Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Investor Competition and the Pricing of
Information Asymmetry
Brian Akins [email protected]
Jeffrey Ng
Rodrigo Verdi [email protected]
Abstract
Whether the information environment affects the cost of capital is a fundamental question in accounting and finance research. In this paper, we study the role of competition among informed investors on the pricing of information asymmetry and its implications for the pricing of information quality. Relying on theories on competition among informed investors and on the pricing of information asymmetry, we hypothesize and find that the pricing of information asymmetry and the pricing information quality decrease with more competition among informed investors. Our results suggest that competition among informed investors has an important effect on how the information environment affects the cost of capital.
First draft: September 2008 Current version: November 2009
JEL classification: G12, G14 Key Words: Information Risk, Information quality, Asset Pricing.
Data Availability: All data is publicly available on WRDS. ___________________________________ All authors are at MIT Sloan School of Management, 50 Memorial Drive E52-325G, Cambridge, MA 02142. We appreciate comments from David Aboody, Paul Asquith, Judson Caskey, Gus DeFranco, Jack Hughes, S.P. Kothari, Christian Leuz, Joe Weber, Paul Zarowin (the editor), two anonymous referees, and seminar participants at the MIT Sloan School of Management (Accounting and Finance workshops) and at the University of California at Los Angeles. We would like to thank Brian Bushee for providing us access to his institutional investor database. We would also like to thank Jefferson Duarte and Lance Young for providing us estimates of the components of PIN.
1
I. Introduction Whether information asymmetry among investors affects the cost of capital is an
important issue in the theoretical (e.g., Diamond and Verrecchia 1991; Easley and O’Hara 2004;
Hughes et al. 2007) and empirical literature (e.g., Brennan and Subrahmanyam 1996; Easley et
al. 2002; Duarte and Young 2009; Mohanram and Rajgopal 2009). In this paper, we study the
role of competition among informed investors on the pricing of information asymmetry. Our key
finding is that the pricing of information asymmetry decreases when there is more competition
among informed investors. This finding is important because it suggests that in the presence of
information asymmetry, more competition among informed investors can lower the cost of
capital. Further, as we describe below, it has implications for a large literature that investigates
the pricing of information quality (e.g., Botosan 1997; Leuz and Verrecchia 2000; Francis et al.
2004, 2005).
The intuition for our hypothesis that the pricing of information asymmetry decreases in
competition among informed investors is briefly described as follows (we describe related
theories and formally develop our hypothesis in Section 2). Theories on investor competition
show that more competition among informed investors results in private information being
incorporated into prices more rapidly, i.e., more competition makes prices more efficient (Foster
and Viswanathan 1993, 1996; Holden and Subrahmanyam 1992, 1994). We argue that
competition among informed investors can then affect the pricing of information asymmetry due
to two reasons: (i) it could reduce trading costs arising from price protection by market makers
against information asymmetry (Kyle 1985; Admati and Pfleiderer 1988; Diamond and
Verrecchia 1991), and (ii) it could reduce the information risk uninformed investors bear when
trading against informed investors (Easley and O’Hara 2004).
2
To measure competition among informed investors, we follow prior literature and
consider institutional investors as informed investors (Arbel and Strebel 1983; Sias and Starks
1997; Bartov et al. 2000; Jiambalvo et al. 2002). Using data on total institutional investor
ownership for each firm, we construct measures of competition based on (i) the number of total
institutional investors, (ii) the percentage of outstanding shares held by total institutional
investors, and (iii) a Herfindahl index of competition that captures both the level and the
distribution of total institutional ownership. Further, recognizing that transient institutional
investors (compared to quasi-indexers and dedicated institutional investors) are the ones most
likely to trade on information (Bushee 1998; Ke and Petroni 2004; Ke and Ramalingegowda
2005), we also construct analogous measures using data on transient institutional investor
ownership.
To measure information asymmetry, we use the information asymmetry component of
bid-ask spreads and the probability of informed trading (PIN) based on decomposition models
developed by Glosten and Harris (1988) and Duarte and Young (2009), respectively. We include
the non-information asymmetry components of spreads and PIN as control variables in our
empirical analyses to increase the confidence that our findings are driven by information
asymmetry. In untabulated analyses, we find that our inferences are unchanged if we use total
spreads and PIN.
To examine the role of competition among informed investors in the pricing of
information asymmetry, we use standard asset pricing regressions. We find significant evidence
that the pricing of information asymmetry decreases in the extent of competition among
informed investors. For instance, the difference in the pricing of the information asymmetry
component of spread between the most competitive quintile of firms and the least competitive
3
quintile ranges from 0.70% to 1.23% per month, depending on the competition measure used.
The results with the information asymmetry component of PIN are in the same direction,
although the economic significance is smaller. Specifically, the differential pricing of
information asymmetry across competition among informed investors ranges from 0.22% to
0.25% per month.
In further analyses, we find that the evidence for the differential pricing of information
asymmetry conditional on competition is the strongest for competition among transient investors
and the weakest (and often insignificant) for competition among dedicated investors. This
evidence is consistent with our conjecture that competition among investors who trade actively
on information (compared to competition among investors who do not trade actively on
information) is likely to have a greater effect in mitigating the pricing of information asymmetry.
To address concerns that our measures of competition might be capturing broader aspects of the
trading environment that are not related to information asymmetry, we also run tests that control
for the cross-sectional variation in the broader trading environment. The results from these tests
indicate that the differential pricing of information asymmetry conditional on competition is
robust to controlling for share turnover, trading volume, and return volatility.
Finally, we examine whether investor competition influences the pricing of information
quality. This literature argues that information quality is priced because poor information quality
is associated with higher information asymmetry and information asymmetry is priced (e.g.,
Botosan 1997; Francis et al. 2004, 2005). To measure information quality, we use accruals
quality and earnings smoothness because these measures have been recently used to examine the
pricing of information quality (Francis et al., 2004, 2005; Core et al. 2008; McInnis 2009). We
also use annual report readability (FOG) developed by Li (2008) as another measure of
4
information quality. We find consistent evidence that the pricing of information quality
decreases when there is more competition among informed investors. This result adds support to
the argument that one reason that information quality is priced is because of information
asymmetry
This study contributes to the literature in a number of ways. First, it draws upon the
theoretical literature to make predictions about the effect of competition among informed
investors on the pricing of information asymmetry. We show that the pricing of information
asymmetry decreases with competition among informed investors, and that the effect is
economically important. While the idea of competition among informed investors has been
discussed in the theoretical literature, to the best of our knowledge, no prior empirical paper has
investigated the outcomes of such competition.
Second, we extend the prior literature that has empirically investigated the pricing of
information asymmetry and information quality. Recent studies provide mixed evidence on
whether information asymmetry/quality is priced (e.g., Easley et al. 2002; Francis et al. 2005;
Core et al. 2008; Duarte and Young 2009; Mohanram and Rajgopal 2009). We show that the
extent of the competition among informed investors has an important role in determining
whether information asymmetry is priced. Stated differently, information asymmetry/quality is
more likely to be priced in trading environments with less competition among informed
investors. Finally, our consistent findings with proxies for information asymmetry and
information quality reinforce the idea that information asymmetry is one mechanism linking
information quality to the cost of capital. By doing so we address one claim in Leuz and
Wysocki (2008, p. 30) that “At present, however, the literature has primarily focused on
5
establishing the link between disclosure and the cost of capital and has provided relatively little
evidence on the mechanism.”
Our paper is related to a concurrent working paper by Armstrong et al. (2009). While
both papers study the effect of competition on the pricing of information asymmetry, there are
some important differences. At the conceptual level, Armstrong et al. study the role of
competition among all investors. In contrast, as our paper relies on theories on competition
among informed investors, we study the role of competition among informed investors in
mitigating the pricing of information asymmetry. Consequently, these papers also differ at the
empirical level by using a non-overlapping set of proxies for competition. Another unique
feature of our paper is that we compare the relation between the pricing of information
asymmetry and competition among different types of informed investors. An interesting result is
that more competition among investors who actively trade base on information has a greater
effect in mitigating the pricing of information asymmetry.
The remainder of the paper is organized as follows. Section II develops our hypothesis.
Section III describes our research design. Sections IV and V present our results on the pricing of
information asymmetry and information quality, respectively. Section VI concludes.
II. Hypothesis development
In this section we develop our hypothesis of the role of competition among privately
informed investors on cost of capital. A seminal theoretical result obtained by Kyle (1985) is that
private information is incorporated into price over time due to the informed trades in the case of
an information monopolist. Several theory papers extend Kyle by introducing more than one
informed investor so that there is competition among informed investors. A key finding is that
6
that competition among informed investors typically results in private information being
incorporated into prices more rapidly and, thereby, increases price informativeness. This finding
has been proven in the case of multiple risk-neutral homogeneously informed traders (Admati
and Pfleiderer 1988; Holden and Subrahmanyam 1992; Foster and Viswanathan 1993), of risk-
averse homogeneously informed traders (Holden and Subrahmanyam 1994), of heterogeneously
informed traders (Admati and Pfleiderer 1988; Foster and Viswanathan 1996), among others. We
build on this literature to hypothesize that the pricing of information asymmetry will decrease
with competition among informed investors.
