· PDF fileLeveling the Playing Field: Financial Regulation and Disappearing Local Bias of...
Transcript of · PDF fileLeveling the Playing Field: Financial Regulation and Disappearing Local Bias of...
Leveling the Playing Field: Financial Regulation and
Disappearing Local Bias of Institutional Investors∗
Gennaro Bernile, University of Miami
Alok Kumar, University of Miami
Johan Sulaeman, Southern Methodist University
July 5, 2012
Abstract – This study examines whether exogenous regulatory shocks induced by Regulation Fair
Disclosure and the Sarbanes-Oxley Act reduce the local bias and the local informational advantage of
institutional investors. We find that, following regulatory changes, the local bias and local informational
advantage of institutional investors around corporate headquarters declines sharply as their selective
access to private information is curtailed. The decline in local institutional ownership is more salient
among firms that had more opaque information environments prior to the regulatory changes. Even at
the aggregate market-level, the degree of informed trading attributed to local investors declines. The
local informational advantage of sell-side equity analysts who are more likely to have selective access
exhibits a similar declining pattern and disappears in the post-regulation period. In contrast, regulatory
changes have minimal impact on the local informational advantage of institutions in economically
relevant regions as they are unlikely to have selective access to management. These findings indicate
that regulatory changes affected institutional behavior primarily by curtailing their selective access to
management rather than adversely affecting their information processing advantages.
∗Please address all correspondence to Alok Kumar, Department of Finance, School of Busi-ness Administration, 514G Jenkins Building, University of Miami, Coral Gables, FL 33124;Phone: 305-284-1882; email: [email protected]. Gennaro Bernile can be reached at 305-284-6690 or [email protected]. Johan Sulaeman can be reached at 214-768-8284 [email protected]. We thank an anonymous referee, Anup Agrawal, Warren Bailey, ScottBauguess, Vidhi Chhaochharia, Larry Fauver, Rocco Huang, George Korniotis, S. P. Kothari, KelvinLaw, Evgeny Lyandres, Kumar Venkataraman, Peter Wysocki, Scott Yonker, and seminar participantsat University of Miami, University of Tennessee, U.S. Securities and Exchange Commission (SEC), and22nd Annual Conference on Financial Economics and Accounting for helpful comments and valuablesuggestions. We are very grateful to Jefferson Duarte and Lance Young for allowing us to use theirPIN decomposition data, and Hongping Tan for sharing the analyst location data. We also thank theFinancial Markets Research Center at Vanderbilt University for providing the effective spread data.We are responsible for all remaining errors and omissions.
Leveling the Playing Field: Financial Regulation and DisappearingLocal Bias of Institutional Investors
Abstract – This study examines whether exogenous regulatory shocks induced by Regulation
Fair Disclosure and the Sarbanes-Oxley Act reduce the local bias and the local informational
advantage of institutional investors. We find that, following regulatory changes, the local bias
and local informational advantage of institutional investors around corporate headquarters de-
clines sharply as their selective access to private information is curtailed. The decline in local
institutional ownership is more salient among firms that had more opaque information envi-
ronments prior to the regulatory changes. Even at the aggregate market-level, the degree of
informed trading attributed to local investors declines. The local informational advantage of
sell-side equity analysts who are more likely to have selective access exhibits a similar declin-
ing pattern and disappears in the post-regulation period. In contrast, regulatory changes have
minimal impact on the local informational advantage of institutions in economically relevant
regions as they are unlikely to have selective access to management. These findings indicate
that regulatory changes affected institutional behavior primarily by curtailing their selective
access to management rather than adversely affecting their information processing advantages.
1. Introduction
Financial regulatory reforms such as the Regulation Fair Disclosure (Reg FD) and the Sarbanes-
Oxley Act (SOX) generated exogenous shocks to the information environments of publicly-traded
firms in the U.S. These regulatory changes were designed to influence the perceived quality of
financial information that firms are required to disclose publicly. Those new rules also affected
the flow and timing of information dissemination.
For example, Reg FD was aimed at “leveling the playing field” among different types of
investors. Its rules were designed to curb selective disclosure of information to market par-
ticipants and mandated the disclosure of all material information to all investors at the same
time. Similarly, SOX mandated the implementation of new rules for financial reporting by pub-
lic companies with the intent to improve its transparency, reliability, and accountability. In
particular, SOX addressed issues such as enhanced financial disclosure, auditor independence,
and corporate governance (Coates (2007)).
In this study, we examine whether financial regulatory reforms affected the local ownership
patterns and local informational advantage of institutional investors.1 Our analysis is motivated
by the recent research in local bias, which demonstrates that institutional investors may have
superior information about local stocks. Consequently, those investors earn higher return from
their local stock investments (e.g., Coval and Moskowitz (1999, 2001), Baik, Kang, and Kim
(2010), Bernile, Kumar, and Sulaeman (2012)).
While the evidence of local bias and superior local performance is robust and economi-
cally significant, the channels through which institutions improve the performance of their local
investments is less clear. For example, do local institutions have selective access to superior pri-
vate information? Or, do they earn superior returns from their local investments by processing
publicly available information more effectively?
If selective access to value relevant information was an important determinant of institutional
preferences for local firms, then Reg FD should reduce the local informational advantage and the
local bias in institutional portfolios.2 Moreover, if local institutions had no other competitive
advantage in processing their privately available information, the local ownership-performance
1Ideally, we would also like to study the impact of regulation on the local bias of retail investors. Unfortunately,data on the portfolio holdings of U.S. retail investors around the period of regulatory changes are not available.
2We do not use the term “local bias” to strictly imply that the local stock preference is “irrational”. The“bias” towards local investments could be induced by a familiarity bias or may arise from informational reasons.
1
relation should decline following the implementation of Reg FD.
Local investors, however, could also have a competitive advantage in exploiting publicly
disclosed information, especially if this information is perceived to be of low quality. For example,
to compensate for low disclosure quality, local investors may be able to directly inspect local
firms and could acquire information about local firms’ operations more easily. They may also be
able to collect information from local sources (e.g., customers and suppliers) at a lower cost. In
principle, various provisions of the Sarbanes-Oxley Act (SOX) affected this potential source of
local informational advantage by mandating improvements in the transparency, reliability, and
accountability of financial reporting.
If proximity to the firm’s headquarters and operations indeed reduces investors’ costs of pro-
cessing lower quality public disclosures, then the local informational advantage would be further
weakened following the passage of SOX. And, by the same logic, the perceived improvement
in the reliability of firms’ financial reports should induce investors to tilt their portfolios more
heavily toward non-local investments, which they may have previously shunned.
We use the natural experiment of regulatory changes in the U.S. to identify whether in-
stitutions are able to generate superior returns from their local investments through selective
access to private information or superior processing of publicly-available information. The two
regulatory changes (Reg FD and SOX) affected the access to private information and the quality
of publicly available information. However, these changes had no impact on investors’ ability
to interpret the publicly available information. If local bias of investors is induced by access
to private information, following the regulatory changes the local bias and local performance of
institutions would decline.
Our empirical analysis focuses on the behavior of local investors around corporate head-
quarters as selective access to private information is likely to be more prevalent. However,
motivated by the finding in Bernile, Kumar, and Sulaeman (2012), who demonstrate that in-
stitutions in economically relevant regions earn superior returns although they are unlikely to
have selective access, we also examine whether investors in regions where firms have economic
rather than physical presence are affected by regulatory changes. This analysis allows us to
determine whether selective access to management or superior information processing abilities
is the dominant channel through which local institutions earn higher returns from their local
investments.
2
To further distinguish between selective access and information processing based explana-
tions, we examine the role of other local market participants such as local sell-side equity ana-
lysts. Specifically, we examine whether sell-side analysts who are more likely to have selective
access to firm management also exhibit a declining trend in local coverage and local forecast
accuracy. In this context, we also investigate whether local analysts facilitate the information
gathering activities of local institutional investors.
We test these hypotheses using the quarterly portfolio holdings of 13(f) institutions during
the 1996 to 2008 period. First, we conduct our analysis in a static setting and examine whether
certain features of the information environment in which a firm operates, including local analyst
coverage, can explain the variation in the level of local ownership. Consistent with our main
conjecture, we find that excess local holdings of institutional investors decline sharply following
the regulatory changes. The average excess local institutional ownership (LOCOWN) is cut by
almost one half. In the pre-regulation period (1996-2000), the average excess LOCOWN is 7.68%
and it drops to 3.98% in the post-regulation period (2003-2008).3 Further, the sharp decline in
local institutional ownership around the adoption of Reg FD and SOX does not continue after
2004. The average excess local institutional ownership stabilizes at a level of about 4% in the
post-SOX period.
This observed pattern in excess local ownership is inconsistent with alternative explanations
of our findings that rely on secular time-series patterns in technological advancement, which
could systematically affect firm transparency or institutional preferences and “level the playing
field”. In particular, secular trends in disclosure practices induced by technological innovations
(e.g., cheaper access to the Internet) are unlikely to explain our findings since it is not obvious
why those trends would suddenly stop in the post-regulation period. Similarly, it would be
hard to explain why a trend in institutional preferences would suddenly stop and coincide
with regulatory changes. In particular, we show that our findings do not reflect the changing
preferences of institutional investors during the rise and fall of the technology sector around the
year 2001.
The investment behavior of non-local investors also changes significantly following the regula-
tory reforms. Specifically, during the post-regulation period, non-local ownership levels increase
3These shifts in excess local ownership mainly reflect changes in actual local ownership levels since the expectedlocal ownership levels do not vary significantly over time.
3
and firm ownership becomes more diffused geographically as the information environment im-
proves. This evidence is consistent with the notion that non-local investors avoid stocks of firms
for which they may be at an informational disadvantage and that the new rules reduce those
concerns.
Next, in a dynamic setting, we test whether the relation between information environment
and local bias varies across the regulatory regimes. To ensure that the causal link is from
firm-level information environment to local institutional ownership and not vice versa, we also
estimate a series of difference-in-difference regressions. These tests are designed to identify
whether the impact of the new regulation on the portfolio decisions of local institutional investors
is greater for firms that operated in a more opaque and less competitive information environment
prior to regulatory changes.
We find that while excess local ownership levels vary inversely with the quality of firms’
information environments prior to Reg FD, this relation is significantly weaker in the post-SOX
regime. Finally, consistent with the information environment causing excess local ownership
prior to Reg FD, the firm-level decline in local ownership is more pronounced among firms that
operated in more opaque and less competitive information environments (e.g., large discretionary
accruals, no major auditor, low liquidity, high skewness, no analyst following) prior to the
reforms.
In the second part of the paper, we investigate whether the ability of institutions to generate
abnormal returns from their local investments and their impact on the trading environment are
affected by the reforms. A natural implication of our arguments is that the relation between
excess local ownership and future stock performance should be weaker in the post-regulation
period. In addition, the strength of the relation between excess local ownership and market-wide
measures of adverse selection should decline.
Consistent with these conjectures, we find that there is a sharp decline in the ability of
institutional investors to generate abnormal returns from their local investments. In the pre-
regulation period, there is a significant positive relation between local ownership levels and
subsequent quarter performance of local stocks. But during the post-SOX period, the infor-
mational advantage of local investors disappears. Further tests using market-wide measures
of informed trading indicate that the relation between local bias and subsequent incidence of
informed trading is weaker in the new regulatory regime.
4
Specifically, we use the decomposition of the probability of informed trading (PIN) measure
proposed in Duarte and Young (2009) and find that there is a strong direct relation between the
adverse selection component of PIN (i.e., adjusted PIN or ADJPIN) and local bias in the pre-
FD regime. However, this relation becomes statistically and economically insignificant following
SOX. By contrast, the relation between local bias and the liquidity component of PIN (i.e., the
estimated probability of symmetric order flow shocks) remains strongly positive and significant
throughout the sample period. These results suggest that while the role of local investors as
liquidity providers was not affected by the new rules, their ability to engage in information-based
trading was significantly curtailed in the post-regulation period.
For robustness, we examine the impact of local ownership on subsequent effective trading
spreads based on the assumption that these spreads are directly related to the adverse selec-
tion component of stock prices (e.g., Eleswarapu, Thompson, and Venkataraman (2004)). Once
again, consistent with the idea that local investors possess an informational advantage, we find
that during the pre-regulation period there is a strong direct relation between local bias and sub-
sequent trading spreads. The local bias-spread relation, however, becomes significantly weaker
in the post-regulation period, and particularly outside of earnings announcement windows where
local investors’ ability to exploit local information sources may have been most valuable.
These results indicate that institutional preferences for local stocks is strongly influenced
by the characteristics of firms’ information environments. The unprecedented changes in the
information environment induced by Reg FD and SOX influence the portfolio decisions of in-
stitutional investors significantly. Prior to Reg FD and SOX, institutions seek local investment
opportunities for which they are likely to enjoy a competitive advantage in gathering and pro-
cessing of information. Following the implementation of regulations aimed at “leveling the
information playing field” and designed to improve the quality and reliability of public disclo-
sures, the informational advantage of local institutions disappears and the bias of their portfolios
toward local firms is greatly reduced. As a result, local institutional ownership declines and firm
ownership becomes more diffused geographically.
In the last part of the paper, we investigate whether regulatory changes affect informational
advantage of local investors by curtailing selective access to private information or reducing the
potential information processing advantages of institutional investors. Specifically, we examine
the impact of regulatory changes on two groups of market participants that are ex ante expected
5
to have different degrees of selective access to management. On the one hand, institutions in
regions where firms have economic rather than physical presence are less likely to have selec-
tive access to management as these economically relevant (ER) regions are often physically far
away from the firm headquarters.4 But on the other hand, sell-side analysts around corporate
headquarters are more likely to have benefited from selective access to firm management.
If the main effect of regulatory changes is to curtail selective access to management, we
expect to observe a weaker effect on institutions in economically relevant states and a stronger
effect among sell-side equity analysts. Consistent with this conjecture, we find that the local
informational advantage of sell-side equity analysts who are more likely to have selective access
exhibits a similar declining pattern and disappears in the post-regulation period. In contrast,
regulatory changes have minimal impact on the local informational advantage of institutions
in economically relevant regions as they are less likely to have selective access to management.
These findings indicate that regulatory changes affected institutional behavior primarily by cur-
tailing their selective access to management rather than by adversely affecting their information
processing advantages.
Beyond this identification method, we use a market-based test to further attempt to identify
whether selective access to private information or superior information processing yields greater
local informational advantage to institutional investors. The evidence from this test suggests
that informed trading ahead of earnings announcements, which are likely to be driven by selective
access to private information, are less likely to be associated with local ownership following the
regulatory reforms. This finding is consistent with the conjecture that financial regulation
curtailed local institutional investors’ preferential access to private information.
Collectively, our findings contribute to both the local bias and the regulation literatures. We
show that the local bias of investors around headquarters is likely to be influenced by selective
access to information rather than superior processing of information. At a more fundamental
level, we show how access to different types of information (private and public) and disclosure
policies affect the behavior of market participants (institutions and sell-side analysts) and market
4We identify states that are economically relevant for a firm through a textual analysis of the firm’s annualaccounting reports. This economic relevance measure of a U.S. state for a firm is based on the citation share ofthe state in the firm’s annual accounting reports. The citation share of a firm-state pair is defined as the numberof times the U.S. state is cited in the relevant sections of the annual financial statement of a firm divided by thetotal number of citations of all U.S. locations. Bernile, Kumar, and Sulaeman (2012) provide additional detailsabout this measure.
6
prices.
While some of these findings may appear “obvious” in light of the stated intent of regula-
tory reforms introduced by Reg FD and SOX, the documented link between financial regulation
and local ownership is novel and makes a fundamental contribution to the local bias litera-
ture. In particular, we demonstrate that access to superior private information is an important
determinant of the local bias phenomenon around corporate headquarters.
Further, in a new economic setting, we show that the regulatory reforms have been effective,
although our analysis does not speak to their efficiency. In particular, the declines in the
level of local bias and local informational advantage of institutional investors indicate that
consistent with their stated intent the new rules substantially changed the competitive landscape
of the information acquisition process. Of course, the consequences of regulatory changes on the
behavior of local investors may have been unintended or unanticipated.
Our paper is related to Bernile, Kumar, and Sulaeman (2012) but there are several key
differences between the two studies. Bernile, Kumar, and Sulaeman (2012) examine whether
economic proximity between firms and investors is more valuable than physical proximity. They
find that institutional preference for local firms that are economically present is stronger than
local firms with physical presence. Further, institutional investors in economically relevant
regions outperform investors around firm headquarters. These findings suggest that institutional
investors in economically relevant regions are relatively more sophisticated and they may be able
to use local information more effectively than local investors around corporate headquarters.
In contrast, the primary focus of the current paper is to examine how regulatory changes
affect the information channels of institutions in headquarters states and economically relevant
states. We exploit the observation that the degree of selective access to firm management dif-
fers at these two locations and examine whether selective access to management or superior
processing of publicly available information is the primary source of local informational ad-
vantage of institutional investors. Overall, our findings complement the evidence in Bernile,
Kumar, and Sulaeman (2012) and provide new insights into the channels through which equity
market participants around corporate headquarters and in economically relevant states gather
information.
The rest of the paper is organized as follows. In the next section, we present a brief review
of the related literature and summarize our main testable hypotheses. In Section 3, we describe
7
our data sources. We present our main empirical findings in Sections 4 and 5. We conclude in
Section 6 with a brief summary of the paper.
2. Related Literature and Testable Hypotheses
Our study is motivated by recent studies in financial regulation and local bias. In this section,
we summarize the related regulation and the local bias literatures and use the findings from
those studies to develop our main testable hypotheses.
2.1. Financial Regulation and the Information Environment
Public companies regularly hold conference calls when their quarterly earnings reports are re-
leased to communicate issues that are not covered in those reports. Until the late 1990s, these
conference calls were typically restricted to equity analysts and investors with substantial capi-
tal. Additionally, communication access to top-level corporate executives was severely restricted
to only these large market players. Companies argued that providing the same level of access
to all investors would be too costly.
