1
Corporate governance efforts of
institutional investors
Thijs de Haan1University of Groningen Faculty of Economics and Business MSc International Financial Management
and MSc Finance
Uppsala University Department of Business Studies
MSc Business and Economics
Supervisor: Marnix Reijenga
13 June 2016
Abstract
This paper studies the effect of institutional ownership on firm-level corporate governance. Rather than focusing on the actions of a subset of institutions, we try to gain an understanding of a world in which different kinds of institutions interact. Using a dataset comprising 3,575 firms from 54 countries over a period of 5 years (2010-2014), we find that quantitative characteristics of institutional investors perform better than qualitative ones in
explaining corporate governance efforts.
JEL classification: G23, G32, G34
Keywords: corporate governance, institutional investors, ownership structure
2
1. Introduction
Today, the securities markets of many countries are dominated by institutional investors. Institutional investors are defined as “professional money managers with discretionary control over assets” (Ferreira and Matos, 2008) and include banks, insurance companies, private equity parties and all sorts of funds (mutual, pension, sovereign wealth). The Boston Consulting Group (BCG) reports that after three years of consecutive growth, total assets under management of these parties now amounts to $74 trillion2 – an all-time high
(BCG, 2015). When it comes to stock ownership, there are major differences between types of institutional investors. For example, hedge funds and private equity firms are known to take large, controlling interests in a select set of firms. Mutual funds and pension funds tend to diversify by holding smaller stakes in a large number of firms.3 Consequently, a single firm may
be held by a few dedicated institutional investors or by many diversified ones (or a combination of these).
Due to its increasing prevalence, ownership by institutional investors (hereafter:
institutional ownership) has become a major topic of interest in academia and in the financial
press. In both of these settings, the question that is being asked is: what are the consequences of institutional ownership for (1) corporate governance and (2) firm value? Obviously this is a broad question. An all-inclusive answer would have to take into account the various types of institutional owners that exist and the amount of shares held by them. It would have to analyze numerous aspects of corporate governance, from the quality of the board of directors to shareholder rights to CEO compensation. Lastly, it would have to verify the existence of a relationship between corporate governance and firm value. In practice, many academic studies and news items focus on one (type of) owner and its impact on either a limited number of corporate governance variables or directly on firm value. For instance, in the 1990s, Californian pension fund CalPERS received considerable attention for its ability to boost firm value through shareholder activism (see Smith, 1996). At the other extreme, in 2007, British hedge fund TCI famously broke up Dutch bank ABN AMRO by acquiring 1% of its shares and calling for a shareholder meeting. This manoeuvre was widely considered to be value-destructive as two of the buying parties (RBS and Fortis) incurred great losses and had to be bailed out by their respective governments. Around this time, academic literature increasingly started to cover the governance efforts of ‘active’ shareholders (Parrino, Sias and Starks, 2003; Brav, Jiang, Partnoy and Thomas, 2008; Del Guercio, Seery and Woidtke, 2008; for a good
2 This number includes assets allocated to equity, but also to other classes such as fixed income. 3 These funds can be divided into active and passive funds. Active funds invest in stocks at the fund
3 roundup see Edmans, 2014). More recently, attention has shifted to the lack of such efforts by passive shareholders. In March 2015, Charlie Munger, vice-chairman at Berkshire Hathaway, warned that index funds4 would become large, permanent owners with “power they are not
likely to use well”. Indeed, index fund giants such as Vanguard rarely take an active stance on corporate governance. An investigation of their “N-PX” filings shows that they have a track record of voting with management at shareholder meetings (The Motley Fool, 2015). Since index funds are growing at the cost of actively managed funds (ICI, 2015), this could be a legitimate cause for concern. Mr. Munger’s warnings have since been echoed in the European financial press and several fund managers have committed to strengthening their corporate governance (FT, 2015a; FD, 2016). Over the same period another traditionally ‘quiet’ investor, the sovereign wealth fund of Norway, announced that it would engage its held firms on issues like board independence (FT, 2015b) and executive pay (FT, 2016). Research on passive forms of institutional ownership is only just starting to appear (for an early example see Appel, Gormley and Keim, 2016).
As it stands, there is no clear-cut answer in regard to the effects of institutional ownership on corporate governance and ultimately on firm value. If news releases are anything to go by, it seems that even the most passively managed institutions are highly dedicated to corporate governance. Yet monitoring is costly, and these investors know that they need to strike a balance between maximizing firm value and keeping their own costs down. Empirical research can help to understand this behavior, but all too often authors choose a subset of institutions to study in isolation, which yields no understanding of the real world in which different types of institutions interact. In this paper, we aim to overcome these issues by reconciling modern empirical studies (which target increasingly specific types of institutions) with classic ownership theory (which treats all investors as homogeneous). Using an ownership database unmatched in its completeness, we find that quantitative characteristics perform better than qualitative ones when it comes to explaining diversity in the corporate governance efforts of institutional investors.
The paper proceeds as follows. Section 2 reviews the academic literature on ownership in general and institutional ownership in particular and presents the paper’s key research question. Section 3 translates this research question into a regression equation and briefly discusses the methodology used. Section 4 describes the data collection and cleaning process and reports descriptive statistics for the final sample. Section 5 contains the main findings, i.e., evidence for the effect of institutional ownership on corporate governance. Section 6 concludes with the paper’s contributions and limitations.