To set the stage for linking competition with the pricing of information asymmetry, first
consider a Kyle (1985) model in which there are a single asset, a single risk neutral privately
informed trader, competitive market makers, and noise traders trading for liquidity reasons. The
asset would be priced in a manner that imposes trading costs, implying an effect on expected
return, i.e., the cost of capital. In this type of model, the revelation of information into the public
domain affects the cost of capital through a reduction in information asymmetry (Diamond and
Verrecchia 1991). Most importantly, an increase in the number of informed traders results in
competition among those traders that would reduce trading costs and thus lower the cost of
capital (Admati and Pfleiderer 1988). Thus, the implication from this theory is that the
information asymmetry component of trading costs would decrease with competition among
informed investors.
Next, consider a competitive market composed of privately informed investors,
uninformed investors, and a finite number of assets as in Easley and O’Hara (2004). In this
model, informed investors use their information advantage to trade with uninformed investors
and hold portfolios more heavily weighted to stocks with positive private information and
7
weighted against stocks with negative private information. The information asymmetry increases
the risk to the uninformed investors, who cannot adjust their portfolios to account for private
information. In equilibrium, information asymmetry is priced to reflect the risk to the
uninformed investors. Once again, to the extent that more competition among informed investors
helps to make prices more informative and mitigate the information disadvantage of the
uninformed investors (Holden and Subrahmanyam 1992, 1993; Foster and Viswanathan 1993,
1996), the pricing of information asymmetry is expected to decrease with more competition.
In light of the above theories, our conjecture is that, ceteris paribus, more competition
among informed investors reduces the impact of information asymmetries on the cost of capital
due to (potentially) two reasons: (i) it could reduce trading costs arising from price protection by
market makers due to information asymmetry (Kyle 1985; Admati and Pfleiderer 1988; Diamond
and Verrecchia 1991), and (ii) it could reduce the information risk uninformed investors bear
when trading against informed investors (Easley and O’Hara 2004).
Hence, our hypothesis, stated in alternative form, is:
H1: The pricing of information asymmetry decreases in competition among informed investors.
III. Research Design
To test our hypothesis, we use cross-sectional asset pricing regressions similar to those
employed in prior research to investigate the effect of information asymmetry on the cross-
section of returns (e.g., Easley et al. 2002; Duarte and Young 2009). In particular, the typical
cross-sectional regression specification used to test for the pricing of information asymmetry is
as follows:
Ri,t+1 = α + Σβj Controlj,t + λ Information Asymmetryi,t + εi,t +1 (1)
8
where Rt+1 is monthly excess return during the 12-month period in year t+1 (in the event of a
delisting, a firm’s delisting return, when available from CRSP, is used as the monthly return),
Control is a set of control variables that are expected to be associated with expected returns and
Information Asymmetry is a proxy for information asymmetry.
To examine whether the pricing of information asymmetry is conditional on the extent of
competition among informed investors, we modify Eq. (1) as follows:
Ri,t+1 = α + Σβj Controlj,t + λ1 Competitioni,t + λ2 Information Asymmetryi,t
+ λ3 Competitioni,t x Information Asymmetryi,t + εi,t+1 (2)
where Competition is a proxy for competition among informed investors. To reduce the effect of
outliers and to ease exposition, we rank Competition into quintiles using the distribution of its
value within the year and then scale the quintile rank so that it ranges from zero to one.
In the above regression, λ2 indicates the pricing of information asymmetry of firms in the
least competitive quintile and λ3 indicates the incremental pricing of information asymmetry as
one moves from the bottom to the top quintile of competition. Based on our hypothesis that the
pricing of information asymmetry is decreasing in the competition among informed investors, we
expect λ3 to be negative. To mitigate cross-sectional dependence in the regression residuals, we
follow the prior literature and estimate Fama-MacBeth (1973) regressions (e.g., Fama and
French 1992; Easley et al. 2002). Specifically, we first run cross-sectional regressions for each
month in the sample. Each reported coefficient is the average of the monthly coefficients. The t-
statistic for each reported coefficient is obtained by dividing the coefficient by the standard error
of the monthly coefficients.
Measures of competition among informed investors
9
In order to test our hypothesis, we need a measure of informed competition. While the
prior theoretical literature has examined issues related to informed competition, we are unaware
of prior attempts in the empirical literature to measure informed competition. Before we describe
our measures, we note that our measures rely on the common assumption in the literature that
institutional investors, as opposed to individual retail investors, are more likely to be the class of
informed investors (e.g., Arbel and Strebel 1983; Sias and Starks 1997; Bartov et al. 2000;
Jiambalvo et al. 2002). For example, Sias and Starks (1997) provide evidence that the returns on
portfolios dominated by institutional investors lead the returns on portfolios dominated by
individual investors, consistent with the hypothesis that institutional trading increases the speed
with which prices reflect market-wide information.
Our first measure of informed competition is the number of institutional investors holding
the firm’s stock (#Inst). This measure is motivated by the theories that examine the effect of
competition among informed investors (e.g., Admati and Pfleiderer 1988; Holden and
Subrahmanyam 1992; Foster and Viswanathan 1993). These theories use the number of informed
investors to represent the extent of the informed competition, where a greater number of
informed investors indicates more informed competition. The second measure of informed
competition is the proportion of institutional stock ownership, i.e., percentage of outstanding
shares held by institutional investors (%Inst). More informed competition might be expected
when there is a higher percentage of institutional investor ownership in a stock because a higher
percentage of institutional investor ownership indicates a higher level of sophistication in the
investor base (Bartov et al. 2000; Jiambalvo et al. 2002).1
1 The use of the number of informed investors (as opposed to the fraction of investors) in theory models is for mathematical tractability. A similar prediction, however, can be developed with the fraction of informed investors. For instance, Edmans and Manso (2009) show that the percentage of informed ownership is negatively associated
10
Our third measure of competition among informed investors captures both the level and
the distribution of total institutional ownership into a Herfindahl index. This index has been used
to measure competition in a variety of settings such as product market competition within the
industry (e.g., Comment and Jarrell 1995; Gande et al. 1999). In the context of competition in the
trading environment of a stock, we propose a measure of the amount of competition in stock i,
HerfInst, as follows:
2,
1
1N
i ji
j i
InvestorHerfInst x
Investors=
⎛ ⎞= − ⎜ ⎟
⎝ ⎠∑
(3)
where Investori,j is the number of shares held by institutional investor j in stock i, Investorsi is the
total shares held by all institutional investors of stock i, and N is the total number of institutional
investors in stock i. Following the standard interpretation of the Herfindahl index, highly
concentrated holdings indicates less competition in the trading environment. Thus, to ease
exposition, we multiply the Herfindahl index by minus one so that a higher value of HerfInst
signifies that there is more competition in the trades of stock i among institutional investors. By
construction, HerfInst could range from -1 to 0, with a number closer to 0 indicating more
competition. When there are no institutional investors for a firm, we assume that there is no
competition among informed investors and assign a value of -1 to the Herfindahl index.
As discussed above, we use institutional investors as a proxy for informed competition.
Prior research, however, has shown that certain types of institutional investors are more likely to
trade on information (e.g., Grinblatt and Keloharju, 2000; Ke and Petroni, 2004; Ke and
Ramalingegowda, 2005). For example, a commonly used institutional investor classification is
the division of institutional investors into transient investors, dedicated investors, and quasi-
with the Kyle’s lambda if one allows the percentage holdings to be correlated with the precision of the information impounded in prices when informed investors trade.
11
indexers that was developed by Bushee (1998). Ke and Petroni (2004) and Ke and
Ramalingegowda (2005) find that transient institutional investors (i.e., institutional investors
who hold small stakes in numerous firms and trade frequently in and out of stocks) trade on
information to make profitable trades. Ke and Petroni (2004) find that transient institutional
investors have information that allows them to predict a break in a string of consecutive quarterly
earnings increases and thereby avoid the economically significant negative stock price response
associated with the break announcement. Ke and Ramalingegowda (2005) provide evidence that
transient institutional investors trade on knowledge about the post-earnings announcement drift.
Hence, in the same vein as our earlier measures based on total institutional investor ownership,
we measure the number of transient institutional investors, #Tran, and percentage of transient
institutional investor ownership, %Tran, and the Herfindahl index for competition among
transient investor, HerfTran.
We construct the above measures of informed competition by using the institutional
investor database provided the author of Bushee (1998), who, in turn, creates his variables using
data on shares held by institutional investors from the CDA/Spectrum Institutional (13f)
Holdings (s34) database available from Thomson Reuters. Specifically, his dataset provides
information on the institutional investors of each firm, as well as the classification of the investor
as transient, dedicated, or quasi-indexing.
