Cheating, emotions, and rationality: an experiment on tax evasionAbstractIntroductionThe experimentExperimental designTwo treatmentsEliciting risk attitudeThe "Physionomics lab" and the self-reportsPredictionsExperimental proceduresResultsSummary statistics and non-parametric analysisDeterminants of the evasion decisionReactions to an auditDiscussion and conclusionAcknowledgementsAppendix. The physionomics laboratory and the skin conductance data analysisReferencesExp Econ (2010) 13: 226–247DOI 10.1007/s10683-010-9237-5Cheating, emotions, and rationality: an experimenton tax evasionGiorgio Coricelli ·Mateus Joffily ·Claude Montmarquette ·Marie Claire VillevalReceived: 12 November 2008 / Accepted: 26 February 2010 / Published online: 19 March 2010© Economic Science Association 2010Abstract The economics-of-crime approach usually ignores the emotional cost andbenefit of cheating. In this paper, we investigate the relationships between emotions,deception, and rational decision-making by means of an experiment on tax evasion.This research was supported by a grant from the Rhône-Alpes Region (CIBLE program).Electronic supplementary material The online version of this article(doi:10.1007/s10683-010-9237-5) contains supplementary material, which is available to authorizedusers.G. Coricelli · M. JoffilyCNRS, Centre de neurosciences cognitives, 67, boulevard Pinel, 69675 Bron Cedex, FranceG. Coricellie-mail: [email protected]. JoffilyIPUB, Federal University of Rio de Janeiro, Av. Venceslau Brás, 71, 22290-140 Rio de Janeiro,Brazile-mail: [email protected]. MontmarquetteCIRANO, Université de Montréal, 2020 University Street, Montreal, H3A 2A5 Quebec, Canadae-mail: [email protected]. Villeval ()Université de Lyon, Lyon, 69007, Francee-mail: [email protected]. VillevalGATE, CNRS, 93, Chemin de Mouilles Ecully, Ecully Cedex 69130, FranceM.C. VillevalIZA, Bonn, GermanyM.C. VillevalCCP,Aarhus,DenmarkCheating, emotions, and rationality: an experiment on tax evasion 227Emotions are measured by skin conductance responses and self-reports. We show thatthe intensity of anticipated and anticipatory emotions before reporting income posi-tively correlates with both the decision to cheat and the proportion of evaded income.The experienced emotional arousal after an audit increases with the monetary sanc-tions and the arousal is even stronger when the evader’s picture is publicly displayed.We also find that the risk of a public exposure of deception deters evasion whereasthe amount of fines encourages evasion. These results suggest that an audit policythat strengthens the emotional dimension of cheating favors compliance.Keywords Deception · Tax evasion ·Emotions · Physiological measures ·ExperimentJEL Classification C91 · C92 · D87 · H261 IntroductionMany situations involve deception, like plagiarism, embellished résumés, or trick-ery in sports. Recent scandals at Enron, for example, remind us that cheating inmarkets is not unusual (Gerschlager 2005). Tax evasion and social fraud provideother examples of substantial deception by economic agents (Slemrod 2007). Thestandard economics-of-crime approach explains deception as resulting from a com-parison between the expected benefit and cost of cheating. The cost of cheatingis however restricted to the monetary consequences of detection, and the cost ofthe very act of cheating is neglected. Yet, recent studies have provided some ev-idence of lie or guilt aversion (Gneezy 2005; Charness and Dufwenberg 2006;Sanchez-Pages and Vorsatz 2007). Regarding tax evasion, if empirical studies con-firm the basic mechanisms of the deterrence models (Allingham and Sandmo 1972,and Yitzhaki 1974), they also reveal that some people comply although the benefitsof cheating clearly outweigh its monetary cost (see Alm 1991; Andreoni et al. 1998;Cowell 1990, and Slemrod 2007, for surveys). This has motivated the exploration ofother dimensions of tax compliance, like tax morale, social norms (Slemrod 1998;Torgler 2007; Kirchler 2007), ethics (Blumenthal et al. 2001), or conformity (Mylesand Naylor 1996; Fortin et al. 2007). We adopt a complementary approach by in-vestigating the influence of emotions on cheating behavior with an application to taxevasion. Economists have not widely explored yet the role of emotions in cheatingbehavior. We aim at contributing to fill this gap.1We claim that both the cognitive deliberation preceding income reporting and therealization of an audit, possibly followed by sanctions, involve emotions. We assumethat emotions associated with reporting, audit, and sanctions inform and influencethe agents’ deliberation and therefore, constitute a driving force of behavior. Various1Although Bentham early incorporated emotions in his theory of utility, economists have only recentlystudied the role of emotions in decision-making (Elster 1996; Loewenstein 2000). They have identifiedtheir influence on reciprocity (Ben-Shakar et al. 2007), punishment (de Quervain et al. 2004), or investmentdecisions (van Winden et al. 2008).228 G. Coricelli et al.emotions might be related to compliance or evasion.2Individuals might experienceboth anticipatory emotions when evaluating the risk associated with underreportingand anticipated emotions when anticipating how bad they will feel if audited andpunished and how good and relieved if not audited.3We also expect that, ex post, the detection of cheating raises emotions which in-tensity is increased when deception is made public. In particular, detected evadersmay feel regret and guilt even when the information is not spread out and shame iftheir cheating behavior is made public.4This is at least what the authorities expectwhen they publicly expose the offenders.5Since it is difficult to cope with a damagedreputation, shame avoidance might have a larger impact on future compliance thanits private counterpart (on such amplification effect, see Kahneman and Miller 1986).Our paper is original in that our laboratory experiment links economic behavior inan income-reporting game to physiological measures of emotions and self-reports ofemotions. In our benchmark treatment, each subject receives an income that is taxedat a proportional rate. He has to decide how much income he is willing to report. Hefaces a probability to be audited, in which case he must pay a penalty for underreport-ing. The picture treatment follows the same rules except that an individual’s