Electronic copy available at: http://ssrn.com/abstract=1117285
Block Ownership, Trading Activity, and Market Liquidity
Department of Finance
513 Cornell Hall
University of Missouri – Columbia
Columbia, MO 65211-2600
Tel: (573) 884-1562
Dennis Y. Chung
Department of Accounting
Simon Fraser University
Burnaby, BC V5A1S6
Tel: (778) 782-4355
Xuemin (Sterling) Yan
Department of Finance
427 Cornell Hall
University of Missouri – Columbia
Columbia, MO 65211-2600
Tel: (573) 884-9708
This Draft: March 2008
* We would like to thank Hendrik Bessembinder (the editor), John Howe, Sandra Mortal, Andy
Puckett, two anonymous referees, and seminar participants at the University of Missouri –
Columbia for helpful comments.
Electronic copy available at: http://ssrn.com/abstract=1117285
Block Ownership, Trading Activity, and Market Liquidity
We examine the impact of block ownership on the firm’s trading activity and
secondary-market liquidity. Our empirical results show that block ownership takes
potential trading activity off the table relative to a diffuse ownership structure and
impairs the firm’s market liquidity. These adverse liquidity effects disappear, however,
once we control for trading activity. Our findings suggest that block ownership is
detrimental to the firm’s market liquidity because of its adverse impact on trading
activity – a real friction effect. After controlling for this real friction effect, we find
little evidence that block ownership has a negative impact on informational friction.
Our results suggest that the relative lack of trading, and not the threat of informed
trading, explains the inverse relation between block ownership and market liquidity.
JEL Codes: G10; G32
Keywords: Block ownership; liquidity; bid-ask spreads; depths; adverse selection
Block ownership plays an increasingly important role in US capital markets. Dlugosz,
Fahlenbrach, Gompers, and Metrick (DFGM) (2006) find that block ownership increased from
21.7 percent of outstanding shares in 1996 to 25 percent in 2001 in their sample of over 1,900
relatively large firms.1 Given the pervasiveness of block ownership, it is important to understand
the role that blockholders play in such areas as firm valuation, corporate decision-making, and
secondary-market liquidity. Several previous studies have investigated the effect of block
ownership on firm valuation and corporate decision-making (e.g., Stulz (1988), Demsetz and
Lehn (1985), Morck, Shleifer, and Vishney (1988), Kole (1995)).2 In contrast, few studies to
date have examined the impact of block ownership on the firm’s market liquidity.3 The purpose
of this paper is to help fill this void by examining the real and informational friction effects of
There are two primary mechanisms through which block ownership can affect the firm’s
secondary-market liquidity: altering the firm’s trading activity or changing its information
environment. Stoll (2000) refers to the first mechanism as a real friction effect, and to the
second mechanism as an informational friction effect. Real friction is defined as “the real
resources used up” in the liquidity-provision process. These order processing and inventory
costs are highly sensitive to trading activity levels. Previous empirical research has shown that
trading volume is negatively related to bid-ask spreads (Benston and Hagerman (1974), Stoll
(2000)) and positively related to depths (Brockman and Chung (1999)). Therefore, block
ownership could affect the real friction component of liquidity by altering the firm’s trading
activity relative to a diffusely-owned firm. In particular, if blockholders trade significantly less
1 We follow prior literature to define blockholders as shareholders who hold five percent or more of a firm’s shares.
2 See Holderness (2003) for a review of the block ownership literature.
3 As discussed below, Heflin and Shaw (2000) is an exception.
than non-blockholders, the reduction in trading activity will increase real friction costs by
spreading fixed real costs over fewer trades.
The second mechanism through which block ownership can affect the firm’s market
liquidity is informational friction.4 Instead of reflecting the real costs of liquidity provision,
informational friction reflects the potential losses of trading against informed traders. Previous
research has shown that market makers widen bid-ask spreads and reduce depths in the presence
of informed traders (Copeland and Galai (1983), Kyle (1985), Glosten and Milgrom (1985)).
The impact of blockholders on the informational friction component of liquidity depends on their
proclivity to use private information while trading against the uninformed. If blockholders
(relative to non-blockholders) frequently trade on private information, then block ownership will
adversely affect market liquidity by increasing informational friction costs. On the other hand, if
legal, regulatory, or internal governance concerns effectively restrict such informed trading, then
block ownership will not adversely affect informational friction costs.
