SlideShare a Scribd company logo
1 of 51
Download to read offline
Leniency Policies
Jaime Plaza Aparicio
July 13, 2015
Bachelor Thesis in Economics
Department of Economics
Chair for Microeconomics and Game Theory
Prof. María Sáez Martí
Jaime Plaza Aparicio
Matriculation-number: 11-732-005
jotaplaza93@gmail.com
Phone: +41786923581
Friesenbergstrasse 16
8055 Zürich
Switzerland
Abstract
This thesis compares leniency programs according to economic theory with real
leniency programs .The analysis focuses mainly on offering leniency in form of a
reward as well as on offering leniency before or/and during investigations. The
Folk Theorem and a competition model with leniency policy (Spagnolo 2005) are
described as well as three different leniency policy papers based on that model.
This thesis also reviews the intentions that lead to the appearance of leniency
programs as well as the main goals of them. The two main leniency programs
(U.S. and E.U.) and the differences between one program and the other are
explained to show how actual leniency programs work. Conclusions lead to real
leniency programs been designed similar to the ones proposed by economic
theory, except for the leniency in form of a reward.
Table of Contents
1. Introduction.............................................................................................1
2. Leniency Programs.................................................................................2
2.1 What is a Leniency Program? .........................................................2
2.2 Why implement Leniency Programs? ...........................................4
2.3 Evolution of Leniency Programs....................................................8
2.3.1 U.S. Leniency Program..............................................................9
2.3.2 E.U. Leniency Program............................................................13
3. Economic Theory..................................................................................19
3.1 Folk Theorem...................................................................................20
3.2 Competition Model with Leniency Programs............................22
3.3 Spagnolo (2005) ...............................................................................24
3.4 Chen & Rey (2013) ..........................................................................30
3.5 Aubert, Kovacic & Rey (2005).......................................................35
4. Conclusion.............................................................................................41
References..................................................................................................44
Leniency Policies Jaime Plaza Aparicio
1
1. Introduction
Leniency programs have changed dramatically the way cartel problems are
tackled. Cartels and similar anti-competitive practices are condemn by
competition authorities all around the world. The Sherman Antitrust Act in the
U.S. and the Article 81 of the European Commission Treaty prohibit activities
that can distort market competition.
Nowadays, a competition authority without a leniency program is rarely
perceived as something feasible. If properly designed, leniency programs can
change the incentive structure of cartel members and provide the means for a
deviation to be profitable. In addition, if programs are effectively planned, they
can also prevent cartel formation.
It is important to highlight that economists started studying leniency programs
and focused not only on the effectiveness of them, but also on their design.
Hence, this thesis focuses on the history and the design of leniency programs.
Directors of the competition authorities designed the programs in a way in which
they do not always harmonize with economic theory. The differences between
both lines of thought are remarkably high in certain aspects, for instance, in
defining the form leniency adopts.
The aim of the thesis is not to decide whether a leniency program is efficient in
itself or not; this work is only trying to assess the differences between the leniency
programs that prevail in “the two worlds”: the real world and the one described
in the economic theory. Nonetheless, it is worth mentioning that the latter partly
agrees with the programs implemented in the real world, and thus this thesis will
explain this “partial” agreement. Economist defend leniency programs and have
some well-established views on certain aspects related to them that have not been
implemented yet in real world programs.
This work is divided in four sections. Section 2 focuses on the history and the
development of leniency programs, and the historical events and the intentions
that motivated politicians to implement such programs are also explained. This
section 2 also explains the two most representative leniency programs (U.S. and
E.U) and the differences between them.
The economic theory is explained in Section 3, where the folk theorem and a basis
model are presented. Here, three major papers based on that model and the most
important concepts of them are exposed. In addition, each of the papers has some
differences to real leniency programs and the work tries to capture them too.
Leniency Policies Jaime Plaza Aparicio
2
Section 4 divides leniency programs into three parts and concludes with
explaining the differences between real leniency programs and the economic
theory through these three parts.
2. Leniency Programs
2.1 What is a Leniency Program?
A leniency program is a tool used by competition authorities to reduce sanctions
or fines for a firm involved in a collusive agreement that self-reports and
provides useful information and evidence. Leniency policies are one of the tools
used by competition authorities in combination with active investigations or
fines in order to deter the collusion among firms in various sectors of the market.
The authority offers a degree of leniency in exchange for active participation in
uncovering wrongdoing.
A leniency program is an automatic agreement between the competition
authority and a wrongdoer conditional on the latter meeting a series of
requirements. These requirements are publicly known, and by fulfilling them,
the agreement is guaranteed. The agreement can be seen as an exchange between
two parties. The authority grants leniency in fines, if wrongdoers report the
collusive agreement and meet certain requirements.
One feature of leniency programs is the clear trade-off between its benefits and
costs. The benefit from inducing a “death threat” to the cartel by giving the
members an escape not as harmful as the expected sanctions (fines or prison) and,
at the same time, by stimulating cartels, members can profit from ending the
illegal activity whilst incurring in less damage than they would without a
leniency program. This underlines the importance of the proper design of the
leniency programs, tipping the balance in the proper direction. This means
neither being too strict nor too generous in terms of the leniency awarded.
