summary and PowerPoint

Make a summary and PowerPoint (A,B)

A. prepare a summary of each paper (3 Article) (including the Libby box)

Focus on summarizing the most salient points of the article.

. After that make sure the summaries address the following questions (I will send them).

B. and make PowerPoint slides for each study.

DOI: 10.1111/1475-679X.12067
Journal of Accounting Research
Vol. 53 No. 1 March 2015
Printed in U.S.A.
Inside the “Black Box” of Sell-Side
Financial Analysts
L A W R E N C E D . B R O W N ,∗ A N D R E W C . C A L L ,†
M I C H A E L B . C L E M E N T ,‡ A N D N A T H A N Y . S H A R P§
Received 7 March 2014; accepted 27 October 2014
ABSTRACT
Our objective is to penetrate the “black box” of sell-side financial analysts by
providing new insights into the inputs analysts use and the incentives they
face. We survey 365 analysts and conduct 18 follow-up interviews covering
a wide range of topics, including the inputs to analysts’ earnings forecasts
and stock recommendations, the value of their industry knowledge, the determinants of their compensation, the career benefits of Institutional Investor
All-Star status, and the factors they consider indicative of high-quality earnings. One important finding is that private communication with management
is a more useful input to analysts’ earnings forecasts and stock recommendations than their own primary research, recent earnings performance, and
∗ Temple University; † Arizona State University; ‡ University of Texas at Austin; § Texas A&M
University.
Accepted by Christian Leuz. We appreciate helpful comments from two anonymous reviewers, Mike Baer, David Bailey, Shuping Chen, Artur Hugon, Stephannie Larocque, Bill
Mayew, Lynn Rees, Kim Ritrievi, Debika Sihi, Nathan Swem, Michael Tang (FARS discussant),
Yen Tong, Senyo Tse, James Westphal, Richard Willis, Yong Yu, and workshop participants at
Colorado State University, Georgetown University, Indiana University, Texas Christian University, Tulane University, the 2013 Southeast Summer Accounting Research Conference
(SESARC), the 2013 Temple University Accounting Conference, and the AAA Financial Accounting and Reporting Section 2014 Midyear Meeting. This paper was a finalist for the 2014
FARS Midyear Meeting best paper award. We are thankful for survey design assistance from
Veronica Inchauste of the Office of Survey Research at the Annette Strauss Institute, and
the excellent research assistance from John Easter, Alexandra Faulk, Emily Hammack, Ashley Loest, Lauren Schwaeble, Sarah Shaffell, and Paul Wong. An online appendix to this
paper can be downloaded at http://research.chicagobooth.edu/arc/journal-of-accountingresearch/online-supplements.
1
C , University of Chicago on behalf of the Accounting Research Center, 2014
Copyright 
2
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
recent 10-K and 10-Q reports. Another notable finding is that issuing earnings forecasts and stock recommendations that are well below the consensus
often leads to an increase in analysts’ credibility with their investing clients.
We conduct cross-sectional analyses that highlight the impact of analyst and
brokerage characteristics on analysts’ inputs and incentives. Our findings are
relevant to investors, managers, analysts, and academic researchers.
JEL codes: G20; G23; G24; G28; M40; M41
Keywords: sell-side analysts; analyst inputs; analyst incentives; private communication; analyst compensation; industry knowledge
1. Introduction
Sell-side financial analysts are of significant interest to academic researchers
because of their prominent role in analyzing, interpreting, and disseminating information to capital market participants. While early research on analysts focused on the statistical properties of their earnings forecasts and on
improving analysts’ expectations models (Fried and Givoly [1982], O’Brien
[1988], Lys and Sohn [1990], Brown [1993]), later research investigated
the investment value of analysts’ earnings forecasts and stock recommendations (Womack [1996], Francis and Soffer [1997], Clement and Tse [2003],
Howe, Unlu, and Yan [2009]). Starting with Schipper [1991] and Brown
[1993], however, researchers have suggested the literature should focus
more on the context within which analysts make their decisions. More recently, Ramnath, Rock, and Shane [2008] and Bradshaw [2011] conclude
that research on the “black box” of analysts’ decision processes is required
for the literature to progress. We penetrate this “black box” by surveying
365 analysts and conducting 18 follow-up interviews to gain insights into
the inputs they use and the incentives they face.1
The inputs we investigate include the determinants of analysts’ earnings
forecasts and stock recommendations; the frequency, nature, and usefulness of their communication with senior management; the valuation models they use to support their stock recommendations; their beliefs about
what constitutes high-quality earnings; and, their perceptions of possible
“red flags” of financial misrepresentation. With respect to incentives, we investigate the determinants of analysts’ compensation, their motivation for
generating accurate earnings forecasts and profitable stock recommendations, and the consequences of issuing unfavorable earnings forecasts and
stock recommendations. While prior research has generally focused on analysts’ incentives to please company management or generate underwriting
business, our findings highlight the strong incentives analysts face to satisfy
their investing clients.
1 Surveys have limitations, such as the potential for response bias, small sample sizes, social
desirability biases, and construct validity issues. However, surveys enable researchers to ask
questions that would be difficult to address with archival data.
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
3
We summarize our main findings here and discuss our detailed results in
section 3. Our findings shed light on the value of private communication
with management as an input to analysts’ decision processes. Soltes [2014]
finds that private communication with management is a valuable source of
information for analysts. We extend Soltes [2014] by providing evidence
that over half of the analysts we survey report that they have direct contact
with the CEO or CFO of the typical company they follow five or more times
a year. We also find that private communication with management is a more
important input to analysts’ earnings forecasts and stock recommendations
than primary research, recent earnings performance, and recent 10-K and
10-Q reports. Further, analysts rate private phone calls as one of the most
useful types of direct contact with management for purposes of generating their earnings forecasts and stock recommendations. Our follow-up
interviews reveal that some analysts avoid asking questions during public
conference calls and use private phone conversations to check the assumptions of their models, to gain qualitative insights into the firm and its industry, and to get other details not explained on public calls. Our findings
provide a deeper understanding of analysts’ communication with management in the post–Regulation Fair Disclosure (Reg FD) environment and
suggest analysts incorporate pieces of nonpublic information from management into a broader “mosaic.”
Institutional Investor (II) surveys regularly find that sell-side analysts’ industry knowledge is extremely valuable to their buy-side clients. We provide evidence that industry knowledge is a very important determinant of sell-side
analysts’ compensation, suggesting brokerage houses provide analysts with
incentives to satisfy their clients’ demand for industry knowledge (Brown
et al. [2014]). We also find that industry knowledge is the single most useful
input to analysts’ earnings forecasts and stock recommendations.
We asked analysts about their perceptions of earnings quality and
their beliefs about potential “red flags” of intentional misreporting. Although Dichev et al. [2013] asked similar questions of the CFOs they
surveyed, users of financial accounting information (analysts) are likely
to have more informative views on financial reporting issues than preparers (CFOs). Specifically, analysts are an important source of information for their investing clients and have incentives to recognize
attributes of high-quality earnings because incorrect assessments of earnings quality could result in economic losses for their clients and have
an adverse effect on their own reputation and compensation. Conversely, CFOs face incentives to manage earnings, which could create a preference for low-quality earnings and bias their responses to
questions about earnings quality (Dechow et al. [2010], Nelson and
Skinner [2013]). In addition, CFOs have other reporting incentives that are
not always consistent with those of investors (Nelson and Skinner [2013]).
For example, Dichev et al. [2013] find that CFOs rate the avoidance of longterm estimates as an important feature of high-quality earnings. However,
the analysts we survey do not believe this factor is an important earnings
4
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
attribute, suggesting CFOs may simply prefer earnings that do not require
additional explanations to external parties (Nelson and Skinner [2013]).
The factors analysts believe are most indicative of high-quality earnings
include that earnings are backed by operating cash flows, are sustainable
and repeatable, reflect economic reality, and reflect consistent reporting
choices over time. While these findings suggest analysts could rein in earnings management before it escalates into more egregious misrepresentations of the financial statements (Schrand and Zechman [2012]), we also
find analysts generally do not focus on detecting fraud or intentional misreporting.
With respect to incentives, our results provide a better understanding of
the nature and structure of analyst compensation. Regulators and investors
have expressed concerns about analysts’ conflicts of interest, and the SEC
and the major U.S. stock exchanges have worked together to fortify the
“Chinese wall” separating the investment banking and research sides of brokerage houses. In spite of these efforts, 44% of our respondents say their
success in generating underwriting business or trading commissions is very
important to their compensation, suggesting conflicts of interest remain a
persistent concern for users of sell-side research.
While many prior studies emphasize II’s annual All-America Research
Team rankings (e.g., Stickel [1992], Leone and Wu [2007], Rees, Sharp,
and Twedt [2014a]), the analysts we survey say broker votes are far more
important to their career advancement.2 Specifically, 83% of analysts indicate that broker votes are very important to their career advancement, while
only 37% say the same about the II rankings. Our findings are consistent
with Maber, Groysberg, and Healy [2014], who find that unlike II rankings,
broker votes translate directly into revenue for analysts’ employers.
We highlight other incentives analysts face. For example, one of their primary motivations for issuing accurate earnings forecasts is to use them as
inputs to their own stock recommendations, revealing that analysts’ forecasts are often a means to an end rather than an end unto themselves.
In addition, analysts report that an increase in their credibility with investing
clients is a more likely consequence of issuing unfavorable earnings forecasts and stock recommendations than many of the negative consequences
discussed in prior research, such as being “frozen out” of the Q&A portion of future conference calls (Mayew [2008]). This finding underscores
analysts’ balancing act of satisfying both company management and their
investing clients.
We conduct cross-sectional analyses that investigate the influence of
analyst characteristics (gender, education, professional certifications,
experience, and All-Star status) and brokerage house characteristics (size,
investment banking activity, and client focus) on analysts’ inputs and incentives. Some of our results help explain findings in the existing literature.