It is important to distinguish between real and informational friction effects because of
their direct implications for asset pricing, corporate governance, and regulation.5 With respect to
asset pricing, real friction will lead to lower prices and higher expected returns to offset the real
costs of trading (Stoll (2000), Amihud (2002)). The impact of informational friction on asset
4 The degree to which aggregate or specific types of blockholders are informed traders is an open empirical question.
Most of the literature dealing with block ownership and stock market returns focuses on the operational and
monitoring aspects of block ownership. Barclay and Holderness (1991) and Bethel, Liebeskind, and Opler (1998)
find abnormally positive returns associated with blockholder purchases. However, both studies reject the idea that
these positive returns are mostly due to superior information. Studies that focus on specific types of blockholders,
including institutions and insiders, have also yielded rather weak results. Bartov, Radhakrishnan, and Krinsky
(2000, p. 43) show that their “tests evaluating the validity of institutional holdings as a proxy for investor
sophistication yield only mixed results.” Bushee and Goodman (2007) find that changes in institutional ownership
are consistent with trading on private information. Perhaps the strongest evidence that some blockholders trade on
superior information comes from studies of insider trading. But even here, Lakonishok and Lee (2001) show that
insider trading abilities are limited to purchases (not sales) of relatively small companies.
5More generally, there is an important causal link between market liquidity and cost of capital (Amihud and
Mendelson (1986), Brennan and Subrahmanyam (1996), Datar, Naik, and Radcliffe (1998), and Easley, Hvidkjaer,
and O’Hara (2002)). Block ownership can indirectly affect the firm’s cost of capital through its impact on liquidity.
prices is less clear since informational friction mainly affects the distribution of wealth between
informed and uninformed investors (Stoll (2000)). With respect to corporate governance and
regulation, evidence of higher informational friction suggests that blockholders not only possess
superior information, but actively trade on it to the detriment of uninformed investors. In
contrast to real frictions, informational frictions could be reduced through stricter corporate
governance and regulatory provisions.
There is little consensus on the relative importance of real and informational frictions in
general, and even less consensus on the block ownership-liquidity relation. At least part of the
difficulty has to do with measurement. In this study, we follow Stoll’s (2000) suggestion that
informational friction can be thought of as (p. 1510) “the difference between total friction (such
as the quoted or effective spread) and real friction.” We disentangle real and informational
effects by first examining the impact of block ownership on the real costs of trading. These real
friction effects are directly related to the firm’s trading activity level (e.g., volume, turnover,
number of trades, trade sizes). After measuring the real friction effects, we then examine the
impact of block ownership on the firm’s market liquidity (e.g., spreads, depths, adverse
selection, price impact) while controlling for the known real friction effects.
Previous empirical studies that investigate block ownership often suffer from biased or
insufficient data. DFGM (2006) show that the widely-available Compact Disclosure database
contains a large number of mistakes that can lead to a significant overstatement in the level of
reported block ownership. One can avoid these Compact Disclosure biases by hand-collecting
block ownership data from source documents. This alternative, however, leads to a relatively
small number of observations and works against standardization and comparability across
studies. In addition, Chordia, Roll, and Subrahmanyam (2001) stress the importance of
analyzing trading activity and market liquidity over multi-year periods. We overcome these
shortcomings by using the multi-year, standardized database generated by DFGM (2006).
Another advantage of the DFGM (2006) database is that it includes three distinct
categories of block ownership: insiders, outsiders, and employee stock ownership plans (ESOPs).
Access to private information is not uniformly distributed across these three blockholder
categories. Inside blockholders are more likely to possess private information than outside
blockholders and ESOPs. Consistent with this argument, Lakonishok and Lee (2001) find that
managerial trades are more informative than large shareholder trades. In this study, we examine
the impact of each blockholder category on informational friction costs after controlling for real
friction costs. If block ownership adversely affects informational friction costs, then this
negative impact should be strongest for inside blockholders.