Measures of information asymmetry
The extensive empirical literature has used various measures of information asymmetry,
with the bid-ask spread and probability of informed trading (PIN) being the more common
measures (e.g., Copeland and Galai 1983; Glosten and Milgrom 1985; Lee et al. 1993; Easley et
12
al. 2002). One issue with these measures is that they also capture other components, specifically
order processing and inventory holding costs for spread and liquidity for PIN, in addition to
information asymmetry. We thus follow prior literature and use the information asymmetry
component of spread and PIN to better capture our construct of interest – information asymmetry
(e.g., Glosten and Harris 1988; Duarte and Young 2009).
We measure the information asymmetry component of the bid-ask spread for NYSE,
AMEX, and NASDAQ firms for the period from 1983 to 2004 using the model of price
formation developed in Glosten and Harris (1988). The empirical implementation of this model
is often used in the literature to estimate the information asymmetry component of the bid-ask
spread (e.g., Brennan and Subrahmanyam 1995, 1996; Verrecchia and Weber 2008).
Specifically, the bid-ask spread is decomposed into an information asymmetry component
(IASpread) and a non-information asymmetry component (NIASpread). We use IASpread as our
first proxy for information asymmetry. In addition, in all tests utilizing this measure we also
include the NIASpread as a control variable to ensure that our results are not driven by the non-
information asymmetry component of the bid-ask spread. More details of the computation of
these measures are presented in Appendix A.
Our second measure of information asymmetry is the information asymmetry component
of PIN developed by Duarte and Young (2009). Duarte and Young (2009) decompose PIN into
an information asymmetry component, adjusted PIN (AdjPIN), and a non-information
asymmetry component, the probability of symmetric order flow shocks (PSOS). They motivate
this decomposition by noting that the sequential trade model on which PIN is based attributes
abnormal trading to private information, but this trading could also result from liquidity shocks.
To address this concern, Duarte and Young (2009) extend the model to account for non-
13
information-based liquidity shocks. We use AdjPIN as our second proxy for information
asymmetry. In addition, as with spreads, when using AdjPIN we control from the non-
information asymmetry component of PIN, the probability of symmetric order-flow shocks
(PSOS). These measures, available for NYSE and AMEX firms from 1983 to 2004, were
provided to us by Duarte and Young. We refer interested readers to Duarte and Young (2009) for
more information underlying the construction of AdjPIN and PSOS.
Other control variables
When discussing our key measures, we have indicated that we add non-information-
asymmetry component of spread and PIN as control variables. In addition, we follow asset
pricing literature and include beta (Beta), market capitalization (Size), and the book-to-market
ratio of equity (BTM) as additional control variables (e.g., Fama and French 1992; Easley et al.
2002). To calculate Beta, we follow Easley et al. (2002) and begin by estimating rolling five-year
Dimson (1979) betas for each firm. For estimation, we require a minimum of 24 monthly return
observations. We then regress these returns on lagged and contemporaneous market returns to
correct for potential bias resulting from non-synchronous trading. Firms are then sorted by these
pre-ranking betas into 40 portfolios at the beginning of each year and rolling five-year Dimson
betas are calculated for each portfolio. Each firm is assigned the beta from its portfolio for that
year. Size is the natural log of the market value of equity measured at the end of the calendar
year. To measure book-to-market, BTM, we obtain the book value of equity for the fiscal year
ending at least three months prior to the calendar end (this ensures that book value is publicly
available at the end of the calendar year). We then divide the book value of equity by the market
value of equity at the end of the calendar year and take the natural logarithm of this ratio.
14
IV. Pricing of information asymmetry
Sample description
Table 1 presents, for each calendar year from 1983 to 2004, the number of firms for
which we are able to compute our measures of information asymmetry. The sample consists of
59,723 (36,071) firm-year observations with available data to compute IASpread (AdjPIN). The
sample with IASpread is larger because spread data is available to compute IASpread for
NASDAQ firms starting from 1987, while the AdjPIN file is comprised of firms from NYSE and
AMEX.
[Insert Table 1 here]
Summary statistics and the correlations for the variables used in our study are provided in
Table 2. Panel A presents the descriptive statistics for our sample. First, we present statistics for
our measures of the information asymmetry and non-information asymmetry components of
spread and PIN. For the spread-based measures, the sample consists of NYSE, AMEX, and
NASDAQ firms for the period from 1983 to 2004 and there are a total of 59,723 firm-year
observations. For the PIN-based measures, the sample consists of NYSE and AMEX, and firms
for the period from 1983 to 2004 and there are a total of 36,071 firm-year observations. The
average information asymmetry component of spread is 0.44%, whereas the average fixed
component is 1.75%. The average probability of information-based trades is 0.18, whereas the
average probability of trades originating because of symmetric order flow shocks is 0.29.
Moving on to our investor competition measures, we find that on average, there are 77.61
(22.93) institutional investors (transient institutional investors) per firm and these investors hold
35% (10%) of the outstanding shares of the firms. The mean investor competition index for
15
institutional investors (transient institutional investors) is -0.19 (-0.44). The distributions for
some of our competition proxies are highly skewed. For example, the number of institutional
investors per firm has a median of 32.25, compared to a mean of 77.61. Given that we convert
the continuous competition measures into ranked measures in our regressions, skewness is
unlikely to be a significant concern. The average market beta is 1.26, indicating that the
systematic market risk of the sample is slightly higher than the market. The average market
capitalization and book-to-market of equity are $19.12 billion and -0.71, respectively.
Panel B presents Pearson correlations above the main diagonal. All the correlations are
significant at the 1% level. IASpread has a positive and significant correlation of 0.32 with
AdjPIN. We also find that the non-information asymmetry components of spread and PIN,
NIASpread and PSOS, have a positive and significant correlation of 0.31. Next, we examine the
properties of our competition measures, which are all positively and significantly correlated with
each other. For example, #Inst has a correlation of 0.51, 0.41, 0.86, 0.33, and 0.55 with %Inst,
HerfInst, #Trans, %Trans, and HerfTrans, respectively.
[Insert Table 2 here]
Cross-sectional asset pricing tests
The next two tables present the regression results of the cross-sectional asset pricing tests
of information asymmetry. Table 3 (Table 4) uses IASpread (AdjPIN) as the measure of
information asymmetry. For comparability across different measures, when we make inferences
about the economic significance from the coefficients of interest, we assume a one standard
deviation difference in the information asymmetry (standard deviations are obtained from Table
2) and examine how this difference translates into expected returns.
16
In the first column of Table 3, we examine whether information asymmetry, as measured
by IASpread, is priced on average. The results indicate that investors expect higher returns when
the information asymmetry component of spread is higher. Specifically, the statistically
significant coefficient on IASpread of 0.25 indicates that a one standard deviation difference in
the information asymmetry component of spread between two stocks translates into a difference
in required return of 0.34% (=1.35 x 0.25) per month (or 4.06% per year).
In the remaining columns, we examine whether the pricing of information asymmetry
decreases in the extent of competition among informed investors. As noted earlier, the
coefficient of interest is λ3, the coefficient on the interaction term between IASpread and a proxy
for competition among informed investors. This coefficient is statistically significant in all the
columns, although only marginally so when using HerfInst as a competition proxy. Specifically,
the coefficients range from -0.52 to -0.91, depending on the measure of competition. This
indicates that the expected return for a one standard deviation difference in information
asymmetry in the most competitive quintile ranges from -0.70% (1.35 x -0.52) to -1.23% (1.35 x
-0.91) per month less than that in the least competitive quintile. This result is consistent with our
hypothesis that the pricing of information asymmetry decreases in the extent of the competition
among informed investors.
We note that the pricing of the non-information-asymmetry component of spread,
NIASpread, also decreases when there is more competition among informed investors.
Conceptually, this component captures the fixed costs per share, specifically inventory holding
cost and order processing cost per share, of market-making. A possible explanation for lower
fixed costs per share is that the trading activities of institutional investors, which is presumably
17
higher when there is more competition among them, helps to lower spread the fixed costs over
more traded shares and thus, lowers fixed costs per share.
[Insert Table 3 here]
In Table 4, we repeat the analyses with AdjPIN as the measure of information asymmetry.
In the first column, the results indicate that AdjPIN is not priced whereas the non-information
asymmetry component of PIN, PSOS is priced. These results are consistent with those in Duarte
and Young (2009). The remaining columns present the regression results conditional on the
extent of competition among informed investors. All of the coefficients on the interaction term
between AdjPIN and Competition are negative, although the results are only statistically
significant when percentage ownership of institutional and transient institutional investors (%Inst
and %Trans) are used to proxy for competition among informed investors. Hence, the evidence
that the pricing of information asymmetry (as proxied by AdjPIN) decreases when there is more
competition among informed investors is not as strong as the results with IASpread. We note that
there is almost no evidence that the pricing of PSOS, which is the liquidity component of PIN,
varies with competition.
