Investor Demand For Sell-Side Research: Lawrence@haas - Berkeley.edu
Investor Demand For Sell-Side Research: Lawrence@haas - Berkeley.edu
Alastair Lawrence
[email protected]
James Ryans
[email protected]
Estelle Sun
[email protected]
June 2016
ACKNOWLEDGMENTS: We are grateful to Yahoo! Inc. for providing the Yahoo Finance web-traffic data.
We also are grateful to stocktargetprices.com for providing the initial web-traffic data used in an earlier version of
this study. We have received helpful comments from Beverly Walther (Editor), two anonymous reviewers, Yonca
Ertimur, Weili Ge, Rick Johnston, Stephannie Larocque, Alina Lerman, Joshua Livnat, Maria Loumioti, Matthew
Lyle, Sarah McVay, Laura Wellman, and from seminar participants at the 2013 AAA FARS Midyear Conference
(San Diego), 2013 AAA Western Region Meeting Doctoral Interchange (San Francisco), and 2013 EAA Annual
Congress (Paris).
ABSTRACT: We use daily page views of analyst estimates, ratings, and target prices on
Yahoo Finance to understand when such users seek sell-side analyst research. Demand for this
information is most pronounced on days with earnings announcements, management guidance,
and All-Star analyst reports. Surprisingly, demand does not increase at Form 10-K and Form
10-Q filings. While the overall demand for analyst estimates is 19.9 percent less than for
analyst ratings and target prices, on earnings announcement and management guidance days
this preference is reversed. Moreover, the demand for analyst information substantially trumps
that of SEC filings and financial statement information.
KEYWORDS: Analyst demand; financial reporting events; analyst ability; retail investors.
In this study we use Yahoo Finance web traffic to provide insights into the value that
analysts provide to these users, both in general and surrounding specific information events.
We further investigate when demand for analyst estimates (EPS and sales) differs from demand
for opinions (ratings and target prices), an important determinant for shaping analysts decision
of when to issue a report and what type of information to include. Despite the prevalence of
analysts in financial markets, academic research is still uncertain concerning the underlying
value of sell-side equity analysts and why they provide certain types of information (Beyer,
obvious method to assess where analysts provide value, direct measures of demand for analyst
information have heretofore been unavailable. 1 As a result, prior research frequently measures
investor demand using the number of analysts following a firm (e.g., Bhushan 1989; OBrien
and Bhushan 1990; Lang and Lundholm 1996; Barth, Kasznik, and McNichols 2001; Lehavy,
Li, and Merkley 2011) and the stock market responses to analyst reports (e.g., Frankel,
Kothari, and Weber 2006; Hugon and Muslu 2010; Altnkl, Balashov, and Hansen 2013).
analyst opinions (buy/sell ratings and target prices) and estimates (earnings, revenue, and
growth) to increase our understanding of: (i) the determinants of investor demand for analyst
information, and (ii) the determinants of investor demand for analyst estimates versus opinions
and target prices. We supplement these analyses by investigating how the demand for analyst
1
Databases such as Institutional Brokers Estimate System (I/B/E/S) have coverage of the supply of the main quantitative
information elements provided by analysts, including EPS and revenue forecasts.
1
information differs from that of financial statement information and SEC filingskey
information sources that analysts process for investorsand the conditions under which
Yahoo Finance (finance.yahoo.com) is the most popular web site for financial
information in the U.S. (Yahoo 2016). It provides firm-specific coverage of all publicly-listed
companies on U.S. exchanges. Because of the large number of daily users, the web-traffic
patterns on Yahoo Finance are likely representative of the U.S. retail investor population, and
not of professional investors. While the predominantly retail clientele of Yahoo Finance limits
our ability to generalize these results to professional investors, our results are nonetheless
interesting as retail investors have been shown to affect prices (e.g., Lee, Shleifer, and Thaler
1991; Barber, Odean, and Zhu 2009; Aboody, Lehavy, and Trueman 2010), and are an
important audience for stock analysts, who frequently appear in the mass media to promote
their research. Additionally, Shiller (2015) argues that the distinction between retail and
professional investors may be subtler than once thought as retail investors get advice from
professional investors, often have substantial resources, are usually educated and intelligent,
Our main findings are as follows. First, demand for analyst information on Yahoo
Finance is highest on days with management guidance, earnings announcements, and Form 8-
Ks. Surprisingly, we do not find increases in demand on days with Form 10-K and Form 10-Q
filings. Specifically, investor searches for analyst information are 181.8, 152.9, and 5.9 percent
higher on days with management guidance, earnings announcements, and Form 8-Ks,
respectively, compared to the overall average level of search. Demand also increases the day
2
investor searches for analyst information are 10.6, 9.4, and 2.4 percent lower than the overall
average level on days with Form 10-K, Form 10-Q, and Form 4 filings, respectively.
The lower demand for analyst information following the release of Forms 10-Ks and
10-Qs appears to reflect sparse retail demand for financial statement information and SEC
filings. For example, the demand for analyst information on non-information event days is on
average over 6 times that of the Yahoo Finance SEC filings pages, and more than double that
of the Yahoo Finance financial statement pages, and on event days reaches up to 20 times that
of SEC filings pages and 4 times that of financial statement pages. These results provide
evidence suggesting that the demand for analyst information significantly trumps that of
financial reporting information and that the average retail investor appears to rely on analysts
to interpret financial reporting information rather than read the actual filings.
Second, while the overall demand for analyst estimates is 19.9 percent lower than the
demand for analyst ratings and target prices, whether investors specifically seek analysts
earnings and sales estimates rather than target prices and recommendations at reporting events
depends on whether the event pertains to earnings-related news. Searches for analyst estimates
are 73.8 percent higher than for analyst ratings and target prices on earnings announcement
days, and 30.8 percent higher on management guidance announcement days, while they are
lower for non-earnings-related events such as Form 8-K, Form 10-K, and Form 10-Q filings.
Beyer et al. (2010, p. 335) highlight that the research is uncertain as to why the practice of
prevalent because investors prefer analyst estimates to recommendations and target prices at
3
Third, firm visibility (e.g., news, extreme returns, initiations and drops of analyst
coverage) is also a significant determinant of investor demand for analyst information. The
demand for analyst information on Yahoo Finance increases between 9 and 15 percent during
days with analyst coverage changes. Media is a significant determinant of demand, supporting
the inference that investors seek analyst guidance to provide resolution to news or possible
uncertainty resulting in heightened visibility. Fourth, even though analyst report narratives are
not directly available on Yahoo Finance, celebrity analysts and attention-grabbing reports are
associated with higher investor demand for analyst information. Specifically, demand for
analyst information increases by 68.2, 39.4, and 8.8 percent on days when celebrity (i.e.,
Institutional Investor All-Star), bold, and more timely analysts, respectively, issue reports; and
investors appear to learn of such analyst reports through the media as suggested by the positive
and significant interactions between such reports and daily media coverage.
The distribution of investor demand is concentrated into relatively few days, as we find
at the firm level that 20 percent of annual analyst page views are concentrated into 5 percent of
days, and 50 percent of annual page views are concentrated into 20 percent of days.
Information event days alone account for a significant fraction of annual page views: if we
consider earnings announcement days and the day following, 7.9 percent of annual analyst
page views occur on 2.2 percent of days. For the average firm, 29.9 percent of firm-days have
media coverage and these days account for 55.3 percent of annual page views. Additionally,
analyst report issuance does not always correspond with abnormal investor demand. Analysts
do issue timely revisions around high-demand and predictable events such as earnings
announcements, but there are opportunities for analysts to revise guidance and estimates after
4
Although our findings provide initial inferences of when investors demand the types of
analyst information presented on Yahoo Finance, we are still limited in assessing how
investors use this information. Moreover, while our inferences reflect the search patterns of
millions of unique daily users on Yahoo Finance, these patterns are likely representative of the
U.S. retail investor population, not of professional investors. Despite these limitations, our
study provides a greater understanding of investor demand for analyst information and
provides novel evidence suggesting the value and importance of analyst information to Yahoo
Finance users.
The remainder of this paper is organized as follows. Section II presents the background
literature and hypotheses development, Section III presents our data and sample, Section IV
Despite research indirectly inferring investors demand for analyst information (e.g.,
Bhushan 1989; OBrien and Bhushan 1990; Barth et al. 2001; Frankel et al. 2006; Hugon and
Muslu 2010; Lehavy et al. 2011; Yezegel 2015), the conditions under which investors seek
analyst information and the type of information sought are still not well understood (Beyer et
al. 2010). Academic findings concerning the underlying value of sell-side analyst information
are mixed. While a large body of literature finds significant investment value from following
analysts ratings and forecasts (e.g., Barber and Loffler 1993; Womack 1996; Asquith,
Mikhail, and Au 2005; Frankel, et al. 2006; Barber, Lehavy, McNichols, and Trueman 2006;
Mikhail, Walther, and Willis 2007), another growing body of research questions the
contribution and the value of analysts (e.g., Dechow and Sloan 1997; Barber, Lehavy,
McNichols, and Trueman 2001; Kothari, Li, and Short 2009; Drake, Rees, and Swanson 2011).
5
The former studies generally find that it is profitable to follow analysts advice or that there is
significant information content in analyst reports; whereas the latter studies suggest that
recommendations are no longer profitable net of trading costs, that analyst reports appear stale
relative to the business press, and that analyst recommendations are contrarian indicators in the
intermediary (e.g., Bushee, Core, Guay, and Hamm 2010; Solomon and Soltes 2012) and even
questions whether analysts provide value over and above the financial press (e.g., Kothari et al.
2009). Furthermore, Altnkl, et al. (2013, p. 2550) provide evidence suggesting that
analysts piggyback their reports on public information from recent events and news about the
firm, while delivering little incremental information, meaning that analysts convert public
information into a forecast revision, which is not very informative beyond the information
itself. However, other research suggests that higher quality analysts are quoted in the financial
press (Bonner, Hugon, and Walther 2007; Rees, Sharp, and Twedt 2015), and that analysts are
actually issuing reports when there are greater opportunities to process information, detect
potential mispricings (Li, Ramesh, Shen, and Wu 2015), respond to investor demand (Yezegel
2015), and provide an important governance role (e.g., Moyer, Chatfield, and Sisneros 1989;
Dyck, Morse, and Zingales 2010; Chen, Harford, and Lin 2015).