As cheap access to conference calls proliferated with the advent of the Internet in the late
1990s, this cost argument became less relevant. In response, the U.S. Securities and Exchange
Commission (SEC) initially proposed the Regulation Fair Disclosure (Reg FD) in December
1999. The proposal included new rules prohibiting publicly-traded companies from selectively
disclosing material non-public information to securities markets professionals ahead of general
public disclosure. The general principle advocated by the new rules was that the disclosure of
material information should be made to all investors at the same time.
A debate occurred before the adoption of the new rules as small investors supported the
regulation while large investors argued that forcing managers to provide equal access to all
investors would lead to less access to value-relevant information for all investors (e.g., Weber
(2000a, 2000b), Shiller (2000), SEC (2000), Hasset (2000), Bushee, Matsumoto, and Miller
(2004), Duarte, Han, Harford, and Young (2008)). While acknowledging some of the concerns,
the SEC approved Regulation FD in October 2000. Although the new rules affect all mar-
ket participants to some degree, their impact may have been amplified for particular classes of
investors who enjoyed selective access to value-relevant information in the pre-Reg FD environ-
8
ment. In particular, if proximity to management facilitated selective access to value relevant
information, curbing of selective disclosures may have reduced the competitive advantage of
local institutional shareholders.
The available empirical evidence on the consequences of Reg FD is somewhat mixed. Several
studies find no significant changes in stock price volatility (e.g., Bailey, Li, Mao, and Zhong
(2003)), analysts’ forecast accuracy (e.g., Heflin, Subramanyam, and Zhang (2003)), and effective
trading spreads (Eleswarapu, Thompson, and Venkataraman (2004)), which suggests that Reg
FD did not adversely affect the overall market quality. Consistent with these findings, Bushee,
Matsumoto, and Miller (2004) conclude that Reg FD did not have a large adverse effect on
the disclosure policies of firms that previously allowed selective access to conference calls and
that small investors benefited from the new rules. Duarte, Han, Harford, and Young (2008),
however, document that Reg FD is associated with a slight increase in the typical firm’s cost of
capital. This finding suggests that the benefits to small investors may have come at the expense
of reduced price efficiency.
Around the time of Reg FD, the public’s confidence in the financial market was considerably
affected by fraudulent accounting scandals at Enron, WorldCom, and several other companies.
For example, in the case of WorldCom, fraudulent accounting practices resulted in over-reporting
of assets by around $11 billion. These highly-publicized scandals exposed significant conflicts
of interest problems in the relation between corporations and their financial auditors. At least
partially motivated by these scandals, several new proposals were introduced to curb fraudulent
accounting practices (Holmstrom and Kaplan (2003), Coates (2007)).
The Sarbanes-Oxley Act (SOX) was enacted in July 2002 with widespread approval by
individual investors and investment groups, and large opposition by companies and industry
groups (Hochberg, Sapienza, and Vissing-Jorgensen (2009)). The proponents of these rules
advocated them as necessary to restore the smooth functioning of capital markets, which had
been undermined by the deteriorating quality of accounting information as attested by the series
of unprecedented accounting scandals. In particular, SOX mandated new rules that would
lead to enhanced financial disclosure, greater auditor independence, and improved corporate
governance with the intent to improve the transparency, reliability, and accountability of firms’
financial reports (Coates (2007)).5
5For example, most notably, Sections 101-109 created the Public Company Accounting Oversight Board
9
While the impact of SOX on overall efficiency is still debatable (e.g., Cohen, Dey, and Lys
(2008)), the available evidence indicates that the enhanced quantity and quality of financial
disclosure advocated by this regulation increased the confidence of the public, especially the
confidence of relatively uninformed investors. These improvements in turn should decrease
investors’ perceived costs of relying on firms’ periodic financial reports.
For example, although they reach different conclusions regarding overall efficiency, the evi-
dence in Zhang (2007), Chhaochharia and Grinstein (2007), Li, Pincus, and Rego (2008), and
Iliev (2010) consistently indicates that SOX requirements are likely to be most beneficial to firms
that operated in more opaque information environments prior to the new rules. These conclu-
sions are supported by the recent analysis in Bauguess, Bernile, Lee, Marietta-Westberg, and
Alexander (2012). They find that although compliance costs remain a concern among smaller
firms, corporate insiders’ views on the benefits of SOX rules largely support the notion that they
improve the quality of financial reporting and control processes and, thus, increase investors’
confidence of investors.
Overall, the supporters of the new regulation argue that both Reg FD and SOX lead to
a more level playing field for relatively uninformed investors. At least to some extent, these
regulations should make the information environment more competitive and reduce the infor-
mation asymmetry among investors. In particular, Reg FD prevents companies from disclosing
information to selected market participants and SOX provides confidence to investors as they
examine the financial reports of publicly-traded firms.
2.2. Financial Regulation and Behavior of Local Institutions
Do these regulatory changes adversely affect the informational advantage of local market partic-
ipants such as institutional investors and sell-side equity analysts? Anecdotal evidence suggests
that physical proximity to firm management is likely to yield informational advantage to local
institutional investors. In fact, in its release of Reg FD, the SEC explicitly mentioned that firms
(PCAOB) to provide for auditors’ oversight; Sections 302, 401-406, 408-409, and 906 mandated new disclosurerules pertaining to internal control systems and officer certifications; Sections 201-209 and 303 further regulatedpublic company auditors and auditor-client relationship; Sections 301, 304, 306, and 407 introduced requirementsfor listed companies pertaining to the composition of audit and control committees, and banned officer loans;Sections 802, 807, 902-905, 1102, 1104, and 1106 introduced criminal penalties for fraudulent misreporting; andSections 806 and 1107 introduced new whistle-blower protections.
10
often disclose non-public information to both securities analysts and institutional investors.6
Beyond this anecdotal evidence, previous research on local and home bias suggests that
excess local holdings of institutional investors may reflect superior local information (e.g., Coval
and Moskowitz (1999, 2001), Baik, Kang, and Kim (2010), Bernile, Kumar, and Sulaeman
(2012)). Similarly, sell-side equity analysts may have relatively better information about local
firms (e.g., Malloy (2005)). Local market participants may be able to extract greater benefits
from local stocks with high information asymmetry either due to selective access to local private
information or because geographical proximity reduces their public information gathering and
processing costs.
In this paper, our main objective is to examine whether the exogenous shocks generated by
regulatory changes (Reg FD and SOX) influenced the ownership patterns and informational ad-
vantage of institutional investors. Our identification strategy uses exogenous regulatory changes
as a natural experiment and relies on the time-series properties of local institutional ownership
and local informational advantage as the information environment experiences these large regu-
latory shocks. We focus on the behavior of institutional investors, but we also examine whether
sell-side equity analysts affect institutional local bias and their local informational advantage.
We organize our empirical analysis around three broad sets of hypotheses. These conjectures
summarize the potential impact of regulatory changes on (i) local ownership levels, (ii) local
information advantage, and (iii) various channels through which local market participants obtain
information about local firms.
6Two subsequent SEC enforcement cases provide evidence of selective disclosure practices that may havebenefited local institutional investors. In November 2001, the SEC contested that the CEO of Siebel Systems,a California-based technology company, disclosed nonpublic information to selected investors at an invitation-only technology conference hosted by Goldman Sachs & Company in California. In response to questions froma Goldman Sachs analyst, the CEO announced that he was optimistic about the firm since the business wasreturning to normal. This announcement was opposite to the negative statements made by the CEO threeweeks earlier. Following the disclosure at the conference, attendees purchased Siebel’s stocks or communicatedthe information to others who purchased Siebel’s shares. Following the conference, Siebel experienced a one-day stock return of about 20% and trading volume twice as its daily average. Thus, investors who attendedthe conference enjoyed a substantial informational advantage. In another relevant enforcement case, the SECcontested in March 2002 that the CEO of Secure Computing (John McNulty) disclosed nonpublic informationabout a significant contract to two portfolio managers in violation of the Reg FD. The SEC took exception tothe fact that Secure actively promoted its stock to institutional investors through in-person presentations, inaddition to a series of conference calls and email exchanges with selected investors. Source: Secure ComputingCorporation and John McNulty, Exchange Act Release No. 46895, 2002 WL 310948 (November 25, 2002).
11
2.2.1 Impact on Local Ownership
Before examining the impact of regulatory shocks on the behavior of local market participants,
we identify the types of stocks where local institutional ownership and local informational ad-
vantage of institutional investors are likely to be stronger. Basic economic intuition suggests
that local investors would gravitate toward firms with high information asymmetry to exploit
their potential local informational advantage. Based on this insight, we posit that:
H1a: If institutions exhibit an abnormal preference for local stocks mainly due to an
informational advantage, firms with lower information quality and less competitive
information environments would have stronger local institutional ownership.
If institutional local preferences at least partially reflect investors’ attempts to exploit the
potentially uneven playing field, regulatory changes designed to level the playing field among
different types of investors would influence the local institutional ownership patterns. In par-
ticular, exogenous adverse shocks to the sources of local informational advantage would induce
institutions to reduce their holdings of local stocks. Our specific conjecture is:
H1b: Following regulatory changes, as the information environment becomes more
competitive, local ownership levels around corporate headquarters would decrease
and the overall firm ownership would become more diffused.
If more opaque and less competitive information environments generate higher levels of local
ownership, then the reduction in local holdings resulting from the new rules should be more
pronounced among certain types of firms. This set would include firms for which the local
information advantage was likely higher prior to the adoption of the new rules and, thus, whose
information environment is more likely to be affected by the regulatory changes. More formally,
we posit that:
H1c: The regulatory changes would most strongly affect the local ownership levels
and the local informational advantage among firms that have the worst information
environments during the pre-regulation period.
12
2.2.2 Impact on Local Informational Advantage
While a more opaque and less competitive information environment may attract investors who
can access local sources of information, this effect on local investors’ portfolios should be greatly
reduced or muted if the new regulatory rules level the competition for information acquisition.
Thus, the relation between firms’ information environments and investors’ portfolio performance
should fundamentally change following the passage of the new rules because the rules greatly
reduce or altogether eliminate at least some sources of local investors’ competitive advantage.
To determine whether financial regulation eliminated the local informational advantage of
institutional investors, we perform a series of market-based tests. First, to test whether the
informational advantage of local investors changes around the regulatory reforms, we estimate
the relation between local investors’ holdings and future stock performance conditional on the
regulatory regime. We posit that:
H2a: If the regulatory shocks have a positive impact on the information dissemina-
tion process and an adverse effect on the ability of local investors to obtain private
information, local ownership levels should contain less information about future stock
returns in the post-regulation period.
Second, based on a similar logic, we perform additional tests to assess the conditional relation
between excess local holdings and two common microstructure measures (PIN and effective
spread) that capture the adverse selection component of stock prices. Similar to the performance
tests, we conjecture that:
H2b: Prior to the implementation of the new rules, institutional local bias would
be more strongly related to the subsequent levels of adverse selection measures.
Following regulatory changes, this relation would weaken.
2.2.3 Impact on Local Informational Channels
In the next set of hypotheses, we investigate which source of informational advantage is more
adversely affected by regulatory changes. Regulatory changes could curtail selective access to
private information and they can also reduce the potential superior information processing ability
of institutional investors. Identifying which of these two informational advantage channels is
13
affected more by regulatory changes is an empirical question. Our next two hypotheses aim to
identify the dominant source of local informational advantage of institutional investors.
The first hypothesis related to information channels is based on the assumption that institu-
tions in regions where firms have economic rather than physical presence are less likely to have
selective access to management. These economically relevant (ER) regions are often physically
far away from the firm headquarters. However, institutions in economically relevant regions
may use bits of locally-generated information that could get lost in the information aggregation
process. Overall, the potential informational advantage of institutions in economically relevant
non-HQ locations is unlikely to emerge from selective access to management.
Our specific conjecture is:
H3a: If regulatory changes affect selective access to private information more than
information processing advantage, local ownership and local informational advan-
tage of institutions would be more strongly affected in headquarters states than in
economically relevant states.
We use an additional identification method to determine whether regulatory changes cur-
tailed selective access to information or reduced the information processing advantages of local
market participants. Specifically, we examine how the coverage of stocks by local sell-side equity
analysts and their forecast accuracy change over time. Previous studies use multiple approaches
to demonstrate that a significant determinant of the forecasting behavior of sell-side equity ana-
lysts is selective access to firm management (e.g., Bailey, Li, Mao, and Zhong (2003), Gintschel
and Markov (2004), Agrawal, Chadha, and Chen (2006)). If regulatory changes affect local
informational advantage through the selective disclosure channel more than the information
processing channel, we expect to find similar declining trend in the local coverage of equity
analysts and their local informational advantage, as their selective access to firm management
is curtailed.
H3b: If regulatory changes affect selective access to private information more than
information processing advantage, local coverage of equity analysts and their local
informational advantage would also decline during the post-regulation period.
Beyond testing this conjecture, our empirical tests examine the interactions among local ana-
lysts and local institutional investors. Specifically, we investigate whether regulatory changes
14
influence the distinct information-gathering activities of local analysts and local institutions
separately or whether local analysts are the primary source of institutions’ local informational
advantage. In this scenarios, the declining local informational advantage of institutions could
merely reflect the declining informational advantage of local analysts.
Last, we use a market-based test to further attempt to identify whether selective access to
private information or superior information processing yields greater local informational advan-
tage to institutional investors. If the effects of regulatory changes are mostly due to the reduction
in selective access, the attenuating effect of regulatory changes on the local ownership-adverse
selection relation should be stronger in the period prior to public disclosures such as earnings
announcements. However, if the effects of regulatory changes are mostly due to a reduction in
local investors’ superior ability to process information after it is disclosed, we should observe
a stronger attenuating effect in the period subsequent to public disclosures. To summarize, we
posit that:
H3c: The local ownership-effective spread relation would become weaker in pre-
disclosure periods if regulatory changes affect selective access to private information
more than information processing advantage of publicly available information.
In the next section, we briefly describe the datasets used to test these empirical predictions.
3. Data and Measures
3.1. Main Data Sources
Our core dataset contains all Compustat firms with available 10-K filings in Electronic Data
Gathering, Analysis, and Retrieval system (EDGAR), with fiscal years ending between 1996 and
2008. There are 47,625 firm-year observations in our core dataset. Our second main dataset
is the quarterly common stock holdings of 13(f) institutions compiled by Thomson Reuters.
We identify the institutional location (zip code) using the Nelson’s Directory of Investment
Managers and by searching the SEC documents and web sites of institutional managers. Our
last main dataset is the research coverage of sell-side equity analysts obtained from Thomson
Reuters’ I/B/E/S database. We augment this dataset with analyst location data.
15
In addition to these main data sources, we use several other standard datasets. We obtain
price, volume, return, and industry membership data from the Center for Research on Security
Prices (CRSP). The firm headquarters location data are from the CRSP-Compustat merged
(CCM) file. We obtain the performance benchmarks for computing characteristic-adjusted stock
returns from Russell Wermers’ web site.7 Data on auditors’ identity and various other firm
attributes are from Compustat.
We also use state-level Presidential elections data to identify the political orientation of
U.S. states.8 We obtain additional state-level demographic characteristics from the U.S. Census
Bureau. Specifically, we consider state population density and the state education level (the
proportion of state population above age 25 that has completed a bachelor’s degree or higher) in
our local ownership regressions. Further, using the religious adherence data from the “Churches
and Church Membership” files available through the American Religion Data Archive (ARDA),
we compute the proportion of Catholics (CATH) and the proportion of Protestants (PROT) in
a state. Using the two religion variables, we define the Catholic-Protestant ratio (CPRATIO)
to capture the relative proportions of Catholics and Protestants in a state. We also measure the
overall religiosity of a region.
3.2. Excess Local Ownership Measure
The level of local institutional ownership is at the core of all our testable hypotheses. Thus,
the first set of key measures used in our empirical analysis are the abnormal local ownership
levels of each firm in the headquarters (HQ) state and in the most economically relevant (ER)
non-headquarters states. Following Korniotis and Kumar (2012), we measure the quarterly local
institutional ownership for each firm-state pair as the ratio of the number of firm shares held
by institutions located within the state and the total institutional share ownership at the end
of the quarter.
To provide a benchmark for comparison, we calculate the percentage weight of local institu-
tions in the aggregate institutional portfolio, which is constructed by combining the portfolios
of all institutions. The weight in the aggregate portfolio represents the expected level of owner-
ship by local institutions if they did not exhibit an abnormal preference for firms with physical
7The web site is http://www.smith.umd.edu/faculty/rwermers/ftpsite/Dgtw/coverpage.htm.8The election data are obtained from David Leip’s web site: http://www.uselectionatlas.org.
16
or economic presence in their state and, therefore, their portfolio weights merely reflect the
“market portfolio” weights. We use the state-level benchmarks to account for the non-uniform
geographical distribution of institutions across the U.S. The local ownership benchmarks are
higher for states with greater concentration of institutional investors.
Using the two measures, we define each firm-state pair’s excess local institutional ownership
level (LOCOWN) as the difference between the quarterly local institutional ownership level
of that firm-state pair and the percentage weight of the state’s institutional investors in the
aggregate institutional portfolio:
LOCOWNij =State j institutional ownership in firm i
Total institutional ownership in firm i
− Dollar value of institutional portfolios in state j
Dollar value of aggregate institutional portfolios(1)
In addition to firm-level local ownership measure, we use an institution-level local bias
measure in the empirical analysis. Motivated by previous local bias studies (e.g., Coval and
Moskowitz (2001), Ivkovic and Weisbenner (2005)), we use the following equation to measure
the local bias of institution i around firm headquarters:
HQ Local Biasi = Weight allocated to firms with headquarters in the state
− Market portfolio weight of firms headquartered in the state. (2)
Here, “weight allocated to firms with headquarters in the state” is the dollar value of the
institutional holdings in firms that are headquartered in the state in which institution i is
located divided by the total dollar value of all of holdings of institution i. The second term is
the benchmark weight and is defined as the total market value of firms headquartered in the
state in which institution i is located divided by the aggregate market value of all firms in the
market portfolio. If institution i follows the prescriptions of the traditional portfolio theory and
holds the well-diversified market portfolio, the HQ local bias measure would be zero.