The idea behind this concept is much older than the first rudimentary leniency
program from the U.S. (1978). This program was the first step to nowadays
corporate leniency programs in U.S. and the rest of the world. The main concept
of inducing mistrust in a group (in this case a cartel) has been very common
throughout human history, especially in such situations as wartime conflicts and
the illegal activities of mafias.
Leniency Policies Jaime Plaza Aparicio
3
Since trust is required in cartels or criminal organizations to build a durable and
profitable relationship, wrongdoers must be certain that individual parties will
not deviate in any way from the agreement. If members do not trust one another,
they cannot be confident that the agreement will be sustained in the future.
Spagnolo (2005 & 2006) highlights the famous “Divide et Impera” strategy
employed by Julio Caesar to defeat his enemies by bringing a few of the opposing
groups across to his side, in order to make it impossible for the enemies to unite
and work together, thereby breaking the opponent’s front. His model from the
2005 work can be interpreted both as an optimal leniency program intended to
deter collusion among firms and to reduce criminal misbehavior among criminal
organizations. Treating cartels as criminal organizations is a very good
counterpoint in the analysis of leniency policies. Differences in the national and
international characteristics of the Antitrust Authority (US) and European
Commission (European Union) make the possible punishments entirely
different. Whereas in both cases colluding firms face monetary sanctions, within
US territory they also face prison sentences.
Spagnolo (2006) names three essential features that cartels and organized crime
share. The first entails cooperation between members of the illegal activity
whereby the classic economic problems of cooperation such as free riding or
moral hazards are of greater importance. These activities usually take part in
several periods of time making it an ongoing relationship, whereby future benefits
and costs deemed important and therefore taken into account. As members
cooperate among themselves in an ongoing relationship, they end up obtaining
evidence of one another’s wrongdoing. The last feature is a result of the first two.
Aside from their tasks in the illegal activity, members must monitor each other.
The extensive information typically obtained from this ongoing relationship,
regarding wrongdoing, is crucial for leniency programs. The aim of competition
authorities when focusing on leniency policies is to give members an incentive to
hand in evidence.
When focusing on cartels and other collusive agreements, it should be noted that
not all information is incriminatory. As in González (2007), a difference can be
made between circumstantial and hard evidence. Whereas circumstantial
evidence is open to interpretation, hard evidence proves without any doubt the
wrongdoing. To this latter group belong, e-mails/letters, records, oral statements
or documents alongside other form of correspondence. Circumstantial evidence
is ambiguous and cannot be taken as incriminatory. Aubert, Rey and Kovacic
Leniency Policies Jaime Plaza Aparicio
4
(2005) cite the example of the woodpulp case. The European Commission
classified the parallel evolution of prices in the woodpulp industry as
incriminatory, and as an evidence of a collusive agreement, but the European
Court of Justice annulled the decision, asserting that price parallelism is
compatible with competitive behavior. In general, observable data without any
hard evidence makes it difficult to prove collusion.
That which is meant by leniency is a full or partial reduction in the fine; however,
there are some regulations that go further and reward agents who report an
illegal activity. The authority rewards agents that provide them with hard
evidence of a collusive agreement. Although the economic analysis defends the
concept that a system of rewards, instead of fine reductions, is more effective, in
reality, such systems are rare. Examples of this include South Korea, which in
2002 implemented a system that grants rewards for individual informants of
collusive agreements, and the US Civil False Claim Act are some examples. The
Act rewards individuals who inform the U.S. government of fraudulent activity.
The frauds reported directly involve the U.S. government as the purchaser of
goods and services or as insurer.1
2.2 Why implement Leniency Programs?
In words of the Director of Criminal Enforcement from the Antitrust Division,
Scott D. Hammond2
: “[…] an effective leniency program will lead cartel members
to confess their conduct to authorities even before an investigation is opened”.
Before the implementation of leniency programs, reductions in sanctions or fines
were in use but they were neither publicly nor general knowledge. Agents
colluding or committing a crime could not take leniency into consideration since
there was no leniency program in place, and trying to negotiate more lenient
terms was fraught with uncertainty.
Antitrust authorities had, and continue to have, other techniques to detect cartels
apart from the leniency policies. These techniques are not available in every
jurisdiction, and sanctions also differ among jurisdictions. Different sanctions can
also increase the fear of detection in stricter regions of the world, making
1
Aubert, Kovacic, and Rey (2005)
2
Hammond (2004)
Leniency Policies Jaime Plaza Aparicio
5
collaboration in those regions more advantageous. The Antitrust Division3
listed
the cooperation of “insiders”, the assistance of skilled FBI agents in interviews,
the collaboration from immigration authorities to track cartel members, and the
execution of search warrants on the offices of the suspected firms as some of the
tools used in their investigations. Despite this, the new version of the Corporate
Leniency Program (1993) has helped to detect a greater number of international
cartels than the combination of aforementioned methods available to the
Antitrust Division.
Spagnolo (2006) remarks that the characteristic of being “ex ante”, general, and
public is what differentiates leniency programs. This makes them different in
terms of their effectiveness and form from previous leniency agreements. It is
“ex-ante” because the program acts a stage prior to other mechanisms such as
plea bargains, which are similar but come into play at the prosecution stage. The
aim of these programs is to encourage the wrongdoers not yet detected to come
forward. It is “general” in so far as any agent facing such situations can adopt
this program and apply for leniency. It is “public” because agents must know the
existence of both the program and the situations in which they can adopt it.