4
2. Literature
Any scientific paper on ownership structure and corporate governance inevitably starts with a reference to Jensen and Meckling (1976)’s Theory of the Firm. More clearly than anyone before them, these authors describe the separation of ownership and control that characterizes the modern corporation. Managers (‘agents’) are hired by shareholders (‘principals’) to run their firm, but as objectives between the two may differ, agency costs are incurred. Corporate governance can then be defined as the range of mechanisms available to shareholders (Shleifer and Vishny, 1997) to minimize agency costs and ultimately maximize shareholder value (Tirole, 2001).
From this point of departure, the literature has tried to resolve two questions: firstly, whether corporate governance truly increases firm value, and secondly, which kinds of owners (‘principals’) govern agents most effectively. Empirical evidence in support of the first question is presented by Gompers, Ishii and Metrick (2003). They find that stronger corporate governance is associated with higher firm value, profits and sales growth. The answer to the second question is not nearly as straightforward. Shareholders have been compared on a wide range of dimensions including size, nationality (domestic or foreign), power (concentrated or dispersed), type (individual, institution, government) and style (activist or absentee). Over time, these dimensions (or: characteristics) have become increasingly specific, and significant effects have been ascribed to virtually all of them. Since most studies include only one or two characteristics at a time, it remains unclear which of them has the greatest impact on the quality of corporate governance. Below, we present the most compelling findings in a roughly chronological order.
Shleifer and Vishny (1986) argue that large shareholders, referring to those holding a significantly large percentage of shares outstanding5, are the only type of owner to have the
ability as well as the incentive to govern firms effectively. Small shareholders lack both the resources to constantly monitor managers and the reason to do so, as their small stake allows them to benefit only marginally from the potential increase in firm value. Due to its simplistic nature, Shleifer and Vishny’s argument can be applied to both institutional and non-institutional investors. It is easy to imagine that for many financial institutions, the theorem might hold. Banks and fund managers are large and wealthy enough to allocate at least some of their resources to monitoring. Furthermore, any increase in firm value due to monitoring is immediately recorded as a gain in value for the fund, which makes for a strong incentive. However, there is also good reason to believe that the theorem might not hold. Many institutions now compete on costs; especially index funds, for whom it is their main selling point (ICI, 2015). It might not be advantageous for them to spend ample resources on
5 monitoring. Even if we look past arbitrarily defined owner types, we can see that the (total) cost of monitoring depends on the number of firms held, while the benefit depends on the size
of the shares taken in those firms. A similar argument is made by Fich, Harford and Tran
(2015). Thus, in its most basal form, corporate governance is more accessible to investors who hold a few large stakes than to those who own many small shares.
In the thirty years following Shleifer and Vishny (1986), a sizable amount of literature has emerged with specific regard to institutional ownership. These papers established the two ways in which institutional investors can exert corporate governance. Firstly, they can engage in shareholder activism6 by proposing governance improvements of their own or voting
favorably on the proposals of other shareholders. Typical governance-related shareholder proposals include those to limit executive pay, require independence of the board of directors, or remove poison pills and golden parachutes (Davis and Kim, 2007). Gillan and Starks (2000) find that institutional investors are particularly successful as shareholder activists, because proposals sponsored by institutions garner more votes than those sponsored by individual investors. In addition to voting for or against issues, institutional investors are sometimes convinced by small activists to withhold their votes altogether in order to put pressure on boards (Del Guercio et al., 2008).
The second way in which institutional institutions can exert corporate governance is by selling shares, or refraining to invest, in companies which they believe to be poorly managed. This behavior is colloquially known as “voting with their feet”. Parrino, Sias and Starks (2003) find that the incentive to sell shares is present within institutions who are concerned with holding prudent stocks and within those who are better informed. Consequently, they find that a drop in institutional ownership increases the likelihood of forced CEO turnover. Both shareholder activism and “voting with their feet” are thus valid options for institutions to improve corporate governance. However, Helwege, Intintoli and Zhang (2012) find that in recent years, the impact of “foot voting” has declined in favor of activism. Of course, part of this change may be attributable to the shift from active to passive funds. Index funds have no way of voting with their feet; their stock holdings are bound by the index that they track. Whether this makes them more likely to engage in shareholder activism is up for debate (Edmans, 2014).
An even narrower branch of literature is concerned with the question of which institutions matter most (Chen, Harford and Li, 2007). Authors in this field study one type of (institutional) owner in great detail, while leaving out others. The type of owner that has received the most attention here is the mutual fund, helped by the fact that they need to publicly disclose their voting behavior, at least in the U.S. Davis and Kim (2007) hypothesize
6 that mutual funds will monitor firms more weakly (i.e., vote in support of their management more often) if the two parties have business ties (i.e., the firm invests in the fund as part of its pension plan). They find that individual voting outcomes are not affected by the presence of business ties, but funds with more business ties do vote more management-friendly on aggregate. Iliev and Lowry (2015) find that mutual funds only evaluate governance issues if they consider the benefits of doing so to be higher than the costs. If not, they will rely on so-called ‘proxy advisory services’. Both of these findings support the hypothesis that diversified investors are less concerned with corporate governance and may opt for a “one-size-fits-all” policy. On the contrary, Appel, Gormley and Keim (2016) find that passive fund ownership is associated with greater board independence, removal of “poison pill” structures, and more equal voting rights – all signs of improved corporate governance. Consequently, they also find a positive effect on firm performance, measured by the return on assets (ROA) and Tobin’s Q. All of the literature described so far focuses on U.S. financial institutions holding predominantly U.S. companies. However, research has also been done on an international level. One particularly extensive paper by Ferreira and Matos (2008) explores the holdings of institutional investors worldwide and their impact on firm value. These authors make a distinction between domestic institutions, which are located in the same country as the firm whose shares they hold, and foreign institutions, which are located in a different country. Taking into account the potential conflict of interest of Davis and Kim (2007), they further divide institutions into grey institutions, which may have business ties with the firms they hold, and independent institutions, which may not. Their results show that firms owned by foreign and independent institutions have higher-than-average firm valuations, while firms primarily owned by domestic and grey institutions do not. The authors attribute this difference to the former group’s superior corporate governance.
