Keeping it real or keeping it simple? Ownership concentration measures compared

Author:Conny Overland, Taylan Mavruk, Stefan Sjögren
Publication Date:01 Sep 2020
Eur Financ Manag. 2020;
© 2019 John Wiley & Sons Ltd.
DOI: 10.1111/eufm.12249
Keeping it real or keeping it simple?
Ownership concentration measures compared
Taylan Mavruk
Conny Overland
Stefan Sjögren
Department of Business Administration,
University of Gothenburg, Gothenburg,
Taylan Mavruk, Department of Business
Administration, University of
Gothenburg, Box 610, SE405 30
Gothenburg, Sweden.
Funding information
VINNOVA and the Centre for Finance at
the University of Gothenburg
We analyze the distributional properties of ownership
concentration measures and find that measures come
from different underlying statistical distributions. Con-
sistent with theory, some measures that are classified
to represent a monitoring dimension have a positive
influence on firm performance; other measures that are
interpreted to represent a shareholder conflict dimen-
sion are negatively related to firm performance. How-
ever, other measures deviate from this pattern, and
therefore, we cannot conclude that simple measures can
replace complicated measures. Some measures are more
suitable for analyzing the relationship between manage-
ment and owners, whereas other measures are more
suitable for analyzing the relationships among owners.
distributional properties, monitoring dimension, ownership concen-
tration measures, shareholder conflict dimension
C18; C46; C81; D74; G34; L25
We are very grateful to the editor, John A. Doukas, and the anonymous referees for insightful comments on earlier
drafts, which have helped to improve the quality of the paper. We thank Uday Rajan, H. Nejat Seyhun, and M. P.
Narayanan at the Ross School of Business, University of Michigan for their valuable discussions, comments, and
suggestions. We thank Jan Bartholdy and other seminar participants at the 4th workshop of the Nordic Corporate
Governance Network, Reykjavik University, the Copenhagen Business School, and at the World Finance Conference, St
Johns University, New York for their comments and suggestions. We also thank Karin S. Thorburn, Shubhashis
Gangopadhyay, Roger Wahlberg, Ted Lindblom, Mattias Hamberg, Martin Holmén, Evert Carlsson, and Tamir Agmon
at the Centre for Finance and Industrial and Financial Management seminars at the University of Gothenburg for their
valuable comments and advice. We thank Renée Adams for her review and valuable comments. We gratefully
acknowledge Euroclear and SIS Ägarservice for providing data and Dennis Leech for making available the power index
algorithms on his webpage. The usual disclaimer applies. This project was conducted with funding from VINNOVA and
the Centre for Finance at the University of Gothenburg.
A crucial matter regarding the empirical study of central corporate governance issues is how the
ownership of corporations should be measured. In the corporate governance literature, we find
measures ranging from simple proxies, such as the largest owner's voting share, to the
computation of advanced power indices based on game theory. Earlier studies have examined
the distributional properties of a selection of different ownership concentration measures (e.g.,
Bøhren & Ødegaard, 2006; Leech, 2002) and compared how they are related to firm
performance (e.g., Edwards & Weichenrieder, 2009). To our knowledge, there has been no study
that takes the full range of ownership measures and compares their underlying statistical
properties. We conduct a more detailed examination of the statistical properties of ownership
measures. This examination increases (a) our knowledge regarding whether the simple
measures found in most empirical research are sufficient to capture the role of ownership or if
more elaborate concentration measures are necessary and (b) our understanding of the
ownership dimensions that are reflected in different concentration measures.
The relationship between ownership concentration and firm performance is a common topic
in studies that make use of ownership measures. These studies yield conflicting results. For
instance, Morck, Nakamura, and Shivdasani (2000), Thomsen and Pedersen (2000), Gedajlovic
and Shapiro (2002), Nguyen, Locke, and Reddy (2015), and Bykova, Molodchik, and Shamilova
(2017) find a positive relationship between ownership concentration and firm performance,
whereas Leech and Leahy (1991), Lehmann and Weigand (2000), and, to some degree,
Claessens, Djankov, Fan, and Lang (2002) and Azar, Schmalz, and Tecu (2018) find that
ownership concentration leads to weaker firm performance. Other studies document a
nonlinear (De Miguel, Pindado, & de la Torre, 2004; McConnell & Servaes, 1990), a mixed
(Gedajlovic & Shapiro, 1998; Gugler, Mueller, & Yurtoglu, 2006; Iwasaki & Mizobata, 2019), or
even a nonexistent (Barontini & Caprio, 2006; Demsetz & Villalonga, 2001) relationship
between ownership concentration and firm performance.
Studies show different results partly because they examine different countries, different time
periods, and use different methods (cf. Wang & Shailer, 2013). Such differences make the
interpretation of the relationship between ownership and performance more difficult (Adams &
Ferreira, 2008). Analyzing ownership concentration, the focus of this study, is also more
difficult when using different concentration measures, as they could capture different aspects of
how ownership drives decisionmaking.
Considering ownership concentration, there are at least two potential causes of conflict that
