Face Value: Trait Impressions, Performance Characteristics, and Market Outcomes for Financial Analysts

• Published date: 01 May 2022
• Author: LIN PENG,SIEW HONG TEOH,YAKUN WANG,JIAWEN YAN
• Date: 01 May 2022
• DOI: http://doi.org/10.1111/1475-679X.12428

DOI: 10.1111/1475-679X.12428

Journal of Accounting Research

Vol. 60 No. 2 May 2022

Printed in U.S.A.

Face Value: Trait Impressions,

Performance Characteristics, and

Market Outcomes for Financial

Analysts

LIN PENG ,∗SIEW HONG TEOH ,†YAKUN WANG ,‡

AND JIAWEN YAN §

Received 1 December 2020; accepted 2 February 2022

∗Zicklin School of Business, Baruch College, City University of New York, and Faculty of

Economics and Darwin College, University of Cambridge; †Anderson School of Management,

University of California, Los Angeles; ‡School of Management and Economics, The Chinese

University of Hong Kong, Shenzhen, and Shenzhen Finance Institute; §SC Johnson College of

Business, Cornell University

Accepted by Douglas Skinner. The paper was previously circulated under the title “Face

Value: TraitInference, Per formance Characteristics, and Market Outcomes for Financial An-

alysts.” We thank an anonymous reviewer, Gauri Bhat, Elizabeth Blankespoor, Donal Byard,

Ed deHaan, Hemang Desai, Phil Dybvig, Rich Frankel, Luzi Hail, Chad Ham, Doug Hanna,

David Hirshleifer, Mingyi Hung, Narasimhan Jegadeesh, Da Ke (discussant), Lisa Koonce,

Mark Lang, Christian Leuz, Edward X. Li, Ningzhong Li, Xinlei Li, Chuchu Liang, Bran-

don Lock, Ben Lourie, Mark Maffett, Xiumin Martin, Arthur Morris, Jing Pan, Devin Shan-

thikumar,Terry Shevlin, Derrell Stice, Alex Todorov,Sorabh Tomar, Marcel Tuijn,Sean Wang,

George Yang, Hayoung Yoon, Huai Zhang, Bohui Zhang, Dexin Zhou, Chenqi Zhu, and con-

ference and seminar participants at the Chicago Quantitative Alliance Annual Conference,

Baruch College, Hong Kong University, Hong Kong University of Science and Technology,

Tsinghua University, The Chinese University of Hong Kong, Shenzhen, Southern Methodist

University, the University of California Irvine, the University of Oregon, Washington Univer-

sity in St. Louis, the University of Zurich, the CICF Conference, and the Midwest Finance

Association Conference for helpful feedback and suggestions. We further thank participants

from the Journal of Accounting Research conference for many helpful comments. Lin Peng ac-

knowledges the Krell research grant and the Keynes Fund for financial support. An online ap-

pendix to this paper can be downloaded at http://research.chicagobooth.edu/arc/journal-

of-accounting-research/online- supplements.

653

© 2022 The Chookaszian Accounting Research Center at the University of Chicago Booth School of

Business

This is an open access article under the terms of the Creative Commons

Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium,

provided the original work is properly cited, the use is non-commercial and no modifications or

adaptations are made.

654 l. peng, s. h. teoh, y. wang, and j. yan

ABSTRACT

Using machine learning–based algorithms, we measure key impressions

about sell-side analysts using their LinkedIn photos. We find that impressions

of analysts’ trustworthiness (TRUST) and dominance (DOM) are positively as-

sociated with forecast accuracy, especially after recent in-person meetings be-

tween analysts and firm managers. High TRUST also enhances stock return

sensitivity to forecast revisions, especially for stocks with high institutional

ownership. In contrast, the impression of analysts’ attractiveness (ATTRACT)

is only positively associated with accuracy for new analysts or when a firm has a

new CEO or CFO. Furthermore, while high DOM helps male analysts’ chances

of attaining All-Star status, it reduces female analysts’ accuracy and the likeli-

hood of winning the All-Star award. In addition, the relation between TRUST

and accuracy is modulated by the disclosure environment and is attenuated

by Regulation Fair Disclosure. Our results suggest that face impressions in-

fluence analysts’ access to information and the perceived credibility of their

reports.

