Here's an idea: Knowledge sharing among competitors to build a critical mass
by
MASSACHUSETTS INSTITUTE
OF TECHNOLOLGY
Tristan Lee Botelho
JUN 02 2015
B.S., Finance and B.A., History
Providence College, 2007
LIBRARIES
Submitted to the MIT Sloan School of Management in Partial Fulfillment of the
Requirements for the Degree of
Master of Science in Management Research
at the
Massachusetts Institute of Technology
January 2015 [we*ru4r
2ay1
D 2015 Massachusetts Institute of Technology. All Rights Reserved.
The author hereby grants to MIT permission to reproduce and to distribute publicly paper
and electronic copies of this thesis document in whole or in part in any medium now known
or hereafter created.
Signature redacted
Signature of Author ...............................................
Tristan L. Botelho
Doctoral Candidate
MIT Sloan School of Management
January 3, 2015
Certified and Accepted by.....................................................................
-
Signature redacted
...........-...
Ezra W. Zuckerman
Professor of Technological Innovation, Entrepreneurship, and Strategic Management
Chair MIT Sloan PhD Program
Thesis Supervisor
1
Here's an idea: Knowledge sharing among competitors to build a critical mass
by
Tristan Lee Botelho
Submitted to the Sloan School of Management on January 3, 2015 in Partial Fulfilment of
the Requirements for the Degree of Master of Science in Management Research
Abstract
Knowledge sharing among competitors is counterintuitive; however, it has been found
to occur in the economy under certain conditions. Recently, in the investment industry,
platforms with the goal of promoting knowledge sharing among investment professionals
have emerged. This knowledge sharing is noteworthy for two reasons: (1) the conditions
that have previously been found to sustain knowledge sharing among competitors are
not present, and (2) it is at odds with the neoclassical efficient-market hypothesis
(EMH). Using a limitation of the EMH framework, I posit that expectations regarding
the strength of the market's efficiency for a stock, measured as the amount of information available about that firm and the level of scrutiny it faces from key market
institutions, plays an important role in sustaining this knowledge sharing. Using a unique
dataset of knowledge sharing among investment professionals, on an online platform,
this study leverages variation in the platform's sharing structure to test this theory. I
find that knowledge sharing is most prominent for stocks that have less information
available and that face a lesser degree of scrutiny from key market institutions. These
findings highlight how market inefficiencies can sustain knowledge sharing among competitors, especially when a critical mass is needed.
Thesis Supervisor: Ezra W. Zuckerman
Title: Alvin J. Siteman (1948) Professor of Entrepreneurship and Strategy, Professor of Technological Innovation, Entrepreneurship, and Strategic Management
2
In 2004, John Schwartz and Whitney Tilson founded the Value Investing Congress (VIC), a
biannual conference for value investors. The VIC brings together investors from around the
globe with the straightforward mission of "providing delegates with immediately actionable
investment recommendations... [and] helping attendees acquire the wisdom they need to understand and profit in the often-irrational market" (Value Investing Congress 2014b). One
investment professional, reflecting on the conference, stated, "When I attend the [VIC], I
know that I will go home with a ton of great investment recommendations and some new
ways of viewing value investing" (Value Investing Congress 2014a). The VIC is one example
of the innovative ventures that have emerged in the investment industry over the past ten
years. A main focus of these ventures is to serve as a platform to facilitate knowledge sharing
among investment professionals from different firms. Specifically, this knowledge sharing involves market participants (or investors) discussing the analysis of historical information that
led them to believe that a security (or stock issued by a firm) is either under- or overvalued.
Knowledge sharing among competitors is puzzling because it can be costly: it is time
consuming, it can reduce competitive advantage, and it does not guarantee a positive return
ex ante. However, there are examples of knowledge sharing among competitors occurring
within the economy (e.g., Appleyard 1996, Fauchart and von Hippel 2008, von Hippel 1987,
Ingram and Roberts 2000, Schrader 1991, cf. Zuckerman and Sgourev 2006). Organizational
and strategy literatures have identified three conditions that often accompany such sharing.
Namely, an expectation of direct reciprocity, a preexisting relationship between those sharing,
and a slow moving industry, which are expected to attenuate the costs related to knowledge
sharing, thus facilitating exchange (Appleyard 1996, Fauchart and von Hippel 2008, von Hippel 1987, Ingram and Roberts 2000, Schrader 1991). Surprisingly, even in the absence of these
conditions, knowledge sharing persists on these platforms in the investment industry, which
leads to my main research question: What explains these instances of knowledge sharing?
Knowledge sharing in the investment industry also directly contradicts the predictions
of the neoclassical finance literature, specifically the efficient-market hypothesis (EMH; Fama
1965, 1970). The EMH assumes that the market instantaneously incorporates all publicly
available information, through an arbitrage and learning process, such that a stock's current
price reflects this information. Therefore, a stock's price should only change when new information is introduced into the market, and mispricings (i.e., the gap between a stock's price
and its value) should be rare. While the EMH has been influential both in research and
practice, recent empirical research provides evidence that markets regularly perform contrary
to the EMH's expectations (e.g., Brunnermeier 2009, LeRoy and Porter 1981, Odean 1998,
Shiller 1981, 2005, Temin and Voth 2004, Thaler 1987, Turco and Zuckerman 2014, White
1990, Zuckerman 2004). Specifically, the level of information available about a firm and the
degree of scrutiny it faces from key market institutions affects how information is incorporated
into its stock's price (e.g., Boehmer and Kelley 2009, Fang and Peress 2009, Yu 2008, Zhang
2006, Zuckerman 1999, 2004). I posit that these market inefficiencies help sustain knowledge
sharing among these competitors, despite the associated costs.
While it may seem obvious that a market participant who finds a mispricing in the
stock market is motivated to act, by buying (or selling) the mispriced stock, the speculative
nature of the market introduces a layer of complexity. Given her capital and time constraints,
the market participant is motivated to act only when she believes that others share this
expectation. In other words, a critical mass is needed to effectively correct mispricings. Inefficiencies in the market make building this critical mass difficult. Thus, I propose that investment professionals who find a mispricing are motivated to share the knowledge that led them
to identify this mispricing in an effort to coordinate with other investment professionals and
build the needed critical mass.
2
In this paper, I test this theory using a unique data set of knowledge sharing among
"buyside" (i.e., hedge fund, mutual fund) investment professionals who are members of a
knowledge sharing platform, the Real Investor's Club (a pseudonym). On this platform, investment professionals, the majority of which are founders or senior members of their organization, share investment recommendations to buy or short sell a stock. They are also required
to supplement this recommendation with a justification for this position, which is visible to
all current and future investment professionals on the platform. The accompanying justification must either be a detailed statement (detailed justification) discussing the thought process
and evidence leading to the given recommendation, with a minimum of 600 words, or a simple
statement (simple justification) supporting their recommendation that can be no longer than
40 words. In my analyses, I consider knowledge sharing to occur only when an investment
professional chooses to include a detailed justification. In contrast with previous research that
has focused on instances of knowledge sharing, this study leverages the existence of the option
to submit a simple justification as a baseline. It is difficult to understand the conditions that
accompany knowledge sharing without the ability to compare knowledge sharing to other
types of information transfer.
Consistent with my theory, I find that knowledge sharing is more likely in the presence
of market inefficiencies. Investment professionals are more likely to engage in knowledge sharing when discussing stocks that have less information available and face a lesser degree of
scrutiny from key market institutions; and when a critical mass is vital, here, short selling.
These findings highlight the role of the market in sustaining knowledge sharing among competitors, specifically, how knowledge sharing may be used to address inefficiencies within a
market. Additionally, these findings shed light on the importance of key institutions as providers of both information and oversight in a market, and how their absence can lead to the
emergence of ventures aimed at filling this void. Lastly, this study introduces a phenomenon
that has received little attention in the organizational and strategy literatures, ventures aimed
3
at facilitating knowledge sharing among competitors. The fact that the majority of the investment professionals in my sample are founders or senior members of their organization,
elucidates the fact that such platforms may play a large role in the competitive and strategic
landscape for organizations going forward.
Conditions for Knowledge Sharing
While the presence and importance of interactions in the economy is not surprising (Baker
1984, Granovetter 1985, Uzzi 1996, 1997, 1999), the motivation for establishing and maintaining interactions among competitors is less clear. Common practical explanations for such
behavior are collusion, collaboration, and knowledge sharing. Of these, the motivations for
knowledge sharing are particularly ambiguous given the associated costs: loss of time, loss of
competitive advantage, and uncertain return. Recent research has aimed to identify the conditions that are often present to help facilitate knowledge sharing among competitors. At one
end of the spectrum, similar organizations who operate in different geographic markets, and
therefore compete for a different consumer, have been found to share very detailed knowledge
with one another (Zuckerman and Sgourev 2006). Other research, focusing on more direct
competitors, has highlighted conditions that reduce the costs of knowledge sharing (Appleyard
1996, Fauchart and von Hippel 2008, von Hippel 1987, Schrader 1991, Stein 2008).
Three conditions in particular have often been discussed: a preexisting relationship
between those sharing, direct reciprocity between those sharing, and a slow moving industry.
A preexisting relationship between those sharing fosters trust, and provides a vehicle for social
sanctioning, thereby reducing informational frictions (Coleman 1988, Greif 1993, Ingram and
Roberts 2000, Stiglitz 1990). Direct reciprocity, the expectation that Actor A shares with
Actor B because Actor A expects that Actor B will initiate sharing in the future, mitigates
the risk of uncertain return (Appleyard 1996, Fauchart and von Hippel 2008, von Hippel 1987,
4
Schrader 1991, Stein 2008). Finally, a slow moving industry ensures that it would be difficult
to implement shared knowledge, hence, protecting the sharer's competitive advantage (Appleyard 1996).
The emergence of platforms broadly connecting investment professionals to facilitate
knowledge sharing is unexpected, given that the conditions often found accompanying
knowledge sharing among competitors are not present. On these platforms, the great majority
of investment professional members are strangers, and they are not able to control the specific
audience that will view their knowledge. Additionally, the speed at which knowledge can be
incorporated in the investment industry is fast. In the investment industry, a market participant who receives knowledge from another market participant can implement this knowledge
almost instantaneously. Most importantly, if the recipient builds on this knowledge, improving
his relative performance, he is not obligated to share this updated knowledge with the sharer.