3.3. Information Environment Proxies
Beyond local ownership levels, most of our testable hypotheses relate to the information envi-
ronment of publicly-traded firms. To characterize a firm’s information environment, motivated
17
by the previous literature, we use multiple firm attributes. Our first set of proxies are market-
based measures of information asymmetry. Specifically, we consider the stock turnover measure
(a proxy for liquidity) as better disclosure policies should reduce information asymmetry among
investors and, therefore, increase liquidity (e.g., Glosten and Milgrom (1985), Leuz and Ver-
recchia (2000)). Our second market-based information environment measure is idiosyncratic
volatility since firms with low quality earnings have higher idiosyncratic return volatility (e.g.,
Diamond and Verrecchia (1991), Rajgopal and Venkatachalam (2011)).9 Last, we consider
idiosyncratic skewness to characterize a firm’s information environment because higher informa-
tion ambiguity induces higher volatility and skewness in stock returns (Epstein and Schneider
(2008)).
Our second set of proxies for a firm’s information environment are based on the presence
of information intermediaries. This choice is based on the assumption that the presence of
reputable auditors and other information intermediaries is associated with more transparent
and competitive information environments (e.g., Healy and Palepu (2001)). For instance, ana-
lyst following is directly related to availability of management forecasts (e.g., Soffer, Walther,
and Thiagarajan (2000)) and disclosure quality (e.g., Lang and Lundholm (1993)). In partic-
ular, Malloy (2005) demonstrates that analysts located near firm headquarters have additional
informational advantage.
Further, auditors provide assurance that financial statements comply with accounting stan-
dards. They also alleviate potential concerns about the quality of information disclosed, as
attested by the evidence that hiring a reputable auditor can be beneficial even when not re-
quired by regulation (e.g., Leftwich (1983), Menon and Williams (1991)).
Our last information environment proxy is based on accounting data. Specifically, following
common practice, we use the absolute modified-Jones discretionary accruals as a proxy for
earnings quality. Higher values of this measure are likely to be associated with greater earnings
management and worse information environments (e.g., Cohen, Dey, and Lys (2008)).
Table 1 provides the summary statistics for the main variables used in the empirical analysis.
We report the key statistics for the full sample, pre-Reg FD period (1996 to 2000), post-Reg FD
9Although low earnings quality can lead to higher idiosyncratic volatility, recent work by Hutton, Marcus,and Tehranian (2009) may indicate otherwise. Hutton et al. find that opaque (low quality) financial reports areassociated with higher R2 and, thus, lower scaled idiosyncratic volatility. While we use the unscaled measureof idiosyncratic volatility in our tests similar to Rajgopal and Venkatachalam (2011), we admit that the recentevidence in Hutton et al. suggests that its interpretation may be ambiguous.
18
and pre-SOX period (2001 to 2002), and post-SOX period (2003-2008). The Appendix provides
brief definitions of all these variables.
4. Empirical Results
4.1. Aggregate Time-Series Patterns in Local Ownership
We begin our empirical analysis by presenting a time-series plot of the local ownership level
around firm headquarters. Figure 1 shows the mean excess local ownership (LOCOWN) com-
puted at the end of each fiscal year using the portfolio holdings of all institutional investors in
our sample. These results are also presented in Table 2.10
We find that although the overall level of institutional ownership exhibits an increasing trend
(see Column (1)), the level of local institutional ownership exhibits a significant decline following
the regulatory changes. Consequently, the average level of nonlocal institutional ownership
increases around these changes. The average local ownership level is consistently around 8%
in each year of the pre-regulation period (see Table 2, Column (2)). It drops sharply to below
6% immediately following the passage of Reg FD and then stabilizes at a level of about 4%
in the post-SOX period. Thus, around the regulatory reforms, the excess local ownership of
institutional investors drops by half. These shifts in excess local ownership mainly reflect changes
in actual local ownership levels since the expected local ownership levels do not vary significantly
over time (see Column (3)).
In absolute terms, the drop in the level of local ownership is larger for smaller firms. Specif-
ically, the local ownership levels for small, medium, and large firms drop from 14.42%, 5.86%,
and 2.98% in 2000 to 8.46%, 2.80%, and 1.91% in 2004, respectively. However, in relative terms,
excess local ownership levels drops by about half across all size groups (see Columns (4) to (6)).
We also observe that there is a structural break in the local ownership time series around the
year 2000. Using Chow (1960) tests to identify potential breakpoints during the 1999 to 2007
period, we find that the most significant breaks occur between 2000 and 2001. Our analysis also
indicates that the break seems to happen earlier for larger stocks than for smaller stocks. This
10Discrepancies in mean LOCOWN across Tables 1 and 2 arise because the sample in Table 1 is restrictedto firm-year observations for which we could obtain information environment variables. For the full sample inTable 2, mean LOCOWN by subperiod is: 7.86% prior to Reg FD; 6.252% after Reg FD but prior to SOX; and4.405% after SOX.
19
evidence is consistent with larger firms complying with RegFD earlier than smaller firms due to
their lower relative fixed cost of conducting open calls and their relative lack of concern about
attracting analysts coverage (Goshen and Parchomovsky (2001)).
When we examine local ownership patterns in states where firms have economic rather than
physical presence (see Table 2, Column (7)), we find that the local ownership levels also exhibit
a declining time trend but the drop is less pronounced.11 This evidence indicates that investors
around headquarters and in economically relevant regions are likely to exhibit an abnormal
preference for local stocks for different reasons. Specifically, the local preference of investors
around headquarters is more likely to be induced by access to private information signals. In
contrast, institutions in economically relevant states may hold local stocks because they are
able to generate abnormal returns from their local investments due to their superior ability to
process public information signals.
We also examine the overall geographical shifts in the composition of firm ownership induced
by regulatory changes. We measure geographical ownership concentration of firms using state-
level institutional ownership data. Specifically, we compute two ownership concentration indices.
The Herfindahl-Hirschman concentration index (HHI) of firm ownership is calculated using state-
level institutional ownership levels across all U.S. states. To account for the “natural” level of
concentration in ownership due to the variation in institutional investors’ presence across the
U.S. states, we use a geographical concentration index (GCI), which captures the “abnormal”
level of ownership concentration.12
We find that firm ownership becomes more diffused geographically following regulatory re-
forms. The mean state-level geographic concentration measure declines from 0.171 in 2000 to
0.115 in 2004, which represents a 32.75% decline between the pre- and post-regulation peri-
ods (see Table 2, Column (10)). Beyond 2004, the GCI measures stabilizes around the 2004
level. This evidence supports the notion that the competitiveness of the information envi-
ronment affects the geographical dispersion in ownership. In particular, non-local institutions
appear significantly more reluctant to invest in distant firms due to their potential information
disadvantage in the pre-regulation period. However, following the implementation of Reg FD
11We compute the change in local ownership around regulatory reforms for headquarters states and economi-cally relevant states. The results are reported in Appendix Table A.1. While the local ownership levels aroundcorporate headquarters decline for almost all U.S. states, the changes in economically relevant states exhibit noconsistent pattern.
12This measure is similar to the industry concentration index of Kacperczyk, Sialm, and Zheng (2005).
20
and SOX, institutional ownership becomes geographically more dispersed due to a decline in
information asymmetry induced by regulatory changes.
Collectively, the time-series patterns in local ownership and concentration measures reveal
that regulatory changes are followed by a decline in local ownership levels and a corresponding
increase in non-local ownership levels. As a result, firm ownership becomes more diffused. While
these time-series trends are consistent with our broad conjecture, we exploit the cross-sectional
variation in the information environment to gather additional support for our key hypotheses.
4.2. Information Environment and Local Ownership
We estimate a series of pooled cross-sectional regressions. To set the stage for subsequent
analyses, we first consider a static setting and examine whether features of the information
environment in which a firm operates affect the level of local ownership. The dependent variable
in these regressions is the excess local ownership in a firm. The main explanatory variables are
firm attributes that characterize the firm’s information environment. In addition, following
Bernile, Kumar, and Sulaeman (2012), we include year and state fixed effects to account for the
effects of other firm-level (e.g, firm size, stock price, firm age, etc.) and state-level factors (e.g.,
education, religiosity, population density, etc.) that may be important determinants of local
ownership patterns.
In these regressions, the level of excess local institutional ownership in a firm is measured at
the end of the quarter of the fiscal year end date, while all explanatory variables are measured
during the same fiscal year. Following Petersen (2009), we compute the t-statistics using stan-
dard errors that are adjusted for clustering by year and state. In most of our tests, the sample
period is from 1996 to 2008.13
The regression estimates from models with constant coefficients are presented in Table 3.
Consistent with our conjecture H1a, we find that local ownership levels are higher for firms with
worse information environments. In particular, firms with a major auditor and analyst coverage
have lower levels of local ownership. Interestingly, local ownership level is not significantly asso-
ciated with coverage by local analysts. This evidence suggests that local institutional ownership
levels are not high merely because local institutions are influenced by local analysts.
13The PIN decomposition data from Duarte and Young (2009) are only available for the 1996 to 2004 period.
21
When we consider sets of information environment proxies separately, most information
environment variables are significant determinants of excess local ownership. However, when
we include all information environment proxies in the same specification, not all information
environment proxies have significant coefficient estimates because some of these variables are
correlated. Among the seven information environment proxies, three measures have strongly
significant coefficient estimates (turnover, major auditor, and analyst coverage), two measures
have marginally significant coefficient estimates (idiosyncratic volatility and discretionary ac-
cruals), and two measures have insignificant coefficient estimates (idiosyncratic skewness and
local analyst coverage).
In economic terms, the coefficient estimates are meaningful. For example, having a major
auditor reduces the local ownership level by 3.17% (see Column (5)). Similarly, a one standard
deviation increase in analyst coverage is associated with a 6.394 × 3.34/100 = 0.21% reduction
in local ownership. Relative to the mean local ownership level of 5.39% for the full sample, these
ownership shifts are economically meaningful.
Taken together, the constant coefficients regression estimates broadly support conjecture
H1a, which posits that local institutions find firms with lower quality information environments
more attractive because such firms likely provide them greater opportunities to exploit their
superior information gathering and processing abilities. In addition, we find that local institu-
tions do not merely rely on local information intermediaries such as sell-side analysts for their
investment decisions.
4.3. Impact of Regulatory Reforms on Local Ownership Levels
Next, in a dynamic setting, we investigate whether financial regulation affects the relation be-
tween the information environment and local institutional preferences. These regression models
with time-varying coefficients test our second conjecture (H1b), which posits that the exogenous
shocks to the competitiveness of the information environment generated by regulatory reforms
such as Reg FD and SOX reduces the local informational advantage of institutional investors.
As a result, local ownership levels decline. These effects are expected to be stronger for firms
with worse information environments (H1c).
The regression estimates from models with time-varying coefficients are presented in Table 4.
Consistent with conjecture H1b, we find that local ownership levels fall after the implementation
22
of Reg FD and SOX (see Column (1)). There is a significant (1.74%) drop in local ownership
following Reg FD. This decline gets amplified (average incremental decline is 3.45%) as both
regulatory changes jointly affect the interactions between firm management and investors.
When we include all variables in the regression specification (see Column (3)), consistent with
Hypothesis H1c, the coefficient estimates of most information environment proxies have the pre-
dicted signs and most estimates are statistically as well as economically significant. Specifically,
we find that firms with worse information environments (i.e., those with low turnover, high
volatility, absence of major auditor, no analyst following, or high discretionary accruals) have
significantly higher local ownership levels prior to the rule changes. However, these relations
become significantly weaker after Reg FD and SOX. Two out of seven information environment
proxies have insignificant coefficient estimates but this evidence is not very surprising as these
variables are correlated.
For example, during the pre-regulation period, all else equal, the presence of a major auditor
reduces the local bias by 7.25%. But, after regulatory changes, presence of a major auditor is
associated with a statistically insignificant reduction of 7.25− 6.13 = 1.12% in local ownership.
Similarly, a one standard deviation increase in idiosyncratic volatility is associated with 0.15 ×6.699 = 1.00% higher local ownership in the pre-regulation period. Again, relative to the mean
local ownership level of 5.39%, this effect is economically meaningful. In the post-regulation
period, this positive relation no longer exists. In economic terms, the effect of the presence of a
major auditor on local ownership levels is the strongest but the other information environment
proxies such as turnover, idiosyncratic volatility, and analyst coverage reveal qualitatively similar
patterns.
This overall weakening of the relation between information environment proxies and local
ownership provides support to our first set of conjectures and highlights the adverse impact of
regulatory changes on local stock preferences of institutional investors. Further, the insignificant
relation between local ownership level and local analyst coverage demonstrates that the excess
local analyst coverage and excess local ownership levels of institutional investors are likely to be
induced by distinct sets of factors.
23
4.4. Identifying Causality Using Difference-in-Difference Tests
The evidence so far is consistent with our conjecture that features of the information environment
affect the level of local institutional ownership because local investors may be able to exploit
their informational advantage more effectively when a firm’s information environment is more
opaque. However, the causality may be in the other direction since a firm’s ownership structure
can also potentially affect its information environment. For example, it is possible that a firm
with high local ownership concentration does not require a major auditor because it can easily
communicate with a large proportion of its investor base directly. Similarly, the demand for
analyst services may be lower when a firm’s local ownership levels are high, which make their
services less valuable and reduces analysts’ incentives to cover the firm.
To ensure that the information environment determines excess local ownership and not
vice versa, we adopt a difference-in-difference experimental approach typical of other studies
that examine the effects of financial regulation (e.g., Bushee, Matsumoto, and Miller (2004),
Chhaochharia and Grinstein (2007)). These tests are designed to identify whether the regula-
tory changes have a greater impact on the investment decisions of local investors among firms
that were less transparent prior to the reforms.
The estimates from difference-in-difference regression models are presented in Table 5. The
analysis in this table focuses on the change in firm-level excess local ownership in the years
following the regulatory changes relative to the local ownership level prior to the regulatory
changes. Since our main goal is to characterize the impact of information environment on
ownership composition, the information environment variables in these regressions are measured
at the end of fiscal year 2000.14 We also include state fixed effects in these regressions to control
for anticipated geographical variation in changes in local ownership levels.
Consistent with our main conjecture and our earlier results, we find that the level of local
ownership decreases after regulatory changes. Further, we find that local ownership decreases
more prominently among stocks with poor information environment prior to the regulatory
changes. This set includes firms with low stock turnover, high discretionary accruals, without a
major auditor, or no analyst following.
For example, the estimate of major auditor in Column (3) suggests that the decline in local
14We obtain qualitatively similar results when we repeat the analysis using the 1996-2000 time-series averagesfor the relevant pre-FD information environment and local ownership variables.
24
bias in the 2001-2002 period relative to its 2000 level for firms without a major auditor is about
2.90% more than the corresponding decline for firms with a major auditor. Moreover, there
is an additional incremental decline of 3.07% during the 2003-2008 period. In sum, following
regulatory changes, firms without a major auditor experience a 6.07% greater decline in local
ownership compared to firms with a major auditor.
The results are qualitatively similar with other information environment proxies but not all
estimates are statistically significant. Similar to the evidence in Table 4, idiosyncratic skewness
yields the weakest results. Further, once we account for the overall level of analyst coverage,
local analyst coverage does not incrementally affect local institutional ownership patterns in an
economically meaningful way.
Overall, these results from difference-in-difference regressions provide additional support for
Hypothesis H1c. They indicate that the information environment of a firm affects its local
ownership and not the reverse.
4.5. Regulation and Local Informational Advantage
In the next set of tests, we examine the relation between local institutional ownership and
portfolio performance. Our main objective is to determine whether the local informational
advantage of institutional investors declines following regulatory reforms that were designed to
curb selective dissemination of information and improve the reliability of financial reports.
We estimate a series of performance regressions where the quarterly stock return in the next
quarter is the dependent variable and local ownership level is the main independent variable.
The results are reported in Table 6. Consistent with our Hypothesis H2a, we find that firms
with higher local ownership earn statistically significant higher returns in the future. The results
are similar irrespective of our choice of raw or characteristics-adjusted performance measure. In
economic terms, a one standard deviation increase in local ownership is on average associated
with a quarterly return increase between 0.41% and 0.71% depending on the return measure and
the benchmark. Compared to the means of the various performance measures, these performance
changes are sizeable and economically meaningful.
When we estimate the performance regression with time-varying coefficients to directly cap-
ture the effects of regulation on ownership-performance relation, we find that prior to Reg FD,
the degree of local excess ownership strongly predicts subsequent quarter raw and abnormal
25
returns (see Table 7). The ownership-performance relation is equally strong in the post-Reg
FD, pre-SOX environment. However, as expected, the relation between excess local ownership
and subsequent stock performance disappears in the post-SOX environment. During this period,
local excess ownership has no ability to predict future returns.
The performance regression estimates suggest that while Reg FD may have curbed the set
of profitable investment opportunities enjoyed by local investors, those investors continued to
earn superior risk-adjusted returns on a smaller set of local investments that they continued to
overweight in their portfolios prior to SOX. However, higher quality and more reliable public
disclosures following SOX reduce the local informational advantage of institutional investors.
Overall, the time-series variation in institutional performance supports Hypothesis H2a,
which posits that regulatory changes increase the competitiveness of the information environ-
ments of firms and reduce the informational advantage of institutions around corporate head-
quarters.
4.5.1 Regulation and Ownership-Adverse Selection Relation
In the next set of tests, we examine the relation between local ownership and stock market-
based measures of adverse selection. Similar to the performance tests, our main objective is to
determine whether excess local ownership is associated with the likelihood of future informed
trading and whether this relation is affected by the implementation of Reg FD and SOX rules. A
natural implication of our arguments is that the local ownership-adverse selection relation should
be positive prior to Reg FD and should become significantly weaker or disappear altogether after
the reforms.
To conduct this analysis we rely on two microstructure constructs intended to proxy for
adverse selection in equity trading. The first of these measures is based on the probability of
informed trading (PIN) introduced by Easley, Hvidkjaer, and O’Hara (2002). This measure has
been subsequently refined by Duarte and Young (2009), who decompose PIN into the probabil-
ity of private information-related trade (i.e., ADJPIN) and the probability that a given trade
happens during a symmetric order flow shock (i.e., PSOS).
The second set of microstructure measures is based on trading spreads. Our main measure
is the effective spread, which is the difference between the executed price and the midpoint of
the bid-ask spread. The effective spread captures the price impact of trades, which is related to
26
the market-maker’s (or liquidity provider’s) expectation of trading against informed agents.
