Underlying risk preferences and analyst risk‐taking behavior

• Published date: 01 July 2020
• Author: Joshua Shemesh,Sean Cleary,Jonathan Jona,Gladys Lee
• DOI: http://doi.org/10.1111/jbfa.12441
• Date: 01 July 2020

DOI: 10.1111/jbfa.12441

Underlying risk preferences and analyst

risk-taking behavior

Sean Cleary1Jonathan Jona2Gladys Lee3Joshua Shemesh4

1Department of Accounting, The Universityof

Melbourne

2FreemanBusiness School, Tulane University

3Department of Accounting, The Universityof

Melbourne

4Department of Banking and Finance, Monash

Business School

Correspondence

JonathanJona, Freeman Business School, Tulane

University,7 McAlister Dr,Ste 509, New Orleans

70118,USA.

Email:jjona@tulane.edu

Fundinginformation

SeanCleary acknowledges the financial support

providedby the University of Melbourne through

theKinsman Scholarship.

Thedata sets used and/or analyzed during the

currentstudy are available from the correspond-

ingauthor on reasonable request.

Abstract

We investigate the relationship between underlying risk prefer-

ences on analysts’ work-related decisions. Specifically, we exam-

ine whether facial width-to-height ratio (fWHR), an innate personal

characteristic that has been linkedto financial risk tolerance, is asso-

ciated with analysts’ stock coverage decisions and the boldness of

their earnings forecasts and stock recommendations. We find that

high-fWHR analysts cover firms with lower earnings predictabil-

ity, and issue bolder forecasts and recommendations. Our findings

shed new light on the black box of analyst decision making, assisting

investment practitioners in evaluating the information content pro-

duced by different types of analysts and understanding the observed

dispersion in analyst forecasts.

KEYWORDS

analysts, analysts forecasts, coverage decisions, facial width-to-

height ratio, forecast boldness, fWHR, innate characteristics, per-

sonal characteristics, risk-taking behavior, risk tolerance, stock

recommendations

JEL CLASSIFICATION

G24, J44

1INTRODUCTION

Most research regarding the relation of analyst characteristics to output has examined skills or variables exogenous

to the analyst, such as brokerage size. Despite extensive research, Bradshaw(2011) and Ramnath, Rock, and Shane

(2008) conclude that much is still unknown and that analyst decision making remains a black box.1,2 We attempt to

1Luoand Xie (2012) support their concerns that after controlling for commonly used forecast accuracy determinants such as general and firm experience, task

complexity,analyst following, the forecast horizon and broker resources, much is still unknown about how analysts form their forecasts. Specifically, without

analystfixed effects, the explanatory power of their regression model is 12.17%, while including analyst fixed effects increases it to 26.11% (Luo & Xie, 2012).

Theirfindings suggest that despite efforts to identify analyst characteristics that affect forecasts, much is as yet unexplained and the black box is still opaque.

2In a more recent study,Brown, Call, Clement, and Sharp (2015) address the ‘Black-Box’ problem using surveys and interviews with sell-side analysts. They

focuson a broad range of factors affecting analysts’ decisions but do not cover innate characteristics that have been found to play an important role in individ-

ualsin financial markets.

J Bus Fin Acc. 2020;47:949–981. wileyonlinelibrary.com/journal/jbfa c

2020 John Wiley & Sons Ltd 949

950 CLEARY ET AL.

shed light into this black box by investigating the relationship between analysts’ facial masculinity,a pre-determined

characteristic used as a proxy for testosterone levelsassociated with financial risk preferences and professional deci-

sions (Apicella et al., 2008; Kamiya, Kim, & Suh, 2016). Specifically,we examine two observable decisions that are likely

to be affected (even if unconsciously so)by an analyst’s innate risk tolerance. First, we investigate whether facial mas-

culinity,as captured by an analyst’s facial width-to-height ratio (fWHR), is associated with the decision to cover riskier

firms and more specifically, firms with less predictable earnings. Second, we investigate whether analysts’ fWHR is

associated with forecast boldness – defined as a deviation from a consensus earnings forecast – and with bold recom-

mendations, defined as strong buy/sellratings. To the best of our knowledge, the current study is the first to examine

the relationship between physical traits of top US analysts coveringU.S. firms.