could affect decisionmaking in firms: one has to do with the relationship between managers
and owners, and the other relates to the interplay between owners. In line with Jensen and
Meckling (1976), utilitymaximizing managers could give rise to agency costs. Such agency costs
can be reduced through increases in ownership concentration since large shareholders have
stronger incentives to engage in costly monitoring. We call this the monitoring dimension.
However, conflicting interests among owners could also curb firm financing decisions and firm
performance, as concentrated ownership paves the way for large shareholders to expropriate
minority shareholders (Burkart, Gromb, & Panunzi, 1997; Claessens et al., 2002; Croci, Doukas,
& Gonenc, 2011; Shleifer & Vishny, 1997). We call this the shareholder conflict dimension.
A concentration measure could be better suited for analyzing one of the two dimensions but
could be worse for the other. For instance, when studying the monitoring dimension, a measure
such as the voting share held by the largest owner could be a reasonable proxy for
concentration, as having selfdealing managers is not in the best interest of shareholders. If,
instead, the shareholder conflict dimension is the main concern, a measure of the largest
shareholder is less likely to contain satisfactory information. The largest shareholder's
opportunity to extract private rent depends not only on the shareholder's own voting share
but also on the influence of the other shareholders. Therefore, previous research may in fact
study different phenomena, despite claiming to study the same. In contrast, we examine what
the distributional properties are of different concentration measures, whether they capture
different dimensions of ownership, and whether they return consistent results when relating
them to commonly used performance measures?
The fact that these conflicts can be of different magnitude can be explained by the separation
of voting rights (control) and cash flow rights. While a higher value of control rights is expected
to increase an owner's ability to extract rents from the firm at the expense of the other owners, a
higher value of cash flow rights might create good incentives for monitoring management.
Thus, one would expect that the controlling ownersalignment with the other owners will
decrease as the voting rights deviates from the cash flow rights (Claessens et al., 2002; Edwards
& Weichenrieder, 2009) and would therefore lead to increased shareholder conflict. The
existence of dual shares should affect the monitoring dimension to a lesser degree.
In our
analysis, to enhance our understanding of the diverse effects of ownership concentration on
corporate governance, we consider the distinction between control and cash flow rights.
Based on a sample of all Swedish companies listed on the Stockholm Stock Exchange (SSE)
between 2000 and 2009 and using ultimate controlling ownersinformation derived from Owners
and Power data (Fristedt & Sundqvist, 20012010) and data on share prices and accounting
variables from Bloomberg, we create 20 ownership concentration variables measuring the
ownership structure for an unbalanced panel of 361 firms. By limiting the data set to include only
Sweden, contextual differences are mitigated because all observations are subject to the same legal
institutions and the same timespecific macro events. Data quality problems are also mitigated as
our data set shows a large crosssectional variation in concentration measures and is free from
incompleteness and nonrandom reductions in sample size.
We carry out three stages of analyses. In the first stage, we examine the distributional properties
of the different ownership measures by using the following: (a) the Spearman rank correlation test
to determine whether the ownership concentration measures rank firms differently; (b) the
Wilcoxon matchedpairs signedrank test to determine whether distributions of the concentration
measures come from the same distribution; and (c) the AndersonDarling test to determine what
distribution best fits each measurement. In the second stage, we perform a principal component
analysis (PCA) and examine whether the two dimensions (monitoring and shareholder conflict) are
discernible in the underlying factors. In the third stage, we relate the ownership concentration
measures to firm performance. Our aim is to providesomeformalreasoningforhowandwhythe
choice of the concentration measure can affect analytical outcomes such as firm performance. We
use different models to carefully examine the endogeneity issues with ownership variables because
the causes of endogeneity in these different ownership measures are uncertain and depend on
which dimension the particular measure represents. We do not expect all models to produce the
same signs for the particular ownership measure; however, our roadmap for the decision criteria is
Edwards and Weichenrieder (2009), finding that the weakest link principle does not perform well, indicate that it cannot be used to identify the effects of
ownership when there is a separation of voting and control rights. They instead suggest and perform a comparison of voting power indices. There is solid
evidence that control benefits do exist (Adams & Ferreira, 2008; Barclay & Holderness, 1989; Bergström & Rydqvist, 1990; Bertrand, Mehta, & Mullainathan,
2002; Dyck & Zingales, 2004; Franks & Mayer, 2001; Gugler & Yurtoglu, 2003; Johnson, La Porta, LopezdeSilanes, & Shleifer, 2000; Lauterbach & Pajuste,
2015; Nenova, 2003; Saggese, Sarto, & Cuccurullo, 2016; Zingales, 1994).
NASDAQ OMX Stockholm.

To continue reading

Request your trial