The main purpose of the study is to increase our understanding of investor demand for
analysts main outputs, using predominantly retail investors daily page views of Yahoo
Finances firm-specific analyst information pages for stocks listed in the U.S. Prior research
argues that the main value proposition of analysts is that they process and interpret information
6
(e.g., Schipper 1991; Lang and Lundholm 1996). Analysts publicize their research to market
their brokerage services to customers and potential customers (e.g., OBrien and Bhushan
1990), and to increase investor recognition and ownership of promoted securities (Merton
1987; Barber and Odean 2008). Analysts also make use of the media to widely disseminate
their research to increase their prestige and compensation (e.g., Groysberg, Healy, and Maber
2011; Rees et al. 2015). Hence, it is important to understand the demand for analyst
information to provide insight into analysts decisions of when to update their reports, the
boldness or timeliness of their forecasts, and the type of information to present, such as
estimates versus ratings and target prices (see Beyer et al. 2010).
information should be most pronounced at the time of the public release of such information
(e.g., Ivkovi and Jegadeesh 2004). Furthermore, demand for analyst information may also
increase prior to predictable information events, as pre-event updates are informative (e.g.,
Kim and Verrecchia 1997; Ivkovi and Jegadeesh 2004) and as investors establish performance
benchmarks for expected information releases (e.g., Drake, Roulstone, and Thornock 2012,
2015). Given that different users of financial information have different abilities to process
information (e.g., Indjejikian 1991, Ball 1992), we expect, especially for the predominantly
retail user base of Yahoo Finance, that analysts are in a position to help these investors process
Beyer et al. (2010) show that among the set of disclosure events for which they test
relative importance, management guidance and earnings announcements are the most price-
relevant, while SEC filings are the least. Additionally, media coverage is more pronounced
7
surrounding earnings and guidance announcements rather than surrounding SEC filings, which
is important given that media coverage generates investor activity (e.g., Engelberg and Parsons
2011). 2 Hence, many SEC filings (e.g., 10-Ks, 10-Qs) may not capture the attention of retail
investors, resulting in limited demand for analyst support even though the information event
may be relatively complex. Our underlying null hypothesis is that investors may not exhibit
greater demand for analyst information at financial reporting events, especially demand not
explained by the overall level of firm-specific information consumption, given that analysts
may simply parrot the news or may not be more skilled at interpreting particular types of
information. If investors demand analyst information at the time of financial reporting events,
we predict cross sectional variation based on the expected information content of the event, for
example, that there is more demand for analyst information at the time of earnings
announcements and management guidance updates compared to more routine SEC filings.
recommendations and target prices, an area that Beyer et al. (2010) note is not well-understood.
Ivkovi and Jegadeesh (2004) separately examine the informativeness of forecast and
recommendation revisions surrounding earnings announcements, and find that both are
informative especially in the days leading up to earnings announcements. Their model does not
seek to predict the relative importance of analyst estimates versus recommendations, and so
our null hypothesis is that there is no difference in demand between the two types of
information surrounding financial reporting events. We predict that investors are more likely to
seek analysts estimates when the reporting event directly pertains to earnings news,
2
Due to limited research examining media coverage surrounding all of the main financial reporting events, in untabulated
analyses we examine the media coverage surrounding all major financial reporting events, finding that media coverage is most
pronounced for earnings announcements, followed by management guidance, Form 10-Ks, Form 10-Qs, Form 8-Ks, and Form
4s, respectively.
8
specifically at earnings announcements and management guidance updates, and that analyst
estimates are less important than recommendations when the reporting event does not
specifically relate to earnings news. Our second hypothesis, in alternative form, is:
H2: Demand for analyst estimates is more pronounced than for analyst opinions
surrounding financial reporting events with earnings news.
collection (e.g., Grevais, Kaniel, and Mingelgrin 2001; Grullon, Kanatas, and Weston 2004;
Barber and Odean 2008) by both potential buyers and sellers of a security. We use four proxies
for visibility that are used in prior studies. First, media coverage of a security raises its
visibility and often presents information that analysts interpret for investors (e.g., Tetlock
2007; Bushee et al. 2010; Engelberg and Parsons 2011; Rees et al. 2015). Consequently, media
coverage is likely to increase demand for analyst information, assuming that the media
coverage itself does not fully satisfy the informational demand of investors. We use changes in
analyst coverage as our second proxy for visibility, as analysts choose to cover securities
specifically for the purpose of increasing investor awareness (OBrien and Bhushan 1990).
Initiation and discontinuation of coverage is associated with future firm performance, and new
coverage tends to be more accurate (McNichols and OBrien 1997). Irani and Oesch (2013)
show that reductions in coverage reflect negatively on future financial reporting quality.
Similarly, retail investor awareness of a security can also be attributed to significant price
movements (e.g., Lee et al. 1991; Barber et al. 2009; Aboody et al. 2010), so we use abnormal
stock returns as our third proxy for visibility. Our fourth proxy for visibility is the change in
short interest. Short sellers garner investor attention as they promote their opinions of
overvaluation in the media (Ljungqvist and Qian 2016). Drake et al. (2011) show that analyst
9
recommendations are not credible when they contrast with the level of short interest, and as a
result, investors may be less likely to seek analyst information when short interest increases.
Moreover, the predominantly retail investor base of Yahoo Finance is likely to be short sale-
constrained, and such investors are unlikely to undertake research to short-sell a security
(Barber and Odean 2008). However, for the subset of investors who already own shares in
firms targeted by short sellers, there may be an increase in demand for analyst information to
While it may be intuitive that many of the foregoing proxies for visibility will be
associated with increases in investor page views for a security on Yahoo Finance, it is not
necessarily obvious that such investor attention will result in an increase in demand for analyst
information after controlling for the overall increase in information consumption. We predict
that the demand for analyst information increases in firm visibility, even after controlling for
Lastly, we predict that the demand for analyst information varies with analyst
characteristics. On the one hand, analyst characteristics may be unknown to the predominantly
retail investor base of Yahoo Finance, or such investors may not respond to situations in which
analyst recommendations are likely to be more informative. On the other hand, prior research
estimates and recommendations. The market response to analyst information is greater for
more-timely analysts (e.g., Trueman 1994; Cooper, Day, and Lewis 2001; Clement and Tse
2003; Gleason and Lee 2003; Mozes 2003; Loh and Stulz 2011), and timely analysts tend to
10
have more experience (Hong, Kubik, and Solomon 2000). Consequently, we predict that timely
Bold recommendations, which are those made further from the consensus, have also
been shown to be more price-relevant (e.g., Jegadeesh and Kim 2010). Hong et al. (2000) build
upon models of reputation and herding (e.g., Scharfstein and Stein 1990; Zwiebel 1995) to
show that career concerns influence forecast boldness, finding that analysts issuing bolder
forecasts are more likely to be terminated. If analysts increase their likelihood of termination
by issuing bold forecasts that provide poor advice, an offsetting benefit of bold forecasts could
be additional visibility for the analyst and career benefits when such calls are accurate. 3
Accordingly, we predict that bold analysts forecasts increase the demand for analyst
information. Finally, as celebrity analysts are by definition more prominent and have their
reports promoted in the media (e.g., Rees et al. 2015), and as their opinions and estimates are
price-relevant (e.g., Bonner et al. 2007; Loh and Stulz 2011), we predict that investor demand
increases when All-Star analysts issue reports. Our fourth hypothesis in alternative form is:
H4: Demand for analyst information is increasing in analyst boldness, timeliness, and
All-Star status.
We obtain proprietary web-traffic data from Yahoo! Inc. for all firm-specific page
views on Yahoo Finance (finance.yahoo.com) from July 1, 2014 to July 1, 2015. 4 Yahoo
3
For example, bold Morgan Stanley analyst Adam Jonas appeared in a prominent Wall Street Journal article upon revising an
extreme target price for Tesla Motors shares downward by $117 from $450 to $333, on a day when Tesla was trading at $286.
Teslas Biggest Bull Slashes His Price Target By More Than $100, by Kristen Scholer, The Wall Street Journal, February 1,
2016.
4
We thank StockTargetPrices.com for providing data utilized in a prior version of this study, and note that the results using
Stock Target Prices data are generally consistent with the results using Yahoo Finance data. The primary benefit to using
11
Finance offers general financial news and market information, and firm-specific information
accessed from each companys summary page. The page view data in this study is different
and more specific than prior studies of web behavior (e.g., Drake et al. 2012) that use web
search engine general-query indexes to infer demand for stock-related information. Our page
views all take place on a finance-specific web site, where users have identified the specific
stock ticker symbol and requested specific types of financial and market information. Yahoo
Finances firm-specific pages are accessed either by searching for a company by name or by
stock ticker symbol. Yahoo Finance records page views for the summary page and for each and
every sub-page the user requests. Our data do not include page views of U.S. stocks made on
non-U.S. versions of Yahoo Finance (e.g., finance.yahoo.co.uk for the United Kingdom), so the
vast majority of page views that we observe are for U.S. users. 5
to 23 different sub-pages, including those for pricing, news, company background, message
board discussion, financial statements, and analyst estimates and opinions, where data for the
sub-pages are available. For the purposes of this study, we are primarily interested in the sell-
side analyst information elements that Yahoo Finance reports for investors, though we are able
consumption of analyst information by measuring both analyst-related page views and non-
Yahoo Finance data is the significantly larger traffic volume and the ability to compare the demand for analyst information to
the demand for other types of financial information.
5
While a user from the United Kingdom could visit the American Yahoo Finance web site, creating a page view that we would
observe, Yahoo indicated in discussions with the authors that international users predominantly use their local version of the
Yahoo Finance web site, and hence, we generally do not observe their page views of U.S. listed stocks.
