Confidence and career choices: an experiment*

• Published date: 01 January 2022
• Author: Kai Barron,Christina Gravert
• Date: 01 January 2022
• DOI: http://doi.org/10.1111/sjoe.12444

Scand. J. of Economics 124(1), 35–68, 2022

DOI: 10.1111/sjoe.12444

Conf‌idence and career choices: an

experiment*

Kai Barron

WZB Berlin, DE-10785 Berlin, Germany

kai.barron@wzb.eu

Christina Gravert

University of Copenhagen, DK-1353 Copenhagen K, Denmark

cag@econ.ku.dk

Abstract

Conf‌idence is often seen as an important determinant of success. However, empirical evidence

regarding the causal effect of conf‌idence on choices is sparse. Using a stylized laboratory

experiment, we examine the effect of an increase in conf‌idence on two important labor market

choices: (i) the amount of ability-contingent earnings risk to take on, and (ii) the subsequent

effort choice. Wef‌ind that increased conf‌idence leads subjects to take on more ability-contingent

earnings risk. However, effort levels are unaffected. Overall, the upward shift in conf‌idence is

detrimental for low-ability workers as a result of high baseline levelsof conf‌idence.

Keywords: Beliefs; career choices; experiment; overconf‌idence; real effort

JEL classif‌ication:C91; D03; M50; J24

1. Introduction

Conf‌idence in one’s own abilities is commonly thought to be an important

determinant of success. A large body of work has studied the channels

through which holding overconf‌ident beliefs might be benef‌icial. For

example, B´enabou and Tirole (2002,2003) provide a discussion of how

a higher level of self-conf‌idence can motivate individuals to work harder,

overcome obstacles, and take benef‌icial risks. Brunnermeier and Parker

*Wewould like to thank TeodoraBoneva,Antonio Cabrales,Rustamdjan Hakimov, Zahra Murad,

Heiner Schuhmacher, Alice Solda, Robert St¨uber, Georg Weizs¨acker,and participantsat THEEM

ESA European Meeting and Nordic Conference in Behavioral and Experimental Economics for

helpful comments and suggestions. This research was funded by the Jan Wallanders and Tom

HedeliusFoundation and the ToreBrowaldhs foundation (grant number P2016-0051:1). K. Barron

gratefully acknowledges f‌inancial support from the Deutsche Forschungsgemeinschaft (through

CRCTRR 190, project number 280092119). C. Gravert gratefully acknowledges support from the

Center for Economic Behavior and Inequality (CEBI) at the Universityof Copenhagen, f‌inanced

by grant no. DNRF134 from the Danish National Research Foundation. Declarations of interest:

none. All mistakes are our own.

C

2021The Authors. The Scandinavian Journal of Economics published by John Wiley& Sons Ltd on behalf of F ¨oreningen

f¨or utgivande avthe SJE.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution

and reproduction in any medium, providedthe original work is properly cited.

36 Conf‌idence and career choices: an experiment

(2005) argue that individuals might hold upward-biased beliefs in order

to enjoy the consumption value of a rose-tinted view of the future, while

K¨oszegi (2006) studies the behavior of individuals who derive ego-utility

from overconf‌ident beliefs. Complementing these theoretical discussions,

a growing body of empirical literature has studied various mechanisms

through which conf‌idence can contribute to an individual’s success. This

literature presents evidence suggesting that more conf‌ident individuals are

better at persuading others that they are of high ability (Burks et al.,2013;

Schwardmann and Van der Weele,2019;Solda et al.,2020), work harder

(Puri and Robinson,2007;Pikulina et al.,2017;Chen and Schildberg-

H¨orisch,2019), and that overconf‌idence is an adaptive evolutionary trait

(Bernardo and Welch,2001;Heifetz et al.,2007;Johnson and Fowler,

2011).

This view that being overconf‌ident is benef‌icial represents a deviation

from the standard Bayesian rational agent perspective; that is, for a standard

economic agent, more accurate beliefs are typically better, particularly in

non-strategic settings. Less accurate beliefs lead to more mistakes, which

results in a loss of utility. A substantial body of theoretical and empirical

evidence has documented examples of the costly mistakes that overconf‌ident

individuals can make, such as exposing themselves to excessive risk

(relative to their risk preferences) in f‌inancial markets (Odean,1998;Barber

and Odean,2001), poor managerial decisions (Malmendier and Tate,2005),

and over-entry into competition (Camerer and Lovallo,1999;Niederle and

Vesterlund,2007;Dohmen and Falk,2011).