2 Buy-side portfolio managers and buy-side analysts assess the value of research services provided by sell-side brokerage houses and allocate research commissions through broker votes.
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
5
For example, we find that female analysts are more motivated to issue
accurate earnings forecasts so they can use them as inputs to their stock recommendations, providing a partial explanation for Kumar’s [2010] result
that female analysts issue superior earnings forecasts. Other cross-sectional
results add texture to our interviews and deepen our understanding of the
main findings. For instance, our finding that analysts at large brokerage
houses are more likely to indicate that private communication with management is a useful input to their stock recommendations is consistent with
a potential information advantage for these analysts (Clement [1999]).
We make several contributions to the literature. A survey allows us to ask
analysts questions about their inputs and incentives that would be difficult
to address with archival data, enabling us to provide the literature with new
insights. Some of our findings strengthen the extant literature. For example, Soltes [2014] uses field evidence from a single large-cap firm to show
that private communication with management is valuable to sell-side analysts. We validate this finding with a broad sample of analysts following
many firms from multiple industries and add context by assessing the value
of private communication relative to other inputs analysts employ.
We also highlight areas where analysts’ survey responses diverge from the
findings of prior research (e.g., the contrast between analysts’ and CFOs’
views on earnings quality), and we provide direction for future research.
For example, we address issues not considered by prior studies, such as the
benefits to analysts of issuing relatively pessimistic earnings forecasts and
stock recommendations. In general, our findings underscore the challenge
analysts face when trying to maintain good relationships with firm management while also satisfying the demands of their investing clients. Our study
is relevant to investors who use analysts’ earnings forecasts and stock recommendations in their investing decisions, managers of companies followed
by analysts, and analysts wishing to benchmark their practices and research
against a broad set of peers.
2. Survey Methodology, Interviews, and Cross-Sectional Analyses
2.1 SUBJECT POOL
Our subject pool consists of sell-side analysts with an equity research report published in Investext during the 12-month period from October 1,
2011, to September 30, 2012. Investext includes more than 150,000 research reports from over 1,000 investment banks and brokerage houses
during our sample period. We recorded the name, email address, phone
number, and employer of every analyst with a sole-authored research report in Investext during this period. Analysts sometimes submit multiauthored (or team) research reports (Brown and Hugon [2009]). Thus, for
every lead analyst who submitted a team report, we identified his or her
most recent team report and collected contact information for every analyst
on that team. This process yielded 3,341 sell-side analysts with very recent
6
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
experience. As a frame of reference, our subject pool is 77.2% of the number of analysts in I/B/E/S who issued an annual earnings forecast for at
least one U.S. firm in 2012.
2.2 SURVEY DESIGN AND DELIVERY
We initially developed a list of survey questions based on our review of
the literature. Our intent was to identify relevant questions that would be
difficult to address using only archival data. After compiling a list of questions, we contacted academic colleagues who are familiar with this literature and asked them to suggest questions they would like to ask a group
of sell-side analysts.3 We received feedback on survey design from a professional survey consultant who contracts with a large public university and
from academic colleagues in various disciplines who are experienced in
conducting surveys. We distributed pilot surveys to several analysts and academic colleagues who helped us assess the reasonableness and presentation
of our questions and the time required to complete the survey. This process helped reduce the possibility that we omitted fundamental questions,
asked unimportant or ambiguous questions, or designed a survey requiring
too much time to complete.
In an effort to address as many topics as possible, we created and administered two related versions of the survey, each containing 14 questions followed by several demographic questions. Both versions of the survey begin
with five identical “common” questions, followed by six similar “twin” questions. In one version, the twin questions are specific to earnings forecasts
(hereafter, EF version); in the other version, the twin questions are specific
to stock recommendations (hereafter, SR version) but are otherwise identical. In each version, the twin questions are followed by three “unique” questions that are loosely related to the theme of either the EF or SR version.
For example, the EF version asks analysts about earnings quality, while the
SR version asks analysts about the valuation models they employ. We asked
a total of 23 questions across the two versions of the survey: 6 specific to
earnings forecasts, 6 specific to stock recommendations, and 11 addressing
analysts’ inputs and incentives in other contexts. The survey instrument is
available in an online appendix.4
We asked the common questions first because we did not want our subjects to think we deemed either earnings forecasts or stock recommendations (depending on which version of the survey they received) to be
particularly important. We asked the twin questions next to ensure that
the responses to these questions would not be influenced by the different
sets of unique questions, which we presented last. With one exception, we
3 Other surveys of financial analysts include Bricker et al. [1995], Barker [1999], and Barker
and Imam [2008].
4 An online appendix to this paper can be downloaded at http://research.
chicagobooth.edu/arc/journal-of-accounting-research/online-supplements.
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
7
randomized the order of the questions presented within each set of questions (common, twin, unique).5 Unless the options had a natural sequence
(e.g., never, once a year, twice a year), we randomized the order of each
question’s options.6 Our survey ended with a series of demographic questions. Demographic characteristics and the correlations among them are
included in the online appendix.
We used Qualtrics.com to deliver the survey via email on January 9, 2013.
Two weeks later, we sent a reminder email to analysts who had not completed the survey.7 We closed the survey on February 6, 2013, four weeks
after our original email. To encourage participation, we told our subjects
we would donate $10,000 multiplied by the response rate to our survey and
that we would allocate the total donation among four charities from which
we allowed the analysts to choose.
We informed analysts that their responses would be held in strict confidence, that no individual response would be reported, and that the survey
should take less than 15 minutes to complete.8 Qualtrics.com assigned each
responding analyst, in alternating fashion, one of the two versions of the
survey. We received a total of 365 responses for a response rate of 10.9%,
which exceeds that of other accounting and finance surveys administered
via email (e.g., Dichev et al. [2013] report a response rate of 5.4%, and
Graham, Harvey, and Rajgopal [2005] report an 8.4% response rate on the
portion of their survey delivered via the internet).
2.3 INTERVIEWS
We asked analysts to provide their phone numbers if they were willing
to be contacted for a follow-up interview. Eighty-two analysts provided their
phone numbers, and we conducted one-on-one interviews with 18 analysts
to gain additional insights beyond those contained within the responses to
our survey.9 We made audio recordings of 13 of these interviews (average
5 In each version of the survey, we asked two “twin” questions about how often research
management exerts upward or downward pressure on analysts’ earnings forecasts (EF version)
or stock recommendations (SR version). Because these two questions are naturally related to
each other, we wanted analysts to answer them in sequence. Therefore, we asked these two
“twin” questions last.
6 See tables A3 and A5 in the online appendix.
7 We used the Kolmogorov-Smirnov test (untabulated) to compare the distribution of demographic characteristics between analysts who responded to the survey early (i.e., before we
sent the reminder email) versus late (i.e., after we sent the reminder email). We cannot reject
the null hypothesis of equal distributions for any characteristic except analyst age, where the
p-value is a marginally significant 0.086 (two-tailed). We did not compare the distribution of
degrees and certifications between early and late responders because analysts can have multiple degrees (e.g., an undergraduate degree in economics and an MBA) and professional
certifications (e.g., CPA and CFA).
8 Excluding 21 analysts who took more than one hour to complete the survey, likely because
of interruptions at work, the mean (median) time the analysts took to complete the survey was
14.1 (12.0) minutes.
9 We conducted 17 interviews by phone and one in person. Before conducting any interviews, we tabulated all the demographic information for each analyst who volunteered to be
8
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
length was 30 minutes, 50 seconds) and took detailed notes on the other
five. The 18 analysts we interviewed represent four of the nine primary industries listed in the survey and six “other” industries: four are female, they
have a median of three to six years of experience both as sell-side analysts
and at their current employer, they follow a median of 16–25 companies,
and 55% of them work at brokerage houses with more than 25 sell-side
analysts.
2.4 CROSS-SECTIONAL ANALYSES
We explore cross-sectional variation in survey responses based on analyst and brokerage house characteristics (Clement [1999]). For each survey
question, we regress analysts’ responses (which usually range from 0 to 6)
on the following 12 characteristics:
Survey Response = ß0 + ß1 Gender + ß2 Accounting + ß3 MBA + ß4 CFA
+ ß5 Experience + ß6 I I AllStar + ß7 StarMine + ß8 WSJ
+ ß9 Broker Size + ß10 I Bank + ß11 Retail Focus
+ ß12 HF Focus + Industry + ε,
(1)
where Survey Response is the analyst’s response to the survey question being
examined. We formally define the independent variables in the appendix.
We obtain values of six independent variables (Gender, Accounting, MBA,
CFA, Experience, and Broker Size) from the results of demographic questions
we pose in the survey. Unlike Gender, Accounting, MBA, and CFA, neither
Experience nor Broker Size is a binary response. To facilitate interpretation
of our results, we create indicator variables for Experience and Broker Size
based on the median response for each variable, allowing for approximately
the same number of analysts to be coded either 0 or 1 (e.g., 7+ years for
Experience; 26+ sell-side analysts for Broker Size).
We hand-collect the data for WSJ, StarMine, and II AllStar to examine
whether award-winning analysts use different inputs or have different incentives from other analysts.10 We define each of these variables based
on award status on the date we administered the survey. Following prior
research (Bradshaw, Huang, and Tan [2014], Rees, Sharp, and Wong
interviewed. Our objective was to interview analysts with a range of demographic characteristics (e.g., gender, experience, primary industry, broker size) that represented the overall
sample. Thus, we interviewed both male and female analysts with varying levels of experience,
representing a variety of primary industries, and from brokerage houses of varying size. Aside
from the demographic information, we did not refer to any individual survey responses when
deciding whom to call or what to ask. No analyst we contacted declined our request for an
interview.