7
2.1 Research question and pseudo-hypotheses
Building on the last set of papers discussed, but integrating them with the theory of Shleifer and Vishny (1986) and subsequent updates, we intend to answer the following research question:
Which institutional investors are most concerned with corporate governance?
We do not make any predictions regarding the answer to this question. We only expect that governance efforts will vary among institutional investors (1) of domestic and foreign origin, (2) of different types, and (3) with different shareholding distributions. For each investor, the following three pseudo-hypotheses can be formulated:
Positive hypothesis: the institutional investor improves corporate governance
Neutral hypothesis: the institutional investor has no effect on corporate governance
Negative hypothesis: the institutional investor impairs corporate governance
The answer to the research question is taken to be the investor who best fulfills the positive hypothesis.
3. Methodology
Research on institutional ownership has thus far been of a quantitative nature. Though we are interested in the mechanisms that owners employ to exert pressure on managers, the literature is still at a stage where we gain most from large sample studies (as opposed to case studies). To quantify the effects of institutional ownership on corporate governance, we use the following regression equations:
𝐶𝐺𝑖𝑡 = 𝛼 + 𝛽 ∗ 𝐼𝑂_𝑇𝑂𝑇𝐴𝐿𝑖𝑡+ 𝛾 ∗ 𝑿 + 𝜖
𝐶𝐺 = 𝛼 + 𝛽1∗ 𝐼𝑂_𝐷𝑂𝑀𝐸𝑆𝑇𝐼𝐶 + 𝛽2∗ 𝐼𝑂_𝐹𝑂𝑅𝐸𝐼𝐺𝑁 + 𝛾 ∗ 𝑿 + 𝜖
𝐶𝐺 = 𝛼 + 𝛽1∗ 𝐼𝑂_𝐴 + 𝛽2∗ 𝐼𝑂_𝐵 + 𝛽3∗ 𝐼𝑂_𝐶 + ⋯ + 𝛾 ∗ 𝑿 + 𝜖
𝐶𝐺 = 𝛼 + 𝛽1∗ 𝐼𝑂_𝐹𝑄1 + 𝛽2∗ 𝐼𝑂_𝐹𝑄2 + 𝛽3∗ 𝐼𝑂_𝐹𝑄3 + 𝛽4∗ 𝐼𝑂_𝐹𝑄4 + 𝛾 ∗ 𝑿 + 𝜖
𝐶𝐺 = 𝛼 + 𝛽1∗ 𝐼𝑂_𝑀𝑄1 + 𝛽2∗ 𝐼𝑂_𝑀𝑄2 + 𝛽3∗ 𝐼𝑂_𝑀𝑄3 + 𝛽4∗ 𝐼𝑂_𝑀𝑄4 + 𝛾 ∗ 𝑿 + 𝜖
8
4. Data
This section provides an overview of the variables used in this study and reports the descriptive statistics for the final sample of firms. As can be seen from the regression equations, the dependent variable is a measure of firm-level corporate governance and the independent variables are based on institutional ownership data. There are also other firm characteristics that may serve as control variables. In order to improve the quality of the data and ensure the validity of the results, a significant number of corrections had to be made and filters had to be applied. Therefore, a considerable part of this section will be spent explaining and justifying these modifications.
4.1 Firm-level governance
As we have seen in the literature, the ‘quality’ of corporate governance within a firm can be assessed by looking at various attributes, such as board independence, executive compensation, and the presence of poison pills. Firms can be scored on their (relative) performance in each of these areas and these scores can then be aggregated to calculate a single ‘corporate governance index’. Aggarwal et al. (2011) obtain 41 individual attribute scores from RiskMetrics and use these to calculate their own index, in which each attribute carries an equal weight. We use an alternative index produced by Thomson Reuters’ ASSET4 division. The ASSET4 database consists of environmental, social and governance (ESG) data; it is the latter in which we are interested. In addition to the overall corporate governance score (CGVSCORE), we obtain individual scores for the quality of the board functions (CGBF), the board structure (CGBS), compensation policy (CGCP), firm vision and strategy (CGVS) and shareholder rights (CGSR). The overall corporate governance score is composed similarly to – and thus allows for comparison with – the index created by Aggarwal et al. (2011). The sub-scores have the potential to yield more detailed findings: using them as the dependent variable (instead of the index), we may be able to tell which aspect of corporate governance institutional investors are most capable of improving. All scores range from 0 to 100, making it easy to interpret the coefficients of the regression. Since the ASSET4 database is smaller than the Orbis database described below, it imposes the first constraint on the number of firms we can study. The initial sample consists of all 4,750 firms for which ASSET4 has at least one year of data in the period 2010 to 2014. Note that this number includes financial companies, which we filter out later.