Similar to the explanations offered by organizational and strategy scholars, in other
contexts, the evidence presented in the economics and finance literature has suggested that
knowledge sharing in the investment industry occurs within small networks (e.g., Cohen,
Frazzini, and Malloy 2008; Duflo and Saez 2002, 2003; Hong, Kubik, and Stein 2004, 2005;
Shiller and Pound 1989). Further, only a small subset of this research has considered the
knowledge sharing platforms, such as the type that is the setting of this study (e.g., Crawford
et al. 2011, Gray 2008, Gray and Kern 2010). For example, Gray and Kern (2010) test if
investment recommendations submitted on knowledge sharing platforms are profitable. After
showing that these recommendations are profitable, they also touch upon the motivation for
these competitors to engage in knowledge sharing, and find varying support for several theories: Competitors share knowledge to receive feedback from peers (cf. Stein 2008); to diversify
their portfolio with the ideas of others (cf. Gray 2008); and, for smaller funds, to attract
additional capital (cf. Dow and Gorton 1994).
5
While this work has made progress in our understanding of knowledge sharing in the
investment industry, it cannot fully explain the observed heterogeneity in this sharing, in
other words, why do competitors engage in knowledge sharing about one stock and not another? The goal of this study is to understand how the expected strength of a stock's arbitrage
and learning process helps sustain knowledge sharing in this competitive industry.
Market Efficiency and the Need for a Critical Mass
The efficient-markets hypothesis (EMH) provides a strong argument for why the interpretation of historical information (i.e., analysis) and the sharing of this knowledge should be
nonexistent. Under its assumptions, the market price for a stock reflects all publicly available
information and the stock's fundamental value. This implies that mispricings in the market
should be extremely unlikely (Fama 1965, 1970). Therefore, this perspective posits that
changes in a stock's price today have no bearing on changes in the stock's price tomorrow,
and its price tomorrow will only change if new information is introduced into the market.
Since it is not possible to predict future information, or the way that this information
will affect current prices, a market participant has a limited set of options. In this model,
market participants act as price takers (or passive investors) who buy and hold stocks that
offer an acceptable income stream. The EMH's focus on new information implies that the
analysis of historical information, and especially sharing this knowledge, is a wasted effort, as
this information it is already incorporated into the stock's current price (Brav and Heaton
2002, Fama 1965, 1970, Malkiel 2003)
The EMH reaches this conclusion through its assumption of an effective arbitrage and
learning process, which is responsible for instantaneously competing away mispricings when
they arise. In this model, arbitragers act as the market's watch dog, continuously monitoring
6
the market with the ability to swoop in to buy (sell) underpriced (overpriced) stocks to eliminate any mispricing. As a result, market participants who employ suboptimal strategies suffer
large losses and are competed out of the market. Others in the market witness this process,
learning from these mistakes, decreasing the probability they reoccur in a future period (see
Brav and Heaton 2002, Zuckerman 2012: 228). This process helps ensure that a stock's price
is aligned with its fundamental (or intrinsic) value.
Though this view has been influential, there is well documented empirical evidence
that suggests that there are weaknesses in the EMH framework, most noticeably in the arbitrage and learning process. In other words, the market has been found to regularly perform
in contradiction to the EMH's expectations, suggesting that mispricings may be more common than expected under the EMH. For example, stocks have been found to demonstrate
excessive volatility, without the introduction of new information (LeRoy and Porter 1981,
Shiller 1981); the market's reaction to new information has been found to be incomplete,
especially for firms with characteristics suggesting a higher likelihood of information uncertainty (Zhang 2006); the concentration of a stock's institutional ownership has been linked to
that stock being price more efficiently (Boehmer and Kelley 2009); and media coverage has
been found to affect how information is incorporated into a stock's price (Fang and Peress
2009, King and Soule 2007). Additionally, the existence of bubbles, where prices increase
beyond a defensible point, based on available information (e.g., the dot-com bubble), suggests
that arbitragers are not always able (or willing) to fix a divergent price-value relationship
(Brunnermeier 2009, Ofek and Richardson 2003, Shiller 2005, Temin and Voth 2004, Turco
and Zuckerman 2014, White 1990).
A mispricing may occur at any point on the continuum of the strength of a stock's
arbitrage and learning process, as markets can never reach complete efficiency (cf. Grossman
and Stiglitz 1980). I posit that the action taken by the market participant, who believes she
7
has found a mispriced stock, depends, in part, on the likely strength of the arbitrage and
learning process of that stock. This is because market participants have a finite amount of
time and capital. Further, given the speculative nature of the market, it is important that a
market participant, who has found a mispricing, expects that others will believe that she is
correct, in order for her to realize a profit-absolute correctness is not sufficient in the shortterm (cf. Ofek and Richardson 2003). In other words, when a mispricing is found it is only
corrected when a critical mass can be formed at this new price.
I posit that if a market participant finds a mispricing when the arbitrage and learning
process is relatively strong, we should expect limited action on her part, beyond buying (or
short selling) the stock. It should only be a matter of time before the market will catch-up
and fix this mispricing, as expected by the EMH, promptly rewarding this market participant.
Here, the costs associated with knowledge sharing do not justify the benefits; therefore,
knowledge sharing should be unlikely. Conversely, when the arbitrage and learning process is
expected to be weak, the market participant who finds a mispricing is motivated to act beyond
buying (or short selling) the stock.
In this scenario, the market is less likely to catch-up in the short-term. Therefore, the
market participant is motivated to attempt and focus the market (i.e., other value oriented
investors) on her price by sharing the knowledge that led to the discovery that the stock was
mispriced. A value investor is one who understands that a stock's price may not reflect its
underlying fundamental value. These market participants are skilled at reviewing the analysis
of others, and must be willing to employ their available capital if convinced. Through
knowledge sharing, the market participant increases the likelihood that she can build a critical
mass by convincing these investment professionals to act similarly by buying (or selling) the
focal stock in order to eliminate this mispricing.
8
Proposition: Given the need to build a critical mass, knowledge sharing is
more likely to occur when the arbitrage and learning process is expected to be
weak.
An example of knowledge sharing leading to the building of a critical mass, occurred
during the Value Investing Congress in 2010. An investment professional, David Einhorn,
presented his investment recommendation to short sell the St. Joe Company, as well as the
analysis that justified this recommendation that the firm was overvalued. The presentation
included a thorough analysis of the St. Joe Company's business, industry, and valuation.
During Einhorn's presentation, the stock's price began to decrease, and it lost approximately
10 percent of its value shortly after his presentation concluded (Selyukh 2010). Through
knowledge sharing, Einhorn was able to build the critical mass at this new, lower price.
A weak arbitrage and learning process
To test this proposition, it is necessary to specify the conditions that likely affect the strength
of a stock's arbitrage and learning process. Though there are no direct measures, the availability of information about a focal firm and the level of scrutiny that it faces from key market
institutions provide good proxies. This is because as the amount of information available
about the firm and the level of scrutiny the firm faces decreases we can expect a greater
degree of uncertainty about the stock's price-value relationship, making a mispricing, or differing opinions regarding the correct price of a stock, more likely (cf. Zuckerman and Rao
2004). This variation, and lack of focus, motivates the use of knowledge sharing to focus the
market on a new price.
The level of information about a firm and scrutiny its faces depends on several key
institutions, such as regulators, critics, and stakeholders. A firm's age is an indicator of the
supply of information for a given firm, with younger firms having less information available.
9
Further, as firms age they have filed more regulatory documents and have developed a consistent track record of performance across various business and economic cycles. Two important critics in this market are "sell side analysts" and the media. Sell side analysts are
responsible for following a set of firms and for issuing periodic reports on these firms to their
company's clients. These reports include an analysis of the firm's historical performance, as
well as the analyst's estimate of the firm's future earnings. Sell side analysts have been found
to play a substantial role in the financial market. For example, coverage from these analysts
has been found to lessen earnings management (Yu 2008), and failure to be covered by these
analysts has been found to lead to a stock price discount (Zuckerman 1999). A second influential critic is the media, whose coverage has been found to have important implications for
a firm's stock price. Firms are often discussed in the media with topics ranging from a firm's
involvement in a given community event to its overall performance. Media coverage, even
when no genuine news is supplied, has been found to affect a stock's price (Fang and Peress
2009). Additionally, substantial (unrelated) media coverage prior to a negative event (e.g., a
protest) has been found to mitigate stock price decline related to this event (King and Soule
2007).
An important stakeholder in this market are institutional investors. These are professional investors who trade in large quantities of stock and are often major holders of a firm's
shares outstanding. These investors have increased their participation in the market over the
past several decades, and have been found to play a large role in a stock's price efficiency
(Boehmer and Kelley 2009). As the availability of information about a firm and the level of
scrutiny it faces from key market institutions decreases there is more variance in a stock's
price-value relationship, increasing the need to build a critical mass. Therefore, we should
expect a greater likelihood of knowledge sharing.
10
Hypothesis 1: Knowledge sharing is more likely as the amount of available
information about a firm and the degree of scrutiny it faces from key market
institutions decreases.
In most market settings, it is difficult for an actor to signal that she believes a certain
product to be mispriced. The U.S. financial market, however, offers a unique mechanism to
handle this scenario, namely short selling. The process of short selling allows an investor who
is pessimistic about the current price of a given stock (i.e., believes it to be overvalued) to
borrow shares of a stock, that they do not own, for a fee, and sell them back to the market.
The investor agrees to return the shares at a later date, along with interest, and any distributions (e.g., dividends) that occur during the borrowing period. The short selling investor
makes a profit when the shares decline to a price that is less than the price at the time he
borrowed the shares (inclusive of fees). Overall, short selling is a risky proposition, especially
when compared to the alternative of buying a stock. When an investor buys common stock,
they own a piece of the firm and their loss is limited to their initial investment. Conversely,
when an investor short sells common stock, they do not own a piece of the firm and they are
faced with the possibility of infinite loss-there is no maximum on the price that the stock
could reach.
The availability of short selling is an important mechanism for an efficient market, as
short selling can be seen as a counter measure to buying and holding a stock (Asquith et al.
2005, Curtis and Fargher 2014). Moreover, the absence of such measures may exacerbate
mispricings, leading to bubbles (e.g., Turco and Zuckerman 2014). However, a necessary condition for successful short selling is the presence of a critical mass of investors who believe the
stock to be overpriced. This is often discussed as a major weakness of short selling (Abreu
and Brunnermeier 2003, Brunnermeier and Nagel 2004). Therefore, while a critical mass is
needed in buying, there is an even greater need for it in the case of short selling, given the
11
risks. In terms of knowledge sharing, we should expect the emphasis on building a critical
mass to lead to a greater likelihood of knowledge sharing.
