.225
0.277**
0.277**
0.211
0.270**
0.271**
0.196
(0.134)
(0.134)
(0.152)
(0.135)
(0.135)
(0.152)
(0.134)
(0.135)
(0.152)
-0.0164
-0.0139
-0.0291
-0.0170
-0.0140
-0.0282
-0.0139
(0.0214)
(0.0209)
(0.0216)
(0.0214)
(0.0209)
(0.0214)
(0.0216)
0.00179
0.00233
0.0167
0.00158
0.00173
0.0153
0.00164
(0.0172)
(0.0170)
(0.0133)
(0.0170)
(0.0168)
(0.0123)
(0.0169)
-0.235**
-0.237**
-0.124
-0.235**
-0.238**
-0.122
-0.240**
(0.102)
(0.102)
(0.119)
(0.101)
(0.101)
(0.119)
(0.101)
Country Institutional and Financial EMBI
x P75=1
Eq.3
-0.0106
-0.0247
(0.0214)
(0.0219)
SC
Sales (Growth)
Coeff.
Eq.2
(0.00339)
(0.00338)
(0.00284)
(0.00340)
(0.00340)
(0.00285)
(0.00340)
(0.00340)
0.265**
0.267**
0.192
0.277**
0.279**
0.233
0.275**
0.279**
0.225
(0.134)
(0.135)
(0.152)
(0.135)
(0.135)
(0.150)
(0.135)
(0.135)
(0.150)
-0.0142
-0.0108
-0.0255
-0.0168
-0.0169
-0.0275
-0.0166
-0.0165
-0.0280
(0.0215)
(0.0213)
(0.0220)
(0.0214)
(0.0210)
(0.0211)
(0.0213)
(0.0211)
(0.0222)
M AN U
Wedge
x P75=1
HI diferrences x P75=1 x Type
Eq.1
Firm Characteristics Size
HI Concentration x P75=1 x Type
0.00157
0.0165
0.00132
0.00132
0.0157
0.00161
0.00241
0.0169
0.00174
0.00255
0.0170
(0.0168)
(0.0126)
(0.0168)
(0.0167)
(0.0122)
(0.0171)
(0.0165)
(0.0124)
(0.0170)
(0.0164)
(0.0131)
-0.243**
-0.126
-0.240**
-0.243**
-0.125
-0.236**
-0.242**
-0.123
-0.237**
-0.242**
-0.125
(0.101)
(0.118)
(0.101)
(0.101)
(0.118)
(0.102)
(0.101)
(0.119)
(0.101)
(0.100)
(0.118)
-5.70e-05** -5.36e-05** -6.40e-05** -5.68e-05** -5.38e-05** -6.49e-05**-5.77e-05*** -5.55e-05** -6.80e-05**-5.79e-05*** -5.57e-05** -6.73e-05** -5.68e-05** -5.29e-05** -5.92e-05** -5.71e-05** -5.30e-05** -6.16e-05** (2.25e-05) (2.26e-05) (2.77e-05) (2.23e-05) (2.23e-05) (2.74e-05) (2.23e-05) (2.22e-05) (2.77e-05) (2.22e-05) (2.21e-05) (2.77e-05) (2.24e-05) (2.26e-05) (2.85e-05) (2.24e-05) (2.25e-05) (2.82e-05)
Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.)
2.161***
2.037***
2.204***
2.200***
2.051***
2.145***
2.137***
1.910***
2.157***
2.150***
1.942***
2.191***
2.172***
2.059***
2.181***
2.160***
2.031***
(0.427)
(0.426)
(0.419)
(0.430)
(0.430)
(0.422)
(0.423)
(0.423)
(0.406)
(0.422)
(0.423)
(0.409)
(0.434)
(0.433)
(0.422)
(0.431)
(0.430)
(0.417)
0.399**
0.396**
0.308
0.407**
0.404**
0.330
0.404**
0.398**
0.307
0.413**
0.409**
0.316
0.405**
0.387**
0.310
0.406**
0.379**
0.292
(0.188)
(0.188)
(0.208)
(0.187)
(0.188)
(0.210)
(0.188)
(0.188)
(0.209)
(0.187)
(0.187)
(0.208)
(0.188)
(0.189)
(0.213)
(0.187)
(0.189)
(0.213)
-0.272
-0.282*
-0.261
-0.293*
-0.300*
-0.275
-0.266
-0.277
-0.246
-0.272
-0.285*
-0.252
-0.279
-0.287
-0.277
-0.273
-0.283
-0.249
(0.172)
(0.171)
(0.178)
(0.176)
(0.175)
(0.181)
(0.171)
(0.170)
(0.176)
(0.171)
(0.170)
(0.177)
(0.176)
(0.175)
(0.181)
(0.174)
(0.173)
(0.180)
-0.166
-0.173
-0.170
-0.165
-0.183
-0.195
-0.166
-0.177
-0.178
-0.164
-0.176
-0.185
-0.167
-0.176
-0.185
-0.166
-0.167
-0.167
(0.214)
(0.213)
(0.240)
(0.213)
(0.212)
(0.237)
(0.213)
(0.212)
(0.236)
(0.212)
(0.212)
(0.236)
(0.214)
(0.213)
(0.235)
(0.214)
(0.213)
(0.236)
-0.00552
-0.00540
-0.00631*
-0.00524
-0.00510
-0.00585
-0.00573
-0.00557
-0.00661*
-0.00559
-0.00547
-0.00638*
-0.00539
-0.00509
-0.00560
-0.00546
-0.00503
-0.00569
Macroeconomic Inflation (cpi)
TE D
Legal rights (w.r.t U.S.)
2.163***
EP
Domestic credit (w.r.t U.S.)
Market cap. / GDP
AC C
(3.63E-03) (3.63E-03) (3.58E-03) (3.65E-03) (3.66E-03) (3.62E-03) (3.62E-03) (3.65E-03) (3.57E-03) (3.63E-03) (3.64E-03) (3.58E-03) (3.66E-03) (3.66E-03) (3.56E-03) (3.65E-03) (3.65E-03) (3.56E-03) 0.00463*** 0.00468*** 0.00455*** 0.00468*** 0.00472*** 0.00456*** 0.00463*** 0.00465*** 0.00443*** 0.00464*** 0.00466*** 0.00445*** 0.00468*** 0.00473*** 0.00465*** 0.00466*** 0.00470*** 0.00462*** (6.97E-04) (6.98E-04) (7.37E-04) (6.99E-04) (6.99E-04) (7.41E-04) (6.94E-04) (6.95E-04) (7.23E-04) (6.93E-04) (6.94E-04) (7.32E-04) (7.02E-04) (7.03E-04) (7.37E-04) (6.98E-04) (6.99E-04) (7.34E-04) GDP growth (pc)
0.0157*** 0.0156*** 0.0160*** 0.0156*** 0.0156*** 0.0160*** 0.0157*** 0.0156*** 0.0167*** 0.0156*** 0.0156*** 0.0165*** 0.0156*** 0.0157*** 0.0160*** 0.0156*** 0.0157*** 0.0161*** (2.61E-03) (2.59E-03) (2.49E-03) (2.60E-03) (2.59E-03) (2.47E-03) (2.60E-03) (2.59E-03) (2.48E-03) (2.60E-03) (2.59E-03) (2.47E-03) (2.60E-03) (2.59E-03) (2.48E-03) (2.61E-03) (2.59E-03) (2.46E-03)
Year dummy (2008) Constant
0.202***
0.202***
0.193***
0.203***
0.202***
0.194***
0.201***
0.201***
0.193***
0.202***
0.202***
0.194***
0.203***
0.204***
0.197***
0.202***
0.203***
(0.0313)
(0.0311)
(0.0306)
(0.0313)
(0.0311)
(0.0306)
(0.0314)
(0.0313)
(0.0305)
(0.0313)
(0.0312)
(0.0305)
(0.0313)
(0.0312)
(0.0309)
(0.0313)
(0.0312)
(0.0308)
2.018***
2.053***
2.026***
2.062***
2.060***
1.873***
1.984***
2.015***
2.047***
1.940***
1.972***
1.993***
2.108***
2.133***
2.093***
2.103***
2.130***
2.045***
48
0.196***
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.3 Tobin’s Q regressions (Cont.)
Coeff.
Variables
Shapley x P75=1
Eq.1
Eq.2
x P75=1 x Type
Eq.3
Coeff.
Contest Index 1 x P75=1
Eq.4
Eq.5
x P75=1 x Type
Eq.6
Coeff.
Contest Index 2
x P75=1
Eq.7
x P75=1 x Type
Eq.8
Eq.9
Regression Statistics
Coeff.
HI Concentration
x P75=1
Eq.10
Eq.11
x P75=1 x Type
Eq.12
Coeff.
RI PT
Presence
HI diferrences
x P75=1
Eq.13
Eq.14
x P75=1 x Type
Eq.15
Coeff.
x P75=1
Eq.16
Eq.17
x P75=1 x Type
Eq.18
Observations
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
R-squared
0.117
0.119
0.135
0.118
0.120
0.137
0.118
0.119
0.136
0.119
0.120
0.137
0.117
0.121
0.139
0.117
0.121
0.135
474
435
474
474
435
474
474
435
474
474
435
474
474
435
474
474
435
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Year dummy
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
0.117
0.119
0.135
0.118
0.120
0.137
0.118
0.119
0.136
-27.30
6.662
72.51
1.112
3.155
68.15
15.26
Robust SE Interaction R2 within Hausman test H0: RE Vs. H1: FE (stat)
SC
474
Sector dummy
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
0.119
0.120
0.137
0.117
0.121
0.139
0.117
0.121
0.135
28.14
104.8
-28.89
-0.879
138.0
-10.50
20.59
136.4
M AN U
Number of id_firm_n
16.30
102.0
29.69
Notes: This table the Tobin´s Q regression looking at the variables of blockholder control: presence, contestability and dispersion. The regression coefficient of each variable of interest are reported in Eq.1, Eq.4, Eq.7, Eq.10, Eq.13 and Eq.16; Columns label by xP75=1 report the interaction with the stock liquidity dummy at 75Th percentile. Columns label as x P75=1 x report the marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) and second blockholder ownership type (three way interaction). All variable definitions are listed in Appendix – table A.1.
AC C
EP
TE D
Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation
49
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.4 Firm values and second blockholder type marginal effects P75=1 x Type Col.1
Family
Local
Foreign Financial State
Col.2
Col.3
0.189
0.0815
Col.4
Col.5
P75=1 x Family Local Foreign Financial Type Col.6 Col.7 Col.8 Col.9 Col.10 Col.11
Presence Blockholder presence
0.171** (0.0721)
0.139 0.410** 0.527**
...
...
...
...
(0.124) (0.0927) (0.0928) (0.164) (0.215)
...
...
...
...
Contestability
Contest. index 2
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
... -0.345** -0.363 -0.247 -0.310** -0.599** -0.751**
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
0.130**
0.106
0.0658
0.132* 0.284**
0.075
...
...
...
(0.103) (0.0785) (0.112) (0.206)
...
...
...
(0.135) (0.222) (0.156)
...
...
...
...
...
...
0.166* 0.0631
0.114
0.180 0.390* -0.120
(0.0899) (0.165) (0.133)
(0.124) (0.202) (0.309)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
... -0.493*** -0.619* -0.276 -0.476**
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
720
116
168
340
86
10
720
-0.593 -2.395***
(0.214)
(0.378)
(0.791)
...
...
...
...
...
...
...
...
...
116
168
340
86
10
EP
Notes: Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation
AC C
... ...
(0.330)
...
…… End of Document ……
50
... ...
(0.263)
...
(0.171) (0.324) (0.227)
... ...
(0.149)
...
TE D
Observations
...
...
(0.0596) (0.0974)
H.I. Differentials
...
...
Dispersion H.I. Concentration
...
SC
Contest. index 1
P75=1 x Family Local Foreign Financial State Type Col.12 Col.13 Col.14 Col.15 Col.16 Col.17 Col.18
...
M AN U
Shapley
State
RI PT
VARIABLES
...
...
...
...
...
...
...
...
...
...
...
...
... -0.437*** -0.538* -0.143 -0.442** -0.727* -1.553 (0.159) (0.308) (0.210) 720
116
168
(0.205) (0.396) (1.241) 340
86
10
ACCEPTED MANUSCRIPT 1
Table 1 - Sample Construction and Representativeness Panel A. Sample Construction. Number of firms in the Sample Total Number of firms (Averages 2003-2011) Average Reported Data-stream Mnemonics Removing non-active firms Removing firms with non-equity instruments Removing banks and financial firms Removing firms with insufficient ownership information Total Sample
Sample Mean Sample Median
Corporate value sample /Mcap. (World Bank) 0.705 0.091 0.777 0.482 0.326 0.211
0.170 0.062 1.132 0.248 0.115 0.071
0.144 0.048 0.778 0.187 0.092 0.091
0.300
0.223
0.432
0.143
0.118
0.404
M AN U
Corporate value - sample /GDP
TE D
Argentina Brazil Chile Colombia México Peru
Total Assets/GDP
SC
Panel B. Sample Representativeness (Averages 2000-2011)
RI PT
4809 1552 1154 937 604 562
EP
Source: Own estimates based on Thomson-one and World Bank Development Indicators
AC C
Table2 Blockholder ownership structure by country (Firm-year observations) Blockholder Structure
Latin America Obs Firms/1 Dist.
Control 1st. shareholder Multiple blockholders Widely held
3,383 3,299 193
226 220 13
Total
6,875
458
0.49 0.48 0.03
Argentina Obs
Brazil Obs
Chile Obs
Colombia Obs
Mexico Obs
Peru Obs
335 114 0
853 1,206 57
604 643 17
184 289 66
653 489 17
754 558 36
449
2,116
1,264
539
1,159
1,348
Notes: 1/ average number of firms with ownership information for the 1997-2011 period. Sources: Own estimates based on Thomson's World-Scope , Economatica only ownership info; countries' security regulators; SECform 20-F for cross-listed companies, and companies' annual reports when applicable
ACCEPTED MANUSCRIPT 2
Table 3 First and second blockholder type by firm ownership structure (Firm-year observations)
Panel B - Ownership share Shareholder Firm Ownership Type Control 1st. shareholder 1st Multiple blockholders Widely held Control 1st. shareholder Multiple blockholders Widely held
664 836 40
1,021 1,140 28
296 317 16
119 50 1
Family Local Foreign Financial State 0.694 0.697 0.678 0.715 0.721 0.274 0.322 0.285 0.317 0.267 0.076 0.082 0.083 0.085 … 0.111 0.164 0.146 0.162 0.060 0.070
0.105 0.154 0.066
0.217 0.147 0.185 0.185 0.075 0.092
TE D
2nd
478 539 95
RI PT
Control 1st. shareholder Multiple blockholders Widely held
State 142 79 0
SC
2nd
Family Local Foreign Financial 585 1,188 702 605 696 1,030 1,086 266 92 47 34 8
M AN U
Panel A - Total Observations Shareholder Firm Ownership Type Control 1st. shareholder 1st Multiple blockholders Widely held
EP
Notes: The table shows the number of firms and ownership belonging to a specific blockholder identity and whether they are first or second largest shareholder, and if they determine the ownership structure among controlling 1st shareholder, multiple blockholders or widely held. For example, 585 firms have as first shareholder a family firm that makes the firm belong to one of controlling shareholder, correspondingly this group of firms has an average ownership of 69%.
AC C
Sources: Own estimates based on Thomson's Worldscope , Economatica (only ownership info; countries' security regulators; SECform 20-F for cross-listed companies, and companies' annual reports when applicable.
ACCEPTED MANUSCRIPT 3
Table 4 Ownership Structure - Blockholder firm All 0.298 0.156 0.086 0.055
All (Max) 0.500 0.500 0.272 0.244
All (Min) 0.101 0.003 0.000 0.000
All (SD) 0.118 0.091 0.053 0.041
Family 0.279 0.146 0.079 0.055
Local 0.295 0.162 0.090 0.059
Foreign 0.300 0.154 0.086 0.057
Financial 0.316 0.185 0.102 0.058
State 0.341 0.185 0.066 0.042
RI PT
Shareholder 1 2 3 4
AC C
EP
TE D
M AN U
SC
Notes: The table shows the ownership structure from the first to the fourth largest shareholder for the sample and by ownership identity. Source: Authors’ estimation.
ACCEPTED MANUSCRIPT 4
Table 5 Performance and Control variables – Summary of Statistics N
min
max
mean
sd
0.08 -1.36
7.03 2.27
1.21 0.12
0.72 0.12
6,682
0.000
1.000
0.494
0.500
6,679 6,681 6,681
0.000 0.000 0.000
1.000 1.000 2.000
0.703 0.381 0.581
0.341 0.315 0.508
6,678 6,678
0.013 0.000
1.000 1.000
0.348 0.234
6,453 6,679 5,993 6,453 4,892 6,265 6,448
-0.258 0.000 -24.32 0.001 0.000 -30.27 0.000
12.650 5.556 23.95 1.437 7.736 31.96 0.967
5.953 1.394 0.02 0.475 0.100 0.096 0.411
1.854 0.331 2.20 0.212 0.399 0.953 0.246
6,682 6,682 4,086 6,682 6,682
64.9 0.055 0.333 0.574 0.333
5,742.0 0.474 0.778 1.000 1.000
482.6 0.199 0.490 0.764 0.716
684.7 0.110 0.154 0.114 0.156
-1.17 9.57 -11.73
31.76 157.00 8.66
6.85 53.12 2.62
4.30 33.29 3.10
SC 0.256 0.264
M AN U
TE D
EP 6,682 6,682 6,682
RI PT
6,916 7,176
AC C
VARIABLES Firm Performance Tobin's Q ROA Presence Blockholder presence dummy Contestability Shapley Contest. Index 1 Contest. Index 2 Dispersion Herfindal Concentration Herfindal Index Finance Size Wedge Sales (Growth) Leverage Beta Free cash flow Tangibility Institutional Country Finance EMBI Domestic credit (w.r.t U.S.) Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.) Macroeconomic Inflation (cpi) Market cap. / GDP GDP growth (Per-capita)
Notes: The table shows the summary statistics for the explanatory variables of interest: presence, contestability, dispersion; and the remaining control variables: country specific financial, institutional and macroeconomic variables. Domestic credit, legal rights index, business freedom, financial freedom and market capitalization are expressed with respect to (w.r.t) the same variable for the United States. All variable definitions are listed in Appendix – table A.1. Source: Authors’ estimation
ACCEPTED MANUSCRIPT 5
Table 6a – Firm value Fixed Effects Regression - Marginal Effects – Voice model Dependent Variable Tobin' Q
Col.1
Col.2
Col.3
Col.4
0.0455 (0.0362)
Contestability Shapley
0.129* (00552) 0.112** (0.0374)
Dispersion Herfindal Index Concentration
-0.11 (0.0794)
Herfindal Index Differentials 3,282
M AN U
Contest. index 2
Col.6
SC
-0.148 (0.0718)
Contest. index 1
Observations
Col.5
RI PT
VARIABLES Presence Blockholder presence
3,282
3,282
AC C
EP
TE D
Significance: *** p < 0.01, ** p < 0.05, * p < 0.1 Source: Authors’ estimation.
3,282
-0.104 (0.0718) 3,282 3,282
ACCEPTED MANUSCRIPT
6
Table 6.b - Firm value Fixed Effects Regression – Average Marginal Effects Voice and Trade models Dependent Variable Tobin' Q x P75=1
x P75=1 x Type
x P75=1
x P75=1 x Type
x P75=1
x P75=1 x Type
x P75=1
x P75=1 x Type
Col.1
Col.2
Col.3
Col.4
Col.5
Col.6
Col.7
Col.8
-0.264**
-0.345**
(0.0909)
(0.0983)
0.135**
0.171**
(0.0561)
(0.0597)
Contest. index 1
0.191**
0.166*
(0.0792)
(0.0859)
Contest. index 2
0.147** (0.0511)
720
866
720
AC C
866
EP
Herfindal Index Differentials
x P75=1
x P75=1 x Type
Col.9
Col.10
Col.11
Col.12
-0.385***
-0.493***
(0.12)
(0.14) -0.391***
-0.437***
(0.113)
(0.146)
866
720
0.130** (0.0552)
TE D
Dispersion Herfindal Index Concentration
Observations
M AN U
Contestability Shapley
x P75=1 x Type
SC
Presence Blockholder presence
x P75=1
RI PT
VARIABLES
866
720
866
720
866
720
Notes: Column P75 = 1 reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) variable in a regression where only these two variables are interacted (two way interaction). Column p75=1 x Type reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) corresponding to a regression where ownership type is included in the interaction specification (three way interaction) as in the full base line regression equations 5 to 7. Average marginal effect estimates the predicted value of the estimated equation for each observed value of the variable of interest and the average effect is the average of all predicted values. For binary variables, it obtains the predicted values for the ones and zeroes and estimates the difference. In the two way interaction the average marginal effect can be calculated by examining the raw coefficients presented in table A.3 from which the single marginal effects of blockholder presence is 0.0245 and the single marginal effect of the interaction between blockholder presence and high liquidity is 0.11 (Table A.3 Eq.2). The combined marginal effect of blockholder presence is the derivative of the dependent variable with respect to presence: ∂Q/∂Presence = + = 0.0245 + 0.11 = 0.135 which is the value in Col.1; Similarly, the single marginal effect of the interaction between the Shapley value and high liquidity is -0.148 (Table A3 Eq.5). Thus overall marginal effect is ∂Q/∂SV = + = −0.116 − 0.148 = −0.264; which is the value reported in Col.3. The remainder marginal effects are obtained in the same way.
ACCEPTED MANUSCRIPT
7
AC C
EP
TE D
M AN U
SC
Heteroscedasticity corrected standard errors in parentheses; Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation.
RI PT
Columns label as x P75=1 x Type report the marginal effect corresponding to the three way interaction between stock liquidity dummy, type of second blockholder and the variable of interest. In this case the average marginal effect implies a nonlinear effect after taking the average of all observed values. Williams (2012) details the procedure undertaken in the statistical analysis software Stata.
ACCEPTED MANUSCRIPT 8
Table 7 - Instrumental variables regressions full over-identified equations (m = 3) Dependent Variable Tobin' Q
VARIABLES Presence Blockholder presence
Eq.1
Eq.2
Eq.3
Eq.4
Eq.5
Eq.6
...
...
...
...
...
(0.914)
...
...
...
...
...
...
-0.148
...
...
(0.161)
...
Contest. index 1
... ...
... ...
0.0713 (0.132)
Contest. index 2
...
...
...
...
...
Dispersion Herfindal Index Concentration
...
...
Herfindal Index Differentials
... ...
... ...
Contestability Shapley
...
...
...
...
...
... ...
... ...
... ...
0.027
...
...
...
(0.0955)
...
...
...
...
-0.184
...
... ...
... ...
(0.211) ...
... -0.199
...
...
...
...