Consistent with Hypothesis H2b, we find that the adverse selection components of stock
prices are on average directly and significantly related to prior year-end excess local ownership
levels across the sample period (see Table 8, Panel A). When we condition on the regulatory
regime, however, we find significant differences in the strength of those relations. The results
reported in Table 8, Panel B indicate that after SOX, the relation between ADJPIN and local bias
is weaker compared to the pre-FD period and becomes statistically insignificant. In contrast,
the relation between PSOS and local ownership is positive and significant prior to Reg FD
and appears to be largely unaffected by the new rules. Therefore, while the positive relation
between PIN and the degree of local ownership reflects both adverse selection and liquidity
considerations prior to Reg FD, it is predominantly driven by liquidity consideration in the
post-SOX environment.
The trading spread-based tests support a similar inference. The results from these tests are
reported in Columns (4) to (6) of Table 8, Panels A and B. In the pre-reform period, there is a
significant positive relation between local ownership and subsequent spreads, and this relation
becomes weaker after the reforms. When the spread measure is separately computed during the
3-day window (i.e., −1:+1) around earnings announcements (SPDEARN) and during the non-
earnings-announcement periods (SPDOTH), the results are qualitatively similar. However, there
is some evidence that the attenuating effects of regulatory changes on local ownership-adverse
selection relation may be more acute during periods of less intense public disclosures.
Taken together, the results from these microstructure-based tests indicate that market par-
ticipants expected local investors to exhibit a higher propensity to engage in informed trading
prior to the regulatory reforms.
4.5.2 Institution-Level Performance Estimates
In our performance analysis, we use the firm as the observation unit because our main objective
is to examine the effects of firm characteristics on the decline in local bias around regulatory
changes. For robustness, we use an alternative method where each institutional investor is the
observation unit. Using this alternative approach, Coval and Moskowitz (2001) reports that
mutual fund managers tend to have a higher portfolio weight assigned to local stocks. In our
next set of tests, we examine the local bias and local performance at the institution level. We
27
repeat the Coval and Moskowitz (2001) analysis for our sample of institutional investors.
Specifically, we calculate the Excess Local Weight of each institutional investor’s quarterly
portfolio snapshot as the percentage of the investor’s portfolio invested in stocks located in
the investor’s state (local stocks) minus the percentage of the “market portfolio” located in the
investor’s state. We compute the average of excess local weights across all institutional investors’
portfolios each quarter, and then report the time-series average of those quarterly averages over
various subperiods in Table 9. The quarterly averaging across institutions is either weighted by
the total dollar value of the institution’s holdings at the beginning of the quarter (Panel A) or
weighted equally (Panel B).
Using holdings-weighted average, we find that the excess local weights of institutional in-
vestors’ portfolios decline by more than 35 percent in the post-regulation period (2003-2008)
relative to the pre-regulation subperiod (1996-2000). These results are similar to the evidence
obtained using firm-level analysis (e.g., see Figure 1).
We also repeat the analysis in Coval and Moskowitz (2001) regarding the performance of
institutional investors’ local and non-local holdings. For each institution with non-zero portfolio
weight in local stocks, we calculate the quarterly characteristics-adjusted returns of its local
and non-local portfolios, as well as the performance differential (local minus non-local). Local
Portfolio is comprised of stocks in the institutional investor’s portfolio that are located in the
investor’s state, while Non-Local Portfolio is comprised of the rest of the institution’s holdings.
The portfolio returns are adjusted using the Daniel, Grinblatt, Titman, and Wermers (1997)
approach to control for variations in size, book-to-market, and past 12-month return.
As reported in Table 9, local holdings of institutional investors outperformed their non-local
holdings by a significant margin prior to regulatory changes (1996-2000). Using dollar-weighted
average, local holdings outperform non-local holdings by about 54 basis points each quarter
or about 2 percent annually. However, the superior performance of local holdings disappears
following the regulatory changes. The performance differential is not significantly different from
zero and the point estimates are generally negative in the post-regulation subperiod. These
results are consistent with our evidence obtained using firm-level observations and provide ad-
ditional support to Hypothesis H2a. Thus, our performance results are not driven by the choice
of unit of aggregation (i.e., firm or institution).
28
4.6. Robustness Checks
In this section, we examine the robustness of our main results. In the first set of tests, we examine
whether our findings indicate an increased awareness against holding employer stocks rather than
regulation-induced decline in local ownership. To alleviate concerns about the incidence of firm
employees’ stock holdings on our findings, we recalculate the excess local ownership measure
after excluding institutional investors that are classified as public or corporate pension funds,
university endowments, or miscellaneous. We repeat our main tests using this more restrictive
measure of local ownership and report the results in columns labeled “No PF” in Tables 4 to
7. Our conclusions remain unaffected and some of the earlier results become more significant
statistically.
Next, we focus on the effect of low priced stocks on our results. Low priced stocks are
typically characterized by more volatile and skewed returns, and are more likely to be subject to
potential microstructure biases (e.g., large bid-ask spread), which may have a disproportionate
impact on our empirical findings. To assess the extent to which our main results are driven
by the unique attributes of low-priced stocks, in Column (6) of Tables 4 and 5, and Column
(7) of Table 7, we report the regression estimates for the subsample of stocks that are priced
above $2. As shown in the tables, we find that our results are fairly robust to imposing this
restriction. This evidence indicates that the patterns of disappearing local ownership and local
informational advantage are not restricted to the subsample of low priced stocks.15
Another potential concern stems from the uneven geographical distribution of firms’ head-
quarters and centers of economic activity across the U.S. The top two states, California and
New York, on average account for approximately 11% of the firms in our sample on an annual
basis. The economic forces at play at these locations may be unique. In particular, there is a
high concentration of technology firms in California, for which information asymmetries may be
more severe. In addition, the pre-regulation period corresponds to the Internet frenzy period,
which could also influence our results. Further, the high concentration of financial institutions
and large brokerage houses in New York suggests that the effects of Reg FD may have stronger
impact on the ability of New York based analysts and institutions to have direct access to
corporations located in the state of New York.
15We also restricted the sample using a $5 price filter and obtain qualitatively similar results, although thepower of our tests diminishes because the sample size decreases by approximately 15%.
29
To ensure that our results do not primarily reflect the unique information environment in
the states of New York and California, in Column (7) of Tables 4 and 5, and Column (8) of
Table 7, we report the regression estimates for subsamples of firms whose headquarters are not
in California or New York. These subsample results are qualitatively similar to the full-sample
results, which indicates that the phenomenon of disappearing local information is not restricted
to the two most business populated states.
As mentioned above, the adoption of Reg FD in October 2000 coincides with the burst
of the tech bubble. To ensure that our results do not reflect the sharp reduction in (local)
institutional ownership in technology firms around this burst, we repeat our main tests by
excluding technology firms and report the results in columns labeled “No Tech” in Tables 4
to 7. Our conclusions remain unaffected and the magnitudes and statistical significances of
regression estimates are similar.
5. Dominant Local Information Channel
Our results so far indicate that exogenous regulatory shocks induced by Reg FD and SOX reduce
both the local bias and the local informational advantage of institutional investors. In this sec-
tion, we investigate whether selective access to management or superior information processing
is the primary channel through which institutions acquire local information. Our strategy is to
examine the impact of regulatory changes on two groups of local market participants that are
ex ante expected to have different degrees of private access to management.
On the one hand, institutions in regions where firms have economic rather than physical
presence are less likely to have selective access to management as these economically relevant
(ER) regions are often physically far away from the firm headquarters.16 On the other hand, sell-
side analysts are more likely to have benefited from private access to firm management. Thus,
if regulatory changes curtail selective access to management, we expect to observe a weaker
effect on institutions in economically relevant states and stronger effect among sell-side equity
analysts. These conjectures are summarized as Hypotheses H3a and H3b.
16See footnote 4 for a definition of economically relevant region.
30
5.1. Impact of Regulation in Economically Relevant States
To test Hypothesis 3a, we examine the ownership shifts and changes in local informational
advantage of institutions in economically relevant (ER) regions following the regulatory changes.
We replace the local ownership and local performance measures computed using institutional
holdings around firm headquarters by corresponding measures computed using the portfolio
holdings of institutions located in the firm’s most economically relevant non-HQ state. The
results are presented in “ER” columns in Tables 2 to 7.
In comparison to the headquarters locations, economically relevant regions exhibit a consid-
erably weaker shift in the local ownership levels. The excess local ownership level exhibits a
mild decline from about 2% to 1% (see Table 2, Column (7)). Since the mean expected level
of ownership in ER states is 8.37%, which does not exhibit much time variation, the total local
ownership drops from about 10% to 9%.
Examining the impact of firm-level information environment on local institutional owner-
ship in ER locations, we find that local ownership in ER states is not systematically higher for
firms that operate in worse information environments. The evidence in Table 3, Column (6)
indicates that two information proxies (analyst coverage and discretionary accruals) are statisti-
cally significant but they portray an inconsistent picture. Further, following regulatory changes,
unlike the headquarters location, the local ownership level at the ER locations does not exhibit
a sharper decline for firms that operate in worse information environments (see Tables 4 and 5,
Column (4)). Last, we find that following regulatory changes the decline in local informational
advantage in ER states is imperceptible (see Tables 6 and 7, Column (5)).
These ER results are different from the evidence obtained using local ownership levels around
headquarters. Consistent with our conjecture that institutions at ER locations are less likely
to have selective access to management, the local institutional ownership levels and local infor-
mational advantage of institutions are minimally affected by regulatory changes. Overall, these
findings provide support to Hypothesis H3a.
5.2. Regulation, Local Analysts, and Local Investors
To further identify the channels through which institutions around headquarters gather informa-
tion about local firms, we study the behavior of local equity analysts and test Hypothesis H3b.
31
If regulatory changes curtail selective access to management, sell-side analysts are likely to ex-
perience a stronger adverse effect as past evidence indicates that they have often benefited from
access to management. This analysis also investigates whether the local informational advan-
tage of institutional investors merely reflect the known local informational advantage of equity
analysts (Malloy (2005)). If institutions merely follow the recommendations of local analysts
and regulatory changes affect the local informational advantage of equity analysts, then follow-
ing the implementation of financial regulation, the local informational advantage of institutional
investors would be reduced and their preferences for local stocks would weaken.
5.2.1 Regulation and the Behavior of Local Analysts
We start by examining analysts’ propensity to cover local stocks. We define the sell-side analyst’s
excess local coverage as the fraction of local stocks covered by the analyst minus the fraction of
local stocks available to the analyst. Unlike institutional investors, there is no obvious weighting
scheme for analyst coverage. Therefore, we examine both equal- and value-weighted fractions.
The last two columns of Table 2 report the annual mean excess local coverage estimates
for both headquarters states and the most economically relevant states. We find that sell-side
analysts around corporate headquarters have excess propensity to provide local coverage and
their excess propensity to cover local stocks declines with the regulatory changes. Prior to
RegFD, we find that the value- (equal-)weighted excess local coverage is about 8% (11%) and,
following SOX, the excess local coverage declines by about two-fifths to 4.5% (value-weighted)
or about 7% (equal-weighted). During this period, the expected local coverage measure does not
exhibit much time variation and has a mean of 7.40%. In contrast, the local analyst coverage
in economically relevant states is not significantly affected by regulatory changes (see Table 2,
Column (12)).
In Table 10, we perform formal statistical tests of this decline by regressing excess local
coverage on indicator variables for FDSOX (2001-2002) and POSTSOX (2003 onwards), analyst
characteristics, and state fixed effects. The analyst characteristics include analyst tenure, past
performance, the number of firms covered, and an indicator variable for analysts employed by
prestigious brokerage houses. The average excess local coverage throughout our sample period
is 9.63 (6.30) percent using equal- (value-) weighting.
Further, the coefficient estimates of FDSOX and POSTSOX suggest that, relative to the
32
pre-regulatory period, excess local coverage declines by 1.24 to 1.75 percentage points during
FDSOX and by 3.21 to 4.68 percentage points after SOX. Focusing on the analysts employed
by prestigious brokerage houses, we find that they are less likely to cover local stocks. However,
their excess local coverage after SOX is similar to that of analysts employed by less prestigious
brokerage houses.
We next examine the forecasting performance of local analysts. In particular, we examine
their forecast accuracy and the stock price impact of revisions in their quarterly earnings fore-
casts. As analysts are evaluated against their peers, we employ two relative accuracy measure
that compares an analyst’s forecast against all forecasts on the same earnings report. The first
measure is whether or not the analyst’s forecast error is below the median error, while the sec-
ond measure is the percentile ranking of the analyst’s forecast error. Following Malloy (2005),
we also use the demeaned absolute forecast error (DAFE) to measure forecast accuracy, where
a negative DAFE implies better than average forecast accuracy and a positive DAFE reflects
below-average forecast accuracy. In addition, we use the short-term market reaction during the
three-day (−1, 0, 1) window around quarterly forecast revisions to measure the informativeness
of analyst forecasts. Specifically, it is calculated as the three-day buy-and-hold return of the
firm’s stock minus the buy-and-hold return of the equal-weighted CRSP index.
Table 11 reports the coefficient estimates from regressing these accuracy and market reaction
measures on a LOCAL indicator variable that takes the value of one if the analyst and the firm
headquarters are located in the same state and zero otherwise, its interaction with indicator
variables for FDSOX (2001-2002) and POSTSOX (2003 onwards), and time-varying analyst
fixed effects to capture variation in analysts’ skill and characteristics.17 We do not include
stand-alone regime variables or time fixed effects in the specifications as they are subsumed by
the time-varying analyst fixed effects. We cluster the errors for all forecasts issued for the same
earnings report as they are correlated due to the way our measures are constructed.
The coefficient estimates in Table 11 suggest that local analysts are more accurate prior
to the introduction of Reg FD, consistent with the findings in Malloy (2005). However, the
superior accuracy of local analysts decline significantly following Reg FD. Adding up the point
estimates of LOCAL and LOCAL × POSTSOX in Columns (1), (2), or (5), we find that superior
17The LOCAL indicator is defined in an analogous manner for economically relevant states. It takes the valueof one if the analyst is located in the economically most relevant state for the firm and, zero otherwise.
33
local performance is at most negligible after the introduction of SOX. In contrast, local analysts
in economically relevant states did not exhibit higher local accuracy during the pre-regulation
period and their local forecast accuracy is not significantly affected by regulatory changes (see
Columns (3) and (4)).
When we estimate the accuracy regressions for different subperiods, similar to the evidence
in Malloy (2005), LOCAL has a significantly negative coefficient estimate in the pre-regulation
period (see Column (6)), and negative but insignificant coefficient estimate during the period
immediately following Reg FD (see Column (7)). However, when we extend the time period
and consider the forecast accuracy of local analysts during a longer 2003-08 period, we find that
LOCAL has a positive but insignificant coefficient estimate. This evidence indicates that local
analysts no longer possess a local informational advantage and, in fact, on average they are less
accurate than nonlocal analysts although this relation is not statistically significant.
Collectively, the results in Tables 10 and 11 are consistent with our conjecture (H3b) that Reg
FD and SOX affect local sell-side equity analysts since prior to the introduction of these financial
regulation, like local institutional investors, they were more likely to rely on selective access to
management in their information-gathering activities. Following the regulatory changes, like
institutional investors, their local informational advantage disappears.
5.2.2 Local Analysts and Local Investors
In this section, we examine whether the local informational advantage of analysts and insti-
tutional investors are related. Specifically, we investigate whether the recommendations from
local analysts drive the local informational advantage of institutional investors. We estimate
the local bias and performance of institutional investors, conditional upon the presence of local
analysts. We sort states into three groups based on the number of analysts in the state: top 5
states, 6th to 20th, and others. This grouping is done to ensure that institutional investors are
spread as evenly as possible across the three groups. within each group, we measure the local
bias of institutional investors as well as their portfolio performance during different subperiods.
The results are presented in Table 12.
We find that the pattern of declining local bias is similar, irrespective of the level of local
analyst coverage (see Panel A). However, the time-series patterns in local performance is quite
different. In states with high number of local analysts, institutional investors do not display
34
superior performance both before and after the regulatory changes. In contrast, institutions
exhibit superior local performance in states with low analyst presence prior to the regulatory
changes. The superior performance of local institutional investors weakens significantly following
the regulatory changes as access to superior local information becomes harder to obtain.
This evidence suggests that local informational advantage of institutional investors do not
merely reflect the local informational advantage of equity analysts. They are likely to compete
for local information prior to the regulatory changes rather than share it. The relative increase
in the local coverage by analysts at prestigious brokerage firms (see Table 10, Columns (2) and
(4)) is also consistent with our conjecture that the local biases of analysts and institutions are
not directly related.
For comparison, we examine the performance of institutional investors among stocks with
local economic presence (see Panel B).18 We find that the superior performance of institutional
investors among these stocks is not significantly affected by the regulatory changes, irrespective
of the level of local analyst coverage. This evidence indicates that the informational advantage
of institutional investors in economically relevant regions is distinct from their local informa-
tional advantage around corporate headquarters. Neither the presence of local analysts nor the
introduction of financial regulation completely eliminated this informational advantage.
5.3. Identifying Selective Access Using a Microstructure Test
In our last set of tests, we use a market-based tests to further identify whether selective access to
firm management is the primary channel through which local investors around corporate head-
quarters obtain a local informational advantage. The evidence in Panels A and B of Table 8
indicates that the attenuating effects of regulatory changes on local ownership-adverse selection
relation may be more acute during periods of less intense public disclosures. This attenuating
effect may be due to the reduction in selective disclosure (i.e., preferential access to firm infor-
mation) provided to local investors. Alternatively, it may be a result of the reduction in local
investors’ superior ability to process information after it is publicly disclosed by the firm. The
former suggests that the attenuating effect should be more severe prior to public disclosure,
while the latter suggests that the effect should more severe following the disclosure.
18The results are reported for the five most economically relevant states but the results are very similar whenwe only consider the most economically relevant states.