4.2 Institutional ownership
9 wealth). Explicitly excluded are individuals7 as well as entities who merely hold stocks without
controlling them (known as ‘nominees’ or ‘custodians’). There are a handful of sources that provide data on institutional ownership, including FactSet (used by Ferreira and Matos (2008) and Aggarwal et al. (2011)) and Thomson Reuters Ownership (used by Appel et al. (2016)). Unfortunately, we could not gain access to either of these databases. Instead, we acquired our data from Bureau van Dijk’s Orbis. Orbis gathers ownership data from a variety of sources: often from FactSet, but sometimes also from stock exchanges, annual reports, and other data providers. This practice should make for a more complete dataset, but for the purpose of this study it yields two additional challenges. Firstly, to separate true institutional ownership from other types of owners, such as dedicated holding companies and the aforementioned nominees. Secondly, to remove any duplicate entries that may have arisen as a result of using multiple data sources.
The same 4,750 firms that are present in ASSET4 are matched to Orbis by using International Securities Identification Numbers (ISINs). At this point, banks and other financial entities are dropped (note: from the list of firms, not the list of shareholders) because their data is incomparable with that of industrials. 3,674 firms remain. Orbis provides us with an exhaustive list of their shareholders’ names, types, countries of origin, and holdings (as a percentage of shares outstanding) in each year from 2010 to 2014. Because we are only interested in institutional ownership, we now try to filter for shareholders from the financial sector and governments. We do this primarily by using shareholders’ industry classifications (NACE) and secondarily by looking at their types (appointed by Orbis). We keep shareholders whose NACE codes start with 64, 65, 66, 70, or 82 (representing the financial sector) or with 84 (representing governments). For shareholders without a NACE code, we keep them if they belong to one of the following types:
Bank (B)
Financial company (F) Insurance company (A) Industrial company (C)
Mutual & Pension Fund/Nominee/Trust/Trustee (E) Foundation/Research Institute (J)
Private equity firm (P) Venture capital (V) Hedge fund (Y)
Public authority, State, Government (S)
7 Unless these individuals invest through a company (like Warren Buffett) or a foundation (like Bill
10 Note again that these types are only used if the shareholder’s NACE code is absent. It is also worth noting that the Foundation, Venture capital and Hedge fund types are rarely used; together they account for less than 1% of firm-year-shareholder observations. To further filter for institutional investors, we exclude shareholdings of more than 50% for the financial types (these are holding companies dedicated to a single firm) and of more than 25% for governments (these are state-owned firms). These upper boundaries are already quite high; institutions such as mutual funds generally hold no more than 10%.
At this point the data, consisting of no less than 957,402 firm-year-shareholder observations, can be downloaded and opened. Of the 16,844 ‘unique’ shareholders in this dataset, we manually identify 2,354 as subsidiaries/aliases of other shareholders. We replace these entries (name, country, and type) by their most common version and delete the duplicates. We keep the first observation, i.e., the one with the highest percentage of shares held8. We also delete those shareholders whose names indicate they are nominees or
custodians9. The data should now be clean, but there are still some dubious (read:
non-institutional) names among the shareholders. To combat this, we remove all shareholders who
do not have any observations sourced by FactSet from the data. This is a significant change,
and unlike the others it is subjective rather than objective. However, it is also very effective (reducing the number of ‘unique’ shareholders from 14,490 to 4,617) and has the benefit of making our dataset more similar to that of Ferreira and Matos (2008) and Aggarwal et al. (2011). The final issue that presents itself is that the ownership data seems to be incomplete in certain years. For instance, Unilever NV has 243 listed shareholders in 2012 and 259 in 2014, but only 109 in 2013. This causes serious distortions if we are to sum the holdings for each year. However, dropping all cases with fewer than K shareholders is not an option, because other firms may simply have a lower number of shareholders. Therefore, we drop firm-year combinations if the number of shareholders in that year is less than 80% of the average number of shareholders in all available years. Returning to the Unilever example, the observations from 2013 are dropped because 109 < 0.8 ∗243+109+259
3 . However, had there been
190 shareholders in 2013 they would have been kept because 190 > 0.8 ∗243+190+259
3 . In the
end we are left with 854,311 firm-year-shareholder observations.
The fully cleaned data is transformed into multiple variables, some of which were introduced by Ferreira and Matos (2008) and Aggarwal et al. (2011) and some of which are
8 The exception is when the top observation contains an approximation (e.g. “<5%”) and the second
observation contains the exact amount (e.g. “4.81%”). In such cases we keep the second observation.
9 Their names are generally too confusing to determine for whom the shares are held, e.g. “HSBC
11 new to this study. First, we sum the percentage of shares held by all institutions (IO_TOTAL) for every firm in every year. We then categorize total institutional ownership in four different ways. The first categorization is based on the nationality of the owner. We sum shares held by institutions domiciled in the same country as the firm (IO_DOMESTIC) and shares held by institutions domiciled in a country differing from that of the firm (IO_FOREIGN). The second distinction is based on the type of owner. Recalling the ten types of owners listed earlier in this section, we sum shares held by banks (IO_B), financial companies (IO_F), insurance companies (IO_A), industrial companies (IO_C), mutual and pension funds (IO_E), foundations (IO_J), private equity (IO_P), venture capital (IO_V), hedge funds (IO_Y), and governments (IO_S).10 The dimensions of nationality and type will later be referred to as the
qualitative factors. The third categorization is based on the number of holdings of the owner.