In terms of economic significance, the -2.73 coefficient on the interaction term between
AdjPIN and %Inst indicates a monthly differential in the required rate of return of 0.25% per
month (2.95% per year) between stocks in the least and most competitive quintiles for a one
standard deviation difference in information asymmetry. The -2.39 coefficient on the interaction
term between AdjPIN and %Trans indicates a monthly differential in the required rate of return
of 0.22% per month (2.58% per year) between stocks in the least and most competitive quintiles
for a one standard deviation difference in information asymmetry.
[Insert Table 4 here]
18
Overall, the results from the cross-sectional asset pricing results in Tables 3 and 4 are
consistent with our hypothesis that the pricing of information asymmetry decreases with
competition among informed investors.
Analysis across Different Types of Institutional Investors
As discussed in Section 3, we consider transient institutional investors as type of
institution investors more likely to actively trade on information and construct measures of
competition among informed investors based on their stock ownership. In this section we discuss
the results for measures of competition constructed based on other types of institutional
investors.
Bushee (1998) uses the ownership concentration and trading activity to classify
institutional investors into three categories following Porter (1992): transient institutional
investors, dedicated institutional investors, quasi-indexers. As noted early, transient institutional
investors have small holdings in many firms and trade frequently. Dedicated institutional
investors have larger, more concentrated holdings. Finally, quasi-indexers have characteristics
that place them in the middle of the two other categories. They generally have less concentrated
holdings like transient investors, but also have longer holding periods, like the dedicated
investors. Because they frequently trade in and out of a variety of stocks, transient institutional
investors to be those who most active in making information-based trades (Ke and Petroni 2004;
Ke and Ramalingegowda 2005). In contrast, dedicated investors are long-term investors who are
less likely to actively trade on information. The activeness of quasi-indexers in making
information-based trades is likely to be somewhere between those of transient institutional
investors and dedicated institutional investors. Thus, we conjecture that competition among
19
quasi-indexers and dedicated institutional investors will have a smaller effect in decreasing the
pricing of information asymmetry, compared to the competition among transient institutional
investors.
To test this conjecture, we reconstruct our measures of competition - number of informed
investors, proportion of informed investors, and Herfindahl index of competition - based on
quasi-indexers institutional investor ownership and dedicated institutional investor ownership.
Specifically, #Ded (#QInd) is the number of shares held by dedicated (quasi-indexing)
institutional investors. %Ded (%QInd) is the percentage of shares outstanding held by dedicated
(quasi-indexing) institutional investors. HerfDed (HerfQInd) is the Herfindahl measure for
dedicated (quasi-indexing) institutional holdings multiplied by -1 so that it is increasing in
market competition. Just like before, the measures are ranked into quintiles using the distribution
of its value within the year and then scale the quintile rank so that it ranges from zero to one.
Table 5 presents the results using these measures. The regression specification in Panel A
(Panel B) is similar to that in Table 3 (Table 4), except that different measures of competition are
used. For brevity, we present only the coefficients on information asymmetry, as well as those on
the interaction terms between competition and information asymmetry.
Similar to Table 3, Panel A uses IASpread as the measure of information asymmetry. The
first three columns of Panel A focuses on the role of competition among dedicated investors in
the pricing of information asymmetry. We observe that there is generally no significant evidence
that the pricing of information asymmetry decreases with competition among dedicated
institutional investors. In contrast, the next three columns, which focus on quasi-indexers,
provide significant evidence that the pricing of information asymmetry decreases with
competition among quasi-indexers, though the results are marginally significant with HerfQInd.
20
Similar to Table 4, Panel B uses AdjPIN as the measure of information asymmetry. Once
again, the first three columns of Panel B indicate that there is no significant evidence that the
pricing of information asymmetry decreases with competition among dedicated institutional
investors. Next, we examine how the pricing of information asymmetry decreases with
competition among quasi-indexers. The results with %QInd indicate that the pricing of
information asymmetry decreases with competition, but not the results with #QInd or HerfQInd.
Overall, the results in Table 5 support our conjecture that the nature of the competition
among investors is important in the effect of competition on the pricing of information
asymmetry. Specifically, competition among investors who are more likely to actively make
information-based trades has a greater effect in mitigating the pricing of information asymmetry.
[Insert Table 5 here]
Controlling for the Broader Trading Environment
As noted earlier, our use of institutional ownership characteristics to develop measures of
informed competition is guided by prior theoretical and empirical literature. However, there is
the concern that our results could be simply capturing cross-sectional variation in the broader
trading environment that is unrelated to competition among informed investors in the trading
environment. The challenge with controlling for the trading environment is that some of the
aspects of the trading environment are likely to be outcomes of the competition among informed
investors. For example, higher stock turnover might result from more competition, and including
these trading characteristics as control variables might not be “over-control” for the effect of
competition. Nevertheless, we examine if our results are robust to attempts to control for the
cross-sectional variation in the general trading environment.
21
To control for cross-sectional variation in the broader trading environment, we use share
turnover (Turnover). The literature has considered share turnover to be a measure of stock
liquidity (Datar et al. 1998), disagreement among investors (D’Avolio 2002), and investor
sentiment (Baker and Wurgler 2006). Specifically, in all our regressions we include share
turnover and its interactions with measures of information asymmetry and non-information
asymmetry. In untabulated analyses we find that our results are robust to the use of alternative
proxies for the broader trading environment such as total trading volume and idiosyncratic
volatility.
Table 6 shows our results after including Turnover and its interactions with measures of
information asymmetry and non-information asymmetry as additional control variables into
Equation (2). For the sake of brevity, we report only the coefficients on the measures of
information asymmetry, share turnover, competition, as well as the interaction terms between
share turnover and information asymmetry and between competition and information
asymmetry. Panel A, which repeats the analyses in Table 3 with the additional control variables,
uses IASpread to measure information asymmetry. The inclusion of Turnover does not
significantly affect our earlier results. Specifically, the pricing of IASpread decreases in the
cross-section with more competition among informed investors, after controlling for the cross-
sectional variation in share turnover. Interestingly, the pricing of information asymmetry does
not vary cross-sectionally with share turnover. This suggests that competition among informed
investors, and not the broader trading environment, drives the cross-sectional variation in the
pricing of information asymmetry.
The results in Panel B, which uses AdjPIN as the measure for information asymmetry,
show that our earlier results in Table 4 are robust to controlling for the cross-sectional variation
22
in the broader trading environment using share turnover. Similar to the results in Table 4, we find
that when competition among informed investors is measured using %Inst and %Trans, the
pricing of AdjPIN decreases in the cross-section with more competition. In fact, the statistical
significance of the interaction terms between %Inst and AdjPIN and between %Trans and
AdjPIN are now at a 1% level, whereas they are significant at a 5% level in Table 4. Moreover,
the coefficient on the interaction term between AdjPIN and competition is now marginally
significant at a 10% level when HerfInst is used to measure competition. Similar to Panel A, we
find that the pricing of information asymmetry does not vary cross-sectionally with share
turnover.
[Insert Table 6 here]
V. Informed competition, information asymmetry, and information quality
Our hypothesis predicts that the pricing of information asymmetry decreases with
competition among informed investors. We test this hypothesis using the information asymmetry
component of spread and PIN. Our goal is to provide evidence with empirical proxies that best
approximate the economic construct information asymmetry.
In this section, however, we examine the implications of our earlier results for a
fundamental issue in the accounting literature that has attracted extensive theoretical and
empirical research: the pricing of information quality. The general prediction in this literature is
that cost of capital is higher when information quality is poorer (e.g., Botosan 1997; Francis et al.
2004, 2005). To the extent that poorer information quality captures higher information
asymmetry, as argued in this literature, the pricing of information quality could also decrease
with more informed competition if the pricing of information asymmetry decreases with more
23
informed competition. Hence, in this section, we investigate whether the pricing of information
quality decreases with competition, under the assumption that information quality proxies for
information asymmetry.
To proxy for information quality (IQ), we use accruals quality (AQ) and earnings
smoothness (Smoothness). These measures have been recently used in the literature with mixed
findings on their association with the cost of capital (Francis et al., 2004, 2005; Core et al. 2008;
McInnis 2009). In addition, we also use a measure of annual report readability (FOG) developed
by Li (2008) as another proxy for information quality. The intuition behind this measure is that,
everything else equal, more syllables per word or more words per sentence make a document
harder to read.
We follow Francis et al. (2005) and estimate AQ as the standard deviation of the firm-
level residuals from the Dechow and Dichev (2002) model as modified by McNichols (2002)
during the years t-5 to t-1. The model is a cross-sectional regression of working capital accruals
on lagged, current, and future cash flows, plus the change in revenue and PPE. All variables are
scaled by average total assets.2 Following Francis et al. (2004), Smoothness is the ratio of firm’s
standard deviation of net income before extraordinary items to its standard deviation of cash
flows from operations (as with AQ all variables are scaled by average total assets). FOG is the
readability index developed by Li (2008). This measure is developed using computational
linguistics based on syntactical textual features (such as words per sentence and syllables per
word) in the 10-K filing. Following this literature and for consistency with our measures of
information asymmetry, we code all information quality proxies such that higher values mean
lower information quality and correspondingly, higher information asymmetry.