35
In Table 8, Panel C, we report results from spread-based tests that allow us to distinguish
between these two potential explanations. Specifically, we report the relation between local own-
ership and spreads in the periods before and after earnings announcements. While local own-
ership is associated with higher spreads both before and after earnings announcement, we find
that the attenuating effects of regulatory changes on local ownership-adverse selection relation
are restricted to the pre-earnings announcement period. This finding suggests that regulatory
changes reduced local investors’ preferential access to firm information, which is one of the main
objectives of Reg FD. The evidence also suggests that informed trading by local institutional
investors prior to regulatory reforms may be driven by preferential access to private information,
which was curbed following the introduction of Reg FD. These results provide support for our
last Hypothesis (H3c).
Overall, we use three different methods to demonstrate that the local informational advantage
of institutional investors is more likely to reflect selective access to firm management prior
to regulatory changes. This type of local informational advantage disappeared following the
implementation of Reg FD, although it took several years before the impact of Reg FD was
fully realized. In contrast, the improvement in the informational environment due to SOX had
minimal impact on the information processing advantage of institutional investors. Specifically,
the informational advantage of institutions in economically relevant regions was not adversely
affected in spite of an improvement in the information environment.
6. Summary and Conclusion
This study examines whether exogenous regulatory shocks induced by Regulation Fair Disclo-
sure and the Sarbanes-Oxley Act reduce the local bias and the local informational advantage
of institutional investors. We also investigate whether regulation affects the informational ad-
vantage of institutions by curtailing selective access to management or adversely affecting their
information processing advantages. Although we focus on the behavior of institutional investors,
we study how the behavior of local analysts affects local institutional ownership patterns and
local performance.
We find that following the adoptions of Reg FD and SOX average firm-level local bias is cut
by about half and firm ownership becomes more diffused geographically. Further, the local in-
36
formational advantage of institutional investors around corporate headquarters declines sharply
as their selective access to private information is curtailed. The decline in local institutional
ownership is more salient among firms that had more opaque information environments prior
to the regulatory changes. Even at the aggregate market-level, the degree of informed trading
attributed to local investors declines.
The local informational advantage of sell-side equity analysts who are more likely to have
selective access exhibits a similar declining pattern and disappears in the post-regulation period.
In contrast, regulatory changes have minimal impact on the local informational advantage of
institutions in economically relevant regions as they are unlikely to have selective access to
management.
Taken together, these findings indicate that regulatory changes affected institutional behavior
primarily by curtailing their selective access to management rather than adversely affecting their
information processing advantages. Our results contribute to both the regulation and the local
bias literatures. We demonstrate that selective access to firm management is an important
determinant of institutional local bias and also show in a new economic setting that financial
regulation can be effective.
These results provide several pointers for future research. For example, while we present
evidence of sharp decline in local ownership levels, a significant level of local institutional pref-
erences remain. While a significant proportion of this “residual” excess local ownership may
be induced by the familiarity, some sophisticated institutional investors might overweight local
stocks even after the leveling of the playing field due to their superior abilities to process publicly
available information. It would be useful to quantify these clientele shifts and examine their
potential impact on asset prices.
It would also be interesting to investigate whether and to what extent the staggered imple-
mentation of SOX-like regulations in several major capital markets around the world affected
home bias – the international counterpart of the local bias phenomenon. If, like in the United
States, the information environment affects both domestic and foreign institutional investors’
participation in equity markets, we should observe a substantial decline in home bias and a
concurrent inflow of foreign investors’ capital in those countries that adopted SOX-like rules.
37
References
Agrawal, A., S. Chadha, and M. A. Chen, 2006, “Who Is Afraid of Reg FD? The Behavior
and Performance of Sell-Side Analysts Following the SEC’s Fair Disclosure Rules,” Journal
of Business, 79, 2811–2834.
Baik, B., J. Kang, and J. Kim, 2010, “Local Institutional Investors, Information Asymmetries,
and Equity Returns,” Journal of Financial Economics, 97, 81–106.
Bailey, W., H. Li, C. X. Mao, and R. Zhong, 2003, “Regulation Fair Disclosure and Earnings
Information: Market, Analyst, and Corporate Responses,” Journal of Finance, 58, 2487–2515.
Bauguess, S., G. Bernile, A. Lee, J. Marietta-Westberg, and C. Alexander, 2012, “The Economic
Effects of SOX Section 404 Compliance: A Corporate Insider Perspective,” Working Paper,
University of Miami.
Bernile, G., A. Kumar, and J. Sulaeman, 2012, “Home Away From Home: Eco-
nomic Relevance and Local Investors,” Working Paper (December); Available at SSRN:
http://ssrn.com/abstract=1711653.
Bushee, B., D. Matsumoto, and G. Miller, 2004, “Managerial and Investor Responses to Dis-
closure Regulation: The Case of Reg FD and Conference Calls,” The Accounting Review, 79,
617–643.
Chhaochharia, V., and Y. Grinstein, 2007, “Corporate Governance and Company Value: The
Impact of the 2002 Governance Rules,” Journal of Finance, 62, 1789–1825.
Chow, G. C., 1960, “Tests of Equality Between Sets of Coefficients in Two Linear Regressions,”
Econometrica, 28, 591–605.
Coates, J., 2007, “The goals and promise of the Sarbanes-Oxley Act,” Journal of Economic
Perspectives, 21, 91–116.
Cohen, D. A., A. Dey, and T. Z. Lys, 2008, “Real and Accrual-Based Earnings Management in
the Pre- and Post-Sarbanes Oxley Periods,” The Accounting Review, 83, 757–787.
38
Coval, J. D., and T. J. Moskowitz, 1999, “Home Bias at Home: Local Equity Preference in
Domestic Portfolios,” Journal of Finance, 54, 2045–2073.
Coval, J. D., and T. J. Moskowitz, 2001, “The Geography of Investment: Informed Trading and
Asset Prices,” Journal of Political Economy, 109, 811–841.
Daniel, K. D., M. Grinblatt, S. Titman, and R. Wermers, 1997, “Measuring Mutual Fund
Performance with Characteristic-Based Benchmarks,” Journal of Finance, 52, 1035–1058.
Dechow, P. M., R. G. Sloan, and A. P. Sweeney, 1995, “Detecting Earnings Management,”
Accounting Review, 70, 193–225.
Diamond, D. W., and R. E. Verrecchia, 1991, “Disclosure, Liquidity, and the Cost of Capital,”
Journal of Finance, 46, 1325–1359.
Duarte, J., X. Han, J. Harford, and L. Young, 2008, “Information Asymmetry, Information
Dissemination and the Effect of Regulation FD on the Cost of Capital,” Journal of Financial
Economics, 87, 24–44.
Duarte, J., and L. Young, 2009, “When 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.
Eleswarapu, V., R. Thompson, and K. Venkataraman, 2004, “The Impact of Regulation Fair
Disclosure: Trading Costs and Information Asymmetry,” Journal of Financial and Quantita-
tive Analysis, 39, 209–225.
Epstein, L. G., and M. Schneider, 2008, “Ambiguity, Information Quality, and Asset Pricing,”
Journal of Finance, 63, 197–228.
Gintschel, A., and S. Markov, 2004, “The Effectiveness of Regulation FD,” Journal of Accounting
and Economics, 37, 293–314.
Glosten, L. R., and P. R. Milgrom, 1985, “Bid, Ask and Transaction Prices in a Specialist Market
with Heterogeneously Informed Traders,” Journal of Financial Economics, 14, 71–100.
39
Goshen, Z., and G. Parchomovsky, 2001, “On Insider Trading, Markets, and “Negative” Prop-
erty Rights in Information,” Virginia Law Review, 87, 1229–1277.
Hasset, K., 2000, Outlaw Selective Disclosure? No, the More Information the Better, The Wall
Street Journal (August 10): A18.
Healy, P. M., and K. G. Palepu, 2001, “Information Asymmetry, Corporate Disclosure, and the
Capital Markets: A Review of the Empirical Disclosure Literature,” Journal of Accounting
and Economics, 31, 405–440.
Heflin, F., K. R. Subramanyam, and Y. Zhang, 2003, “Regulation FD and the Financial Infor-
mation Environment: Early Evidence,” The Accounting Review, 78, 1–37.
Hochberg, Y., P. Sapienza, and A. Vissing-Jorgensen, 2009, “A Lobbying Approach to Evaluat-
ing the Sarbanes-Oxley Act of 2002,” Journal of Accounting Research, 47, 519–583.
Holmstrom, B., and S. Kaplan, 2003, “The State of U.S. Corporate Governance: What’s Right
and What’s Wrong?,” Journal of Applied Corporate Finance, 15, 8–20.
Iliev, P., 2010, “The Effect of SOX Section 404 Compliance on Audit Fees, Earnings Quality
and Stock Prices,” Journal of Finance, 65, 1163–1196.
Ivkovic, Z., and S. Weisbenner, 2005, “Local Does as Local Is: Information Content of the
Geography of Individual Investors’ Common Stock Investments,” Journal of Finance, 60,
267–306.
Kacperczyk, M., C. Sialm, and L. Zheng, 2005, “On the Industry Concentration of Actively
Managed Equity Mutual Funds,” Journal of Finance, 60, 1983–2011.
Korniotis, G. M., and A. Kumar, 2012, “State-Level Business Cycles and Local Return Pre-
dictability,” Journal of Finance, Forthcoming.
Kumar, A., 2009, “Who Gambles in the Stock Market?,” Journal of Finance, 64, 1889–1933.
Lang, M., and R. Lundholm, 1993, “Cross-sectional Determinants of Analyst Ratings of Corpo-
rate Disclosures,” Journal of Accounting Research, 31, 246–271.
40
Leftwich, R., 1983, “Accounting Information in Private Markets: Evidence from Private Lending
Agreements,” The Accounting Review, 58, 23–42.
Leuz, C., and R. E. Verrecchia, 2000, “The Economic Consequences of Increased Disclosure,”
Journal of Accounting Research, 38.
Li, H., M. Pincus, and S. O. Rego, 2008, “Market Reaction to Events Surrounding the Sarbanes-
Oxley Act of 2002 and Earnings Management,” Journal of Law and Economics, 51, 111–134.
Malloy, C. J., 2005, “The Geography of Equity Analysis,” Journal of Finance, 60, 719–755.
Menon, K., and D. D. Williams, 1991, “Auditor Credibility and Initial Public Offerings,” The
Accounting Review, 66, 312–332.
Newey, W. K., and K. D. West, 1987, “A Simple, Positive semi-Definite Heteroskedasticity and
Auto-Correlation Consistent Variance-Covariance Matrix,” Econometrica, 55, 703–708.
Petersen, M. A., 2009, “Estimating Standard Errors in Finance Panel Data Sets: Comparing
Approaches,” Review of Financial Studies, 22, 435–480.
Rajgopal, S., and M. Venkatachalam, 2011, “Financial Reporting Quality and Idiosyncratic
Return Volatility,” Journal of Accounting and Economics, 51, 1–20.
SEC, 2000, Selective Disclosure and Insider Trading, Release No. 33-7881 and 34-43154; Avail-
able at http://www.sec.gov/rules/final/33-7881.htm.
Shiller, R., 2000, Outlaw Selective Disclosure? Yes, Markets Must be Fair, The Wall Street
Journal (August 10): A18.
Soffer, L. C., B. R. Walther, and S. R. Thiagarajan, 2000, “Earnings Pre-Announcement Strate-
gies,” Review of Accounting Studies, 5, 5–26.
Weber, J., 2000a, Full Disclosure For All, Business Week (September 18): 106.
Weber, J., 2000b, Why Should Analysts be Told More Than You? Business Week (March 13):
42.
Zhang, I., 2007, “Economic Consequences of the Sarbanes-Oxley Act of 2002,” Journal of Ac-
counting and Economics, 44, 74–115.
41
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20080
2
4
6
8
10
12
14
16
Exc
ess
Loc
al O
wne
rshi
p
Year
All FirmsSmall FirmsLarge Firms
Figure 1. Excess local ownership around firm headquarters. This figure shows the mean excess equal-weighted local ownership time series for the full sample of firms. Excess local ownership is the state institutionalownership of a firm’s HQ state minus that state’s average state institutional ownership. Small and large firmssubsamples contain firms in the lowest and highest firm size quartiles, respectively. Additional details about thevariables are available in the Appendix.
42
Table
1
Sum
mary
Sta
tist
ics
Thi
sta
ble
repo
rts
the
sum
mar
yst
atis
tics
for
mai
nva
riab
les
used
inth
eem
piri
cal
anal
ysis
.T
heva
riab
les
are
defin
edin
the
App
endi
x.T
hefu
llsa
mpl
epe
riod
isfr
om19
96to
2008
.P
RE
SOX
peri
odin
clud
esfir
m-fi
scal
year
sen
ding
star
ting
in19
96an
den
ding
prio
rto
orin
cale
ndar
year
2000
.FD
SOX
peri
odin
clud
esfir
m-fi
scal
year
sen
ding
inca
lend
arye
ars
2001
and
2002
.P
OST
SOX
peri
odin
clud
esfir
m-fi
scal
year
sen
ding
inca
lend
arye
ar20
03on
war
d.T
hede
finit
ions
ofal
lvar
iabl
esar
epr
ovid
edin
the
App
endi
x.*,
**,an
d**
*in
dica
test
atis
tica
lsi
gnifi
canc
eat
the
1%,5%
,an
d10
%le
vels
,res
pect
ivel
y.
Full
Sam
ple:
1996
-200
8P
RE
SOX
:19
96-2
000
FD
SOX
:20
01-0
2P
OST
SOX
:20
03-0
8
Var
iabl
eSh
ort
Nam
eN
Mea
nM
edia
nSt
dD
evM
ean
Med
ian
Mea
nM
edia
nM
ean
Med
ian
Owne
rshi
p,C
once
ntra
tion
,an
dPer
form
ance
Mea
sure
s
Loc
alO
wne
rshi
pH
QLO
CO
WN
47,6
255.
390
0.30
016
.870
6.76
00.
660
5.25
00.
190
4.27
00.
140
Loc
alO
wne
rshi
pE
RR
ER
LO
CO
WN
44,0
401.
300
−0.3
4012
.590
1.77
0−0
.330
1.26
0−0
.370
0.93
0−0
.330
Ow
ners
hip
Her
finda
hlH
HI
47,6
250.
246
0.19
90.
174
0.26
40.
213
0.25
10.
202
0.22
90.
190
Ow
ners
hip
Geo
.C
onc.
GC
I47
,625
0.13
10.
075
0.16
10.
151
0.09
90.
142
0.08
10.
110
0.05
8
Nex
tQ
uart
erR
etur
n:
Raw
Ret
urn
RAW
RE
T44
,546
3.29
00.
700
32.5
705.
110
0.85
01.
500
−0.2
702.
200
0.90
0
CR
SP-E
WA
dj.
EW
RE
T44
,546
−0.3
80−2
.760
31.0
50−0
.550
−4.4
30−0
.490
−2.2
10−0
.170
−1.8
00
CR
SP-V
WA
dj.
VW
RE
T44
,546
2.34
0−0
.450
31.5
602.
670
−1.9
803.
240
1.28
01.
680
0.00
0
DG
TW
Cha
r.A
dj.
CA
DJR
ET
38,0
621.
200
−1.0
5029
.320
1.71
0−1
.550
1.86
00.
020
0.48
0−1
.020
Info
rmat
ion
Env
iron
men
tVar
iabl
es
Tur
nove
rT
UR
N47
,625
1.40
60.
933
1.43
21.
247
0.79
01.
495
0.95
91.
512
1.07
5
Idio
.V
olat
ility
IVO
L47
,625
9.56
27.
762
6.69
910
.103
8.50
012
.981
11.2
427.
987
6.44
4
Idio
.Sk
ewne
ssIS
KE
W47
,625
0.13
90.
133
0.70
60.
164
0.16
90.
159
0.14
10.
111
0.09
9
Maj
orA
udit
orM
AJA
UD
47,6
250.
847
1.00
00.
360
0.92
51.
000
0.91
71.
000
0.75
81.
000
Ana
lyst
Cov
erag
eA
NC
OV
47,6
256.
006
4.00
06.
394
5.62
74.
000
5.67
04.
000
6.43
85.
000
Num
.of
Ana
lyst
sN
UM
AN
A47
,625
0.82
11.
000
0.38
30.
812
1.00
00.
814
1.00
00.
831
1.00
0
Abs
.D
isc.
Acc
rual
sA
BSD
AC
C38
,678
0.27
30.
071
0.64
40.
125
0.05
50.
418
0.13
50.
361
0.07
9
Con
tinued
...
43
Full
Sam
ple:
1996
-200
8P
RE
SOX
:19
96-2
000
FD
SOX
:200
1-02
PO
STSO
X:20
03-0
8
Var
iabl
eSh
ort
Nam
eN
Mea
nM
edia
nSt
dD
evM
ean
Med
ian
Mea
nM
edia
nM
ean
Med
ian
Oth
erFir
mC
hara
cter
istics
log(
Mar
ket
Cap
)M
CA
P47
,625
12.8
2312
.760
1.94
112
.544
12.4
5012
.491
12.4
7113
.169
13.0
98
Boo
k-To-
Mar
ket
BM
47,5
530.
583
0.47
20.
584
0.58
20.
458
0.67
20.
536
0.55
50.
468
Fin
anci
alLev
erag
eFIN
LE
V47
,500
0.18
00.
115
0.20
10.
193
0.13
60.
184
0.11
10.
168
0.10
2
Fir
mA
geA
GE
47,6
2514
.848
9.00
015
.507
13.6
158.
000
14.1
198.
000
16.1
3411
.000
6mLag
ged
Ret
urn
RE
T6M
46,2
960.
083
0.03
40.
419
0.11
70.
054
0.04
40.
008
0.06
60.
026
Stoc
kP
rice
PR
ICE
47,6
2521
.884
17.3
6518
.180
21.1
8916
.726
20.9
6715
.620
22.7
7318
.484
Lot
tery
-Typ
eSt
ock
LO
TT
47,6
040.
335
0.00
00.
472
0.32
50.
000
0.33
30.
000
0.34
40.
000
Loca
lA
ttri
bute
s
Edu
cati
onE
DU
47,6
2525
.490
26.1
003.
701
24.7
5225
.320
26.0
9226
.600
25.9
2026
.600
Pop
.D
ensi
tyP
OP
DE
N47
,625
313.
783
217.
998
455.
044
305.
148
211.
515
319.
088
217.
998
319.
392
217.
998
Cat
h.-P
rot.