We divide our empirical analysis into two main sections. In the first section, we conduct
an analysis of the real frictions caused by block ownership by evaluating the impact of block
ownership on the firm’s trading activity including turnover, number of trades, and average trade
size. We find that block ownership significantly reduces the firm’s trading activity in the cross-
section relative to a diffuse ownership structure. Most of the reduced turnover caused by block
ownership results from a reduction in the number of trades, as opposed to changes in the average
In our second section, we investigate the blockholder-liquidity relation – a relation that
potentially involves both real and informational frictions. We analyze the impact of block
ownership on the firm’s market liquidity, including bid-ask spreads, depths, adverse selection
components, and price impact. We find that block ownership significantly increases the firm’s
quoted and effective bid-ask spreads, adverse selection costs, and Amihud’s (2002) price impact
measure. Block ownership also significantly reduces the firm’s market depth. These adverse
liquidity effects, however, either disappear or are reversed after controlling for blockholders’
direct impact on trading activity. Thus, our results suggest that block ownership impairs the
firm’s market liquidity by reducing trading activity and not by increasing asymmetric
information costs. Separate analyses of inside and outside block ownership yields results similar
to those for aggregate block ownership.
In a related study, Cao, Field, and Hanka (2004) disentangle real and informational
frictions by examining changes in the firm’s trading activity versus bid-ask spreads around IPO
lockup expirations. At the expiration of an IPO lockup, there is a large-scale entry of informed
(insider) traders into the market. Cao, Field, and Hanka (2004) distinguish between changes in
the firm’s trading activity (i.e., volume, number of trades, and average trade size) and changes in
the firm’s bid-ask spreads. They show that lockup expirations are associated with a significant
increase in the firm’s trading activity, but little if any change in bid-ask spreads. The implication
is that any increase in the firm’s informational friction is offset by a decrease in the firm’s real
Heflin and Shaw (2000) find that higher block ownership for both managers and non-
managers leads to wider spreads, thinner depths, and higher adverse selection costs. They
attribute these findings to informational frictions caused by differentially informed insiders
(blockholders) and outside investors. Although there are similarities between their study and
ours, there are also significant differences. They use a hand-collected sample of 260 firms
during 1988. Our sample includes 1,225 firms spanning a six-year period from 1996 to 2001.
While the primary focus of their study is on informational friction effects, we examine in
considerable detail the real friction effects of block ownership. The main difference, however, is
our evidence and conclusion that block ownership affects market liquidity principally through its
impact on real frictions, and not informational frictions.
In another related study, Rubin (2007) finds that institutional ownership affects the
liquidity-ownership relation more than inside ownership.6 He shows that while liquidity is
positively related to the level of institutional ownership (consistent with Gompers and Metrick
(2001) and Bennett, Sias, and Starks (2003)), it is negatively related to the concentration of
institutional ownership. However, he is unable to determine whether the inverse liquidity-
concentration relation is due to real friction effects or informational friction effects (or both).
In the next section, we describe the data used in our study and discuss our methods of
analysis. In section III, we present our empirical findings and analysis. In section IV, we
conclude our study.
II. Data and Methods of Analysis
A. Data and Sample Description
Our sample includes the six-year period beginning in 1996 and ending in 2001. The
block ownership data are from the blockholding database constructed by DFGM (2006).7 In
addition to reading original proxy statements, DFGM (2006) use a filtering process designed to
6 Rubin’s (2007) “trading hypothesis” corresponds to our real friction effect (i.e., liquidity differences are mainly
due to trading activity differences), and his “adverse selection hypothesis” corresponds to our informational friction
effects (i.e., liquidity differences are mainly due to asymmetric information differences).