We simply count the frequency with which an institution appears in the dataset and then classify it as being in the lowest quartile (0-25%) of owners, the second (25-50%), the third (50-75%), or the highest (75-100%). Consequently, we sum the percentage of shares held by first-quartile owners (IO_FQ1), second-quartile owners (IO_FQ2), etc. To give an indication of what these quartiles mean: all shareholders in the first quartile have fewer than 500 total holdings (approx. 100 per year), while those in the fourth quartile boast more than 5,000 (approx. 1,000 per year). Lastly, our fourth categorization is based on the typical size of holdings of the owner. We calculate the median percentage of shares held by an institution (in all firms) and again classify it as being in a certain quartile, then sum the percentage of shares held by first-quartile owners (IO_MQ1), second-quartile owners (IO_MQ2), etc. Following Shleifer and Vishny (1986), the dimensions of frequency and median should be able to give an indication of the average involvement of a shareholder. We will refer to these dimensions as the quantitative factors.
4.3 Firm characteristics
Firm-level control variables are represented through a single ‘economic score’ also obtained from ASSET4. Much like the corporate governance score, the economic score is a weighted average of various attributes, in this case of economic performance, shareholder loyalty and client loyalty. It includes most of the variables that show up as being statistically significant in Aggarwal et al. (2011)’s regressions, such as firm size (total assets), leverage (debt-to-equity ratio) and profitability (return on assets). The advantage of using this score rather than its individual elements is that it is available for all firms. To illustrate, the debt-to-equity ratio is known for roughly half of the firms in our dataset. Explicitly requiring this as a control variable would therefore cut the number of observations to be used in our regressions
10 We also experimented with using the industry (NACE) code as an indication of the type of owner.
12 in half. On the contrary, every firm has an ASSET4 economic score, which is always composed from the maximum amount of data points possible. Since we are not interested in measuring the effect of firm-level characteristics, but only in controlling for them, the economic score is perfect for our needs.
4.4 Descriptive statistics
The final sample of merged Orbis and ASSET4 data contains 14,005 observations for 3,575 companies, meaning there are on average 4 years of data available for each company. 11,212 observations, or 80% of total, are from countries also studied by Aggarwal et al. (2011). These are the United States, Canada, the United Kingdom, France, Germany, Switzerland, Italy, Spain, the Netherlands, Sweden, Finland, Belgium, Norway, Denmark, Greece, Austria, Ireland, Portugal, Japan, Australia, Hong Kong, Singapore, and New Zealand11. In addition,
our dataset contains information on companies from Taiwan, Korea, South Africa, India and Brazil. To maintain the ease of comparison with Aggarwal et al. (2011), we do not include these countries in the descriptive statistics. However, we do include them in our regressions. Over the next few pages, Figure 1 shows the percentage of shares held by domestic and foreign institutions, Figure 2 shows the shares held by institutions with different amounts of holdings (referred to as ‘frequency’), and Figure 3 shows the shares held by institutions with different holding sizes (referred to as ‘median’) 12. Our graphs differ from those of Aggarwal et al. (2011)
in that they are based on average holdings over a period of five years (2010-2014), while theirs consist of a snapshot of the distribution in one year. The numbers supporting the graphs can be found in Tables A2, A3, and A4 of the Appendix.
11 Roughly in order of number of observations by continent.
12 Attentive readers will notice that we fail to include a graph of shares held by institutions of different
13 Figure 1. Figure 2. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% US CA UK FR DE CH IT ES NL SE FI BE NO DK GR AT IE PT JP AU HK SG NZ
Institutional ownership by nationality
Domestic Foreign 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% US CA UK FR DE CH IT ES NL SE FI BE NO DK GR AT IE PT JP AU HK SG NZ
Institutional ownership by frequency
14 Figure 3.
Table A2 of the Appendix tells us that worldwide, an average of 52% of tradable shares is now in the hands of institutional investors. This is a substantial increase from the average of 27.9% found by Ferreira and Matos (2008). In Figure 1 (and 2 and 3) we see how this figure varies per country. Institutional holdings upward of 60% are recorded in the U.S. and the U.K., while they are less than 20% in Greece, Japan, and Italy. The division of shares between domestic and foreign institutions seems to be skewed slightly in favor of domestic investors, who take 30.8% of ownership (versus 21.9% by foreigners). However, Figure 1 suggests that these are to a large extent the same (American) investors, given the size of the domestic share in the U.S. and that of the foreign share elsewhere. In fact, the only non-U.S. country in which domestic institutional investors outweigh foreign institutional investors is Sweden. Table A3 further shows that shares are predominantly held by institutions classified by Orbis as ‘banks’. Insurance companies are somewhat common in the U.K. and Finland. Investing governments (e.g., sovereign wealth funds) play a niche role in Norway and Sweden. Table A4 and Figures 2 and 3 are a bit more complicated. Here we observe that owners from the upper two quartiles (frequency or median) generally account for more than half of total institutional ownership, which makes sense13. The interesting countries are those in which low-frequency or small-stake
investors collectively own most of the shares, as is the case in Scandinavia (Sweden, Finland, and Denmark) and Australia and New Zealand.