Researchers investigate coverage choice and boldness because they are key components of analysts’ decisions

(Bradshaw,2011; Ramnath et al., 2008), particularly in regards to risk tolerance (Kadous, Mercer, & Thayer,2009). The

effects of these decisions are critical to firms, peer analysts and investors (Clement & Tse,2005; Yu, 2008). Coverage

choice is an essential decision, as it provides oversight and facilitates information distribution, thereby affecting the

corporate production of information and a company’s external financing selections (Chang,Dasgupta, & Hilary, 2006;

Yu, 2008).3We examinethe relationship between earnings predictability and the likelihood that a firm is covered by

a high, relative to low, fWHR analyst. When an analyst covers firms with less-predictable earnings, the likelihood of

inaccurate forecasts is increased, which could result in negative career consequences (Das, Levine, & Sivaramakrish-

nan, 1998). We expect that a high-fWHR analyst compared to a low-fWHR analyst is more likely to coverfirms with

low-earnings predictability.

Bold forecasts are commonly defined as forecasts that deviate from the consensus (Clement & Tse, 2005). Such

forecasts are generally considered risky, as more negative consequences exist for an inaccuratebold forecast com-

pared to an inaccurate non-bold forecast (Kadous et al., 2009). We expect a positiverelationship between fWHR and

the likelihood of issuing a bold forecast.

With the benefit of hindsight, we then ask whether high-fWHR analysts, that we find to be bolder in less predictable

firms,can produce forecasts that eventually turn out to be more accurate. The answer to this question is ex- ante unclear.

While risk has clear implications for coverage and boldness decisions, increased risk taking may translateinto an ana-

lyst’s higher or lower accuracy.Todemonstrate, consider the extreme case of an analyst that only trusts a private signal

and ignores the informativeness of the aggregated consensus. In such a case, a bold forecast is more likely to turn out

wrong. Our results suggest that the tendency of high-fWHR analysts to go against the herd does not improve their

accuracy.That said, fWHR has limited value in identifying analysts that will consistently generate more accurate fore-

casts. We show, for example, that high-fWHR analysts do not always display risk-taking behavior. Furthermore, our

results regarding accuracy do not providecredible real-time value to market participants in evaluating individual fore-

casts as they are announced. Accuracy can only be measured ex-post once realized earnings are announced, typically

one year after the forecast decision. We thus argue that market participants would benefit more from the ability to

evaluate analyst decisions in real time. As such, we focus our hypotheses on coverage choice and boldness, both of

which are observable at the time of forecast announcements.

Our sample consists of analysts honored by StarMine in their Top Stock Picker database.4We focus on star

analysts because they have more discretion in coverage choices and can afford to make bolder forecasts (Clarke,

Khorana, Patel, & Rau, 2007; Fang & Yasuda, 2009). Our sample consists of 465 star analysts nominated in 68

industries for the calendar years 2010–2015. Our first hypothesis examines whether high-fWHR analysts cover

firms with less-predictable earnings. Using 26,818 analyst-firm-year observations, we find support for this prediction.

Our second hypothesis examines whether high-fWHR analysts issue bolder forecasts. Using a sample of 20,128

3For example, Yu(2008) finds that firms covered by more analysts have less discretionary accruals. Easley and O’hara (2004) find that firms with greater

analyst coveragetend to have greater abnormal returns and have a lower cost of capital, as investor uncertainty is decreased. Analyst coverage also affects

externalfinancing decisions since firms with lower analyst coverage havehigher leverage, and issue equity less often, but in greater amounts (Chang, Dasgupta,

&Hilary, 2006).

4ThomsonReuters produces the StarMine rankings of the top 3 analysts across 59 industries based on I/B/E/S data (Thomson Reuters, 2015).

CLEARY ET AL.951

analyst forecasts, we find evidence consistent with this hypothesis. Our results suggest that fWHR, a static facial cue

associated with one’s propensity for aggressive behavior, is an important factor that contributes to analyst decision

making. We suggest that heterogeneity in analysts’ risk preferences helps to explainthe properties of their forecasts.