7 DFGM (2006) construct their database beginning with firms covered by the Investor Responsibility Research
Center (IRRC). The IRRC database provides board of director information and governance details for roughly
1,500 firms each year, including the components of the S&P 500 and other large corporations listed in such
publications as Forbes, Fortune, and Businessweek. Companies with multiple-class stocks are eliminated, leaving
them with roughly 1,300 firms per year over the 1996-2001 sample period. Next, DFGM (2006) collect ownership
data from Compact Disclosure. They use only the data that Compact Disclosure obtains directly from proxy
statements since ownership data based on insider trading has been shown to be problematic for these purposes
(Anderson and Lee (1997)). They compare the ownership data from Compact Disclosure with original proxy
statements for every firm in the sample, even if Compact Disclosure shows no block ownership. This process
corrects Compact Disclosure’s two main biases – overlapping beneficial ownership and the treatment of preferred
determine actual beneficial ownership.8 We use the CRSP database to obtain stock returns, share
prices, number of shares outstanding, and trading volume. We use the Trade and Quote (TAQ)
database to construct the number of trades, average trade size, quoted and effective spreads,
quoted depth, and adverse selection components of the bid-ask spread.9 We exclude firms for
which trading and liquidity data are not available. In our main analysis, we include only firms
that are traded in NYSE or AMEX. We exclude Nasdaq firms and firms that switch exchanges
due to market microstructure differences.
Our main variable of interest, block ownership, includes three mutually-exclusive and
exhaustive categories; insiders, outsiders, and employee stock ownership plans (ESOPs). Inside
blocks include the blockholdings of officers, directors, and affiliated entities. DFGM’s (2006)
definition of an affiliated entity includes “any individual, trust, or company whose voting
outcome is partially influenced, but not completely controlled, by an officer or director of the
company.” Examples include shares owned by retired officers or directors, shares held in a trust
controlled by officers or directors, or shares owned by another business entity that has a specific
business relationship with the firm. Inside block ownership, therefore, includes all shares that
are directly or indirectly controlled by officers and directors of the firm. Outside blocks include
all blockholdings that are not held by insiders or through ESOPs.
In addition to the block ownership data, we obtain or construct various independent and
dependent variables from CRSP and TAQ databases. Our first set of dependent variables
correspond to real friction effects, including turnover (trading volume divided by the number of
8 These rules deal with such issues as the definition of beneficial ownership using voting power versus investment
power, the inclusion of shares that can be acquired within 60 days, and the treatment of temporary ownership from a
recent merger. See DFGM’s (2006) appendices A and B for additional details. See Chetty and Saez (2005) and
Cronqvist and Fahlenbrach (2007) for other studies that use the DFGM database.
9 We follow Chordia, Roll, and Subrahmanyam (2001, 2002) to purge the following trade and quote data: trades out
of sequence, trades and quotes before the open or after the close, quotes not originated on the primary exchange,
negative trades or quotes or spreads, quotes with spread greater than $4 or 20% of the midquote.
shares outstanding), number of trades, and trade size (share trading volume divided by the
number of trades). Our second set of dependent variables represent different aspects of market
liquidity, including relative quoted bid-ask spreads, relative effective bid-ask spreads, quoted
depths. We define the quoted bid-ask spread as the quoted ask price minus the quoted bid price
scaled by their midpoint. Our quoted depth measure is simply the number of shares available at
the inside quoted bid and ask prices. We define the effective bid-ask spread as two times the
absolute value of the difference between the transaction price and the quoted midpoint, scaled by
the quoted midpoint. All trading activity and liquidity variables are averages across all trading
days of each calendar year.
Our third set of dependent variables include the adverse selection component of the bid-
ask spread and Amihud’s (2002) price impact measure. We define and present three alternative
spread decomposition models in the next section. We follow Amihud’s (2002) definition of a
price-impact illiquidity measure by dividing the absolute value of daily stock returns by daily
Our control variables include price, volatility, firm size, a dummy variable for S&P 500
Index inclusion, and institutional ownership. Price is the average daily closing price, volatility is
the standard deviation of daily returns, and firm size is the share price times total number of
shares outstanding. We define S&P 500 as a dummy variable that takes the value of one if the
firm is included in the S&P 500 Index during the calendar year. We define institutional
ownership as the fraction of total shares outstanding held by 13F institutions at the end of the
previous calendar year, and obtain these values from Thomson Financial.
10 We exclude zero-volume days in our calculation of Amihud’s illiquidity measure. Zero-volume days represent
only about 0.1% of our sample. Therefore, this exclusion has a negligible effect on our results.
B. Adverse Selection Models
We use three methods to estimate the adverse selection component of the bid-ask spread
(i.e., Glosten and Harris (1988), Huang and Stoll (1997), and Lin, Sanger, and Booth (1995)) as
described below.11 We use these three approaches since recent research shows that different
models capture different aspects of adverse selection (Van Ness, Van Ness, and Warr (2001)).