Hypothesis 2: Knowledge sharing increases when a stock is judged to be overvalued.
Methods
Empirical Context
The setting for this research is the Real Investor's Club (RIC, a pseudonym), a private online
platform that brings together buyside (e.g., hedge fund, mutual fund) investment professionals
with the goal of openly sharing relevant investment recommendations. These are individuals
who analyze securities, such as common stock, with the goal of investing in these securities
on behalf of their organization, as opposed to sell side analysts, who analyze securities with
the goal of disseminating their opinion to a client base. Prospective investment professional
members must apply for entry, which is used by RIC as a safeguard, to help ensure that those
that are given entry are buyside investment professionals. On the platform, an investment
professional's name and their place of employment are visible to other investment professionals
on the platform.
Investment recommendations focus on how a stock (e.g., common stock) is expected to
perform over a future, finite, period. This means that these investment professionals are not
discussing a previous investment that worked out well, or poorly, but a current opportunity.
When an investment professional decides to submit an investment recommendation for a
stock, they must include certain basic information: a recommendation (e.g., buy or sell); a
price target, the price they expect the stock to reach; and an investment horizon, the estimated time it will take for this price target to be reached (e.g., one year). They are also
12
required to include a justification for this recommendation, which is visible to all current and
future investment professionals on the platform. The accompanying justification must either
be a detailed statement (detailed justification) discussing the thought process and valuation
leading to the given recommendation, with a minimum of 600 words, or a simple statement
(simple justification), supporting their recommendation, with a maximum of 40 words.
The vast majority of funds represented on RIC are hedge funds, however, mutual funds,
private equity funds, and other types of investment management funds are represented in the
sample. Most investment professionals who use RIC are employed at smaller funds, however,
large funds are also represented. This skewed distribution is consistent with the hedge fund
industry, where only about 10 percent of hedge funds manage more than $500 million
(Citigroup 2012). In terms of demographics, of the 25 percent who voluntarily submitted their
age, investment professionals using this platform are nearly 36 years old on average and mostly
male. Additionally, over 50 percent of the investment professionals who list their job title are
portfolio managers or in other top management positions.
I conducted over twenty unstructured interviews with investment professionals, twelve
of which were members of RIC, to gain a more in-depth understanding of the investment
industry as well as knowledge sharing within the industry. In my conversations with investment professionals who use RIC, when asked about what led them to join, their reasoning
was almost always the following: to be part of a community of value investors. Many also
expressed gratitude to the community, stating that the investment recommendations that
they read on this platform affected their own view of their portfolio and their investment
strategy. Junior-level investment professionals often stated that they discussed their decision
to join RIC with their firm's management. Therefore, the decision to join these knowledge
sharing platforms can be seen as a strategic decision on the part of the organization.
13
Of those interviewed, who were not part of at least one knowledge sharing platform, two
common reasons were given for this lack of participation. First, some stated that their firm
did not allow detailed justifications for their investments to be shared outside of the firm.
One Director of Research stated that it was important to his firm to keep this type of information away from competing firms. Second, some investment professional stated that they
viewed their knowledge as too valuable to share. Similar to Shiller and Pound (1989), these
investment professionals stated if they engage in knowledge sharing it is only with a select
few.
Defining knowledge sharing: Investment recommendation write-ups
I define knowledge sharing as occurring when an investment professional submits an investment recommendation with a detailed justification. Variation in the sharing structure is important because a baseline is needed for analysis. The majority of extant research on
knowledge sharing focuses on the conditions that are often present conditional on knowledge
sharing taking place. This method of analysis introduces the possibility of measurement error,
as it is unclear if these conditions are also present when knowledge sharing does not occur. A
strength of this setting, and my empirical approach, is the ability to compare cases of
knowledge sharing to a baseline.
Investment recommendations that include a simple justification have an average word
count of approximately 23.65. A sample of simple justifications can be found in Figure 1.
What is immediately evident is that these simple justifications (Figure 1) offer tenuous insight
into the thought process that went into making the specific recommendation. Little to no real
knowledge is being shared, beyond the investment professional's absolute opinion about the
focal stock.
[Figure 1 Here]
14
Conversely, when an investment professional chooses to include a detailed justification, much
more rigor is presented. On average, detailed justifications have a word count greater than
1,400 words. Though the content of these write-ups vary, they commonly include: an interpretation of supporting information, gleaned from meetings with management (and investor
calls), recent news, and company reports (e.g., quarterly filing-10Q); an analysis of comparable companies, such as competitors; a discussion of macroeconomic and industry trends;
and a discussion pertaining to the valuation. During interviews, investment professionals often
stated that these justifications took months to research and hours to write, with one interviewee claiming that writing the investment recommendation took him about 12 hours to
complete.
When asked about the differences between using the simple justification and the detailed justification one investment professional stated, "I see them as two completely different
vehicles, [the detailed justification] lets me fix the market by sharing my due diligence with
the community while the [the simple justification] lets me make a call." The "fix[ing]" of the
market speaks directly to the main proposition of this paper; as the expectation of a weak
arbitrage and learning process increases, we expect the likelihood of knowledge sharing to also
increase because this knowledge sharing provides an opportunity to build the needed critical
mass to address market inefficiencies.
The motivation for choosing to include a simple justification is less clear. Those that
were interviewed were in agreement with the above quote, regarding "mak[ing] a call." In
other words, they valued being able to have proof that they were correct (i.e., they knew a
stock was mispriced). Many investment professionals stated that they avoid reading investment recommendations that include a simple justification, as these do not offer them any
benefit. Importantly, investment professionals are not guided by the platform in terms of
15
which justification type to use. Thus, it is up to the individual investment professional to
proceed as they see fit.
Sample
The analyses utilize unique data collected from RIC, namely the investment recommendations
submitted by investment professionals on this platform, as well as information about the
individual investment professionals. The dataset covers all submitted recommendations pertaining to common stock listed on a U.S. exchange, submitted between 2009 and 2013. The
sample includes 16,019 investment recommendations by 3,829 investment professionals. Of
this total, 22.95 percent (or 3,676 investment recommendations) were submitted with a detailed justification, by 1,523 unique investment professionals, with the remainder (12,343 investment recommendations) submitted with a simple justification, by 3,372 unique investment
professionals. Approximately 23.90 percent of the investment professionals who gain access to
RIC submit only one investment recommendation (using either justification type). Of those
who submit more than one investment recommendation, about 55.50 percent submit more
than three investment recommendations. The content included in a detailed justification is
closely monitored by RIC employees to ensure a minimal level of quality is achieved.
The data from RIC will be supplemented with data about the firms/stocks featured in
the investment recommendation. Financial market data are from the Center for Research in
Security Prices (CRSP), sell side analyst coverage data are from I/B/E/S, and institutional
ownership data are from Thomson-Reuters Institutional Holdings (13F) Database. The form
13F reports the equity holdings by investment managers that have assets under management
of at least $100 million, but does not include short selling positions.
16
Dependent Variable
The main dependent variable for all analyses will be the indicator variable, Knowledge Sharing. It takes the value of one for an investment recommendation that was submitted with a
detailed justification and zero if the investment recommendation was submitted with a simple
justification. Without the choice between justification types it would be difficult to conduct
an analysis linking the likelihood of knowledge sharing to the expected strength of the arbitrage and learning process of a given stock.
Measuring the strength of the arbitrage and learning process
There is an inherent difficulty in measuring the expected strength of the arbitrage and learning process (or the efficiency) of a given stock, as many factors are at play, and no one measure
captures it completely. To address this challenge, multiple measures related to the availability
of information for a given firm and the level of scrutiny it receives from key market institutions
are used as proxies. The goal of this approach is to not overly emphasize the coefficient of any
one measure, but instead interpret the results collectively. Specifically, the analyses will utilize
the following measures: Firm Size, Firm Age, Sell Side Coverage, Institutional Ownership
Concentration, and Media Coverage. For consistency with the presented hypotheses, the inverse of each of these measures, not including Institutional Ownership Concentration, will be
used. For the short selling analysis, the variable Short, will be used, and takes the value of
one when an investment recommendation is to short sell the focal stock.
Firm Size ($ billions). Firm Size is calculated as the market capitalization (share
price*shares outstanding) of the stock featured in the investment recommendation, on the
day prior to the recommendation being posted, using the shares outstanding reported in the
previous quarter. A firm's size has been found to be correlated with other measures related
to the speed at which a stock's price incorporates information (Atiase 1985, Grant 1980).
17
Also, larger firms are more visible and diversified, therefore, these firms are discussed more
frequently and by a broader audience (Fang and Peress 2009). Conversely, smaller firms are
less visible and less prominent in public discussion, therefore, we can logically expect that
information acquisition will be most costly for this subset of firms (Grossman and Miller 1988,
Merton 1987). However, the fact that a firm's size is likely correlated with many unrelated
variables, making the collection of other, more direct, measures necessary.
Firm Age (years). Firm Age is calculated as the difference between the year the investment recommendation was written and the initial year the stock was covered in the CRSP
database (i.e., the firm's IPO year). Firm Age is used to approximate how much history the
firm has that can be analyzed-an overall sense of information availability. It is assumed that
the relationship between time and information is positively correlated. As time passes, there
is more information and certainty about a firm's: strategy, industry, leadership, and performance. Additionally, given that these are public firms, older firms have also submitted more
financial documentation (e.g., quarterly reports) to the Securities and Exchange Commission
(SEC).
Sell Side Coverage (count). Sell Side Coverage is calculated as the sum of the number
of unique earnings estimates for the stock featured in the investment recommendation, in each
of the four quarters prior to that recommendation.' When a firm goes public, sell side analysts
can initiate coverage of the firm. This set of analysts will issue periodic reports about the
stock. These reports routinely include historical information, industry outlooks, earnings estimates, and other analyses. For this measure, if three analysts cover a stock in each of the
last four quarters (3*4 = 12), and two analysts cover that same stock in two of the last four
quarters (2*2 = 4), this measure will equal 16 (12 + 4). If any of these analysts were to
1This variable takes a value of missing if there is no reported coverage. Including these as instances of no coverage
biases the results favorably towards the hypothesized direction.
18
update their initial estimate in a given quarter it would not affect this measure. As this
coverage increases information about the firm reaches a broader audience. Greater coverage
has been found to be directly related to a firm's information availability (e.g., Bushman et al.