12
Yahoo Finance has two firm-specific pages displaying the main quantitative outputs
analysts provide other information, both quantitative and qualitative, in their detailed reports
and in conversations with their clients and the media. The first page, Analyst Opinion, provides
consensus recommendations and target prices (data provided by Thomson/First Call), as well
as analyst-specific ratings history (data provided by Briefings.com). The second page, Analyst
Estimates, provides consensus EPS, revenue, and long-term growth estimates (data provided
by Thomson/First Call). The analyst data elements presented on Yahoo Finance are updated
daily. Figures 2 and 3 provide examples of the respective firm-specific analyst pages. As with
most major financial providers, Yahoo Finance presents the foregoing analyst outputs on a
consensus basis and does not provide access to the full research reports. An exception to the
consensus presentation is made for recent analyst opinion upgrade and downgrade history,
which is given in detail. Analysts make extensive use of media to promote their research
(Bonner et al. 2007; Rees et al. 2015), and as a result we expect that prominent analysts and
attention-grabbing reports will generate investor demand for analyst information on Yahoo
Finance, enabling us to study the impact of celebrity, bold, and timely analysts, even if report
Our traffic data consists of daily page views for all U.S. publicly traded stocks,
sum, Analyst Total (AT_PAGEVIEW). These measures capture the level of demand for analyst
information, consistent with prior studies of consumer demand (e.g., Ludvigson 2004; Parker,
13
Souleles, Johnson, and McClelland 2013). To address the possibility that demand for analyst
control for the demand for other information using other page views (OT_PAGEVIEW, i.e.
non-analyst, firm-specific page views). Hence, we can illustrate both the baseline increase in
demand for analyst information as well as the increase in demand not explained by demand for
other information.
We obtain all financial, market, news information, and analyst characteristics from
FactSet. These data include: quarterly return on assets, leverage, biweekly short interest, daily
stock prices, market capitalization, book-to-market ratio, total returns, analyst reports, news
management earnings-per-share guidance, and analysts EPS forecasts from detailed broker
files. We obtain EDGAR filing dates from the SECs daily EDGAR index files. See the
Sample Selection
Our sample period begins July 1, 2014 and continues to July 1, 2015. We match Yahoo
Finance data with FactSet based on the ticker symbol. Our main analyses include weekend
days, with abnormal returns set to zero and financial variables set to the preceding Fridays
value. As a robustness test, we exclude weekend days from our analyses and our results are
unchanged. We match firms to their EDGAR filings using FactSets primary CIK for each
security. Our initial sample includes 1,843,176 firm-day observations for 5,036 unique firms.
We exclude firms with no analyst following during the entire sample period, as these firms
have no analyst pages on Yahoo Finance for which to record page views, reducing our sample
to 1,612,596 firm-days for 4,406 unique firms. After excluding observations with missing data
14
items, we obtain a final sample of 1,353,030 firm-days for 4,068 unique firms. The final
sample contains 213,551 EDGAR filings, 425,063 analyst report records, and 1,113,529 news
headlines.
IV. RESULTS
Descriptive Statistics
Table 1 provides descriptive statistics for the variables in our analyses. Page view
counts are measured by firm-day, and these values are normalized by the mean daily page view
counts for all U.S. listed firms provided by Yahoo Finance. The mean daily total page views is
therefore 1.0 for our sample period, prior to dropping observations that do not meet our data
approximately 1.9 percent of the average firm-day page view volume is for Analyst Total page
has a mean value of 0.008. The mean of AO_PAGEVIEW is statistically (p < 0.01) higher than
the mean of AE_PAGEVIEW, suggesting that investors have a general preference for target
prices and ratings over earnings and sales estimates, with the overall demand for Analyst
Estimates being 19.9 percent lower than for Analyst Opinions. We control for the overall
demand for firm-specific financial information by observing all other page views (non-analyst
page views) on the Yahoo Finance firm-specific pages (OT_PAGEVIEW), which has a mean
value of 1.098, exceeding the expected value of 0.981 (1 0.019), as the observations
eliminated due to missing data had fewer page views on average than the entire Yahoo Finance
15
U.S. stock universe. We use the natural logarithm of one plus the page view counts in our
regressions, and even after taking logarithms, the resulting page views are right-skewed.
The mean value of AT_PAGEVIEW is 0.019, comparable to the 75th percentile value of
0.015. The 95th percentile value is 0.073, and the maximum value is 7.668. In untabulated
analyses, we find that for the average firm, approximately 50 (20) percent of annual analyst
page views occur in 72 (18) days out of the sample period of 365 days. EARN_ANN has a mean
value of 0.011, equal to the expected value of 0.011 (4 / 365). We also separately examine
bundled earnings announcements, i.e., those with management guidance issued on the same
our sample, 15.2 percent of earnings announcements are bundled with guidance
(2,256/14,827). 8K has a mean value of 0.020, indicating that firms on average release 7.3 8-Ks
during the year (0.020 x 365), not counting those released surrounding earnings
announcements. FORM4 has a mean value of 0.039, indicating that firms on average report
0.010, indicating that one percent of firm-days have an increase in the number of analysts
following, and the mean value of DROP is 0.010, indicating that one percent of firm-days have
a decrease in the number of analysts following. The median number of news headlines is zero,
the 75th percentile is one headline (0.693 in log form), and the 95th percentile is four headlines
(1.609 in log form). We use three measures of analyst characteristics. The mean value of 0.007
16
for ALL_STAR indicates that All-Star analyst reports are observed on only 0.7 percent of firm-
days. 6 BOLD has a mean value of 0.007 and TIMELY has a mean value of 0.012.
Table 2 provides the Pearson correlations for the primary variables used in our
analyses. Analyst Total page views (AT_PAGEVIEW) are positively correlated with the
demand for financial information in general (OT_PAGEVIEW; corr. = 0.81), media coverage
(MEDIA; corr. = 0.28), and analyst report issuance (REPORT; corr. = 0.28). The correlation
between Analyst Opinion page views (AO_PAGEVIEW) and Analyst Estimates page views
(AO_PAGEVIEW) is 0.92, highlighting that investors usually concurrently search for both
analyst opinions (target prices and ratings) and estimates. Of the control variables, size has the
Table 3 presents a univariate analysis of the difference in mean values of Analyst Total
page views (AT_PAGEVIEW), conditioned on the information event, visibility, and analyst
characteristic variables. In all cases, the differences in means are statistically significant, with
t-statistics greater than 4.00 for all conditioning variables. The increase in Analyst Total page
views is most pronounced for earnings announcements with guidance (EARN_ANN_G), with
an increase of 386.4 percent compared to days when there are no earnings announcements, and
6
While it appears that we observe the majority of All-Star analyst reports, certain brokers restrict their identities in Factset,
thus we are unable to flag every All-Star analyst report. For example, with Apple Inc., we observe three out of four All-Star
analysts. We expect that the presence of some unidentified All-Star analyst reports will likely bias against rejecting the null
hypothesis.
17
Total page views of 245.3 percent. The effect for all earnings announcements (EARN_ANN),
with and without guidance changes, is an increase in Analyst Total page views of 270.4
percent. Increases in Analyst Total page views are also significant for Form 10-Ks (38.2
percent), 10-Qs (19.5 percent), 8-Ks (66.7 percent), Form 4s (58.3 percent), and management
guidance (311.6 percent). For the visibility measures, firm-days with changes in analyst
coverage have significant increases in analyst page views compared to other days. Increases in
analyst coverage (INITIATION) correspond to increases in Analyst Total page views of 170.8
percent, and days with coverage drops (DROP) have increases in Analyst Total page views of
119.4 percent. For firm-days with at least one media article (MEDIA_DUM equals one),
Analyst Total page views increase by 177.2 percent. The increase in Analyst Total page views
is 29.5 percent when there is an increase in short interest, and a 3.9 percent increase when
abnormal return is positive versus negative. When we consider analyst characteristics, there is
a 409.5 percent increase on days when All-Star analysts issue reports, a 284.5 percent increase
on days when bold analysts issue reports, and a 244.2 percent increase on days when timely
analysts issue reports, suggesting that investors are aware of analyst report issuance, with a
page views conditioned on information event days, and find that searches for analyst estimates
are 73.8 percent higher than for analyst ratings and target prices on earnings announcement
days, and 30.8 percent higher on management guidance announcement days, while they are
lower for non-earnings-related news events such as Form 8-Ks (27.6 percent lower), Form 10-
18
Analyst Page Views Surrounding Information Events
Figure 4 plots the percent change in Yahoo Finance page views relative to earnings
announcement events, as well as for the other information events we study. Analyst Estimates
page views are shown in Panel A, Analyst Opinion page views are shown in Panel B, and
Other Page Views are shown in Panel C. These plots illustrate an increase in demand for
analyst information on the event day (day 0). For earnings announcements, demand noticeably
increases the day prior to the event day, increasing significantly on the event day, and persists
for several days following the event day. Increases in demand are most pronounced for
updatesand are less pronounced for other disclosures. The increase in demand for Analyst
Estimates pages is more pronounced than for Analyst Opinion and Other pages (all firm-
specific page views less analyst page views) surrounding these information events.
analyst information, and provides evidence relating to H1, which proposes that investor
demand for analyst information is more pronounced surrounding financial reporting events.
We regress daily Analyst Total page views (AT_PAGEVIEW) on indicator variables for the
announcements without guidance (EARN_ANN_NG), Form 8-Ks, Form 10-Ks, Form 10-Qs,
Form 4s, and changes in management guidance (CHG_GUIDE). In Columns (1) to (3), we
estimate the increase in Analyst Total page views without including a control for other
information page views (OT_PAGEVIEW). This provides an estimate of the overall increase in
19
demand. After controlling for the demand for other information by including OT_PAGEVIEW
as an independent variable, we are able to examine the increase in Analyst Total page views
that are orthogonal to the demand for other information. 7 In all specifications, we control for
the normal level of firm-specific demand for analyst information by including the average level
of firm-Analyst Total page views over the trailing 30 days (AT_PAST30). 8 We also include
controls for MCAP, ROA, LEV, BTM, and PAST_RET_VOL. We cluster standard errors by firm
In Column (1), we consider independent variables for the information event day only
(time subscript t), and the coefficient on EARN_ANN is 0.0260 and significant (p < 0.01),
indicating that there is an average increase of 152.9 percent (0.0260 / 0.017) over the mean
0.0010 and significant (p < 0.01), indicating an increase in Analyst Total page views of 5.9
percent. The coefficient on 10K is -0.0018 (p < 0.01), and the coefficient on 10Q is -0.0016 (p
< 0.01) indicating a decrease in Analyst Total page views of 10.6 and 9.4 percent at Form 10-K
and Form 10-Q filings, respectively, when those filings are not accompanied by an earnings
announcement, controlling for firm characteristics. The coefficient on FORM4 is -0.0004 (p <
0.05), implying a decrease in Analyst Total page views of 2.4 percent. The coefficient on
increase in Analyst Total page views of 181.8 percent on non-earnings announcement days
when management issues updated guidance. Other page views (OT_PAGEVIEW, coeff. =
7
Inferences are more pronounced when we do not control for other OT_PAGEVIEW, consistent with overall demand for
financial information driving some of the demand for analyst information.