(0.200)
-0.216***
-0.201***
-0.199***
-0.199***
-0.205***
-0.204***
(0.0560)
(0.0469)
(0.0470)
(0.0471)
(0.0475)
(0.0473)
0.647 (0.421)
0.0263 (0.0906)
-0.0590 (0.0523)
-0.0596 (0.0551)
-0.0806* (0.0476)
-0.0932* (0.0488)
1.42E-03
-3.55E-03
-3.47E-03
-3.32E-03
-3.58E-03
-3.57E-03
TE D
Sales (Growth)
M AN U
Wedge
Leverage Beta
(5.36E-03) 0.375**
(2.72E-03) 0.240**
(2.74E-03) 0.235*
(2.73E-03) 0.238**
(2.73E-03) 0.243**
(2.76E-03) 0.243**
(0.175)
(0.120)
(0.120)
(0.120)
(0.120)
(0.120)
-0.0441 (0.0307)
-0.0252 (0.0199)
-0.0232 (0.0204)
-0.0237 (0.0204)
-0.0284 (0.0200)
-0.0291 (0.0204)
0.105**
0.021
0.018
0.018
0.020
0.020
(0.0441) -0.0441
(0.0126) -0.110
(0.0119) -0.111
(0.0116) -0.114
(0.0125) -0.110
(0.0127) -0.110
(0.147)
(0.108)
(0.109)
(0.109)
(0.108)
(0.108)
-5.84e-05*
-6.13e-05**
-6.35e05***
-6.27e05***
-5.86e-05**
-6.02e-05**
(3.40e-05) 3.059***
(2.38e-05) 2.042***
(2.38e-05) 1.951***
(2.38e-05) 1.961***
(2.40e-05) 2.069***
(2.40e-05) 2.054***
(0.712)
(0.333)
(0.324)
(0.324)
(0.343)
(0.333)
0.431** (0.207)
0.331* (0.173)
0.319* (0.173)
0.325* (0.173)
0.322* (0.175)
0.310* (0.174)
-1.109**
-0.291*
-0.264*
-0.258*
-0.305**
-0.295**
(0.505) 0.0560
(0.152) -0.165
(0.144) -0.165
(0.144) -0.167
(0.153) -0.147
(0.149) -0.143
(0.247)
(0.180)
(0.181)
(0.181)
(0.181)
(0.182)
EP
Lag FCF ratio
AC C
Lag tangibility
SC
...
...
Finance Size
RI PT
1.640*
Country Finance and Institutional EMBI
Domestic credit (w.r.t U.S.) Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.)
ACCEPTED MANUSCRIPT 9
Table 7 - Instrumental variables regressions full over-identified equations (m = 3) (Cont.) Dependent Variable Tobin' Q
VARIABLES Macroeconomic Inflation (cpi)
Eq.1
Eq.2
Eq.3
Eq.4
Eq.5
Eq.6
-4.43E-03 (3.77E-03)
-4.84E-03 (3.74E-03)
-4.70E-03 (3.74E-03)
-4.16E-03 (3.83E-03)
-4.31E-03 (3.81E-03)
Market cap. / GDP
0.0051***
0.00451***
0.0044***
0.0044***
0.0046***
0.0045***
GDP growth (pc)
(8.78E-04) 0.0150***
(6.50E-04) 0.0159***
(6.45E-04) 0.0159***
(6.46E-04) 0.0158***
(6.49E-04) 0.0158***
(6.45E-04) 0.0159***
(3.83E-03)
(2.60E-03)
(2.63E-03)
(2.63E-03)
(2.61E-03)
(2.61E-03)
0.203***
0.192***
0.192***
0.192***
0.194***
0.193***
(0.0383)
(0.0287)
(0.0288)
(0.0288)
(0.0289)
(0.0289)
2,689 426
2,689 426
2,689 426
2,689 426
2,689 426
2,689 426
No
No
No
No
No
No
Underidentification. p-value Valid inst Hansen J. H_0: Overidentification
No Yes
No Yes
No Yes
Yes
Yes
Yes
Yes
Yes
Slope 0.132
Slope 0.128
Slope 0.129
Slope 0.132
Slope 0.130
9.482
9.375
9.378
9.531
9.481
2.45
0.04
0.21
0.44
1.08
[0.118]
[0.849]
[0.977]
[0.644]
[0.506]
[0.298]
3.3
60.9
65.0
60.2
60.7
39.4
Yes Slope …. 6.140
8.17E-04
7.1
101.3
108.2
106.6
76.5
60.6
[0.029]
[0.000]
[0.000]
[0.000]
[0.000]
[0.000]
AC C
Valid inst Hansen J. p-value
No Yes
TE D
Endogeneity p-value Weak inst. H_0: Weak identification Underidentification. H_0: Underidentification
No Yes
EP
Endogeneity H_0: x_j is not endogenous Vs. x_j is endogenous
No Yes
SC
Observations Number of id_firm_n Sector dummy Year dummy Robust SE HAC Interaction R2 centered F-stat
M AN U
Year dummy (2008)
RI PT
3.66E-03 (7.03E-03)
2.860
9.221
8.472
8.511
9.462
9.381
[0.091]
[0.010]
[0.015]
[0.014]
[0.009]
[0.009]
Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 Notes: Estimation results for equations 1 to 6 using instrumental variables to check endogeneity of the dependent variable. HAC stands for heterosckedastic and autocorrelation consistent standard errors. Each regression uses 3 join instruments: i) the lag of blockholder presence, contestability and dispersion definition; ii) operating income volatility; iii) cash flow ratio volatility. Slope interactions refers to presence, contestability and dispersion variables times 75th percentile of stock liquidity dummy. Regressions are controlled by second blockholder type dummies. All variable definitions are listed in Appendix – Table A.1. Source: Authors’ estimation.
ACCEPTED MANUSCRIPT 10
APPENDIX Table A.1 Definition of Variables Source
Definition
Variable Ownership Blockholder Presence
Shareholder 2
Equity share of the 2nd largest blockholder
Shareholder 3
Equity share of the 3rd largest blockholder
Shareholder 4
Equity share of the 4th largest blockholder
Wedge
Shapley Value to control rights largest shareholder ratio
ROA Contestability Shapley Contestability Index 1 Contestability Index 2 Wedge Dispersion
SC
Anual Reports-Thomson Reuters
The Shapley Value solution for the largest shareholders in a three-voting game The share of the second to the largest blockholder ratio.
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
The share of the second and third blockholders to the largest blockholder ratio. Shapley Value to control rights largest shareholder s1 Herfindahl concentration index
H.I Differences
The sum of the squares of the differences between the first and the second H.I Con.
Leverage Stock Beta
Free Cash Flow
Tangibility
AC C
Sales (growth)
EP
H.I Concentration
Finance Size
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Own estimates
Market value/book value: [stock market capitalization + long term liabilities + book value of prefered stocks (if any) + minority interest]/[book value of assets] The ratio of net profits after tax to total assets
TE D
Tobin's Q
M AN U
Performance
Anual Reports-Thomson Reuters
RI PT
Shareholder 1
Dummy variable for firms with blockholder as an investor with equity rights (direct votes) with less than 50% and more than 10% ownership in a listed company Equity share of the largest blockholder
Natural Logarithm of total asssets Percentage of sales growth compared to previous year. The ratio of total liabilities to total assets Standard measure of systemic risk for each firm’s stock with respect to the market. It was measure for stocks which have been active in the stock market for more than 180 days in a given year. Free Cash Flow (FCF) to equity ratio Is defined as EBITDA minus tax expenditure minus interest expenditures to total firm equity. This is an indicator for firm short term liquidity. Total property, plant and equipment divided by total assets
Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
ACCEPTED MANUSCRIPT 11
Table A.1 Definition of Variables (Cont.) Source
Definition
Business Freedom (w.r.t.US)
Financial Freedom (w.r.t.US)
Measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and thus facilitate lending. Is an overall indicator of the efficiency of government regulation of business. The score for each country is a number between 0 and 100, with 100 equaling the freest business environment. The score is based on 10 factors, all weighted equally, using data from the World Bank’s Doing Business study.
The World Bank — world development indicators
Is a measure of banking efficiency as well as a measure of independence from government control and interference in the financial sector. Index from 0 to 10.
Index of economic freedom
Inflation as measured by the consumer price index reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services Market capitalization (also known as market value) is the share price times the number of shares outstanding, divided by GDP Annual percentage growth rate of GDP per capita based on constant local currency.
The World Bank — world development indicators
RI PT
Legal Rights (w.r.t.US)
Anual Reports-Thomson Reuters The World Bank — world development indicators
Market Cap/ GDP
AC C
EP
GDP Growth (pc)
TE D
Macroeconomic Inflation (CPI)
Index of economic freedom
SC
Domestic Credit (w.r.t.US)
Emerging Markets Bonds Index. Measures financial risk in each country. Expressed in basis points Domestic credit to private sector: refers to financial resources provided to the private sector, such as through loans, purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment
M AN U
Variable Institutional EMBI
The World Bank — World development indicators The World Bank — world development indicators
ACCEPTED MANUSCRIPT 12
Table A.2 Identification - empirical equation Tobin's Q and Shapley Value Fixed Effects Regressions
Contestability Shapley
0.217*** (0.0804) Firm financial and specific variables Size Wedge
Leverage Beta Lag free cash flow ratio
AC C
Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.)
Fin. freedom (w.r.t U.S.)
-0.148 (0.110)
-0.116 (0.112)
0.0780** (0.0391) -0.133 (0.0940) 0.00535 (0.00417) 0.245* (0.130) 0.0120 (0.0209) 0.0160** (0.00754) -0.559*** (0.119)
-0.217*** (0.0577) 0.00756 (0.0796) -0.00150 (0.00340) 0.277** (0.135) -0.0170 (0.0214) 0.00158 (0.0170) -0.235** (0.101)
-0.218*** (0.0576) 0.0181 (0.0798) -0.00150 (0.00339) 0.277** (0.135) -0.0140 (0.0209) 0.00173 (0.0168) -0.238** (0.101)
-5.68e05** (2.23E-05)
-5.38e05** (2.23E-05)
-6.49e05** (2.74E-05)
2.204*** (0.430) 0.407** (0.187)
2.200*** (0.430) 0.404** (0.188)
2.051*** (0.422) 0.330 (0.210)
-0.293* (0.176) -0.165 (0.213)
-0.300* (0.175) -0.183 (0.212)
-0.275 (0.181) -0.195 (0.237)
EP
Country institutional variables
Domestic credit (w.r.t U.S.)
0.199* (0.120)
TE D
Lag tangibility
EMBI
+ Turnover interaction x P75 = 1
Marginal Effect
M AN U
Sales (Growth)
+ Country and Macro controls
+ Turnover and Type x p75 x Type
RI PT
+ Firm specific controls
-0.264** (0.123)
SC
Coeff
VARIABLES
-0.0114 (0.152)
-0.206*** (0.0585) 0.0321 (0.0811) -0.00369 (0.00287) 0.211 (0.152) -0.0282 (0.0214) 0.0153 (0.0123) -0.122 (0.119)
Marginal Effect
-0.345** (0.135)
ACCEPTED MANUSCRIPT 13
Table A.2 Identification - empirical equation Tobin's Q and Shapley Value (Cont.) Fixed Effects Regressions
Market cap. / GDP GDP growth (pc) Year dummy (2008) Constant
1.068***
0.939***
(0.0565)
(0.242)
6,305
4,426
0.005 No
0.031 No
No
No
Yes
Yes
Interaction
Level + slope
Number of id_firm_n
548
-5.24E-03
-5.10E-03
(3.65E-03) 0.00468***
(3.66E-03) 0.00472***
(6.99E-04)
(6.99E-04)
0.0156*** (2.60E-03)
0.0156*** (2.59E-03)
0.0160*** (2.47E-03)
0.203***
0.202***
0.194***
Marginal Effect
+ Turnover and Type x p75 x Type
Marginal Effect
-5.85E-03
(3.62E-03) 0.00456*** (7.41E-04)
(0.0313) 2.062***
(0.0311) 2.060***
(0.0306) 1.873***
(0.346)
(0.346)
(0.368)
3,282
3,280
866
2,699
720
0.118 No
0.120 No
No
0.137 No
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Level + slope
Level + slope
Level + slope
TE D
Observations R-squared Sector dummy Year dummy Robust SE
+ Turnover interaction x P75 = 1
RI PT
Macroeconomic variables Inflation (cpi)
+ Country and Macro controls
SC
+ Firm specific controls
M AN U
Coeff
VARIABLES
Level + slope
Level + slope
Level + slope
475
474
474
435
AC C
EP
Notes: The table reports the identification process of the Tobin’s Q empirical equation for the case of the Shapley Value. It begins with a simple regression and then regressions add the firm-specific, country controls, and the interacting terms of stock liquidity and second blockholder identity. Columns Marginal effects show the marginal effect of the immediate left column. Robust standard errors in parentheses. Significance: *** p < 0.01, ** p < 0.05, * p<0.1
ACCEPTED MANUSCRIPT
14
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within)
Variables Presence Blockholder presence
x P75=1
Eq.1
Eq.2
Eq.3
0.0455 (0.0479)
0.0245 (0.0482)
-0.0144 (0.0914)
Contestability Shapley
Contest Index 1
Coeff.
x P75=1
x P75=1 x Type
Eq.4
Eq.5
Eq.6
-0.148 (0.110)
-0.116 (0.112)
-0.0114 (0.152)
Contest. index 1
H.I. Differences Interaction stock liquidity dummy -0.0919 (0.0969) 0.203*
(0.0558)
(0.123)
0.266** (0.132)
-0.148** (0.0751)
-0.352* (0.180)
EP
P75 x contest_index 2 |
AC C
p75 x contest_index 1
P75 x hi_concentration
0.0548 (0.0603)
TE D
-0.0992** (0.0419) 0.110**
P75 x Shapley
P75 x hi_differences
Eq.7
Eq.8
Eq.9
0.100 (0.0732)
Coeff. Eq.10
HI Concentration x P75=1 Eq.11
x P75=1 x Type Eq.12
Coeff.
x P75=1
x P75= Typ
Eq.13
Eq.14
E
-0.110
-0.0546
-0.0
(0.120)
(0.124)
(0.2
0.0569 (0.0474)
0.1 (0.1
-0.330*** (0.108)
-0.5 (0.3
-0.121 (0.125)
M AN U
Dispersion H.I. Concentration
P75 x blockholder presence
x P75=1
0.129* (0.0683)
Contest. index 2
P75 dummy
Coeff.
Contest Index 2 x P75=1 x Type
RI PT
Coeff.
Shapley x P75=1 x Type
SC
Presence
-0.0762 (0.0488)
-0.0364 (0.115)
0.0909 (0.0764)
0.184 (0.174)
0.112**
0.0977*
-0.0464
(0.0502)
(0.0545)
(0.0815)
-0.0725 (0.0463)
-0.0527 (0.106)
0.0495
0.153
(0.0487)
(0.101)
ACCEPTED MANUSCRIPT
15
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within) (Cont.)
Leverage Beta L.fcf_r L.tangibility
-0.218*** (0.0582) -0.0480 (0.0617) -0.00143 (0.00339) 0.282** (0.134) -0.0164 (0.0214) 0.00179 (0.0172) -0.235** (0.102)
Country Institutional and Financial EMBI -5.70e-05** (2.25e-05) Domestic credit (w.r.t U.S.) 2.163*** (0.427) Legal rights (w.r.t U.S.) 0.399** (0.188) Buss. freedom (w.r.t U.S.) -0.272 (0.172) Fin. freedom (w.r.t U.S.) -0.166 (0.214) Macroeconomic Inflation (cpi) -0.00552 (3.63E-03) Market cap. / GDP 0.00463*** (6.97E-04) GDP growth (pc) 0.0157*** (2.61E-03) Year dummy (2008) 0.202*** (0.0313)
Coeff. Eq.4
-0.221*** (0.0581) -0.0426 (0.0617) -0.00141 (0.00338) 0.285** (0.134) -0.0139 (0.0209) 0.00233 (0.0170) -0.237** (0.102)
-0.214*** (0.0598) -0.0522 (0.0607) -0.00362 (0.00287) 0.225 (0.152) -0.0291 (0.0216) 0.0167 (0.0133) -0.124 (0.119)
-0.217*** (0.0577) 0.00756 (0.0796) -0.00150 (0.00340) 0.277** (0.135) -0.0170 (0.0214) 0.00158 (0.0170) -0.235** (0.101)
-5.36e-05** (2.26e-05) 2.161*** (0.426) 0.396** (0.188) -0.282* (0.171) -0.173 (0.213)
-6.40e-05** (2.77e-05) 2.037*** (0.419) 0.308 (0.208) -0.261 (0.178) -0.170 (0.240)
-5.68e-05** (2.23e-05) 2.204*** (0.430) 0.407** (0.187) -0.293* (0.176) -0.165 (0.213)
-0.00540 (3.63E-03) 0.00468*** (6.98E-04) 0.0156*** (2.59E-03) 0.202*** (0.0311)
-0.00631* (3.58E-03) 0.00455*** (7.37E-04) 0.0160*** (2.49E-03) 0.193*** (0.0306)
x P75=1
Coeff.
Eq.5
Eq.6
Eq.7
-0.218*** (0.0576) 0.0181 (0.0798) -0.00150 (0.00339) 0.277** (0.135) -0.0140 (0.0209) 0.00173 (0.0168) -0.238** (0.101)
-0.00524 (3.65E-03) 0.00468*** (6.99E-04) 0.0156*** (2.60E-03) 0.203*** (0.0313)
Contest Index 2 x P75=1
x P75=1 x Type
Coeff.
Eq.8
Eq.9
Eq.10
RI PT
Eq.3
H x P75=1
x P75=1 x Type
C
Eq.11
Eq.12
E
-0.206*** (0.0585) 0.0321 (0.0811) -0.00369 (0.00287) 0.211 (0.152) -0.0282 (0.0214) 0.0153 (0.0123) -0.122 (0.119)
-0.217*** (0.0578) -0.0413 (0.0599) -0.00149 (0.00339) 0.270** (0.134) -0.0139 (0.0216) 0.00164 (0.0169) -0.240** (0.101)
-0.217*** (0.0579) -0.0365 (0.0600) -0.00141 (0.00339) 0.271** (0.135) -0.0106 (0.0214) 0.00157 (0.0168) -0.243** (0.101)
-0.201*** (0.0572) -0.0654 (0.0561) -0.00366 (0.00284) 0.196 (0.152) -0.0247 (0.0219) 0.0165 (0.0126) -0.126 (0.118)
-0.217*** (0.0573) -0.0262 (0.0599) -0.00138 (0.00339) 0.265** (0.134) -0.0142 (0.0215) 0.00132 (0.0168) -0.240** (0.101)
-0.217*** (0.0573) -0.0215 (0.0599) -0.00133 (0.00338) 0.267** (0.135) -0.0108 (0.0213) 0.00132 (0.0167) -0.243** (0.101)
-0.200*** (0.0573) -0.0481 (0.0564) -0.00346 (0.00284) 0.192 (0.152) -0.0255 (0.0220) 0.0157 (0.0122) -0.125 (0.118)
-5.38e-05** (2.23e-05) 2.200*** (0.430) 0.404** (0.188) -0.300* (0.175) -0.183 (0.212)
-6.49e-05** (2.74e-05) 2.051*** (0.422) 0.330 (0.210) -0.275 (0.181) -0.195 (0.237)
-5.77e-05*** (2.23e-05) 2.145*** (0.423) 0.404** (0.188) -0.266 (0.171) -0.166 (0.213)
-5.55e-05** (2.22e-05) 2.137*** (0.423) 0.398** (0.188) -0.277 (0.170) -0.177 (0.212)
-6.80e-05** (2.77e-05) 1.910*** (0.406) 0.307 (0.209) -0.246 (0.176) -0.178 (0.236)
-5.79e-05*** (2.22e-05) 2.157*** (0.422) 0.413** (0.187) -0.272 (0.171) -0.164 (0.212)
-5.57e-05** (2.21e-05) 2.150*** (0.423) 0.409** (0.187) -0.285* (0.170) -0.176 (0.212)
-6.73e-05** (2.77e-05) 1.942*** (0.409) 0.316 (0.208) -0.252 (0.177) -0.185 (0.236)
-0.00510 (3.66E-03) 0.00472*** (6.99E-04) 0.0156*** (2.59E-03) 0.202*** (0.0311)
-0.00585 (3.62E-03) 0.00456*** (7.41E-04) 0.0160*** (2.47E-03) 0.194*** (0.0306)
-0.00573 (3.62E-03) 0.00463*** (6.94E-04) 0.0157*** (2.60E-03) 0.201*** (0.0314)
-0.00557 (3.65E-03) 0.00465*** (6.95E-04) 0.0156*** (2.59E-03) 0.201*** (0.0313)
-0.00661* (3.57E-03) 0.00443*** (7.23E-04) 0.0167*** (2.48E-03) 0.193*** (0.0305)
-0.00559 (3.63E-03) 0.00464*** (6.93E-04) 0.0156*** (2.60E-03) 0.202*** (0.0313)
-0.00547 (3.64E-03) 0.00466*** (6.94E-04) 0.0156*** (2.59E-03) 0.202*** (0.0312)
-0.00638* (3.58E-03) 0.00445*** (7.32E-04) 0.0165*** (2.47E-03) 0.194*** (0.0305)
SC
Sales (Growth)
Eq.2
Contest Index 1 x P75=1 x Type
TE D
Wedge
x P75=1
Eq.1
EP
Firm Characteristics Size
Coeff.
AC C
Variables
Shapley x P75=1 x Type
M AN U
Presence
ACCEPTED MANUSCRIPT
16
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within) (Cont.)
Regression Statistics Observations R-squared Number of id_firm_n Sector dummy Year dummy Robust SE
Eq.2
Eq.3
Coeff. Eq.4
x P75=1
Coeff.
Eq.5
Eq.6
Eq.7
Contest Index 2 x P75=1
x P75=1 x Type
Coeff.
Eq.8
Eq.9
Eq.10
HI Concentration x P75=1
x P75=1 x Type
Coeff.
x P75
Eq.11
Eq.12
Eq.13
Eq.14
2.018*** (0.349)
2.053*** (0.350)
2.026*** (0.379)
2.062*** (0.346)
2.060*** (0.346)
1.873*** (0.368)
1.984*** (0.346)
2.015*** (0.347)
2.047*** (0.369)
1.940*** (0.340)
1.972*** (0.342)
1.993*** (0.367)
2.108*** (0.347)
3,282 0.117 474 No No Yes Level + slope 4.63E-04 9.69E-03 0.117
3,280 0.119 474 No No Yes Level + slope 4.85E-04 1.01E-02 0.119
2,699 0.135 435 No No Yes Level + slope 7.69E-06 5.20E-03 0.135
3,282 0.118 474 No No Yes Level + slope 4.85E-04 9.90E-03 0.118
3,280 0.120 474 No No Yes Level + slope 4.83E-04 1.02E-02 0.120
2,699 0.137 435 No No Yes Level + slope 5.87E-06 5.51E-03 0.137
3,282 0.118 474 No No Yes Level + slope 5.10E-04 1.02E-02 0.118
3,280 0.119 474 No No Yes Level + slope 5.09E-04 1.06E-02 0.119
2,699 0.136 435 No No Yes Level + slope 1.14E-05 4.66E-03 0.136
3,282 0.119 474 No No Yes Level + slope 5.19E-04 1.04E-02 0.119
3,280 0.120 474 No No Yes Level + slope 5.22E-04 1.08E-02 0.120
2,699 0.137 435 No No Yes Level + slope 1.30E-05 4.96E-03 0.137
3,282 0.117 474 No No Yes Level + slope 4.40E-04 9.55E-03 0.117
-27.30
6.662
72.51
1.112
3.155
68.15
15.26
16.30
102.0
29.69
28.14
104.8
-28.89
TE D
Interaction R2 overall R2 between R2 within Hausman test H0: RE Vs. H1: FE (stat)
x P75=1
Eq.1
SC
Constant
Coeff.