2005, Bushman and Smith 2001, Francis et al. 1997, Lang and Lundholm 1996), with an
increased level of scrutiny leading to beneficial outcomes in terms of corporate governance
(Yu 2008).
Institutional Investor Concentration (percentage). Institutional ownership is the
ratio of the number of shares owned by institutional investors to the total number of shares
outstanding for a focal firm. Institutional Investor Concentrationdichotomizes this measure
into two separate variables: Institutional Investor Concentration (Percent, Top 5), which
measures the ratio of the total number of shares held by the five largest institutional investors
to the total number of shares outstanding for a focal firm, and Institutional Investor Concentration (Percent, Other), measures the ratio of the total number of shares held by the institutional investors outside of the five largest to the total number of shares outstanding for a
focal firm.2 These measures are calculated using the institutional ownership data from the
quarter prior to the focal investment recommendation.
For example, if a firm has 100 shares outstanding with 80 of these shares being held by
nine different institutional investors, then the institutional ownership of that firm is 80 percent. If the five largest institutional investors own 60 of those 80 shares then Institutional
Investor Concentration (Percent, Top 5) is equal to 60 percent (60/100) and Institutional
Investor Concentration (Percent, Other) is equal to 20 percent (20/100). Based on recent
evidence, the concentration levels of this ownership among a small subset of institutional
investors has been found to have direct implications for pricing (Boehmer and Kelley 2009).
There are instances that institutional ownership is reported as being greater than 100 percent. As discussed by Asquith and colleagues (2005), there are legitimate reasons for this counterintuitive result. In this sample, institutional
ownership was capped at 100 percent. Results are robust to the removal of these observations.
2
19
Specifically, as institutional ownership is increasingly concentrated among few institutional
investors, prices for that stock are less efficient.
Media Attention (Count). Media Attention is measured as the number of articles
posted about the focal firm in the month preceding the investment recommendation. Article
counts were hand collected from a leading investment focused website that both aggregates
and publishes articles about publicly traded firms. A greater amount of media attention
should lead to an increase in the availability of information about a firm and an increase in
the level of scrutiny that a firm faces. Media coverage has been found to affect a firm's stock
price, even when no genuine news is supplied (Fang and Peress 2009). Further, substantial
(unrelated) media coverage in a previous period has been found to attenuate the effect of
negative future events, such as protests (King and Soule 2007).
Control Variables
Variables at the investment professional- and investment recommendation-level will be utilized
as controls. At the investment professional-level, these measures include: education, ranking
of both undergraduate and graduate institution, and the investment professional's physical
location. At the investment recommendation-level investment horizon, industry fixed effects,
and year fixed effects will be included.
For undergraduate education, the 2013 US News College Ranking (U.S. News & World
Report 2014b) was used to match an investment professional's undergraduate institution to
its ranking. This was also completed for graduate education. For US business schools, the
2013 US News MBA Ranking was used (U.S. News & World Report 2014a) and the 2013
FinancialTimes Global MBA Ranking (Financial Times 2014) was used for non-US business
schools. Investment professionals were grouped into four categories: Top ranked (under)graduate institution (for a ranking of 1-10), Mid ranked (under)graduate institution (for a ranking
20
of 11-50), Bottom ranked (under)graduate institution (for a ranking of 51-100), and Unranked
(under)graduate institution (for a ranking greater than 100 or missing). Additionally, the
dichotomous variable No Grad is coded as one if no graduate school is listed. Controlling for
education is important because certain institutions, especially graduate programs, may train
investors in a certain style.
Given that an investment professional's city has been found to affect investment choice
(e.g., Hong et al. 2000) location is included as a control. Two indicator variables were created:
Major City and Non- US to control for geographical variation. Major City represents large
metropolitan cities in the United States, and is coded one for all investment professionals
from these cities. Some of these cities include: Boston, Chicago, New York City, and San
Francisco. Additionally, the indicator variable Non- US is coded one for all investment professionals located in a city outside of the United States. At the investment recommendationlevel, certain investment horizons, industries, or time periods may be more suitable for
knowledge sharing. Therefore, the indicator variable for recommendations of Less than one
Year is used. Additionally, industry fixed effects and year fixed effects, for the year that the
investment recommendation was submitted, are included in all models. Table 1 provides summary statistics for each of the key variables, separated by justification type.
[Table 1 Here]
Empirical Model
Hypothesis 1 will evaluate how the availability of information and the level of scrutiny from
key market institutions effect the likelihood of knowledge sharing. I estimate the following
linear probability model:
Knowledge Sharingijt = 83Information/Scrutinyi + yXi + 8j + At + cijt,
21
where Knowledge Sharing is a dichotomous variable that takes the value of one when the
investment recommendation was submitted with a detailed justification, where i indexes the
investment recommendation, the unit of analysis; j indexes the industry; and t indexes the
year. Xi is a vector of controls; 8i represents 24 Global Industry Classification Standard
(GICS) industry group subcodes, and an indicator for a missing GICS group subcode, and At
is a vector of year dummies that control for time trends. Robust standard errors are clustered
at the investment professional-level given the possibility that the choice of stocks to analyze
may be correlated within investment professional. While the outcome variable is bounded
between 0 and 1, a linear probability model is used to ease the interpretation of the coefficients
as a change in probability. To evaluate hypothesis 2, I estimate a similar linear probability
model as the one above. Here, I introduce the indicator variable Short to help explain the
likelihood of knowledge sharing if a short selling recommendation is submitted.
Results
Figure 2 presents preliminary evidence that investment professionals utilize knowledge sharing
when discussing smaller firms, in terms of market capitalization. The average market capitalization of a stock discussed using a detailed justification is approximately $7.3 billion, almost
one-third of the size of the average market capitalization of the stocks listed on the Standard
& Poor's 500 (S&P 500; $19.1 billion, at the midpoint of the period under study). A plausible
explanation for this result could be that individuals who participate in such platforms prefer
stocks with a small market capitalization, regardless of availability of information or level of
scrutiny associated with that stock. Comparing the market capitalization of the firms discussed across these two justifications-detailed versus simple justifications-begins to address
these concerns.
[Figure 2 Here]
22
From Figure 2, it is also evident that Firm Size differs between justification types. The
average market capitalization of recommended stocks that use a simple justification is about
$18.8 billion, approximately two and one-half times larger than those that utilized a detailed
justification (p < 0.001), and about the same size as the average market capitalization of the
stocks listed on the S&P 500. This provides additional evidence that investment professionals
are utilizing the justifications for different types of stocks. Specifically, in the fast moving
investment industry, these investment professionals, in the absence of direct reciprocity and
preexisting ties, engage in knowledge sharing when recommending smaller firms.
More rigorous analysis, using the inverse of Firm Size, Sell Side Coverage, Firm Age, and
Media Attention, lead to similar conclusions (Table 2 and Table 3). As firm size decreases
there is a higher likelihood of knowledge sharing (Table 2, M1). For example, a firm with the
market capitalization of $0.5 billion (1.0/0.5 * 0.003 = 0.006) is twice as likely to be associated
with knowledge sharing relative to a firm with the market capitalization of $1.0 billion (1.0/1.0
* 0.003 = 0.003).
Other, more robust, measures also suggest the hypothesized effect, even when controlling
for firm size. Sell Side Coverage is a strong predictor of knowledge sharing, with evidence
suggesting that of knowledge sharing is more likely for stocks with less coverage from sell side
analysts (Table 2, M2). An examination of the effect of Firm Age leads to a similar conclusions
(Table 2, M3): Investment recommendations for a stock that has recently IPO'ed are significantly more likely to involve knowledge sharing than are recommendations for a firm that has
been public for 10 years (7.2 versus 0.7 percent, respectively).
While defining a firm's age as the number of years it has been public has benefits, such
as it accurately measures the firm's interaction with regulation and reporting requirements, a
fair criticism is that some firms have a long private history. These firms would therefore have
extensive information that would be available to the market, which is not being taken into
23
account by the current measure of a firm's age. For example, calculating a firm's age in this
manner treats a firm such as Rackspace, Inc., which was founded in 1998, as the same "age"
as Visa, Inc., which was founded in 1958, because they both had their initial public offering
in 2008. Clearly, given its additional 40 years in operation, we would expect that Visa, Inc.
would have a greater amount of information available and that it would face a greater degree
of scrutiny relative to Rackspace, Inc.
[Table 2 Here]
To address this concern, I collected founding years using a dataset made available by
Laura Field and Jay Ritter (2014). Using these data, I was able to exactly match 53 percent
of the sample and identify the founding year for these firms. I then defined Firm Age (Founding) as the number of years the firm has been in operation, at the time of the investment
recommendation. Using this alternative specification, the effect of a firm's age nearly triples
in magnitude to 0.21 (p < 0.01, Table Al, AM1) relative to the original Firm Age coefficient
(Table 2, M3). To ensure that this increase in magnitude was due to this alternative definition
of a firm's age, and not the reduction of the sample, I reduced the sample to only those firms
that have non-missing value for Firm Age (Founding). The coefficient only marginally increases, from 0.072 (Table 2, M3) to 0.086 (Table Al, AM2); therefore, the increase in magnitude is not caused by the reduction of the sample. Instead, and consistent with my proposed
theory, it is evident that the availability of all types of information is very important in the
decision to engage in knowledge sharing.
I also find that the concentration of a firm's institutional ownership, among a small
set of institutional investors, to be a strong predictor of knowledge sharing (Table 3, M4). A
greater dispersion among institutional investors implies that a greater pool of "smart money,"
a term often used to refer to institutional investors, is monitoring the stock. Institutional
investors are influential in the market given that they hold a larger share of voting power.
24
This allows them to meet with top management, and motivates them to follow the firm much
more closely than typical retail investors (i.e., individual investors). Therefore, as more institutional investors are interested in a given stock, it is more likely that the stock's current
price is closely related to its value, and that the strength of that stock's arbitrage and learning
process is strong.
[Table 3 Here]
Similarly, Media Attention affects the likelihood of knowledge sharing (Table 3, M5).
Approximately, 68.4 percent of firms in the sample received no attention in the month prior
to the recommendation. These firms were given a value of 1, to allow for an inverse to be
taken. Knowledge sharing was about 10.7 percent more likely when the focal firm received at
most one instance of Media Attention, in the month prior to the investment recommendation.
The combination of these results offer strong support for hypothesis 1: that knowledge sharing
is more likely as the level of available information and the degree of scrutiny from key market
institutions decreases.