8
For July 2014, the normal level of Analyst Total page views is calculated using only the available days of data rather than 30
days given the web-traffic data begins on July 1, 2014. For example, the average level of Analyst Total page views on July 16,
2014, would be computed using only 15 days. We find all the main inferences are robust to dropping observations from July
2014 from the sample.
20
0.0347, p < 0.01), the average level of Analyst Total page views for the past 30-days
(AT_PAST30, coeff. = 0.6025, p < 0.01), and the past returns volatility (PAST_RET_VOL,
coeff. = -0.3291, p < 0.01) are the most significant control variables affecting Analyst Total
page views. These findings indicate that the concurrent demand for other information and
recent demand for analyst information are positive predictors of current demand for analyst
information, while stock return volatility is negatively associated with demand for analyst
information. 9 The reduction in page views for Forms 10-K and 10-Q may be a result of recent
(captured by AT_PAST30), and may also reflect that investors are unaware of the timing of
SEC filings. Furthermore, analysts may be slow to update their outputs after these filings, and
therefore investors benefit little from seeking analyst information on the day of the filing itself.
In Column (2) we examine the demand for analyst information in the three days
surrounding the information events. For the information event indicator variables, the subscript
t-1 indicates the demand the day prior to the information event and the subscript t+1 indicates
the demand the day following the information event. Increases in Analyst Total page views
leading up to the event and after the event vary depending on the event variable. For earnings
announcements, the demand is elevated at the p < 0.01 level the day prior to the earnings
announcement (EARN_ANNt-1) with a coefficient of 0.0078 and the day after (EARN_ANNt+1)
with a coefficient of 0.0141. This corresponds to an increase in Analyst Total page views of
44.9 percent over normal levels on the day prior to the earnings announcement, and an increase
in demand of 82.9 percent on the day after the earnings announcement. Changes in guidance
which are not released in conjunction with earnings announcements (CHG_GUIDE) show little
9
The coefficient on PAST_RET_VOL is positive and significant (p < 0.01) when we do not control for OT_PAGEVIEW.
21
increase in Analyst Total page views in the day prior to the guidance change (coeff. = 0.0023, p
< 0.10). 10 The increase in Analyst Total page views the day following the guidance change is
167.6 percent greater than normal levels (coeff. = 0.0285, p < 0.01), higher than the demand the
day after earnings announcements. This suggests that non-earnings announcement guidance
changes, which are often associated with profit warnings and material acquisitions or
divestitures, are events for which investors exhibit significant demand for analyst assistance to
interpret. The remaining information event variables do not have economically significant
In Column (3), we examine Analyst Total page views separately for earnings
that earnings announcements bundled with earnings guidance contain more information that
may be more complicated for investors to process, the coefficient on EARN_ANN_G (coeff. =
0.0353, p < 0.01) is significantly greater (p < 0.01) than the coefficient on EARN_ANN_NG
(coeff. = 0.0244, p < 0.01). These coefficients correspond to an increase in Analyst Total page
views of 207.7 percent for bundled earnings announcements, but 143.5 percent for unbundled
coefficients on the information events in Column (3) are equal. We can reject the pairwise
equality of coefficients at the p < 0.01 level for all cases except for the coefficients between
10
Some of the guidance events that were not issued concurrently with the earnings announcement were found to be recorded
by FactSet on the trading date rather than the announcement date. We manually checked the accuracy of all such dates,
identified approximately 18 percent of observations that were off by one day, and corrected these dates for our analyses.
22
Table 5 provides evidence supporting H2, which predicts that the demand for analyst
estimates is more pronounced than for analyst opinions surrounding earnings-news financial
reporting events. We examine investor demand separately for analyst estimates via the Analyst
Estimates page (AE_PAGEVIEW) in Column (1), analyst opinions via the Analyst Opinion
page (AO_PAGEVIEW) in Column (2), and the difference between Analyst Estimates and
Analyst Opinion page views (AE AO_PAGEVIEW) in Column (3), in order to understand
how demand for the type of analyst output varies based on the information event.
On non-information event days, the demand for analyst opinion information is typically
greater, as the intercept term on AE AO_PAGEVIEW is negative and significant (p < 0.01).
greater than the demand for analyst opinion information. Comparing the coefficients on
EARN_ANNt between Columns (1) and (2), we see that the increase in demand for the Analyst
Estimates page is 311.3 percent (0.0249 / 0.008) on the event day, whereas the increase for the
announcements. Comparing the coefficients on CHG_GUIDEt between Columns (1) and (2),
we see that the increase in demand for the Analyst Estimates page is 306.2 percent on the event
day, whereas the increase for the Analyst Opinion page is only 103.0 percent. The difference
between AE_PAGEVIEW and AO_PAGEVIEW for changes in guidance in Column (3) gives a
coefficient on CHG_GUIDE of 0.0134 (p < 0.01), indicating a significantly greater demand for
23
analyst estimate information at guidance updates. Coefficients on the other information events
are all negative and significant, although their magnitudes are much lower than for earnings
announcements and changes in management guidance, indicating that demand for analyst
control variables in Column (3) indicate that the demand for estimates over target prices and
recommendations is increasing in firm size (MCAP, coeff. = 0.0002, p < 0.01), profitability
(ROA, coeff. = 0.0002, p < 0.05), growth (BTM, coeff. = -0.0001, p < 0.05), and past return
volatility (PAST_RET_VOL, coeff. = 0.0405, p < 0.01), and decreasing in leverage (LEV, coeff.
= -0.0003, p < 0.01), consistent with the notion that equity valuations are more sensitive to
In Table 6, Panel A, we study the impact of firm visibility and analyst characteristics on
the demand for analyst information, to test H3, which predicts that demand is increasing in
firm visibility, and H4, which predicts that demand is increasing in analyst boldness,
timeliness, and All-Star status. Table 6 has three columns which study the effect of firm
visibility and analyst ability measures on the dependent variables: Analyst Total page views
(2), and Analyst Opinion page views (AO_PAGEVIEW) in Column (3), respectively. To proxy
for visibility, we utilize the initiation (INITIATION) and dropping (DROP) of coverage by
analysts, the logarithm of one plus the number of media headlines (MEDIA), changes in short
interest (CHG_SHORT), and positive and negative abnormal returns (ABRET_P, ABRET_N).
To proxy for analyst ability, we use an indicator if an Institutional Investor All-Star ranked
24
analyst issues a report (ALL_STAR), an indicator if a bold analyst issues a report (BOLD), and
an indicator if a timely analyst issues a report (TIMELY). We use the same set of control
variables as in Tables 4 and 5 with the addition of the number of reports issued (REPORT).
The overall results from Table 6 indicate that the visibility and analyst ability measures
have a similar direction for all three dependent variables. We first consider the analyst
associated with both initiations (INITIATION, coeff. = 0.0016, p < 0.01) and coverage
terminations (DROP, coeff. = 0.0025, p < 0.01), and the effect is significantly higher for
coverage terminations versus initiations (F-value = 12.64, p < 0.01). These results indicate an
increase in Analyst Total page views of 9.4 percent at coverage initiations, and 14.7 percent at
coverage drops. Higher levels of media reports (MEDIA, coeff. = 0.0016, p < 0.01), and both
abnormal positive returns (AB_RET_P, coeff. = 0.0943, p < 0.01) and abnormal negative
returns (AB_RET_N, coeff. = -0.1005, p < 0.01) are also associated with higher
(CHG_SHORT, coeff. = -0.0243, p > 0.10), which may be due to the bi-weekly nature of short
interest reporting, or that investors do not seek analyst estimates and opinions to help explain
changes in short interest. The results for Analyst Estimates page views in Column (2) and
Analyst Opinion page views in Column (3) are similar, with the exception of changes in short
interest in Column (2), (CHG_SHORT, coeff. = -0.0214, p < 0.05), which indicates that
investor demand for analyst estimate information is greater on days when short interest was
analyst report being issued on a day (ALL_STAR, coeff. = 0.0116, p < 0.01) is associated with
25
an increase in demand for analyst information of 68.2 percent (0.0116 / 0.017), after
controlling for other information and visibility variables. The increase in demand for bold
analysts is 39.4 percent (0.0067 / 0.017). The increase in demand for timely analysts is 8.8
percent (0.0015 / 0.017). The results for Analyst Estimates page views in Column (2) and
Analyst Opinion page views in Column (3) produce similar inferences, except that the demand
for timely analyst information is limited to analyst estimate information in Column (2) (coeff. =
0.0020, p < 0.01), as the coefficient on TIMELY in Column (3) for Analyst Opinion page views
is insignificant.
Since the media provides a channel for disseminating analyst information, we interact
our proxies for analyst ability (ALL_STAR, BOLD, and TIMELY) with the media attention
variable (MEDIA) in Table 6, Panel B. Consistent across all three dependent variables, the
increase in investor demand for analyst information occurs when All-Star, bold, and timely
analyst reports are accompanied by higher levels of media reports (ALL_STAR * MEDIA, coeff.