Spagnolo also highlights codification also as a useful instrument of these new
programs. Codification helps to reduce some of the main problems of these pre-
program situations; namely, discretionality and uncertainty. If firms want to
apply for leniency but do not know the amount of leniency offered by the
authority, or are afraid of possible changes in jurisdiction, calculating the value
of reporting will be difficult.
Spagnolo (2006) also summarizes the goals of well-designed leniency programs.
The first goal is the prevention of cartel formation, deterring the formation of
cartels by inducing mistrust in the cartel structure from the onset, so that the
probability of members losing confidence increases. The second goal is oriented
to cartels that were formed before, or in spite of, the adoption of the program.
The authority achieves its detection of cartels by reaching an agreement with one
of the members and obtains the information from this source. These two concepts
are very important in terms of the analysis. For this framework cartel deterrence,
and only cartel deterrence, means to prevent a cartel from being formed because
the leniency program makes the illegal activity unprofitable. Cartel detection
means that the authority discovers a cartel as a result of one of the members
reporting it through the leniency program.
3
Hammond (2004)
Leniency Policies Jaime Plaza Aparicio
6
With the implementation of leniency programs, applying for leniency becomes
more predictable. Transparency, publicity, anonymity or generality are some of
the features that new well-designed programs fulfil that helps their
predictability. To assist its implementation, firms should know the existence and
workings of these programs. Speeches in workshops and conferences increase
the program knowledge between its potential “users”; namely, the corporations
and their directors, officers and employees.
Scott D. Hammond, former Director of Criminal Enforcement of the Antitrust
Division of the US, explains in one of his speeches the three fundamental
cornerstones for the success of a leniency program: threat of severe sanctions,
perceiving a high risk of detection, and high transparency and predictability of
the program.
If the program is well-designed, it should succeed in real world. Although
competition authorities consider the leniency program their most powerful and
successful deterrence weapon, and that with its implementation the number of
convicted cartels and fines has increased, there is a need to be cautious when
taking into considering the programs. The increase in leniency applications,
cartel detection and fines, after they have been introduced is clear and
undeniable, but to characterize the program as a complete cartel detector is
perhaps an overly sweeping claim.
Since the number of active cartels reigning in the market is unknown, the increase
in cartel conviction cannot be solely attributed to the leniency programs. The rise
could come from the existence of a greater number of cartels in the market or, as
the authorities suggest, from leniency programs, but a clear statement cannot be
formulated. When adopting a program with full deterrence and detection of
cartels, the authority ought not to receive any leniency application. This is clearly
not the case, since the leniency applications rates have increased with the
implementation and updates of the programs.
Even assuming this success comes directly from leniency programs, another
issue arises. The design of the leniency program may have played a role, which
meaning that a different design could have performed even better than the one
implemented. It can also be argued that leniency programs have changed nothing
in cartel deterrence and detection (same probability of detection), but simply that
the number of active cartels have increased, and that it is this that has resulted in
a notable increase in cartel convictions.
Leniency Policies Jaime Plaza Aparicio
7
The introduction of leniency programs has increased the number of convicted
cartels4 and has also radically changed the way in which the authorities discover
cartels. The deterrence effects are difficult to quantify and qualify, but the
analysis is easier for determining the effects of detection.
Detection effects can be seen in the applications for leniency received by the
competition authorities, as for example with HENKEL, a company that produces
chemical products such as detergents or shower gels. HENKEL has applied for
leniency three times in Spain, but also in Italy, in France, in Germany and at the
European Commission. The program has proved very profitable for HENKEL’s
interest, and it subsequently assisted several competing authorities with the
discovery of cartels. In Spain alone HENKEL has saved over 15 million Euros.
There are undoubtedly other companies, which, like HENKEL would be willing
to apply for it or that have already applied. With the application comes the
investigation in the offices of the reported firms; the emails, videos and other
exchange of information between the wrongdoers; the knowledge of the practices
used by them to fix the price or exchange information and these not only help the
authority with regards to the reported cartel, but the authority also learns about
those involved firms in the cartel and their bad practices. Multinational
companies that collude in one particular market may also have other collusive
agreements in other markets, which proved to be the case with Henkel and its
cartel’s colleagues. If HENKEL and the other companies take turns in reporting
and their agreement remains unbroken in spite of the reports, the competition
authority will lose with regards to the leniency program and should change the
incentive structure of it.
A leniency program, when it is well designed so that firms do not systematically
benefit from reporting, can prove greatly beneficial to the authority through the
information and knowledge that it acquires about the running of cartels. This
only happen in the worst case-scenario, whereas the program could actually
deter and detect cartels at a very high rate.
Despite the success and evolution of leniency programs, competition authorities
did not dare to set leniency as a reward. High fine reductions or fine cancellations
are the form leniency takes in the policies of the authorities. If the authority is
willing to refrain from levying the full fine against a wrongdoer, why should it
not go further by setting a reward? Through its leniency policies the authority
4
Spartling (1998), Delrahim (2004)
Leniency Policies Jaime Plaza Aparicio
8
reveals that discovering an entire collusive agreement compensates the
condonation of the sanction to a confessed wrongdoer. If the authority is willing
to pay that price for cartel detection, why do not increase this effect by setting a
reward instead of a reduction/cancellation?