It is also interesting to note that investment professionals from top undergraduate institutions were about 3 percent less likely to engage in knowledge sharing than those at midranked institutions. Further, investment professionals from bottom ranked graduate institutions were about 13 to 15 percent more likely to engage in knowledge sharing than those from
mid-ranked graduate institutions. Taken together, these results suggest that a possible statusbased mechanism may be at play, where investment professionals from more prestigious institutions are either more attracted to a certain set of stocks or are less dependent on the benefits
associated with knowledge sharing-a possibility for future research.
To test hypothesis 2, we would like to test how knowledge sharing is utilized when a short
selling recommendation is submitted. Across all of the model specifications (Table 4) it is
25
evident that knowledge sharing is significantly more likely for short selling investment recommendation, as compared to a long (or buy) investment recommendation. Short selling recommendations are 9 to 10 percent more likely to involve knowledge sharing, therefore, providing
support for hypothesis 2. An explanation for this result is that the added risks of short selling,
regarding the magnified need for a critical mass, increases the investment professionals' motivation to engage in knowledge sharing.
[Table 4 Here]
It is also possible that stocks that are overvalued are more likely to be associated with
a weaker arbitrage and learning process. Therefore, to disentangle this explanation, from the
expectation that short selling's reliance on a critical mass induces knowledge sharing, I introduce interactions into the model specification. Specifically, I introduce interactions between
Short and the measures of information availability and level of scrutiny (Table 5). Of the
interactions introduced, knowledge sharing is more likely to be associated with a short selling
recommendation as Media Attention decreases (Table 5, M12). A possible explanation for this
significant interaction effect is that wider media coverage may shield a firm from negative
events (cf. King and Soule 2007), therefore, short selling opportunities become less valuable
when the firm is substantially featured in the media.
The lack of significance across other interactions suggests that knowledge sharing is
most salient when a critical mass is a necessity, here, short selling. The amount of information
available about a firm or the level of scrutiny it faces from key institutions does not influence
the likelihood that an investment professional will engage in knowledge sharing when they
submit an investment recommendation to short sell a stock.
[Table 5 Here]
26
The aggregate of these results provides strong evidence for the main proposition of this
paper, specifically, that given the need to build a critical mass, knowledge sharing is more
likely when the arbitrage and learning process for a given stock is expected to be weak.
Therefore, when market inefficiencies are more likely investment professionals are more willing
to incur costs related to knowledge sharing.
Selection Issue
An important concern is that certain investment professionals self-select into knowledge sharing. Specifically, since there is no guidance regarding when an investment recommendation
should include a detailed justification versus a simple justification, certain types of investment
professionals may select into providing detailed justifications, whereas other types of investment professionals may select into providing simple justifications. Additionally, there may be
concerns regarding unobserved investment professional heterogeneity that cannot be controlled for in the above models. While these concerns would not exclude the above knowledge
sharing hypotheses, it would open the door to possible alternative explanations.
To address this concern, investment professionals were identified who utilized both justification types, within the sample, and the sample was then reduced to include only this subset.
Further, individual fixed effects were included in all models for a within-investment professional analysis. The analyses presented in Table 2 and Table 3 were replicated in Table 6 and
the analyses presented in Table 4 were replicated in Table 7, using this new sample and
specification. These do not differ from the main results, neither economically nor statistically
(Table 6 and Table 7). The evidence still strongly suggests that investment professionals chose
to bear the costs of knowledge sharing when recommending firms that had less available
information and faced a lesser degree of scrutiny from key market institutions; and when a
critical mass was especially needed.
27
[Table 6 Here]
[Table 7 Here]
Market Manipulation
It may also be the case that investment professionals engage in knowledge sharing to artificially increase (or decrease) the price of stocks that they own. Though this behavior is unlikely
given that it is illegal, and that the investment professionals who participate are identified by
their name and their firm, it is still worth addressing. In the empirical finance literature, $5
per share is often used as the cutoff to ensure that results are not driven by very small firms,
illiquid stocks, or a bid-ask price bounce (Zhang 2006: 111). The sample was restricted to
those investment professionals who utilized both justification types, and submitted investment
recommendations for stocks with a price per share of over $5. The results are robust to this
specification (Table 8); they do not differ either statistically or substantively from the above
results. Therefore, we can conclude that these results are not driven by either small or illiquid
stocks that are vulnerable to manipulation.
[Table 8 Here]
Discussion
Prior research has provided evidence that, while counterintuitive given the potential costs,
knowledge among competitors does occur. This extant research has identified common conditions that are often present that help attenuate these costs (Appleyard 1996, Fauchart and
von Hippel 2008, von Hippel 1987, Ingram and Roberts 2000, Schrader 1991). Recently, in
the investment industry, despite the absence of these conditions, there has been an emergence
28
of platforms aimed at facilitating knowledge sharing among competitors. I theorize and provide evidence for an additional conditions that helps sustains knowledge sharing, in this context. Drawing on the limitations of the efficient-market hypothesis (EMH; Fama 1965, 1970)
I theorize that inefficiencies in a market increases the difficulty of building a critical mass. In
these cases, the market actors who realize an inefficiency are motivated to engage in knowledge
sharing to attempt to focus the market, thereby building a critical mass
I empirically test this theory using unique data from an online platform, whose goal is
to facilitate knowledge sharing among investment professionals. I find that knowledge sharing
is more likely when a weak arbitrage and learning process is expected, measured as firms with
less information available and that face a lesser degree of scrutiny from key market institutions. These results are robust to the use of multiple measures. Additionally, when a critical
mass is a necessity, here, short selling, there is an increased likelihood of knowledge sharing,
regardless of the amount of information available about the firm or the level of scrutiny it
faces.
Contributions
These findings have several important implications for both research and practice. First,
these findings elucidate an additional condition that sustains knowledge sharing among competitors. Specifically, the need for a critical mass, given market inefficiencies, may motivate
knowledge sharing among competitors. A recent example of such behavior, outside of the
investment industry, comes from Tesla Motors, Inc. In June 2014, Elon Musk announced the
company's decision to no longer enforce their patents (Musk 2014). This was an attempt to
convince others in the industry to embrace Tesla's technology, in other words, a chance for
Tesla to build the needed critical mass around its technology (Kaufman 2014). In his letter,
Musk (2014) describes that though Tesla identified a market inefficiency, and subsequently
29
patented this technology to ensure a competitive advantage, that others were needed to eliminate the market's inefficiency. Only days after this announcement, some of Tesla's major
competitors were already planning to discuss this technology with Tesla (Kaufman 2014).
Second, these findings suggest that it is important to view the EMH as an ideal, and not
as a perfect model for market outcomes. While these findings do not invalidate the EMH, nor
do I attempt to claim that the EMH is wrong, they suggest that it would be more productive
for future research to understand how shortcomings in the EMH elicit (or suppress) market
action. Along this line, this study shows that when the EMH is expected to face larger obstacles market action may take on an unexpected form. These findings join other work in economic and organizational sociology that attempts to move beyond classic models of the market to better understand market action and its implications more broadly (e.g., Baker 1984,
Granovetter 1985, Lounsbury and Hirsch 2010, Zuckerman 1999, 2004).
Third, these findings provide evidence that the absence of key institutions may lead to
the emergence of other ventures to fill this void. Here, platforms aimed at facilitating social
interaction were formed and operated as substitute when there was a lack of information and
oversight from traditional institutions. This new type of organizational form has important
consequences for organizations more broadly, specifically, they aim at connecting competitors
in order to help them improve their overall performance. Given that the majority of the
investment professionals on the platform are senior members of their organization, highlights
the fact that such platforms may play a large role in the competitive and strategic landscape
of organizations going forward.
30
Limitations and Future Directions
While the financial market is one of the most important institutions in the economy, and
understanding processes for the individuals and firms that are a part of the market has enormous implications, this study may be limited by its generalizability to other industries.
Though there are recent examples from of knowledge sharing from other contexts, discussed
above, it may be difficult to identify the applicable measures related to information availability, the level of external scrutiny, and the need for a critical mass in other industries. Additionally, some may argue that the ability to take a position in a stock before engaging in
knowledge sharing offers the investment professional a form of protection against the loss of
their competitive advantage. However, in interviews with investment professionals who did
not participate in any knowledge sharing platform, many stated that such in-depth knowledge
sharing revealed too much about their investment process, and that the ability for others to
improve upon their due diligence, without the guarantee of reciprocation, was too risky of a
proposition.
I see two fruitful avenues for further research in terms of knowledge sharing: First, a focus
on other examples of knowledge sharing in order to gain a deeper understanding of how it is
evolving, and to identify the boundary conditions of the phenomenon. Second, extending this
study by focusing on a more macro unit of analysis in order to gain a better understanding
of what firm policies and strategies lead some firms to embrace knowledge sharing but others
to avoid it. For example, instead of viewing knowledge sharing as a potential threat to superior
performance, there may be instances where firms embrace these platforms given the potential
opportunity for enhancing the performance of involved employees (cf. Ingram and Roberts
2000). It is possible that management in many firms who cannot guarantee reciprocity, or do
not have preexisting connections, are more apt to abstain from knowledge sharing given the
cost. However, in the presence of market inefficiencies and need for a critical mass this may
31
be an ill-advised strategy. A future study may focus on this question by leveraging exogenous
variation that leads to some firms engaging in knowledge sharing, but not other firms.
Finally, in terms of platforms similar to the setting for this research, there is still much
to be known. For example, what are the common antecedents to a similar platform being
created in another industry? Crowd funding platforms offer an interesting opportunity for
research. They offer similarities to the knowledge sharing platforms in the investment industry: Crowd funding platforms emerged as a direct response to market inefficiencies across
various industries. Future research can explore how the availability of information and level
of scrutiny in certain industries affects the likelihood of obtaining crowd funding.
32
References
Abreu, D., M. K. Brunnermeier. 2003. Bubbles and Crashes. Econometrica 71(1) 173-204.
Appleyard, M. M. 1996. How Does Knowledge Flow? Interfirm Patterns in the Semiconductor
Industry. Strateg. Manag. J. 17 137-154.
Asquith, P., P. A. Pathak, J. R. Ritter. 2005. Short interest, institutional ownership, and
stock returns. J. Financ. Econ. 78(2) 243-276.
Atiase, R. K. 1985. Predisclosure Information, Firm Capitalization, and Security Price Behavior Around Earnings Announcements. J. Account. Res. 23(1) 21-36.
Baker, W. E. 1984. The Social Structure of a National Securities Market. Am. J. Sociol. 89(4)
775-811.