= 0.0166, p < 0.01; BOLD * MEDIA, coeff. = 0.0086, p < 0.01; TIMELY * MEDIA, coeff. =
0.0018, p < 0.01). The results for Analyst Estimates page views in Column (2) and Analyst
Opinion page views in Column (3) produce similar inferences. These results support the role of
media in disseminating analyst information to the Yahoo Finance audience and increasing
investor demand, primarily for All-Star analysts, and to a lesser extent for bold and timely
analysts. Overall, the findings in Table 6 provide support for H3 and H4 that demand for
26
Demand for Analyst Information versus Demand for Other Financial Information
In Table 7, we examine the relative demand for analyst information compared to other
statistics for the mean number of page views of the major categories of Yahoo Finance firm-
specific pages, for all days and on information event days. The Summary page is the first page
that a Yahoo Finance user sees after searching for a firm, and therefore represents the greatest
number of page views on average (0.802). Moreover, it is the page that investors refresh
throughout the day to observe price changes. The second most popular firm-specific page is the
Yahoo Finance message board (mean page views = 0.197). This finding is consistent with prior
literature highlighting the popularity of stock message boards among retail investors (e.g.,
Wysocki 1999; Tumarkin and Whitelaw 2001; Antweiler and Frank 2004). Analyst
information is the third most popular type of information (mean page views = 0.019),
compared to other types of financial information on information event days to understand the
extent to which Yahoo Finance users seek analyst information more than primary sources of
6.3 times greater than the demand for SEC filings information, and increasing to 17.4 times
greater on earnings announcement days (EARN_ANN), 19.9 times greater on bundled earnings
announcement and management guidance days (EARN_ANN_G), and 19.6 times greater on
management guidance days (CHG_GUIDE). These results indicate that Yahoo Finance users
are significantly more likely to investigate analyst information surrounding important financial
reporting events than the underlying filings. We find similar but less pronounced relative
27
demand for analyst information compared to financial statement information and news
information. It is important to note that the demand for news information could be understated
given that some news information is also reported on the Summary page. In untabulated tests,
we also test whether the demand for analyst information is significantly greater than the
demand for SEC filings, financial statement information, and news information. The mean
ratios of total analyst page views to page views of these other types of information are
significantly (p < 0.01) greater than one. Overall, these findings suggest a fairly sparse retail
demand for financial statement information and SEC filings, especially when compared to the
In Table 8, we show that analyst information is still relevant to investors, even when
new analyst reports are not being issued. This finding may indicate three non-mutually-
exclusive situations. First, investors may be seeking analyst information to interpret an event
and are unsatisfied as no updated information is available. Second, investors may find utility in
comparing the event-news to analysts prior expectations, even if there are no updates. Third,
analysts may have reviewed the event and determined that no update to their prior analysis is
required, hence the non-issuance of a report may also be informative to investors. To examine
this question, in Panel A we tabulate the number of firm-day observations with a high level of
page views (two times the normal level of Analyst Total page views for the firm over the
trailing 30 days, which represents the top 11.0 percent of firm-days by Analyst Total page view
traffic), conditioned on no analyst reports being issued on the high page view day (day t) or the
following day (day t+1). Of 149,012 firm-days with high page views on day t, 91,996 (61.7
28
percent) have no analyst reports issued on day t or t+1, indicating that in most instances of
high investor demand for analyst information, there are no current or immediately forthcoming
In Panel B, we examine whether these high-demand days are associated with the
information events included in our study. We find that analyst reports are generated much more
frequently for certain events. Analysts provide timely reports for 95.4 percent of high-traffic
earnings announcements days (541 no-report-days versus 11,213 report-days), 94.0 percent of
high-traffic guidance change days. On the other hand, analysts only provide timely reports for
29.6 percent of high-traffic non-earnings announcement 10-Q days, 31.9 percent of high-traffic
Form 4 days, 39.4 percent of high-traffic non-earnings announcement 10-K days, 51.3 percent
of high-traffic 8-K days, and 58.8 percent of high-traffic media days. These results indicate that
while analysts seem to provide reports for predictable value-relevant events such as earnings
announcements and guidance changes, they are not providing reports for some unpredictable
yet value-relevant events such as media coverage, Form 4, and Form 8-K filings. Analysts may
require time to process such events (e.g., Li et al. 2015; Bradshaw, Wang, and Zhou 2015), and
hence do not always satisfy investor demand. In terms of sheer number of event days, the
largest opportunity for analysts to meet unmet demand for analyst information is after media
coverage. Specifically, we find 23,917 firm-days in our sample with media coverage and high
V. CONCLUSION
This study examines Yahoo Finance page views of analyst outputs to directly examine
investor demand for analyst estimates, ratings, and target prices. Our empirical evidence
indicates that in terms of financial reporting events, demand is most pronounced at earnings
29
announcements and management guidance updates, and less pronounced on days with Form
10-K and 10-Q filings that do not have accompanying earnings announcements in a
surrounding three-day window. Demand for analyst information is increasing in firm visibility
when All-Star, bold, and timely analysts issue reports. While investors display a preference for
target prices and ratings over earnings and sales estimates, this preference is reversed on days
with earnings-related news. Moreover, the demand for analyst information is substantially
higher than for other financial reporting information, indicating that the average Yahoo
Finance user relies on analysts to interpret financial reporting information. Overall, the
findings provide novel evidence indicating the value and importance of analyst information to
30
REFERENCES
Aboody, D., R. Lehavy, and B. Trueman. 2010. Limited attention and the earnings announcement returns
of past stock market winners. Review of Accounting Studies 15 (2): 317-344.
Altnkl, O., V. Balashov, and R. Hansen. 2013. Are analysts forecasts informative to the general
public? Management Science 59 (11): 2550-2565.
Antweiler, W., and M. Frank. 2004. Is all that talk just noise? The information content of internet stock
message boards? Journal of Finance 59 (3): 1259-1294.
Asquith, P., M. Mikhail, and A. Au. 2005. Information content of equity analyst reports. Journal of
Financial Economics 75 (2): 245-282.
Barber, B., and D. Loffler. 1993. The Dartboard column: Secondhand information and price pressure.
Journal of Financial and Quantitative Analysis 28 (2): 273-284.
, R. Lehavy, M. McNichols, and B. Trueman. 2001. Can investors profit from the prophets?
Security analyst recommendations and stock returns. Journal of Finance 56 (2): 531-563.
, , , and . 2006. Buys, holds, and sells: The distribution of investment banks'
stock ratings and implications for the profitability of analysts' recommendations. Journal of
Accounting & Economics 41 (1-2): 87-117.
, and T. Odean. 2008. All that glitters: The effect of attention and news on the buying behavior of
individual and institutional investors. Review of Financial Studies 21 (2): 785-818.
, , and Zhu. 2009. Do retail trades move markets? Review of Financial Studies 22 (1): 151-
186.
Barth, M., R. Kasznik, and M. McNichols. 2001. Analyst coverage and intangible assets. Journal of
Accounting Research 39 (1): 1-34.
Beyer, A., D. Cohen, T. Lys, and B. Walther. 2010. The financial reporting environment: Review of the
recent literature. Journal of Accounting & Economics 50 (2-3): 296-343.
Bhushan, R. 1989. Firm characteristics and analyst following. Journal of Accounting & Economics 11 (2-
3): 255-274.
Bonner, S., A. Hugon, and B. Walther. 2007. Investor reaction to celebrity analysts: the case of earnings
forecast revisions. Journal of Accounting Research 45 (3): 481-513.
Bradshaw, M., X., Wang, and D. Zhou. 2015. Analysts assimilation of soft information in the financial
press. Working paper, Boston College.
Bushee, B., J. Core, W. Guay, and S. Hamm. 2010. The role of the business press as an information
intermediary. Journal of Accounting Research 48 (1): 1-19.
Chen, T., J. Harford, and C. Lin. 2015. Do analysts matter for governance? Evidence from natural
experiments. Journal of Financial Economics 115 (2): 383-410.
31
Clement, M., and S. Tse. 2003. Do investors respond to analysts forecast revisions as if forecast accuracy
is all that matters? The Accounting Review 78 (1): 227-249.
Cooper, R., T. Day, and C. Lewis. 2001. Following the leader: a study of individual analysts earnings
forecasts. Journal of Financial Economics 61 (3): 383-416.
Dechow, P., and R. Sloan. 1997. Returns to contrarian investment strategies: Tests of naive expectations
hypotheses. Journal of Financial Economics 43 (1): 3-27.
Drake, M., L. Rees, and E. Swanson. 2011. Should investors follow the prophets or the bears? Evidence
on the use of public information by analysts and short sellers. The Accounting Review 86 (1): 101-
130.
, D. Roulstone, and J. Thornock. 2012. Investor information demand: Evidence from Google
searches around earnings announcements. Journal of Accounting Research 50 (4): 1001-1040.
Dyck, A., A. Morse, and L. Zingales. 2010. Who blows the whistle on corporate fraud? Journal of
Finance 65 (6): 2213-2253.
Engelberg, J., and C. Parsons. 2011. The causal impact of media in financial markets. Journal of Finance
66 (1): 67-97.
Frankel, R., S.P. Kothari, and J. Weber. 2006. Determinants of the informativeness of analyst research.
Journal of Accounting & Economics 41 (1-2): 29-54.
Gleason, C., and C. Lee. 2003. Analyst forecast revisions and market price discovery. The Accounting
Review 78 (1): 193-225.
Grevais, S., R. Kaniel, and D. Mingelgrin. 2001. The high-volume return premium. Journal of Finance 56
(3): 877-919.
Groysberg, P., P. Healy, and D. Maber. 2011. What drive sell-side analyst compensation at high-status
investment banks. Journal of Accounting Research 49 (4) 969-1000.
Grullon, G., G. Kanatas, and J. Weston. 2004. Advertising, breadth of ownership, and liquidity. Review of
Financial Studies 17 (2): 439-461.
Hong, H., J. Kubik, and A. Solomon. 2000. Security analysts' career concerns and herding of earnings
forecasts. RAND Journal of Economics 31 (1): 121-144.
Huang, A., R. Lehavy, A. Zang, and R. Zheng. 2015. Analyst information discovery and interpretation
roles: A topic modeling approach. Working paper, HKUST.
Hugon, A., and V. Muslu. 2010. Market demand for conservative analysts. Journal of Accounting &
Economics 50 (1): 42-57.
Indjejikian, R. 1991. The impact of costly information interpretation on firm disclosure decisions. Journal
of Accounting Research 29 (2): 277-301.
32
Irani, R., and D. Oesch. 2013. Montioring and corporate disclosure: Evidence from a natural experiment.
Journal of Financial Economics 109 (2): 398-418.
Ivkovi, Z., and N. Jegadeesh. 2004. The timing and value of forecast and recommendation revisions.
Journal of Financial Economics 73 (3): 433-463.
Jegadeesh, N., and W. Kim. 2010. Do analysts herd? An analysis of recommendations and market
reactions. Review of Financial Studies 23 (2): 901-937.
Kim, O., and R. Verrecchia. 1997. Pre-announcement and event-period private information. Journal of
Accounting and Economics 24 (3): 395-419.
Kothari, S.P., X. Li, and J. Short. 2009. The effect of disclosures by management, analysts, and business
press on cost of capital, return volatility, and analyst forecasts: A study using content analysis.
The Accounting Review 84 (5): 1639-1670.
Lang, M., and R. Lundholm. 1996. Corporate disclosure policy and analyst behavior. The Accounting
Review 71 (4): 467-492.