JEL codes: D83, G14, G24, G28, G41, J16, M41, M48

Keywords: machine learning; facial recognition; trait impressions; analysts;

gender discrimination; EPS forecasts; All-Star Analysts; forecast revision; so-

cial interactions

1. Introduction

Humans form first impressions about other people from their faces

spontaneously within milliseconds.1We refer to these judgments as face

impressions. Face impressions have powerful effects on visual attention,

trait inferences, social judgments, and social interactions (Hugenberg and

Wilson [2013], Todorov [2017]). For example, people judged as more at-

tractive or trustworthy have more positive legal, political, and labor market

outcomes.2There is also growing evidence that face impressions have capi-

tal market consequences.3

The evidence that face impressions have consequential outcomes mo-

tivates our study of face impressions of sell-side analysts. Information

1See, for example, Asch [1946], Hassin and Trope [2000], Willis and Todorov [2006], Bar,

Neta, and Linz [2006], and Todorov [2017].

2Attractive people win more legal cases (Zebrowitz and McDonald [1991]), earn a beauty

premium (Mobius and Rosenblat [2006]), have higher teaching ratings (Hamermesh and

Parker [2005]), and are more successful in online dating (Finkel et al. [2012]). Trustworthy-

looking people are more likely to win elections (Todorov et al. [2005]) and move up the

corporate ladder (Linke, Saribay, and Kleisner [2016]).

3Examples include studies of borrower trustworthiness on loan terms (Duarte, Siegel, and

Young [2012]), CEO competence on compensation (Graham, Harvey, and Puri [2017]), an-

alyst beauty on forecast accuracy (Cao et al. [2020], Li et al. [2020]), and auditor trustwor-

thiness on auditor tenure and audit fees (Hsieh et al. [2020]). Another set of studies ex-

tracts trait impressions from body language using videos to examine entrepreneurial funding

(Blankespoor, Hendricks, and Miller [2017], Huang et al. [2020]; Hu and Ma [2020]).

face value of financial analysts 655

possessors (firm insiders, industry experts, analyst peers) and clients (in-

vestors, buy-side analysts) form perceptions about analysts via social interac-

tions. We study whether and how perceptions about analysts are associated

with analyst outcomes to obtain insights into the role of impressions in in-

formation acquisition and information dissemination in capital markets.

Most face impression studies use human ratings of faces to measure one

or a few specific personality traits selected by the researchers, such as beauty

and competence. However, people form a multitude of impressions from

observing faces. We apply recent social psychology models and artificial in-

telligence (AI) machine learning (ML) techniques to extract a set of key

factors that comprehensively measure a multitude of impressions gained

from observing faces.

We first use facial recognition software to identify a large set of facial

features from the LinkedIn profile photographs of U.S. sell-side analysts.

We then apply the ML algorithms to obtain empirical measures for three

key face impressions or Face Factors: trustworthiness (TRUST), attractiveness

(ATTRACT), and dominance (DOM). Weexamine the associations of these

factors with analyst forecast accuracy, stock return sensitivity to analyst

forecast revisions, and analyst attainment of All-Star status. We further

explore face factor effects across different situational contexts to provide

additional insights about the potential channels for impression effects.

Cognitive psychology suggests that the wide range of personality traits

in face impressions can be reduced to these three key dimensions, which

together capture a substantial amount (72%) of the variation in face im-

pressions (Oosterhof and Todorov [2008], Sutherland et al. [2013]). The

TRUST factor reflects multiple traits related to the observer’s perceptions

about the observed’s intention to help or hurt the observer. The DOM fac-

tor concerns perceptions about the ability of the observed to carry out

intentions. Oosterhof and Todorov [2008] suggest that humans have de-

veloped these perceptions through natural selection over our evolutionary

history as shortcuts in cognitive processing to appraise threats in our social

environment to aid survival. Regarding ATTRACT, Sutherland et al. [2013]

suggest that the factor is associated with unconscious fitness and mate se-

lection cues.4

Perceptions about the personality traits formed from interactions with

the financial analysts affect the analysts’ information access. We hypothesize

that information possessors may be more willing to share information with

analysts they trust more (high-TRUST analysts), who they perceive as more

effective and efficient in processing and disseminating information (high-

DOM analysts), or who they find more attractive or novel (high-ATTRACT

4The TRUST factor loads highest on traits such as approachability, trustworthiness, and

degree of smile, the DOM factor loads highest on dominance, sexual dimorphism, and con-

fidence (Vernon et al. [2014]), and the ATTRACT factor loads highest on attractiveness and

health (Sutherland et al. [2013]).