Unpacking Negligence Liability: Experimentally Testing the Governance Effect

• Author: Theodore Eisenberg,Christoph Engel
• DOI: http://doi.org/10.1111/jels.12099
• Published date: 01 March 2016
• Date: 01 March 2016

Unpacking Negligence Liability:

Experimentally Testing the

Governance Effect

Theodore Eisenberg and Christoph Engel*

Arguably, if a court holds a defendant liable for negligently inflicting harm on the plaintiff,

this intervention combines three effects: (1) the court specifies the normative expectation,

(2) the court expresses dissatisfaction with the plaintiff’s behavior, for example, her level of

activity, and (3) the court obliges the defendant to compensate the plaintiff. In the field, it

would be close to impossible to disentangle the three effects, or to investigate how they

interact with intrinsic reticence to inflict harm on a passive outsider. We therefore go to the

lab. We do not find an effect of intrinsic morality. However, the intervention has a separate

significant effect on each of the three channels.

I. RESEARCH QUESTION

Whenever a court rules against a defendant in a typical torts case, the court (1) deter-

mines that there was harm to a third person, (2) defines a standard against which the

defendant’s behavior is to be measured, (3) determines that the defendant violated that

standard and expresses society’s disapproval, and (4) obliges the defendant to compen-

sate the plaintiff.

While classic law and economics emphasizes the extrinsic incentive created by the

obligation to pay, interpreted as a monetary sanction, a potential tortfeasor might also

be dissuaded from committing a tort by the moral intuition of not harming a third per-

son, by the tendency to adhere to a norm, or by the desire to avoid blame. In the field,

all these factors are conflated. We therefore design an experiment to disentangle these

different sources of compliance, and study their interaction. We look at both first-round

effects and the dynamic over time.

From the perspective of classic law and economics (Posner 2011; Cooter & Ulen

2012), the threat of liability is a constraint (Shavell 1980, 1987). Would-be tortfeasors

*Address correspondence to Christoph Engel, Director, Max Planck Institute for Research on Collective Goods,

Kurt-Schumacher-Straße 10, D 53113 Bonn, Germany; email: engel@coll.mpg.de.

We gratefully acknowledge helpful comments by Giancarlo Spagnolo and by three anonymous referees, as

well as the audiences at the Conference on Empirical Legal Studies at Berkeley, the Hebrew University Law

School in Jerusalem, the Max Planck Institute for Human Development in Berlin, and the Max Planck Institute

for Research on Collective Goods at Bonn. We are most grateful to Lukas Kießling for programming the experi-

ment in zTree, and to Florian Knauth for running the experiment in the lab.

116

Journal of Empirical Legal Studies

Volume 13, Issue 1, 116–152, March 2016

are assumed to possess perfect cognitive abilities, and to be exclusively motivated by per-

sonal profit. For such an individual, the prospect of liability changes the optimization

problem. If liability is newly imposed, or if the extent of liability is changed, rational

actors exhibit a marginal reaction.

The typical situation in the field is not one individual intentionally harming

another. Another situation is much more common. A person engages in some activity

that yields a profit. But for some outsider, this activity is risky. The first individual could

reduce the risk, but this is costly. It either has an out-of-pocket cost, or it has an oppor-

tunity cost in that the first individual would have to reduce her level of activity. In such

a setting, law and economics scholars advocate legal intervention, but they do not call

for containing risk at whatever cost. Rather, they propose to hold the first individual

liable only if she has exercised less than optimal care, or has chosen a higher than opti-

mal activity level, for that matter, with the optimum defined as marginal social benefit

equalizing marginal social cost. In the world of rational choice models, the obligation to

pay damages is anticipated. Victims never have to go to court since would-be tortfeasors

exhibit the prescribed level of care, or level of activity (Shavell 1987).

This is certainly an elucidating perspective on tort law. However, it is also narrow.

If lawyers say that the defendant has been negligent, they also mean that the defendant

did not intend to damage the plaintiff. In light of prevalent moral intuitions, knowingly

imposing harm on an innocent outsider is not the same as exposing this outsider to the

risk of suffering harm. Legal doctrine distinguishes between intentional harm and inten-

tional negligence, and considers the latter to be less blameworthy. In the first situation,

detriment is certain, while it is only a possibility in the second situation. Vignette studies

demonstrate that this doctrinal distinction reflects moral intuitions of law’s subjects

(Mueller et al. 2012). Yet exposing bystanders to risk is not morally desirable or neutral.