Aubert, Kovacic, Rey (2005) name certain possible implantation issues for
rewards. A limited budget can impede the possible reward but the reward can
be paid through the fines imposed on the other cartel members. There may be
public condemnation of rewarding wrongdoers, especially at a time when a
country is in recession. A third issue is the additional collusion incentives that
rewards produce. If with a fine reduction firms can adopt collusive strategies to
exploit the leniency authority, a reward can increase incentives to collude and
loot the competition authority.
2.3 Evolution of Leniency Programs
The aim of the previous sections is to explain what changes for the wrongdoers
with the introduction of the programs, the characteristic of ideal programs and
the trade-offs and mistrust induced by them. This section focus on how real
leniency programs are designed, their evolution the years that have passed since
the first programs were adopted and the differences between the main leniency
programs.
Figure 2 shows the year of adoption of a leniency program in each country and
the percentage of countries from the total with leniency program in that year. In
less than two decades most of the countries have adopted a leniency program.
These programs can greatly differ between countries and can be more or less
generous or restrictive. Since the revision of the Corporate Leniency Policy in 1993
more competition authorities in national and international regulations have
adopted a leniency program for firms.
Since, in this section, the approach is going to focus on how real programs are
designed there must be clarity about the users of the program. When wrongdoer
or agents was mentioned, it primarily referred to firms.5
Despite the existence of
individual leniency policies, the majority of leniency programs are oriented
towards corporations. This work focuses principally on leniency treatments to
5
For further analysis whenever wrongdoer or a similar term is mentioned is meant firms. Whereas the
term individuals, and only it, will be used for persons.
Leniency Policies Jaime Plaza Aparicio
9
firms; nevertheless, in some jurisdictions the leniency to corporations implies the
leniency treatment extends to individuals.
Leniency programs have changed with many useful revisions, but the pioneer is
the U.S. The U.S. program has been the model for the rest of the programs. The
main two competition authorities with regards to leniency regulation are the
Antitrust Authority of the U.S. and the European Commission of the European
Union (E.U). The following sections will focus on both leniency programs and
their differences and similarities.
2.3.1 U.S. Leniency program
The first version of the Corporate Leniency Policy dates from 1978 and was
introduced by the U.S. Department of Justice (DoJ). Within the DoJ the Antitrust
Division is the responsible for the policy. Wrongdoers reporting illegal activity
before an investigation had been launched were eligible to obtain leniency.
Amnesty6
was not granted automatically and the Division had great deal of
prosecutorial discretion in the process. The program was not very appealing and
this resulted in a very poor number of applications (only one application per
6
For this section, as in S. Hammond (2004), the terms amnesty, leniency and immunity will refer only to
a full elimination of the fine and not to partial reductions.
Figure 1: Borrell, García, Jiménez (2013)
Leniency Policies Jaime Plaza Aparicio
10
year) and it was unable to discover an international cartel. Both theorists and
directors of competition agreed the program had some errors in its design. As
Spagnolo (2006) highlights, the program was neither transparent nor
“automatic”.
Given that the program was unable to bring firms to apply for leniency, the
Division revised the Corporate Leniency Policy in 1993. The aim was to make it
easier and make applying for leniency and cooperating more attractive to firms.
Three aspects, as stated by several members of the Division7
, changed in the
revision that exist and have made the program what it is today.
i) Automatic amnesty if there is no investigation already launched.
ii) Amnesty could still be awarded even if firms cooperate during an
investigation.
iii) Every individual director, officer and employee from the firm that obtains
leniency is protected from criminal prosecution.
The application rate quickly jumped to one application per month and reached
rates near 20 applications per year8
. With this new face the Corporate Leniency
Policy became the most effective tool in detecting international cartel cases.
Branches from the same cartel came to the Antitrust Division within the same
weeks, and even within the same day9
, but only the first to come was granted
leniency.
Spagnolo (2006) comments on the huge impact of the revision in the program.
Together with the increase in the application rate, the magnitude of the penalties
imposed increased dramatically. Supported by OECD reports that state that the
increase in the use of leniency programs is powered by the substantial increase
in penalties, Spagnolo asserts the two forces operate together. Spagnolo also
defends the notion that the higher quality and quantity of evidence in hands of
the Division was crucial for impose higher sanctions. Finally, higher sanctions
make the risk of getting convicted greater and therefore the program proves more
appealing.
7
Spratling (1998), Hammond (2000), Delrahim (2004).
8
Hammond (2004)
9
It did not happen in the U.S. but HENKEL and other company applied the same day for leniency;
namely, the day the leniency program was first adopted in Spain.
Leniency Policies Jaime Plaza Aparicio
11
This new version of the program, still currently in use, is divided into four parts10
.
Leniency before an investigation has begun, alternative requirements for
leniency, leniency for corporate directors & employees, and leniency procedure.
A firm involved in an illegal activity11
, looking to be awarded with full immunity,
that is not known by the Division ought to meet several requirements. The firm
must i) be the first from an unknown cartel to report illegal activity; ii) take quick
and effective action to end its participation in the illegal activity; iii) completely
report to the Division and cooperate with it throughout the investigation; iv)
confess as a truly corporate act; v) where possible make restitution to injured
parties; and vi) have participated in the illegal activity without coercing other
parties to take part in it, which means, among other things, being neither the
leader or originator of the illegal activity.