Rat
ioC
PR
AT
IO47
,625
0.49
90.
521
0.10
80.
505
0 .52
30.
495
0.52
00.
497
0.52
0
Rel
igio
sity
RE
L47
,625
1.42
60.
956
1.23
61.
405
0.95
61.
493
1.02
01.
423
0.89
0
Rep
ublic
anR
EP
UB
47,6
250.
731
1.00
00.
443
0.81
51.
000
0.70
61.
000
0.66
81.
000
Mic
rost
ruct
ure
Var
iabl
es
PIN
PIN
12,4
830.
177
0.15
10.
096
0.18
50.
159
0.17
30.
145
0.15
20.
123
Adv
.Se
lect
ion
AD
JPIN
1248
30.
145
0.12
70.
070
0.15
10.
135
0.14
10.
121
0.12
50.
108
Liq
uidi
tyP
SOS
12,4
830.
257
0.21
60.
136
0.26
30.
219
0.25
90.
223
0.23
10.
193
Effe
ctiv
eSp
read
SPR
EA
D35
,057
0.40
40.
186
0.54
10.
617
0.39
90.
384
0.18
80.
256
0.09
0
Aro
und
Ear
ning
sSP
DE
AR
N35
,057
0.42
20.
199
0.58
10.
626
0.40
80.
414
0.20
60.
275
0.11
0
Exc
l.E
arni
ngs
SPD
OT
H35
,057
0.40
30.
185
0.54
10.
616
0.39
80.
383
0.18
70.
255
0.09
0
Ana
lyst
-Rel
ated
Mea
sure
s
Hig
hPer
form
ance
HIG
HP
ER
F23
,566
0.34
50.
333
0.20
80.
346
0.33
30.
331
0.31
60.
351
0.33
3
Ten
ure
TE
NU
RE
23,5
6623
.626
17.0
0020
.218
23.3
5918
.000
23.3
3217
.000
24.1
8318
.000
Pre
stig
ious
Bro
ker
PB
RO
KE
R23
,566
0.33
40.
000
0.47
20.
342
0.00
00.
376
0.00
00.
302
0.00
0
Num
.of
Fir
ms
Cov
ered
NFC
OV
ER
23,5
6610
.995
10.0
006.
175
11.1
1710
.000
10.3
2910
.000
11.1
5111
.000
Dem
eane
dA
bs.Fo
re.E
rr.
DA
FE
786,
605
−0.0
93−0
.008
0.34
4−0
.132
−0.0
05−0
.172
−0.0
08−0
.025
−0.0
12
Mar
ket
Rea
ctio
n,U
pR
ev.
RE
TU
P28
3,00
10.
345
0.15
27.
370
0.22
20.
085
−0.1
89−0
.097
0.59
50.
232
Mar
ket
Rea
ctio
n,D
n.R
ev.
RE
TD
N26
4,19
9−1
.072
−0.6
768.
350
−1.2
66−0
.820
−1.1
56−0
.740
−0.8
59−0
.560
44
Table 2
Annual Local Ownership and Ownership Concentration Estimates
This table reports annual averages for local ownership, ownership concentration, and local analyst coveragemeasures. All measures are multiplied by 100 to improve readability. For comparison, in Column (1), wereport the mean equal-weighted institutional ownership (IO) across all CRSP firms. Column (2) reports thefull-sample annual excess local ownership estimates for firm headquarters (HQ) states. To provide a benchmark,the mean expected level of local ownership is reported in Column (3). Columns (4) to (6) report the excesslocal ownership estimates for small, mid-sized, and large firms, respectively. Column (7) reports the excess localownership estimates for most economically relevant (ER) states. In Columns (8) to (10), we report the meanstate-level ownership concentration indices. Specifically, Columns (8) and (9) report the Herfindahl-Hirschmanconcentration index (HHI) of state-level institutional ownership across all U.S. states. HH1 is based on the fullsample, while HQ states are excluded when we compute HH2. Column (10) reports the geographical concentrationindex (GCI) of state-level institutional ownership across all U.S. states adjusted for the natural level of ownershipconcentration resulting from the geographical clustering of institutional investors. In Columns (11) and (12), wereport the excess local analyst coverage around corporate headquarters and in the most economically relevantstates. The definitions of all variables are provided in the Appendix.
HQ Local Ownership (2-6) ER (7) Ownership Conc. (8-10) Loc An Cov (11-12)
IO Excess Exp SmlCap MidCap LgCap All HHI1 HHI2 GCI HQ ER
Year (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
1996 26.89 8.02 7.37 14.65 5.76 3.77 1.62 27.76 32.96 15.92 12.35 2.10
1997 27.96 7.87 7.35 13.73 5.88 3.94 1.75 26.56 31.37 14.97 12.97 1.47
1998 28.41 7.64 7.19 13.47 5.76 3.73 2.02 27.15 32.16 16.12 11.94 3.31
1999 28.27 7.90 7.58 14.31 5.50 3.51 1.70 27.44 32.00 16.23 11.34 2.78
2000 29.07 7.86 8.00 14.42 5.86 2.98 2.11 27.90 32.26 17.14 10.94 2.87
2001 30.56 6.80 7.84 12.37 5.05 2.91 1.75 26.01 30.61 15.64 10.44 2.96
2002 33.19 5.75 7.77 10.68 4.18 2.36 0.96 25.70 29.99 14.77 10.83 1.69
2003 36.09 5.53 7.46 10.53 3.74 2.26 1.18 23.39 26.57 13.38 10.55 1.70
2004 40.71 4.38 7.52 8.46 2.80 1.91 0.57 23.84 27.22 11.50 7.32 2.95
2005 42.70 4.28 7.59 8.29 3.06 1.55 0.52 23.64 27.16 11.20 7.13 2.80
2006 44.53 3.65 7.63 7.21 2.59 1.20 0.88 22.73 25.84 10.22 6.18 2.59
2007 48.46 3.66 7.84 6.61 3.11 1.30 1.16 23.09 26.30 10.54 5.53 1.77
2008 48.03 3.96 7.77 7.46 2.98 1.45 1.17 23.72 27.81 11.34 5.49 1.91
45
Table 3
Information Environment and Excess Local Ownership:Estimates From Constant Coefficients Specifications
This table reports pooled cross-sectional regression coefficient estimates and corresponding t-statistics for therelation between firm-level excess local ownership (LOCOWN) and the characteristics of the firm informationenvironment, controlling for time varying firm and/or state characteristics and year and state fixed effects. Thelevel of firm local excess institutional ownership is measured as of the fiscal year end, and all explanatory variablesare measured during the same fiscal year. In Columns (1) to (5), we use the portfolio holdings of institutions inthe headquarters (HQ) state while in Column (6) we use the holdings of institutions located in most economicallyrelevant (ER) states. The definitions of all variables are provided in the Appendix. Following Petersen (2009),the t-statistics reported in parentheses are based on standard errors clustered by year and state. The sampleperiod is from 1996 to 2008.
Dependent Variable: Excess Local Ownership (in %)
Firm Headquarters (HQ) ER
Independent Variable (1) (2) (3) (4) (5) (6)
Information Environment Variables
TURN −0.44 −0.27 0.05
(−6.2) (−3.7) (0.9)
IVOL 0.05 0.04 −0.01
(1.8) (1.8) (−0.6)
ISKEW 0.22 0.07 0.00
(1.5) (0.5) (0.0)
MAJAUD −2.87 −3.17 0.07
(−3.9) (−6.4) (0.2)
ANCOV −3.80 −3.31 −1.16
(−10.8) (−7.4) (−3.4)
ABSDACC RANK 0.26 0.16 −0.38
(2.0) (1.9) (−4.2)
LOCANCOV −0.98 0.08 0.10
(−3.6) (0.4) (0.6)
Continued . . .
46
Firm Headquarters (HQ) ER
Independent Variable (1) (2) (3) (4) (5) (6)
Other Firm Characteristics
MCAP RANK −2.37 −1.77 −2.48 −2.33 −1.68 −0.82
(−10.6) (−7.6) (−11.2) (−11.5) (−8.7) (−8.9)
FINLEV −0.89 −0.45 −0.70 −1.22 −0.771 0.544
(−1.3) (−0.6) (−1.0) (−2.3) (−1.5) (1.4)
AGE 0.01 0.00 0.02 0.017 0.001 −0.008
(2.2) (0.3) (3.8) (2.7) (0.3) (−1.9)
BM −0.11 −0.03 −0.13 −0.35 −0.16 0.25
(−0.4) (−0.1) (−0.5) (−1.1) (−0.5) (1.4)
RET6M 0.10 −0.03 0.22 0.13 −0.15 0.04
(0.3) (−0.1) (0.7) (0.5) (−0.6) (0.2)
PRICE 0.01 0.00 0.01 −0.00 0.00 −0.01
(1.7) (0.7) (1.1) (−0.1) (0.0) (−2.1)
LOTT −0.40 −0.07 −0.09 −0.11 −0.26 −0.09
(−2.1) (−0.3) (−0.4) (−0.5) (−1.0) (−0.4)
Local Attributes
EDU −4.37 −4.31 −4.40 −2.84 −2.77 0.79
(−6.2) (−6.2) (−6.2) (−3.0) (−2.9) (2.2)
POPDEN 0.07 0.07 0.07 0.03 0.04 −0.02
(3.2) (3.1) (3.2) (0.8) (0.9) (−1.0)
CPRATIO 5.56 5.50 5.63 9.44 8.58 2.26
(2.3) (2.4) (2.3) (1.5) (1.3) (0.9)
REL −28.45 −28.34 −27.01 −28.06 −31.57 10.15
(−1.1) (−1.1) (−1.1) (−1.2) (−1.3) (0.8)
REPUB −0.54 −0.66 −0.50 −0.82 −0.96 −0.52
(−1.2) (−1.5) (−1.1) (−1.5) (−1.7) (−1.5)
State FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
N 37,876 37,876 37,876 37,782 37,782 35,214
Adj. R2 0.113 0.121 0.112 0.117 0.127 0.065
47
Table 4
Information Environment and Excess Local Ownership:Estimates From Time-Varying Coefficients Specifications
This table reports pooled cross-sectional regression coefficient estimates and corresponding t-statistics for therelation between firm-level excess local ownership (LOCOWN) and the characteristics of the firm informationenvironment conditional on the regulatory regime. We also control for time-varying firm and state characteristicsand consider regime as well as state fixed effects. The level of firm local excess institutional ownership is measuredas of the fiscal year end, and all explanatory variables are measured during the same fiscal year. FDSOX(POSTSOX) is an indicator variable equal to one for fiscal years ending in calendar years 2001 and 2002 (2003and onward). In Columns (1) to (3) and (5) to (8), we use the portfolio holdings of institutions in the headquarters(HQ) state while in Column (4) we use the holdings of institutions located in most economically relevant (ER)states. The excess local ownership dependent variable in Column (5) is recalculated after excluding institutionalinvestors that are classified as public or corporate pension funds, university endowments, or miscellaneous. Theresults in Columns (6) through (8) are based on samples that exclude stocks priced below $2, firms located inCalifornia or New York states, and high-tech firms, respectively. The definitions of all variables are provided inthe Appendix. Following Petersen (2009), the t-statistics reported in parentheses are based on standard errorsclustered by year and state. The sample period is from 1996 to 2008.
48
Dependent Variable: Excess Local Ownership (in %)
Firm Headquarters (HQ) ER Robustness (5-8)
No PF Pr>$2 No CA/NY No Tech
Independent Variable (1) (2) (3) (4) (5) (6) (7) (8)
FDSOX −1.74 −2.32 −2.77 −0.80 −2.28 −2.97 −3.34 −3.13
(−5.0) (−2.5) (−2.3) (−0.6) (−1.9) (−1.8) (−1.8) (−2.1)
POSTSOX −3.45 −6.07 −6.10 −2.09 −5.46 −5.63 −7.48 −6.35
(−16.4) (−5.5) (−3.2) (−1.7) (−3.0) (−2.8) (−4.4) (−3.1)
TURN −0.67 −0.56 0.05 −0.71 −0.44 −0.94 −0.71
(−6.5) (−3.2) (0.4) (−3.3) (−2.8) (−4.1) (−3.4)
TURN × FDSOX 0.22 0.06 0.07 0.28 −0.03 0.08 −0.32
(2.1) (0.3) (0.6) (1.0) (−0.1) (0.3) (−0.9)
TURN × POSTSOX 0.51 0.46 −0.04 0.48 0.39 0.80 0.71
(3.6) (2.2) (−0.3) (1.9) (2.0) (3.1) (2.9)
IVOL 0.15 0.15 −0.02 0.18 0.12 0.15 0.15
(2.6) (3.0) (−0.6) (3.8) (2.5) (2.3) (2.8)
IVOL × FDSOX −0.13 −0.10 −0.01 −0.15 −0.05 −0.13 −0.08
(−1.8) (−1.5) (−0.4) (−1.8) (−0.8) (−1.6) (−1.2)
IVOL × POSTSOX −0.22 −0.24 0.03 −0.25 −0.24 −0.22 −0.21
(−3.6) (−3.9) (0.9) (−3.9) (−3.9) (−2.9) (−3.2)
ISKEW 0.12 −0.03 −0.13 −0.02 −0.00 −0.01 −0.10
(1.2) (−0.1) (−0.7) (−0.1) (−0.0) (−0.1) (−0.4)
ISKEW × FDSOX −0.85 −0.42 0.35 −0.14 −0.47 0.12 −0.35
(−2.8) (−1.0) (1.3) (−0.3) (−1.0) (0.3) (−0.9)
ISKEW × POSTSOX 0.34 0.37 0.18 0.38 0.28 0.34 0.63
(2.4) (1.3) (0.8) (1.2) (0.9) (1.2) (2.1)
MAJAUD −6.88 −7.25 0.43 −7.86 −6.65 −8.83 −7.01
(−12.8) (−7.2) (0.7) (−7.2) (−6.3) (−6.0) (−7.6)
MAJAUD × FDSOX 1.87 1.69 −1.50 1.82 1.74 1.96 1.96
(3.5) (1.3) (−1.4) (1.1) (1.2) (0.9) (1.1)
MAJAUD × POSTSOX 6.26 6.13 −0.40 6.47 5.59 6.81 5.63
(10.4) (5.7) (−0.5) (5.6) (4.8) (4.2) (5.4)
Continued . . .
49
Firm Headquarters (HQ) ER Robustness (5-8)
No No No
PF Pr>$2 CA/NY Tech
Independent Variable (1) (2) (3) (4) (5) (6) (7) (8)
ANCOV −4.01 −3.91 −1.91 −3.67 −3.37 −4.37 −4.36
(−8.0) (−5.6) (−3.8) (−6.0) (−4.5) (−6.6) (−5.9)
ANCOV × FDSOX 1.86 1.98 1.10 1.68 1.47 4.01 2.11
(3.7) (1.2) (1.0) (0.9) (0.9) (4.3) (1.1)
ANCOV × POSTSOX 0.49 0.96 1.42 0.22 0.76 2.54 0.72
(0.8) (0.8) (1.9) (0.2) (0.7) (2.8) (0.6)
ABSDACC RANK 0.56 0.50 −0.39 0.26 0.46 0.15 0.27
(1.7) (2.4) (−2.7) (1.2) (2.2) (0.8) (1.4)
ABSDACC RANK × FDSOX −0.33 −0.17 0.09 0.02 −0.22 0.18 0.01
(−1.0) (−0.6) (0.3) (0.1) (−0.7) (0.6) (0.0)
ABSDACC RANK × POSTSOX −0.66 −0.73 0.00 −0.35 −0.64 −0.20 −0.43
(−1.9) (−2.9) (0.0) (−1.4) (−2.5) (−1.0) (−1.7)
LOCANCOV 0.57 −0.05 0.44 0.61 0.54 0.77
(1.1) (−0.2) (0.3) (0.7) (0.8) (1.5)
LOCANCOV × FDSOX −0.55 0.16 −0.75 −0.38 −0.60 −0.75
(−0.6) (0.4) (−0.6) (−0.7) (−0.4) (−0.6)
LOCANCOV × POSTSOX −0.83 0.34 −0.63 −0.61 −0.56 −1.18
(−1.0) (0.8) (−0.4) (−0.7) (−0.6) (−1.6)
Firm Controls No Yes Yes Yes Yes Yes Yes Yes
State Controls No Yes Yes Yes Yes Yes Yes Yes
State FE Yes Yes Yes Yes Yes Yes Yes Yes
N 38,834 37,876 37,782 35,212 37,782 36,101 26,025 28,695
Adj. R2 0.149 0.133 0.133 0.067 0.115 0.130 0.149 0.128
50
Table 5
Change in Local Ownership and Pre-Regulation Information Environment
This table reports pooled cross-sectional regression coefficient estimates and corresponding t-statistics for therelation between firm-level changes in excess local ownership relative to end of fiscal year 2000 level of the samemeasure and the characteristics of the firm information environment at the end of fiscal year 2000. We alsocontrol for time-varying firm and state characteristics and consider regime as well as state fixed effects. The levelof firm local excess institutional ownership is measured as of the fiscal year end, and all explanatory variablesare measured during the same fiscal year. FDSOX (POSTSOX) is an indicator variable equal to one for fiscalyears ending in calendar years 2001 and 2002 (2003 and onward). In Columns (1) to (3) and (5) to (8), we usethe portfolio holdings of institutions in the headquarters (HQ) state while in Column (4) we use the holdingsof institutions located in most economically relevant (er) states. The excess local ownership dependent variablein Column (5) is recalculated after excluding institutional investors that are classified as public or corporatepension funds, university endowments, or miscellaneous. The results in Columns (6) through (8) are basedon samples that exclude stocks priced below $2, firms located in California or New York states, and high-techfirms, respectively. The definitions of all variables are provided in the Appendix. Following Petersen (2009), thet-statistics reported in parentheses are based on standard errors clustered by year and state. The sample periodis from 1996 to 2008.