It still is bad, only not as bad as doing so intentionally.

1

Moral intuitions do not only matter in a deontological perspective. Lately, this

observation has fueled a whole literature on the commitment effect of promises (Char-

ness & Dufwenberg 2006; Vanberg 2008; Reuben et al. 2009; Charness & Dufwenberg

2010; Ellingsen et al. 2010; Ederer & Stremitzer 2014). To the extent that the law

matches the moral intuitions of its subjects, it may reduce enforcement efforts. If the

individual cares sufficiently about her own morality, she may be willing to live up to the

legal standard even if there is no enforcement at all.

Practical court rulings combine all three components: the court defines a behav-

ioral standard; the court expresses that the defendant has violated this standard and

explains that, thereby, the defendant has violated the normative expectations of the law;

and, using this justification, the court obliges the defendant to compensate the plaintiff

for the damage she has caused. This potentially combines four motivating forces: the

individual’s intrinsic inhibition to harm a passive outsider, normativity, blame, and a

1

See only §§ 281--283 and § 302, Restatement of the Law 2d, Torts. We are not interested in, and do not study,

an alternative definition of negligence: the actor had overlooked the risk for the outsider. Legal orders also treat

deliberate exposure to harm as negligence.

117Unpacking Negligence Liability

monetary sanction. It is the purpose of this experiment to disentangle these four forces,

and to demonstrate in which ways they contribute to the forward-looking effect of negli-

gence liability.

These four motives may be independent of each other (economists would then

call them additively separable). If so, the need for enforcing liability is the smaller the

more pronounced morally grounded inhibition to impose harm, the stronger sensitivity

to normative expectations of the (legal) community, and the stronger the disciplining

effect of blame. However, intrinsic morality, normativity, blame, and sanctions may also

be substitutes. If so, legal intervention crowds out moral inhibition, at least partly

(Bowles & Polania-Reyes 2012). Or the motives may be complements: they reinforce

each other. We have reason to expect that all three motives affect the choice of activity

level. However, we do not have cogent theoretical reasons to expect these effects to be

substitutes, complements, or not to interact with each other. We therefore do not have

a directed hypothesis.

Morality is generally believed to be a personality trait. In principle, it can there-

fore not be induced ad hoc. Yet there are situations in which moral inhibition to harm

an outsider is pointless. This is the case if the participant plays against a computer. We

exploit this possibility and repeat the experiment, but replace the passive player with a

machine. Comparing these additional treatments with the main experiment, we can

find out to which degree knowing that one exposes another human individual to a risk

of severe harm is critical for the remaining effects to play themselves out.

Rational actors do not need to learn their preferences. They possess a complete

preference function. All they need is information about the environment. A typical tort

relationship is unidirectional. There is a tortfeasor and a victim.

2

The victim is passive.

Therefore, a rational actor needs no information about the victim. All she needs is

information about the prospect of gain and the risk of having to pay compensation. All

this information is available right from the start of the experiment, even if the same

potential tortfeasor repeatedly comes into a situation that exposes passive outsiders to a

risk of harm. In principle, actors guided by morality need no learning either. When

they first face the tradeoff between gain for themselves and potential harm to a victim,

they dissolve the conflict between their individual moral standard and profit according

to the weight they attach to either element of their utility. Yet the repeated experience

of being responsible for a passive outsider, of being reminded of the social optimum, of

being blamed, or of having to compensate a victim may have an additional effect in

either direction. It may be that the individual gradually overcomes the temptation of

selfishness, and aims at living up to her own moral standards, with moral expectations

of society as expressed in official statements, and becomes more and more sensitive to

the experience of having to pay damages. Or any of these forces may lose their power if

they are called upon repeatedly. This is why we repeatedly expose our participants to

the conflict between individual profit and potential harm to a passive outsider. For

2

We bracket the possibility of contributory negligence.

118 Eisenberg and Engel