10 WSJ analysts are selected based on the profitability of their recommendations. StarMine
analysts are awarded based on both the profitability of their recommendations and the accuracy of their earnings forecasts. II All-America Research analysts are selected based on votes by
institutional investors.
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
9
[2014b]), we use Thomson One Banker to determine whether analysts’ employers provide underwriting of debt or equity issuances (I Bank). We code
the last two indicator variables, Retail Focus and HF Focus, based on the survey responses compiled in table 12, to capture the extent to which retail
investing clients and hedge funds are important to the analyst’s employer.
We include industry fixed effects based on the primary industry the analyst
covers.
For brevity, we report all cross-sectional results in the online appendix.
We limit our discussion of cross-sectional results in the text to those that
are significant at the 5% level or better, briefly summarizing the results we
consider most interesting.
3. Results and Interview Responses
We organize the results based on the primary themes of our survey.
Tables 1 through 7 address the inputs analysts use in their decisions. Specifically, tables 1 and 2 relate to general inputs to analysts’ earnings forecasts
and stock recommendations, table 3 pertains to analyst direct contact with
management, and tables 4 to 7 present results relating to analysts’ assessments of financial reporting quality. Tables 8 through 13 address the incentives analysts face. Specifically, tables 8 and 9 report on the determinants
of analysts’ career success, tables 10 and 11 present responses to questions
about factors that influence analysts’ earnings forecasts and stock recommendations, and tables 12 and 13 relate to other incentives analysts face.
In the first column of each table, we report the choices for each question
based on the average ratings from the analysts. We also test whether the
average rating for a given choice exceeds the average rating of the other
choices, and, in the second column, we report the rows corresponding to
a significant difference at the 5% level, using Bonferroni-Holm–adjusted
p-values to correct for multiple comparisons. The final two columns indicate the percentage of respondents who rate each choice near the top and
bottom of the scale. In panel B of the four tables that contain the “twin”
questions (tables 1, 3, 10, and 11), the middle column further reports the
results of a t-test of the null hypothesis that the average rating is the same
across both the EF and SR versions of the survey.
3.1 FREQUENCY AND CORRELATIONS OF DEMOGRAPHIC CHARACTERISTICS
(ONLINE APPENDIX)
Among the analysts responding to the survey, the most commonly
covered “primary” industries are banking/finance/insurance (15.1%),
transportation/energy (14.5%), technology (12.3%), and retail/wholesale
(9.3%).11 Of those stating “other,” 29 analysts indicated health care, making it the fifth most covered industry (7.9%). Nearly half cover only one
11 We follow Dichev et al. [2013] in our choice of industries.
10
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
industry, and the median and modal analyst follows 16–25 firms. The vast
majority of our respondents are male and under 50 years of age. Almost half
have either an MBA or an undergraduate degree in economics or finance.
More than a third are CFAs, but less than 4% are CPAs. Approximately half
have been sell-side analysts for at least six years, have worked for their employer for at least three years, and work for a brokerage house with more
than 25 analysts. For comparative purposes, we provide statistics for all analysts in I/B/E/S during 2012. The primary difference between our sample
and I/B/E/S analysts is that our sample analysts follow more firms, suggesting I/B/E/S potentially excludes some firms that analysts follow.12
3.2 GENERAL INPUTS
One limitation of the existing literature is researchers’ inability to observe the inputs that shape analysts’ outputs (Ramnath, Rock, and Shane
[2008], Bradshaw [2011]). We asked survey questions with the goal of shedding light on the inputs analysts use when forming their earnings forecasts
and stock recommendations.
3.2.1. How Useful Are the Following for Determining Your Earnings Forecasts/
Stock Recommendations? (Table 1). While II surveys regularly find that industry knowledge is highly valued by analysts’ buy-side clients, little evidence
exists regarding the importance of industry knowledge to sell-side analysts.
Table 1 reveals that industry knowledge is the single most useful input to
both analysts’ earnings forecasts (panel A) and their stock recommendations (panel B). Industry knowledge includes understanding the industry’s key trends and technologies; its supply chains, distribution models,
and margins; and its customers, labor, and management teams. Consistent
with evidence from archival research that industry knowledge is an important strength of sell-side analysts (Piotroski and Roulstone [2004], Kadan
et al. [2012]), our respondents indicate that industry knowledge is the most
useful input to their earnings forecasts and stock recommendations.
Private communication with management is another useful input to
analysts’ earnings forecasts and stock recommendations, underscoring
the importance of analysts’ access to management. While prior research
demonstrates that private communication with management is valuable
to sell-side analysts (Soltes [2014]), we document that it is even more
useful to analysts than their own primary research, the firms’ recent
earnings performance, and the recent 10-K or 10-Q reports. Analysts at
the largest brokerage houses indicate that private communication with
management is a more useful input to their stock recommendations than
12 We unambiguously identified 209 of our sample analysts in I/B/E/S, and we compared
the number of firms these analysts say they follow with the number that I/B/E/S reports they
followed in January 2013 (immediately before we administered the survey). Sixty-four analysts
report following more firms than I/B/E/S suggests, while only 21 analysts report following
fewer firms than I/B/E/S indicates.
5.15
4.70
4.67
4.65
4.22
4.18
4.16
3.96
2.16
2.06
1.72
Average Rating
2–11
5–11
5–11
5–11
9–11
9–11
9–11
9–11
11
11
–
Significantly
Greater Than
79.35
65.76
61.96
61.41
46.45
41.30
42.39
46.20
7.07
7.07
3.80
Very Useful
(5 or 6)
0.54
3.26
1.63
1.63
2.73
3.26
4.89
14.13
36.41
42.39
46.74
Not Useful
(0 or 1)
% of Respondents Who Answered
(Continued)
Column 1 reports the average rating, where higher values correspond to greater usefulness. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating usefulness
of 5 or 6 (0 or 1).
Total possible N = 184
(1) Your industry knowledge
(2) Private communication with management
(3) Earnings conference calls
(4) Management’s earnings guidance
(5) Quality or reputation of management
(6) Recent earnings performance
(7) Recent 10-K or 10-Q
(8) Primary research (e.g., channel checks, surveys, etc.)
(9) Other analysts’ earnings forecastsa
(10) Your stock recommendationa
(11) Recent stock price performance
Responses
Panel A: Summary statistics for the EF version
TABLE 1
Survey Responses to the Question: How Useful Are the Following for Determining Your Earnings Forecasts (Stock Recommendations)?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
11
Total possible N = 181
2–11
4–11
4–11
5–11
10–11
10–11
10–11
10–11
10–11
11
–
4.56
4.21
3.98
3.92
3.90
3.87
3.27
1.56
Significantly
Greater Than
5.31
4.92
4.84
Average
Rating
34.25
32.60
38.67
33.70
21.11
2.22
5.50∗∗∗
1.86∗
1.72∗
5.93∗∗∗
9.69†††
4.08∗∗∗
56.67
2.67†††
50.28
83.43
73.33
72.22
1.83†
19.10†††
0.99
1.45
Very Useful
(5 or 6)
EF vs. SR
3.87
4.97
9.39
6.63
15.56
54.44
6.08
1.67
0.00
1.67
4.44
Not Useful
(0 or 1)
% of Respondents Who Answered
The wording of these responses is different across the two versions of the survey because one version refers to earnings forecasts (panel A) and the other version refers to stock
recommendations (panel B).
Column 1 reports the average rating, where higher values correspond to greater likelihood. Column 2 reports the results of t-tests of the null hypothesis that the average rating
for a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the other items
at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 reports the results of a t-test of the null hypothesis that the average rating
is the same across both the earnings forecast and stock recommendation versions of the survey. ∗∗∗ , ∗∗ , and ∗ (††† ,†† , and † ) indicate that the average rating in the EF (SR) version of
the survey is significantly larger at the 1%, 5%, and 10% level, respectively. Column 4 (5) presents the percentage of respondents indicating usefulness of 5 or 6 (0 or 1).
a
(1) Your industry knowledge
(2) Your earnings forecasta
(3) Private communication with
management
(4) Quality or reputation of
management
(5) Primary research (e.g., channel
checks, surveys, etc.)
(6) Earnings conference calls
(7) Recent earnings performance
(8) Recent 10-K or 10-Q
(9) Management’s earnings guidance
(10) Recent stock price performance
(11) Other analysts’ stock
recommendationsa
Responses
Panel B: Summary statistics for the SR version
T A B L E 1—Continued
12
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
13
other analysts do, offering a possible explanation for Ertimur, Sunder, and
Sunder’s [2007] finding that analysts at large brokerage houses issue more
profitable stock recommendations.
More than 70% of analysts indicate that their own earnings forecasts are
a very useful input to their stock recommendations, consistent with our
evidence in table 10 that analysts’ most important motivation for issuing accurate earnings forecasts is to use them as inputs to their own stock recommendations. Our findings reveal that analysts’ earnings forecasts are useful
not only as a stand-alone output but also as an input to their stock recommendations.
Stock prices are a leading indicator of future earnings (Beaver,
Lambert, and Morse [1980], Basu [1997]), and prior research indicates analysts’ earnings forecasts do not fully reflect the information in prior stock
price changes (Lys and Sohn [1990], Abarbanell [1991]). Similarly, our respondents indicate that recent stock price performance is not particularly
useful for determining their earnings forecasts.
Although analysts generally report that other analysts’ earnings forecasts
(stock recommendations) are not useful for determining their own earnings forecasts (stock recommendations), some interviewees said they sometimes examine other analysts’ reports.13 One said the main reason his team
looks at other analysts’ estimates is to remove stale earnings forecasts from
the consensus. Another reported, “Some analysts are just better than others, so I watch them more closely. If I notice that they’re very light on an
estimate, then it gives me pause. I say, ‘Why am I 10 cents above this guy?’
And I go back and look, and I say, ‘Am I still comfortable that I did it right?’
I’m not going to change it, but I am going to double-check. This isn’t an
idiot, and he’s 10 cents below me. Why is that?”