Because risk preferences may be correlated with the likelihoodof becoming a star, we apply a two-stage methodol-

ogy in which we estimate this likelihood in the first stage, and then control for it in the second stage. Our results show

that implications of our study are not limited to elite (star) analysts and could be extended to the general analyst com-

munity. In particular,we argue that fWHR could be used to uncover analyst characteristics regardless of star status,

which makes it relevantfor investors, brokerage houses and prospective analyst careers in general.

We also examine the behavior of up-and-coming stars. fWHR has been found to have a lower impact on behavior

for high, compared to low, status individuals (Goetz et al., 2013; Welker,Goetz, & Carré, 2015). In the case of analysts

selected as stars at some point in their career,we proxy for status by whether they are currently reigning. Consistent

with prior studies, we find non-linearity in the relationship between fWHR and risk taking, which stems from variation

in the subject’s status.

Our study makes the following contributions. First, it provides further evidence for the importance of innate char-

acteristics in evaluating information content generated by analysts and advances our understanding of observed dis-

persion in analysts’ forecasts. Our findings help researchers and investors to better understand analyst forecasts and

coverage decisions. In particular, our findings help investmentmanagers to better weigh forecasts in making invest-

ment decisions. Importantly, our findings allow market participants to form a more informed evaluation of analysts’

releases at an early stage in their career (e.g., rookie analysts) when they still lack a trackrecord.

While prior literature has focused on acquired traitssuch as experience, or factors exogenous to the analyst such as

broker resources, we contribute to the literature by showing that innate personal traits are associated with decision

making. Recent literature has started to recognize the importance of innate analyst characteristics in explaining the

variation in forecast accuracy (Barron, Byard, & Liang, 2013; Luo & Xie, 2012). While such studies largely focus on

gender effects (Green, Jegadeesh, & Tang,2009; Kumar, 2010; Li, Sullivan, Xu, & Gao, 2013b), we put forth an innate

characteristic that explains heterogeneity within gender,which is noteworthy given that the vast majority of analysts

are male. Our findings are also useful for financial analysts by increasing their awareness of – and reducing the perils

of – their own behavioral biases (De Bondt & Thaler,1990; Mokoaleli-Mokoteli, Taffler, & Agarwal, 2009).

Our findings also contribute to emerging interdisciplinary approaches and complement many studies in both

the biology and accounting literature that find high-fWHR individuals make riskier choices in financial markets.

For example, Cesarini, Johannesson, Lichtenstein, Sandewall, and Wallace (2010) find that approximately 25% of

individual variation in financial decision making is due to genetic variation. Prior literature on innate characteristics

has also focused on CEOs (see for example Dikolli, Mayew,& Steffen, 2012; Graham, Harvey, & Puri, 2013; Jia, Lent,

& Zeng, 2014), and money managers (Brown, Lu, Ray, & Teo, 2018; Lu & Teo,2018). The current study is one of the

first to extend this literature by focusing on analysts. With regard to analysts’ innate characteristics, a concurrent

paper by He, Yin, Zeng, Zhang, and Zhao (2019) uses data from China to test whether fWHR is associated with analyst

accuracy.5Heet al. (2019) consider fWHR as a proxy for status seeking and achievement drive and argue that fWHR is

thus correlated with an analyst’s effort and employer-allocated resources. They report, for instance, that high-fWHR

analysts are more likely to conduct corporate site visits, consistent with other findings relating fWHR with access to

better resources (Brunell et al., 2008).6The authors also report that analysts with higher fWHR are more likely to be

selected as star analysts, albeit using a Chinese award, which is quite different from StarMine. We complement their

findings in showing that fWHR affects the most influential analysts covering the largest US firms. Our findings hold

5Chan, Wang,and Wang (2016) seem to have similar findings. They report that analysts with greater fWHR are more likely to issue a ‘lowball’ forecast right

before earnings announcements. The marketreacts less to the forecast revisions they issue, particularly when the date of a forecast is close to the date of an

earningsannouncement.

6Relatedly, Cao, Guan, Li, and Yang(2019) find that attractive analysts attain better job performance through privileged access to information from firm

management.