A firm reporting the illegal activity after the beginning of the investigation has a
further opportunity to achieve immunity. The Division grants immunity to a firm
that reports even if the investigation has already started, but it must not have
incriminatory evidence at that point of the investigation. This firm must i) be the
first to report and qualify for leniency from the members of the illegal activity
reported; ii) report at a time when the Division has no evidence against the
company that serves for the conviction of it; iii) take quick and effective action to
end its participation in the illegal activity; iv) completely report to the Division
and cooperate with it helping advance the investigation; v) confess as a truly
corporate act; and vi) where possible, make restitution to injured parties. Other
than that the Division must determine that awarding leniency to that firm, in
consideration of the illegal activity, will not be unfair to others. Being the leader
or having coerce others to participate as well as reporting after an investigation
has started can jeopardize the granting of leniency by the Division.
If a firm qualifies for leniency and meets all requirements its directors, officers
and employees, having admitted their involvement in the illegal activity, will
receive leniency, which means that they cannot be charged criminally. Qualifying
for leniency can save the corporation millions of dollars and also eliminates the
threat of prosecution that would have persisted by continuing with the illegal
activity.
10
Corporate Leniency Policy (1993)
11
For sake of clarity, when in previous sections the term “cartel” was mentioned, it does not meant only
cartels but every other form of collusive illegal activity disturbing competition that is condemned by the
Antitrust Division.
Leniency Policies Jaime Plaza Aparicio
12
Qualifying for leniency also entailed certain costs for the firms. Besides
interrupting the typically profitable illegal activity, firms could be the target of
federal and state-treble damage claims based on their involvement in the cartel.
As mentioned by Spagnolo (2006), applicant firms could find themselves not only
liable for three the damages suffered by customers but also thrice that suffered
by the costumers of their conspirator.
With a new legislation approved in 2004, Criminal Penalty Enhancement and
Reform Act, that tries to solve this cost, the Division wanted to give firms a
greater incentive to report. The new legislation reduces potential damages in
private lawsuits to single damages based on their own role in the cartel for the
receiver of leniency, whilst other participants remain fully liable for treble
damages12
. As stated by Spagnolo (2006) this new legislation increases the
potential liability for colluding firms that do not apply for leniency. In words of
the Deputy Assistant Attorney General of the Antitrust Division M. Delrahim13:
“the other cartel participants remain fully liable for treble damages based on
harm caused by the entire conspiracy.”
Within the leniency program there are some mechanisms that make reporting
even more attractive to firms involved and which could result in more cartels
being convicted. The Amnesty Plus Program encourages firms involved in ongoing
investigations to seek out amnesty in other markets in which they compete. By
qualifying for leniency (the Amnesty) in other uninvestigated market, and,
precisely, by collaborating with the Division in the discovery of this new cartel
firms will get a reduction (the Plus) on the fine imposed for the participation in
the first cartel.
The Penalty Plus Program works in the opposite direction to the Amnesty Plus
Program. If a firm, which is been investigated and participates in cartels in other
markets does not profit from the opportunity given by the Amnesty Plus and
denies any other illegal activity, it will suffer harsh consequences if the Division
discovers its implication in any other cartel. That which is meant by the Plus is
that if the lie is discovered, the firm will not be condemned with the “regular”
fine or jail sentence of each cartel, but the Division will pursue a harder level of
sanction.
12
Delrahim (2004)
13
Delrahim (2004)
Leniency Policies Jaime Plaza Aparicio
13
Through its directors, the Division has espoused the success of the program. Its
work has changed a lot with the revision of the program in 1993 and such
statements are very useful in illustrating how the revision has affected the
behavior of the firms with respect to the leniency program.
In 1994 the Division also published the Individual Leniency Policy.14
The program
is oriented to individuals that approach the Division representing themselves,
and not a corporation, and report an illegal antitrust activity. In the case of
individuals, leniency means freedom from any criminal charges. As with the
Corporate Leniency Policy, individuals must meet several requirements that run
along similar lines to those required by the corporations.
Currently, when the individual applies for leniency and reports an illegal
antitrust activity i) the Antitrust Division must have not received any information
about the illegal activity being reported; ii) the report must be complete and
collaboration throughout the investigation must be offered to the Division; and
iii) the participation in the illegal activity must be free from coercion to other
parties.
Firms usually run a race against the other cartel members to apply first for
leniency. With the introduction of the Individual Leniency Policy, they also run a
race versus their own employees. One of their employees could approach the
Division, win the race and qualify for leniency leaving the firm defenceless.
Directors of the Division15 have suggested that the utility of the program lies in
the number of corporate applications it produces. Firms, afraid of losing the race
against their employees, could apply for leniency as a corporation, securing
leniency for both the corporation and all its employees. The Division values this
more than the possibility of a great number of individual leniency applications.
2.3.2 E.U. Leniency Program
The old continent took longer to implement a leniency program and revised it
more than the U.S. The different versions of the E.U. programs take up more
pages than those of the U.S., which are clearer and easier to read.
14
Individual Leniency Policy (1994)
15
Hammond (2004)
Leniency Policies Jaime Plaza Aparicio
14
Notably, the E.U. Leniency program is oriented towards firms involved in cartels
that operate in more than one country of the European Union. If the cartel only
operates in one country then firms must seek leniency from the competing
authority of that particular country.
In 1996, three years after the revision of the U.S. program, the European
Commission published the Commission Notice on the non-imposition or reduction of
fines in cartel cases16
. This notice explained the conditions that firms applying for
a reduction should fulfil.
