Preemptive bribery and incomplete information: Does prior knowledge matter?

• Published date: 01 June 2017
• Author: Ajit Mishra,Andrew Samuel
• DOI: http://doi.org/10.1111/jpet.12238
• Date: 01 June 2017

Received: 6 June 2016 Accepted: 4 October 2016

DOI: 10.1111/jpet.12238

ARTICLE

Preemptive bribery and incomplete information:

Does prior knowledge matter?

Ajit Mishra1Andrew Samuel2

1Universityof Bath

2LoyolaUniversity Maryland

AjitMishra, Department of Economics,

Universityof Bath, Claverton Down, UK

(a.mishra@bath.ac.uk).

AndrewSamuel, Department of Economics,

LoyolaUniversity Maryland, 4501 N.

CharlesStreet, Baltimore, MD 21210

(asamuel@loyola.edu).

Wewould like to thank Dilip Mookherjee,

EmmanuelDechenaux, Paul Pecorino, and semi-

narparticipants at the Southern Economic Associ-

ationmeetings (2014) for insightful conversations

onthis topic.

This paper studies bribery between a firm and a supervisor who

monitors the firm for regulatory compliance. Bribery occurs preemp-

tively, that is before the supervisor exerts costly effort to discover

the firm’s level of noncompliance and collect evidence for success-

ful prosecution. In contrast to previous papers, preemptive bribery

is modeled as a Bayesian signaling game because the supervisor is

uninformed about the firm’s level of noncompliance. We show that

under normal informational assumptions, some (possibly all) firms

always engage in preemptive bribery. However, if knowledge of the

firm’s level of noncompliance has implications for the supervisor’s

ability to collect evidence and prosecute (prior knowledge), pre-

emptive bribery can be completely eliminated. Results which apply

to preemptive bribery under complete information do not apply

here.

1INTRODUCTION

It is well recognized that corruption and bribery undermine enforcement in various regulatory settings. In particu-

lar, collusion between the supervisor who is in charge of enforcement, and the agent who is being regulated, dilutes

enforcement. However, bribery is often undertaken in highly uncertain environments with incomplete information.

These informational considerations can put natural limits on the nature and scope of bribery,for example, because the

briber does not know which official to bribe or how much to offer. Empirically using cross-country data, Lambsdorff

(2007) finds that uncertainty regarding the size of the bribe can reduce corruption. However,a theoretical explanation

for whyincomplete information places limits on bribery has not been precisely established. This is largely because most

of the literature assumes that bribery occurs under complete information. Thus, the objective of this paper is to exam-

ine the impact of these informational considerations on the likelihood of bribery. Furthermore,we wish to examine

whether the regulator can safely ignore certain types of bribery because asymmetric information makesthese unlikely

to occur.

Toachieve our objective, we consider a model of preemptive bribery where bribery can occur before the supervi-

sor has carried out inspections and the supervisor is relatively uninformed about the firm (agent). Although there is a

sizeable literature on corruption and enforcement, most of the focus is on ex post bribery; that is, collusion which takes

Journal of Public Economic Theory.2017;19:658–673. wileyonlinelibrary.com/journal/jpet c

2017 Wiley Periodicals,Inc. 658

MISHRA AND SAMUEL 659

place after the supervisor has collected evidence regarding the firm’s noncompliance.1Mookherjee and Png (1995),

for example, provide a detailed analysis of the conditions under which ex post bribery will occur and how it must be

deterred (see also Polinsky & Shavell, 2001). Distinguishing between ex post and preemptive bribery is important for

our project because incomplete information does not have strategic significance for ex post bribery—the supervisor is

supposed to have conducted the inspection and therefore is in possession of evidence. The supervisor might still be

uninformed about certain aspects, but the complete information set-up is not highly restrictive.

In contrast, with preemptive collusion the supervisor accepts a bribe in order to not carry out any inspection (Bac,

1998; Bag, 1997; Motta, 2009; Samuel, 2009). Therefore, it is more likely to be sensitive to informational consider-

ations and assumptions. Indeed, preemptive bribery differs from ex post bribery in two crucial aspects, which com-

plicates the analysis of its incentives and deterrence. First, preemptive bribery is more likely to occur because the

coalition of the firm and the supervisor saves on the inspection costs by engaging in preemptive bribery. Further-

more, because hard evidence has not been gathered, there is no risk of that evidence “leaking” and being used to

prosecute the firm and the supervisor in the future. Second, as we have discussed, preemptive bribery takes place

with a relatively uninformed supervisor who does not know the extent of the firm’s noncompliance. In this scenario,

a firm may be unwillingto bribe the supervisor, because the bribe may signal information about the extent to which the

firm is noncompliant. Arguably then, this asymmetric information may be sufficient to disrupt the bargaining process,

thereby,preventing preemptive bribery from occurring. Given that avoiding costly evidence gathering encourages preemp-

tive bribery, while incomplete information discourages it, we wish to examine how these two effects impact the incidence of

preemptive bribery in equilibrium. Toour knowledge, ours is the first attempt at fully understanding the implications of

asymmetric information for preemptive bribery.

Toachieve this goal, we develop a model of bribery in a regulatory setting where polluting firms can be either low-

waste or high-waste firms. Different levels of waste attract different mandated fines. The government hires a super-

visor who can easily observe whether a firm is nonpolluting or polluting, but cannot distinguish between low- and

high-waste firms.2Todistinguish between low- and high-waste firms, the supervisor must exert costly effort to gather

evidence regarding the firm’s waste level. Conditional on successfully obtaining evidence, the supervisor is required

to report this waste level to the regulator who imposes a fine on the firm based on this report. In the absence of any

evidence, a firm is treated as a nonpolluting firm by the legal process. We assume that supervisors are corruptible and

may accept a bribe in exchange for not reporting the firm’s levelof waste to the regulator. If bribery occurs preemp-

tively,the supervisor accepts a bribe in exchange for not gathering evidence regarding the firm’s type. The supervisor

can commit to not investigate and seek hard information or evidence.

Unlike ex post bribery, preemptive bribery occurs before the supervisor observes whether the firm’s actual type is

known. Therefore, we set up the preemptive bribe game as a signaling game where firms offer bribes to the supervi-

sor and the size of the bribe can potentially signal the firm’s type. Using the concept of Bayesian Nash equilibrium, we

determine whether preemptive bribery can occur in equilibrium. As is well known, signaling games admit a plethora

of equilibria; therefore, we characterize the entire set of (BayesianNash) equilibria of this game. Although we identify

several equilibria, we show that there existsa unique (separating) equilibrium which satisfies standard refinement cri-

teria. In this equilibrium, high-waste and low-waste firms offer different bribes which are accepted with differentprobabilities.

As a result, high-waste types always engage in preemptivebribery and escape prosecution, and low-waste types are sometimes

prosecuted. Thus, our first result is that informational asymmetry has limited deterrence and the presence of asymmet-

ric information alone is not sufficient to prevent preemptivecollusion completely.

Nevertheless, the standard signaling framework does not fully capture the informational complexities of preemp-

tive bribery. Specifically,in many cases the preemptive bribing process can reveal some information about the firm’s

level of waste, and this information can makethe supervisor’s investigation more effective at a later stage in the game.

1Empiricallypreemptive bribery is quite common. See Barron and Olken (2009) for an example where both ex post and ex an te bribery takeplace in the trucking

industry.

2This information structure is often used to capture the hierarchyof information. The firm has the finest, the supervisor has a coarser, and the regulatorhas

thecoarsest information structure. For example, see Celik (2009).