First, we follow Glosten and Harris (1988) and estimate the following decomposition model:
u)V δ(Q) θ(ΔQΔP
where Pt = Pt – Pt-1 and Qt = Qt – Qt-1; Qt is the indicator for trade type at time t and takes a
value of +1 if the trade is a buyer-initiated transaction and –1 if the trade is a seller-initiated
transaction;12 Vt is the transaction size at time t; is the transitory component of the bid-ask
spread; is the adverse selection component of the bid-ask spread. As in Glosten and Harris
(1988), our estimated adverse selection coefficients are expressed in units of 1,000-share lots.
Second, we follow Huang and Stoll (1997) and estimate the following regression:
where Mt+1 = Mt+1 – Mt; Mt is the quoted bid-ask spread midpoint at time t; St-1/2 is the half
spread which is half the difference between the quoted ask and bid prices; Qt is the indicator for
11 We estimate each of the three adverse selection models using firm-month data. We then use the monthly averages
of adverse selection estimates over each year as our annual measure. We also estimate each model using firm-year
data and obtain similar results.
12 We classify transactions as buyer- or seller-initiated using the Lee and Ready (1991) algorithm: if a trade occurs
above (below) the mid-point of the prevailing quote, it is classified as a buyer- (seller-) initiated trade. If a trade
occurs at the mid-point of the prevailing quote, it is signed based on the tick test. Werner (2003) shows that the Lee
and Ready (1991) algorithm can misclassify a significant percent of market orders because buyer-initiated (seller-
initiated) market orders are often executed below (above) the bid-ask midpoint. This potential shortcoming,
however, does not affect most of our trading activity or liquidity variables including turnover, number of trades,
quoted spread, effective spread, quoted depth, and price impact. The misclassification of buyer- versus seller-
initiated transactions only affects our adverse selection cost estimates.
trade type at time t and takes a value of +1 if the trade is a buyer-initiated transaction and –1 if
the trade is a seller-initiated transaction; is the combined adverse selection and inventory
holding cost component of the bid-ask spread.
And third, we use Lin, Sanger, and Booth’s (1995) approach to estimate the following
bid-ask spread decomposition model:
e) (z λ ΔM
where Mt+1 = Mt+1 – Mt; Mt is the quoted bid-ask spread midpoint at time t; zt = Pt – Mt; is the
adverse selection component of the bid-ask spread; and e is a normally distributed error term.
III. Empirical Results
A. Summary Statistics and Univariate Tests
Table 1 reports summary statistics for the full sample (Panel A) and by year (Panel B).
The block ownership figures in Panel A show that, on average, 2.29 blockholders control 23.07
percent of company shares. When we break these figures down into blockholder types, we find
that inside blockholders control 5.36 percent of company shares on average; ESOPs own an
average of 1.24 percent of company shares; and outside blockholders have the highest company
ownership with an average of 16.47 percent. Turning to the real friction (i.e., trading activity)
measures, we show that the average number of trades per day is 390, the average daily share
volume is 640,000, and the average annual turnover is 104.90 percent. The average trade size is
Next, we present summary statistics for three liquidity measures, including relative quoted
spreads, relative effective spreads, and depths. The average relative quoted bid-ask spread is 0.36
percent. As expected, relative effective bid-ask spreads are uniformly lower than their quoted
spread counterparts. The average relative effective bid-ask spread is 0.24 percent. According to
Stoll (2000), these bid-ask spread figures capture both real friction costs and informational friction
costs. The average quoted depth is 4,960 shares. Our sample firms have an average market
capitalization of $7.70 billion, price of $34.86, annual volatility of 40.62 percent, and institutional
ownership of 57.53 percent.
In Panel B, we report average values for block ownership, trading activity, liquidity, and
control variables by year. Total block ownership and the number of blockholders have both
increased over our sample period. Much of this increase is due to the rise in outside block
ownership. Although there is some time variation in block ownership over our sample period, we
find considerably more time variation in real friction costs. The number of trades, trading volume,
and turnover increased substantially from 1996 to 2001, while average trade sizes decreased.