Contest Index 1 x P75=1 x Type
M AN U
Variables
Shapley x P75=1 x Type
RI PT
Presence
EP
Notes: This table the Tobin´s Q regression looking at the variables of blockholder control: presence, contestability and dispersion. The regression coefficient of each variable of interest are reported in Eq.1, Eq.4, Eq.7, Eq.10, Eq.13 and Eq.16; Columns label by xP75=1 report the interaction with the stock liquidity dummy at 75Th percentile. Columns label as x P75=1 x report the marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) and second blockholder ownership type (three way interaction). All variable definitions are listed in Appendix – table A.1.
AC C
Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation
2.1 (
4. 1.
17
ACCEPTED MANUSCRIPT
TE D
M AN U
SC
RI PT
Figure 1 - Firm type, marginal effect.
AC C
EP
Source: Own estimates. Notes: The figure shows the marginal effect of the variables of interest (blockholder presence, contestability and ownership concentration) whenever the high concentration and the identity blockholder are equal to one. These figures allow distinguishing the effect of the blockholder upon performance when identity is studied.
…….End of document…..
ACCEPTED MANUSCRIPT
Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America
RI PT
Highlights • The study examines the relationship between firm value and blockholder control in three dimensions: presence, contestability and dispersion. • Blockholder contestability becomes effective on firms with high liquid stocks.
SC
• The existence and type of a second blockholder within firms with multiple blockholder structures becomes a critical factor in explaining the effect on firm value.
AC C
EP
TE D
M AN U
• Results show that blockholder voice (monitoring) versus exit (trade) models are complementary rather than substitute mechanisms for firm corporate governance.
S1059-0560(17)30406-9
DOI:
10.1016/j.iref.2017.05.012
Reference:
REVECO 1432
To appear in:
International Review of Economics and Finance
Received Date: 21 November 2015 Revised Date:
10 May 2017
Accepted Date: 18 May 2017
Please cite this article as: Pombo C. & Taborda R., Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America, International Review of Economics and Finance (2017), doi: 10.1016/j.iref.2017.05.012. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Resubmitted version.R2
This version May 8 - 2017
SC
Abstract
RI PT
Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America
JEL Classification: G32, G34.
M AN U
This paper examines the relationship between firm value and blockholders in Latin America. Econometric results for a comprehensive data set of more than 550 firms listed in the six largest stock markets of the region support a positive effect of variables measuring the existence, contestability, dispersion and identity of blockholder on performance (Tobin’s Q) only for highly liquid stocks. The identity of the second largest blockholder (family, foreign, financial or the State) emerges as critical for such effects. The study supports that the voice and exit models are complementary rather than substitute mechanisms for firm corporate governance.
AC C
EP
TE D
Keywords: Multiple large blockholders, stock liquidity, firm value, corporate governance, Latin America
1
ACCEPTED MANUSCRIPT Resubmitted version.R2
1. Introduction Recent empirical literature on blockholder ownership has focused on the presence of multiple
RI PT
large shareholders and their interactions, and how these can affect firm management either through direct intervention—monitoring—or through share trading. Edmans and Manso (2011), for example, argue that the presence of multiple blockholders creates two conflicting governance effects. In the first, known as the voice mechanism or shareholder activism, multiple
SC
blockholders serve “as a commitment device to reward or punish the manager ex post for his actions.” By contesting control, blockholders can align interests to implement either profitable
M AN U
projects or simply to monitor managers. The second effect, known as the exit mechanism, occurs in firms with multiple blockholder structures and sees dispersed (and small) blockholders punish the largest shareholder, or management, by exiting the firm, i.e. by trading their shares, thus affecting firm value. This approach to corporate governance highlights that, by possessing better
TE D
information about firm value, smaller blockholders can use the exit strategy as a device to discipline the behavior of the largest blockholder. On the one hand, when blockholders contest control, a positive effect on monitoring appears, since the second or third blockholders are given
EP
more equal voting rights in circumstances where, due to other mechanisms, they cannot exert absolute control. These mechanisms may include the issuing of non-voting shares or multiple
AC C
share classes, pyramidal and cross-share ownership structures, or the existence of disproportionate board representation (Villalonga and Amit, 2010). On the other hand, a negative effect of blockholder ownership may result from the possibility that small shareholders might be subject to expropriation by large shareholders, turning the agency problem from one of controlling blocks to controlling minorities. In such circumstances firm management responds to the interests of the controlling owners, reducing firm value when
2
ACCEPTED MANUSCRIPT Resubmitted version.R2
large shareholders derive private benefits from related party transactions. In addition, an excess of monitoring by blockholders induces a loss of value, since management may refrain from certain profitable initiatives when a vigilante approach is followed by large shareholders.
RI PT
The paradigm of the benefits of a widespread ownership structure, mainly fostered after the prevalent, but not precisely right, view of diffuse ownership in the United States, led the academic agenda to study corporate governance in settings were large shareholders dominate the
SC
landscape.1 At the same time, this view defined a policy agenda towards developing countries in favor of increasing ownership dispersion as corporate governance mechanism. This perspective
M AN U
has evolved in line with the reality of blockholders as the dominant ownership structure in corporate finance. Moshirian et al. (2014) summarize how large ownership is also a prevalent corporate structure in developed countries. The growth of institutional investors worldwide during the last four decades as large shareholders explains in part the presence of blockholder
TE D
and investor activism within developed capital markets (Agrawal et al., 2011). Under this new reality, whereby widespread ownership has come to be used as a mechanism for increasing corporate governance and ultimately corporate value in developing countries, public
EP
policy is at a crossroads. Where to go? Should corporate governance be increased or ownership structures disseminated? The former approach has been widely accepted and implemented in
AC C
developing countries (McGee, 2009). Whether or not such positive initiatives exist in practice, beneficial effects on corporate value are not guaranteed. The latter approach, on the other hand, cannot be warranted, as a policy interventions aimed at reducing the concentration ownership are hardly an economic policy option. The highly concentrated ownership structure of public and private firms in Latin America provides a useful scenario for an empirical investigation of the 1
An example of the argument that ownership is widespread is provided by the work of Becht (2001), while Holderness (2009) has been pivotal in debunking this view.
3
ACCEPTED MANUSCRIPT Resubmitted version.R2
association between firm value and ownership in developing countries. Leaving aside the option of dispersed ownership or tighter corporate governance laws or guidelines, the remaining question is how intermediate ownership structures relate to firm value.
RI PT
Several papers have addressed some of these issues. The blockholder entry and market response was studied by Park et al. (2008), showing how the acquirer’s profile shapes abnormal returns differently. Ali et al. (2017) establish how better governance influence liquidity in
SC
Australian listed firms. And extending this work, Nadarajah et al. (2016) establish how more liquid firms have lower leverage once corporate governance mechanisms are accounted for.
M AN U
Closely related to Latin American experience with corporate governance, the work of Cueto (2013) examines corporate governance and ownership concentration and the effect upon firm value. Given that high ownership concentration is prevalent in the region, the author suggests that to minimize its negative effect, firms engage in corporate governance and institutional
TE D
investors engage in monitoring and restraint asset appropriation.
The paper contributes to the literature on multiple blockholders in empirically addressing two analytical alternatives that can shed light on approaches to increase firm value under a setting of
EP
high ownership concentration in emerging markets. First, we claim that blockholder direct monitoring or voice is an effective channel for reducing shareholders’ agency costs and
AC C
improving firm corporate governance but its effectiveness is conditional to stock liquidity. In other words, contestability among blockholders is only effective and shareholders’ exit strategy credible for high liquid stocks within financial markets where capital deepening is a structural issue. Second, we provide evidence that in firms with multiple blockholder ownership structure but without absolute control by the largest shareholder, the type of the second blockholder is central in explaining firm valuation.
4
ACCEPTED MANUSCRIPT Resubmitted version.R2
Ownership structures in Latin America led us to hypothesize that the second blockholder is crucial in this connection and that the situation described for the region might also elucidate the case of other emerging markets. Nearly 50% of the firms in this study have a controlling
RI PT
blockholder with equity rights above 50%, while fewer than 5% of firms within the sample have no kind of blockholder at all (i.e. all their shareholders have less than 10% of equity). The remaining 45% of firms in the sample exhibit multiple blockholders without direct control, but
SC
that are still of considerable size (more than 10% of equity), meaning that there is scope for forming controlling coalitions among the largest blockholders. The frequency distribution of
M AN U
ownership within this last group of firms shows that when the largest voting block colludes with the second largest, the coalition represents around 45% of direct votes, a figure that rises to 54% when the third top blockholder is included. High ownership concentration implies low separation ratios between ownership and control. Empirical country-level case studies on corporate
TE D
ownership in Latin America, based on real sector firms, show that separation ratios between ownership and control are close to 0.85 (Gutierrez and Pombo, 2009). This means that a controlling shareholder will need 42% of equity rights if they are to obtain control over a firm’s
EP
voting rights. Thus, focusing the empirical analysis on the role of blockholder coalitions and their effects on firm value, measured by direct votes, is fully justified by the circumstances on
AC C
the ground.2
Our work extends prior results of benchmark studies of multiple blockholders and firm value. Maury and Pajuste (2005) based on a sample of 136 Finnish firms for the 1993-2003 period, found that contestability and blockholder dispersion reduces firm value. They only report the regression coefficients and direct effect through Shapley coalitional value, the Herfindal 2
Indirect estimation of the wedge between ownership and control puts it at 0.75 - similar to the results of direct measurement. The indirect wedge indicator is defined as the Shapley value solution of a four-voting oceanic game to largest shareholder cash flow rights (Guedes and Loureiro, 2006).
5
ACCEPTED MANUSCRIPT Resubmitted version.R2
concentration and the differences in shareholder rights indices. There is no discussion on the role of stock liquidity. Similar results were found by Laeven and Levine (2008) for a sample of 1657 European firms but these results were limited to estimates for a cross section. The authors report
RI PT
a negative premium of cash flow rights dispersion between the largest and second largest blockholder on firms’ Tobin’s Q without the presence of a controlling owner with majority control rights (ownership share > 0.50). They did not consider any discussion on the role of
SC
market liquidity and blockholder trading as a blockholders’ disciplinary device. Konijn et al. (2011) study the influence of blockholder presence, size and dispersion on firms’ Tobin’s Q
M AN U
based on a sample of 3772 firm-year observations of US listed companies. They found a negative correlation between blockholder presence and size on firm value. They disentangle the blockholder effect produced by insiders and outsiders, where the former have a positive effect of firm value offset by the negative effect of outsiders. They control for stock liquidity in their
TE D
regression but they do not model blockholder contestability. Their results contradict what has been found for Continental Europe and Asia arguing that such results might depend on different regional and institutional features that need more research.
EP
The results of this study on the relationship between blockholders and firm value takes into account the dimensions of presence, contestability, control rights dispersion, and type of
AC C
blockholder. The empirical design presents a baseline regression that takes into account the full set of interactions among stock liquidity and blockholder ownership structure (refined in terms of presence, contestability, and dispersion). All results are presented in terms of the marginal effect of blockholder ownership on firm value conditioning for liquidity and blockholder type. In previous studies analyzing blockholder structure and corporate value, Latin America has not been studied in depth. The region is a classic example of emerging economies in terms of
6
ACCEPTED MANUSCRIPT Resubmitted version.R2
economic and financial development. Based on a comprehensive dataset of 562 Latin American corporations listed in the six largest stock markets of the region from 1997 to 2011, the econometric results support a positive effect from variables measuring contestability, dispersion
RI PT
and type of blockholder on Tobin’s Q only for high liquid stocks. In this sense, we report that the presence of multiple blockholders implies a premium of 4.5% on firms’ Tobin’s Q. Nonetheless, this premium increases by up to 13.5% after controlling for the interacting term of stock
SC
liquidity, and it rises to 17.1% if we expand interaction for the second blockholder type dummy variables (family, local, foreign, financial and state). Under this specification, we found that
M AN U
regardless of whether the second blockholder is a financial firm or foreign investor, there is a statistically and economically significant effect on firm value.
The effect of contestability and dispersion on firm value is in line with the presence effect when liquidity is considered. The overall marginal effect increases and becomes statistically robust
TE D
after including the interacting terms of stock liquidity and second blockholder type. Thus, the approach to address the blockholder effect goes beyond presence or size, and expands to shareholder dispersion and the ability to form coalitions and ultimately become a driving force
EP
behind firm value.
Specification tests for linear panel data models favored fixed effects within estimates in
AC C
contrast to the transformation of random effects. We include several control variables at firm level and we also control for country macroeconomic, financial, and institutional variables that are sensitive to local and foreign investors’ systematic risk perceptions. The remainder of the document is structured as follows: Section 2 presents the theoretical framework and development of the working hypotheses; section 3 analyzes the data and
7
ACCEPTED MANUSCRIPT Resubmitted version.R2
variables included in the empirical model; section 4 presents the econometric analysis; and section 5 concludes.
RI PT
2. Theoretical Framework and Hypotheses 2.1 Blockholders contest for control
Theoretical models and empirical studies in corporate governance have addressed settings with more than one large blockholder, circumstances that allow some level of contestability to emerge
SC
on the basis of coalitions that may be formed. Two competing views have been proposed. The first refers to the positive effect of blockholder presence, due to their monitoring role and the
M AN U
second to the benefits of control when large groups of shareholders form coalitions in order to conduct tunneling strategies that enable them to divert a firm’s cash flow. Regarding the first view, Bloch and Hedge (2001) analyzed a firm’s governance framework in which two large blockholders and dispersed shareholders interact in the shareholders’ general
TE D
meeting. These authors show that control contestability between two blockholders that are seeking to attract the votes of small shareholders tends to reduce rent extraction. Bennedsen and Wolfenson (2000) developed a model of firm control where firm founders dilute their (initial)
EP
control by distributing proportional equity holdings to several large shareholders, allowing them to form a coalition, through which, in turn, they can obtain control and curb private benefit
AC C
extraction. Gomes and Novaes (2006) model corporate governance according to two scenarios. The first occurs when insiders, or the initial founders of a firm, face investment opportunities that are difficult to evaluate to the extent that sharing control creates bargaining problems. As a result, ownership structures in which there is
only one large shareholder monitoring
management represent the most efficient system for protecting minority shareholders. However, they also find that in weak legal environments, the optimal ownership structure involves shared
8
ACCEPTED MANUSCRIPT Resubmitted version.R2
control between multiple blockholders. In these circumstances, the presence of multiple blockholders is positively associated with firm value. The second view concerns the benefits of control that accrue when large groups of shareholders
RI PT
form coalitions in order to conduct tunneling strategies that enable them to divert cash flow. Zwiebel (1995) argues that investors may choose to hold a significant block equity in a firm because it confers them partial benefits of control, especially in those cases where the next
SC
blockholder is insufficiently large and they feel the need to form a coalition.
The theoretical model of control contestability had its origins in La Porta et al. (2002), who
M AN U
modeled managerial rent extraction, and in Maury and Pajuste (2005), who extended the baseline model to reveal contestability behavior more explicitly among multiple large shareholders. The model has two main assumptions: that the existence of multiple large shareholders can reduce profit diversion (i.e., that control contestability is value-enhancing), and that diversion of profits
hypotheses of this paper:
TE D
by the controlling coalition is costly. The theoretical work of those studies supports the first two
EP
H1: Firm value (performance) increases with blockholder presence and higher contestability of power.
AC C
H2: If the probability of forming a controlling coalition between the largest shareholder and the next largest blockholders increases, rent diversion grows to the benefit of the coalition block, thus lowering firm value (performance).
2.2 The role of the second large blockholder Empirical research on corporate governance suggests that the type of controlling shareholder present in a given company is essential to understanding the link between ownership structure, agency costs and firm value. Claessens et al. (2002) find that the expropriation of minority shareholders is more likely when the largest controlling shareholder is a family or the state. The
9
ACCEPTED MANUSCRIPT Resubmitted version.R2
underlying assumption is that different types of blockholders have distinct incentives and abilities to monitor the controlling shareholder. Empirical results from regional or country case studies on the marginal effects of blockholder identity and firm value confirm that the kind of
RI PT
second blockholder is vitally important when it comes to contesting the agency costs of controlling owners. For instance, Jara-Bertin et al. (2008), based on sample of 1,200 European corporations, find that in firms where the two largest shareholders are family blocks decrease
SC
firm valuation.. Similarly, of Attig et al. (2009) who studied a sample of 1,252 East Asian public companies find that the presence of either a family or the state as second largest shareholder
M AN U
represents a value premium of between 0.23 and 0.96 on firms Tobin’s Q. In contrast, there is no significant effect if the second blockholder is a widely held firm.
Recent research on family firms has focused on analyzing the effects of the presence of multiple large shareholders on firm value. In two related works on family listed firms in Spain,
TE D
Sacristan et al. (2012, 2015) find that there is a premium of 3.5% on a firm´s return on assets (ROA) when a it has a second blockholder.
Research on the role of institutional investors, mainly from developed capital markets, shows a
EP
consistent positive effect of the presence of this type of investors on firm governance, efficiency and performance. Moreover, it is expected that independent institutional owners –e.g. investment
AC C
firms or banks- are more prone to behave as investors with high levels of control and to act as watchdogs of management because of their low monitoring costs and their capacity to collect information and implement risk management protocols. These investors have less natural potential for developing business relations with the corporations (Almazan et al. 2005). International evidence shows that the premium for the presence of institutional ownership is 0.12 on the Tobin’s Q ratio of a firm (Ferreira and Matos, 2008). Recent studies for Latin America
10
ACCEPTED MANUSCRIPT Resubmitted version.R2
note the premium there as 0.08 (Pombo and De-la-Hoz, 2015). These conceptual elements and findings support the third hypothesis of this study: H3: The presence of an independent second blockholder enhances firm performance.
RI PT
2.3 The role of market liquidity and blockholder voice/trade mechanisms
A liquid stock market provides the opportunity for non-shareholders to become shareholders, it encourages management compensation, reduces managerial opportunism, and promotes
SC
investment. Blockholders’ direct intervention and management monitoring—voice models— might be effective whether market liquidity reduces the cost of monitoring and discloses more
M AN U
private information on assets value or not. The main argument by which blockholder contest can act as a powerful internal firm governance mechanism is refined once market liquidity is studied. This argument can be traced back to Hirschman (1970) who investigated under what conditions the voice option prevails over the exit option and vice versa in shaping the optimal behavior of
TE D
business firms, organizations, and economic institutions. Theoretical models in the corporate finance literature predict different outcomes depending on whether stock liquidity enhances blockholder intervention or not. Maug (1988) develops an extensive theoretical game where a
EP
large shareholder (LSH) buys a firm’s shares and then decides to monitor or trade. He argues that liquidity can enhance monitoring because it allows the LSH to buy additional stakes at a price
AC C
that is lower than the expected equity price after blockholder intervention. In the same vein, Kahn and Winton (1998) within the context of institutional investors as blockholders, develop a two stage dynamic game where stock liquidity for some cases will help investors to monitor and, in other cases, to follow the Wall Street rule and sell their stakes. Liquidity on traded stocks might foster intervention when informed blockholders can exploit the ability to speculate in a given period. In period + 1, minorities are willing to sell to
11
ACCEPTED MANUSCRIPT Resubmitted version.R2
blockholders under adverse liquidity shocks and expected blockholder losses. These models highlight the existing tradeoff between voice and exit models. Dasgupta and Piascentino (2015) developed a moral hazard model built on the
RI PT
complementary hypothesis between blockholder intervention and trading suggested by Hirschman (1970). Their model is based on the fact that many blockholders are asset managers—i.e., mutual funds, hedge funds, pension funds—and therefore have different
SC
preferences in terms of long- and short-term portfolios. In their model, they consider active monitoring a feasible strategy to be used by asset managers when they are constrained by
M AN U
investors’ cash flows. Their model predicts that active voice is an equilibrium strategy when the threat of exit is credible and therefore the firm management’s best response is to exert effort to improve firm performance and better corporate governance. The results of this model provide an analytical explanation to the empirical behavior that mutual funds are less active than hedge
TE D
funds.
On the empirical side, Falkenstein (1996) suggested that investors (in particular mutual funds) prefer to invest in stocks with high visibility and low transaction costs. One of the arguments
EP
used to explain this behavior is that traders will choose stocks with high levels of visibility in the press and that have been listed for a long time. There is a well-grounded theoretical approach
AC C
supporting the association between market liquidity and firm performance. A liquid stock market provides an opportunity for non-shareholders to become shareholders, encourages management compensation, reduces managerial opportunism and promotes investment. The most important result of Fang et al. (2009) investigation of stock liquidity and firm performance is that firms with liquid stocks perform better in the market. The mechanism by which liquidity increases performance involves improved information on market prices and managerial compensation,
12
ACCEPTED MANUSCRIPT Resubmitted version.R2
This is particularly powerful for industries with high levels of business uncertainty. Chung et al. (2010) find that firms with better corporate governance typically display better financial performance. However, causality runs from corporate governance to higher liquidity. This result
RI PT
suggests that firms alleviate trading issues (liquidity) by improving their corporate governance record. The mechanism provided for such a dynamic is the improvement of financial and operational transparency, which decreases information asymmetries between current and outside
SC
investors.
Roosenboom et al. (2014), based on takeover data, show a negative relationship between lower
M AN U
stock liquidity and acquirers’ returns, particularly on those firms with a single blockholder. For the cases of multiple blockholders the correlation turns positive because block trades and exit become feasible. Their empirical results suggest that lower liquidity is related with withdrawal from takeover bids that would trigger negative announcements, higher CEO turnover, and lower
TE D
premium. Edmans et al. (2013) also address the issue of liquidity and governance finding a positive effect of stock liquidity on blockholder governance. Liquidity and block formation are linked, since the latter is favored by high liquidity, while the former increases the likelihood of
EP
blockholders governing through voice and exit. Hence, firm performance increases3. Latin American business environment, the issue of liquidity is of particular interest because
AC C
stock turnover is structurally low. For instance, the stock turnover median of all companies that belong to the S&P Latin America index was less than 20 basis points between 2000 and 2012. Representative companies of S&P 500 in the US, for the same time span, have a stock turnover
3
For more details, see Edmans (2014) for a comprehensive critical review on voice and exit models from theoretical and empirical standpoints.
13
ACCEPTED MANUSCRIPT Resubmitted version.R2
of above 150 basis points.4 The above theoretical elements and the regional financial markets structure statistics in the region lead us to state the following hypothesis:
RI PT
H4. Multiple blockholder contestability and voice intervention becomes effective in firms with high liquid stocks. Hence, this hypothesis implies that blockholders’ voice and exit disciplinary mechanisms are complements rather than substitutes; especially in markets such as the Latin American one where
SC
the presence of multiple blockholders is a common structure and market liquidity is structurally low.