2 Wysocki (2008) proposes a modified version of accruals quality to address some potential construct validity issues with AQ. In untabulated analyses we have re-estimated the regressions in Table 7 using Wysopcki’s accrual quality measure and have found that the pricing of this measure decreases with all our measures of competition.
24
Table 7 presents the results of the cross-sectional asset pricing tests examining whether
the pricing of information quality varies cross-sectionally with competition. The regression
specification follows Eq. (2), except that we now replace the information asymmetry measures
with measures of information quality. Unlike the earlier regressions, we do not include a non-
information asymmetry component in the regression because unlike spread and PIN, prior
literature does not provide guidance on how to decompose information quality measures into
information asymmetry and non-information-asymmetry components. For brevity, we only
tabulate the results with measures of competition based on transient institutional investor
ownership. The results are essentially the same with measures of competition based on total
institutional investor ownership.
The first three columns documents how the pricing of AQ varies with competition among
informed investors. The next three columns and the final three columns report the results with
Smoothness and FOG, respectively. Similar to our earlier regressions, the coefficient on the
interaction term of IQ and Competition is the coefficient of interest in each of the columns. A
statistically significant negative coefficient on this interaction term indicates that more
competition is associated with less pricing of information quality.
To the extent that poorer information quality proxies for higher information asymmetry,
the results support our hypothesis that the pricing of information asymmetry decreases in
competition among informed investors. From the first three columns, we observe that the
coefficient on the interaction term between AQ and #Trans (%Trans, HerfTrans) is -13.87 (-7.56,
-13.59). This indicates that for a one standard deviation (which equals 0.03) difference in AQ, the
monthly difference in the expected return in the most competitive quintile is 0.42% (0.23%,
25
0.41%) less than in the least competitive quintile.3 In the next three columns, the coefficient on
the interaction term between Smoothness and #Trans (%Trans, HerfTrans) is -0.34 (-0.28, -0.45).
This indicates that for a one standard deviation (which equals 0.52) difference in Smoothness, the
monthly difference in the expected return in the most competitive quintile is 0.18% (0.15%,
0.23%) less than in the least competitive quintile. Finally, in the last three columns, the
coefficient on the interaction term between FOG and #Trans (%Trans, HerfTrans) is -0.17 (-
0.21, -0.16). This indicates that for a one standard deviation (which equals 1.41) difference in
FOG, the monthly difference in the expected return in the most competitive quintile is 0.24%
(0.30%, 0.23%) less than in the least competitive quintile. Overall, the results in Table 7 are
present support for the joint hypothesis that the pricing of information quality decreases with
competition and that information quality proxies for information asymmetry.
[Insert Table 7 here]
VI. Conclusion
The issue of whether information asymmetry is priced has been of significant academic
interest. In this paper, we re-examine this question by emphasizing an important aspect of capital
markets with information asymmetry - competition among informed investors. While prior
empirical literature has investigated whether information asymmetry is priced on average, it has
not studied whether there is cross-sectional variation in the pricing conditional on the extent of
the competition among informed investors. Relying on theories that highlight that more
competition among informed investors leads to more informative prices, our study examines the
3 To estimate the economic significance of the interaction term, we multiply the coefficient on the interaction term by the standard deviation of the respective IQ variable. For example, the economic significance of the interaction term on AQ and #Trans equals 0.42% (=0.03 x -13.87).
26
variation in the pricing of information asymmetry conditional on the extent of competition
among informed investors.
In our study, we use the information asymmetry components of bid-ask spread and PIN
as proxies for information asymmetry. To measure the degree of competition among informed
investors, we use a variety of proxies based on institutional investor ownership, with the
underlying assumption that institutional investors are more likely to be the informed investors.
While the theories on competition among informed investors typically depict competition in
terms of the number of informed investors, we also develop alternative proxies that take into
account the size of the ownership by informed investors and the distribution of shares among
them.
Consistent with our hypothesis, we show that the pricing of information asymmetry
decreases with competition among informed investors, and that the effect is economically
important. We then explore its implications on the accounting literature that examines whether
information quality, as a proxy for information asymmetry, is priced. We repeat our analyses by
replacing our measures of information asymmetry proxies with measures of information quality,
and we find similar results. That is, our results indicate that the pricing of information quality
also decreases in competition among informed investors.
Our results suggest that future research investigating the effects of the information
environment should consider the level of competition among informed investors in the trading
environment. A direct implication of our findings is that in face of information asymmetry, firms
could potentially reduce their cost of capital by encouraging more competition among informed
investors through higher institutional ownership and/or more even distribution of shares among
institutional investors. An indirect implication of our findings is that efforts to mitigate
27
information asymmetry such as increased corporate disclosure and transparent financial reporting
might have greater cost of capital effects in markets (either within a single country or across
different countries) characterized by less competition among informed investors.
28
Appendix A. Decomposition of spread
The following regression specification to obtain the parameters required to decompose
spread:
i,s 0 i,s 1 i,s i,s 0 i,s 1 i,s i,sΔPrice =C ΔTrade +C ΔTrade TradeSize +Z Trade +Z Trade TrdSize +ε (A1)
where for the trade at time s for firm i, ΔPrice is the change in trade price scaled by the previous
trade price, TradeSize is the number of shares traded, and Trade is an indicator that is equal to +1
if the trade is classified as buyer-initiated and -1 if the trade is seller-initiated.
A brief description of the intuition underlying Eq. (A1) is as follows: Glosten and Harris
(1988) indicate that for a round-trip transaction, the non-information asymmetry component is
given by 2(C0 + C1TradeSize) and the information asymmetry component of the bid-ask spread
is given by 2(Z0 + Z1TradeSize), with the estimated spread being the sum of the two components.
The first component allows market makers to generate revenue from a seemingly random order
flow to cover inventory holding and order processing costs, as well as provide monopoly profits.
It is a transitory component because it is unrelated to the underlying value of the securities. The
second component assumes that order flows will be correlated with future price changes. It arises
because rational market makers in a competitive environment will widen the spread in response
to information asymmetry. Scaling of price changes by the previous price, i.e., the use of
intraday return, facilitates cross-sectional comparability in the extent of information asymmetry
across firms (Armstrong et al. 2009).
Econometrically, it can be observed from Eq. (A1), the key distinction between the
information asymmetry component and the non-information asymmetry component is that the
coefficients for the non-information asymmetry component is based on ΔTrade, while the
information asymmetry component is based on Trade. The intuition for the difference is as
29
follows. The non-information asymmetry component assumes that market makers generate
revenue using random switching between buyer- and seller-initiated trades to “buy low and sell
high” on average. ΔTrade captures the idea that when a buy (sell) order is filled, market makers
raise bid and/or ask prices to increase the probability that the next order will be a sell (buy) to
maintain inventory. Price changes, which reflect the compensation to the market makers, reverse
on average (i.e., the effect of trades on prices is transitory). The information asymmetry
component captures the idea that buy orders (i.e., Trade = 1) cause “true” prices to rise by (Z0 +
Z1TradeSize) while sell orders (i.e., Trade = -1) cause them to fall by -(Z0 + Z1TradeSize). Buy
and sell orders cause a permanent effect on prices since they are due to a change in expectations
of firm value. Eq. (A1) provides the regression coefficients, Z0, Z1, C0, and C1. For trade size, we
compute the average trade size (AvgTradeSize).
We compute IASpread and NIASpread using the intraday data from the Institute for the
Study of Security Markets database (ISSM) and NYSE Trade and Quotes database (TAQ). ISSM
provides the data for NYSE and AMEX firms from 1983 to 1992 and NASDAQ firms from 1987
to 1992. TAQ provides the data for NYSE, AMEX, and NASDAQ firms from 1993 to 2004.
Prior to using the data, the intraday data is cleaned following the procedure discussed in Ng et al.
(2008). All the intraday data in each year is then used to compute annual measures of IASpread
and NIASpread.
30
References
Admati, A., and P. Pfleiderer. 1988. A theory of intraday patterns: Volume and price variability, Review of Financial Studies 1: 3-40.
Arbel, A., and P. Strebel. 1983. Pay attention to neglected firms. Journal of Portfolio
Management 9, 37-42. Armstrong, C., J. Core, D. Taylor, and R. Verrecchia. 2009. Asset pricing in imperfect markets.
Working paper, the Wharton School of the University of Pennsylvania. Baker, M., and J. Wurgler. 2006. Investor sentiment and the cross-section of stock returns.