Consistent with Goldstein and Kavajecz (2000) and Bessembinder (2003), our liquidity measures,
including relative quoted and effective spreads and quoted depths, decreased over our sample
period. The decrease in spreads implies that stocks became more liquid, while the decrease in
depths implies the opposite. Investors could buy and sell at lower costs, but for fewer shares.
With the exception of average price, our control variables generally increased in value over our
In Table 2, we examine the relation between block ownership and the firm’s trading
activity and market liquidity by using a univariate portfolio approach. We divide all sample stocks
with non-zero block ownership into quintile portfolios (from Q1-low to Q5-high) every year. We
group sample stocks with no block ownership into a separate portfolio (Q0-none). In the two
rightmost columns, we perform difference-in-means tests between high and zero block ownership
portfolios (Q5-Q0), and between high and low block ownership portfolios (Q5-Q1).
We find that trading activity declines significantly as we move from the zero or lowest
block ownership portfolios (Q0 or Q1) to the highest block ownership portfolio (Q5). The average
number of trades decreases monotonically from 860 (520) for the zero (lowest) block ownership
portfolio to 130 for the highest block ownership portfolio. The differences in the number of trades
(Q5-Q0) and (Q5-Q1) are highly significant with t-values in excess of 13. Share volume displays
the same pattern, decreasing monotonically from 1.41 (0.86) million for the zero (lowest) block
ownership portfolio to 0.21 million for the highest block ownership portfolio. We find a similar
overall pattern across block ownership portfolios for turnover; that is, the highest block ownership
portfolio has lower turnover than the zero and lowest block ownership portfolios. But unlike the
monotonic decline in the number of trades, turnover displays an inverted U-shape pattern (with
the right-side foot lower than the left-side foot). Overall, these univariate tests are consistent with
our hypothesis that block ownership generally reduces trading activity.
Next, we examine the relation between block ownership and the firm’s market liquidity.
We find a monotonically increasing relation between block ownership and relative bid-ask
spreads, both quoted and effective. The relative effective bid-ask spread, for example, increases
from 0.16 (0.19) percent in the zero (lowest) block ownership portfolio to 0.34 percent in highest
block ownership portfolio. We find a similar, though inverse, monotonic relation between block
ownership and quoted depth. The average depth decreases from 6,220 (5,590) shares in the zero
(lowest) block ownership portfolio to 3,760 shares in the highest block ownership portfolio. These
results suggest that block ownership impairs market liquidity. However, the extent to which these
liquidity patterns are caused by real friction effects (e.g., trading volume, number of trades)
remains an open empirical question. Lastly, consistent with our results on bid-ask spreads, we
find a monotonically increasing relation between block ownership and price impact, as well as
between block ownership and each of our adverse selection estimates.
B. Real Friction Effects: Turnover, Number of trades, and Trade Size
In this section, we examine the impact of block ownership on real friction effects (i.e.,
trading activity) in a multivariate setting, holding constant firm size, price, volatility, S&P 500
Index inclusion, and institutional ownership (IO). We fit the following cross-sectional regression
ε BlockβIOβ 500&Sβ
β cap) log(MarketββActivity Trading
Our real friction trading activity variables (Trading Activity) include turnover, number of trades,
and trade size. Our block ownership variable (Block) refers to firm i’s aggregate block ownership.
We use lagged values for block ownership and institutional ownership and contemporaneous
values for all other variables in our regressions.13 Specifically, market capitalization, price, and
trading activity variables are averages over the current year. Volatility is estimated using daily
returns during the same year. Institutional and block ownership variables are measured at the end
of the previous year. Parallel to regression model (4), we replace Block with its constituent parts
to analyze the incremental effects of inside, outside, and ESOP block ownership. We normalize
the block ownership variables by their respective cross-sectional standard deviations in each year.
This allows us to make meaningful comparisons across the coefficients. We estimate the cross-
sectional regression model (4) using time-series averages for both dependent and independent
variables. Specifically, for each firm we first compute its time-series averages for all variables in
13 We also estimated our regressions using contemporaneous block ownership values. The results based on
contemporaneous values are qualitatively similar to those reported herein.