Lee, C., A. Shleifer, and R. Thaler. 1991. Investor sentiment and the closed-end fund puzzle. Journal of
Finance 46 (1): 75-109.
Lehavy, R., F. Li, and K. Merkley. 2011. The effect of annual report readability on analyst following and
the properties of their earnings forecasts. The Accounting Review 86 (3): 1087-1115.
Li, E., K. Ramesh, M. Shen, and J. Wu. 2015. Do analyst stock recommendations piggyback on recent
corporate news? An analysis of regular-hour and after-hours revisions. Journal of Accounting
Research 53 (4): 821-861.
Ljungqvist, A., and W. Qian. 2016. How constraining are limits to arbitrage? Working paper, New York
University.
Loh, R., and R. Stulz. 2011. When are analyst recommendation changes influential? Review of Financial
Studies 24 (2): 593-627.
Ludvigson, S. 2004. Consumer confidence and consumer spending. Journal of Economic Perspectives 18
(2): 29-50.
McNichols, M., and P. OBrien. 1997. Self-selection and analyst coverage. Journal of Accounting
Research 35 (Supplement): 167-199.
Merton, R. 1987. A simple model of capital market equilibrium with incomplete information. Journal of
Finance 42 (3): 483-510.
Mikhail, B., B. Walther, and R. Willis. 2007. When security analysts talk, who listens? The Accounting
Review 82 (5): 1227-1253.
Moyer, R.C., R. Chatfield, and P. Sisneros. 1989. Security analyst monitoring activity: Agency costs and
information demands. Journal of Financial and Quantitative Analysis 24 (4): 503-512.
33
Mozes, H. 2003. Accuracy, usefulness and the evaluation of analysts forecasts. International Journal of
Forecasting 19 (3): 417-434.
OBrien, P., and R. Bhushan. 1990. Analyst following and institutional ownership. Journal of Accounting
Research 28 (Supplement): 55-76.
Parker, J., N. Souleles, D. Johnson, and R. McClelland. 2013. Consumer spending and the economic
stimulus payments of 2008. American Economic Review 103 (6): 2530-2553.
Rees, L., N. Sharp, and B. Twedt. 2015. Whos heard on the Street? Determinants and consequences of
financial analyst coverage in the business press. Review of Accounting Studies 20 (1): 173-209.
Scharfstein, D., and J. Stein. 1990. Herd behavior and investment. American Economic Review 80 (3):
465-479.
Shiller, R. 2015. Irrational exuberance. 3rd Ed. Princeton, N.J.: Princeton University Press.
Solomon, D., and E. Soltes. 2012. Managerial control of business press coverage. Working paper,
University of Southern California.
Tetlock, P. 2007. Giving content to investor sentiment: The role of media in the stock market. Journal of
Finance 62 (3): 1139-1168.
Trueman, B. 1994. Analyst forecasts and herding behavior. Review of Financial Studies 7 (1): 97-124.
Turmarkin, R., and R. Whitelaw. 2001. News or noise? Internet postings and stock prices. Financial
Analysts Journal 57 (3): 41-51.
Womack, K. 1996. Do brokerage analysts recommendation have investment value? Journal of Finance
51 (1): 137-167.
Wysocki, P. 1999. Cheap talk on the web: The determinants of postings on stock message boards.
Working paper, University of Michigan.
Yezegel, A. 2015. Why do analysts revise their stock recommendations after earnings announcements?
Journal of Accounting and Economics 59 (2-3): 163-181.
Zwiebel, J. 1995. Corporate conservatism and relative compensation. Journal of Political Economy 103
(1): 1-25.
34
APPENDIX
Variable Definitions
35
Variable Definition Data Source
AO_PAST30t The natural log of one plus firm i's total number of Yahoo Finance
normalized trailing 30-day average Analyst Opinion
page views from Yahoo Finance on day t-1. Page views
are normalized by the mean daily page view counts for
all U.S. listed firms provided by Yahoo Finance during
our sample period. Only the available days of page
views are used during July 2014 rather than 30 days as
the web-traffic data starts on July 1st 2014;
_ The natural log of one plus firm i's total number of Yahoo Finance
normalized Analyst Estimates and Analyst Opinion page
views from Yahoo Finance on day t. Page views are
normalized by the mean daily page view counts for all
U.S. listed firms provided by Yahoo Finance during our
sample period;
AT_PAST30t The natural log of one plus firm i's total number of Yahoo Finance
normalized trailing 30-day average Analyst Estimates
and Analyst Opinion page views from Yahoo Finance
on day t-1. Page views are normalized by the mean daily
page view counts for all U.S. listed firms provided by
Yahoo Finance during our sample period. Only the
available days of page views are used during July 2014
rather than 30 days as the web-traffic data starts on July
1st 2014;
"1" if a bold analyst issues an EPS forecast for firm i on FactSet:
day t, "0" otherwise. An analyst is considered a bold FE_BROKER_ESTIMATE
analyst if s/he is ranked in the top 25% of all analysts in
terms of the average boldness score using the data
between 2012/1/1 and 2014/6/30. Boldness score is
calculated following Hong, Kubik, and Solomon (2000).
Specifically, for each firm is fiscal period k earnings
forecast between 2012/1/1 and 2014/6/30, we calculate a
measure of consensus forecast
,,, for each analyst
a, where -a is the set of all analysts other than analyst a
who produce an EPS forecast for firm i in on day t.
Analyst as forecast boldness for firm is fiscal period k
earnings forecast on day t is the absolute value of the
difference between ,,, and
,,, . A ranking
,,, =1, 2. N is then assigned to each analyst-
forecast and the boldest analyst receives the first rank,
and the least bold analyst receives the highest rank. We
then calculate a boldness score for analyst as forecast
for firm is fiscal period k on day t, ,,, = 100
,,, 1
100 . Lastly, the boldness
,, 1
score per analyst-forecast is averaged among all
forecasts per analyst to calculate the average boldness
score for each analyst. Higher values correspond to
more bold analysts.
The ratio of book value of equity to market FactSet: FF_PBK
capitalization for firm i on day t;
(continued)
36
Variable Definition Data Source
_ "1" if there is a change in management guidance for EPS FactSet: FE_GUIDANCE
or revenue, annual or quarterly, for firm i between day t FF_EPS_RPT_DATE
and day t-1, "0" otherwise and if the firm has an
earnings announcement on days t-3 to t+3;
_+ "1" if CHG_GUIDE equals "1" on day t+j, j= -1,1, and FactSet: FE_GUIDANCE
zero otherwise; FF_EPS_RPT_DATE
_ The short interest scaled by shares outstanding on day t FactSet:
minus the short interest scaled by shares outstanding for FSI_SI_ANY_EXCHG
firm i on day t-1; FF_COM_SHS_OUT
"1" if there is a decrease in the number of analysts FactSet:
following firm i between day t and day t-1, "0" FE_ESTIMATE:NEST
otherwise;
_ "1" if firm i announces its earnings on day t, "0" FactSet: FF_EPS_RPT_DATE
otherwise;
_+ "1" if EARN_ANN equals "1" on day t+j, j= -1,1, and FactSet: FF_EPS_RPT_DATE
zero otherwise;
__ "1" if firm i announces its earnings and has a change in FactSet: FE_GUIDANCE
management guidance on day t, "0" otherwise; FF_EPS_RPT_DATE
__ "1" if firm i announces its earnings and does not have a FactSet: FE_GUIDANCE
change in management guidance on day t, "0" otherwise; FF_EPS_RPT_DATE
4 "1" if firm i files one or more Form 4 on day t, "0" EDGAR Daily Index File
otherwise and if the firm has an earnings announcement
on days t-3 to t+3;
4+ "1" if FORM4 equals "1" on day t+j, j= -1,1, and zero EDGAR Daily Index File
otherwise;
"1" if there is an increase in the number of analysts FactSet:
following firm i between day t and day t-1, "0" FE_ESTIMATE:NEST
otherwise;
The ratio of debt to total assets for firm i on day t; FactSet: FF_DEBT_ASSETS
The natural log of firm i's market capitalization on day t; FactSet:
P_MARKET_VAL_SEC
The natural log of one plus the total number of media FactSet:
articles that mention firm i on day t; NEWS_HEADLINE_COUNT
_ "1" if there is at least one media article that mentions FactSet:
firm i on day t, "0" otherwise; NEWS_HEADLINE_COUNT
_ The natural log of one plus firm i's total number of Yahoo Finance
normalized other page views from Yahoo Finance on
day t. Other page views is the difference between total
page views and total Analyst Estimates/Opinion page
views. Page views are normalized by the mean daily
page view counts for all U.S. listed firms provided by
Yahoo Finance during our sample period;
__ The standard deviation of firm i's daily returns over the FactSet: P_TOTAL_RETURN
past 6 months ending on day t-1;
The natural log of one plus the total number of analyst FactSet:
reports for firm i on day t; FE_BROKER_ESTIMATE
(continued)
37
Variable Definition Data Source
The ratio of net income to total assets for firm i on day t; FactSet: FF_ROA
and,
38
FIGURE 1
Example of a Yahoo Finance Firm-Specific Homepage
This figure illustrates an example of a typical Yahoo Finance (finance.yahoo.com) firm-specific homepage,
providing summary stock market and headline information, as well as access to another 23 firm-specific pages
containing financial, market, and analyst data.
39
FIGURE 2
Example of a Yahoo Finance Firm-Specific Analyst Opinion Web Page
This figure illustrates an example of a typical Yahoo Finance Analyst Opinion firm-specific web page, which
provides recommendation and target price summary statistics, as well as details of recent broker-specific ratings
changes.
40
FIGURE 3
Example of a Yahoo Finance Firm-Specific Analyst Estimates Web Page
This figure illustrates an example of a typical Yahoo Finance Analyst Estimates firm-specific web page, which
provides consensus estimates of earnings, revenue, EPS, and growth.