13 The upper quartiles contain the investors with more or larger stakes than those in the lower quartiles.
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% US CA UK FR DE CH IT ES NL SE FI BE NO DK GR AT IE PT JP AU HK SG NZ
Institutional ownership by median
15
5. Results
In this section, the main results of the study are presented. The analysis is divided over several panels, the first three of which differ in terms of independent variables and the last of which differs only in terms of the independent variable. All reported regressions are carried out as pooled OLS regressions with country, industry (NACE), and year dummies. As a robustness check, we also performed panel regressions with firm and time fixed effects, the results of which we address in the conclusion. In all of the tables below, standard errors have been corrected for heteroskedasticity and clustering.
Table 1 presents the effects of total, domestic, and foreign institutional ownership on the corporate governance index. This panel is methodologically comparable to Table 4A of Aggarwal et al. (2011) and indeed, the results are very similar: we find a statistically significant positive effect of both domestic and foreign institutional ownership on corporate governance, with a slightly higher coefficient for the foreign institutions. However, in our model, unlike in theirs, the domestic institutional ownership effect remains significant if we include both variables in the same regression, as is done in column (4). This makes it hard for us to support their conclusion that foreign institutions are more important than domestic institutions when it comes to improving corporate governance.
Table 1. Dependent variable: corporate governance index (CGVSCORE)
(1) (2) (3) (4) IO_TOTAL 0.165*** (0.009) IO_DOMESTIC 0.091*** (0.011) 0.132*** (0.011) IO_FOREIGN 0.184*** (0.012) 0.211*** (0.012) ECNSCORE 0.247*** (0.005) 0.251*** (0.005) 0.246*** (0.005) 0.246*** (0.005)
Country dummies Yes*** Yes*** Yes*** Yes***
Industry dummies Yes*** Yes*** Yes*** Yes***
Year dummies Yes Yes Yes Yes
N 13,978 13,978 13,978 13,978
R² 0.756 0.750 0.754 0.757
16 that of bank ownership (IO_B). The extremely high coefficient of foundational ownership (IO_J) is biased due to the fact that there are only a few foundations in our dataset. In column (2), this type of owner, as well as industrial companies (IO_C) and hedge funds (IO_Y) are removed. Based on column (2), it seems that state ownership (IO_S) has the strongest positive effect on corporate governance, whereas venture capital (IO_V) is the only type of institution that manages to impair corporate governance. However, these findings should not be accepted without question, since they are highly dependent on the way owners are classified by Orbis. Table 2. Dependent variable: corporate governance index (CGVSCORE)
(1) (2) IO_A 0.051 (0.034) 0.053 (0.034) IO_B 0.240*** (0.017) 0.243*** (0.017) IO_C 0.173** (0.068) IO_E 0.073*** (0.026) 0.086*** (0.025) IO_F 0.059 (0.042) 0.066 (0.042) IO_J 10.274*** (2.074) IO_P 0.138*** (0.029) 0.140*** (0.029) IO_S 0.515*** (0.056) 0.527*** (0.056) IO_V -0.422* (0.224) -0.387* (0.224) IO_Y 0.008 (0.208) ECNSCORE 0.238*** (0.005) 0.239*** (0.005)
Country dummies Yes*** Yes***
Industry dummies Yes*** Yes***
Year dummies Yes Yes
N 13,978 13,978
R² 0.758 0.757
17 This runs counter to our theory that diversified investors are less concerned with corporate governance and opt for a “one-size-fits-all” policy. The bottom half of Table 3 concerns the median share. Here all of the columns state that the effect on corporate governance is greater for institutions that make smaller investments. At first glance this may also seem to refute theory; we expect greater involvement from owners with larger shares. However, it becomes clear once we realize that the independent variable refers to the (average) share held in all firms, not in a single one. Thus the conclusion is that it is good for corporate governance when investors with small stakes in other firms increase their share in the firm in question (because they will then devote most of their attention to this firm).