Dependent Variable: Change in Excess Local Ownership relative to FYE 2000 (in %)
Firm Headquarters (HQ) ER Robustness
No PF Pr>$2 No CA/NY No Tech
Independent Variable (1) (2) (3) (4) (5) (6) (7) (8)
FDSOX −2.94 −15.33 −14.79 −0.06 −5.12 −20.02 −10.31 −21.04
(−5.6) (−2.0) (−1.9) (−0.0) (−0.7) (−2.9) (−1.2) (−1.9)
POSTSOX −4.01 −15.16 −14.56 −1.89 −3.87 −20.53 −6.89 −21.42
(−9.3) (−2.0) (−1.9) (−0.3) (−0.6) (−3.0) (−0.9) (−2.0)
TURN 0.32 0.39 −0.41 0.54 0.34 0.38 0.56
(3.8) (4.2) (−2.4) (4.3) (2.8) (2.9) (2.2)
TURN × POSTSOX 0.44 0.44 −0.22 0.69 0.4 0.39 0.85
(3.5) (3.4) (−1.6) (6.3) (3.3) (3.1) (3.8)
IVOL 0.09 0.08 0.07 0.02 0.08 0.01 0.04
(3.0) (2.7) (1.8) (0.7) (2.8) (0.2) (0.8)
IVOL × POSTSOX 0.06 0.06 0.05 −0.03 0.04 0.03 0.01
(2.5) (2.0) (1.7) (−1.2) (1.1) (0.8) (0.2)
ISKEW 0.01 0.05 −0.58 0.49 0.04 −0.14 −0.27
(0.1) (0.2) (−1.3) (1.2) (0.1) (−0.3) (−0.6)
ISKEW × POSTSOX −0.15 −0.26 0.16 −0.02 −0.34 −0.46 −0.95
(−1.6) (−1.5) (0.6) (−0.1) (−1.8) (−2.2) (−3.5)
Continued . . .51
Firm Headquarters (HQ) ER Robustness
No No No
PF Pr>$2 CA/NY Tech
Independent Variable (1) (2) (3) (4) (5) (6) (7) (8)
MAJAUD 3.15 2.90 −3.18 3.17 2.29 4.59 5.14
(2.2) (2.1) (−2.5) (2.2) (1.5) (2.2) (2.6)
MAJAUD × POSTSOX 2.99 3.07 −1.92 3.13 3.81 2.33 5.32
(4.5) (4.7) (−2.3) (4.3) (5.3) (2.4) (5.8)
ANCOV 0.58 0.98 −0.9 2.06 0.84 2.01 0.27
(1.0) (1.0) (−0.9) (2.3) (1.1) (2.3) (0.3)
ANCOV × POSTSOX 0.05 0.05 −1.37 0.56 −0.50 0.40 0.75
(0.6) (0.1) (−2.4) (1.0) (−1.1) (0.7) (0.3)
ABSDACC RANK −0.74 −0.72 0.44 −0.54 −0.65 −0.10 −0.09
(−1.9) (−1.9) (1.5) (−1.4) (−1.7) (−0.4) (−0.2)
ABSDACC RANK × POSTSOX −1.21 −1.36 0.98 −1.25 −1.14 −0.31 −1.26
(−5.1) (−5.0) (5.3) (−4.7) (−4.5) (−2.1) (−3.7)
LOCANCOV 0.00 0.34 0.36 0.07 −0.08 0.05
(0.0) (0.9) (0.8) (0.2) (−0.2) (0.1)
LOCANCOV × POSTSOX 0.14 −0.65 0.47 0.09 −0.54 −0.38
(0.3) (−1.6) (1.1) (0.2) (−2.2) (−1.4)
Firm Controls No Yes Yes Yes Yes Yes Yes Yes
State Controls No Yes Yes Yes Yes Yes Yes Yes
State FE Yes Yes Yes Yes Yes Yes Yes Yes
N 16,476 16,074 16,045 16,045 16,045 15,256 10,806 11,837
Adj. R2 0.030 0.050 0.053 0.040 0.068 0.059 0.066 0.081
52
Table 6
Local Ownership and Local Performance:Estimates From Constant Coefficients Specifications
This table reports pooled cross-sectional regression coefficient estimates and corresponding t-statistics for therelation between firm-level returns in the quarter following the fiscal year end and the level of excess localinstitutional ownership as of the fiscal year end. We also control for time-varying firm and state characteristicsand consider regime as well as state fixed effects. The definitions of all variables are provided in the Appendix.Following Petersen (2009), the t-statistics reported in parentheses are based on standard errors clustered by yearand state. The sample period is from 1996 to 2008.
Dependent Variable: Subsequent Quarter Return (in %)Firm Headquarters (HQ) ER
RAWRET EWRET VWRET CADJRET CADJRETIndependent Variable (1) (2) (3) (4) (5)LOCOWN 3.65 3.78 4.21 2.45 0.98
(2.6) (2.6) (2.8) (2.4) (0.8)
Other Firm CharacteristicsMCAP RANK −0.57 −0.71 −0.59 −0.09 −0.21
(−1.8) (−2.3) (−2.1) (−0.4) (−1.0)
FINLEV −0.67 −1.22 −0.57 −3.20 −3.21(−0.3) (−0.5) (−0.2) (−1.7) (−1.4)
AGE 0.04 0.04 0.04 0.01 0.02(1.4) (1.7) (1.5) (0.6) (0.8)
BM 4.00 3.39 3.68 3.13 2.7(2.0) (1.7) (1.8) (2.6) (2.1)
RET6M 0.00 0.37 −0.15 −0.74 −0.76(0.0) (0.2) (−0.1) (−1.3) (−1.2)
PRICE −0.06 −0.06 −0.05 −0.02 −0.02(−1.5) (−1.5) (−1.4) (−0.9) (−0.7)
LOTT 0.77 0.92 0.84 0.71 0.48(0.6) (0.7) (0.7) (0.7) (0.5)
Local AttributesEDU 0.71 0.75 0.65 0.18 −2.31
(0.4) (0.5) (0.4) (0.1) (−3.4)
POPDEN 0.00 0.01 0.00 −0.04 −0.01(0.0) (0.1) (0.0) (−0.8) (−0.1)
CPRATIO −4.37 −5.65 −3.90 1.76 0.95(−0.3) (−0.4) (−0.3) (0.2) (0.2)
REL −14.19 −11.56 −19.46 −33.94 −12.56(−0.7) (−0.6) (−1.1) (−1.8) (−0.5)
REPUB −1.88 −1.56 −1.84 −1.23 −0.67(−0.7) (−0.6) (−0.7) (−0.6) (−0.8)
State FE Yes Yes Yes Yes YesYear FE Yes Yes Yes Yes YesN 46,433 46,433 46,433 40,035 37,645Adj. R2 0.057 0.012 0.034 0.006 0.006
53
Table 7
Local Ownership and Local Performance:Estimates From Time-Varying Coefficients Specifications
This table reports pooled cross-sectional regression coefficient estimates and corresponding t-statistics for therelation between firm-level returns in the quarter following the fiscal year end and the level of excess localinstitutional ownership as of the fiscal year end conditional on the regulatory regime. We also control for time-varying firm and state characteristics and consider regime as well as state fixed effects. FDSOX (POSTSOX) isan indicator variable equal to one for fiscal years ending in calendar years 2001 and 2002 (2003 and onward).We interact these indicator variables with LOCOWN, but do not include the standalone variables since theyare subsumed by the year fixed effects. In Columns (1) to (4) and (6) to (9), we use the portfolio holdingsof institutions in the headquarters (HQ) state while in Column (5) we use the holdings of institutions locatedin most economically relevant (ER) states. The excess local ownership independent variable in Column (6)is recalculated after excluding institutional investors that are classified as public or corporate pension funds,university endowments, or miscellaneous. The results in Columns (7) to (9) are based on samples that excludestocks priced below $2, firms located in California or New York states, and high-tech firms, respectively. We useequal-weighted returns in all four robustness tests. The definitions of all variables are provided in the Appendix.Following Petersen (2009), the t-statistics reported in parentheses are based on standard errors clustered by yearand state. The sample period is from 1996 to 2008.
Dependent Variable: Subsequent Quarter Return (in %)
Firm Headquarters (HQ) ER Robustness (using EWRET)
RAW EW VW CADJ CADJ No No No
RET RET RET RET RET PF Pr>$2 CA/NY Tech
Independent Variable (1) (2) (3) (4) (5) (6) (7) (8) (9)
LOCOWN 5.40 5.76 6.25 3.99 0.75 7.26 4.84 3.54 2.86
(2.1) (2.1) (2.3) (2.4) (0.5) (2.5) (2.2) (2.0) (1.8)
LOCOWN × FDSOX 0.74 0.15 −0.02 0.36 −3.68 −1.84 0.81 3.74 3.66
(0.2) (0.1) (−0.0) (0.2) (−2.3) (−0.6) (0.3) (1.4) (2.0)
LOCOWN × POSTSOX −6.69 −7.14 −7.25 −5.50 2.68 −9.41 −5.92 −10.85 −5.74
(−1.9) (−2.0) (−2.0) (−2.5) (0.9) (−2.8) (−1.9) (−4.5) (−2.5)
Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes
State Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes
State FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
N 46,433 46,433 46,433 40,035 37,645 46,433 44,660 31,269 35,159
Adj. R2 0.058 0.013 0.034 0.007 0.006 0.013 0.012 0.020 0.018
54
Table 8
Local Ownership and Informed Trading:Estimates From Constant and Time-Varying Coefficients Specifications
This table reports pooled cross-sectional regression coefficient estimates and corresponding t-statistics for therelation between firm-level measures of informed trading in the year following the fiscal year end and the level ofexcess local institutional ownership as of the fiscal year end. In Panels B and C this relation is conditioned on theregulatory regime. We also control for time-varying firm and state characteristics and include year as well as statefixed effects. FDSOX (POSTSOX) is an indicator variable equal to one for fiscal years ending in calendar years2001 and 2002 (2003 and onward). We interact these indicator variables with LOCOWN, but do not includethe standalone variables since they are subsumed by the year fixed effects. Among the dependent variables,PIN is the probability of private information related trade from Easley, Hvidkjaer, and O’Hara (2002) model.ADJPIN and PSOS are PIN’s components from Duarte and Young (2009). ADJPIN is the probability of privateinformation related trade from the extended model in Duarte and Young (2009), while PSOS is the probabilitythat a given trade happens during a symmetric order flow shock. Effective Spread (SPREAD) is the differencebetween the actual execution price and the midpoint of market quotes. We provide separate measurements ofthe spread around earnings announcements (SPDEARN) and excluding earnings announcements (SPDOTH).The definitions of all variables are provided in the Appendix. Following Petersen (2009), the t-statistics reportedin parentheses are based on standard errors clustered by year and state. The sample is restricted to 1996 to 2004period due to the availability of PIN components.
Panel A: Estimates Using Constant Coefficients Specifications
Dependent Variable: Subsequent Year PIN or Spread (in %)
PIN-Based Spreads-Based
PIN ADJPIN PSOS SPREAD SPDOTH SPDEARN
Independent Variable (1) (2) (3) (4) (5) (6)
LOCOWN 5.65 2.15 3.73 0.35 0.35 0.38
(10.0) (4.8) (9.9) (10.6) (10.6) (8.4)
Firm Controls Yes Yes Yes Yes Yes Yes
State Controls Yes Yes Yes Yes Yes Yes
State FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
N 12,406 12,406 12,406 34,785 34,785 34,785
Adj. R2 0.321 0.486 0.476 0.524 0.523 0.474
55
Panel B: Estimates Using Time-Varying Coefficients Specifications
Dependent Variable: Subsequent Year PIN or Spread (in %)
PIN-Based Spreads-Based
PIN ADJPIN PSOS SPREAD SPDOTH SPDEARN
Independent Variable (1) (2) (3) (4) (5) (6)
LOCOWN 3.59 2.24 3.23 0.46 0.46 0.46
(4.2) (4.6) (6.2) (7.8) (8.1) (8.1)
LOCOWN × FDSOX 2.20 −0.43 −0.22 −0.17 −0.18 −0.10
(1.6) (−0.6) (−0.3) (−3.6) (−3.8) (−1.1)
LOCOWN × POSTSOX 4.40 −1.43 0.25 −0.19 −0.20 −0.15
(3.0) (−1.7) (0.3) (−3.0) (−3.1) (−1.4)
Firm Controls Yes Yes Yes Yes Yes Yes
State Controls Yes Yes Yes Yes Yes Yes
State FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
N 12,406 12,406 12,406 34,785 34,785 34,785
Adj. R2 0.326 0.507 0.495 0.528 0.528 0.478
Panel C: Estimates Using Time-Varying Coefficients Specifications Around Earnings Announcements
Dependent Variable: Spread Around Earnings Announcement (in %)
Pre-Announcement Post-Announcement
(−5:−1) (−10:−1) (−20:−1) (+1:+5) (+1:+10) (+1:+20)
Independent Variable (1) (2) (3) (4) (5) (6)
LOCOWN 0.37 0.37 0.36 0.32 0.32 0.31
(5.9) (5.9) (5.7) (5.0) (5.2) (5.2)
LOCOWN × FDSOX −0.20 −0.22 −0.22 −0.04 −0.04 −0.04
(−2.2) (−2.6) (−2.6) (−0.4) (−0.5) (−0.5)
LOCOWN × POSTSOX −0.20 −0.21 −0.18 −0.06 −0.07 −0.07
(−2.2) (−2.3) (−2.1) (−0.6) (−0.7) (−0.8)
Firm Controls Yes Yes Yes Yes Yes Yes
State Controls Yes Yes Yes Yes Yes Yes
State FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
N 41,001 41,007 41,009 41,055 41,059 41,060
Adj. R2 0.413 0.437 0.453 0.432 0.454 0.462
56
Table 9
Local Bias and Local Performance: Institution-Level Estimates
This table reports the abnormal fraction of institutional investors’ equity holdings invested in local stocks aswell as the average performance of institutional investors’ local and non-local holdings. Excess Local Weight isthe percentage of the investor’s portfolio invested in stocks located in the investor’s state (local stocks) minusthe percentage of the “market portfolio” located in the investor’s state. For each institution with non-zeroportfolio weight in local stocks, we calculate the quarterly characteristics-adjusted returns of its local and non-local portfolios, as well as the performance differential (local minus non-local). Local Portfolio is comprised ofstocks in the institutional investor’s portfolio that are located in the investor’s state, while Non-Local Portfolio iscomprised of the rest of the institution’s holdings. The portfolio returns are adjusted for the following three stockcharacteristics using the Daniel, Grinblatt, Titman, and Wermers (1997) method: size, book-to-market, and past12-month return. We take the average of excess weights and adjusted returns across all institutions each quarter,and then report the time-series average of those quarterly averages. The quarterly averaging across institutionsis either weighted by the total dollar value of the institution’s holdings at the beginning of the quarter (PanelA) or weighted equally (Panel B). The t-statistics reported in parentheses are adjusted for autocorrelation andheteroscedasticity following the Newey and West (1987) method. The sample period is from 1996 to 2008.
Panel A: Holdings-Weighted Average
PRESOX FDSOX POSTSOX
Measure 1996-2000 2001-2002 2003-2008 2001-2008
Excess Local Weight 3.54% 2.71% 2.24% 2.35%
Char. Adj. Quarterly Returns
Local Portfolio 1.11% −0.32% −0.20% −0.23%
Non-Local Portfolio 0.58% 0.05% −0.08% −0.04%
Local − Non-Local 0.54% −0.38% −0.12% −0.19%
(3.5) (−1.0) (−0.7) (−1.0)
Panel B: Equal-Weighted Average
PRESOX FDSOX POSTSOX
Measure 1996-2000 2001-2002 2003-2008 2001-2008
Excess Local Weight 5.93% 5.17% 4.22% 4.45%
Char. Adj. Quarterly Returns
Local Portfolio 1.08% 0.12% −0.08% −0.03%
Non-Local Portfolio 0.83% 0.04% 0.03% 0.04%
Local − Non-Local 0.27% 0.06% −0.13% −0.08%
(2.9) (0.3) (−1.1) (−0.8)
57
Table 10
Local Analyst Coverage Regression Estimates
This table reports the estimates from local coverage regressions of sell-side equity analysts. The dependentvariable is the excess coverage of local stocks, which is the difference between the fraction of local stocks in ananalyst’s stock coverage portfolio and the fraction of local stocks that the analyst can potentially cover withinthe CRSP universe. In the first two columns, we use equal-weighted fractions, while in the last two columns, weuse value-weighted fractions. The independent variables include indicator variables for FDSOX (2001-2002) andPOSTSOX (2003 onward), analyst characteristics (tenure, past performance, and number of firms covered), andan indicator variable for analysts employed by prestigious brokerage houses. All specifications include state fixedeffects. The definitions of all variables are provided in the Appendix. Following Petersen (2009), the t-statisticsreported in parentheses are based on standard errors clustered by year and state. The sample period is from1996 to 2008.
Dependent Variable: Excess Coverage of Local Stocks (in %)
Equal-Weighted Value-Weighted
Independent Variable (1) (2) (3) (4)
FDSOX −1.24 −2.08 −1.75 −2.22
(−2.7) (−3.1) (−1.8) (−2.0)
POSTSOX −4.68 −5.79 −3.21 −4.64
(−5.6) (−8.6) (−3.8) (−7.0)
PBROKER −1.83 −3.31 −1.38 −3.00
(−2.8) (−4.5) (−1.9) (−4.9)
PBROKER × FDSOX 2.33 1.39
(2.5) (1.2)
PBROKER × POSTSOX 3.41 4.43
(2.3) (2.7)
HIGHPERF 0.54 0.52 1.23 1.19
(0.6) (0.6) (1.3) (1.2)
TENURE 0.00 0.00 0.01 0.01
(0.3) (0.4) (0.5) (0.6)
NFCOVER −0.17 −0.17 −0.13 −0.14
(−5.1) (−5.0) (−3.9) (−3.9)
State FE Yes Yes Yes Yes
N 23, 569 23,569 23,567 23,567
Adj. R2 0.209 0.210 0.231 0.233
Mean of Excess Coverage 9.63 9.63 6.30 6.30
58
Table
11
LocalFore
cast
Accura
cy
Regre
ssio
nEst
imate
s
Thi
sta
ble
repo
rts
the
esti
mat
esfr
omfo
reca
stac
cura
cyan
dm
arke
tre
acti
onre
gres
sion
s.In
Col
umns
(1)
to(8
),th
ede
pend
ent
vari
able
ison
eof
the
mea
sure
sof
anal
yst
quar
terl
yea
rnin
gsfo
reca
ster
ror,
whi
lein
Col
umns
(9)
and
(10)
,th
ede
pend
ent
vari
able
isth
esh
ort-
term
mar
ket
reac
tion
mea
sure
com
pute
dse
para
tely
for
upw
ard
(RE
TU
P)
and
dow
nwar
d(R
ET
DN
)fo
reca
stre
visi
ons.