3.1. Sources and data characteristics
M AN U
3. Data and method
This study uses an unbalanced panel of more than 550 non-financial firms and 6,785 firm-year data points from six Latin American countries (Argentina, Brazil, Chile, Colombia, Mexico and
TE D
Peru) from 1997 to 2011 (15 years). The main source of the dataset was Thomson’s Datastream platform of financial information; Worldscope platform was used to gather information on shareholding records. Other primary sources comprised databases of financial regulators in each
EP
country sampled, as well as the annual reports of each firm. In particular, the shareholder signatures for Chile, Colombia Peru and Brazil were obtained from the Economatica database
AC C
and the respective local regulatory agencies.5 Ownership information for Mexico and Argentina was also compiled using: i) U.S. Securities and Exchange Commission form 20F for cross-listed companies and, ii) annual reports compiled by the Mexican stock exchange. Table 1 (Panel A) summarizes the construction of the sample and its representativeness. The data extraction began with 4,809 firms. After removing non-active firms and those for which no 4
These estimates are based on data collected from Thomson’s Datastream. Superintendencia de Valores y Seguros (Chile), Superintendencia del Mercado de Valores (Perú), Comissao de Valores Mobiliarios (Brazil), Superintendencia Financiera (Colombia). Comisión Nacional Bancaria y de Valores (México), and Comisión Nacional de Valores (Argentina).
5
14
ACCEPTED MANUSCRIPT Resubmitted version.R2
financial information or equity instruments were available, the number of firms fell to 558. The sample is representative because the selected firms imply more than 40% of total corporate
… Table 1 here…
RI PT
market capitalization reported by the World Bank for each country (Panel B).
Firms are gathered into three broad ownership brackets. Firms whose largest blockholder owns more than 50% of its stocks are named “Firms with absolute control”, while firms whose largest
SC
blockholder holds fewer than 10% of stocks are termed “Widely held firms”. Those in between, whose largest shareholder owns between 50% and 10% of shares are named “Firms with
M AN U
multiple blockholders”. This breakdown corresponds well to the purpose of the study, reflecting the highly concentrated ownership among first shareholders previously reported in Chong and Lopez-de-Silanes (2005). Table 2 summarizes the sample distribution according to the structure of blockholders. On average, firms with multiple blockholders -that is, whose largest shareholder
TE D
holds more than 10% and less than 50% of equity rights- make up about half the sample. In most of the remaining firms, at least 50% of company equity is controlled by a single shareholder. Around 13 companies, or 3% of the sample, are widely held. Colombia is the only country in
EP
which widely held firms account for more than 5% of the sample. This ownership distribution is similar to that reported by Leaven and Levine (2008) in their study of 1,657 European firms,
AC C
confirming that multiple blockholders without absolute control and high ownership concentration are common across large capital markets. … Table 2 here …
Table 3 shows the distribution within the sample of the first and second largest shareholders by firm ownership structure and blockholder identity. Panel A depicts the number of firm-year observations. Two aspects should be mentioned in relation to these figures. First, local firms –
15
ACCEPTED MANUSCRIPT Resubmitted version.R2
whether limited liability or incorporated are the main source of large shareholders, followed by firms owned by families or by financial institutions. The state is the final source of blockholders. Second, this observation might seem to contradict the findings of previous studies on ownership
RI PT
and control that families are the first source of large and/or controlling shareholders (La Porta et al., 1998; Claessens et al., 2002; Faccio and Lang, 2002). This phenomenon may be explained by cases where families use (country specific) alternative legal structures that are not observable in
SC
the information collected, resulting in an underestimation of the importance of family ownership. For instance, families might constitute a trust fund managed by a financial institution. Whenever
M AN U
this is the case it is the trust fund that is recorded as the shareholder.
Panel B depicts the average ownership of the first and second blockholder by identity and firm ownership structure. There are a few features worth mentioning. First, ownership participation of second blockholders is remarkably similar regardless of the nature of the blockholders in
TE D
question and of the firm’s ownership structure. For instance, for the sample of firms with multiple blockholders the largest shareholder has around 29% of equity rights regardless of whether it belongs to a family, a local company, a foreign firm or a financial institution. The
EP
same is the case for the second largest shareholder, whose equity rights range from 14.5% to 18.5%. Second, in firms with multiple blockholders there is scope for the second voting block to
AC C
contest the largest shareholder or to form a coalition in order to obtain absolute control. This prediction is independent of whether the second blockholder is a family, foreign or a financial institution. The ownership sum of both shareholder blocks is close to 50% in all cases. Eventually, all second blockholders will become pivotal in the formation of a coalition. … Table 3 here …
16
ACCEPTED MANUSCRIPT Resubmitted version.R2
The ownership structure of the firms with multiple blockholders is of particular interest. In any attempt to characterize the ability of shareholders to form coalitions in cases where there is no controlling shareholder, ownership distribution is highly relevant. Table 4 shows how, on
RI PT
average, the first two shareholders account for 45% of ownership, and 54% if they engage with the third blockholder. This holds true for all firms and for all types of blockholder (family, local,
shareholder wishes to engage in a coalition attempt. … Table 4 here…
SC
foreign, financial institution or state) regardless of the blockholder with which the largest
M AN U
3.2 Blockholder variables: presence, contestability, and dispersion
This study defines a blockholder as an investor with equity rights (direct votes) representing less than 50% and more than 10% of ownership in a listed company. Thus, blockholder presence is a dummy variable for firms fulfilling this shareholder characteristic. Three measures for
TE D
blockholder contestability are explored: i) the Shapley value of the top three shareholders; ii) the ratio of the second shareholder to the first and iii) the ratio of the second and third shareholders to the first. The Shapley value, as a solution to a cooperative game, measures the probability for
EP
the largest shareholder to form a coalition with either of the next two largest shareholders.6 By construction, the Shapley value is one for all firms where the largest shareholder has more than
AC C
50% of the firm’s equity rights, because it implies absolute shareholder control. The second measure is the contestability index 2nd to 1st shareholder defined as the ratio of the shares of the second blockholder to the shares of the largest one, that is: 1 =
6
ℎ 2 ℎ ℎ (1) ℎ 1 ℎ ℎ
We follow Leech’s (2002) construction of oceanic finite games to compute power indices.
17
ACCEPTED MANUSCRIPT Resubmitted version.R2
The index approaches one when the direct votes of the top two blockholders are similar. The closer to one means the second blockholder can contest the power of the first, therefore the optimal strategy is to engage in a coalition. The farthest from one, means that the first
RI PT
shareholder can advance in decision making without the second largest. This indicator grows in relevance for firms without absolute control and for multiple blockholders where the largest block is able to engage in a coalition with the second shareholder, achieving a controlling
SC
position. The third measure, contestability index 2nd and 3rd to 1st shareholder, proxies the
shareholder. It is defined as: 2 =
M AN U
ability of the second plus the third largest shareholders to challenge and monitor the largest
ℎ 2 + 3 ℎ ℎ (2) ℎ 1 ℎ ℎ
Dispersion estimates are defined using two Herfindahl-Hirschman Index (HHI) type measures. First, the standard HHI is constructed using the first to fourth shareholder ownership:
(3)
TE D
=
!
where is the share of ith shareholder´s shares.
EP
A higher HHI concentration indicates the presence of a large shareholder, and his inherited control; this is an obvious result for the segment of firms with a controlling shareholder, about
AC C
50% of the sample described above. For the 45% of the sample where non-controlling large shareholders have been identified the HHI concentration will always be lower than for firms with controlling shareholders.
The second dispersion estimate highlights the difference between subsequent shareholders in order to measure their dispersion: ""# = 18
( − % )
!
ACCEPTED MANUSCRIPT Resubmitted version.R2
where is the share of ith shareholder´s shares, and % other shareholder different from i. The HHI difference is a measure of shareholder distribution. The higher the index is the larger is the distance (in shares) between shareholders. In conceptual terms the HHI differences reflect the
RI PT
inequality in ownership. 3.3 Firm performance and control variables
In line with existing empirical research on firm value and multiple blockholder ownership
SC
(Claessens et al., 2002; Faccio and Lang, 2002; Mishra, 2011; Attig et al., 2009), several variables were considered to measure firm performance and financial characteristics. We relied
M AN U
in this study on Tobin’s Q as the primary proxy for firm performance, which follows the definition proposed by Black et al. (2006) who define the variable as the ratio between market value and the book value of assets. Market value is the sum of the market value of common stock and preferred stocks (if any), plus the book value of long-term liability and minority interest.7
TE D
Several financial variables are included in the econometric estimation, with the purpose of capturing the behavior and characteristics of firms appropriately, and particularly those related to blockholder structure. The natural logarithm of assets (deflated using the Consumer Price Index
EP
(CPI)) is used as a proxy of firm size. It is expected that a negative coefficient will be observed in the estimation results and that large and long-established firms will demonstrate slower market
AC C
dynamics than small and young companies. This variable can also capture a moral hazard effect easily observable in small firms, which are harder for external investors to monitor and where, in the absence of a watchdog, reckless management can divert resources. Wedge is measured as the ratio between Shapley value and the cash flow rights of the largest shareholder − ! −. Following Guedes and Loureiro (2006) this measure is used as an approach 7
We did alternative estimations using ROA instead of firms Tobin´s Q as performance variable. The econometric results are nor qualitatively different from the reported ones in this section. We used two definitions of ROA: i) EBITDA to book value of assets and ii) Operating Income to book value of assets.
19
ACCEPTED MANUSCRIPT Resubmitted version.R2
to identify shareholders voting rights in cases where data on direct ownership is available. Sales growth is measured as the annual growth rate of operating income. This variable is appropriate for measuring investment opportunities. Firms with better growth opportunities are expected to
RI PT
grow faster. Consequently, a positive relationship is expected between this variable and the value and performance measures.
Leverage is measured as the ratio of the book value of total liabilities to total assets. A high
SC
leverage indicates a scenario of financial stress affecting the firm, and challenges to its managerial decisions. A low value indicates financial leeway. Existing empirical evidence
M AN U
suggests that leverage may have either a positive disciplinary effect on the free use of cash flow by management, or a negative one if it increases the probability of bankruptcy or of a firm’s aggregate financial risk.
Stock beta is the standard measure of systemic risk for a firm’s stock with respect to the
TE D
market; it measures shares that have been actively traded in the stock market for more than 180 days in a given year. The Free Cash Flow (FCF) to equity ratio is defined as EBITDA minus expenditure on tax and interest to total firm equity. This is an indicator of a firm’s short term
EP
liquidity. Tangibility is measured as the ratio between plant, property and equipment, and total assets. Lower asset tangibility signals that a firm’s cash flow is presumed to be generated by
AC C
intangibles such as know-how or branding, which implies high firm market value. The expected relationship with the dependent variable is negative. Finally, country macroeconomic and institutional variables are included in the empirical baseline equation. Latin American countries share the same legal origin (French), and are therefore classified within the empirical literature of corporate governance as having similar investor protection standards. The institutional variables mainly deal with a country’s existing
20
ACCEPTED MANUSCRIPT Resubmitted version.R2
ability to undertake business initiatives, which ultimately have an effect on firms’ market value. Most of the variables fulfill the purpose of representing the financial market power to promote private enterprise. The included macro and country-institutional variables are the Emerging
RI PT
Markets Bond Index (EMBI), Legal rights, business freedom, financial freedom, per-capita GDP growth, inflation rate, and the corporate market capitalization to GDP ratio.8
Table 5 displays the summary of statistics of the above discussed firm performance and control
SC
variables that are included in the econometric baseline estimating equations.
4. Econometric analysis
M AN U
….Insert Table 5 here …
We estimate three baseline regression specifications in order to identify the predicted associations of blockholder interactions on firm market value. The three types of equations are consistent with the previous treatment of blockholder presence, contestability and dispersion. In
TE D
all three, the effect of liquidity and of the identity of the second blockholder is included as independent variables and interactions (level and slope). The equation to estimate the
EP
blockholder presence effect is:
Tobin Qit = β 0 + β1 Pr esenceit + β 2 Liquidityit + β3 Identityl + β 4 Pr esenceit × Liquidityit + β5 Pr esenceit × Identityl + β 6 Liqudityit × Identityl
AC C
+ β 7 Pr esenceit × Liquidity × Identity + β5 X1it + β 6 X 2 j + ε it
(5)
The equation used to estimate the contestability effect is:
8
Definitions of all variables included in the empirical model are summarized in the appendix (Table A.1).
21
ACCEPTED MANUSCRIPT Resubmitted version.R2
Tobin Qit = β 0 + β1Contestabilityit + β 2 Liquidityit + β 3 Identityl + β 4Contestabilityit × Liquidityit + β5Contestabilityit × Identityl + β 6 Liquidityit × Identityl + β 7Contestabilityit × Liquidity × Identity + β5 X1it + β 6 X 2 j + ε it
(6)
Tobin Qit = β 0 + β1Contestabilityit + β 2 Liquidityit + β 3 Identityl
RI PT
The equation used to estimate the dispersion effect is:
+ β 4Contestabilityit × Liquidityit + β5Contestabilityit × Identityl + β 6 Liquidityit × Identityl + β 7Contestabilityit × Liquidity × Identity + β5 X1it + β 6 X 2 j + ε it
(7)
SC
where: &' , &( , &) &* are a vector of coefficients of length − 1 = 4 corresponding to the
− 5 identity values: family, local, foreign, financial institution, and state (family is used as the
M AN U
base group); and the interaction with Liquidity and the variables of interest. The vector - . is the set of firm-specific control variables and - / is the set of country institutional and macro control variables. Subscript i indexes firms while subscript t indexes time.
Stock liquidity is a dummy variable for those firms whose stock turnover belongs in the top
TE D
75th percentile. The set of firms belonging to the top quarter most traded firms in each country are also included in Latin American stock indices such as S&P Latin America 40, which comprises 70% of market capitalization of the region from Brazil, Chile, Colombia, Mexico and
EP
Peru. Therefore, if we use the 75th percentile as the cutoff point, we acknowledge that we are examining highly traded stocks in the region, which are also of international relevance. In
AC C
addition, we used alternative near cutoff points such as the 80th or 70th percentiles and the econometric results remained the same.9 Identity is a dummy variable for the identity of the second largest blockholder. The coefficients of interest are those associated to presence, contestability and dispersion, which shows the association between the different blockholder interests and firm market value. Furthermore, what
9
Thus, the usage of the top quartile is a standard measure for academics and financial analysts alike.
22
ACCEPTED MANUSCRIPT Resubmitted version.R2
matters here is the marginal effect associated with the corresponding interaction whenever they are “active”, in the case of liquidity: &! , & , &( , &* whenever Liquidity and Identity are not zero. The interactions are included in the equation in such a way that they modify the constant and
RI PT
slope in order to fully capture the effect of the firm’s liquidity and of the identity of the shareholder.
Given the longitudinal structure of the data, the econometric estimation falls within the fixed
SC
effects transformation for panel data structure, since the Hausman specification test rejected the null of Random Effects (RE) specification as the true model, and because of the need to control
M AN U
for unobserved fixed variables that are inherent in the estimating equation.
4.1 Econometric results: identification, liquidity and blockholder contestability marginal effects.
This section reports the main findings of this study. The first point worth to highlight is the
TE D
ability of the set of interactions gathering high liquid stocks and blockholder identity in the baseline equations become pivotal in the findings associated to the hypotheses of this study. The econometric results are the result of the identification of a simple equation where Tobin’s Q is
EP
regressed against the variable of interest, adding the firm and country control variables discussed above in batches. This process has proven useful when coefficients produce meaningful results
AC C
following the inclusion of control variables and when the estimation method is improved by using within−fixed effects (FE) rather than Ordinary Least Squares (OLS). This strategy helps overcome the two basic estimator biases suspected to be present in the econometric results: unobservable and omitted variables. Table A.2 in the appendix displays an example of the identification of the empirical baseline regression looking at the marginal effect of the Shapley value on firm’s Tobin’s Q. The reported marginal effects of the Shapley value display the
23
ACCEPTED MANUSCRIPT Resubmitted version.R2
expected sign (negative) and become statistically significant when the full set of controls and interactions are included. In addition, we observe that overall regression goodness of fit increases along with the specifications. Hence, adding the suggested control variables minimizes the
RI PT
omitted variable bias in our estimations.
Table 6a summarizes the marginal effects of blockholder presence, contestability and dispersion on firm Tobin’s Q without any of the interacting terms. These coefficients are related
SC
to the voice model. Table 6b displays the average marginal effects taking into account the interactions. These estimates capture the mechanism of how the blockholder variables respond
M AN U
when exit is observed through high liquidity. The full regression equations are displayed in the Appendix (Table A.3).
Several results are worth highlighting. First, across all estimates we observe that marginal effects for blockholder presence, contestability and dispersion on firm value become
TE D
economically and statistically significant when regressions include the liquidity interaction terms. This result validates the argument that voice and exit are complementary governance mechanisms within firms with multiple blockholders listed in low liquid markets (H4).
EP
Second, blockholder presence implies a premium on firm value of 0.045 units on Tobin´s Q [Eq.1 Table 6a]. This premium is three times larger when market liquidity is considered, rising to
AC C
0.135 units, and to 0.17 when the second blockholder identity interaction is included [Eq1. and Eq2 Table 6b]. Regression coefficients become significant at the 5% level. The positive effect suggests that the mere presence of blockholders increases value to shareholders. In value terms, a $10 million book asset firm and Tobin ratio close to 1 will be valued at $10.45m if there are blockholders with respect to absolute control firms. Now, if stock turnover is high (i.e., the firm belongs to the top quartile), this implies a value of $11.35m.
24
ACCEPTED MANUSCRIPT Resubmitted version.R2
The estimated size effect is greater than reported by Laeven and Levine (2008) for European firms, whose results show that there is a premium on Tobin’s Q of 0.037 units of the control rights of the largest shareholder for firms that only have multiple blockholders. The above results
RI PT
validate H1, which states that blockholder presence is value enhancing.
Third, the Shapley value marginal effects on firm value are all negative as expected. That is, as long as the probability to form a controlling coalition between the largest block with the
SC
subsequent blockholders increases, the firm value decreases. Marginal effect ranges from −0.14 to −0.34 when regressions include liquidity and second blockholder interactions. This finding is
M AN U
particularly important because if the probability of forming a controlling coalition raises 10 points (i.e. from 0.4 to 0.5) Tobin’s Q decreases by 0.034 units. In value terms, if a firm shows a Tobin’s Q ratio of 1 and corporate capitalization is 10 million dollars, rent diversion of a potential coalition will reduces asset value by 340k dollars. This result is in accordance with
TE D
Maury and Pajuste (2005) who report a −0.20 regression coefficient and by Gutierrez and Pombo (2009) who regression coefficients of −0.30 and −0.58 for medium and high liquid stocks for 140 Colombian firms. These results provide powerful validation for H2, which proposed a negative
AC C
coalition.
EP
effect on firm value due to potential tunneling and rent diversion by the blockholders’ controlling
Given that the Contestability Indices are inversely related to the Shapley value, the former display a positive association with firm value. They also capture the potential control exerted either by the second or the next two blockholders on the largest one. The marginal effect ranges from 0.11 to 0.13 for the voice model (Col.3 and Col.4, Table 6a). However, this effect becomes larger when regressions are controlled by stock liquidity and by the identity of the second blockholder. The marginal effects lie between 0.13 and 0.19 (Col.5 to Col.8, Table 6b). For the
25
ACCEPTED MANUSCRIPT Resubmitted version.R2
latter case, a change in one standard deviation in the contestability index (0.32) the Tobin’s Q will rise by 0.061 units. Fourth, regarding dispersion measures, the regression coefficients are displayed in the last two
RI PT
(four) columns in table 6a (6b). The HHI captures blockholders’ cash flow rights dispersion. The regression coefficients are negative related (as expected) with firm Tobin’s Q indicating that as dispersion increases there is more room for tunneling by the controlling shareholders. The
SC
marginal effect ranges from 0.39 and 0.49 and becomes statistically significant for the overall regressions with the interacting terms (Col.9 to Col.12, table 6b) suggesting that a change in one
M AN U
standard deviation on blockholders dispersion (0.26) will decrease Tobin’s Q by an average of 0.11 units. This result reinforces the one derived using the Shapley value. In fact the lower the control power of the second or third blockholder the greater the rent capture by the controlling voting-block.
TE D
The last set of results supports the third and fourth working hypotheses, in the sense that the second blockholder is a key element in improving the internal governance of firms, particularly in emerging markets such as Latin America. The consequence for firms with illiquid, seldom
EP
traded stocks, probably retained with the sole purpose of complying with stock exchange rules, is that the absence of such an internal corporate governance mechanism – i.e., blockholder
AC C
contestability− increases firm agency costs and deteriorates investor protection. …Insert Tables 6a and 6b here…
4.2 Identification results: second blockholder identity marginal effects The results presented in the previous section showed how the second blockholder is important to firm value and becomes a key player exerting direct monitoring on the largest shareholder or possibly resorting to the exit mechanism (voice) in order to discipline firm management for
26
ACCEPTED MANUSCRIPT Resubmitted version.R2
highly liquid stocks. It becomes crucial, therefore, to distinguish the identity of the second blockholder if firm value is to be maximized. This section is devoted to disentangling whether a particular kind of blockholder is more effective than others in enforcing managerial control.
RI PT
As described in the data section, the second blockholder is divided by their identity among family, local firm, foreign investor, financial institution or the state. In the second set of results the ownership effect by type of second blockholder is disaggregated. Figure 1 summarizes the
SC
marginal effect of each variable of interest, disentangled by the identity of the second largest shareholder. We have chosen to present the marginal effects in graphical form instead of using a
M AN U
table mainly for illustrative purposes. The coefficient display in a graph allows us to see the coefficient size, its confidence interval, and it allows us to contrast all of these elements with each of the indicators of blockholders’ presence, contestability, and dispersion measures included in the study. There are a total of 36 marginal effects that are displayed in the appendix
TE D
(Table A.4). For instance, the estimates of the baseline model of the overall marginal effect of the Shapley Value of firm Tobin’s Q is −0.345 (Table 6b) once regressions take into account the full set of interactions of stock liquidity and second blockholder type. Looking at Fig. 1, we can
EP
see that this negative effect is largest when the probability of conforming dominant coalition with the State as second blockholder is −0.75 within a confidence interval which lies between
AC C
−1.5 to −0.05. Hence, in this case, rent diversion is maximized.
….. Insert Figure 1 here…
The principal results may be summarized as follows. First, the positive and significant effect of the type second blockholder on firm value, for high liquid stock firms, is observed when they are financial firms, foreign investors or the state. The extent of the marginal effects of these types of blockholders is greater than that of family and local firms. Second, after separating by
27
ACCEPTED MANUSCRIPT Resubmitted version.R2
identity the findings show a substantial differential effect. The change between the mere fact of being a second large shareholder and it being of a value enhancing kind, such as a financial firm, is not trivial. For the case of blockholder presence, the corresponding marginal effect changes
RI PT
from 0.19 when he/she is family, to 0.41 or to 0.52 whether a financial firm or the state is the second blockholder.
Similar results are observed for the contestability indices. The capacity of the second
SC
blockholder to contest the largest blockholder is only significant for financial firms; the marginal effect of contestability highlights this role, increasing from 0.06 for family to 0.39 for financial
M AN U
firms. Third, controlling coalitions (Shapley value) will harm firm performance considerably. In this case, the marginal effect on Tobin’s Q decreases from −0.24 when the second voting block is a local firm to −0.59 and −0.75 when a financial firm or the state is involved in the coalition. Fourth, the role of the state as second blockholder deserves special attention. Many firms in the
TE D
sample were formerly state owned enterprises −some belonging to the power and energy sector − that had been privatized in the 1990s. In terms of market capitalization and size, privatized companies might be categorized as belonging to the subgroup of high stock turnover. In the
EP
shallow Latin American stock markets these firms have become major new players. The role of the state as second blockholder within other private firms is positive according to our findings.