Applying for a fine reduction or leniency in this first version came with
uncertainty. Despite some discussion, the notice set some basics for the following
versions of the program. This first version only allowed for firms to obtain the
total fine reduction providing they reported before an investigation had been
launched and when the Commission did not have sufficient information to prove
the existence of a cartel.
Firms had to meet several conditions. Ending its participation in the illegal
activity; being the first to report decisive evidence; providing the Commission
with all relevant information available as well as extending the cooperation
through all the investigation; having participated in the cartel without being the
instigator or coercing other parties; and reporting prior to all other firms involved
in that cartel were the important requirements that firms had to meet. If a firm
met all this conditions, it could have benefited from a reduction of at least 75% of
the fine that it could have received without the program.
If firms come to report after the investigation has started, but meet all the
aforementioned conditions, it still could have benefited from a reduction of
between 50% and 75% of that fine. The Commission also gave reductions of fines
between 10% and 50%, if firms that did not meet the named conditions cooperate
in the investigation. Firms from 4317
different cartels applied for leniency between
1996 and 2014, with the years 2001 and 2002 being when most firms applied for
a reduction.
This first version lacked clarity, particularly with regards to the type of
information that was deemed necessary to report and the precise nature of the
process. Firms applying for a reduction were not sure about the amount
discounted from the fine. This problem also appeared in the first version of the
U.S. program in 1978. The Commission had a lot of discretion in the process
resulted in lot of uncertainty regarding applying for it.
16
Commission Notice (1996)
17
Borrell, Jiménez, Ordóñez (2014)
Leniency Policies Jaime Plaza Aparicio
15
For these reasons the Commission revised the Commission Notice of 1996, and in
2002 made the Commission Notice on immunity from fines and reduction of fines in
cartel cases18
public. They admitted that, according to their judgement, the 1996
version was effective but could be improved by increasing its transparency and
certainty. One desired feature of the program was to clarify the conditions (not
for a fine reduction, but) for the immunity from fines.
This new version modified some important aspects of the former one. The
Commission granted immunity, as opposed to a high fine reduction, if the firm
was first in reporting the illegal activity to the authority, and providing that they
did not already have sufficient evidence to incriminate the cartel. Conditions like
not having coerce any other party, ending the participation in the illegal activity
from the point the cooperation starts or cooperating through the entire process
were also in this second version.
This second version introduces several novelties. It was open to both instigators
and ringleaders. Only parties that coerced others were restricted from applying.
It introduced the possibility to apply in hypothetical terms. In this kind of
application, the reporting firm must present a list of evidence it proposes to
reveal at a later agreed date. The documents shown are free from sensitive
information and are used to illustrate the nature of the evidence. The
Commission could withdraw the immunity, or fine reduction at any moment
during the investigation if the firm fails to meet the conditions required.
In this second version, firms that do not meet the mentioned conditions (a basic
requirement, or that apply after an investigation has started) can apply for a fine
reduction, but are rejected from the possibility of immunity. These firms must
report the collusive agreement with evidence that adds significant value to the
evidence already available to the Commission. The Commission Notice (2002) tries
to clarify what is meant by significant value in order to reduce any uncertainty
and to increase the transparency of the program. Firms that report after the first
can still benefit from a reduction in fines. The first firm will get a reduction of
between 30% - 50%, the second firm between 20% - 30% and, at best, any firm
thereafter will obtain a maximum reduction of 20% or less. This version of the
program has been used by firms in relation to 29 different cartels since 200219
.
This second version still retained the basic elements introduced by the first
version in 1996 but increases transparency and certainty by clarifying its
conditions, the nature of the information required and by implementing new
mechanisms, such as a proper understanding of the application in hypothetical
terms. All this to make it easier for firms to end their participation in cartels since,
18
Commission Notice (2002)
19
Borrell, Jiménez, Ordoñez (2014)
Leniency Policies Jaime Plaza Aparicio
16
as stated in their Notices, the Commission is aware of the existence of firms
willing to finish its participation in the illegal activities.
In 2006 the Commission published the Commission Notice on immunity from fines
and reduction of fines in cartel cases (2006) going a step further in its work to increase
the transparency and clarity of the program and implement new mechanisms to
facilitate the report.
The fact that the program does not lack transparency and certainty is very
important. Reporting their involvement is a great risk for firms since it leaves
them exposed. Firms declare to the competition authority and confirm their
involvement in an illegal activity that has proved detrimental to consumers and
had great social costs. It is evident that the greater that a firm perceives the risk
of making a report backfiring, the more likely they are to back away and refrain
from doing so.
The 2006 version tries to eliminate uncertainty and provide applicants with
greater guidance owing to the evident shortcomings in former versions of the
program. The information handed to the Commission when seeking for leniency
must be sufficiently substantial that the Commission can carry out a targeted
inspection pertaining to that cartel, or can find an infringement of Article 81 EC20
.
Article 81 EC states the behaviours that are incompatible with competition or
common markets. Agreements of firms fixing prices (directly or indirectly),
limiting the production, or sharing sources of supply among other practices are
considered as conforming to such behaviour.
With regards to the reductions of fines, the information handed to the authority
will be more valid, if it does not need corroboration. Flexibility in the process,
restringing evidence manipulation or obligating the continuous cooperation for
both immunity and fine reduction applicants are some of the new features of
the program.