The
inde
pend
ent
vari
able
sin
clud
ean
indi
cato
rva
riab
leLO
CA
Lan
dit
sin
tera
ctio
nsw
ith
FD
SOX
(200
1-20
02)an
dP
OST
SOX
(200
3on
war
d).
InC
olum
ns(4
)an
d(5
),th
eLO
CA
Lin
dica
torva
riab
leis
set
toon
eif
the
anal
yst
islo
cate
din
the
stat
eth
atis
econ
omic
ally
mos
tre
leva
ntfo
rth
efir
m.
Inot
her
colu
mns
,LO
CA
Lis
set
toon
eif
the
anal
yst
and
the
firm
head
quar
ters
are
inth
esa
me
stat
e.W
edo
not
incl
ude
the
stan
d-al
one
regi
me
vari
able
ssi
nce
they
are
subs
umed
byth
eti
me-
vary
ing
anal
yst
fixed
effec
ts.
The
defin
itio
nsof
allv
aria
bles
are
prov
ided
inth
eA
ppen
dix.
Follo
win
gPet
erse
n(2
009)
,the
t-st
atis
tics
repo
rted
inpa
rent
hese
sar
eba
sed
onst
anda
rder
rors
clus
tere
dby
firm
and
quar
ter.
The
sam
ple
peri
odis
from
1996
to20
08.
Dep
ende
ntV
aria
ble:
Fore
cast
Err
oror
Mar
ket
Rea
ctio
n(i
n%
)
Fore
cast
Err
or(1
-8)
Mar
ket
Rea
ctio
n(9
-10)
HQ
HQ
ER
ER
Full
PR
ESO
XFD
SOX
PO
STSO
X
ER
RM
DN
ER
RP
CT
LE
RR
MD
NE
RR
PC
TL
DA
FE
DA
FE
DA
FE
DA
FE
RE
TU
PR
ET
DN
Inde
pend
ent
Var
iabl
e(1
)(2
)(3
)(4
)(5
)(6
)(7
)(8
)(9
)(1
0)
LO
CA
L0.
81−0
.34
−0.0
40.
12−0
.44
−0.4
4−0
.18
0.13
0.19
0.02
(2.6
)(−
2.6
)(−
0.1
)(0
.8)
(−1.9
)(−
1.9
)(−
0.4
)(0
.8)
(2.2
)(0
.2)
LO
CA
L×
FD
SOX
−0.5
90.
13−0
.43
0.49
0.26
0.21
−0.0
6
(−1.0
)(0
.5)
(−0.7
)(1
.7)
(0.6
)(1
.2)
(−0.4
)
LO
CA
L×
PO
STSO
X−1
.15
0.27
0.44
−0.1
20.
57−0
.19
−0.1
0
(−2.5
)(1
.3)
(0.9
)(−
0.5
)(2
.0)
(−1.9
)(−
0.8
)
Ana
lyst
×Q
uart
erFE
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
N81
1,66
981
1,66
981
1,66
981
1,66
978
6,60
533
3,16
311
9,73
933
3,70
328
3,00
126
4,19
9
Adj
.R2
0.15
60.
164
0.15
60.
164
0.12
10.
107
0.16
70.
138
0.33
60.
350
59
Table 12
Local Bias and Local Performance: Institution-Level Estimates,Conditional Upon Local Analyst Presence
This table reports the local bias and local performance of institutional investors, sorted by the presence of localequity analysts. States are sorted into three groups based on the number of local analysts in the state: top5 states, 6th-20th states, and the rest of the states. For each group, we report the excess local holdings andthe average performance of institutional investors’ local and non-local holdings. Excess Local HQ Weight isthe percentage of the investor’s portfolio invested in stocks whose HQ are located in the investor’s state (localHQ stocks) minus the percentage of the “market portfolio” located in the investor’s state. For each institutionwith non-zero portfolio weight in local stocks, we calculate the quarterly characteristics-adjusted returns of itslocal and non-local portfolios, as well as the performance differential (local minus non-local). Local Portfolio iscomprised of stocks in the institutional investor’s portfolio whose HQ are located in the investor’s state, whileNon-Local Portfolio is comprised of the rest of the institution’s holdings. The portfolio returns are adjusted forthe following three stock characteristics using the Daniel, Grinblatt, Titman, and Wermers (1997) method: size,book-to-market, and past 12-month return. We take the average of excess weights and adjusted returns acrossall institutions each quarter, and then report the time-series average of those quarterly averages. The quarterlyaveraging across institutions is weighted by the total dollar value of the institution’s holdings at the beginningof the quarter. Panel B examines local ER1−5 stocks rather than local HQ stocks. Local ER1−5 stocks arethose for which the investor’s state is one of the five most economically relevant states. The definitions of allvariables are provided in the Appendix. The t-statistics reported in parentheses are adjusted for autocorrelationand heteroscedasticity following the Newey and West (1987) method. The sample period is from 1996 to 2008.
Panel A: Institutions Around Firm Headquarters
Quarterly Char. Adj. Returns (in %)
Analyst Presence Time Period Excess ER Wt (in %) Local HQ Non-Local HQ−NL (t-stat)
High (Top 5 States) 1996-2000 3.04 −0.55 0.52 −1.07 (−1.8)
2001-02 1.75 0.04 −0.22 0.26 (0.4)
2003-08 1.84 −0.38 −0.06 −0.32 (−0.9)
2001-08 1.82 −0.28 −0.10 −0.18 (−0.5)
Medium (States 6-20) 1996-2000 3.32 1.35 0.48 0.87 (1.2)
2001-02 2.64 0.15 0.04 0.21 (0.7)
2003-08 2.16 0.16 0.04 0.12 (0.6)
2001-08 2.28 0.18 0.04 0.14 (0.6)
Low (Other States) 1996-2000 3.01 1.48 0.70 0.78 (2.9)
2001-02 1.93 −0.50 0.19 −0.69 (−1.7)
2003-08 2.23 −0.38 0.18 −0.56 (−1.0)
2001-08 2.15 −0.41 0.18 −0.60 (−1.3)
60
Panel B: Institutions in Economically-Relevant States
Quarterly Char. Adj. Returns (in %)
Analyst Presence Time Period Excess ER Wt (in %) Local ER Non-Local ER−NL (t-stat)
High (Top 5 States) 1996-2000 2.84 1.55 0.52 1.03 (3.2)
2001-02 3.00 −0.17 −0.22 0.04 (0.2)
2003-08 2.61 0.66 −0.06 0.72 (2.2)
2001-08 2.71 0.45 −0.10 0.55 (2.7)
Medium (States 6-20) 1996-2000 4.38 1.48 0.48 1.00 (3.5)
2001-02 3.78 0.77 0.04 0.73 (2.9)
2003-08 3.75 0.28 0.04 0.24 (1.4)
2001-08 3.76 0.48 0.04 0.44 (1.9)
Low (Other States) 1996-2000 5.37 0.99 0.70 0.29 (1.0)
2001-02 5.40 0.77 0.19 0.57 (2.1)
2003-08 5.10 0.54 0.18 0.36 (1.2)
2001-08 5.17 0.60 0.18 0.41 (1.7)
61
Appendix: Variable Definitions
Ownership, Concentration, and Performance Measures
HQ State is the U.S. state or the District of Columbia in which the firm headquarters are
located. ER State is the economically most relevant state for a firm and refers to the U.S.
location that is not the firm HQ state but has the highest number of citations in a given
firm-year financial statement (10-K), as defined in Bernile, Kumar, and Sulaeman (2012). HQ
Excess Ownership (LOCOWN) is the state institutional ownership of a firm’s HQ state minus
that state’s average institutional ownership. ER Excess Ownership (ERLOCOWN) is the state
institutional holdings of a firm’s most economically relevant state minus that state’s average state
ownership. State Ownership is the state’s institutional investors’ share in the total institutional
ownership. Ownership Herfindahl is the Herfindahl-Hirschman concentration index (HHI) of
state-level institutional ownership across all U.S. states. Ownership GCI is the concentration
of state-level institutional ownership across all U.S. states adjusted for the “natural” level of
ownership concentration resulting from the geographical clustering of institutional investors.
Next Quarter Return (RAWRET) is the raw stock return in the first quarter following the firm
fiscal year end date. We adjust this raw return using equal-weighted CRSP (EWRET), value-
weighted CRSP (VWRET), and characteristics-based (CADJRET, Daniel, Grinblatt, Titman,
and Wermers (1997)) benchmarks. Institutional ownership (IO) in a firm is the total number of
shares owned by all 13(f) institutions as a proportion of the total number of shares outstanding
for the firm.
Information Environment Variables
Turnover (TURN) is trading volume divided by shares outstanding. Idiosyncratic Volatility
(IVOL) and Idiosyncratic Skewness (ISKEW) are calculated from residuals of annual market-
model regressions of monthly stock returns. Major Auditor (MAJAUD) is an indicator variable
equal to one if the firm hires a major auditor (AU code 01 to 08 in Compustat) in the relevant
fiscal year. Analyst Coverage (ANCOV) is an indicator variable equal to one if the firm is fol-
lowed by at least one equity analyst during the relevant fiscal year according to IBES records.
Number of Analysts (NUMANA) is the number of unique equity analysts providing earnings
forecasts or recommendations for the company stock during the relevant fiscal year. Absolute
Discretionary Accruals (ABSDACC) is the absolute value of the residual from the modified
Jones model of discretionary accruals estimated separately for each year/two-digit SIC code
combination, following Dechow, Sloan, and Sweeney (1995). Local Analyst Coverage (LOCAN-
COV) is an indicator variable equal to one if the firm is followed by at least one equity analyst
located in the firm’s HQ state during the relevant fiscal year. LOCAL is a dummy variable that
62
is set to one if the firm headquarters and the analyst are located in the same state and, zero
otherwise. It is defined in an analogous manner for economically relevant states.
Other Firm Characteristics and Local Attributes
Log(Market Cap) (MCAP) is the stock price multiplied by the number of shares outstanding.
Book-to-Market (BM) is the ratio of book and market equity. Financial Leverage (FINLEV)
is defined as (Total Long Term Debt + Preferred Equity Liquidation Value)/(Market Value of
Equity at Fiscal Year End + Total Long Term Debt + Preferred Equity Liquidation Value).
Firm Age (AGE) is the number of years since a firm’s first appearance in the CRSP database
(i.e., since it becomes a publicly traded firm). 6m Lagged Return (RET6M) is the raw stock
return in the six months leading up to the beginning of the institutional ownership measurement
quarter. Stock Price (PRICE) is the stock price at the beginning of the institutional ownership
measurement quarter. Following Kumar (2009), Lottery-Type Stock (LOTT) is an indicator
variable for firms that are not in the bottom third of volatility, the bottom third of skewness,
or the top third of price. Among the local attributes, Education (EDU) is the fraction of
college graduates in the state. Population Density (POPDEN) is the state’s population divided
by its land area. Catholic-Protestant Ratio (CPRATIO) is the ratio of Catholic adherents to
Protestant adherents in the state. Religiosity (REL) is the fraction of religious adherents in the
state. Republican (REPUB) is the percentage of the state’s registered voters that voted for the
presidential candidate from the Republican party in the last election.
Microstructure Variables
PIN is the probability of private information related trade from Easley, Hvidkjaer, and O’Hara
(2002) model. ADJPIN and PSOS are PIN’s components from Duarte and Young (2009).
ADJPIN is the probability of private information related trade from the extended model in
Duarte and Young (2009), while PSOS is the probability that a given trade happens during a
symmetric order flow shock. Effective Spread (SPREAD) is the difference between the actual
execution price and the midpoint of market quotes. We measure the spread in the [−1:+1]
window around earnings announcements (SPDEARN) and excluding earnings announcements
(SPDOTH).
Analyst-Related Measures
High Performance (HIGHPERF) is the fraction of analyst forecasts in the past year with below-
median forecast errors. Tenure (TENURE) is the number of quarters since the analyst’s first
appearance in the I/B/E/S dataset. Prestigious Broker (PBROKER) is an indicator variable
63
that takes the value of one if the analyst is employed by a prestigious brokerage house. Number
of Firms Covered (NFCOVER) is the number of stocks covered by the analyst in a particular
quarter. The excess coverage of local stocks (EXCOVLOC) is the difference between the fraction
of local stocks in an analyst’s stock coverage portfolio and the fraction of local stocks that the
analyst can potentially cover within the CRSP universe. Equal-weighted excess local coverage
is defined as the number of covered local stocks divided by the number of all covered stocks
minus the number of available local stocks divided by the number of all stocks. Value-weighted
excess local coverage is defined as the total market capitalization of covered local stocks divided
by the total market capitalization of all covered stocks minus the total market capitalization
of all local stocks divided by the total market capitalization of all stocks. Error Below Median
(ERRMDN) is an indicator variable that is set to one if the analyst forecast error is below
the median forecast error of all analyst forecasts for the same firm-quarter. Error Percentile
(ERRPCTL) is the percentile ranking of a given analyst forecast error within the universe of
all forecasts for the same firm-quarter. Following Malloy (2005), demeaned absolute forecast
error (DAFE) is calculated as the absolute forecast error (AFE) for an analyst’s forecast of a
particular firm’s annual earnings minus the mean absolute forecast error for that earnings report.
Absolute forecast error is calculated as the absolute value of an analyst’s latest forecast, minus
the actual earnings, as a percentage of the stock price one-year prior to the beginning of the fiscal
year. Market Reaction (MKTREACT) is the buy-and-hold abnormal return over the three-day
(−1, 0, 1) window around quarterly forecast revisions, calculated as the three-day buy-and-hold
return of the firm’s stock, minus the buy-and-hold return of the equal-weighted CRSP index.
We compute this separately for upward (RETUP) and downward (RETDN) forecast revisions.
64
Table A.1
State-Level Local Bias Estimates
This table reports the local bias for each state in the sub-periods before and after the introduction of Reg FD andSarbanes-Oxley. Missing rows indicate that either there is no publicly-traded firm headquartered in the state orthe state is not an economically relevant state for any publicly-traded firm. The definitions of all variables areprovided in the Appendix. The sample period is from 1996 to 2008.
Headquarters State Economically Relevant StatePREREGFD POSTSOX Change PREREGFD POSTSOX Change
State (1996-2000) (2003-2008) (1996-2000) (2003-2008)AL 8.13% 4.74% −3.39% 0.25% 0.33% 0.08%AK 0.08% 0.12% 0.05%AZ 3.06% −0.08% −3.14% 0.28% 0.64% 0.36%AR 3.18% 2.73% −0.45% 0.68% 0.35% −0.33%CA 12.08% 8.03% −4.05% 9.42% 10.09% 0.68%CO 3.71% 0.02% −3.69% 0.30% 0.32% 0.02%CT 2.30% 4.09% 1.78% 2.29% 8.67% 6.38%DE 3.96% 2.28% −1.68% 1.79% 1.06% −0.73%DC 0.23% 0.58% 0.35% 1.52% 0.36% −1.16%FL 4.75% 3.50% −1.26% 4.94% 1.45% −3.48%GA 1.83% 4.53% 2.70% 0.45% 0.66% 0.21%HI 16.69% 6.48% −10.21% −0.02% 0.05% 0.07%IL 3.00% 1.44% −1.56% 0.18% 4.71% 4.53%IN 7.33% 4.25% −3.07% 0.59% 0.13% −0.46%IA 11.17% 1.22% −9.96% 1.29% 3.12% 1.83%KS 2.00% 3.46% 1.46% 1.46% 1.66% 0.21%KY 14.47% 5.51% −8.96% 1.04% 1.06% 0.02%LA 2.91% 2.47% −0.44% 1.02% 0.40% −0.62%ME 15.62% 13.05% −2.57% 0.91% 0.61% −0.31%MD 5.60% 5.02% −0.58% 2.44% 4.48% 2.03%MA 2.60% −1.22% −3.81% 7.33% 9.18% 1.85%MI 8.94% 3.34% −5.60% 0.98% 0.12% −0.86%MN 19.75% 7.78% −11.97% 4.21% 1.99% −2.22%MS 16.11% 5.21% −10.90% 0.15% 0.02% −0.13%MO 6.17% 2.58% −3.59% 1.95% 1.41% −0.54%MT 0.04% 0.07% 0.03%NE 3.14% 2.48% −0.66% 13.17% 18.32% 5.15%NV 0.81% 0.65% −0.16%NH −0.11% 0.84% 0.94% 2.39% 0.18% −2.21%NJ 3.72% 2.70% −1.02% 2.41% 1.27% −1.15%NM −0.01% 0.13% 0.14%NY 6.75% 2.66% −4.08% 12.74% 12.75% 0.02%NC 10.81% 1.48% −9.33% 2.79% 2.20% −0.59%OH 13.85% 5.33% −8.52% 2.89% 0.65% −2.25%OK 13.80% 1.63% −12.17% 0.04% 0.12% 0.07%OR 5.82% 2.19% −3.63% 4.12% 1.23% −2.89%PA 6.98% 3.91% −3.06% 0.78% 2.41% 1.62%RI 4.42% 1.75% −2.66% 15.81% 0.30% −15.51%SC 17.40% 3.18% −14.22% 0.51% 1.83% 1.32%TN 3.69% 1.95% −1.74% 4.73% 1.46% −3.28%TX 2.06% 3.18% 1.12% −0.81% 0.44% 1.25%UT 7.62% 0.01% −7.61% 0.19% 0.09% −0.10%VT 7.81% 1.14% −6.66%VA 13.26% 7.08% −6.18% 4.63% 5.50% 0.87%WA 6.18% 3.11% −3.08% 2.38% 0.37% −2.02%WV 30.12% 23.21% −6.92% 3.51% 8.23% 4.72%WI 16.30% 5.78% −10.52% 11.02% 8.31% −2.71%Average 8.12% 3.83% −4.29% 2.82% 2.60% −0.22%
65