One analyst stated, “You keep an eye on the outliers, because a lot of
times if people do have a contrarian opinion, it’s interesting to see how
they’re thinking about it.” Another analyst said, “We don’t care about other
analysts’ stock ratings. We never look. But we do care about where estimates come out after the quarter, especially for new companies . . . If we’re
off, and we don’t have a non-consensus view on something, we ask, ‘OK,
why are we this low?’ And usually there’s a reason why, and that’s OK. But
if there’s not, it’s a red flag to us that maybe we’re overlooking part of the
story or making an error.” Consistent with prior research on herding in
analyst earnings forecasts (Trueman [1994], Welch [2000], Clement and
Tse [2005]), our comparison of responses to the twin questions reveals that
13 One inherent difficulty with surveys is that respondents may be reluctant to disclose the
full extent of certain beliefs or practices if they perceive that such disclosure could result in
an unfavorable portrayal of them or their profession. Despite evidence of herding behavior
among sell-side analysts in the literature, our respondents give other analysts’ earnings forecasts and stock recommendations low ratings in terms of their usefulness as inputs to their
own forecasts and recommendations. We cannot rule out the possibility that analysts biased
their responses downward to avoid appearing to rely heavily on other analysts.
14
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
analysts find other analysts’ earnings forecasts more useful than other analysts’ stock recommendations.14
3.2.2. How Often Do You Use the Following Valuation Models to Support Your
Stock Recommendations? (Table 2). Consistent with Bradshaw [2004], most
analysts state that they very frequently rely on price-earnings (P/E) or priceearnings-growth (PEG) models to support their stock recommendations.
Reliance on P/E or PEG models implies that analysts’ earnings forecasts
are a key factor in their valuation models, consistent with our result in table
10 that analysts’ most important motivation for issuing accurate earnings
forecasts is to use their forecasts as an input to their stock recommendations. We also find that most analysts frequently use cash flow models but
use the other five models much less frequently.
3.3 COMMUNICATION WITH MANAGEMENT
Although prior research examines the role of analysts’ communication
with company management (Chen and Matsumoto [2006], Ke and Yu
[2006], Soltes [2014]), several important questions remain unanswered,
such as the usefulness of private communication with management relative
to other inputs analysts employ, the frequency of analysts’ communication
with management, and the relative usefulness of different venues for contact with management. We asked analysts several questions to address these
issues.
3.3.1. How Often Do You Have Direct Contact with the CEO or CFO of the Typical Company You Cover? (Online Appendix). Among our responding analysts,
98.4% say they have direct contact with the CEO or CFO of the typical firm
they cover at least once a year, and 53.2% have direct contact at least five
times a year. Although our interviewees said Reg FD was a “game changer”
that profoundly affected the way management communicates with analysts,
several stated that managers are more accessible now than when Reg FD
was first implemented. One analyst described the changes from the pre–
Reg FD period to today as follows: “There was a lot of backroom chatter
before Reg FD. Now management has figured out how to ‘paper things up’
[with an 8-K]. So now we’re almost back to where we were pre–Reg FD, but
not quite because that backroom chatter is shut down. It’s just now it’s not
in the backroom; it’s everywhere.”
14 To determine whether we can reliably compare answers to twin questions in the EF and
SR versions of our survey, we test whether respondents to the two versions of the survey provide similar answers to the five common questions discussed earlier. The respondents to the EF
and SR versions of the survey provide virtually identical answers to the five common questions.
Specifically, for each of the 39 choices available in these questions, we compare the average
rating between the EF respondents and the SR respondents. Untabulated t-tests reveal no significant differences between the two groups at the 1% level, no significant differences between
the two groups at the 5% level, and only three significant differences between the two groups
at the 10% level. Establishing the similarity of these two groups of analysts enables us to reliably
compare answers to twin questions in the two versions of our survey.
Total possible N = 181
Price/earnings (P/E) or Price/earnings growth
(PEG) model
Cash flow model
Dividend discount model
A model based on earnings momentum or
earnings surprises
Economic value added (EVA) model
Residual income model
A model based on stock price and volume
patterns
1.34
1.14
0.67
4.37
1.76
1.53
4.42
Average
Rating
7
7
–
3–7
5–7
7
3–7
Significantly
Greater Than
7.73
4.97
2.76
60.22
12.22
9.44
61.33
Very Frequently
(5 or 6)
69.06
69.61
83.43
12.15
53.67
62.22
12.15
Very Infrequently
(0 or 1)
Column 1 reports the average rating, where higher values correspond to greater frequency. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating frequency
of 5 or 6 (0 or 1).
(5)
(6)
(7)
(2)
(3)
(4)
(1)
Responses
% of Respondents Who Answered
TABLE 2
Survey Responses to the Question: How Often Do You Use the Following Valuation Models to Support Your Stock Recommendations?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
15
16
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
Another analyst reported that buy-side clients believe the insights of sellside analysts are more valuable when analysts have direct contact with management: “Regardless of Reg FD, investors value analysts’ direct contacts
with management more than anything. As an analyst, if I call up a money
manager, a hedge fund, whoever, and I’ve got a call to make on a stock,
and I’m able to say, ‘Hey, by the way, we were able to spend 20–30 minutes
talking to senior management,’ boom! Their ears are just straight up.”
One analyst provided an interesting anecdote about the extent to which
some brokerage houses go in order to understand how to read cues from
management in the post–Reg FD environment: “We had an FBI profiler
come in, and all the analysts and portfolio managers spent four hours with
this profiler trying to understand how to read management teams, to tell
when they’re lying, to tell when they were uncomfortable with a question.
That’s how serious this whole issue has become.” Although the evidence in
section 3.4.2 suggests analysts do not focus on uncovering intentional misrepresentation in the financial statements, this interview anecdote is consistent with recent empirical research suggesting senior management’s vocal
cues can be used to assess firms’ future prospects (Mayew and Venkatachalam [2012]).
3.3.2. How Useful Are the Following Types of Direct Contact with Management
for the Purpose of Generating Your Earnings Forecasts/Stock Recommendations?
(Table 3). More than 66% (72%) of analysts report that private phone calls
are a very useful source of direct contact with management for the purpose of generating their earnings forecasts (stock recommendations), reinforcing our findings that analyst communication with management is
both frequent (section 3.3.1) and useful (section 3.2.1). Analysts say private
phone calls with management are at least as useful as other venues examined by recent research, including earnings conference calls, company investor day events, and conferences sponsored by brokerage houses (Green
et al. [2014], Kirk and Markov [2014], Mayew, Sharp, and Venkatachalam
[2013]).
Our cross-sectional evidence reveals that analysts for whom hedge funds
are an important client are more likely to indicate that private phone calls
with management are useful for their earnings forecasts.15 If private phone
calls with managers provide analysts with an information advantage, our
results suggest analysts catering to hedge funds are likely to make superior
earnings forecasts.
We used our interviews to inquire into the nature, timing, and content of
analysts’ private phone calls with management. Consistent with the results
of our survey, our interviewees reported having private phone calls with
15 Solomon and Soltes [2013] find that hedge funds are more likely than other investors to
benefit from private meetings with managers, which they attribute to hedge funds’ superior
ability to process the information disclosed in private meetings or to their having possession
of other information that makes the discussions in meetings especially valuable.
4.71
4.60
4.36
4.34
4.19
4.13
3.55
3.14
Average
Rating
3–8
3–8
7–8
7–8
7–8
7–8
8
–
Significantly
Greater Than
66.48
58.79
50.00
46.96
46.15
48.90
26.92
21.43
Very Useful
(5 or 6)
7.69
7.69
5.49
2.76
7.14
10.44
9.34
20.33
Not Useful
(0 or 1)
% of Respondents
Who Answered
Column 1 reports the average rating, where higher values correspond to greater likelihood. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating usefulness
of 5 or 6 (0 or 1).
(Continued)
Total possible N = 182
(1) Private phone calls with management
(2) The Q&A portion of earnings conference calls
(3) Company investor day events
(4) Management’s presentation on earnings conference calls
(5) Company or plant visits
(6) Road shows
(7) Industry conferences
(8) Conferences sponsored by your employer
Responses
Panel A: Summary statistics for the EF version
TABLE 3
Survey Responses to the Question: How Useful Are the Following Types of Direct Contact with Management for the
Purpose of Generating Your Earnings Forecasts (Stock Recommendations)?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
17
Total possible N = 181
Private phone calls with management
Company or plant visits
Road shows
Company investor day events
The Q&A portion of earnings conference calls
Industry conferences
Conferences sponsored by your employer
Management’s presentation on earnings conference calls
4.98
4.79
4.59
4.34
4.00
3.76
3.74
3.66
Average
Rating
3–8
4–8
5–8
5–8
8
–
–
–
Significantly
Greater Than
Very Useful
(5 or 6)
72.38
65.56
58.33
48.07
36.44
28.73
32.60
27.07
EF vs. SR
1.86†
3.89††
2.86†††
0.10
4.42∗∗∗
1.42
3.48†††
4.76∗∗∗
3.31
3.33
3.33
2.76
4.42
4.97
10.50
6.63
Not Useful
(0 or 1)
% of Respondents Who
Answered
Column 1 reports the average rating, where higher values correspond to greater usefulness. Column 2 reports the results of t-tests of the null hypothesis that the average rating
for a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the other items
at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 reports the results of a t-test of the null hypothesis that the average rating
is the same across both the earnings forecast and stock recommendation versions of the survey. ∗∗∗ , ∗∗ , and ∗ (††† ,†† , and † ) indicate that the average rating in the EF (SR) version of
the survey is significantly larger at the 1%, 5%, and 10% level, respectively. Column 4 (5) presents the percentage of respondents indicating usefulness of 5 or 6 (0 or 1).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Responses
Panel B: Summary statistics for the SR version
T A B L E 3—Continued
18
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
19
senior management—most often the CFO—at least quarterly.16 Many analysts said companies schedule analyst “call-backs” immediately after their
public earnings conference calls: one-on-one, private calls from the CFO,
who answers additional questions from individual analysts.
