Does Delegation Undermine Accountability? Experimental Evidence on the Relationship Between Blame Shifting and Control

• Date: 01 June 2015
• Author: Adam Hill
• Published date: 01 June 2015
• DOI: http://doi.org/10.1111/jels.12074

Does Delegation Undermine

Accountability? Experimental Evidence on

the Relationship Between Blame Shifting

and Control

Adam Hill*

A small but growing literature in experimental economics finds that principals can shift

responsibility for blameworthy behavior to agents, even when those agents are effectively

powerless. Prior work in this field measures blameworthy behavior only indirectly, however.

It uses modified dictator games to measure attributions of blame for inequitable allocations

of wealth. Yet participants might find inequitable allocations of wealth not blameworthy.

Thus, such indirect measures leave open the possibility that prior work is not measuring

blame shifting at all. This article corrects for a crucial shortcoming by providing a direct

measure of blame-shifting behavior. It reports and discusses first-of-its-kind experimental

evidence that shows that principals can delegate to powerless intermediaries in order to

evade blame.

I. Introduction

Can principals delegate the authority to take immoral or antisocial actions to their agents

in order to evade responsibility for the consequences of the decision? A burgeoning line of

research suggests that principals can shift responsibility to their agents, even where the

principal exerts significant control over the agent’s decision (Hamman et al. 2010; Coffman

2011; Bartling & Fischbacher 2012). One of the main results, for example, shows that in the

context of a dictator game, principals face less punishment by delegating a decision that

produces inequitable payoffs to an agent, even if the agent is powerless to choose the

equitable outcome (Grossman & Oexl 2012). Thus, on the basis of this literature it appears

that principals can, in fact, employ agents to evade responsibility.

However, these dictator games measure immoral or antisocial behavior only

indirectly. These studies interpret inequitable allocations in a dictator as immoral or

*Ph.D. Candidate, University of California at Berkeley, 2240 Piedmont Ave., Berkeley, CA 94704; email:

adh@berkeley.edu.

This research was supported by grants from the UC-Berkeley Experimental Social Sciences Lab and the Berkeley

Empirical Legal Studies Fellowship. Special thanks to Jennifer Arlen, Stephen Galoob, William H. J. Hubbard,

Christopher Kutz, Janice Nadler, Anne Joseph O’Connell, Avani Mehta Sood, participants of the 2014 Berkeley

Empirical Legal Studies Colloquium, Journal of Empirical Legal Studies editors, and two anonymous referees for valuable

advice and guidance.

bs_bs_banner

Journal of Empirical Legal Studies

Volume 12, Issue 2, 311–339, June 2015

311

antisocial behavior, and they interpret third-party decisions to deduct money from the

agents’ allocations as “punishment” (Erat 2013). Both methodological decisions pose dif-

ficulties. These studies provide no evidence that participants in fact construe inequitable

payouts as immoral or antisocial. Indeed, Sanjiv Erat suggests that inequitable payouts

might be straightforwardly prosocial.1While not all of the studies make available their

instructions to participants, those that do make the instructions available do not charac-

terize inequitable payouts as immoral or antisocial. Although researchers versed in, for

example, public goods literature might be strongly inclined to interpret inequitable results

as blameworthy or antisocial (Hamman et al. 2011), one would need further evidence to

conclude that the players themselves take inequitable payouts as blameworthy. In short, the

experimental economic games that have thus far driven this burgeoning literature have

failed to provide direct measures of players’ beliefs that inequitable dictator game distri-

butions are in fact blameworthy.

This article addresses this shortcoming in the literature. It provides the first study of

delegation and blame to directly test third-party observers’ beliefs about the blameworthi-

ness of the principals’ and agents’ behaviors. Instead of relying on indirect measures of the

immorality or antisocial nature of certain behaviors, the experiments reported here directly

measure third-party observers’ evaluations of the blameworthiness of the principals’ and

agents’ behavior. Using vignettes, the experiments reported here operationalize the behav-

ior of principals and agents in the context of legislative delegation to administrative

agencies. After reading the vignettes, participants in the study rated the blameworthiness of

the actors in several treatments, including a treatment in which the principal delegates to

a powerless intermediary agent. Even in these cases, where the agent is effectively powerless

to change the outcome, participants blame principals significantly less than in cases where

the principal brings about the outcome directly. Accordingly, this article provides the first

clean evidence that principals can delegate a blameworthy decision to a powerless inter-

mediary and nonetheless evade responsibility.

A. Legislative Delegation of Policy-Making Authority

This article studies the relationship between delegation and blame in the context of

legislative delegation of policy-making authority to administrative agencies. In such situa-

tions, we are often confronted with the question of how voters allocate blame when

Congress and an agency are jointly involved in creating unpopular policy. Ideally, rational

voters with perfect information would attribute blame in proportion to the relative respon-

sibility of each institution (Cutler 2004). The complexity of contemporary policy making,

however, makes certain types of information material to the process of blame attribution

1Erat’s own study sought to solve this problem of measuring immoral or antisocial behavior by shifting from a dictator

game to a sender-agent-receiver game. In Erat’s study, the principal engages in deception. This approach does not

solve the problem, however, since deception might well be warranted and in any event we receive no direct evidence

of what the participants make of the deception under the circumstances. Erat acknowledges as much: “Whether or

not deception is immoral is admittedly a debatable, and much debated, point”(Erat 2013:274). Yet despite acknowl-

edging that whether deception is immoral is much debated, Erat notes that his paper “assume[s] that people do

perceive deception to be an immoral act.” Id.

312 Hill