AC C
However, this set of results should be accepted with care, given the small number of observations. For instance, in the case of Chile the development bank that is a state owned firm is an active shareholder in mining and domiciliary public services companies. The findings presented above highlight the importance of financial and foreign firms as second blockholders, reinforcing the positive effect on firm performance, in contrast to family and local ownership. These results are in line with the findings of the institutional investors and foreign
28
ACCEPTED MANUSCRIPT Resubmitted version.R2
investment literature. In general terms, it is expected that these two investor types will behave independently, providing better governance standards. This outcome supports H3, which predicts that the presence of an independent second blockholder enhances firm value.
RI PT
4.3 Endogeneity check
This section summarizes the endogeneity analysis conducted on the results that have just been discussed. The purpose is to determine whether the results hold to misspecifications tests due to a Empirical literature on
SC
potential endogeneity effect embedded in the estimation results.
corporate finance stresses the potential endogeneity between firm value and ownership structure
M AN U
(Renders and Gaeremynck, 2006; Bascle, 2008; Larcker and Rusticus, 2010; Roberts and Whited, 2013). To address these issues, we relied on instrumental variable estimations for the fixed effects regressions. The main endogeneity issue in the econometric estimation regards simultaneity; that is, large shareholders can pick stocks on high valued firms and, at the same
corporate value.
TE D
time, blockholder intervention and monitoring improves firm governance mechanisms and
Three variables were used as instrument for the blockholder ownership variables: (1) The
EP
blockholder definition −i.e. presence, contestability and dispersion− lagged one period; (2) The free cash flow standard error of the last three periods; and (3) The operating income standard
AC C
error of the last three periods. Each variable was used individually, as well as jointly in the IV estimation, using robust standard errors and serial correlation correction upon the estimates. The choice of lagged variables as an endogenous regressor (0 ) in a linear model = &1 + &! ! +. . +&0 0 + 3 as an instrument (4) is based on the assumption that satisfies the relevance
and the exclusion conditions, that is, #5(4, 0 ) ≠ 0 and #5(4, 8) = 0. Both conditions imply that in a linear regression the instrument z only has a relationship with the dependent variable
29
ACCEPTED MANUSCRIPT Resubmitted version.R2
through the relationship of the endogenous regressor 0 (Roberts and Whited, 2013). The relevance condition on the instrumented equation 0 = 91 + 9! ! +. . . +90%! 0%! + :; 4; + < (8)
RI PT
requires that : ≠ 0; The relevance and exclusion conditions ensure the correct identification of the original linear model. Thus, the lagged variable as an instrument variable is closely related to current value of the potential endogenous regressor and does not necessarily explain the
SC
dependent variable in the second stage estimation. Such requirements are met by the instrumental variables chosen, of which the corresponding statistical tests are reported at the bottom of the IV
M AN U
estimation results.
Following Himmelberg et al. (1999), we use the free cash flow volatility and operating income volatility as instruments. Unobserved firm heterogeneity comes from several sources such as managerial discretion or investor risk aversion. The former deals with intangible assets as a
TE D
source of discretional spending, as it is harder to monitor by shareholders and the blockholder ownership required to monitor rent diversion and managerial perk consumption would need to be greater.10 The second source concerns firms’ operating risk. All other things being equal, the
EP
higher a firm’s cash flow or operating income volatility, the less likely it is that multiple blockholder ownership will be present. These variables have the ability to isolate the potential
AC C
endogeneity between ownership structure and performance thanks to their ability to signal the estate of the company to potential investors. Such volatility informs the investors’ decision to increase his ownership stake or not. At the same time firm market value will have a lower response to such volatility.
10
We do not account for a good proxy that captures firm cash flow related with intangibles.
30
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 7 reports the IV regression results for joint instruments. The regression equations replicate the estimates of blockholder presence, contestability and dispersion with interactions. The focus is on the full estimation including high stock turnover firms and taking into account
RI PT
the identity of the second blockholder. In this case the table displays the full regressions, including the set of firm-specific financial controls, country institutional and macro variables. The principal result of this exploration of endogeneity is that the set of tests used to validate the
SC
IV analysis favor the original estimation and suggest that the IV estimation results should not be taken as definitive. To begin with, there is no statistical evidence of endogeneity. The tests listed
M AN U
at the bottom of the table accept the null hypothesis that the explanatory variables are not endogenous. The weakest case is the one corresponding to presence of blockholders, which rejects the null hypothesis at 10% significance; the remaining estimations reject the case of endogeneity at high levels of statistical significance. From this finding the remaining tests
TE D
comply with the standard procedure when estimating IV regressions. The weak instruments test is also rejected (high values of the statistic). Therefore, the set of instruments become a valid set of variables to uncover an endogenous process within the estimation. The under-identification
EP
test is also rejected, with zero probability on the null hypothesis of under-identification. Finally, the valid instrument test (Hansen J) is accepted, since we have one candidate variable for
AC C
endogeneity and three exogenous instruments (as tested above) … Insert table 7 here….
5. Conclusions
This article examined the relationship between multiple blockholder ownership (largest shareholders between 50% and 10%, as opposed to majority shareholders with more than 50%) and firm value. The results show that via different measures of presence, contestability and
31
ACCEPTED MANUSCRIPT Resubmitted version.R2
dispersion, blockholder presence and contestability are positively associated with market firm value within a representative sample of real sector listed firms across Latin America for the 1997 – 2011 period. The finding results show that voice models are limited if there is no further
RI PT
improvement on stock market liquidity. In this sense, blockholder direct monitoring and contestability among voting blocks, become effective internal corporate governance mechanisms within high traded stocks. That is, voice and trade models are complementary to governance
SC
mechanisms rather than substitutes for them. Further, the study evaluates whether the identity of the second largest blockholder enhances blockholder contestability when Tobin’s Q is already
M AN U
controlled by stock liquidity and finds that firms with financial or foreign second largest blockholders are those where the previous result strongly holds. This is not to say that family, local and state second largest blockholders are not important, but financial and foreign blockholders are those with the strongest effects on firm performance.
TE D
These findings extend previous literature on blockholder contestability considering the full set of interacting terms when firm value regressions control for stock liquidity and the second blockholder identity. The implications of these results from a standpoint of financial
EP
development highlight the role of financial deepening in emerging markets. On the one hand, by increasing market liquidity through greater firm listing and stock float ratios is dependent on
AC C
ownership concentration that might still present a natural restriction for further financial deepening. But on the other hand, the entry of large shareholders such as local or foreign institutional and other financial investors has a significant impact on firm performance by raising corporate governance standards and fostering the development of equity markets. These results coincide with studies on institutional investor heterogeneity and firm value.
Acknowledgements
32
ACCEPTED MANUSCRIPT Resubmitted version.R2
We would like to thank the editor (Carl R Chen) and one anonymous referee for their comments raised during the reviewing process that helped us to improve the paper’s theoretical framework and the empirical design. We want to thank to Luis Gutierrez, for helpful comments
RI PT
to previous versions of this paper and María Camila de la Hoz for her helpful research assistance, the comments by the discussant at the World Finance Conference-2016 at NY, and seminar
SC
participants at the research workshop at Universidad de los Andes School of Management.
References
M AN U
Ali S., Liu B., & Su J.J. (2017). Corporate governance and stock liquidity dimensions: panel evidence from pure order-driven Australian market. International Review of Economics and Finance. Forthcoming. Almazan A., Hartzell, J., & Starks, L. (2005). Active institutional shareholders and cost of monitoring: evidence from executive compensation. Financial Management, 34, 5-34. Attig, N., El Ghoul, S. & Guedhami, O. (2009). Do multiple large shareholders play a corporate governance role? Evidence from East Asia, The Journal of Financial Research, 32, 395-422. Bascle, G. (2008). Controlling for endogeneity with instrumental variables in strategic management research, Strategic Organization: 6(3), 285–327
TE D
Becht, M. (2001). Beneficial ownership in the United States. In F. Barca & M. Becht (Eds.), The control of corporate Europe: Chapter 11, pp. 285–298. Oxford University Press. Bennedsen, M. & D. Wolfenson (2000). The balance of power in closely held corporations. Journal of Financial Economics, 58: 113-139
EP
Bloch F. & Hege U. (2003). Multiple shareholders and control contests. MPRA Paper 42286, University Library of Munich, Germany.
AC C
Claessens S., Djankov S., Fan J. & Lang L. (2002). Disentangling the incentive and entrenchment effects of large shareholders. Journal of Finance, 57: 2741–2771 Chong, A. & Lopez-de-Silanes, F. (Eds.) (2005) Privatization in Latin America, myths and realities, Washington: Stanford University Press - World Bank. Chung K. H., Elder J. & Kim J. C., (2010) Corporate governance and liquidity. Journal of Financial and Quantitative Analysis, 45 (2): 265-291. Cueto, D. (2013). Substitutability and complementarity of corporate governance mechanisms in Latin America. International Review of Economics and Finance, 25: 310-325. Dasgupta A, & Piacentino G. (2015) The Wall Street Walk when blockholders compete for flows. Journal of Finance, 70(6): 2853-2896 De-la-Hoz M.C. & Pombo C. (2016) Institutional investor heterogeneity and firm valuation: Evidence from Latin America. Emerging Markets Review, 26, 197-221
33
ACCEPTED MANUSCRIPT Resubmitted version.R2
Edmans A., Fang V.W., & Zur E. (2013) The effect of liquidity on governance, Review of Financial Studies, 26(6), 1443–82 Edmans A., & Manso, G. (2011) Governance through trading and intervention: A theory of multiple blockholders. Review of Financial Studies, 24(7), 2395–428
RI PT
Edmans, A. (2014) Blockholders and corporate governance, Annual Review of Financial Economics, 6, 23-50 Faccio, M., & Lang, L. (2002)The ultimate ownership of western European corporations. Journal of Financial Economics, 65, 365–395 Falkensein, E.G. (1996) Preferences for stock characteristics as revealed by mutual fund portfolio holdings, Journal of Finance, 51(1): 111–135
M AN U
SC
Fang V.W., Noe T., Tice S. (2009). Stock market liquidity and firm value, Journal of Financial Economics, 94(1), 150–69 Ferreira, M., & Matos, P., (2008) The colors of investors’ money: The role of institutional investors around the world. Journal of Financial Economics, 88, 499–533 Gomes, A., & Novaes, W. (2006) Sharing of control versus monitoring as corporate governance mechanisms, Working paper, May 2006 Guedes J., & Loureiro, G. (2006) Estimating the expropriation of minority shareholders: Results from a new empirical approach, European Review of Finance, 12: 421-448 Gutiérrez, L. H., & Pombo, C. (2009). Corporate ownership and control contestability in emerging markets: The case of Colombia. Journal of Economics and Business, 61, 112-139
TE D
Heritage Foundation (2013). Index of economic freedom, Web site: http://www.heritage.org/index/. Accessed: December, 2014 Hirschman, A.O. (1970). Exit, voice, and loyalty: Responses to decline in firms, organizations, and states. Cambridge, MA: Harvard Univ. Press.
EP
Himmelberg, C., Hubbard, R., & Palia, D. (1999). Understanding the determinants of managerial ownership and the link between ownership and performance, Journal of Financial Economics, 53: 353–384 Holderness, C.G. (2009). The myth of diffuse ownership in the United States. Review of Financial Studies, 22(4), 1377–1408.
AC C
Konijn, S. J, Kräussl, R., & Lucas, A. (2011). Blockholder dispersion and firm value. Journal of Corporate Finance, 17, 1330–1339. Jara-Bertin, M., López, F. & López-de-Foronda O. (2008). The contest to the control in European family firms: How other shareholders affect firm value. Corporate Governance: An International Review, 16: 146-159. La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and Vishny, R.W. (1999). Corporate ownership around the world. Journal of Finance, 59, 471–517. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R. (2002). Investor protection and corporate valuation, Journal of Finance, 57, 1147-1170. Laeven, L., & Levine, R. (2008). Complex ownership structures and corporate valuations, Review of Financial Studies, 21 (2): 579-604
34
ACCEPTED MANUSCRIPT Resubmitted version.R2
Larcker, D. F. & Rusticus, T. O. (2010). On the use of instrumental variables in accounting research, Journal of Accounting and Economics, 49(3), 186–205. Leech, D. (2002). Designing the Voting System for the Council of the European Union, Public Choice, 113(3-4), 437–64.
RI PT
McGee, R. W. (2009). Corporate governance in developing economies. Springer. Maug, E. (1998). Large shareholders as monitors: Is there a trade-off between liquidity and control?, Journal of Finance, 53(1), 65–98. Maury, B., Pajuste A. (2005). Multiple Large Shareholders and Firm Value, Journal of Banking and Finance, 29, 1813-1834.
SC
Mishra, D.R. (2011). Multiple large shareholders and corporate risk taking: Evidence from East Asia, Corporate Governance: An International Review, 19, 507-528.
M AN U
Moshirian, F., Nguyen T. T. and Zhang, B. (2014). Large shareholders and firm value: An international analysis. SSRN Scholarly Paper ID 2407845, Social Science Research Network Nadarajah, S., Ali, S., Liu B., & Haung, A. 2016. Stock liquidity, corporate governance and leverage: new panel evidence. Pacific-Basin Finance Journal. Forthcoming. Park Y.W., Zekiye S., & Moon H.S. (2008). Large outside blockholders as monitors: evidence from partial acquisitions. International Review of Economics and Finance, 17, 529-545. Renders, A. & Gaeremynck, A. (2006). Corporate governance and performance: controlling for sample selection bias and endogeneity. Working Paper 0606, Katholieke Universitiet Leuven
TE D
Roosenboom, P., Schlingemann, F.P., & Vasconcelos, M. (2014). Does stock liquidity affect the incentives to monitor: Evidence from corporate takeovers. Review of Financial Studies, 27(8): 2392–433. Roberts, M. and Whited, T. (2013). Endogeneity in empirical corporate finance. In G. Constantinides, Harris, M. and Stulz, R. (Eds.), Handbook of the economics of finance, Volume 2A, Chapter 7, pp. 493–572. Elsevier.
EP
Sacristán-Navarro, M., Gómez-Ansón, S. & Cabeza-García, L. (2011). Family ownership and control, the presence of other large shareholders, and firm performance: Further evidence, Family Business Review, 24(1), 71-93. Sacristán-Navarro, M., Gómez-Ansón, S. & Cabeza-García, L. (2015). The Company you keep: The effect of other large shareholders in family firms, Corporate Governance: An International Review, 23 (3), 216-233.
AC C
Villalonga, B. & Amit, R. (2010). Family control of firms and industries. Financial Management 39(3), 863–904. World Bank (2008). World development indicators. Web site, http://data.worldbank.org/indicator. Accessed: December, 2014 Williams R., (2012). Using the margins command to estimate and interpret adjusted predictions and marginal effects. The Stata Journal, 12 (2), 308-331. Zwiebel, J. (1995). Block investment and partial benefits of corporate control, Review of Economic Studies, 62: 161-185
35
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 1 - Sample Construction and Representativeness Panel A. Sample Construction. Number of firms in the Sample Total Number of firms (Averages 2003-2011) Average Reported Datastream Mnemonics Removing non-active firms Romoving firms with non-equity instruments Removing banks and financial firms Removing firms with insufficient ownership information Total Sample
Panel B. Sample Representativeness (Averages 2000-2011)
Sample Mean Sample Median
0.300 0.143
SC
0.170 0.062 1.132 0.248 0.115 0.071
0.223 0.118
0.432 0.404
TE D
Argentina Brazil Chile Colombia México Peru
Corporate value - Corporate value sample /GDP sample /Mcap. (World Bank) 0.144 0.705 0.048 0.091 0.778 0.777 0.187 0.482 0.092 0.326 0.091 0.211
M AN U
Total Assets/GDP
RI PT
4809 1552 1154 937 604 562
Source: Own estimates based on Thomson-one and World Bank Development Indicators
Table2 Blockholder ownership structure by country (Firm-year observations) Latin America Obs Firms/1 3,383 226 3,299 220 193 13
EP
Blockholder Structure
AC C
Control 1st. shareholder Multiple blockholders Widely held Total
6,875
Dist. 0.49 0.48 0.03
Argentina Obs 335 114 0
Brazil Obs 853 1,206 57
449
2,116
458
Chile Colombia Mexico Obs Obs Obs 604 184 653 643 289 489 17 66 17 1,264
539
1,159
Peru Obs 754 558 36 1,348
Notes: 1/ average number of firms with ownership information for the 1997-2011 period. Sources: Own estimates based on Thomson's World-Scope , Economatica only ownership info; countries' security regulators; SEC-form 20-F for cross-listed companies, and companies' annual reports when applicable.
36
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 3 First and second blockholder type by firm ownership structure (Firm-year observations)
2nd
Control 1st. shareholder Multiple blockholders Widely held
478 539 95
664 836 40
1,021 1,140 28
RI PT
Panel A - Total Observations Shareholder Firm Ownership Type Family Local Foreign Financial State Control 1st. shareholder 585 1,188 702 605 142 1st Multiple blockholders 696 1,030 1,086 266 79 Widely held 92 47 34 8 0 296 317 16
119 50 1
2nd
Control 1st. shareholder Multiple blockholders Widely held
M AN U
SC
Panel B - Ownership share Shareholder Firm Ownership Type Family Local Foreign Financial State Control 1st. shareholder 0.694 0.697 0.678 0.715 0.721 1st Multiple blockholders 0.274 0.322 0.285 0.317 0.267 Widely held 0.076 0.082 0.083 0.085 0.111 0.164 0.146 0.162 0.060 0.070
0.105 0.154 0.066
0.217 0.147 0.185 0.185 0.075 0.092
TE D
Notes: The table shows the number of firms and ownership belonging to a specific blockholder identity and whether they are first or second largest shareholder, and if they determine the ownership structure among controlling 1st shareholder, multiple blockholders or widely held. For example, 585 firms have as first shareholder a family firm that makes the firm belong to one of controlling shareholder, correspondingly this group of firms has an average ownership of 69%. Sources: Own estimates based on Thomson's Worldscope , Economatica (only ownership info; countries' security regulators; SEC-form 20-F for cross-listed companies, and companies' annual reports when applicable.
All 0.298 0.156 0.086 0.055
All (Max) All (Min) 0.500 0.101 0.500 0.003 0.272 0.000 0.244 0.000
AC C
Shareholder 1 2 3 4
EP
Table 4 Ownership Structure - Blockholder firm All (SD) 0.118 0.091 0.053 0.041
Family 0.279 0.146 0.079 0.055
Local 0.295 0.162 0.090 0.059
Foreign Financial 0.300 0.316 0.154 0.185 0.086 0.102 0.057 0.058
State 0.341 0.185 0.066 0.042
Notes: The table shows the ownership structure from the first to the fourth largest shareholder for the sample and by ownership identity. Source: Authors’ estimation.
37
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 5 Performance and Control variables – Summary of Statistics min
max
mean
0.08 -1.36
7.03 2.27
1.21 0.12
6,682
0.000
1.000
0.494
6,679 6,681 6,681
0.000 0.000 0.000
1.000 1.000 2.000
0.703 0.381 0.581
6,678 6,678
0.013 0.000
1.000 1.000
0.348 0.234
M AN U
6,453 6,679 5,993 6,453 4,892 6,265 6,448
0.72 0.12
0.500
0.341 0.315 0.508
SC
6,916 7,176
sd
RI PT
N
0.256 0.264
-0.258 0.000 -24.32 0.001 0.000 -30.27 0.000
12.650 5.556 23.95 1.437 7.736 31.96 0.967
5.953 1.394 0.02 0.475 0.100 0.096 0.411
1.854 0.331 2.20 0.212 0.399 0.953 0.246
6,682 6,682 4,086 6,682 6,682
64.9 0.055 0.333 0.574 0.333
5,742.0 0.474 0.778 1.000 1.000
482.6 0.199 0.490 0.764 0.716
684.7 0.110 0.154 0.114 0.156
6,682 6,682 6,682
-1.17 9.57 -11.73
31.76 157.00 8.66
6.85 53.12 2.62
4.30 33.29 3.10
AC C
EP
TE D
VARIABLES Firm Performance Tobin's Q ROA Presence Blockholder presence dummy Contestability Shapley Contest. Index 1 Contest. Index 2 Dispersion Herfindal Concentration Herfindal Index Finance Size Wedge Sales (Growth) Leverage Beta Free cash flow Tangibility Institutional Country Finance EMBI Domestic credit (w.r.t U.S.) Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.) Macroeconomic Inflation (cpi) Market cap. / GDP GDP growth (Per-capita)
Notes: The table shows the summary statistics for the explanatory variables of interest: presence, contestability, dispersion; and the remaining control variables: country specific financial, institutional and macroeconomic variables. Domestic credit, legal rights index, business freedom, financial freedom and market capitalization are expressed with respect to (w.r.t) the same variable for the United States. All variable definitions are listed in Appendix – table A.1. Source: Authors’ estimation
38
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 6a – Firm value Fixed Effects Regression - Marginal Effects – Voice model Dependent Variable Tobin' Q
Col.1
Col.2
Col.3
Col.5
0.0455 (0.0362)
Contestability Shapley
0.129* (00552) 0.112** (0.0374)
Dispersion Herfindal Index Concentration
-0.11 (0.0794)
Herfindal Index Differentials 3,282
M AN U
Contest. index 2
Col.6
SC
-0.148 (0.0718)
Contest. index 1
Observations
Col.4
RI PT
VARIABLES Presence Blockholder presence
3,282
AC C
EP
TE D
Significance: *** p < 0.01, ** p < 0.05, * p < 0.1 Source: Authors’ estimation.
3,282
39
3,282
-0.104 (0.0718) 3,282 3,282
Resubmitted version.R2
ACCEPTED MANUSCRIPT
Table 6.b - Firm value Fixed Effects Regression – Average Marginal Effects Voice and Trade models Dependent Variable Tobin' Q
VARIABLES
x P75=1 x P75=1 x x P75=1 x P75=1 x x P75=1 x P75=1 x x P75=1 x P75=1 Type Type Type x Type Col.1 Col.2 Col.3 Col.4 Col.5 Col.6 Col.7 Col.8
x P75=1
x P75=1 x x P75=1 x P75=1 x Type Type Col.9 Col.10 Col.11 Col.12
Presence
Contest. index 1
0.191** 0.166* (0.0792) (0.0859)
0.147** 0.130** (0.0511) (0.0552)
SC
Contest. index 2 Dispersion Herfindal Index Concentration
866
720
866
M AN U
0.385*** 0.493*** (0.12) (0.14)
Herfindal Index Differentials Observations
RI PT
Blockholder presence 0.135** 0.171** (0.0561) (0.0597) Contestability Shapley 0.264** 0.345** (0.0909) (0.0983)
720
866
720
866
720
866
720
0.391*** 0.437*** (0.113) (0.146) 866 720
Notes: Column P75 = 1 reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) variable in a regression where only these two variables are interacted (two way interaction). Column p75=1 x Type reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) corresponding to a regression where ownership type is included in the interaction specification (three way interaction) as in the full base line regression equations 5 to 7.