Journal of Finance 61: 1645-1680. Bartov, E., S. Radhakrishnan, and I. Krinsky. 2000. Investor sophistication and patterns in stock
returns. The Accounting Review 75: 43-63. Botosan, C. 1997. Disclosure level and the cost of equity capital, The Accounting Review 72: 323-349. Brennan, M., and A. Subrahmanyam. 1995. Investment analysis and pricing information in
securities markets. Journal of Financial Economics 38: 361-381. Brennan, M., and A. Subrahmanyam. 1996. Market microstructure and asset pricing: On the
compensation for illiquidity in stock returns. Journal of Financial Economics 41: 441-464.
Bushee, B. 1998. The influence of institutional investors on myopic R&D investment behavior.
The Accounting Review 73: 305–333. Comment, R., and G. Jarrell. 1995. Corporate focus and stock returns. Journal of Financial
Economics 37: 67-87. Copeland, T., and D. Galai. 1983. Information effects on the bid-ask spread. Journal of Finance
38: 1457-1469. Core, J., W. Guay, and R. Verdi. 2008. Is accruals quality a priced risk factor? Journal of
Accounting and Economics 46: 2-22. D’Avolio, G. 2002. The market for borrowing stock. Journal of Financial Economics 66: 271-
306. Datar, V., N. Naik, and R. Radcliffe. 1998. Liquidity and stock returns: an alternative test.
Journal of Financial Markets 1: 203-219.
31
Dechow, P. and I. Dichev. 2002. The quality of accruals and earnings: The role of accrual estimation errors. The Accounting Review 77: 35-59.
Diamond, D., and R. Verrecchia. 1991. Disclosure, liquidity and the cost of capital. Journal of
Finance 46: 1325-1359. Dimson, E., 1979. Risk measurement when shares are subject to infrequent trading. Journal of
Financial Economics 7: 197-226. Duarte, J., and L. Young. 2009. Why is PIN priced? Journal of Financial Economics 91: 119-
138. Easley, D., S. Hvidkjaer, and M. O'Hara. 2002. Is information risk a determinant of asset
returns? Journal of Finance 57: 2185-2221. Easley, D., and M. O’Hara. 2004. Information and the cost of capital. Journal of Finance 59:
1553-1583. Edmans, A., and G. Manso. 2009. Governance through exit and voice: A theory of multiple
blockholders. MIT working paper. Fama, E., and K. French. 1992. The cross-section of expected stock returns. Journal of Finance
47: 427–465. Fama, E., and J. MacBeth. 1973. Risk, return and equilibrium: empirical tests. Journal of
Political Economy 81: 607-636. Foster, F. and S. Viswanathan. 1993. The effect of public information and competition on trading
volume and price volatility. Review of Financial Studies 6: 23-56. Foster, F., and S. Viswanathan. 1996. Strategic trading when agents forecast the forecasts of
others. Journal of Finance 51: 1437-1478. Francis, J., R. LaFond, P. Olsson, and K. Schipper. 2004. Costs of equity and earnings attributes.
The Accounting Review 79: 967–1010. Francis, J., R. LaFond, P. Olsson, and K. Schipper. 2005. The market pricing of accruals quality.
Journal of Accounting and Economics 39: 295-327. Gande, A., M. Puri, and S. Anthony. 1999. Bank entry, competition, and the market for corporate
securities underwriting. Journal of Financial Economics 54: 165-195. Glosten, L., and L. Harris. 1988. Estimating the components of the bid/ask spread. Journal of
Financial Economics 21: 123-142.
32
Glosten, L., and P. Milgrom. 1985. Bid, ask, and transaction prices in a specialist market with hetergeneously informed traders. Journal of Financial Economics 14: 71-100.
Grinblatt, M., and M. Keloharju, 2000. The investment behavior and performance of various
investor types: a study of Finland’s unique data set. Journal of Financial Economics 55: 43-67.
Guedhami, O., and J. Pittman. 2006. Ownership concentration in privatized firms: The role of
disclosure standards, auditor choice, and auditing infrastructure. Journal of Accounting Research 44: 889-929.
Guriev, S., and A. Rachinsky. 2005. The role of oligarchs in Russian capitalism. Journal of
Economic Perspectives 19: 131-150. Herfindahl, O. 1950. Concentration in the U.S. steel industry. Unpublished doctoral dissertation,
Columbia University. Holden, C., and A. Subrahmanyam. 1992. Long-lived private information and imperfect
competition, Journal of Finance 47: 247-270. Holden, C., and A. Subrahmanyam. 1994. Risk aversion, imperfect competition and long lived
information. Economics Letters 44: 181-190. Hughes, J., J. Liu, and J. Liu. 2007. Information, diversification and the cost of capital. The
Accounting Review 82: 705-729. Jimbalvo, J., S. Rajgopal, and M. Venkatachalam. 2002. Institutional ownership and the extent
to which stock prices reflect future earnings. Contemporary Accounting Research 19: 117-145.
Ke, B., and K. Petroni. 2004. How informed are actively trading institutional investors?
Evidence from their trading behavior before a break in a string of consecutive earnings increases. Journal of Accounting Research 42: 895-927.
Ke, B., and S. Ramalingegowda. 2005. Do institutional investors exploit the post-earnings
announcement drift? Journal of Accounting and Economics 39: 25-53. Kyle, A. 1985. Continuous auctions and insider trade, Econometrica 53: 1315-1335. Lambert, R., C. Leuz, and R. Verrecchia. 2007. Accounting information, disclosure, and the cost
of capital, Journal of Accounting Research 45: 385-420. Lee, C., B. Mucklow, and M. Ready. 1993. Spreads, depths, and the impact of earnings
information: An intraday analysis. Review of Financial Studies 6: 345-374.
33
Leuz, C., and R. Verrecchia. 2000. The economic consequences of increased disclosure. Journal of Accounting Research 38: 91-124.
Leuz, C., and P. Wysocki. 2008. Economic consequences of financial reporting and disclosure
regulation: A review and suggestions for future research. Working paper, the Booth School at University of Chicago, MIT Sloan School of Management.
Li, F. 2008. Annual reporting readability, current earnings, and earnings persistence. Journal of
Accounting and Economics 45: 221-247. McInnis, J. 2009. Are smoother earnings associated with a lower cost of equity capital? The
Accounting Review (forthcoming). McNichols, M. 2002. Discussion of the quality of accruals and earnings: The role of accrual
estimation errors. The Accounting Review 77: 61–69. Mohanram, P., and S. Rajgopal. 2009. Is information risk (PIN) priced? Journal of Accounting
and Economics 47: 226-243. Ng, J., T. Rusticus, and R. Verdi. 2008. Implications of transaction costs for the post-earnings
announcement drift. Journal of Accounting Research 46: 661-696. Porter, M. 1992. Capital choices: Changing the way America invests in industry. Boston, MA:
Council on Competitiveness/Harvard Business School. Sias, R. and L. Starks. 1997. Return autocorrelation and institutional investors. Journal of
Financial Economics 46: 103-131. Verrecchia, R., and J. Weber. 2008. Firm performance measures and adverse selection costs.
Working paper, the Wharton School of the University of Pennsylvania, MIT Sloan School of Management.
Wysocki, P. 2008. Assessing earnings and accruals quality: U.S. and international evidence.
Working paper, MIT Sloan School of Management.
34
Table 1: Observations by Year
The number of firm-year observations for each information asymmetry proxy (IASpread and AdjPIN) is listed. IASpread is the scaled information asymmetry component of spread calculated according to the modified Glosten and Harris (1988) methodology. AdjPIN is Duarte and Young’s (2008) adjusted PIN measure representing the information asymmetry component of PIN (Easley et al. 2002).