Boehmer, E., E. K. Kelley. 2009. Institutional Investors and the Informational Efficiency of
Prices. Rev. Financ. Stud. 22(9) 3563-3594.
Brav, A., J. B. Heaton. 2002. Competing Theories of Financial Anomalies. Rev. Financ. Stud.
15(2) 575-606.
Brunnermeier, M. K. 2009. Bubbles: Entry in New Palgrave Dictionary of Economics. Ed
Steven Durlauf Lawrence Blume.
Brunnermeier, M. K., S. Nagel. 2004. Hedge Funds and the Technology Bubble. J. Finance
59(5) 2013-2040.
Bushman, R. M., J. D. Piotroski, A. J. Smith. 2005. Insider Trading Restrictions and Analysts' Incentives to Follow Firms. J. Finance 60(1) 35-66.
Bushman, R. M., A. J. Smith. 2001. Financial accounting information and corporate governance. J. Account. Econ. 32(1-3) 237-333.
Citigroup. 2012. Hedge Fund Industry Snapshot. Available at: https://www.citibank.com/mss/products/investor_ svcs/prime finance/business_ advisory/docs/hf monthly-may12_primefin.pdf [Accessed November 1, 2014].
Cohen, L., A. Frazzini, C. Malloy. 2008. The Small World of Investing: Board Connections
and Mutual Fund Returns. J. Polit. Econ. 116(5) 951-979.
Coleman, J. S. 1988. Social Capital in the Creation of Human Capital. Am. J. Sociol. 94 S95S120.
Crawford, S., W. Gray, B. Johnson, R. Price. 2011. Do Buy-side Recommendations Have
Investment Value? Unpubl. Manuscr.
Curtis, A., N. L. Fargher. 2014. Does Short Selling Amplify Price Declines or Align Stocks
with Their Fundamental Values? Manag. Sci.
33
Dow, J., G. Gorton. 1994. Arbitrage Chains. J. Finance 49(3) 819-849.
Duflo, E., E. Saez. 2002. Participation and investment decisions in a retirement plan: the
influence of colleagues' choices. J. Public Econ. 85(1) 121-148.
Duflo, E., E. Saez. 2003. The Role of Information and Social Interactions in Retirement Plan
Decisions: Evidence from a Randomized Experiment. Q. J. Econ. 118(3) 815-842.
Fama, E. F. 1965. The Behavior of Stock-Market Prices. J. Bus. 38(1) 34-105.
Fama, E. F. 1970. Efficient Capital Markets: A Review of Theory and Empirical Work. J.
Finance 25(2) 383-417.
Fang, L., J. Peress. 2009. Media Coverage and the Cross-section of Stock Returns. J. Finance
64(5) 2023-2052.
Fauchart, E., E. von Hippel. 2008. Norms-Based Intellectual Property Systems: The Case of
French Chefs. Organ. Sci. 19(2) 187-201.
Field, L., J. R. Ritter. 2014. Field-Ritter Dataset of Company Founding Dates.
Financial Times. 2014. Global MBA Ranking 2013. Available at: http://rankings.ft.com/businessschoolrankings/global-mba-ranking-2013 [Accessed September 17, 2014].
Francis, J., J. Douglas Hanna, D. R. Philbrick. 1997. Management communications with securities analysts. J. Account. Econ. 24(3) 363-394.
Granovetter, M. 1985. Economic Action and Social Structure: The Problem of Embeddedness.
Am. J. Sociol. 91(3) 481-510.
Grant, E. B. 1980. Market Implications of Differential Amounts of Interim Information. J.
Account. Res. 18(1) 255-268.
Gray, W. 2008. Information exchange and the limits of arbitrage. Unpubl. Manuscr.
Gray, W. R., A. Kern. 2010. Do hedge fund managers identify and share profitable ideas.
Unpubl. Manuscr.
Greif, A. 1993. Contract Enforceability and Economic Institutions in Early Trade: The Maghribi Traders' Coalition. Am. Econ. Rev. 83(3) 525-548.
Grossman, S. J., M. H. Miller. 1988. Liquidity and Market Structure. J. Finance 43(3) 617633.
Grossman, S. J., J. E. Stiglitz. 1980. On the Impossibility of Informationally Efficient Markets.
Am. Econ. Rev. 70(3) 393-408.
Von Hippel, E. 1987. Cooperation between rivals: Informal know-how trading. Res. Policy
16(6) 291-302.
34
Hong, H., J. D. Kubik, J. C. Stein. 2004. Social Interaction and Stock-Market Participation.
J. Finance 59(1) 137-163.
Hong, H., J. D. Kubik, J. C. Stein. 2005. Thy Neighbor's Portfolio: Word-of-Mouth Effects in
the Holdings and Trades of Money Managers. J. Finance 60(6) 2801-2824.
Hong, H., T. Lim, J. C. Stein. 2000. Bad News Travels Slowly: Size, Analyst Coverage, and
the Profitability of Momentum Strategies. J. Finance 55(1) 265-295.
Ingram, P., P. W. Roberts. 2000. Friendships among Competitors in the Sydney Hotel Industry. Am. J. Sociol. 106(2) 387-423.
Kaufman, A. C. 2014. Tesla's Clever Patent Move Is Already Paying Off. Huffington Post.
Available at: http://www.huffingtonpost.com/2014/06/16/tesla-patent-superchargerstation_n_5500724.html [Accessed November 2, 2014].
King, B. G., S. A. Soule. 2007. Social Movements as Extra-Institutional Entrepreneurs: The
Effect of Protests on Stock Price Returns. Adm. Sci. Q. 52(3) 413-442.
Lang, M. H., R. J. Lundholm. 1996. Corporate Disclosure Policy and Analyst Behavior. Account. Rev. 71(4) 467-492.
LeRoy, S. F., R. D. Porter. 1981. The Present-Value Relation: Tests Based on Implied Variance Bounds. Econometrica 49(3) 555-574.
Lounsbury, M., P. M. Hirsch. 2010. Markets on Trial: The Economic Sociology of the U.S.
FinancialCrisis. Emerald Group Publishing.
Malkiel, B. G. 2003. The Efficient Market Hypothesis and Its Critics. J. Econ. Perspect. 17(1)
59-82.
Merton, R. C. 1987. A Simple Model of Capital Market Equilibrium with Incomplete Information. J. Finance 42(3) 483-510.
Musk, E. 2014. All Our Patent Are Belong To You I Blog I Tesla Motors. Available at:
http://www.teslamotors.com/blog/all-our-patent-are-belong-you [Accessed November
2, 2014].
Odean, T. 1998. Are Investors Reluctant to Realize Their Losses? J. Finance 53(5) 17751798.
Ofek, E., M. Richardson. 2003. DotCom Mania: The Rise and Fall of Internet Stock Prices.
J. Finance 58(3) 1113-1137.
Schrader, S. 1991. Informal technology transfer between firms: Cooperation through information trading. Res. Policy 20(2) 153-170.
35
Selyukh, A. 2010. St. Joe stock hits 52-wk low on Einhorn's comments. Reuters. Available at:
http://www.reuters.com/article/2010/10/14/stjoe-shares-idUSN1414590320101014
[Accessed November 2, 2014].
Shiller, R. J. 1981. Do Stock Prices Move Too Much to be Justified by Subsequent Changes
in Dividends? Am. Econ. Rev. 71(3) 421-436.
Shiller, R. J. 2005. IrrationalExuberance. Random House LLC.
Shiller, R. J., J. Pound. 1989. Survey evidence on diffusion of interest and information among
investors. J. Econ. Behav. Organ. 12(1) 47-66.
Stein, J. C. 2008. Conversations among Competitors. Am. Econ. Rev. 98(5) 2150-2162.
Stiglitz, J. E. 1990. Peer Monitoring and Credit Markets. World Bank Econ. Rev. 4(3) 351366.
Temin, P., H.-J. Voth. 2004. Riding the South Sea Bubble. Am. Econ. Rev. 94(5) 1654-1668.
Thaler, R. H. 1987. Amomalies: The January Effect. J. Econ. Perspect. 1(1) 197-201.
Turco, C., E. Zuckerman. 2014. So You Think You Can Dance? Lessons from the US Private
Equity Bubble. Sociol. Sci. 1 81-101.
U.S. News & World Report. 2014a. 2013 Best Business Schools. Available at: http://gradschools.usnews.rankingsandreviews.com/best-graduate-schools/top-businessschools/mba-rankings [Accessed September 17, 2014].
U.S. News & World Report. 2014b. 2013 Best Colleges Ranking. Available at: http://colleges.usnews.rankingsandreviews.com/best-colleges [Accessed September 17, 2014].
Uzzi, B. 1996. The Sources and Consequences of Embeddedness for the Economic Performance
of Organizations: The Network Effect. Am. Sociol. Rev. 61(4) 674-698.
Uzzi, B. 1997. Social Structure and Competition in Interfirm Networks: The Paradox of Embeddedness. Adm. Sci. Q. 42(1) 35-67.
Uzzi, B. 1999. Embeddedness in the Making of Financial Capital: How Social Relations and
Networks Benefit Firms Seeking Financing. Am. Sociol. Rev. 64(4) 481-505.
Value Investing Congress. 2014a. New York 2013 Highlights. Available at: http://www.valueinvestingcongress.com/the-congress/past-congress-highlights/2013/newyork/
[Accessed September 17, 2014].
Value Investing Congress. 2014b. Our Mission. Available at: http://www.valueinvestingcongress.com/the congress/our-mission/ [Accessed September 17, 2014].
White, E. N. 1990. The Stock Market Boom and Crash of 1929 Revisited. J. Econ. Perspect.
4(2) 67-83.
36
Yu, F. (Frank). 2008. Analyst coverage and earnings management. J. Financ. Econ. 88(2)
245-271.
Zhang, X. F. 2006. Information Uncertainty and Stock Returns. J. Finance 61(1) 105-137.
Zuckerman, E. W. 1999. The Categorical Imperative: Securities Analysts and the Illegitimacy
Discount. Am. J. Sociol. 104(5) 1398-1438.
Zuckerman, E. W. 2004. Structural Incoherence and Stock Market Activity. Am. Sociol. Rev.
69(3) 405-432.
Zuckerman, E. W. 2012. Market efficiency: a sociological perspective. Oxf. Handb. Sociol.
Finance 223.
Zuckerman, E. W., H. Rao. 2004. Shrewd, crude or simply deluded? Comovement and the
internet stock phenomenon. Ind. Corp. Change 13(1) 171-212.