AC C
EP
TE D
Average marginal effect estimates the predicted value of the estimated equation for each observed value of the variable of interest and the average effect is the average of all predicted values. For binary variables, it obtains the predicted values for the ones and zeroes and estimates the difference. In the two way interaction the average marginal effect can be calculated by examining the raw coefficients presented in table A.3 from which the single marginal effects of blockholder presence is 0.0245 and the single marginal effect of the interaction between blockholder presence and high liquidity is 0.11 (Table A.3 Eq.2). The combined marginal effect of blockholder presence is the derivative of the dependent variable with respect to presence: ∂Q/∂Presence = &! + & = 0.0245 + 0.11 = 0.135 which is the value in Col.1; Similarly, the single marginal effect of the interaction between the Shapley value and high liquidity is -0.148 (Table A3 Eq.5). Thus overall marginal effect is ∂Q/∂SV = &! + & = −0.116 − 0.148 = −0.264; which is the value reported in Col.3. The remainder marginal effects are obtained in the same way. Columns label as x P75=1 x Type report the marginal effect corresponding to the three way interaction between stock liquidity dummy, type of second blockholder and the variable of interest. In this case the average marginal effect implies a nonlinear effect after taking the average of all observed values. Williams (2012) details the procedure undertaken in the statistical analysis software Stata. Heteroscedasticity corrected standard errors in parentheses; Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation.
40
Resubmitted version.R2
ACCEPTED MANUSCRIPT
TE D
M AN U
SC
RI PT
Figure 1 - Firm type, marginal effect.
AC C
EP
Source: Own estimates. Notes: The figure shows the marginal effect of the variables of interest (blockholder presence, contestability and ownership concentration) whenever the high concentration and the identity blockholder are equal to one. These figures allow distinguishing the effect of the blockholder upon performance when identity is studied.
41
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 7 - Instrumental variables regressions full over-identified equations (m = 3) Dependent Variable Tobin' Q
Eq.1
Eq.2
...
...
...
...
...
...
...
...
...
-0.148
...
...
...
...
...
(0.161)
...
...
...
...
...
...
0.0713
...
...
...
...
...
(0.132)
...
...
...
...
...
...
...
...
Herfindal Index Differentials
... ...
...
...
(0.0955)
...
...
...
SC
...
...
...
-0.184
...
...
...
...
(0.211)
...
...
...
...
...
-0.199
...
...
...
...
(0.200)
-0.199***
-0.199***
(0.0560)
(0.0469)
(0.0470)
(0.0471)
(0.0475)
(0.0473)
0.647
0.0263
-0.0590
-0.0596
-0.0806*
-0.0932*
-0.205***
-0.204***
(0.421)
(0.0906)
(0.0523)
(0.0551)
(0.0476)
(0.0488)
1.42E-03
-3.55E-03
-3.47E-03
-3.32E-03
-3.58E-03
-3.57E-03
(5.36E-03) (2.72E-03)
(2.74E-03)
(2.73E-03)
(2.73E-03)
(2.76E-03)
0.235*
0.238**
0.243**
0.243**
TE D
AC C
Lag tangibility
...
-0.201***
0.240**
(0.175)
(0.120)
(0.120)
(0.120)
(0.120)
(0.120)
-0.0441
-0.0252
-0.0232
-0.0237
-0.0284
-0.0291
(0.0307)
(0.0199)
(0.0204)
(0.0204)
(0.0200)
(0.0204)
EP
Lag FCF ratio
...
0.027
-0.216***
0.375**
Beta
RI PT
...
Dispersion Herfindal Index Concentration
Leverage
Eq.6
...
Contest. index 2
Sales (Growth)
Eq.5
1.640*
Contest. index 1
Wedge
Eq.4
(0.914)
Contestability Shapley
Finance Size
Eq.3
M AN U
VARIABLES Presence Blockholder presence
0.105**
0.021
0.018
0.018
0.020
0.020
(0.0441)
(0.0126)
(0.0119)
(0.0116)
(0.0125)
(0.0127)
-0.0441
-0.110
-0.111
-0.114
-0.110
-0.110
(0.147)
(0.108)
(0.109)
(0.109)
(0.108)
(0.108)
Country Finance and Institutional EMBI -5.84e-05* -6.13e-05** -6.35e-05*** -6.27e-05*** -5.86e-05** -6.02e-05** (2.38e-05)
(2.38e-05)
(2.40e-05)
(2.40e-05)
Domestic credit (w.r.t U.S.)
3.059***
2.042***
1.951***
1.961***
2.069***
2.054***
(0.712)
(0.333)
(0.324)
(0.324)
(0.343)
(0.333)
Legal rights (w.r.t U.S.)
0.431**
0.331*
0.319*
0.325*
0.322*
0.310*
(0.207)
(0.173)
(0.173)
(0.173)
(0.175)
(0.174)
Buss. freedom (w.r.t U.S.)
-1.109**
-0.291*
-0.264*
-0.258*
-0.305**
-0.295**
Fin. freedom (w.r.t U.S.)
(3.40e-05) (2.38e-05)
(0.505)
(0.152)
(0.144)
(0.144)
(0.153)
(0.149)
0.0560
-0.165
-0.165
-0.167
-0.147
-0.143
(0.247)
(0.180)
(0.181)
(0.181)
(0.181)
(0.182)
42
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table 7 (Cont.) - Instrumental variables regressions over-identified equations Dependent Variable Tobin' Q
Year dummy (2008) Observations Number of id_firm_n Sector dummy Year dummy Robust SE HAC Interaction R2 centered F-stat
Eq.5
Eq.6
3.66E-03
-4.43E-03
-4.84E-03
-4.70E-03
-4.16E-03
-4.31E-03
(7.03E-03) (3.77E-03)
(3.74E-03)
(3.74E-03)
(3.83E-03)
(3.81E-03)
0.0051*** 0.00451***
0.0044***
0.0044***
0.0046***
0.0045***
(8.78E-04) (6.50E-04)
(6.45E-04)
(6.46E-04)
(6.49E-04)
(6.45E-04)
0.0150*** 0.0159***
0.0159***
0.0158***
0.0158***
0.0159***
(3.83E-03) (2.60E-03)
(2.63E-03)
(2.63E-03)
(2.61E-03)
(2.61E-03)
0.203***
0.192***
0.192***
0.192***
0.194***
0.193***
(0.0383)
(0.0287)
(0.0288)
(0.0288)
(0.0289)
(0.0289)
2,689
2,689
2,689
2,689
2,689
2,689
426
426
426
426
426
426
No
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Slope
Slope
Slope
Slope
Slope
0.132
0.128
0.129
0.132
0.130
9.482
9.375
9.378
9.531
9.481
2.445
0.0364
0.000817
0.214
0.443
1.083
0.118
0.849
0.977
0.644
0.506
0.298
3.339
60.86
65.02
60.22
60.70
39.35
7.117
101.3
108.2
106.6
76.48
60.64
Yes Slope -0.475 6.140
TE D
EP
Underidentification. H_0: Underidentification
Eq.4
Yes
Endogeneity H_0: x_j is not endogenous Vs. x_j is endogenous No Endogeneity p-value Weak inst. H_0: Weak identification
Eq.3
RI PT
GDP growth (pc)
Eq.2
SC
Market cap. / GDP
Eq.1
M AN U
VARIABLES Macroeconomic Inflation (cpi)
0.0285
0
0
0
0
0
2.860
9.221
8.472
8.511
9.462
9.381
Valid inst Hansen J. p-value
0.0908
0.00995
0.0145
0.0142
0.00882
0.00918
AC C
Underidentification. p-value Valid inst Hansen J. H_0: Overidentification
Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 Notes: Estimation results for equations 1 to 6 using instrumental variables to check endogeneity of the dependent variable. HAC stands for heterosckedastic and autocorrelation consistent standard errors. Each regression uses 3 join instruments: i) the lag of blockholder presence, contestability and dispersion definition; ii) operating income volatility; iii) cash flow ratio volatility. Slope interactions refers to presence, contestability and dispersion variables times 75th percentile of stock liquidity dummy. Regressions are controlled by second blockholder type dummies. All variable definitions are listed in Appendix – Table A.1. Source: Authors’ estimation.
43
ACCEPTED MANUSCRIPT Resubmitted version.R2
APPENDIX Table A.1 Definition of Variables Ownership Blockholder Presence
Source
Definition
Variable
Shareholder 2
Equity share of the 2nd largest blockholder
Shareholder 3
Equity share of the 3rd largest blockholder
Shareholder 4
Equity share of the 4th largest blockholder
Wedge
Shapley Value to control rights largest shareholder ratio
ROA
Market value/book value: [stock market capitalization + long term liabilities + book value of prefered stocks (if any) + minority interest]/[book value of assets] The ratio of net profits after tax to total assets
Contestability Shapley Contestability Index 1 Contestability Index 2
Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters
M AN U
Tobin's Q
Anual Reports-Thomson Reuters
Own estimates
SC
Performance
RI PT
Shareholder 1
Dummy variable for firms with blockholder as an investor with Anual Reports-Thomson Reuters equity rights (direct votes) with less than 50% and more than 10% ownership in a listed company Equity share of the largest blockholder Anual Reports-Thomson Reuters
The Shapley Value solution for the largest shareholders in a three-voting game The share of the second to the largest blockholder ratio.
Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters
The share of the second and third blockholders to the largest blockholder ratio. Shapley Value to control rights largest shareholder s1
Anual Reports-Thomson Reuters
H.I Concentration
Herfindahl concentration index
Anual Reports-Thomson Reuters
H.I Differences
The sum of the squares of the differences between the first and the second H.I Con.
Anual Reports-Thomson Reuters
Size
Natural Logarithm of total asssets
Anual Reports-Thomson Reuters
Sales (growth)
Percentage of sales growth compared to previous year.
Anual Reports-Thomson Reuters
Leverage
The ratio of total liabilities to total assets
Anual Reports-Thomson Reuters
Wedge
Standard measure of systemic risk for each firm’s stock with Anual Reports-Thomson Reuters respect to the market. It was measure for stocks which have been active in the stock market for more than 180 days in a given year. Free Cash Flow (FCF) to equity ratio Is defined as EBITDA minus Anual Reports-Thomson Reuters tax expenditure minus interest expenditures to total firm equity. This is an indicator for firm short term liquidity. Total property, plant and equipment divided by total assets Anual Reports-Thomson Reuters
AC C
Stock Beta
EP
Finance
TE D
Dispersion
Free Cash Flow
Tangibility
44
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.1 Definition of Variables (Continued) Source
Definition
Variable Institutional
Domestic Credit (w.r.t.US)
Emerging Markets Bonds Index. Measures financial risk in each Anual Reports-Thomson Reuters country. Expressed in basis points Domestic credit to private sector: refers to financial resources The World Bank — world provided to the private sector, such as through loans, development indicators purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment
RI PT
EMBI
Measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and thus facilitate lending. Business Freedom (w.r.t.US) Is an overall indicator of the efficiency of government regulation of business. The score for each country is a number between 0 and 100, with 100 equaling the freest business environment. The score is based on 10 factors, all weighted equally, using data from the World Bank’s Doing Business study. Financial Freedom (w.r.t.US) Is a measure of banking efficiency as well as a measure of independence from government control and interference in the financial sector. Index from 0 to 10. Macroeconomic
The World Bank — world development indicators
Inflation (CPI)
The World Bank — world development indicators
AC C
EP
GDP Growth (pc)
Inflation as measured by the consumer price index reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services Market capitalization (also known as market value) is the share price times the number of shares outstanding, divided by GDP Annual percentage growth rate of GDP per capita based on constant local currency.
TE D
Market Cap/ GDP
M AN U
SC
Legal Rights (w.r.t.US)
45
Index of economic freedom
Index of economic freedom
The World Bank — World development indicators The World Bank — world development indicators
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.2 Identification - empirical equation Tobin's Q and Shapley Value Fixed Effects Regressions VARIABLES
Coeff
+ Firm + Country + Turnover Marginal + Turnover Marginal specific and Macro interaction Effect and Type x Effect controls controls x P75 = 1 p75 x Type
Contestability Shapley
AC C
EP
TE D
M AN U
SC
RI PT
0.217*** 0.199* -0.148 -0.116 -0.264** -0.0114 -0.345** (0.0804) (0.120) (0.110) (0.112) (0.123) (0.152) (0.135) Firm financial and specific variabless Size 0.0780** -0.217*** -0.218*** -0.206*** (0.0391) (0.0577) (0.0576) (0.0585) Wedge -0.133 0.00756 0.0181 0.0321 (0.0940) (0.0796) (0.0798) (0.0811) Sales (Growth) 0.00535 -0.00150 -0.00150 -0.00369 (0.00417) (0.00340) (0.00339) (0.00287) Leverage 0.245* 0.277** 0.277** 0.211 (0.130) (0.135) (0.135) (0.152) Beta 0.0120 -0.0170 -0.0140 -0.0282 (0.0209) (0.0214) (0.0209) (0.0214) Lag free cash flow ratio 0.0160** 0.00158 0.00173 0.0153 (0.00754) (0.0170) (0.0168) (0.0123) Lag tangibility -0.559*** -0.235** -0.238** -0.122 (0.119) (0.101) (0.101) (0.119) Country institutional variables EMBI -5.68e-05** -5.38e-05** -6.49e-05** (2.23E-05) (2.23E-05) (2.74E-05) Domestic credit (w.r.t U.S.) 2.204*** 2.200*** 2.051*** (0.430) (0.430) (0.422) Legal rights (w.r.t U.S.) 0.407** 0.404** 0.330 (0.187) (0.188) (0.210) Buss. freedom (w.r.t U.S.) -0.293* -0.300* -0.275 (0.176) (0.175) (0.181) Fin. freedom (w.r.t U.S.) -0.165 -0.183 -0.195 (0.213) (0.212) (0.237) Macro economic variables Inflation (cpi) -5.24E-03 -5.10E-03 -5.85E-03 (3.65E-03) (3.66E-03) (3.62E-03) Market cap. / GDP 0.00468*** 0.00472*** 0.00456*** (6.99E-04) (6.99E-04) (7.41E-04) GDP growth (pc) 0.0156*** 0.0156*** 0.0160*** (2.60E-03) (2.59E-03) (2.47E-03) Year dummy (2008) 0.203*** 0.202*** 0.194*** (0.0313) (0.0311) (0.0306) Constant 1.068*** 0.939*** 2.062*** 2.060*** 1.873*** (0.0565) (0.242) (0.346) (0.346) (0.368) Observations 6,305 4,426 3,282 3,280 866 2,699 720 R-squared 0.005 0.031 0.118 0.120 0.137 Sector dummy No No No No No No No Year dummy No No No No No No No Robust SE Yes Yes Yes Yes Yes Yes Yes Level + Level + Level + Level + Level + Level + Level + Interaction slope slope slope slope slope slope slope Number of id_firm_n 548 475 474 474 435
Notes: The table reports the identification process of the Tobin’s Q empirical equation for the case of the Shapley Value. It begins with a simple regression and then regressions add the firm-specific, country controls, and the interacting terms of stock liquidity and second blockholder identity. Columns Marginal effects show the marginal effect of the immediate left column. Robust standard errors in parentheses. Significance: *** p < 0.01, ** p < 0.05, * p<0.1
46
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within) Shapley
Coeff.
Variables
x P75=1 x Type
x P75=1
Eq.1
Eq.2
Eq.3
Contest Index 1
Coeff.
x P75=1 x Type
x P75=1
Eq.4
Eq.5
Eq.6
Coeff.
Contest Index 2 x P75=1 x Type
x P75=1
Eq.7
Eq.8
Eq.9
x P75=1
Eq.10
Eq.11
Coeff.
HI diferrences x P75=1 x Type
x P75=1
Coeff.
x P75=1 x Type
x P75=1
Eq.12
Eq.13
Eq.14
Eq.15
Eq.16
Eq.17
Eq.18
0.0455
0.0245
-0.0144
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.0479)
(0.0482)
(0.0914)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
Contestability
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
Shapley
...
...
...
-0.148
-0.116
-0.0114
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.110)
(0.112)
(0.152)
...
...
...
...
...
...
...
...
...
...
...
...
Contest. index 1
...
...
...
...
...
...
0.129*
0.100
-0.121
...
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.0683)
(0.0732)
(0.125)
...
Contest. index 2
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
SC
Presence
Dispersion
...
...
...
...
...
...
...
H.I. Concentration
...
...
...
...
...
...
...
...
...
...
...
...
...
...
H.I. Differences
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.0992**
-0.0919
...
0.0548
0.266**
...
...
(0.0419)
(0.0969)
...
(0.0603)
(0.132)
...
Interaction stock liquidity dummy
0.110**
0.203*
...
...
...
(0.0558)
(0.123)
...
...
P75 x Shapley
...
...
...
...
...
...
...
p75 x contest_index 1
...
...
...
...
...
...
...
...
P75 x contest_index 2
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
P75 x hi_differences P75 x hi_concentration
...
...
...
...
...
...
...
0.0977*
-0.0464
...
...
...
...
...
...
(0.0502)
(0.0545)
(0.0815)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.110
-0.0546
-0.0827
...
...
...
...
...
...
...
...
(0.120)
(0.124)
(0.238)
...
...
...
...
...
...
...
...
...
...
...
-0.104
-0.0462
-0.0324
...
...
...
...
...
...
...
...
(0.105)
(0.106)
(0.234)
-0.0762
-0.0364
...
-0.0725
-0.0527
...
0.0569
0.173*
...
0.0281
0.124
(0.0488)
(0.115)
...
(0.0463)
(0.106)
...
(0.0474)
(0.102)
...
(0.0376)
(0.0779) ...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.148**
-0.352*
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.0751)
(0.180)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
0.0909
0.184
...
...
...
...
...
...
...
...
... ...
...
...
(0.0764)
(0.174)
...
...
...
...
...
...
...
...
EP
...
...
...
...
...
...
0.0495
0.153
...
...
...
...
...
...
...
...
...
...
...
...
(0.0487)
(0.101)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
-0.344***
-0.506
...
...
...
...
...
...
...
...
...
...
...
...
(0.0992)
(0.311)
...
...
...
...
...
...
...
...
...
...
-0.330***
-0.536*
...
...
...
...
...
...
...
...
...
...
...
...
...
(0.108)
(0.324)
...
...
...
AC C
P75 x blockholder presence
...
0.112**
...
TE D
P75 dummy
...
M AN U
Blockholder presence
Coeff.
HI Concentration x P75=1 x Type
RI PT
Presence
47
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.3 Tobin’s Q regressions (Cont.) Presence Coeff.
Variables
Shapley x P75=1
x P75=1 x Type
Coeff.
Contest Index 1 x P75=1
x P75=1 x Type
Coeff.
Contest Index 2
x P75=1
x P75=1 x Type
Coeff.
Leverage Beta L.fcf_r L.tangibility
Coeff.
x P75=1
x P75=1 x Type
Eq.5
Eq.6
Eq.7
Eq.8
Eq.9
Eq.10
Eq.11
Eq.12
Eq.13
Eq.14
Eq.15
Eq.16
Eq.17
Eq.18
-0.218***
-0.221***
-0.214***
-0.217***
-0.218***
-0.206***
-0.217***
-0.217***
-0.201***
-0.217***
-0.217***
-0.200***
-0.217***
-0.221***
-0.212***
-0.217***
-0.222***
-0.211***
(0.0582)
(0.0581)
(0.0598)
(0.0577)
(0.0576)
(0.0585)
(0.0578)
(0.0579)
(0.0572)
(0.0573)
(0.0573)
(0.0573)
(0.0580)
(0.0580)
(0.0583)
(0.0581)
(0.0579)
(0.0582)
-0.0480
-0.0426
-0.0522
0.00756
0.0181
0.0321
-0.0413
-0.0365
-0.0654
-0.0262
-0.0215
-0.0481
-0.0759
-0.0685
-0.0931
-0.0803
-0.0726
-0.0949*
RI PT
Eq.4
(0.0617)
(0.0617)
(0.0607)
(0.0796)
(0.0798)
(0.0811)
(0.0599)
(0.0600)
(0.0561)
(0.0599)
(0.0599)
(0.0564)
(0.0585)
(0.0580)
(0.0565)
(0.0579)
(0.0574)
(0.0562)
-0.00143
-0.00141
-0.00362
-0.00150
-0.00150
-0.00369
-0.00149
-0.00141
-0.00366
-0.00138
-0.00133
-0.00346
-0.00153
-0.00161
-0.00407
-0.00156
-0.00154
-0.00381
(0.00339)
(0.00338)
(0.00287)
(0.00340)
(0.00339)
(0.00287)
(0.00339)
(0.00339)
(0.00284)
(0.00284)
0.282**
0.285**
0.225
0.277**
0.277**
0.211
0.270**
0.271**
0.196
(0.134)
(0.134)
(0.152)
(0.135)
(0.135)
(0.152)
(0.134)
(0.135)
(0.152)
-0.0164
-0.0139
-0.0291
-0.0170
-0.0140
-0.0282
-0.0139
(0.0214)
(0.0209)
(0.0216)
(0.0214)
(0.0209)
(0.0214)
(0.0216)
0.00179
0.00233
0.0167
0.00158
0.00173
0.0153
0.00164
(0.0172)
(0.0170)
(0.0133)
(0.0170)
(0.0168)
(0.0123)
(0.0169)
-0.235**
-0.237**
-0.124
-0.235**
-0.238**
-0.122
-0.240**
(0.102)
(0.102)
(0.119)
(0.101)
(0.101)
(0.119)
(0.101)
Country Institutional and Financial EMBI
x P75=1
Eq.3
-0.0106
-0.0247
(0.0214)
(0.0219)
SC
Sales (Growth)
Coeff.
Eq.2
(0.00339)
(0.00338)
(0.00284)
(0.00340)
(0.00340)
(0.00285)
(0.00340)
(0.00340)
0.265**
0.267**
0.192
0.277**
0.279**
0.233
0.275**
0.279**
0.225
(0.134)
(0.135)
(0.152)
(0.135)
(0.135)
(0.150)
(0.135)
(0.135)
(0.150)
-0.0142
-0.0108
-0.0255
-0.0168
-0.0169
-0.0275
-0.0166
-0.0165
-0.0280
(0.0215)
(0.0213)
(0.0220)
(0.0214)
(0.0210)
(0.0211)
(0.0213)
(0.0211)
(0.0222)
M AN U
Wedge
x P75=1
HI diferrences x P75=1 x Type
Eq.1
Firm Characteristics Size
HI Concentration x P75=1 x Type
0.00157
0.0165
0.00132
0.00132
0.0157
0.00161
0.00241
0.0169
0.00174
0.00255
0.0170
(0.0168)
(0.0126)
(0.0168)
(0.0167)
(0.0122)
(0.0171)
(0.0165)
(0.0124)
(0.0170)
(0.0164)
(0.0131)
-0.243**
-0.126
-0.240**
-0.243**
-0.125
-0.236**
-0.242**
-0.123
-0.237**
-0.242**
-0.125
(0.101)
(0.118)
(0.101)
(0.101)
(0.118)
(0.102)
(0.101)
(0.119)
(0.101)
(0.100)
(0.118)
-5.70e-05** -5.36e-05** -6.40e-05** -5.68e-05** -5.38e-05** -6.49e-05**-5.77e-05*** -5.55e-05** -6.80e-05**-5.79e-05*** -5.57e-05** -6.73e-05** -5.68e-05** -5.29e-05** -5.92e-05** -5.71e-05** -5.30e-05** -6.16e-05** (2.25e-05) (2.26e-05) (2.77e-05) (2.23e-05) (2.23e-05) (2.74e-05) (2.23e-05) (2.22e-05) (2.77e-05) (2.22e-05) (2.21e-05) (2.77e-05) (2.24e-05) (2.26e-05) (2.85e-05) (2.24e-05) (2.25e-05) (2.82e-05)
Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.)