Several analysts discussed the importance of these follow-up calls. One
analyst suggested the order of calls is based on the analysts’ valuations of
the company: “Management will call the analysts who are at the low end
of their valuation, if they want the stock to move up. By the order in which
management calls analysts, they can move the consensus to where they want
it to be.”17
Another analyst explained the benefits of private calls as follows: “In private conversations with management, you get details that they’re not necessarily going to go into on a public call with investors. They might be more
willing to share that with us because we can then go to clients and say, ‘This
is our understanding of the situation. This is what the company says; this is
what we think.’ It’s a way for them to broadcast. We’re sort of like a megaphone for them.”
Another said, “We ask for qualitative thoughts and insights into industry
trends or specific business lines, just so that we’re also double-checking our
own thought processes and that our models are solid.” Consistent with empirical evidence (Mayew and Venkatachalam [2012], Hobson, Mayew, and
Venkatachalam [2012]), one analyst reported, “The CEO and CFO, you
can read their body language—even on the phone—and get a feel for how
optimistic they are or how realistic something might be. And it’s really that
kind of information you’re looking for—it’s not something specific that
they wouldn’t tell someone else.” This same analyst went on to say, “For
the calls around the earnings calls, a lot of management teams want to call
all the analysts and say, ‘Did you understand what happened? Do you have
any questions? Was anything confusing about the results themselves? Before you write your note, are you thinking badly about this? Can we maybe
talk with you about it so you don’t think so badly about it?’” Finally, another
analyst described the information discussed on the private calls as follows,
“It’s not nonpublic material information; it’s clarification of points. They
help you digest the information a little bit better.” Thus, our interviewees
suggested that the follow-up calls they receive from management after public earnings conference calls are a valuable source of information.
16 In contrast, Solomon and Soltes [2013] report that the investor relations officer and the
CEO of a single mid-cap company were more likely than the CFO to meet with institutional
investors in one-on-one meetings.
17 It is plausible that managers use a similar technique to walk down analysts’ earnings forecasts ( Richardson, Teoh, and Wysocki [2004], Libby et al. [2008]). In this scenario, a manager
seeking to lower the consensus forecast would first call the analyst with the highest earnings
forecast, pointing out, among other things, that every other analyst has a lower forecast. Following this initial call, the manager would then call the analyst with the next-highest forecast
and use a similar line of reasoning to encourage the analyst to lower his or her forecast, and
so on.
20
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
In spite of restrictions on selective disclosure enacted through Reg FD
in October 2000, our findings are consistent with a provision of Reg FD
that allows managers to disclose immaterial information to an analyst that
“helps the analyst complete a ‘mosaic’ of information that, taken together,
is material” (Securities and Exchange Commission [2000]). In other words,
information analysts obtain privately from management can become useful
within the context of other information the analyst already possesses. Thus,
while our findings do not constitute direct evidence of violations of Reg
FD, they do show that information conveyed in private conversations with
management is extremely valuable to sell-side analysts in the post–Reg FD
environment.18
Although academic research finds evidence consistent with the notion
that analysts who ask questions on earnings conference calls are either
highly favored by management (Mayew [2008]) or possess superior information about the firm ( Mayew, Sharp, and Venkatachalam [2013]), some
analysts told us they purposely avoid asking questions on public conference
calls. One analyst stated, “There are three things that can happen when
you ask a question on an earnings conference call: one, you sound like a
complete idiot; two, they give you no information at all; and three, you get
a really insightful answer except you’ve just shared it with all your competition. So I don’t ask questions on calls.”
A comparison of responses to these twin questions reveals that the Q&A
portion of earnings conference calls and management’s presentation on
earnings conference calls are more useful for generating earnings forecasts
than stock recommendations. In contrast, company or plant visits, road
shows, and conferences sponsored by their employers are more useful for
generating stock recommendations than earnings forecasts.
3.4 ASSESSMENTS OF FINANCIAL REPORTING QUALITY
Recent survey evidence sheds light on the perspective of CFOs regarding earnings quality (Dichev et al. [2013]). However, CFOs’ views on this
topic are likely influenced by financial reporting concerns. For example,
CFOs have incentives related to compensation, litigation risk, or the firm’s
stock price, which could create a preference for managed earnings and bias
their responses to questions about earnings quality (Dechow et al. [2010],
Nelson and Skinner [2013]). In contrast, analysts are an important source
of information for their investing clients (Brown et al. [2014]) and have
incentives to identify attributes of high-quality earnings, because incorrect
assessments of earnings quality could result in economic losses for their
clients and have an adverse effect on their own reputation and compensation. Thus, because analysts’ views on earnings quality are likely to be
more informative than those of financial statement preparers (Nelson and
18 This evidence is similar to what Solomon and Soltes [2013] report with respect to the
value of private meetings with management to institutional investors.
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
21
Skinner [2013]), we asked analysts for their views on various financial reporting issues.
3.4.1. How Important Are the Following to Your Assessment of Whether a Company’s “Quality” of Reported Earnings Is High? (Table 4). Analysts respond that
“high-quality” earnings are backed by operating cash flows (Sloan [1996]),
are sustainable and repeatable, reflect economic reality, and reflect consistent reporting choices over time. In contrast to the views of CFOs surveyed by Dichev et al. [2013] who rate avoidance of long-term estimates as
an important factor in assessing earnings quality (2nd of 12 choices), analysts rate it much lower (10th of our 12 choices). This finding underscores
Nelson and Skinner’s [2013] concern that CFOs’ preference for earnings
that are free of long-term estimates may reflect their bias toward earnings
that are easy to explain to external parties rather than representing the
views of users of accounting information.
The lowest rated responses are that earnings are less volatile than operating cash flows and that the company is audited by one of the Big 4. Analysts
view a Big 4 audit as relatively unimportant (12th of 12 choices), contrasting with research that indicates a Big 4 audit is associated with high-quality
earnings (Khurana and Raman [2004], Behn, Choi, and Kang [2008]).
However, analysts who primarily follow companies with Big 4 auditors may
not view a Big 4 auditor as a distinguishing feature.
Our cross-sectional evidence shows that, consistent with their training,
analysts with a bachelor’s degree in accounting are more likely to consider
a Big 4 audit a sign of high-quality earnings. II All-Stars, who receive votes
from buy-side analysts and portfolio managers for providing the best equity research, are less likely than other analysts to believe many of the constructs the literature associates with high-quality earnings (e.g., earnings
are backed by operating cash flows, are sustainable and repeatable, are less
volatile than operating cash flows, and are predictive of future cash flows
and earnings) are important.
3.4.2. To What Extent Do You Believe the Following Indicate Management Effort
to Intentionally Misrepresent the Financial Statements? (Table 5). We asked analysts about the extent to which they believe potential “red flags” of misreporting indicate management effort to intentionally misrepresent financial
statements. Financial statement users, such as analysts and investors, are
likely to have more informative views on this topic than financial statement
preparers because CFOs often have incentives to manage earnings and may
have biased views of the indicators of financial misrepresentation (Nelson
and Skinner [2013]). Although prior research suggests recent management
turnover, consistently meeting or beating earnings targets, management
wealth being closely tied to stock price, and recent auditor turnover are
signals of financial misrepresentation (e.g., Krishnan and Krishnan [1997],
Desai, Hogan, and Wilkins [2006], Efendi, Srivastava, and Swanson [2007],
Myers, Myers, and Skinner [2007]), these items received relatively low
4.67
4.46
4.44
4.42
4.29
4.05
3.85
3.78
3.63
3.21
2.90
2.62
Average Rating
5–12
6–12
6–12
6–12
7–12
9–12
10–12
10–12
10–12
11–12
–
–
64.29
56.04
57.69
56.04
49.45
46.70
38.25
36.07
31.69
24.58
16.67
15.38
2.20
3.85
3.20
3.30
3.85
10.99
9.29
7.10
9.84
15.64
24.44
29.12
Not Important
(0 or 1)
% of Respondents Who Answered
Very Important
(5 or 6)
Column 1 reports the average rating, where higher values correspond to greater importance. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating importance
of 5 or 6 (0 or 1).
Total possible N = 183
(1) Earnings are backed by operating cash flows
(2) Earnings are sustainable and repeatable
(3) Earnings reflect economic reality
(4) Earnings reflect consistent reporting choices over time
(5) Company managers have high integrity or moral character
(6) Earnings are free from one-time or special items
(7) Earnings can predict future cash flows
(8) Company has strong corporate governance
(9) Earnings can predict future earnings
(10) Earnings are not highly dependent on long-term estimates
(11) Earnings are less volatile than operating cash flows
(12) Company is audited by a Big 4 auditor
Responses
Significantly
Greater Than
TABLE 4
Survey Responses to the Question: How Important Are the Following to Your Assessment of Whether a Company’s “Quality” of Reported Earnings Is High?