Table 3. Dependent variable: corporate governance index (CGVSCORE)
(1) (2) (3) (4) (5) IO_FQ1 0.099*** (0.018) 0.055*** (0.018) IO_FQ2 0.283*** (0.026) 0.105*** (0.026) IO_FQ3 0.319*** (0.022) 0.197*** (0.022) IO_FQ4 0.366*** (0.021) 0.280*** (0.021) ECNSCORE 0.254*** (0.005) 0.249*** (0.005) 0.242*** (0.005) 0.243*** (0.005) 0.241*** (0.005)
Country dummies Yes*** Yes*** Yes*** Yes*** Yes***
Industry dummies Yes*** Yes*** Yes*** Yes*** Yes***
Year dummies Yes Yes Yes Yes* Yes
N 13,978 13,978 13,978 13,978 13,978 R² 0.749 0.751 0.752 0.755 0.757 (6) (7) (8) (9) (10) IO_MQ1 0.531*** (0.046) 0.323*** (0.047) IO_MQ2 0.316*** (0.032) 0.103*** (0.034) IO_MQ3 0.238*** (0.024) 0.062** (0.025) IO_MQ4 0.237*** (0.019) 0.203*** (0.014) ECNSCORE 0.251*** (0.005) 0.251*** (0.005) 0.252*** (0.005) 0.244*** (0.005) 0.246*** (0.005)
Country dummies Yes*** Yes*** Yes*** Yes*** Yes***
Industry dummies Yes*** Yes*** Yes*** Yes*** Yes***
Year dummies Yes Yes Yes Yes* Yes
N 13,978 13,978 13,978 13,978 13,978
R² 0.751 0.750 0.750 0.755 0.757
18 influence on board functions, board structure and compensation policy (all illustrated by a high explanatory power, R²) and not as much influence on firm vision and strategy or shareholder rights. Within the three well-explained governance attributes we see a pattern: the largest (most frequent) institutions have the highest coefficients and thus wield the most influence. Table 4. Dependent variables: board functions (CGBF), board structure (CGBS), compensation policy (CGCP), firm vision and strategy (CGVS), shareholder rights (CGSR)
(1) CGBF (2) CGBS (3) CGCP (4) CGVS (5) CGSR IO_FQ1 0.073*** (0.019) 0.043** (0.019) 0.122*** (0.019) -0.234*** (0.026) 0.111*** (0.028) IO_FQ2 0.118*** (0.028) 0.151** (0.027) 0.148*** (0.027) -0.161*** (0.039) 0.114*** (0.042) IO_FQ3 0.109*** (0.023) 0.130*** (0.023) 0.049** (0.023) 0.181*** (0.034) 0.222*** (0.037) IO_FQ4 0.217*** (0.022) 0.249*** (0.022) 0.139*** (0.021) 0.130*** (0.032) 0.197*** (0.034) ECNSCORE 0.119*** (0.006) 0.087*** (0.005) 0.087*** (0.005) 0.526*** (0.007) 0.133*** (0.008)
Country dummies Yes*** Yes*** Yes*** Yes*** Yes***
Industry dummies Yes*** Yes*** Yes*** Yes*** Yes***
Year dummies Yes Yes Yes Yes* Yes
N 13,978 13,978 13,978 13,978 13,978
R² 0.737 0.754 0.741 0.506 0.307
6. Conclusion
With this paper, we have taken the literature on institutional ownership and corporate governance forward by taking it a step back. Rather than focusing on the actions of a subset of institutions, we have tried to gain an understanding of a world in which different kinds of institutions interact. Using a dataset unmatched in its completeness, comprising 3,575 firms from 54 countries over a period of 5 years (2010-2014), we have studied the effect of institutional ownership on firm-level corporate governance. In doing so, we have contrasted the popular qualitative factors of nationality and type with quantitative factors derived from the classic blockholder theory of Shleifer and Vishny (1986).
19 domestic institutions, though the latter do so as well and the difference between the two is not as pronounced as in previous studies. The current setup of the study also does not allow us to see whether the greater governance efforts of foreign institutional investors are due to a lack of business ties or due to other yet to be determined factors. In terms of owner types, we find that the optimal investor is surprisingly often of the ‘government’ type. However, we would argue that their governance efforts are the result of their quantitative characteristics and not something intrinsic or unique to governments.
The paper is subject to a fair number of limitations. Firstly, while the dataset is more complete than that used by any other study in this field, it is not of equally high quality. Orbis’ internal classification of types leaves something to be desired. Data availability varies per year and seems to be a lot better for U.S. institutions than for non-U.S. institutions. These issues limited the potential for the qualitative factors to explain the research question. Secondly, the statistical evidence in the thesis is not on the same level as that of rivaling papers. Our pooled OLS regression matches the findings of these papers well, but a regression with fixed effects left our ownership variables with no significance. We suspect this is due to the fact that the independent variable (corporate governance index) is highly unique to every firm, and thus when firm fixed effects are introduced, the other variables provide no added value.
The aforementioned are issues that future research in this field could address. In addition, we stimulate researchers to try and integrate the quantitative and qualitative features of owners rather than contrast them like we did. One way to do this might be by using interaction variables. Once the relations between institutional investors and firms have been properly developed, research could venture into case studies to find out the exact mechanisms that investors employ to exert corporate governance.
References
Aggarwal, R., Erel, I., Ferreira, M., Matos, P., 2011. Does governance travel around the world? Evidence from institutional investors. Journal of Financial Economics 100, 154-181.
Appel, I., Gormley, T., Keim, D., 2016. Passive investors, not passive owners. Journal of
Financial Economics, forthcoming. Available online at
dx.doi.org/10.1016/j.jfineco.2016.03.003.
The Boston Consulting Group, 2015. Global Asset Management 2015: sparking growth with go-to-market excellence. Available online at bcgperspectives.com.
Brav, A., Jiang, W., Partnoy, F., Thomas, R., 2008. Hedge fund activism, corporate governance, and firm performance. Journal of Finance 63, 1729-1775.
20 Davis, G., Kim, E., 2007. Business ties and proxy voting by mutual funds. Journal of Financial Economics 85, 552-570.
Del Guercio, D., Seery, L., Woidtke, T., 2008. Do boards pay attention when institutional investor activists
‘‘just vote no’’? Journal of Financial Economics 90, 84-103.
Edmans, A., 2014. Blockholders and corporate governance. Annual Review of Financial Economics 6, 23-50.
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.