41
FIGURE 4
Panel A: Analyst Estimates Page Views Surrounding Information Events
This figure plots investor demand for analyst information and other information relative to key information events:
earnings announcements, management guidance changes, Forms 8-K, 4, 10-K, and 10-Q. Management guidance
changes, Forms 8-K, 4, 10-K, and 10-Q are those that do not coincide with earnings announcements. Panel A
illustrates mean abnormal (percent change from sample mean) Analyst Estimates page views (consensus EPS,
revenue and long term growth) relative to these events, and we note that the increase in page views is highest for
guidance changes and earnings announcements. The increase in demand is visible the day prior to the event for
earnings announcements, and excess demand continues for several days after the event in the case of guidance
changes and earnings announcements. Panel B illustrates mean Analyst Opinion page views (consensus ratings,
target prices, and detailed recent rating changes), and we note that the increase in page views surrounding
information events is less pronounced compared to the increase for Analyst Estimates. Panel C illustrates the
42
increase in demand for other page views (all page views less Analyst Estimates and Analyst Opinions). Again the
demand for other information is highest for earnings announcements and guidance changes, though the increase is
not as pronounced as for Analyst Estimate page views.
43
TABLE 1
Descriptive Statistics
Std.
N Mean Min 5% 25% 50% 75% 95% Max
Dev.
Page Views (normalized by the mean total page views per day)
_ (unlogged) 1,353,030 0.019 0.057 0.000 0.000 0.002 0.005 0.015 0.073 7.668
_ (unlogged) 1,353,030 0.010 0.032 0.000 0.000 0.001 0.003 0.008 0.042 3.071
_ (unlogged) 1,353,030 0.008 0.026 0.000 0.000 0.001 0.002 0.007 0.032 4.754
_ (unlogged) 1,353,030 1.098 6.962 0.000 0.019 0.072 0.186 0.571 4.031 1507.760
_ 1,353,030 0.017 0.044 0.000 0.000 0.002 0.005 0.015 0.070 2.160
_ 1,353,030 0.010 0.027 0.000 0.000 0.001 0.003 0.008 0.041 1.404
_ 1,353,030 0.008 0.022 0.000 0.000 0.001 0.002 0.007 0.032 1.750
_ 1,353,030 0.394 0.573 0.000 0.019 0.070 0.170 0.452 1.616 7.319
Information Events
_ 1,353,030 0.011 0.104 0.000 0.000 0.000 0.000 0.000 0.000 1.000
__ 1,353,030 0.002 0.041 0.000 0.000 0.000 0.000 0.000 0.000 1.000
__ 1,353,030 0.009 0.096 0.000 0.000 0.000 0.000 0.000 0.000 1.000
10 1,353,030 0.001 0.038 0.000 0.000 0.000 0.000 0.000 0.000 1.000
10 1,353,030 0.003 0.050 0.000 0.000 0.000 0.000 0.000 0.000 1.000
8 1,353,030 0.020 0.138 0.000 0.000 0.000 0.000 0.000 0.000 1.000
4 1,353,030 0.039 0.194 0.000 0.000 0.000 0.000 0.000 0.000 1.000
_ 1,353,030 0.000 0.017 0.000 0.000 0.000 0.000 0.000 0.000 1.000
Visibility
1,353,030 0.010 0.101 0.000 0.000 0.000 0.000 0.000 0.000 1.000
1,353,030 0.010 0.097 0.000 0.000 0.000 0.000 0.000 0.000 1.000
1,353,030 0.325 0.575 0.000 0.000 0.000 0.000 0.693 1.609 2.833
_ 1,353,030 0.299 0.458 0.000 0.000 0.000 0.000 1.000 1.000 1.000
_ 1,353,030 0.000 0.001 -0.020 0.000 0.000 0.000 0.000 0.000 0.021
__ 1,353,030 0.007 0.013 0.000 0.000 0.000 0.000 0.009 0.033 0.077
__ 1,353,030 -0.007 0.012 -0.069 -0.032 -0.009 -0.001 0.000 0.000 0.000
Analyst Characteristics
_ 1,353,030 0.007 0.085 0.000 0.000 0.000 0.000 0.000 0.000 1.000
1,353,030 0.007 0.086 0.000 0.000 0.000 0.000 0.000 0.000 1.000
1,353,030 0.012 0.109 0.000 0.000 0.000 0.000 0.000 0.000 1.000
(continued)
44
Std.
N Mean Min 5% 25% 50% 75% 95% Max
Dev.
Controls
1,353,030 0.153 0.387 0.000 0.000 0.000 0.000 0.000 1.099 3.932
1,353,030 20.979 1.853 17.129 18.054 19.617 20.916 22.219 24.217 25.506
1,353,030 -0.018 0.193 -1.070 -0.425 -0.009 0.024 0.063 0.151 0.288
1,353,030 0.232 0.206 0.000 0.000 0.041 0.200 0.376 0.618 0.936
1,353,030 0.543 0.416 0.022 0.085 0.247 0.442 0.735 1.277 2.411
__ 1,353,030 0.023 0.013 0.008 0.009 0.014 0.019 0.028 0.049 0.077
This table presents the firm-day descriptive statistics for all variables. The total unique number of firms is 4,068. See the Appendix for variable definitions.
45
TABLE 2
Correlation Matrix
Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25)
(1) AT_PAGEVIEW -
(2) AO_PAGEVIEW 0.98 -
(3) AE_PAGEVIEW 0.97 0.92 -
(4) EARN_ANN 0.11 0.06 0.16 -
(5) 10K 0.01 0.01 0.00 0.00 -
(6) 10Q 0.00 0.00 0.00 -0.01 0.00 -
(7) 8K 0.04 0.04 0.03 -0.01 0.01 0.02 -
(8) FORM4 0.04 0.04 0.04 -0.02 0.03 0.02 0.06 -
(9) CHG_GUIDE 0.02 0.01 0.03 0.00 0.00 0.00 0.07 0.00 -
(10) INITIATION 0.07 0.06 0.07 0.08 0.00 0.00 0.01 0.00 0.01 -
(11) DROP 0.05 0.04 0.04 0.01 0.00 0.00 0.02 0.01 0.01 -0.01 -
(12) MEDIA 0.28 0.26 0.28 0.30 0.07 0.09 0.24 0.34 0.04 0.08 0.05 -
(13) CHG_SHORT 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -
(14) AB_RET_P 0.06 0.05 0.06 0.04 0.00 0.00 0.02 -0.01 0.01 0.02 0.00 0.03 0.00 -
(15) AB_RET_N -0.05 -0.04 -0.05 -0.04 0.00 0.01 -0.01 0.02 -0.01 -0.02 0.00 0.00 0.00 0.30 -
(16) ALL_STAR 0.13 0.12 0.15 0.14 0.00 0.00 0.01 0.00 0.02 0.10 0.02 0.12 0.00 0.02 -0.02 -
(17) BOLD 0.10 0.08 0.11 0.16 0.00 0.00 0.01 -0.01 0.05 0.14 0.02 0.12 0.00 0.05 -0.04 0.17 -
(18) TIMELY 0.10 0.08 0.12 0.20 0.00 0.00 0.01 -0.01 0.06 0.16 0.02 0.15 0.00 0.05 -0.04 0.23 0.41 -
(19) REPORT 0.28 0.26 0.28 0.21 0.01 0.01 0.06 0.03 0.05 0.20 0.06 0.31 0.00 0.05 -0.04 0.32 0.32 0.39 -
(20) OT_PAGEVIEW 0.81 0.80 0.74 0.10 0.01 0.01 0.06 0.06 0.02 0.07 0.05 0.32 0.01 0.11 -0.10 0.12 0.09 0.10 0.31 -
(21) MCAP 0.35 0.33 0.31 0.00 0.00 0.00 0.00 0.05 0.00 0.05 0.04 0.22 0.00 -0.13 0.17 0.09 0.03 0.05 0.27 0.33 -
(22) ROA 0.06 0.04 0.08 0.00 0.00 0.01 -0.01 0.02 0.00 0.01 0.01 0.04 0.00 -0.13 0.15 0.03 0.01 0.02 0.08 -0.10 0.39 -
(23) LEV 0.05 0.05 0.03 0.00 0.00 -0.01 0.02 -0.01 0.00 0.02 0.02 0.04 0.00 -0.04 0.03 0.02 0.00 0.00 0.06 0.03 0.20 0.12 -
(24) BTM -0.07 -0.06 -0.07 0.00 0.00 0.00 -0.01 -0.04 0.00 -0.01 0.00 -0.04 0.00 0.03 -0.05 -0.02 0.00 -0.02 -0.02 -0.08 -0.24 0.05 0.01 -
(25) PAST_RET_VOL 0.01 0.01 0.00 0.00 0.00 -0.01 0.01 -0.03 0.00 -0.01 -0.01 -0.05 0.00 0.23 -0.26 -0.03 0.00 -0.02 -0.07 0.18 -0.50 -0.53 -0.15 0.06 -
This table presents Pearson correlation coefficients for all firm-day observations. The insignificant correlation coefficients are bolded. See the Appendix for
variable definitions.
46
TABLE 3
Univariate Analysis of Analyst Total Page Views
AT_PAGEVIEWt AT_PAGEVIEWt
Event N Mean Event N Mean Ratio in Means Diff. in Mean t-val
Information Events
_ =1 14,827 0.063 _ =0 1,338,203 0.017 3.704 0.046 127.73
__ =1 2,256 0.084 __ =0 1,350,774 0.017 4.864 0.067 72.89
__ =1 12,571 0.059 __ =0 1,340,459 0.017 3.453 0.042 107.29
10 =1 2,000 0.024 10 =0 1,351,030 0.017 1.382 0.007 6.81
10 =1 3,459 0.021 10 =0 1,349,571 0.017 1.195 0.003 4.57
8 =1 26,400 0.029 8 =0 1,326,630 0.017 1.667 0.012 42.32
4 =1 53,115 0.027 4 =0 1,299,915 0.017 1.583 0.010 51.46
_ =1 372 0.072 _ =0 1,352,658 0.017 4.116 0.054 23.98
Visibility
=1 14,053 0.047 =0 1,338,977 0.017 2.708 0.029 79.21
=1 12,862 0.038 =0 1,340,168 0.017 2.194 0.021 53.25
MEDIA_DUMt=1 404,093 0.032 MEDIA_DUMt=0 948,937 0.011 2.772 0.020 251.98
_ >0 47,778 0.022 _ <=0 1,305,252 0.017 1.295 0.005 25.04
__ >0 652,850 0.018 __ <0 700,180 0.017 1.039 0.001 8.89
Analyst Characteristics
_ =1 9,826 0.086 _ =0 1,343,204 0.017 5.095 0.070 158.25
=1 10,049 0.066 =0 1,342,981 0.017 3.845 0.049 111.60
=1 16,214 0.058 =0 1,336,816 0.017 3.442 0.041 120.52
This table presents the univariate analysis of Analyst Total page views (_ ) by information events, visibility, and analyst characteristics. Ratio in
Means refers to the ratio of mean analyst total page views when the indicator equals one scaled by the ratio of mean analyst total page views when the indicator
equals zero. T-val refers to t-statistics obtained from two-sided tests of differences in means. See the Appendix for variable definitions.