22
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
Total possible N = 180
Company has weak corporate governance
Company has a material internal control weakness
Large or frequent one-time items or special items
Large gap between earnings and operating cash flows
Company recently restated earnings
Company consistently reports smooth earnings
Deviations from industry or peer norms
Management is overconfident and/or overly optimistic
Recent auditor turnover
Management wealth is closely tied to stock price
Company consistently meets or beats earnings targets
Recent management turnover
3.55
3.55
3.47
3.09
2.93
2.88
2.85
2.83
2.77
2.64
2.48
2.34
Average Rating
4–12
4–12
4–12
10–12
11–12
11–12
11–12
11–12
12
–
–
–
31.84
29.05
29.05
19.32
16.11
17.78
14.53
16.67
12.78
13.33
12.29
7.22
11.17
12.29
15.08
20.45
20.56
23.33
21.23
23.33
28.33
28.33
30.17
34.44
Do Not Believe
(0 or 1)
% of Respondents Who Answered
Strongly Believe
(5 or 6)
Column 1 reports the average rating, where higher values correspond to greater likelihood. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating likelihood
of 5 or 6 (0 or 1).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
Responses
Significantly
Greater Than
TABLE 5
Survey Responses to the Question: To What Extent Do You Believe the Following Indicate Management Effort to Intentionally Misrepresent the Financial Statements?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
23
24
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
ratings from analysts.19 Instead, analysts consider weak corporate governance, internal control deficiencies, and large one-time or special items
to be more indicative of financial reporting irregularities. When we compare analysts’ responses with the responses of CFOs of public companies
regarding red flags of misreporting (Dichev et al. [2013]), it is evident that
managers and analysts have widely divergent views.20
In follow-up interviews, we asked analysts directly about their attention to
“red flags” of potential misreporting. Most responded that they exert little
effort trying to determine whether firms misreport earnings. Prior research
provides evidence that sell-side analysts play a role in uncovering corporate
fraud (Dyck, Morse, and Zingales [2010]), but analysts say it is not their
job to look for earnings manipulation (Abarbanell and Lehavy [2003]). Financial misrepresentation is often difficult to detect, and analysts’ buy-side
clients value industry-level insights above all other services sell-side analysts
provide; therefore, sell-side analysts are unlikely to have incentives to try to
uncover firm-specific financial misrepresentation.
On the topic of intentional financial misrepresentation, one analyst said
he “takes the financial statements at face value,” because it is extremely difficult to uncover intentional misconduct. Another said, “It’s up to the auditor to catch that . . . If they were able to fool the auditor into a clean audit
opinion, I’m never going to be able to catch it just from the information
that’s in a Q or a K.” Another analyst said that, if a company has audited
financial statements, “It’s somebody else’s job to figure out if the information they’re giving us is correct. We have to take that on faith.” We note,
however, that our collective evidence does not imply that analysts ignore
more benign forms of earnings management (e.g., within-GAAP discretion
to manage earnings). Indeed, as discussed in section 3.4.1, analysts prefer earnings that are backed by operating cash flows, that are sustainable,
and that reflect economic reality, suggesting analysts could actually rein in
earnings management before it escalates into more egregious misrepresentations of the financial statements (Schrand and Zechman [2012]).
Our cross-sectional evidence provides additional evidence that analysts
are not a strong line of defense against financial reporting irregularities.
II All-Stars and analysts employed at large brokerage houses are less likely
than other analysts to be concerned with many common signs of financial
statement misrepresentation, suggesting uncovering intentional financial
misrepresentation is not a priority for even highly regarded analysts.
19 If analysts are complicit in the “numbers game” that results in companies consistently
meeting or beating earnings targets, they may be reluctant to respond that consistently meeting or beating earnings targets is a “red flag” of intentional financial misrepresentation.
20 For example, analysts rate material internal control weakness as an important red flag
of misreporting (2nd of 12 choices), but internal control weaknesses do not make the list of
20 types of red flags mentioned by CFOs in Dichev et al.’s table 14. Moreover, that the company consistently meets or beats earnings targets receives little support from analysts (11th of
12 choices) but strong support from CFOs as a red flag of misreporting (3rd of 20 choices).
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
25
3.4.3. How Likely Are You to Take the Following Actions if You Observe a “Red
Flag” of Management Effort to Intentionally Misrepresent the Financial Statements?
(Online Appendix). The two most common actions analysts take when observing a “red flag” of management effort to intentionally misrepresent financial statements are to seek additional information from management
and to seek additional information from nonmanagement sources. While
not nearly as prevalent as the first two actions, more than half of our surveyed analysts say they are very likely to revise their stock recommendations
and earnings forecasts downwards after observing a “red flag” of intentional
misrepresentation. The only action analysts say they are unlikely to take is
to cease covering the firm.
3.4.4. How Often Do You Exclude the Following Components of GAAP Earnings When Forecasting Street Earnings? (Table 6). A majority of analysts very
frequently exclude extraordinary items, discontinued items, restructuring
charges, and asset impairments when forecasting “street earnings,” but
most include amortization, changes in working capital, and depreciation
in these forecasts. These findings shed light on the earnings components
analysts include in their forecasts and are of interest given the importance
of “street” earnings as a determinant of stock prices (Bradshaw and Sloan
[2002]).
3.4.5. Do You Exclude Components of GAAP Earnings from Your Forecast of
“Street” Earnings for the Following Reasons? (Table 7). The primary reason analysts exclude components of GAAP earnings from their forecasts of “street”
earnings is their belief that the component is nonrecurring. In addition,
nearly half say they exclude components of GAAP earnings because of their
desire to improve earnings forecast accuracy.
3.5 DETERMINANTS OF ANALYSTS’ CAREER SUCCESS
In contrast to archival studies that must infer analysts’ incentives from
observed statistical associations, we asked analysts directly about the factors
that determine their compensation and the importance of various analyst
rankings for their career advancement.
3.5.1. How Important Are the Following to Your Compensation? (Table 8). II
surveys suggest institutional investors highly value sell-side analysts’ industry
knowledge, so it is reasonable for brokerage houses to compensate sell-side
analysts for the industry knowledge they provide to institutional investors,
their most important clients (see table 12). Indeed, sell-side analysts rate industry knowledge and their standing in analyst rankings or broker votes as
the most important determinants of their compensation.21 Broker votes are
a process whereby buy-side portfolio managers and buy-side analysts vote to
21 Our cross-sectional evidence reveals that experienced analysts are more likely to state
that industry knowledge is important to their compensation, in contrast to MBAs, who are less
likely to make this statement.
Total possible N = 183
Extraordinary items
Discontinued items
Restructuring charges
Asset impairments
Cumulative effect of accounting changes
Nonoperating items
Stock option expense
Amortization
Changes in working capital
Depreciation
4.81
4.60
4.34
4.17
3.67
3.63
2.35
1.90
1.41
1.28
Average Rating
3–10
4–10
5–10
5–10
7–10
7–10
8–10
10–11
–
–
71.04
63.74
57.69
55.74
41.11
39.78
25.41
17.78
12.78
11.80
4.92
9.34
8.79
13.11
17.78
18.78
48.07
56.67
66.67
70.79
Very Infrequently
(0 or 1)
% of Respondents Who Answered
Very Frequently
(5 or 6)
Column 1 reports the average rating, where higher values correspond to greater frequency. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating frequency
of 5 or 6 (0 or 1).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Responses
Significantly
Greater Than
TABLE 6
Survey Responses to the Question: How Often Do You Exclude the Following Components of GAAP Earnings When Forecasting “Street” Earnings?
26
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
2–5
4–5
–
–
–
3.41
3.27
3.09
Significantly
Greater Than
4.51
3.86
Average Rating
37.22
36.11
36.11
61.33
49.72
Very Frequently
(5 or 6)
22.22
24.44
31.11
7.18
14.92
Very Infrequently
(0 or 1)
Column 1 reports the average rating, where higher values correspond to greater frequency. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating frequency
of 5 or 6 (0 or 1).
Total possible N = 181
(1) Because you believe the component is “nonrecurring”
(2) Because you believe excluding the component improves your
earnings forecast accuracy
(3) Because you want to be consistent with management guidance
(4) Because you want to be consistent with other sell-side analysts
(5) Because you want to be consistent with communication from
I/B/E/S, First Call, Zacks, or S&P
Responses
% of Respondents Who Answered
TABLE 7
Survey Responses to the Question: Do You Exclude Components of GAAP Earnings from Your Forecast of “Street” Earnings for the Following Reasons?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
27
3–9
5–9
5–9
5–9
7–9
8–9
9
–
–
4.95
4.73
4.73
4.69
4.17
4.14
3.94
3.65
3.59
Average Rating
24.10
72.18
66.85
63.54
63.99
38.95
44.63
35.08
44.20
7.76
1.93
4.97
2.21
3.60
2.76
7.16
5.52
20.17
Not Important
(0 or 1)
% of Respondents Who Answered
Very Important
(5 or 6)
Column 1 reports the average rating, where higher values correspond to greater importance. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating importance
of 5 or 6 (0 or 1).
Total possible N = 363
(1) Your industry knowledge
(2) Your standing in analyst rankings or broker votes
(3) Your accessibility and/or responsiveness
(4) Your professional integrity
(5) Your written reports
(6) Your relationship with management of the companies you follow
(7) The profitability of your stock recommendations
(8) Your success at generating underwriting business or trading
commissions
(9) The accuracy and timeliness of your earnings forecasts
Responses
Significantly
Greater Than
TABLE 8
Survey Responses to the Question: How Important Are the Following to Your Compensation?
28
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
29
assess the value of research services from sell-side brokerage houses and to
determine how to allocate research commissions. Two-thirds of analysts indicate their standing in analyst rankings or broker votes is very important,
while fewer than 5% say it is not important to their compensation. Experienced analysts and analysts from large brokerage houses are more likely to
say their standing in analyst rankings or broker votes is important to their
compensation, consistent with evidence that these analysts are more likely
to become II All-Stars (Rees, Sharp, and Wong [2014b]).
Industry knowledge is potentially important to analysts’ compensation
for several reasons. First, providing buy-side analysts with industry knowledge helps sell-side analysts generate broker votes. Second, analysts who
are industry experts are more likely to develop investment banking relationships, which are important to their employers. Third, institutional
investors highly value sell-side analysts’ industry knowledge, suggesting brokerage houses likely reward industry experts in an effort to prevent them
from being hired away by competitors.