Zuckerman, E. W., S. V. Sgourev. 2006. Peer Capitalism: Parallel Relationships in the U.S.
Economy. Am. J. Sociol. 111(5) 1327-1366.
37
Figures and Tables
Figure 1
Others in the industry are trading at 14 times price-to-earnings. [Firm] is only
trading at 10!
|With home building picking back up they should be able to capitalize
IThe entire industry is undervalued and [FIRM] is has the highest upside.
Note: Each row represents an example of a simple justification. This reproduction is paraphrased from actual data and removes any reference to real
firms.
38
I
Figure 2
Avg. Market Cap. of all Invest. Reco. by Quarter
2009 to 2013
A
I'
I'
I1
'I
I
/
/
/
I'
I
-~ .-
-
-
-
~
-
-
-
-
-
-
--
A
/
C>
F---0
Simple Justification
5
V
Detailed Justification
15
10
Quarter
20
Note: Market capitalization is averaged across investment recommendations
within quarter and within justification type from 2009 to 2013. The horizontal line represents the overall sample mean for the respective justification
39
Table 1: Summary Statistics of Variables
Panel A: Detailed Justification
Variable
Obs.
Mean
Std. Dev.
Min.
Max.
Key Variables
Firm Size
Sell Side Coverage
Firm Age
Media Attention
Instit. Investor Concen. (Per., Top 5)
Instit. Investor Concen. (Per., Other)
Recommendation: Short
3,676
3,233
3,676
3,676
3,305
3,305
3,676
7.27
51.10
17.09
1.39
0.29
0.36
0.01
4.00
0.00
0.00
0.16
31.34
42.35
17.56
7.58
.13
.21
0.36
0.00
594.86
272.00
88.00
183.00
.0(
0.80
1.00
Control Variables-Recommendation-Level
Time Frame: Less than one Year
3,676
0.40
0.49
0.00
1.00
3,676
3,676
0.74
0.05
0.44
0.22
0.00
0.00
1.00
1.00
3,676
3,676
3,676
3,676
3,676
3,676
3,676
0.23
0.11
0.00
0.28
0.03
0.04
0.56
0.42
0.32
0.47
0.45
0.17
0.19
0.50
0.00
0.00
0.00
0.00
0.00
0.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
Obs.
Mean
Std. Dev.
Mi.
Max.
Key Variables
Firm Size
Sell Side Coverage
Firm Age
Media Attention
Instit. Investor Concen. (Per., Top 5)
Instit. Investor Concen. (Per., Other)
Recommendation: Short
12,343
11,324
12,343
12,343
11,012
11,012
12,343
18.822
65.24
19.246
1.94
0.27
0.40
0.06
49.13
47.23
19.17
7.78
0.12
0.19
0.24
0.00
4.00
0.00
0.00
0.00
0.00
0.00
657.98
272.00
88.00
147.00
1.00
0.81
1.00
Control Variables-Recommendation-Level
Time Frame: Less than one Year
12,343
0.15
0.36
0.00
1.00
12,343
12,343
0.72
0.45
0.26
0.00
0.00
1.00
1.00
12,343
12,343
12,343
12,343
12,343
12,343
12,343
0.28
0.45
0.29
0.46
0.45
0.12
0.15
0.49
0.00
0.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
Control Variables-Investment Professional-Level
Location: Majoy City
Location: Non-US
Education (Ref: Mid-Rank)
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
0.32
0.0(
0.00
Panel B: Simple Justification
Variable
Control Variables-Investment Professional-Level
Location: Majoy City
Location: Non-US
Education (Ref: Mid-Rank)
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
40
0.07
0.09
0.31
0.29
0.01
0.02
0.59
0.00
0.00
0.00
0.00
0.00
Table 2: Linear Probability Model, regressing Knowledge Sharing on measures
of information availability and scrutiny. Unit of analysis is investment professional-idea.
M1
M2
0.072
Location: Majoy City
Location: Non-US
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate
Rank: Unranked
Graduate Rank: No Grad.
Constant
IK-Square Adj.
Observations
*
+
*
-0.009
-0.010
(0.022)
0.1.27
(0.054)
*
(0.043)
-0.020
(0.020)
0.110
(0.026)
0.140
16,019
*
(0.020)
0.127
(0.058)
0.039
(0.040)
-0.019
(0.019)
0.090
(0.026)
0.058
0.149
14,543
*
+
*
*
*
0.003
(0.001)
0.203
(0.014)
0.024
(0.014)
-0.034
(0.024)
-0.034
(0.013)
0.020
(0.024)
-0.015
(0.013)
0.010
(0.022)
0.130
(0.054)
0.060
(0.042)
-0.019
(0.020)
0.100
(0.026)
0.142
16,019
*
*
*
Time Frame: Less than one Year
0.003
(0.001)
0.198
(0.014)
0.025
(0.014)
-0.024
(0.023)
-0.035
(0.013)
0.018
(0.022)
-0.015
(0.014)
*
+
0.003
(0.001)
0.204
(0.014)
0.025
(0.014)
-0.033
(0.024)
-0.034
(0.013)
0.021
(0.024)
-0.014
(0.013)
(0.016)
**
*
Firm Size
*
Firm Age
*
0.637
(0.088)
*
Sell Side Coverage
M3
.Note: Miodels contain year and industry flxed ettects.
Robust Standard errors clustered at the investment professional-level are in parentheses.
Significance Levels: + p
0.10, * p < 0.05, ** p ! 0.01, *** p < 0.001.
41
Table 3: Linear Probability Model, regressing Knowledge Sharing on
measures of information availability and scrutiny. Unit of
analysis I investment professional-idea.
M4
Instit. Investor Concen. (Per., Top 5)
Instit. Investor Concen. (Per., Other)
Media Attention
0.168
(0.031)
-0.141
(0.024)
M5
* **
* **
0.107
(0.012)
0.003
(0.001)
0.204
(0.014)
Firm Size
*
+
0.002
**
(0.001)
Time Frame: Less than one Year
0.196
(0.014)
Location: Majoy City
0.033
*
0.025
(0.014)
(0.014)
Location: Non-US
--0.026
-0.030
(0.024)
(0.024)
Undergraduate Rank: Top
-0.035
**
-0.033
(0.013)
(0.013)
Undergraduate Rank: Bottom
0.022
0.021
(0.023)
(0.024)
Undergraduate Rank: Unranked
0.014
-0.013
(0.013)
(0.013)
Graduate Rank: Top
-0.002
-0.011
(0.021)
(0.022)
Graduate Rank: Bottom.
0.146
**
0.129
(0.054)
(0.055)
Graduate Rank: Unranked
0.056
0.047
(0.040)
(0.043)
Graduate Rank: No Grad.
-0.013
-0.021
(0.019)
(0.020)
Constant
0.107
***
0.013
(0.028)
(0.028)
A-Square Adj.
U. 143
U.145
Observations
14,317
16,019
Note: Models contain year and industry tixed ettects.
Robust Standard errors clustered at the investment professional-level are in parentheses.
Significance Levels: + p <0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.
42
Table 4: Linear Probability Model, regressing Knowledge Sharing on Short Selling.
Unit of analysis is investment professional-idea.
M7
M6
Sell Side Coverage
0.091
(0.018)
0.096
*
(0.018)
0.661
(0.088)
*4**
Instit. Investor Concen. (Per., Top 5)
Time Frame: Less than one Year
Location: Majoy City
Location: Non-US
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
Constant
* **
0.160
(0.030)
Instit. Investor Concen. (Per., Other)
Firm Size
0.089
(0.018)
0.003
(0.001)
0.191
(0.014)
0.024
(0.014)
--0.=32
(0.024)
-0.034
(0.013)
0.020
(0.024)
0.003
*
*
(0.001)
0.184
*
+
(0.014)
0.025
(0.014)
-0.023
+
(0.023)
*
-0.m15
-0.035
(0.024)
-0.035
(0.013)
-0.014
(0.013)
-0.0N3
(0.021)
0.148
(0.055)
0.048
(0.040)
-0.013
(0.019)
0.129
(0.059)
0.041
(0.040)
-0.01.9
(0.019)
0.092
**
0.110
(0.028)
It-Square Adj.
0.143
0.147
Observations
16,01.9
14,543
14,317
Note: Models contain year and industry fxed effects.
Robust Standard errors clustered at the investment professional-level are in parentheses.
Significance Levels: + p < 0.10, * p < 0.05, ** p <_ 0.01, *** p < 0.001.
43
***
**
0.021
(0.020)
(0,025)
U. 153
**
(0.023)
-0.01g
**
0.182
(0.014)
0.033
(0.014)
-0.024
(0.013)
0.00)
(0.022)
0.015
(0.013)
-0.010
0.002
(0.001)
*
(0.013)
(0.022)
0.129
(0.055)
0.059
(0.043)
-0.020
(0.020)
0.112
(0.026)
-0.137
(0.024)
*
Recommendation: Short
M8
**
Table 5: Linear Probability Model, regressing Knowledge Sharing on Short Selling.
Unit of analysis is user-idea.
M9
Recommendation: Short
Firm Age
0.085
(0.020)
0.064
(0.016)
M10
***
***
0.100
M12
Ml I
*
0.061
(0.044)
0.011
(0.033)
(0.021)
0.669
(0.090)
Sell Side Coverage
0.099
(0.012)
Media Attention
0.156
(0.031)
0.142
(0.024)
Instit. Investor Concen. (Per., Top 5)
Instit. Investor Concen. (Per., Other)
Short X Firm Age
Short .X Sell Side Coverage
0.006
(0.048)
-0.133
(0.359)
0.109
(0.039)
Short X Media Attention
0.021
(0.110)
Short X Instit. Inv. Concen. (Per., Top 5)
Short X Instit. Inv. Concen. (Per., Other)
Firm Size
Time Frame: Less than one Year
Location: Majoy City
Location: Non-US
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
0.058
(0.087)
0.190
(0.014)
0.023
(0.014)
0.033
(0.024)
-0.035
(0.013)
0.0N9
(0.023)
(0.015
(0.013)
-0.010
(0.021)
0.003
(0.001)
0.184
(0.014)
*
(0.014)
0.023
(0.023)
**-0.035
(0.013}
0.017
***
0.024
(0,014)
0.0N9
-0.035
(0.013)
0.021
(0.013)
-0.012
-0.003
(0.021)
(0.021)
-0.014
*0.130
(0.056)
0.057
(0 043)
-0.021
-0.019
(0.019)
0.091
(0.025)
U. 152
14,543
***
(0.020)
0.023
(0.028)
0.149
R-Square Adj.