47
TABLE 4
Analyst Page Views and Information Events
Panel A: Regressions
(1) (2) (3)
Dependent Variable AT_PAGEVIEWt
Coeff. t-stat Coeff. t-stat Coeff. t-stat
Intercept 0.0441*** 9.77 0.0433*** 9.76 0.0441*** 9.78
Information Events
_1 0.0078*** 22.48
_ 0.0260*** 25.04 0.0269*** 25.22
_+1 0.0141*** 18.01
__ 0.0353*** 14.14
__ 0.0244*** 21.94
81 -0.0003 -1.62
8 0.0010*** 3.54 0.0013*** 4.75 0.0010*** 3.56
8+1 0.0005** 2.23
101 -0.0014*** -3.23
10 -0.0018*** -5.14 -0.0014*** -4.23 -0.0018*** -5.12
10+1 -0.0016*** -3.84
101 -0.0006* -1.71
10 -0.0016*** -4.31 -0.0009** -2.50 -0.0016*** -4.34
10+1 -0.0013*** -3.73
41 -0.0001 -0.68
4 -0.0004** -2.19 -0.0001 -0.32 -0.0004** -2.17
4+1 -0.0007*** -3.89
_1 0.0023* 1.71
_ 0.0309*** 6.38 0.0314*** 6.44 0.0309*** 6.39
_+1 0.0285*** 6.40
Controls
OT_PAGEVIEWt 0.0347*** 23.28 0.0341*** 23.04 0.0346*** 23.26
AT_PAST30t 0.6025*** 16.69 0.6086*** 17.08 0.6029*** 16.71
-0.0020*** -10.06 -0.0020*** -10.05 -0.0020*** -10.07
0.0113*** 9.93 0.0111*** 9.90 0.0113*** 9.92
-0.0000 -0.06 -0.0000 -0.09 -0.0000 -0.07
-0.0011*** -3.34 -0.0011*** -3.35 -0.0011*** -3.30
__ -0.3291*** -16.78 -0.3234*** -16.75 -0.3288*** -16.76
This table presents the relation between investor demand for total analyst information and information events. The
demand for total analyst information (_ ) is measured by the natural log of one plus firm i's total
number of normalized Analyst Estimates and Analyst Opinion page views from Yahoo Finance on day t. Page views
48
are normalized by the mean daily page view counts for all U.S. listed firms provided by Yahoo Finance during our
sample period. Panel A presents the regression results. Column (1) includes six information events on day t:
earnings announcements, changes in management guidance, Forms 10-K, 10-Q, 8-K, and 4. Column (2) further
includes the day before and the day after the information events. For example, _+ , j=-1, 1 equals one if
firm i has an earnings announcement on day t+j, and zero otherwise. Column (3) splits the indicator for earnings
announcements (_ ) into observations with changes in management EPS guidance (__ ) and
without changes in management EPS guidance (__ ) on day t.*, **, *** indicate significance at the
0.10, 0.05, and 0.01 levels, respectively, using two-tailed tests. T-statistics and p-values are calculated using
clustered standard errors by firm. Week fixed effects and day-of-week fixed effects are included in the regressions.
Panel B presents the p-values for the F-tests on the pairwise differences of the coefficients for information events in
Panel A Column (3). See the Appendix for variable definitions.
49
TABLE 5
Analyst Estimate/Opinion Page Views and Information Events
This table presents the relation between investor demand for analyst estimates (Column 1) and analyst opinions
(Column 2), and information events. The demand for analyst estimates ( _ )/opinions
( _ ) is measured by the natural log of one plus firm i's total number of normalized Analyst
Estimate/Analyst Opinion page views from Yahoo Finance on day t. Page views are normalized by the mean daily
page view counts for all U.S. listed firms provided by Yahoo Finance during our sample period. Column (3) uses the
difference between analyst estimates page views and the analyst opinions page views as the dependent variable and
tests whether the investor demand on information event days differs between analyst estimates and analyst opinions
page views. PAST30t refers to AE_PAST30t, AO_PAST30t, and AE-AO_PAST30t, for Columns (1)-(3), respectively.
*, **, *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively, using two-tailed tests. T-statistics and
p-values are calculated using clustered standard errors by firm. Week fixed effects and day-of-week fixed effects are
included in the regressions. See the Appendix for variable definitions.
50
TABLE 6
Analyst Page Views, Visibility, and Analyst Characteristics
51
Panel B: Interactions with Media
(1) (2) (3)
Dependent Variable _ _ _
Coeff. t-stat Coeff. t-stat Coeff. t-stat
Intercept 0.0472*** 10.48 0.0248*** 10.87 0.0299*** 10.41
Visibility
0.0017*** 4.06 0.0008*** 3.38 0.0010*** 3.85
0.0024*** 6.06 0.0012*** 5.48 0.0016*** 5.85
0.0013*** 8.10 0.0006*** 8.24 0.0007*** 7.50
_ -0.0253 -1.36 -0.0221** -2.14 -0.0087 -0.77
__ 0.0930*** 13.74 0.0505*** 15.87 0.0520*** 12.02
__ -0.0988*** -15.73 -0.0558*** -17.30 -0.0542*** -14.15
Analyst Characteristics
_ -0.0068*** -6.32 -0.0046*** -6.54 -0.0041*** -5.92
-0.0032*** -3.39 -0.0013** -2.04 -0.0031*** -4.98
-0.0013** -2.42 0.0002 0.75 -0.0019*** -5.75
_ * 0.0166*** 10.72 0.0115*** 10.80 0.0089*** 9.01
* 0.0086*** 6.03 0.0057*** 5.69 0.0053*** 5.89
* 0.0018** 2.38 0.0011** 2.22 0.0010** 2.18
Information Events
_ 0.0147*** 22.34 0.0178*** 29.26 -0.0020*** -6.30
10 -0.0008*** -2.98 -0.0006*** -4.33 -0.0002 -0.86
10 -0.0033*** -8.40 -0.0022*** -9.88 -0.0014*** -6.12
8 -0.0027*** -6.49 -0.0018*** -7.76 -0.0012*** -4.77
4 -0.0015*** -6.16 -0.0011*** -8.45 -0.0006*** -3.97
_ 0.0163*** 3.62 0.0151*** 4.85 0.0025 1.03
Controls
REPORT 0.0041*** 12.47 0.0021*** 13.53 0.0025*** 10.37
OT_PAGEVIEWt 0.0321*** 21.76 0.0142*** 19.94 0.0181*** 19.67
PAST30t 0.6190*** 17.96 0.6511*** 17.47 0.6732*** 18.41
-0.0022*** -10.83 -0.0012*** -11.21 -0.0014*** -10.77
0.0108*** 10.19 0.0062*** 11.11 0.0048*** 8.19
-0.0002 -0.33 -0.0005** -1.96 0.0002 0.64
-0.0014*** -4.42 -0.0010*** -6.31 -0.0006*** -3.14
__ -0.3559*** -18.86 -0.1643*** -16.70 -0.2219*** -19.73
Firm clustering Yes Yes Yes
Week FE Yes Yes Yes
Day-of-week FE Yes Yes Yes
Adj 2 0.81 0.72 0.80
N 1,353,030 1,353,030 1,353,030
This table presents the relation between investor demand for analyst estimates and opinions, and visibility and
analyst characteristics. Columns (1)-(3) show the results for the demand for total analyst information, analyst
estimates, and analyst opinions, respectively. The demand for total analyst information ( _ ) is
measured by the natural log of one plus firm i's total number of normalized Analyst Estimates and Analyst Opinion
page views from Yahoo Finance on day t. Page views are normalized by the mean daily page view counts for all
U.S. listed firms provided by Yahoo Finance during our sample period. Panel A presents regressions where there are
no media and analyst characteristics interactions. Panel B presents regressions where there are media and analyst
characteristic interactions in order to test whether the analyst characteristic findings are more pronounced in the
presence of media. PAST30t refers to AT_PAST30t, AE_PAST30t, and AO_PAST30t, for Columns (1)-(3),
respectively. *, **, *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively, using two-tailed tests.
T-statistics and p-values are calculated using clustered standard errors by firm. Week fixed effects and day-of-week
fixed effects are included in the regressions. See the Appendix for variable definitions.
52
TABLE 7
Descriptive Statistics of Page Views by Page Views Category
Panel B: Mean Value of Ratios of Total Analysts Page Views to Other Page Views
N ANALYST to SEC FILINGS ANALYST to FINANCIAL STATEMENTS ANALYST to NEWS
_ 1,353,030 6.6 2.3 3.9
_ =1 14,827 17.4 4.3 4.2
__ =1 2,256 19.9 4.2 3.9
__ =1 12,571 17.0 4.3 4.2
10 =1 2,000 8.2 2.4 4.1
10 =1 3,459 7.5 2.1 3.6
8 =1 26,400 9.7 2.4 3.8
4 =1 53,115 9.9 2.4 4.3
_ =1 372 19.6 3.1 4.1
__ =1 1,257,450 6.3 2.3 3.9
This table presents the descriptive statistics of page views for each information event by page views category. Panel A presents the mean value of page views for
the top eight categories ranked by the mean value of page views including summary, message board, total analyst, key statistics, historical prices, options, profile,
53
and all other. Panel B presents the mean value of ratios of total analysts page views to SEC filings, financial statement, and news page views, respectively. All
ratios are significantly different from 1 (p < 0.01). See the Appendix for variable definitions.
54
TABLE 8
High Page Views and Analyst Reports
This table reports the number of firm-day observations for the High Page Views and Non-High Page Views samples. The High
Page Views sample refers to observations with Analyst Total page views on day t (_ ) that are greater than two
times the trailing 30-day average Analyst Total page views on day t-1 (AT_PAST30t). Panel A shows the number of High Page
Views firm-day observations with or without analyst reports on day t and day t+1. Panel B shows the distribution of High Page
Views and Non-High Page Views firm-day observations by each major information event day. See the Appendix for variable
definitions.
55