Although its average rating is relatively low, 44% of analysts say their success at generating underwriting business or trading commissions is very
important to their compensation. This result suggests conflicts of interest
remain a persistent issue for a substantial number of sell-side analysts.22 Finally, although the accuracy and timeliness of analysts’ earnings forecasts
and the profitability of their stock recommendations receive relatively low
average ratings, 35% and 24% of our respondents, respectively, say they are
very important determinants of their compensation. Retail-focused analysts
are more likely to say the accuracy and timeliness of their earnings forecasts are important to their compensation, suggesting that they are more
motivated to make accurate and timely earnings forecasts.
3.5.2. How Important Are the Following Analyst Rankings for Your Career Advancement? (Table 9). Although much of the prior literature on analyst rankings emphasizes the II All-America Research Team awards (Stickel [1992],
Cox and Kleiman [2000], Leone and Wu [2007], Rees, Sharp, and Twedt
[2014a]), analysts indicate that broker or client votes are significantly more
important to their career advancement than the II awards.23 More than
22 Jack Grubman (2013), the highest paid sell-side analyst on Wall Street before being permanently banned from the securities industry for simultaneously advising both firms and investors, recently suggested the analyst industry has changed in form but not in substance. As
an example, he says that, prior to the reforms of the past decade, an investment banker and
a research analyst would hold a single meeting with management in an attempt to secure the
firm’s underwriting business. Now, he says, there are two meetings instead of one—one meeting in which the investment banker meets with management to try to gain the firm’s underwriting business and another meeting in which the research analyst meets with management
and makes another pitch for the underwriting business.
23 In cross-sectional tests, we find that, although II All-Stars are more likely than other analysts to state that being an II All-Star is important for their career advancement, both II
All-Stars and non–II All-Stars consider broker votes to be equally important for their career
advancement.
5.13
3.28
2.48
2.32
1.48
Average Rating
2–5
3–5
5
5
–
82.74
37.29
15.15
10.74
3.02
7.12
28.45
35.26
37.19
59.89
Not Important
(0 or 1)
% of Respondents Who Answered
Very Important
(5 or 6)
Buy-side portfolio managers and buy-side analysts award broker or client votes to sell-side brokerages based on the value of the research the brokerages’ analysts provide. Column
1 reports the average rating, where higher values correspond to greater importance. Column 2 reports the results of t-tests of the null hypothesis that the average rating for a given
item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding items
at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating importance of 5
or 6 (0 or 1).
Total possible N = 365
(1) Broker or Client votes
(2) Institutional Investor’s All-American Research Team
(3) The Wall Street Journal’s Survey of Award Winning Analysts
(4) Star Mine Analyst Awards
(5) Zacks All-Star Analyst Ratings
Responses
Significantly
Greater Than
TABLE 9
Survey Responses to the Question: How Important Are the Following Analyst Rankings for Your Career Advancement?
30
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
31
twice as many analysts indicate that broker or client votes are important for
career advancement (83%) than say the same thing about II status (37%).
Researchers who seek to obtain more powerful tests of analyst rankings
should use broker or client votes in lieu of II awards, if they are able to
access the relevant data.
The analysts we interviewed explained why broker or client votes are
so important for their career advancement. Broker votes translate directly
into revenue from the sell-side analysts’ clients to their employers ( Maber,
Groysberg, and Healy [2014]), and several stated that their bonuses are directly affected by broker votes. One analyst stated, “The part to me that’s
shocking about the industry is that I came into the industry thinking [success] would be based on how well my stock picks do. But a lot of it ends
up being ‘What are your broker votes?’” Another analyst said, “Broker votes
have become very important in this business, not necessarily just to the analysts, but to the sales and trading part of the equation, too.” Another analyst
remarked, “Broker votes translate into revenue for my firm. They directly
impact my compensation and directly impact my firm’s compensation.” Going further, the analyst stated: “25% of the allocation of our bonus pool is
based on broker votes.” These comments highlight analysts’ incentives to
satisfy their investing clients (Firth et al. [2013]).
We also asked analysts about the benefits of being an II All-Star. One
analyst described it as “your external stamp of approval” and, consistent
with prior research, said that, because the II results are visible to outsiders,
“Your access to management teams is greatly increased by your II ranking” (Mayew [2008], Soltes [2014]). Another said, “The II rankings . . . give
you significant leverage within your own firm” because II-rated analysts can
easily find employment elsewhere. In summary, analysts indicate that, although broker votes are more important than II rankings for their career
advancement, both forms of recognition provide analysts with valuable benefits (Groysberg, Healy, and Maber [2011]).
3.6 INFLUENCES ON EARNINGS FORECASTS AND STOCK RECOMMENDATIONS
Although academic researchers and market participants focus heavily
on analysts’ earnings forecasts (Mikhail, Walther, and Willis [1999], Hong
and Kubik [2003], Call, Chen, and Tong [2009]) and stock recommendations ([Womack [1996], Francis and Soffer [1997], Bradshaw [2004]), relatively little is known about analysts’ motivation for issuing accurate earnings
forecasts and profitable stock recommendations. In addition to examining these issues, we consider the consequences to analysts who issue belowconsensus earnings forecasts and stock recommendations and the internal
pressures they face to alter their research outputs.
3.6.1. How Important Are the Following in Motivating You to Accurately
Forecast Earnings/Make Profitable Stock Recommendations? (Table 10). Consistent with research suggesting analysts’ stock recommendations are more
profitable when supported by accurate earnings forecasts (Loh and Mian
2–7
3–7
4–7
6–7
–
–
–
4.45
3.94
3.40
3.04
2.82
2.72
Significantly
Greater Than
4.77
Average
Rating
32.42
23.63
14.92
18.13
59.34
40.88
66.48
Very Important
(5 or 6)
17.03
21.43
22.10
28.02
6.04
8.84
3.30
Not Important
(0 or 1)
% of Respondents Who Answered
Column 1 reports the average rating, where higher values correspond to greater importance. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating importance
of 5 or 6 (0 or 1).
(Continued)
Total possible N = 182
(1) Your earnings forecast as an input to your stock
recommendationa
(2) Demand from your clients
(3) Your reputation with management of the companies you
follow
(4) Your standing in analyst rankings
(5) Your job security
(6) Your compensation
(7) Your job mobility
Responses
Panel A: Summary statistics for the EF version
TABLE 10
Survey Responses to the Question: How Important Are the Following in Motivating You to Accurately Forecast Earnings (Make Profitable Stock Recommendations)?
32
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
Total possible N = 181
Demand from your clients
Your standing in analyst rankings
Your compensation
Your job security
Your reputation with management of the
companies you follow
Your job mobility
Your stock recommendation as an input
to your earnings forecasta
3.29
2.99
4.34
3.92
3.78
3.65
3.44
Average
Rating
–
–
2–7
5–7
5–7
6–7
7
Significantly
Greater Than
53.04
47.51
43.33
39.23
29.44
30.39
25.14
2.94†††
10.32∗∗∗
Very Important
(5 or 6)
0.69
2.71†††
5.11†††
3.25†††
2.93∗∗∗
EF vs. SR
19.89
23.46
8.84
13.81
17.22
17.68
13.89
Not Important
(0 or 1)
% of Respondents Who Answered
The wording of these responses is different across the two versions of the survey because one version refers to earnings forecasts (panel A) and the other version refers to stock
recommendations (panel B).
Column 1 reports the average rating, where higher values correspond to greater importance. Column 2 reports the results of t-tests of the null hypothesis that the average rating
for a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the other items
at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 reports the results of a t-test of the null hypothesis that the average rating
is the same across both the earnings forecast and stock recommendation versions of the survey. ∗∗∗ , ∗∗ , and ∗ (††† ,†† , and † ) indicate that the average rating in the EF (SR) version of
the survey is significantly larger at the 1%, 5%, and 10% level, respectively. Column 4 (5) presents the percentage of respondents indicating usefulness of 5 or 6 (0 or 1).
a
(6)
(7)
(1)
(2)
(3)
(4)
(5)
Responses
Panel B: Summary statistics for the SR version
T A B L E 1 0—Continued
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
33
34
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
[2006], Ertimur, Sunder, and Sunder [2007]), our surveyed analysts say
their single most important motivation for issuing accurate earnings forecasts
is for use as inputs to their stock recommendations. Female analysts are
more likely to be motivated to issue accurate earnings forecasts for this
purpose, consistent with evidence that they issue more accurate earnings
forecasts than male analysts do (Kumar [2010]).
Demand from their clients is analysts’ most important motivation for
making profitable stock recommendations and their second most important motivation for issuing accurate earnings forecasts. Analysts’ concerns
about their standings in analyst rankings, compensation, job security, and
job mobility are more important for motivating them to make profitable
stock recommendations than to make accurate earnings forecasts.
3.6.2. How Likely Are the Following Consequences to You of Issuing Earnings
Forecasts/Stock Recommendations That Are Well Below the Consensus? (Table 11).
Collectively, the seven choices for the EF version of the survey received the
lowest ratings of any question in our survey. The only response where more
analysts believe the outcome is very likely than believe it is very unlikely is an
increase in investing clients’ perception of the analyst’s credibility (the only
favorable consequence we presented to the analysts). In contrast, far fewer
analysts say the loss of access to management is very likely than say it is
very unlikely. Our cross-sectional analyses suggest CFAs are less concerned
about many negative repercussions of issuing earnings forecasts well below
the consensus, suggesting they may be more likely to make bold, pessimistic
forecasts.