Whereas in the 2002 version the novelty lay in the application in hypothetical
terms, in 2006 the Commission introduced the marker system. Firms can apply
by sending only limited information that must be completed in the future, but
they secure their approach position to the Commission. This means that a firm
can arrive first and report with limited information that must be completed upon
in the future and, in doing so, secure its place first for immunity. Since 2006 firms
have applied for leniency with regards to 9 different cartels21.
20
European Commission Treaty
21
Borrell, Jiménez, Ordóñez (2014)
Leniency Policies Jaime Plaza Aparicio
17
Each new version of the program has come with a fewer number of sentences
sanctioning cartels by the authority. As stated in section 2.2, a reason for this
could be that program is becoming more efficient with each consecutive revision.
However, this might also be explained through firms that have fear of incurring
even greater fines. This requires for further analysis and interpretation that goes
beyond the parameters of this work.
2.3.3 Differences between programs
The two main Leniency programs have some differences both in the prosecution
and the design of the program. Spagnolo (2006) names some of the differences
between the U.S. and the E.U. program.
Ringleaders
The leniency program in the E.U. is not restricted to ringleaders, whereas the U.S.
the program does not allow them to apply for leniency. This fact can create
differences in the formation of cartels. Firms might perceive taking leader
positions as being overly risky. Furthermore, it raises the question of how the
will define the role of the ringleader. With unclear criteria, firms might be
unwilling to enter to cartel or, perhaps more clearly stated, one might posit that
no firm is likely to want to take a leading role since that means rejecting any
possible future leniency agreement.
Milder leniency
If the information reported by a firms adds additional value to the investigation,
regardless of the order of the report, the European Commission awards the
informant with some form of leniency. At the very least, a reduction of the
imposed fine can be achieved by reporting significant information. The U.S.
program denies this possibility. It only allows the first reporting firm to obtain
full immunity. This is a very important point. The “winner takes it all” approach
from the U.S. increase the incentives to report first. The position of the European
Commission22
is that collaboration has an intrinsic value that must be awarded
with immunity or a reduction in fines. In the E.U. reducing sanctions not only to
the first firm makes that strategies like “wait and report, if somebody reports”
increase its value against “report first”. If the difference in terms of leniency and
fine reduction between reporting first and second increases, the incentive to
report first will also increase. Spagnolo also comments that this approach from
the E.U. will lead to lower fines paid by cartel.
22
Commission Notice (2002)
Leniency Policies Jaime Plaza Aparicio
18
Immunity
Whereas, in the U.S., the first reporting firm obtains automatic leniency, in the
E.U. firms reporting after an investigation has started may obtain leniency,
depending on the quality of the information. By quality is meant the novelty of
the information provided which that adds to that already possessed by the
Commission. In the E.U. there is a link between the value of the information
reported and the awarding of leniency. Spagnolo argues that there is in fact no
difference. The U.S. program sets conditions about the information reported.
These conditions determine that the information reported must have high value
to the Division either because it proves the existence of an unknown cartel or else
because it helps the Division in the investigation of a cartel where the Division
does not a have lot of evidence.
Individual liability
The Individual Leniency Policy complements the corporate program in the U.S. The
possibility of individuals, employees, directors or officers, obtaining leniency in
the U.S. can generate agency problems in the firms. The E.U. does not offer such
possibility so that individuals cannot approach the Commission seeking leniency
for themselves, but for the corporation as a whole.
One more abstract reason, which complements itself with individual liability, is
the difference in the perceived risk of the two programs because of the possible
sanctions as stated in Hammond (2000 & 2004). The Antitrust Division is a
national institution, whereas the European Commission comprehends a high
number of countries. Cartel activity is treated as an administrative offense in the
E.U. and the highest possible sanction is a very high fine. Nevertheless, cartel
activity is treated as a criminal offense in the U.S. With this explanation, it is clear
that, since the E.U. cannot incarcerate individuals, the fines imposed by the
Commission have to be much larger than the ones imposed by the Antitrust
Division. The Commission has to increase the level of fines. The other alternative
would not prove attractive to the applicants.
In the U.S., firms that do not obtain leniency often pay fines equal to 30% of more
of the revenue generated by the sale of the “cartelled” product during the entire
lifetime of the cartel23
. On the other hand firms unable to obtain any reduction in
the E.U. have to pay a fine approximating 10% of the worldwide turnover for
every product sold by that firm24
. It is clear that the difference between obtaining
leniency and a reduction could save the company a lot of money.
23
Hammond (2004)
24
Hammond (2000)
Leniency Policies Jaime Plaza Aparicio
19
Choosing between high fines and incarceration may be an easy decision, as
appears to have been the case for a lot directors of corporations which, according
to Hammond (2004), operated in cartel in Europe, Asia and other places around
the world but avoided operating in a cartel in the U.S. In some of these cases the
U.S. market was potentially the most profitable market for them but they refuse
to operate there, and the reason given by the directors of the cartel firms for the
avoidance of the U.S. on the cartel map was the possibility of getting caught and
going to jail in the U.S.
Hammond (2004) states that the leniency program of the E.U. is a very good one
for a jurisdiction that does not have criminal sanctions. Even so, he considers that
a jurisdiction with criminal sanctions and individual liability will always be more
effective at inducing amnesty applications. The fear induced by the U.S. program
thanks to the possible incarceration speaks for this statement.