Year IASpread AdjPIN
1983 1,498 1,725 1984 1,512 1,640 1985 1,396 1,580 1986 1,335 1,525 1987 2,437 1,461 1988 2,399 1,492 1989 2,412 1,492 1990 2,374 1,520 1991 2,502 1,540 1992 2,866 1,574 1993 2,085 1,792 1994 2,281 1,813 1995 2,821 1,863 1996 3,015 1,829 1997 3,620 1,790 1998 3,841 1,844 1999 3,849 1,725 2000 3,603 1,635 2001 3,541 1,589 2002 3,486 1,579 2003 3,462 1,549 2004 3,388 1,514
Total 59,723 36,071
35
Table 2: Summary Statistics Panel A presents summary statistics on the variables of interest used in this study. Panel B shows the Pearson correlations for these variables along with the respective p-values. All variables are calculated over the calendar year. IASpread is the scaled information asymmetry component of spread calculated according to the modified Glosten and Harris (1988) methodology. NIASpread is the non-information asymmetry proportion of spread. AdjPIN is Duarte and Young’s (2008) adjusted PIN measure representing the information asymmetry component of PIN (Easley et al. 2002). PSOS is the probability of symmetric order flow shock based on Duarte and Young (2008). #Inst (#Trans) is the number of shares held by institutional (transitory) investors. %Inst (%Trans) is the percentage of shares outstanding held by institutional (transitory) investors. HerfInst (HerfTrans) is the Herfindahl measure for institutional (transitory) holdings multiplied by -1 so that it is increasing in competition. Beta is the post-ranking Dimson (1979) beta calculated using 40 portfolios formed on five-year rolling pre-ranking individual firm betas. Size is the natural log of the year end market value of equity. BTM is the book to market ratio, which is the natural log of year end market value of equity divided by the book value of equity known three months prior to the calendar year end. Variables are winsorized at the 1st and 99th percentiles. Panel A: Descriptive statistics Variable Mean STD P25 Median P75 Measures of information and non-information asymmetry IASpread (%) 0.44 1.35 0.09 0.21 0.44 NIASpread (%) 1.75 2.67 0.30 0.82 2.13 AdjPIN 0.18 0.09 0.12 0.16 0.22 PSOS 0.29 0.17 0.17 0.23 0.36 Measures of competition #Inst 77.61 118.41 11.50 32.25 94.50 %Inst 0.35 0.25 0.14 0.32 0.55 HerfInst -0.19 0.21 -0.25 -0.11 -0.05 #Trans 22.93 38.87 2.50 8.25 27.00 %Trans 0.10 0.11 0.02 0.06 0.15 HerfTrans -0.44 0.33 -0.74 -0.33 -0.14 Control variables Beta 1.26 0.49 0.92 1.18 1.49 Size 19.12 2.06 17.61 18.96 20.51 BTM -0.71 0.95 -1.23 -0.64 -0.13
36
Table 2: continued Panel B: Correlation matrix IASpread NIASpread AdjPIN PSOS #Inst %Inst HerfInst #Trans %Trans HerfTrans IASpread 1 0.33 0.32 0.33 -0.14 -0.19 -0.22 -0.13 -0.14 -0.24 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 NIASpread 1 0.20 0.31 -0.31 -0.43 -0.45 -0.29 -0.33 -0.52 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 AdjPIN 1 0.37 -0.48 -0.44 -0.47 -0.45 -0.36 -0.57 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 PSOS 1 -0.35 -0.43 -0.54 -0.29 -0.28 -0.61 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 #Inst 1 0.51 0.41 0.86 0.33 0.55 <.0001 <.0001 <.0001 <.0001 <.0001 %Inst 1 0.61 0.51 0.76 0.72 <.0001 <.0001 <.0001 <.0001 HerfInst 1 0.37 0.42 0.74 <.0001 <.0001 <.0001 #Trans 1 0.47 0.54 <.0001 <.0001 %Trans 1 0.59 <.0001 HerfTrans 1
37
Table 3: Pricing of Information Asymmetry – IASpread
The following Fama-MacBeth regression is estimated with monthly returns from 1984 through 2005 (717,145 firm-months). Rt+1 = α + Σ βj Controls j,t + λ1t Competitioni,t + λ2t IASpread i,t + λ3t Competitioni,t x IASpread i,t + εi,t+1 where the dependent variable, Rt+1, is a monthly realized return in year t+1. All independent variables are from year t. IASpread is the scaled information asymmetry component of spread calculated according to the modified Glosten and Harris (1988) methodology. NIASpread is the non-information asymmetry proportion of spread. Competition is either #Inst, %Inst, HerfInst, #Trans, %Trans, or HerfTrans. #Inst (#Trans) is the number of shares held by institutional (transitory) investors. %Inst (%Trans) is the percentage of shares outstanding held by institutional (transitory) investors. HerfInst (HerfTrans) is the Herfindahl measure for institutional (transitory) holdings multiplied by -1 so that it is increasing in competition. Competition measures are ranked into quintiles and scaled so that their quintile rankings range from zero to one. Beta is the post-ranking Dimson (1979) beta calculated using 40 portfolios formed on five-year rolling pre-ranking individual firm betas. Size is the natural log of the year end market value of equity. BTM is the book to market ratio, which is the natural log of year end market value of equity divided by the book value of equity known three months prior to the calendar year end. The Fama-MacBeth (1973) t-statistics are below the coefficient estimates in parentheses. Significance levels are based on two-tailed tests. ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.
Competition Proxy #Inst %Inst HerfInst #Trans %Trans HerfTrans
Intercept -1.09 2.15* 0.11 0.56 1.29 -0.71 0.57 (-1.18) (1.92) (0.12) (0.62) (1.13) (-0.75) (0.58)
Beta 0.02 0.00 0.00 0.02 -0.01 0.00 0.02 (0.06) (0.00) (0.01) (0.04) (-0.03) (0.01) (0.07)
Size 0.10* -0.10 0.02 -0.01 -0.05 0.07 -0.01 (1.89) (-1.54) (0.40) (-0.13) (-0.72) (1.33) (-0.09)
BTM 0.32*** 0.28*** 0.32*** 0.32*** 0.30*** 0.32*** 0.32*** (3.22) (2.81) (3.22) (3.22) (3.05) (3.27) (3.28)
Competition 1.38*** 0.73*** 0.84*** 1.05*** 0.37* 0.83*** (4.38) (3.54) (4.31) (3.45) (1.80) (3.78)
NIASpread 0.20*** 0.24*** 0.24*** 0.24*** 0.22*** 0.22*** 0.24*** (3.13) (3.74) (3.46) (3.73) (3.50) (3.28) (3.63)
IASpread 0.25** 0.33*** 0.37*** 0.32*** 0.36*** 0.40*** 0.33*** (2.28) (2.72) (2.76) (2.59) (2.90) (3.08) (2.61)
Competition x NIASpread -0.45*** -0.30*** -0.42*** -0.35*** -0.11 -0.42*** (-3.27) (-2.97) (-3.55) (-2.67) (-1.02) (-3.54)
Competition x IASpread -0.75** -0.75*** -0.52* -0.91*** -0.68*** -0.83*** (-2.22) (-2.87) (-1.69) (-2.66) (-2.58) (-2.66)
Adj-R2 (%) 4.03 4.36 4.35 4.30 4.39 4.40 4.33
38
Table 4: Pricing of Information Asymmetry – AdjPIN
The following Fama-MacBeth regression is estimated with monthly returns from 1984 through 2005 (433,621 firm months). Rt+1 = α + Σ βj Controls j,t + λ1t Competitioni,t + λ2t AdjPIN i,t + λ3t Competitioni,t x AdjPIN i,t + εi,t+1 where the dependent variable, Rt+1, is a monthly realized return in year t+1. All independent variables are from year t. AdjPIN is Duarte and Young’s (2008) adjusted PIN measure representing the information asymmetry component of PIN. PSOS is the probability of symmetric order flow shock based on Duarte and Young (2008). Competition is either #Inst, %Inst, HerfInst, #Trans, %Trans, or HerfTrans. #Inst (#Trans) is the number of shares held by institutional (transitory) investors. %Inst (%Trans) is the percentage of shares outstanding held by institutional (transitory) investors. HerfInst (HerfTrans) is the Herfindahl measure for institutional (transitory) holdings multiplied by -1 so that it is increasing in competition. Competition measures ranked into quintiles and are scaled so that their quintile rankings range from zero to one. Beta is the post-ranking Dimson (1979) beta calculated using 40 portfolios formed on five-year rolling pre-ranking individual firm betas. Size is the natural log of the year end market value of equity. BTM is the book to market ratio, which is the natural log of year end market value of equity divided by the book value of equity known three months prior to the calendar year end. The Fama-MacBeth (1973) t-statistics are below the coefficient estimates in parentheses. Significance levels are based on two-tailed tests. ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.