Fich, E., Harford, J., Tran, A., 2015. Motivated monitors: the importance of institutional investors’ portfolio weights. Journal of Financial Economics 118, 21-48.
The Financial Times, 26 April 2015. Compulsory stewardship by passive managers moves closer. Available online at ft.com/cms/s/0/3917d0d0-e812-11e4-894a-00144feab7de.html. The Financial Times, 10 August 2015. Norway’s active attitude to corporate governance. Available online at ft.com/cms/s/0/54fea644-3f50-11e5-b98b-87c7270955cf.html.
The Financial Times, 1 May 2016. Norway’s oil fund to target high executive pay votes. Available online at ft.com/cms/s/0/aeb270ec-0f63-11e6-bb40-c30e3bfcf63b.html.
Het Financieele Dagblad, 31 March 2016. Maalt de indexbelegger wel om het lot van ‘zijn’ bedrijven? [Does the index investor care about the fate of ‘his’ companies?]. Available online at fd.nl/beurs/1145460/maalt-de-indexbelegger-wel-om-het-lot-van-zijn-bedrijven.
Gillan, S., Starks, L., 2000. Corporate governance proposals and shareholder activism: the role of institutional investors. Journal of Financial Economics 57, 275-305.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of Economics 118, 107-156.
Helwege, J., Intintoli, V., Zhang, A., 2012. Voting with their feet or activism? Institutional investors’ impact on CEO turnover. Journal of Corporate Finance 18, 22-37.
Iliev, P., Lowry, M., 2015. Are mutual funds active voters? Review of Financial Studies 28, 446-485.
Investment Company Institute, 2015. 2015 Investment Company Fact Book. Available online at icifactbook.org.
Jensen, M., Meckling, W., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305-360.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. Journal of Political Economy 106, 1113-1155.
The Motley Fool, 30 March 2015. A billionaire’s warning on index funds. Available online at
21 Parrino, R., Sias, R., Starks, L., 2003. Voting with their feet: institutional ownership changes around forced CEO turnover. Journal of Financial Economics 68, 3-46.
Shleifer, A., Vishny, R., 1986. Large shareholders and corporate control. Journal of Political Economy 94, 461-488.
Shleifer, A., Vishny, R., 1997. A survey of corporate governance. Journal of Finance 52, 737-783.
Smith, M., 1996. Shareholder activism by institutional investors: evidence from CalPERS. Journal of Finance 51, 227-252.
Tirole, J., 2001. Corporate governance. Econometrica 69, 1-35.
Appendices
Table A1. Overview of variables.
Name Description Scale Source CGVSCORE Corporate governance index 0 – 100 ASSET4
CGBF Board functions 0 – 100 ASSET4
CGBS Board structure 0 – 100 ASSET4
CGCP Compensation policy 0 – 100 ASSET4
CGVS Firm vision and strategy 0 – 100 ASSET4
CGSR Shareholder rights 0 – 100 ASSET4
IO_TOTAL Percentage of shares held by
institutional investors 0 – 100 Orbis
IO_DOMESTIC Percentage of shares held by domestic
institutional investors 0 – 100 Orbis
IO_FOREIGN Percentage of shares held by foreign
institutional investors 0 – 100 Orbis
IO_A / IO_B / IO_C / … Percentage of shares held by insurance
companies (A), banks (B), … 0 – 100 Orbis
IO_FQ1 / IO_FQ2 / … Percentage of shares held by institutional investors from the nth quartile in terms of firms held
0 – 100 Orbis
IO_MQ1 / IO_MQ2 / … Percentage of shares held by institutional investors from the nth quartile in terms of median share
0 – 100 Orbis
ECNSCORE Economic score 0 – 100 ASSET4 Table A2. Institutional ownership by country and by owner’s nationality.
Country N IOT IOD IOF
22 IT 124 20.2% 2.1% 18.1% ES 141 30.3% 7.4% 22.9% NL 119 44.7% 6.9% 37.8% SE 174 49.1% 30.9% 18.2% FI 112 36.4% 13.6% 22.8% BE 82 21.4% 1.0% 20.4% NO 88 40.8% 8.4% 32.4% DK 72 29.2% 7.8% 21.4% GR 43 16.1% 1.6% 14.5% AT 47 25.1% 4.2% 20.9% IE 51 48.9% 0.8% 48.1% PT 33 27.2% 7.4% 19.8% JP 1654 18.5% 6.0% 12.5% AU 1043 29.5% 12.2% 17.3% HK 431 21.5% 0.3% 21.2% SG 128 20.0% 1.4% 18.6% NZ 45 29.0% 5.3% 23.7% World 11212 52.7% 30.8% 21.9% Table A3. Institutional ownership by country and by owner’s type.
Country N IOA IOB IOC IOE IOF IOJ IOP IOS IOV IOY
23
SG 128 2.2% 8.3% 0.6% 2.3% 1.2% 0.0% 3.1% 2.0% 0.0% 0.3%
NZ 45 4.5% 9.3% 0.8% 5.5% 2.3% 0.0% 1.7% 4.9% 0.0% 0.0%
World 11212 6.6% 21.2% 1.9% 9.4% 4.5% 0.1% 6.1% 2.5% 0.2% 0.2%
Table A4. Institutional ownership by country and by owner’s frequency and median.
24 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% US CA UK FR DE CH IT ES NL SE FI BE NO DK GR AT IE PT JP AU HK SG NZ