Observations
16,01.9
16,019
Note: Models contain year and industry fixed effects.
Robust Standard errors clustered at the investment professional-level are In parentheses.
Significance Levels: + p < 0.10, * p ! 0.05, ** p < 0.01, *** P < 0.001.
44
0.024
(0.024)
**
(0.013)
--0.010
(0.020)
(0.040)
0.145
(0.024)
-0.034
(0.013)
0.033
(0.014)
(0,023)
0.014
(0.013)
(OM042)
(0.026)
4-
(0.001)
0.183
(0.014)
(0.023)
0.061
**
*
-0.028
**
0.002
**
0.189
(0.014)
-0.0102
-0.019
0.003
(0.001)
(0.022)
(0.055)
0.103
*
0.024
+-
0.129
(0.059)
0.041
(0.020)
Constant
*
0.003
(0.001)
0.148
(0.055)
0.048
(0.040)
-0.013
(0.019)
0.113
(0.028)
.147
14,317
*5*
Table 6: Linear Probability Model, regressing Knowledge Sharing on measures of information availability
and scrutiny. Sample restricted to only those investment profeesionals who utilized both Justifications.
Unit of analysis Is investment professional-idea.
R2
R5
Media Attention
Firm Size
01,
0.003
(0.001)
0.254
(0.017)
0.356
(0.305)
0.100
(0.305)
-0.017
(0.266)
0.093
(0.406)
-0.203
(0.405)
-1.129
(0.581)
-0.688
(0.373)
-0.993
(0.647)
-0.682
0.811
(0.437)
0.002
*
(0.001)
Time Frame: Less than one Year
***
0.245
*
(0.018)
Location: Majoy City
0.268
(0.321)
Location: Non-US
0.014
(0.320)
Undergraduate Rank: Top
-0,015
(0.266)
Undergraduate Rank: Bottom
0.084
(0.405)
Undergraduate Rank: Unranked
-0.129
(0.416)
Graduate Rank: Top
+
1.203
(0.626)
Graduate Rank: Bottom
+
-0.809
*
(0.412)
Graduate Rank: Unranked
-1.070
(0.688)
Graduate Rank: No Grad.
**
0.843
**
Constant
+
0.972
*
(0.464)
R-Square Adj.
U.1W
0.131
Observations
5,760
5,136
Note: Models contain year, industry, and investment protessional tixed ettects.
Standard errors are in parentheses.
Significance Levels: + p < 0.10, * p ! 0.05, ** P < 0.01, *** p < 0.001.
*
0.003
(0.001)
0.253
(0.017)
0.367
(0.305)
0.083
(0.305)
--0.020
(0.266)
0.083
(0.406)
-0.206
(0.405)
.L150
(0.580)
-0.679
(0.373)
-0.998
(0.647)
-0.684
0.806
(0.437)
U..131
5,760
*
*
*
+
*
+
0.002
(0.001)
0.245
(0.018)
0.355
(0.306)
0.111
(0.305)
-0.042
(0.267)
0.022
(0.407)
-0.286
(0.406)
1.081
(0.590)
-0.563
(0.392)
-0.965
(0.655)
--0.620
0.798
(0.449)
U.128
5,158
*
*
+
0.116
(0.025)
0.003
(0.001)
0.252
(0.017)
0.355
(0.304)
0.153
(0.304)
0.023
(0.266)
0.078
(0.405)
-0.243
(0.404)
1.182
*
0.200
(0.058)
0.143
*
(0.025)
Instit. Investor Concen. (Per., Other)
*
Instit. Investor Concen. (Per., Top 5)
*
0.057
Firm Age
*
0.391
(0.139)
R4
*
Sell Side Coverage
R3
*
RI
(0.580)
-0.733
(0.373)
-0.969
(0.646)
*
-0.696
+
0.745
(0.436)
U.134
5,760
**
Table 7: Linear Probability Model, regressing Knowledge Sharing on Short Selling.
Sample restricted to only those investment professionals who utilized both justifications.
Unit of analysis is investment professional-idaea.
Sell Side Coverage
*
Location: Majoy City
Location: Non-US
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
Constant
R-Square Adj.
U.132
0.002
(0.001)
0.234
(0.018)
0.297
(0.320)
0.034
(0.320)
0.019
(0.265)
0.057
(0.404)
(0.171
(0.416)
*
1.229
*
1.118
+
(0.411.)
(0.589)
0.561
(0.392)
*
(0.687)
0.823
(0.307)
**
(0.655)
0.615
(0.279)
*
Time Frame: Less than one Year
j
0.977
0.814
*
*
*
*
-0.044
0.001
(0.406)
(0.36
(0.4:06)
(0.625)
*
+
0.792
**
(0.267)
1.126
1.028
*
(0.464)
U.153
5,136
5,760
Observations
Note: Models contain year, ndustry. and investment protessional tixed ettects.
Standard errors are in parentheses.
Significance Levels: + p < 0.10, * p K 0.05, ** p K 0.01, *** p S 0.001.
46
0.196
(0.058)
-0.142
(0.041)
0.002
(0.001)
0.235
(0.018)
0.385
(0.306)
0.134
(0.305)
+
Instit. Investor Concen. (Per., Other)
0.003
(0.001)
0.243
(0.017)
0.392
(0.305)
0.127
(0.304)
.0.020
(0.266)
0.069
(0,405)
(0.250
(0.405)
-1.85
(0.580)
0.695
(0,373)
-.1079
(0,647)
-0.689
(0.261)
0.839
(0.437)
**
**
(Per., Top 5)
Firm Size
0.067
(0.023)
*
Instit. Investor Concen.
0.080
(0.024)
0,412
(0.139)
*
*
+
0,077
(0.022)
Recommendation: Short
R8
R7
R6
(0.449)
U.14U
5,158
Table 8: Linear Probability Model, regressing Knowledge Sharing on measures of information availability and scrutiny.
Sample restricted to only those investment professionals who utilized both justifications and with stock
price greater than $5.00 per share. Unit of analysis is investment professIonal-idea.
RIO
0.357
R11
R12
R13
R14
*
R,9
Sell Side Coverage
(0.169)
Firm Age
0.074
**
Time Frame: Less than one Year
Location: Majoy City
Location0.40
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
Constant
(0.377)
(0.025
(0.264)
-0.080
(0.415)
0.219
(0.402)
-0.713
(0.622)
-0.603
(0.395)
0.589
(0.683)
-0.765
(0.267)
0.91.9
0.005
(0.003)
*
*
0.230
+
*
(0.020)
-0.243
(0,391)
-0.500
(0.390)
*
*
-0.434
*
(0.438)
(0.463)
k-Square Adj.
0.132
0.129
Observations
4,988
4,532
Note: Models contain year, industry, and investment protessional fixed effects.
Standard errors are in parentheses.
Significance Levels: 4 p < 010, * p L 0.05, ** p < 0.01, *** p < 0.001.
0.910
(0.438)
0.134
4,988
**
*
(0.378)
0.389
(0.377)
-0.039
(0.265)
-0.121
(0.416)
-0.298
(0.403)
-0.680
(0,632)
-0.492
(0.419)
-0.575
(0.693)
-0.738
(0.289)
0.897
(0.453)
0.132
4,470
0.006
(0.002)
*
0.229
*
1
(0,377)
(0.377)
-0.027
(0,264)
-0.086
(0.415)
0.222
(0.402)
-0.736
(0.621)
-0.596
(0.395)
-0.591
(0.682)
-0.766
(0.267)
0.007
(0.002)
(0.020
- 0 .1 7
0.141
(0.265)
-0.083
(0.416)
0.137
(0.415)
-0.683
(0.663)
-0.603
(0.431)
0.559
(0.721)
-0.824
(0.306)
0.974
*
(0.018
-0.021
**
0.007
(0.002)
0.236
0.234
*
*
(0.018)
(0.(19)
-0.156
(0.377)
0.40t
(0.376)
-0.020
(0.264)
-0.085
(0.415)
0.253
(0.401)
-0.759
(0.621)
-0.648
(0.395)
-O.103
(0.377)
0.345
(0.376)
-0,028
(0.264)
-0.120
(0.415)
0.286
(0.401)
-0.786
(0.621)
-0.590
(0.395)
0.703
(0.682)
-0.768
(0.266)
-0.566
*
*
(0.682)
-0.778
(0.266)
0.859
(0,437)
0.135
4.988
**
*
0.954
(0.437)
0.136
4,988
*
0.107
(0.024)
0.008
(0.002)
0.221
*
0.007
(0.002)
0.237
(0.018)
-0.15A
(0.377)
*
Firm Size
*
0.104
(0.026)
Recormrnendation: Short
*
Media Attention
*
Instit. Investor Concen. (Per., Other)
+
0.229
(0.063)
-0.080
(0.047)
*
(0.026)
Instit. Investor Concen. (Per., Top 5)
Appendix
Table Al: Linear Probability Model, regressing Knowledge Sharing on two
definitions of firm age. Unit of analysis is investment professional-idea.
AM2
AMI
0.003
Firm Size
Time Frame: Less than one Year
Location: Majoy City
Location: Non-US
Undergraduate Rank: Top
Undergraduate Rank: Bottom
Undergraduate Rank: Unranked
Graduate Rank: Top
Graduate Rank: Bottom
Graduate Rank: Unranked
Graduate Rank: No Grad.
Constant
(0.001)
0.196
(0.017)
0.035
(0.016)
-0.036
(0.026)
-0.023
(0.015)
0.016
(0.026)
0.005
(0.016)
0.007
(0.026)
0.158
(0.071)
0.074
(0.044)
0.008
(0.024)
0.054
(0.032)
**
*
*
0.003
(0.001)
0.196
(0.017)
0.034
**
*
0.086
(0.024)
Firm Age
*
**
*
0.206
(0,074)
(0.016)
-0.038
(0.025)
*
-
-0.023
(0.015)
0.015
(0.026)
0.005
(0.016)
0.006
(0.026)
0.160
(0.071)
0.077
(0.044)
-0.008
(0.024)
0.056
(0.032)
*
Firm Age (Founding)
4
4
0.149
0.149
H-Square Adj.
8,487
8,487
Observations
Note: Models contain year and industry fixed effects.
Robust Standard errors clustered at the investment professional-level are in parentheses.
0.01, *** p < 0.001.
0.05, ** p
0.10, * p
Significance Levels: + p
48