2.161***
2.037***
2.204***
2.200***
2.051***
2.145***
2.137***
1.910***
2.157***
2.150***
1.942***
2.191***
2.172***
2.059***
2.181***
2.160***
2.031***
(0.427)
(0.426)
(0.419)
(0.430)
(0.430)
(0.422)
(0.423)
(0.423)
(0.406)
(0.422)
(0.423)
(0.409)
(0.434)
(0.433)
(0.422)
(0.431)
(0.430)
(0.417)
0.399**
0.396**
0.308
0.407**
0.404**
0.330
0.404**
0.398**
0.307
0.413**
0.409**
0.316
0.405**
0.387**
0.310
0.406**
0.379**
0.292
(0.188)
(0.188)
(0.208)
(0.187)
(0.188)
(0.210)
(0.188)
(0.188)
(0.209)
(0.187)
(0.187)
(0.208)
(0.188)
(0.189)
(0.213)
(0.187)
(0.189)
(0.213)
-0.272
-0.282*
-0.261
-0.293*
-0.300*
-0.275
-0.266
-0.277
-0.246
-0.272
-0.285*
-0.252
-0.279
-0.287
-0.277
-0.273
-0.283
-0.249
(0.172)
(0.171)
(0.178)
(0.176)
(0.175)
(0.181)
(0.171)
(0.170)
(0.176)
(0.171)
(0.170)
(0.177)
(0.176)
(0.175)
(0.181)
(0.174)
(0.173)
(0.180)
-0.166
-0.173
-0.170
-0.165
-0.183
-0.195
-0.166
-0.177
-0.178
-0.164
-0.176
-0.185
-0.167
-0.176
-0.185
-0.166
-0.167
-0.167
(0.214)
(0.213)
(0.240)
(0.213)
(0.212)
(0.237)
(0.213)
(0.212)
(0.236)
(0.212)
(0.212)
(0.236)
(0.214)
(0.213)
(0.235)
(0.214)
(0.213)
(0.236)
-0.00552
-0.00540
-0.00631*
-0.00524
-0.00510
-0.00585
-0.00573
-0.00557
-0.00661*
-0.00559
-0.00547
-0.00638*
-0.00539
-0.00509
-0.00560
-0.00546
-0.00503
-0.00569
Macroeconomic Inflation (cpi)
TE D
Legal rights (w.r.t U.S.)
2.163***
EP
Domestic credit (w.r.t U.S.)
Market cap. / GDP
AC C
(3.63E-03) (3.63E-03) (3.58E-03) (3.65E-03) (3.66E-03) (3.62E-03) (3.62E-03) (3.65E-03) (3.57E-03) (3.63E-03) (3.64E-03) (3.58E-03) (3.66E-03) (3.66E-03) (3.56E-03) (3.65E-03) (3.65E-03) (3.56E-03) 0.00463*** 0.00468*** 0.00455*** 0.00468*** 0.00472*** 0.00456*** 0.00463*** 0.00465*** 0.00443*** 0.00464*** 0.00466*** 0.00445*** 0.00468*** 0.00473*** 0.00465*** 0.00466*** 0.00470*** 0.00462*** (6.97E-04) (6.98E-04) (7.37E-04) (6.99E-04) (6.99E-04) (7.41E-04) (6.94E-04) (6.95E-04) (7.23E-04) (6.93E-04) (6.94E-04) (7.32E-04) (7.02E-04) (7.03E-04) (7.37E-04) (6.98E-04) (6.99E-04) (7.34E-04) GDP growth (pc)
0.0157*** 0.0156*** 0.0160*** 0.0156*** 0.0156*** 0.0160*** 0.0157*** 0.0156*** 0.0167*** 0.0156*** 0.0156*** 0.0165*** 0.0156*** 0.0157*** 0.0160*** 0.0156*** 0.0157*** 0.0161*** (2.61E-03) (2.59E-03) (2.49E-03) (2.60E-03) (2.59E-03) (2.47E-03) (2.60E-03) (2.59E-03) (2.48E-03) (2.60E-03) (2.59E-03) (2.47E-03) (2.60E-03) (2.59E-03) (2.48E-03) (2.61E-03) (2.59E-03) (2.46E-03)
Year dummy (2008) Constant
0.202***
0.202***
0.193***
0.203***
0.202***
0.194***
0.201***
0.201***
0.193***
0.202***
0.202***
0.194***
0.203***
0.204***
0.197***
0.202***
0.203***
(0.0313)
(0.0311)
(0.0306)
(0.0313)
(0.0311)
(0.0306)
(0.0314)
(0.0313)
(0.0305)
(0.0313)
(0.0312)
(0.0305)
(0.0313)
(0.0312)
(0.0309)
(0.0313)
(0.0312)
(0.0308)
2.018***
2.053***
2.026***
2.062***
2.060***
1.873***
1.984***
2.015***
2.047***
1.940***
1.972***
1.993***
2.108***
2.133***
2.093***
2.103***
2.130***
2.045***
48
0.196***
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.3 Tobin’s Q regressions (Cont.)
Coeff.
Variables
Shapley x P75=1
Eq.1
Eq.2
x P75=1 x Type
Eq.3
Coeff.
Contest Index 1 x P75=1
Eq.4
Eq.5
x P75=1 x Type
Eq.6
Coeff.
Contest Index 2
x P75=1
Eq.7
x P75=1 x Type
Eq.8
Eq.9
Regression Statistics
Coeff.
HI Concentration
x P75=1
Eq.10
Eq.11
x P75=1 x Type
Eq.12
Coeff.
RI PT
Presence
HI diferrences
x P75=1
Eq.13
Eq.14
x P75=1 x Type
Eq.15
Coeff.
x P75=1
Eq.16
Eq.17
x P75=1 x Type
Eq.18
Observations
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
3,282
3,280
2,699
R-squared
0.117
0.119
0.135
0.118
0.120
0.137
0.118
0.119
0.136
0.119
0.120
0.137
0.117
0.121
0.139
0.117
0.121
0.135
474
435
474
474
435
474
474
435
474
474
435
474
474
435
474
474
435
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Year dummy
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
0.117
0.119
0.135
0.118
0.120
0.137
0.118
0.119
0.136
-27.30
6.662
72.51
1.112
3.155
68.15
15.26
Robust SE Interaction R2 within Hausman test H0: RE Vs. H1: FE (stat)
SC
474
Sector dummy
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
Yes Level + slope
0.119
0.120
0.137
0.117
0.121
0.139
0.117
0.121
0.135
28.14
104.8
-28.89
-0.879
138.0
-10.50
20.59
136.4
M AN U
Number of id_firm_n
16.30
102.0
29.69
Notes: This table the Tobin´s Q regression looking at the variables of blockholder control: presence, contestability and dispersion. The regression coefficient of each variable of interest are reported in Eq.1, Eq.4, Eq.7, Eq.10, Eq.13 and Eq.16; Columns label by xP75=1 report the interaction with the stock liquidity dummy at 75Th percentile. Columns label as x P75=1 x report the marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) and second blockholder ownership type (three way interaction). All variable definitions are listed in Appendix – table A.1.
AC C
EP
TE D
Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation
49
ACCEPTED MANUSCRIPT Resubmitted version.R2
Table A.4 Firm values and second blockholder type marginal effects P75=1 x Type Col.1
Family
Local
Foreign Financial State
Col.2
Col.3
0.189
0.0815
Col.4
Col.5
P75=1 x Family Local Foreign Financial Type Col.6 Col.7 Col.8 Col.9 Col.10 Col.11
Presence Blockholder presence
0.171** (0.0721)
0.139 0.410** 0.527**
...
...
...
...
(0.124) (0.0927) (0.0928) (0.164) (0.215)
...
...
...
...
Contestability
Contest. index 2
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
... -0.345** -0.363 -0.247 -0.310** -0.599** -0.751**
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
0.130**
0.106
0.0658
0.132* 0.284**
0.075
...
...
...
(0.103) (0.0785) (0.112) (0.206)
...
...
...
(0.135) (0.222) (0.156)
...
...
...
...
...
...
0.166* 0.0631
0.114
0.180 0.390* -0.120
(0.0899) (0.165) (0.133)
(0.124) (0.202) (0.309)
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
... -0.493*** -0.619* -0.276 -0.476**
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
720
116
168
340
86
10
720
-0.593 -2.395***
(0.214)
(0.378)
(0.791)
...
...
...
...
...
...
...
...
...
116
168
340
86
10
EP
Notes: Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation
AC C
... ...
(0.330)
...
…… End of Document ……
50
... ...
(0.263)
...
(0.171) (0.324) (0.227)
... ...
(0.149)
...
TE D
Observations
...
...
(0.0596) (0.0974)
H.I. Differentials
...
...
Dispersion H.I. Concentration
...
SC
Contest. index 1
P75=1 x Family Local Foreign Financial State Type Col.12 Col.13 Col.14 Col.15 Col.16 Col.17 Col.18
...
M AN U
Shapley
State
RI PT
VARIABLES
...
...
...
...
...
...
...
...
...
...
...
...
... -0.437*** -0.538* -0.143 -0.442** -0.727* -1.553 (0.159) (0.308) (0.210) 720
116
168
(0.205) (0.396) (1.241) 340
86
10
ACCEPTED MANUSCRIPT 1
Table 1 - Sample Construction and Representativeness Panel A. Sample Construction. Number of firms in the Sample Total Number of firms (Averages 2003-2011) Average Reported Data-stream Mnemonics Removing non-active firms Removing firms with non-equity instruments Removing banks and financial firms Removing firms with insufficient ownership information Total Sample
Sample Mean Sample Median
Corporate value sample /Mcap. (World Bank) 0.705 0.091 0.777 0.482 0.326 0.211
0.170 0.062 1.132 0.248 0.115 0.071
0.144 0.048 0.778 0.187 0.092 0.091
0.300
0.223
0.432
0.143
0.118
0.404
M AN U
Corporate value - sample /GDP
TE D
Argentina Brazil Chile Colombia México Peru
Total Assets/GDP
SC
Panel B. Sample Representativeness (Averages 2000-2011)
RI PT
4809 1552 1154 937 604 562
EP
Source: Own estimates based on Thomson-one and World Bank Development Indicators
AC C
Table2 Blockholder ownership structure by country (Firm-year observations) Blockholder Structure
Latin America Obs Firms/1 Dist.
Control 1st. shareholder Multiple blockholders Widely held
3,383 3,299 193
226 220 13
Total
6,875
458
0.49 0.48 0.03
Argentina Obs
Brazil Obs
Chile Obs
Colombia Obs
Mexico Obs
Peru Obs
335 114 0
853 1,206 57
604 643 17
184 289 66
653 489 17
754 558 36
449
2,116
1,264
539
1,159
1,348
Notes: 1/ average number of firms with ownership information for the 1997-2011 period. Sources: Own estimates based on Thomson's World-Scope , Economatica only ownership info; countries' security regulators; SECform 20-F for cross-listed companies, and companies' annual reports when applicable
ACCEPTED MANUSCRIPT 2
Table 3 First and second blockholder type by firm ownership structure (Firm-year observations)
Panel B - Ownership share Shareholder Firm Ownership Type Control 1st. shareholder 1st Multiple blockholders Widely held Control 1st. shareholder Multiple blockholders Widely held
664 836 40
1,021 1,140 28
296 317 16
119 50 1
Family Local Foreign Financial State 0.694 0.697 0.678 0.715 0.721 0.274 0.322 0.285 0.317 0.267 0.076 0.082 0.083 0.085 … 0.111 0.164 0.146 0.162 0.060 0.070
0.105 0.154 0.066
0.217 0.147 0.185 0.185 0.075 0.092
TE D
2nd
478 539 95
RI PT
Control 1st. shareholder Multiple blockholders Widely held
State 142 79 0
SC
2nd
Family Local Foreign Financial 585 1,188 702 605 696 1,030 1,086 266 92 47 34 8
M AN U
Panel A - Total Observations Shareholder Firm Ownership Type Control 1st. shareholder 1st Multiple blockholders Widely held
EP
Notes: The table shows the number of firms and ownership belonging to a specific blockholder identity and whether they are first or second largest shareholder, and if they determine the ownership structure among controlling 1st shareholder, multiple blockholders or widely held. For example, 585 firms have as first shareholder a family firm that makes the firm belong to one of controlling shareholder, correspondingly this group of firms has an average ownership of 69%.
AC C
Sources: Own estimates based on Thomson's Worldscope , Economatica (only ownership info; countries' security regulators; SECform 20-F for cross-listed companies, and companies' annual reports when applicable.
ACCEPTED MANUSCRIPT 3
Table 4 Ownership Structure - Blockholder firm All 0.298 0.156 0.086 0.055
All (Max) 0.500 0.500 0.272 0.244
All (Min) 0.101 0.003 0.000 0.000
All (SD) 0.118 0.091 0.053 0.041
Family 0.279 0.146 0.079 0.055
Local 0.295 0.162 0.090 0.059
Foreign 0.300 0.154 0.086 0.057
Financial 0.316 0.185 0.102 0.058
State 0.341 0.185 0.066 0.042
RI PT
Shareholder 1 2 3 4
AC C
EP
TE D
M AN U
SC
Notes: The table shows the ownership structure from the first to the fourth largest shareholder for the sample and by ownership identity. Source: Authors’ estimation.
ACCEPTED MANUSCRIPT 4
Table 5 Performance and Control variables – Summary of Statistics N
min
max
mean
sd
0.08 -1.36
7.03 2.27
1.21 0.12
0.72 0.12
6,682
0.000
1.000
0.494
0.500
6,679 6,681 6,681
0.000 0.000 0.000
1.000 1.000 2.000
0.703 0.381 0.581
0.341 0.315 0.508
6,678 6,678
0.013 0.000
1.000 1.000
0.348 0.234
6,453 6,679 5,993 6,453 4,892 6,265 6,448
-0.258 0.000 -24.32 0.001 0.000 -30.27 0.000
12.650 5.556 23.95 1.437 7.736 31.96 0.967
5.953 1.394 0.02 0.475 0.100 0.096 0.411
1.854 0.331 2.20 0.212 0.399 0.953 0.246
6,682 6,682 4,086 6,682 6,682
64.9 0.055 0.333 0.574 0.333
5,742.0 0.474 0.778 1.000 1.000
482.6 0.199 0.490 0.764 0.716
684.7 0.110 0.154 0.114 0.156
-1.17 9.57 -11.73
31.76 157.00 8.66
6.85 53.12 2.62
4.30 33.29 3.10
SC 0.256 0.264
M AN U
TE D
EP 6,682 6,682 6,682
RI PT
6,916 7,176
AC C
VARIABLES Firm Performance Tobin's Q ROA Presence Blockholder presence dummy Contestability Shapley Contest. Index 1 Contest. Index 2 Dispersion Herfindal Concentration Herfindal Index Finance Size Wedge Sales (Growth) Leverage Beta Free cash flow Tangibility Institutional Country Finance EMBI Domestic credit (w.r.t U.S.) Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.) Macroeconomic Inflation (cpi) Market cap. / GDP GDP growth (Per-capita)
Notes: The table shows the summary statistics for the explanatory variables of interest: presence, contestability, dispersion; and the remaining control variables: country specific financial, institutional and macroeconomic variables. Domestic credit, legal rights index, business freedom, financial freedom and market capitalization are expressed with respect to (w.r.t) the same variable for the United States. All variable definitions are listed in Appendix – table A.1. Source: Authors’ estimation
ACCEPTED MANUSCRIPT 5
Table 6a – Firm value Fixed Effects Regression - Marginal Effects – Voice model Dependent Variable Tobin' Q
Col.1
Col.2
Col.3
Col.4
0.0455 (0.0362)
Contestability Shapley
0.129* (00552) 0.112** (0.0374)
Dispersion Herfindal Index Concentration
-0.11 (0.0794)
Herfindal Index Differentials 3,282
M AN U
Contest. index 2
Col.6
SC
-0.148 (0.0718)
Contest. index 1
Observations
Col.5
RI PT
VARIABLES Presence Blockholder presence
3,282
3,282
AC C
EP
TE D
Significance: *** p < 0.01, ** p < 0.05, * p < 0.1 Source: Authors’ estimation.
3,282
-0.104 (0.0718) 3,282 3,282
ACCEPTED MANUSCRIPT
6
Table 6.b - Firm value Fixed Effects Regression – Average Marginal Effects Voice and Trade models Dependent Variable Tobin' Q x P75=1
x P75=1 x Type
x P75=1
x P75=1 x Type
x P75=1
x P75=1 x Type
x P75=1
x P75=1 x Type
Col.1
Col.2
Col.3
Col.4
Col.5
Col.6
Col.7
Col.8
-0.264**
-0.345**
(0.0909)
(0.0983)
0.135**
0.171**
(0.0561)
(0.0597)
Contest. index 1
0.191**
0.166*
(0.0792)
(0.0859)
Contest. index 2
0.147** (0.0511)
720
866
720
AC C
866
EP
Herfindal Index Differentials
x P75=1
x P75=1 x Type
Col.9
Col.10
Col.11
Col.12
-0.385***
-0.493***
(0.12)
(0.14) -0.391***
-0.437***
(0.113)
(0.146)
866
720
0.130** (0.0552)
TE D
Dispersion Herfindal Index Concentration
Observations
M AN U
Contestability Shapley
x P75=1 x Type
SC
Presence Blockholder presence
x P75=1
RI PT
VARIABLES
866
720
866
720
866
720
Notes: Column P75 = 1 reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) variable in a regression where only these two variables are interacted (two way interaction). Column p75=1 x Type reports the average marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) corresponding to a regression where ownership type is included in the interaction specification (three way interaction) as in the full base line regression equations 5 to 7. Average marginal effect estimates the predicted value of the estimated equation for each observed value of the variable of interest and the average effect is the average of all predicted values. For binary variables, it obtains the predicted values for the ones and zeroes and estimates the difference. In the two way interaction the average marginal effect can be calculated by examining the raw coefficients presented in table A.3 from which the single marginal effects of blockholder presence is 0.0245 and the single marginal effect of the interaction between blockholder presence and high liquidity is 0.11 (Table A.3 Eq.2). The combined marginal effect of blockholder presence is the derivative of the dependent variable with respect to presence: ∂Q/∂Presence = + = 0.0245 + 0.11 = 0.135 which is the value in Col.1; Similarly, the single marginal effect of the interaction between the Shapley value and high liquidity is -0.148 (Table A3 Eq.5). Thus overall marginal effect is ∂Q/∂SV = + = −0.116 − 0.148 = −0.264; which is the value reported in Col.3. The remainder marginal effects are obtained in the same way.
ACCEPTED MANUSCRIPT
7
AC C
EP
TE D
M AN U
SC
Heteroscedasticity corrected standard errors in parentheses; Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation.
RI PT
Columns label as x P75=1 x Type report the marginal effect corresponding to the three way interaction between stock liquidity dummy, type of second blockholder and the variable of interest. In this case the average marginal effect implies a nonlinear effect after taking the average of all observed values. Williams (2012) details the procedure undertaken in the statistical analysis software Stata.
ACCEPTED MANUSCRIPT 8
Table 7 - Instrumental variables regressions full over-identified equations (m = 3) Dependent Variable Tobin' Q
VARIABLES Presence Blockholder presence
Eq.1
Eq.2
Eq.3
Eq.4
Eq.5
Eq.6
...
...
...
...
...
(0.914)
...
...
...
...
...
...
-0.148
...
...
(0.161)
...
Contest. index 1
... ...
... ...
0.0713 (0.132)
Contest. index 2
...
...
...
...
...
Dispersion Herfindal Index Concentration
...
...
Herfindal Index Differentials
... ...
... ...
Contestability Shapley
...
...
...
...
...
... ...
... ...
... ...
0.027
...
...
...
(0.0955)
...
...
...
...
-0.184
...
... ...
... ...
(0.211) ...
... -0.199
...
...
...
...
(0.200)
-0.216***
-0.201***
-0.199***
-0.199***
-0.205***
-0.204***
(0.0560)
(0.0469)
(0.0470)
(0.0471)
(0.0475)
(0.0473)
0.647 (0.421)
0.0263 (0.0906)
-0.0590 (0.0523)
-0.0596 (0.0551)
-0.0806* (0.0476)
-0.0932* (0.0488)
1.42E-03
-3.55E-03
-3.47E-03
-3.32E-03
-3.58E-03
-3.57E-03
TE D
Sales (Growth)
M AN U
Wedge
Leverage Beta
(5.36E-03) 0.375**
(2.72E-03) 0.240**
(2.74E-03) 0.235*
(2.73E-03) 0.238**
(2.73E-03) 0.243**
(2.76E-03) 0.243**
(0.175)
(0.120)
(0.120)
(0.120)
(0.120)
(0.120)
-0.0441 (0.0307)
-0.0252 (0.0199)
-0.0232 (0.0204)
-0.0237 (0.0204)
-0.0284 (0.0200)
-0.0291 (0.0204)
0.105**
0.021
0.018
0.018
0.020
0.020
(0.0441) -0.0441
(0.0126) -0.110
(0.0119) -0.111
(0.0116) -0.114
(0.0125) -0.110
(0.0127) -0.110
(0.147)
(0.108)
(0.109)
(0.109)
(0.108)
(0.108)
-5.84e-05*
-6.13e-05**
-6.35e05***
-6.27e05***
-5.86e-05**
-6.02e-05**
(3.40e-05) 3.059***
(2.38e-05) 2.042***
(2.38e-05) 1.951***
(2.38e-05) 1.961***
(2.40e-05) 2.069***
(2.40e-05) 2.054***
(0.712)
(0.333)
(0.324)
(0.324)
(0.343)
(0.333)
0.431** (0.207)
0.331* (0.173)
0.319* (0.173)
0.325* (0.173)
0.322* (0.175)
0.310* (0.174)
-1.109**
-0.291*
-0.264*
-0.258*
-0.305**
-0.295**
(0.505) 0.0560
(0.152) -0.165
(0.144) -0.165
(0.144) -0.167
(0.153) -0.147
(0.149) -0.143
(0.247)
(0.180)
(0.181)
(0.181)
(0.181)
(0.182)
EP
Lag FCF ratio
AC C
Lag tangibility
SC
...
...
Finance Size
RI PT
1.640*
Country Finance and Institutional EMBI
Domestic credit (w.r.t U.S.) Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.) Fin. freedom (w.r.t U.S.)