What is relatively undisputed is that there exists a wealth of evidence

documenting the widespread existence of overconf‌idence (see, e.g., Moore

and Healy,2008) and that overconf‌idence is one of the most commonly

studied behavioral biases (see, e.g., B´enabou and Tirole,2016). Despite

this prevalence and prominence of overconf‌idence, causal evidence that

assesses how an exogenous shift in conf‌idence affects behavior is relatively

scarce.1Such causal evidence is essential for evaluating whether increasing

an individual’s conf‌idence is benef‌icial or harmful in a specif‌ic domain. Any

observed correlation between individual heterogeneity in overconf‌idence

and behavior or outcomes could simply be driven by other unobserved

characteristics that generate the behavior or outcomes.

Furthermore, standard economic models typically postulate that behavior

in the presence of dynamic uncertainty operates as follows: new information

arrives →Abeliefs are updated according to Bayes rule →Bchoices are

1Some notable exceptions to this include papers that create exogenous variation in one’s belief

in oneself (relative to the truth) by varying the feedback that individuals receive, such as in

Schwardmann and Vander Weele (2019)andChen and Schildberg-H ¨orisch (2019), o r by varying

the mechanism used for selection into the experiment (e.g., Camerer and Lovallo,1999).

C

2021The Authors. The Scandinavian Journal of Economics published by John Wiley& Sons Ltd on behalf of F ¨oreningen

f¨or utgivande avthe SJE.

K. Barron and C. Gravert 37

then prescribed by these posterior beliefs. There has been considerable effort

invested recently in investigating the f‌irst relationship (→A), namely in

testing the descriptive validity of the way belief formation has traditionally

been modeled (see, e.g., Grether,1980;Eil and Rao,2011;M¨obius et al.,

2014;Heger and Papageorge,2018;Benjamin,2019;Coutts,2019;Barron,

2021). However, less attention has been paid to the second relationship

(→B), namely testing whether a shift in beliefs actually translates into a

shift in choices in the way that traditional models predict. The literature

that has studied this relationship empirically reveals a complex picture,

documenting many instances in which beliefs do not affect behavior in

the manner predicted by the standard model (see, e.g., Costa-Gomes and

Weizs¨acker,2008;Duffy and Tavits,2008;Ivanov et al.,2010;Chetty

and Saez,2013;Oster et al.,2013;Golman et al.,2017;Huck et al.,

2018;Fischer and Sliwka,2018;Lergetporer et al.,2018;Berkes et al.,

2019;Bettinger et al.,2020;Haaland and Roth,2019;Thaler,2020). In

Section 2, we provide a more detailed discussion of this literature. This

body of work highlights the importance of empirically testing the causal

relationship between beliefs and actions in different contexts, so that we

can learn why a shift in (measured) beliefs has a large effect on behavior

in some contexts and no effect in others. Additionally, these types of studies

can serve as a litmus test in the evaluation of a particular model, since they

provide a reduced-form test of a key comparative static (i.e., the causal

effect of beliefs on actions).

In this paper, we contribute to this endeavor of studying how a shift

in beliefs translates into changes in behavior. To do this, we use a stylized

career choice task to study several channels through which an exogenous

shift in beliefs can operate. We develop a simple theoretical framework and

show empirically how an upward shift in conf‌idence can causally inf‌luence

decision-making regarding the subject’s preferred payment scheme, effort

provision, and resulting earnings.

The following illustrative example provides the intuition for how an

increase in conf‌idence can affect decision-making in a labor market setting.

Imagine a computer programmer who is about to graduate from college (let

us call her Taylor). Taylor is well educated. She has the choice between

two types of jobs: (1) a job at a mid-sized company that will pay her a

f‌ixed wage, and (2) working at a start-up where her earnings will depend

heavily on her performance. She will earn far more at the start-up if she

is better than the average competing programmer who is also graduating

from college, and far less if she is worse.

Assume that Taylor, like many other people, believes that she is better

than average (see, amongst others Kruger,1999;Burson et al.,2006;Healy

and Moore,2007;Moore and Healy,2008;Benoˆıt et al.,2015). This belief

can be inf‌luential for two of her decisions. First, she needs to choose the

C

2021The Authors. The Scandinavian Journal of Economics published by John Wiley& Sons Ltd on behalf of F ¨oreningen

f¨or utgivande avthe SJE.