Competition Proxy #Inst %Inst HerfInst #Trans %Trans HerfTrans
Intercept 0.29 2.35 0.36 0.79 2.07 0.31 1.09 (0.24) (1.64) (0.30) (0.65) (1.40) (0.25) (0.84)
Beta -0.19 -0.22 -0.22 -0.21 -0.23 -0.20 -0.21 (-0.62) (-0.74) (-0.72) (-0.69) (-0.79) (-0.66) (-0.70)
Size 0.03 -0.11 0.02 -0.01 -0.09 0.02 -0.03 (0.53) (-1.46) (0.27) (-0.20) (-1.18) (0.38) (-0.48)
BTM 0.23** 0.19** 0.23** 0.22** 0.20** 0.23** 0.22** (2.48) (2.05) (2.42) (2.40) (2.18) (2.47) (2.40)
Competition 1.41*** 0.70*** 0.88*** 1.31*** 0.48** 1.00*** (3.45) (2.74) (3.16) (3.39) (2.08) (3.33)
PSOS 0.62*** 0.91*** 0.89*** 0.95*** 0.93*** 0.79*** 0.96*** (2.65) (2.79) (2.84) (2.81) (2.84) (2.61) (2.89)
AdjPIN 0.26 0.92 0.83 0.68 0.85 0.71 0.74 (0.42) (1.43) (1.28) (1.05) (1.37) (1.13) (1.16)
Competition x PSOS -0.59 -0.72 -1.11* -0.87 -0.45 -1.05 (-0.95) (-1.27) (-1.76) (-1.36) (-0.78) (-1.59)
Competition x AdjPIN -2.19 -2.73** -2.28 -1.99 -2.39** -2.10 (-1.25) (-2.35) (-1.45) (-1.14) (-2.00) (-1.31)
Adj-R2 (%) 3.42 3.74 3.80 3.67 3.78 3.80 3.72
39
Table 5: Competition among other Types of Institutional Investors This table presents the regressions that examine how the pricing of information asymmetry varies with measures of competition constructed using dedicated institutional investor ownership and quasi-indexer institutional investor ownership. The regression specification in Panel A (Panel B) is similar to that in Table 3 (Table 4), except that different measures of competition are used. Information asymmetry and non-information-asymmetry are proxied by IASpread and NIASpread (AdjPIN and PSOS), respectively, in Panel A (Panel B). Competition is either #Ded, %Ded, HerfDed, #Trans, %Trans, or HerfTrans. #Ded (#QInd) is the number of shares held by dedicated (quasi-indexing) institutional investors. %Ded (%QInd) is the percentage of shares outstanding held by dedicated (quasi-indexing) institutional investors. HerfDed (HerfQInd) is the Herfindahl measure for dedicated (quasi-indexing) institutional holdings multiplied by -1 so that it is increasing in market competition. Competition measures are ranked into quintiles and scaled so that their quintile rankings range from zero to one. For brevity, we do not tabulate the coefficients on Beta, Size, and BTM. We also do not tabulate the coefficients on NIASpread, PSOS, as well as on the interactions of these variables with Competition. All variables are defined in Table 2. The Fama-MacBeth (1973) t-statistics are below the coefficient estimates in parentheses. Significance levels are based on two-tailed tests. ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively. Panel A – IASpread
Competition Proxy #Ded %Ded HerfDed #QInd %QInd HerfQInd
IASpread 0.31** 0.31** 0.32*** 0.36*** 0.37*** 0.34*** (2.27) (2.20) (2.59) (2.90) (2.62) (2.69)
Competition x IASpread -0.32 -0.32 -0.52* -0.91*** -0.64** -0.55* (-1.04) (-1.35) (-1.69) (-2.66) (-2.31) (-1.88)
Panel B – AdjPIN
Competition Proxy #Ded %Ded HerfDed #QInd %QInd HerfQInd
AdjPIN 0.93 0.78 0.68 0.85 0.90 0.67 (1.36) (1.12) (1.05) (1.37) (1.35) (1.01)
Competition x AdjPIN -2.27 -1.09 -2.28 -1.99 -2.96*** -2.35 (-1.38) (-1.10) (-1.45) (-1.14) (-2.44) (-1.42)
40
Table 6: Controlling for the Trading Environment
This table presents the regressions that examine how the pricing of information asymmetry varies with competition after controlling for the broader trading environment. The regression specification is similar to that in Table 3. We add additional controls for trading environment by including Turnover, and its interactions with proxies for information asymmetry and non-information-asymmetry. Turnover is ranked into quintiles based on the distribution within the year. The quintile ranks are then scaled to range from zero to one. Information asymmetry and non-information-asymmetry are proxied by IASpread and NIASpread (AdjPIN and PSOS), respectively, in Panel A (Panel B). Competition is measured as either #Inst, %Inst, HerfInst, #Trans, %Trans, or HerfTrans. For brevity, we do not tabulate the coefficients on Beta, Size, BTM. We also do not tabulate the coefficients on NIASpread, PSOS, as well as on the interactions of these variables with Competition and Turnover. Except for Turnover, all variables are defined in Table 2. The Fama-MacBeth (1973) t-statistics are below the coefficient estimates in parentheses. Significance levels are based on two-tailed tests. ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.
Panel A – IASpread
Competition Proxy #Inst %Inst HerfInst #Trans %Trans HerfTrans
Turnover -0.66** -0.57* -0.56* -0.75*** -0.66** -0.65** (-2.33) (-1.90) (-1.91) (-2.65) (-2.19) (-2.22)
Competition 1.89*** 1.00*** 1.15*** 1.73*** 0.78*** 1.30*** (6.46) (4.40) (5.53) (6.17) (3.70) (5.98)
IASpread 0.29** 0.34*** 0.30** 0.29** 0.34*** 0.28** (2.30) (2.60) (2.33) (2.33) (2.67) (2.26)
Turnover x IASpread 0.15 0.14 0.10 0.23 0.27 0.13 (0.38) (0.38) (0.26) (0.58) (0.70) (0.33)
Competition x IASpread -1.06** -0.93*** -0.73** -1.22*** -0.87*** -1.01** (-2.55) (-2.82) (-2.08) (-2.82) (-2.77) (-2.49)
Panel B – AdjPIN
Competition Proxy #Inst %Inst HerfInst #Trans %Trans HerfTrans
Turnover -0.37 -0.24 -0.23 -0.54 -0.26 -0.30 (-1.12) (-0.77) (-0.70) (-1.59) (-0.76) (-0.93)
Competition 1.60*** 0.79*** 0.92*** 1.69*** 0.69*** 1.15*** (3.41) (2.85) (2.88) (3.71) (2.58) (3.34)
AdjPIN 0.03 0.10 -0.02 -0.10 0.09 -0.04 (0.06) (0.17) (-0.03) (-0.17) (0.15) (-0.08)
Turnover x AdjPIN 1.63 1.70 1.40 1.97 1.92 1.26 (1.01) (0.99) (0.86) (1.17) (1.08) (0.77)
Competition x AdjPIN -2.96 -3.89*** -2.91* -3.02 -3.84*** -2.75 (-1.52) (-2.92) (-1.69) (-1.50) (-2.75) (-1.56)
41
Table 7: Pricing of Information Quality
This table presents the regressions that examine how the pricing of information quality (IQ) varies with competition. The regression specification is identical to that in Table 3. Information quality is proxied by AQ, Smoothness, and FOG. AQ is the standard deviation calculated over a five year period of a firm’s residuals from an annual estimation of the modified Dechow-Dichev (2002) model (Francis et al. 2005). Smoothness is the ratio of firm’s standard deviation of net income before extraordinary items divided by beginning total assets, to its standard deviation of cash flows from operations divided by beginning total assets. FOG is the measure of financial statement readability developed in Li (2008). The regressions for AQ, Smoothness, and FOG use 546,882, 546,882, and 220,721 firm-months. Competition is measured as either #Trans, %Trans, or HerfTrans, with the quintile rankings scaled to range from zero to one. All other variables are defined in Table 2. The Fama-MacBeth (1973) t-statistics are below the coefficient estimates in parentheses. Significance levels are based on two-tailed tests. ***, **, and * denotes significance at the 1%, 5%, and 10% levels, respectively.
IQ = AQ IQ = Smoothness IQ = FOG
#Trans %Trans HerfTrans #Trans %Trans HerfTrans #Trans %Trans HerfTrans Intercept 5.84*** 2.70*** 3.98*** 5.68*** 2.75*** 3.89*** 3.97** 0.32 1.89
(5.60) (3.14) (4.41) (5.10) (2.97) (4.04) (2.27) (0.21) (1.25)
Beta -0.06 0.02 -0.01 -0.11 0.03 -0.05 0.32 0.39 0.36 (-0.20) (0.06) (-0.04) (-0.31) (0.07) (-0.16) (0.58) (0.71) (0.64)
Size -0.29*** -0.09* -0.18*** -0.27*** -0.09* -0.16*** -0.32*** -0.13 -0.19** (-4.66) (-1.81) (-3.40) (-4.04) (-1.65) (-2.92) (-3.25) (-1.35) (-2.36)
BTM 0.26*** 0.32*** 0.30*** 0.27*** 0.32*** 0.31*** 0.26 0.31* 0.30* (2.87) (3.48) (3.30) (3.02) (3.51) (3.43) (1.62) (1.87) (1.88)
Competition 1.77*** 0.21 1.05*** 1.27*** 0.01 0.70*** 4.39*** 3.99*** 3.44** (5.88) (1.24) (4.90) (4.35) (0.09) (3.22) (2.59) (2.67) (2.06)
IQ 5.41** 3.68 5.37** 0.24** 0.23* 0.31** 0.15** 0.17** 0.15* (2.42) (1.64) (2.46) (2.04) (1.95) (2.57) (1.97) (2.55) (1.95)
IQ x Competition -13.87*** -7.56*** -13.59*** -0.34** -0.28* -0.45*** -0.17* -0.21*** -0.16* (-4.12) (-2.69) (-4.35) (-2.15) (-1.91) (-2.87) (-1.91) (-2.74) (-1.77)
Adj-R2 (%) 3.61 3.65 3.58 3.44 3.51 3.42 4.43 4.49 4.39