ACCEPTED MANUSCRIPT 9
Table 7 - Instrumental variables regressions full over-identified equations (m = 3) (Cont.) Dependent Variable Tobin' Q
VARIABLES Macroeconomic Inflation (cpi)
Eq.1
Eq.2
Eq.3
Eq.4
Eq.5
Eq.6
-4.43E-03 (3.77E-03)
-4.84E-03 (3.74E-03)
-4.70E-03 (3.74E-03)
-4.16E-03 (3.83E-03)
-4.31E-03 (3.81E-03)
Market cap. / GDP
0.0051***
0.00451***
0.0044***
0.0044***
0.0046***
0.0045***
GDP growth (pc)
(8.78E-04) 0.0150***
(6.50E-04) 0.0159***
(6.45E-04) 0.0159***
(6.46E-04) 0.0158***
(6.49E-04) 0.0158***
(6.45E-04) 0.0159***
(3.83E-03)
(2.60E-03)
(2.63E-03)
(2.63E-03)
(2.61E-03)
(2.61E-03)
0.203***
0.192***
0.192***
0.192***
0.194***
0.193***
(0.0383)
(0.0287)
(0.0288)
(0.0288)
(0.0289)
(0.0289)
2,689 426
2,689 426
2,689 426
2,689 426
2,689 426
2,689 426
No
No
No
No
No
No
Underidentification. p-value Valid inst Hansen J. H_0: Overidentification
No Yes
No Yes
No Yes
Yes
Yes
Yes
Yes
Yes
Slope 0.132
Slope 0.128
Slope 0.129
Slope 0.132
Slope 0.130
9.482
9.375
9.378
9.531
9.481
2.45
0.04
0.21
0.44
1.08
[0.118]
[0.849]
[0.977]
[0.644]
[0.506]
[0.298]
3.3
60.9
65.0
60.2
60.7
39.4
Yes Slope …. 6.140
8.17E-04
7.1
101.3
108.2
106.6
76.5
60.6
[0.029]
[0.000]
[0.000]
[0.000]
[0.000]
[0.000]
AC C
Valid inst Hansen J. p-value
No Yes
TE D
Endogeneity p-value Weak inst. H_0: Weak identification Underidentification. H_0: Underidentification
No Yes
EP
Endogeneity H_0: x_j is not endogenous Vs. x_j is endogenous
No Yes
SC
Observations Number of id_firm_n Sector dummy Year dummy Robust SE HAC Interaction R2 centered F-stat
M AN U
Year dummy (2008)
RI PT
3.66E-03 (7.03E-03)
2.860
9.221
8.472
8.511
9.462
9.381
[0.091]
[0.010]
[0.015]
[0.014]
[0.009]
[0.009]
Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 Notes: Estimation results for equations 1 to 6 using instrumental variables to check endogeneity of the dependent variable. HAC stands for heterosckedastic and autocorrelation consistent standard errors. Each regression uses 3 join instruments: i) the lag of blockholder presence, contestability and dispersion definition; ii) operating income volatility; iii) cash flow ratio volatility. Slope interactions refers to presence, contestability and dispersion variables times 75th percentile of stock liquidity dummy. Regressions are controlled by second blockholder type dummies. All variable definitions are listed in Appendix – Table A.1. Source: Authors’ estimation.
ACCEPTED MANUSCRIPT 10
APPENDIX Table A.1 Definition of Variables Source
Definition
Variable Ownership Blockholder Presence
Shareholder 2
Equity share of the 2nd largest blockholder
Shareholder 3
Equity share of the 3rd largest blockholder
Shareholder 4
Equity share of the 4th largest blockholder
Wedge
Shapley Value to control rights largest shareholder ratio
ROA Contestability Shapley Contestability Index 1 Contestability Index 2 Wedge Dispersion
SC
Anual Reports-Thomson Reuters
The Shapley Value solution for the largest shareholders in a three-voting game The share of the second to the largest blockholder ratio.
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
The share of the second and third blockholders to the largest blockholder ratio. Shapley Value to control rights largest shareholder s1 Herfindahl concentration index
H.I Differences
The sum of the squares of the differences between the first and the second H.I Con.
Leverage Stock Beta
Free Cash Flow
Tangibility
AC C
Sales (growth)
EP
H.I Concentration
Finance Size
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Own estimates
Market value/book value: [stock market capitalization + long term liabilities + book value of prefered stocks (if any) + minority interest]/[book value of assets] The ratio of net profits after tax to total assets
TE D
Tobin's Q
M AN U
Performance
Anual Reports-Thomson Reuters
RI PT
Shareholder 1
Dummy variable for firms with blockholder as an investor with equity rights (direct votes) with less than 50% and more than 10% ownership in a listed company Equity share of the largest blockholder
Natural Logarithm of total asssets Percentage of sales growth compared to previous year. The ratio of total liabilities to total assets Standard measure of systemic risk for each firm’s stock with respect to the market. It was measure for stocks which have been active in the stock market for more than 180 days in a given year. Free Cash Flow (FCF) to equity ratio Is defined as EBITDA minus tax expenditure minus interest expenditures to total firm equity. This is an indicator for firm short term liquidity. Total property, plant and equipment divided by total assets
Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
Anual Reports-Thomson Reuters Anual Reports-Thomson Reuters
ACCEPTED MANUSCRIPT 11
Table A.1 Definition of Variables (Cont.) Source
Definition
Business Freedom (w.r.t.US)
Financial Freedom (w.r.t.US)
Measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and thus facilitate lending. Is an overall indicator of the efficiency of government regulation of business. The score for each country is a number between 0 and 100, with 100 equaling the freest business environment. The score is based on 10 factors, all weighted equally, using data from the World Bank’s Doing Business study.
The World Bank — world development indicators
Is a measure of banking efficiency as well as a measure of independence from government control and interference in the financial sector. Index from 0 to 10.
Index of economic freedom
Inflation as measured by the consumer price index reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services Market capitalization (also known as market value) is the share price times the number of shares outstanding, divided by GDP Annual percentage growth rate of GDP per capita based on constant local currency.
The World Bank — world development indicators
RI PT
Legal Rights (w.r.t.US)
Anual Reports-Thomson Reuters The World Bank — world development indicators
Market Cap/ GDP
AC C
EP
GDP Growth (pc)
TE D
Macroeconomic Inflation (CPI)
Index of economic freedom
SC
Domestic Credit (w.r.t.US)
Emerging Markets Bonds Index. Measures financial risk in each country. Expressed in basis points Domestic credit to private sector: refers to financial resources provided to the private sector, such as through loans, purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment
M AN U
Variable Institutional EMBI
The World Bank — World development indicators The World Bank — world development indicators
ACCEPTED MANUSCRIPT 12
Table A.2 Identification - empirical equation Tobin's Q and Shapley Value Fixed Effects Regressions
Contestability Shapley
0.217*** (0.0804) Firm financial and specific variables Size Wedge
Leverage Beta Lag free cash flow ratio
AC C
Legal rights (w.r.t U.S.) Buss. freedom (w.r.t U.S.)
Fin. freedom (w.r.t U.S.)
-0.148 (0.110)
-0.116 (0.112)
0.0780** (0.0391) -0.133 (0.0940) 0.00535 (0.00417) 0.245* (0.130) 0.0120 (0.0209) 0.0160** (0.00754) -0.559*** (0.119)
-0.217*** (0.0577) 0.00756 (0.0796) -0.00150 (0.00340) 0.277** (0.135) -0.0170 (0.0214) 0.00158 (0.0170) -0.235** (0.101)
-0.218*** (0.0576) 0.0181 (0.0798) -0.00150 (0.00339) 0.277** (0.135) -0.0140 (0.0209) 0.00173 (0.0168) -0.238** (0.101)
-5.68e05** (2.23E-05)
-5.38e05** (2.23E-05)
-6.49e05** (2.74E-05)
2.204*** (0.430) 0.407** (0.187)
2.200*** (0.430) 0.404** (0.188)
2.051*** (0.422) 0.330 (0.210)
-0.293* (0.176) -0.165 (0.213)
-0.300* (0.175) -0.183 (0.212)
-0.275 (0.181) -0.195 (0.237)
EP
Country institutional variables
Domestic credit (w.r.t U.S.)
0.199* (0.120)
TE D
Lag tangibility
EMBI
+ Turnover interaction x P75 = 1
Marginal Effect
M AN U
Sales (Growth)
+ Country and Macro controls
+ Turnover and Type x p75 x Type
RI PT
+ Firm specific controls
-0.264** (0.123)
SC
Coeff
VARIABLES
-0.0114 (0.152)
-0.206*** (0.0585) 0.0321 (0.0811) -0.00369 (0.00287) 0.211 (0.152) -0.0282 (0.0214) 0.0153 (0.0123) -0.122 (0.119)
Marginal Effect
-0.345** (0.135)
ACCEPTED MANUSCRIPT 13
Table A.2 Identification - empirical equation Tobin's Q and Shapley Value (Cont.) Fixed Effects Regressions
Market cap. / GDP GDP growth (pc) Year dummy (2008) Constant
1.068***
0.939***
(0.0565)
(0.242)
6,305
4,426
0.005 No
0.031 No
No
No
Yes
Yes
Interaction
Level + slope
Number of id_firm_n
548
-5.24E-03
-5.10E-03
(3.65E-03) 0.00468***
(3.66E-03) 0.00472***
(6.99E-04)
(6.99E-04)
0.0156*** (2.60E-03)
0.0156*** (2.59E-03)
0.0160*** (2.47E-03)
0.203***
0.202***
0.194***
Marginal Effect
+ Turnover and Type x p75 x Type
Marginal Effect
-5.85E-03
(3.62E-03) 0.00456*** (7.41E-04)
(0.0313) 2.062***
(0.0311) 2.060***
(0.0306) 1.873***
(0.346)
(0.346)
(0.368)
3,282
3,280
866
2,699
720
0.118 No
0.120 No
No
0.137 No
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Level + slope
Level + slope
Level + slope
TE D
Observations R-squared Sector dummy Year dummy Robust SE
+ Turnover interaction x P75 = 1
RI PT
Macroeconomic variables Inflation (cpi)
+ Country and Macro controls
SC
+ Firm specific controls
M AN U
Coeff
VARIABLES
Level + slope
Level + slope
Level + slope
475
474
474
435
AC C
EP
Notes: The table reports the identification process of the Tobin’s Q empirical equation for the case of the Shapley Value. It begins with a simple regression and then regressions add the firm-specific, country controls, and the interacting terms of stock liquidity and second blockholder identity. Columns Marginal effects show the marginal effect of the immediate left column. Robust standard errors in parentheses. Significance: *** p < 0.01, ** p < 0.05, * p<0.1
ACCEPTED MANUSCRIPT
14
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within)
Variables Presence Blockholder presence
x P75=1
Eq.1
Eq.2
Eq.3
0.0455 (0.0479)
0.0245 (0.0482)
-0.0144 (0.0914)
Contestability Shapley
Contest Index 1
Coeff.
x P75=1
x P75=1 x Type
Eq.4
Eq.5
Eq.6
-0.148 (0.110)
-0.116 (0.112)
-0.0114 (0.152)
Contest. index 1
H.I. Differences Interaction stock liquidity dummy -0.0919 (0.0969) 0.203*
(0.0558)
(0.123)
0.266** (0.132)
-0.148** (0.0751)
-0.352* (0.180)
EP
P75 x contest_index 2 |
AC C
p75 x contest_index 1
P75 x hi_concentration
0.0548 (0.0603)
TE D
-0.0992** (0.0419) 0.110**
P75 x Shapley
P75 x hi_differences
Eq.7
Eq.8
Eq.9
0.100 (0.0732)
Coeff. Eq.10
HI Concentration x P75=1 Eq.11
x P75=1 x Type Eq.12
Coeff.
x P75=1
x P75= Typ
Eq.13
Eq.14
E
-0.110
-0.0546
-0.0
(0.120)
(0.124)
(0.2
0.0569 (0.0474)
0.1 (0.1
-0.330*** (0.108)
-0.5 (0.3
-0.121 (0.125)
M AN U
Dispersion H.I. Concentration
P75 x blockholder presence
x P75=1
0.129* (0.0683)
Contest. index 2
P75 dummy
Coeff.
Contest Index 2 x P75=1 x Type
RI PT
Coeff.
Shapley x P75=1 x Type
SC
Presence
-0.0762 (0.0488)
-0.0364 (0.115)
0.0909 (0.0764)
0.184 (0.174)
0.112**
0.0977*
-0.0464
(0.0502)
(0.0545)
(0.0815)
-0.0725 (0.0463)
-0.0527 (0.106)
0.0495
0.153
(0.0487)
(0.101)
ACCEPTED MANUSCRIPT
15
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within) (Cont.)
Leverage Beta L.fcf_r L.tangibility
-0.218*** (0.0582) -0.0480 (0.0617) -0.00143 (0.00339) 0.282** (0.134) -0.0164 (0.0214) 0.00179 (0.0172) -0.235** (0.102)
Country Institutional and Financial EMBI -5.70e-05** (2.25e-05) Domestic credit (w.r.t U.S.) 2.163*** (0.427) Legal rights (w.r.t U.S.) 0.399** (0.188) Buss. freedom (w.r.t U.S.) -0.272 (0.172) Fin. freedom (w.r.t U.S.) -0.166 (0.214) Macroeconomic Inflation (cpi) -0.00552 (3.63E-03) Market cap. / GDP 0.00463*** (6.97E-04) GDP growth (pc) 0.0157*** (2.61E-03) Year dummy (2008) 0.202*** (0.0313)
Coeff. Eq.4
-0.221*** (0.0581) -0.0426 (0.0617) -0.00141 (0.00338) 0.285** (0.134) -0.0139 (0.0209) 0.00233 (0.0170) -0.237** (0.102)
-0.214*** (0.0598) -0.0522 (0.0607) -0.00362 (0.00287) 0.225 (0.152) -0.0291 (0.0216) 0.0167 (0.0133) -0.124 (0.119)
-0.217*** (0.0577) 0.00756 (0.0796) -0.00150 (0.00340) 0.277** (0.135) -0.0170 (0.0214) 0.00158 (0.0170) -0.235** (0.101)
-5.36e-05** (2.26e-05) 2.161*** (0.426) 0.396** (0.188) -0.282* (0.171) -0.173 (0.213)
-6.40e-05** (2.77e-05) 2.037*** (0.419) 0.308 (0.208) -0.261 (0.178) -0.170 (0.240)
-5.68e-05** (2.23e-05) 2.204*** (0.430) 0.407** (0.187) -0.293* (0.176) -0.165 (0.213)
-0.00540 (3.63E-03) 0.00468*** (6.98E-04) 0.0156*** (2.59E-03) 0.202*** (0.0311)
-0.00631* (3.58E-03) 0.00455*** (7.37E-04) 0.0160*** (2.49E-03) 0.193*** (0.0306)
x P75=1
Coeff.
Eq.5
Eq.6
Eq.7
-0.218*** (0.0576) 0.0181 (0.0798) -0.00150 (0.00339) 0.277** (0.135) -0.0140 (0.0209) 0.00173 (0.0168) -0.238** (0.101)
-0.00524 (3.65E-03) 0.00468*** (6.99E-04) 0.0156*** (2.60E-03) 0.203*** (0.0313)
Contest Index 2 x P75=1
x P75=1 x Type
Coeff.
Eq.8
Eq.9
Eq.10
RI PT
Eq.3
H x P75=1
x P75=1 x Type
C
Eq.11
Eq.12
E
-0.206*** (0.0585) 0.0321 (0.0811) -0.00369 (0.00287) 0.211 (0.152) -0.0282 (0.0214) 0.0153 (0.0123) -0.122 (0.119)
-0.217*** (0.0578) -0.0413 (0.0599) -0.00149 (0.00339) 0.270** (0.134) -0.0139 (0.0216) 0.00164 (0.0169) -0.240** (0.101)
-0.217*** (0.0579) -0.0365 (0.0600) -0.00141 (0.00339) 0.271** (0.135) -0.0106 (0.0214) 0.00157 (0.0168) -0.243** (0.101)
-0.201*** (0.0572) -0.0654 (0.0561) -0.00366 (0.00284) 0.196 (0.152) -0.0247 (0.0219) 0.0165 (0.0126) -0.126 (0.118)
-0.217*** (0.0573) -0.0262 (0.0599) -0.00138 (0.00339) 0.265** (0.134) -0.0142 (0.0215) 0.00132 (0.0168) -0.240** (0.101)
-0.217*** (0.0573) -0.0215 (0.0599) -0.00133 (0.00338) 0.267** (0.135) -0.0108 (0.0213) 0.00132 (0.0167) -0.243** (0.101)
-0.200*** (0.0573) -0.0481 (0.0564) -0.00346 (0.00284) 0.192 (0.152) -0.0255 (0.0220) 0.0157 (0.0122) -0.125 (0.118)
-5.38e-05** (2.23e-05) 2.200*** (0.430) 0.404** (0.188) -0.300* (0.175) -0.183 (0.212)
-6.49e-05** (2.74e-05) 2.051*** (0.422) 0.330 (0.210) -0.275 (0.181) -0.195 (0.237)
-5.77e-05*** (2.23e-05) 2.145*** (0.423) 0.404** (0.188) -0.266 (0.171) -0.166 (0.213)
-5.55e-05** (2.22e-05) 2.137*** (0.423) 0.398** (0.188) -0.277 (0.170) -0.177 (0.212)
-6.80e-05** (2.77e-05) 1.910*** (0.406) 0.307 (0.209) -0.246 (0.176) -0.178 (0.236)
-5.79e-05*** (2.22e-05) 2.157*** (0.422) 0.413** (0.187) -0.272 (0.171) -0.164 (0.212)
-5.57e-05** (2.21e-05) 2.150*** (0.423) 0.409** (0.187) -0.285* (0.170) -0.176 (0.212)
-6.73e-05** (2.77e-05) 1.942*** (0.409) 0.316 (0.208) -0.252 (0.177) -0.185 (0.236)
-0.00510 (3.66E-03) 0.00472*** (6.99E-04) 0.0156*** (2.59E-03) 0.202*** (0.0311)
-0.00585 (3.62E-03) 0.00456*** (7.41E-04) 0.0160*** (2.47E-03) 0.194*** (0.0306)
-0.00573 (3.62E-03) 0.00463*** (6.94E-04) 0.0157*** (2.60E-03) 0.201*** (0.0314)
-0.00557 (3.65E-03) 0.00465*** (6.95E-04) 0.0156*** (2.59E-03) 0.201*** (0.0313)
-0.00661* (3.57E-03) 0.00443*** (7.23E-04) 0.0167*** (2.48E-03) 0.193*** (0.0305)
-0.00559 (3.63E-03) 0.00464*** (6.93E-04) 0.0156*** (2.60E-03) 0.202*** (0.0313)
-0.00547 (3.64E-03) 0.00466*** (6.94E-04) 0.0156*** (2.59E-03) 0.202*** (0.0312)
-0.00638* (3.58E-03) 0.00445*** (7.32E-04) 0.0165*** (2.47E-03) 0.194*** (0.0305)
SC
Sales (Growth)
Eq.2
Contest Index 1 x P75=1 x Type
TE D
Wedge
x P75=1
Eq.1
EP
Firm Characteristics Size
Coeff.
AC C
Variables
Shapley x P75=1 x Type
M AN U
Presence
ACCEPTED MANUSCRIPT
16
Table A.3 Tobin’s Q regressions and blockholder presence, contestability and dispersion - Fixed Effects (within) (Cont.)
Regression Statistics Observations R-squared Number of id_firm_n Sector dummy Year dummy Robust SE
Eq.2
Eq.3
Coeff. Eq.4
x P75=1
Coeff.
Eq.5
Eq.6
Eq.7
Contest Index 2 x P75=1
x P75=1 x Type
Coeff.
Eq.8
Eq.9
Eq.10
HI Concentration x P75=1
x P75=1 x Type
Coeff.
x P75
Eq.11
Eq.12
Eq.13
Eq.14
2.018*** (0.349)
2.053*** (0.350)
2.026*** (0.379)
2.062*** (0.346)
2.060*** (0.346)
1.873*** (0.368)
1.984*** (0.346)
2.015*** (0.347)
2.047*** (0.369)
1.940*** (0.340)
1.972*** (0.342)
1.993*** (0.367)
2.108*** (0.347)
3,282 0.117 474 No No Yes Level + slope 4.63E-04 9.69E-03 0.117
3,280 0.119 474 No No Yes Level + slope 4.85E-04 1.01E-02 0.119
2,699 0.135 435 No No Yes Level + slope 7.69E-06 5.20E-03 0.135
3,282 0.118 474 No No Yes Level + slope 4.85E-04 9.90E-03 0.118
3,280 0.120 474 No No Yes Level + slope 4.83E-04 1.02E-02 0.120
2,699 0.137 435 No No Yes Level + slope 5.87E-06 5.51E-03 0.137
3,282 0.118 474 No No Yes Level + slope 5.10E-04 1.02E-02 0.118
3,280 0.119 474 No No Yes Level + slope 5.09E-04 1.06E-02 0.119
2,699 0.136 435 No No Yes Level + slope 1.14E-05 4.66E-03 0.136
3,282 0.119 474 No No Yes Level + slope 5.19E-04 1.04E-02 0.119
3,280 0.120 474 No No Yes Level + slope 5.22E-04 1.08E-02 0.120
2,699 0.137 435 No No Yes Level + slope 1.30E-05 4.96E-03 0.137
3,282 0.117 474 No No Yes Level + slope 4.40E-04 9.55E-03 0.117
-27.30
6.662
72.51
1.112
3.155
68.15
15.26
16.30
102.0
29.69
28.14
104.8
-28.89
TE D
Interaction R2 overall R2 between R2 within Hausman test H0: RE Vs. H1: FE (stat)
x P75=1
Eq.1
SC
Constant
Coeff.
Contest Index 1 x P75=1 x Type
M AN U
Variables
Shapley x P75=1 x Type
RI PT
Presence
EP
Notes: This table the Tobin´s Q regression looking at the variables of blockholder control: presence, contestability and dispersion. The regression coefficient of each variable of interest are reported in Eq.1, Eq.4, Eq.7, Eq.10, Eq.13 and Eq.16; Columns label by xP75=1 report the interaction with the stock liquidity dummy at 75Th percentile. Columns label as x P75=1 x report the marginal effect of the variable of interest and its interaction with high liquidity (P75 dummy) and second blockholder ownership type (three way interaction). All variable definitions are listed in Appendix – table A.1.
AC C
Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimation
2.1 (
4. 1.
17
ACCEPTED MANUSCRIPT
TE D
M AN U
SC
RI PT
Figure 1 - Firm type, marginal effect.
AC C
EP
Source: Own estimates. Notes: The figure shows the marginal effect of the variables of interest (blockholder presence, contestability and ownership concentration) whenever the high concentration and the identity blockholder are equal to one. These figures allow distinguishing the effect of the blockholder upon performance when identity is studied.
…….End of document…..
ACCEPTED MANUSCRIPT
Stock liquidity and second blockholder as drivers of corporate value: Evidence from Latin America
RI PT
Highlights • The study examines the relationship between firm value and blockholder control in three dimensions: presence, contestability and dispersion. • Blockholder contestability becomes effective on firms with high liquid stocks.
SC
• The existence and type of a second blockholder within firms with multiple blockholder structures becomes a critical factor in explaining the effect on firm value.
AC C
EP
TE D
M AN U
• Results show that blockholder voice (monitoring) versus exit (trade) models are complementary rather than substitute mechanisms for firm corporate governance.