In the SR version of the survey, “an increase in your investing clients’
perception of your credibility” and “loss of access to management” are the
two most likely consequences of issuing below-consensus stock recommendations. The fact that analysts perceive that the issuance of below-consensus
stock recommendations improves their standing with investing clients underscores analysts’ need to please not only the management of the companies they cover but also their investing clients.24
A comparison of responses between the two versions of the survey indicates that analysts believe issuing a below-consensus stock recommendation is more likely to lead to a loss of access to management than
is issuing a below-consensus earnings forecast, possibly because issuing
below-consensus earnings forecasts makes it easier for management to report a positive earnings surprise (Brown [2001], Richardson, Teoh, and
Wysocki [2004], Graham, Harvey, and Rajgopal [2005], Ke and Yu [2006],
Libby et al. [2008]). Our cross-sectional results reveal that female analysts
are less concerned about lower bonus/compensation if they issue stock
24 The upward bias in analysts’ stock recommendations (Womack [1996], Barber et al.
[2001], Chen and Matusmoto [2006], Mayew [2008]) is consistent with analysts’ perception
that a loss of access to management is a potential consequence of issuing below-consensus
stock recommendations.
2–7
3–7
6–7
6–7
6–7
–
–
2.53
2.21
1.94
1.92
0.76
0.74
Significantly
Greater Than
3.16
Average
Rating
1.63
1.09
7.61
6.01
16.48
13.59
21.43
Very Likely
(5 or 6)
77.72
78.80
47.28
43.17
32.97
43.48
18.13
Very Unlikely
(0 or 1)
% of Respondents Who Answered
Column 1 reports the average rating, where higher values correspond to greater likelihood. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating likelihood
of 5 or 6 (0 or 1).
(Continued)
Total possible N = 184
(1) An increase in your investing clients’ perception of your
credibility
(2) Loss of access to management
(3) Being “frozen out” of the Q&A portion of future conference
calls
(4) Damage to your employer’s business relationship with
buy-side clients who hold stock in the firm
(5) Damage to your employer’s business relationship with the
company
(6) Promotion less likely
(7) Lower bonus/compensation
Responses
Panel A: Summary statistics for the EF version
TABLE 11
Survey Responses to the Question: How Likely Are the Following Consequences to You of Issuing an Earnings Forecast (Stock Recommendation) that Is Well Below the Consensus?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
35
3–7
3–7
5–7
6–7
6–7
–
–
3.24
2.62
2.35
2.26
1.04
0.97
Significantly
Greater Than
3.55
Average
Rating
6.67
2.78
0.56
1.99††
2.26††
1.73†
24.44
12.78
3.95†††
4.00†††
15.00
26.55
2.38††
0.71
Very Likely
(5 or 6)
EF vs. SR
68.89
72.78
32.78
40.56
17.78
26.67
9.04
Very Unlikely
(0 or 1)
% of Respondents Who Answered
Column 1 reports the average rating, where higher values correspond to greater likelihood. Column 2 reports the results of t-tests of the null hypothesis that the average rating
for a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the other items
at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 reports the results of a t-test of the null hypothesis that the average rating
is the same across both the earnings forecast and stock recommendation versions of the survey. ∗∗∗ , ∗∗ , and ∗ (††† ,†† , and † ) indicate the average rating in the EF (SR) version of the
survey is significantly larger at the 1%, 5%, and 10% level, respectively. Column 4 (5) presents the percentage of respondents indicating usefulness of 5 or 6 (0 or 1).
Total possible N = 180
(1) An increase in your investing clients’ perception of
your credibility
(2) Loss of access to management
(3) Damage to your employer’s business relationship
with the company
(4) Being “frozen out” of the Q&A portion of future
conference calls
(5) Damage to your employer’s business relationship
with buy-side clients who hold stock in the firm
(6) Lower bonus/compensation
(7) Promotion less likely
Responses
Panel B: Summary statistics for the SR version
T A B L E 1 1—Continued
36
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
37
recommendations below the consensus. In addition, analysts with a hedge
fund focus are less likely to believe issuing stock recommendations well below the consensus will result in lower compensation or damage to their
employer’s relationship with buy-side clients, perhaps due to their clients’
unique ability to execute short positions and profit from analysts’ negative
ratings.25
Several analysts said a good relationship with management is critical to
succeed as a sell-side analyst (Francis and Philbrick [1993]). One interviewee described an experience where company management canceled an
already-scheduled road show with the analyst immediately after the analyst
lowered his stock recommendation for the company. Another responded,
“If I’ve got a sell rating on a stock, is that company really going to want
to come attend a conference we’re hosting? Is that company really going
to give me three days to go market with them in New York? No, they’re
not. So you have to factor that in.” One analyst stated, “When a company
cuts you off, not only do you lose the information value of that [access],
but you actually lose revenue. The company won’t come to your conference; therefore, your conference is going to be less important. Clients pay
a boatload for that access.” Another candidly told us, “Most of the sell-side is
worried more about what management thinks of them than they are about
whether they’re doing a good job for investors.” Finally, one analyst said,
“It’s a needle you have to thread sometimes, between being intellectually
honest yet not offensive. It’s always in the back of your mind, because one
of the biggest things the buy-side compensates sell-side research firms for
is corporate access: road shows, meetings, access to management teams. So
you obviously want to keep an amicable relationship with the companies
that you follow.”
Our findings highlight an important conflict in sell-side research.
Whereas issuing earnings forecasts and stock recommendations that are
well below the consensus increases analysts’ credibility with investing
clients, it can also damage analysts’ relationships with managers of the firms
they follow.
3.6.3. How Often Does Research Management Pressure You to Issue an Earnings Forecast That Is Lower Than (Exceeds) What Your Own Research Would
Support? (EF Version) How Often Does Research Management Pressure You to Issue a Stock Recommendation Vhat Is Less Favorable (More Favorable) than What
Your Own Research Would Support? (SR Version) (Online Appendix). Pressure
related to issuing earnings forecasts or stock recommendations is not
pervasive within analysts’ own firms. The vast majority of analysts have
never experienced pressure from research management to alter their earnings forecasts or stock recommendations. Consistent with the positive bias
in analysts’ recommendations (Barber et al. [2006]), we find research
25 Cross-sectional tests also reveal that II All-Stars are less likely to miss a promotion due to
a stock recommendation that is well below the consensus.
38
L. D. BROWN, A. C. CALL, M. B. CLEMENT, AND N. Y. SHARP
management is more likely to pressure analysts to raise rather than to lower
their stock recommendations.26
In response to questions about why research management pressures sellside analysts, one interviewee explained: “Something like two-thirds of our
clients are long-only shops. So even if you have a sell, the best the client
can do is either own less of it or just not own it. They can’t do much with
a sell rating; unless they’re a hedge fund, they can’t profit directly from it.”
Another analyst put it simply: “There are lots of constituencies that analysts
have to answer to, and none of them likes an under-perform.”
Consistent with the literature suggesting analyst impartiality is influenced
by investment banking relationships or trading incentives of the firm at
which the analyst is employed (Lin and McNichols [1998], Michaely and
Womack [1999], Lin, McNichols, and O’Brien [2005], Cowen, Groysberg,
and Healy [2006], Ljungqvist, Marston, and Wilhelm [2006]), one interviewee said, “Equity analysts . . . are very, very reluctant—even after the Spitzer
rules—to upset the investment bankers, because the investment bankers
bring in so much more profitability . . . They certainly realize that the success of their company is tied to the performance of this much highermargin business than the business that they’re part of.”
3.7 OTHER INCENTIVES
Finally, we explore two other incentives that shape sell-side research: the
importance of various investing clients to analysts’ employers and analysts’
motivation to initiate coverage of a firm.
3.7.1. How Important Are the Following Clients to Your Employer? (Table 12).
Hedge funds and mutual funds are the two most important clients to analysts’ employers, and retail brokerage clients are the least important. These
responses suggest that most analysts focus on addressing the needs of large,
institutional investors, rather than the needs of small, individual investors
(De Franco, Lu, and Vasvari [2007]).
3.7.2. How Important Are the Following in Your Decision to Cover a Given
Company? (Table 13). Table 13 reveals that client demand for information
about the company is the most important determinant of analysts’ coverage decisions, with less than 1% of analysts saying this factor is not important to their coverage decision. Earnings predictability is among the
least important determinants. Although prior archival research suggests
disclosure quality (Lang and Lundholm [1996]) and company profitability
(McNichols and O’Brien [1997]) are important factors in analysts’ coverage decisions, these items receive relatively low ratings from our respondents. Our findings suggest analyst coverage is largely driven by a desire to
satisfy client demand, with relatively little consideration given to financial
26 A t-test indicates that research management exerts more downward pressure on earnings
forecasts than on stock recommendations.
Total possible N = 362
Hedge funds
Mutual funds
Defined-benefit pension funds
Insurance firms
Endowments and foundations
High net-worth individuals
Retail brokerage clients
5.26
5.24
3.61
3.31
2.96
2.41
1.89
3–7
3–7
4–7
5–7
6–7
7
–
Significantly
Greater Than
81.49
80.11
36.84
29.89
22.22
18.23
13.30
2.21
1.66
16.62
20.67
26.39
41.61
51.52
Not Important
(0 or 1)
% of Respondents Who Answered
Very Important
(5 or 6)
Column 1 reports the average rating, where higher values correspond to greater importance. Column 2 reports the results of t-tests of the null hypothesis that the average rating for
a given item is not different from the average rating of the other items. We report the rows for which the average rating significantly exceeds the average rating of the corresponding
items at the 5% level, and use Bonferroni-Holm–adjusted p-values to correct for multiple comparisons. Column 3 (4) presents the percentage of respondents indicating importance
of 5 or 6 (0 or 1).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Responses
Average
Rating
TABLE 12
Survey Responses to the Question: How Important Are the Following Clients to Your Employer?
INSIDE THE “BLACK BOX” OF SELL-SIDE FINANCIAL ANALYSTS
39
Total possible N = 365
Client demand for information about the company
The similarity of the company to other companies you
follow
The stock’s trading volume
The stock’s market capitalization
The company’s growth prospects
The composition of the company’s investor base
The company’s disclosures
The company’s corporate governance
The company’s profitability
The company’s investment banking relationship with
your employer
Other sell-side analysts cover the company
The predictability of the company’s earnings
2.54
2.43
4.16
4.05
3.98
3.32
2.90…