3. Economic Theory
After the Antitrust Division revised the Corporate Leniency Policy in 1993 and three
years later with the adoption by the E.U. of its own leniency program, the
programs became well-known. Firms started applying for the program and
economists started studying the program. Economists tried to figure out whether
the programs set the proper incentives to the applicants or not. The programs
have to be designed in a way that they benefit the authority and not every cartel
member, but the leniency receiver. Concepts like “winner takes it all” or
“exploitable programs” were on the table. Economists have tried to produce
models which approximate real world situations in order to see how much
leniency has to be offered, who can apply for that leniency, when participants
can apply for leniency and, even, if offering leniency was the right decision.
Economists also have considered going a step further and not offering leniency
but a reward, whereas competition authorities did not consider this possibility.
This thesis chooses certain models and findings from an extensive bibliography
of papers that deal with leniency programs. The models shown in the papers are
based on an infinitely repeated interaction among firms. By proposing
environments with and without leniency programs and defining fines applied
and resources available to the competition authority, the authors take some
conclusions on how well-designed leniency programs should be designed.
Taking into account the idealized world described in the models, the aim of this
Leniency Policies Jaime Plaza Aparicio
20
Figure 2: Prisoner's Dilemma
section is to compare the results obtained by the authors with the “real world”
programs. Even some statements by directors of the Antitrust Division on how
leniency should be are challenged by the economists. There is also a debate
among economists regarding the design of leniency programs.
Most of the models on leniency programs are based on the infinitely repeated
interaction among firms; therefore, it is necessary to explain how this assumption
“changes the game” dramatically. Large established corporations operating in
multiple markets around the world meet each other and exchange information
or agree on market shares. Even if one of these agreements is detected by a
competition authority, the firm will probably pay the fine and make new
decisions on further collusion or collaboration, but certainly will not disappear.
These big corporations will make those decisions throughout all their lifetime.
The Folk Theorem is an important concept to be emphasized when dealing with
infinitely repeated interactions.
3.1 Folk Theorem
A game with a unique Nash Equilibrium played a finite number of times by
rational players will yield the same result in every period; namely, the Nash
Equilibrium. Looking at the Prisoner’s Dilemma, (D,D) is the unique Nash
Equilibrium. However, both players would be better off playing (C,C). Making
credible that both will play C is not possible, unless there are mechanisms to
punish a deviation (enforceable contracts) or the interaction lasts for a finite
number of periods. If the time horizon changes to an infinitely repeated
interaction, players will want to play (C,C) under certain conditions.
Leniency Policies Jaime Plaza Aparicio
21
The Bertrand Competition will better illustrate this effect. There is a market with
two firms that produce perfect substitutes. The firm that sets the lower price wins
the whole market. Both have the same production costs per unit of c. They
interact with each other T periods of time with T = ∞. In each period of time, the
firms choose prices simultaneously. Choices may depend on former decisions,
i.e., on the prices set in the previous periods. A firm i will want to maximize the
presented discounted value of its payoffs, i.e. ∑ 𝛿 𝑡−1𝑇
𝑡=1 𝛱𝑖
(𝑝𝑖𝑡, 𝑝𝑗𝑡) .
The payoff of a firm i depends on the payoff of both firms i and j. Furthermore,
(p1, p2) = (c, c) is the only Nash Equilibrium in the one-shot game where the
incentive to deviate outweighs every other strategy. The monopoly price pm can
be sustained as an equilibrium price through the trigger strategy “set pm in the
first period and continue with pm as long as both firms have set pm, otherwise
set p = c”. With this strategy, each firm obtains half of the monopoly profit each
period. The present discounted value of this strategy is
𝛱 𝑚
2
( 1 + 𝛿 + 𝛿2
… ) =
𝛱 𝑚
2
1
1− 𝛿
.
The alternative for a firm is to slightly underbid the monopoly price and to win
the entire market. If a firm underbids pm in the first period, it can count on
competition, and a payoff of 0, for the remaining periods. Even so, this firm will
obtain the entire monopoly profit in the first period. Therefore the present
discounted value of this strategy is 𝛱 𝑚
.
The second strategy is a deviating strategy. For this section and the next ones, a
deviating firm will be a firm that was initially involved in a collusive agreements,
but deviates from the agreement setting a market variable which is not in line
with collusion and making higher profits with it. In the Bertrand Competition a
deviating firm will underbid the collusive price.
A comparison of the value of both strategies shows that the trigger strategy is
better off when 𝛿 ≥
1
2
. Then
𝛱 𝑚
2
1
1− 𝛿
≥ 𝛱 𝑚
→ 𝛿 ≥
1
2
.
Applying this reasoning, any other price between c and pm can be an equilibrium
price.
Leniency Policies Jaime Plaza Aparicio
22
Figure 3: Folk Theorem in the Bertrand Competition
The Folk Theorem states that every pair of profits Π1 > 0 and Π2 > 0 so that
Π1 + Π2 ≤ Πm is an equilibrium payoff in each period for δ sufficiently close to 1.
In other words the grey zone contains every sustainable per-period payoff for δ
close to 1. This reasoning can be applied for the infinitely repeated interaction
between firms under a leniency program. If the competition authority does not
have a leniency program and does not carry out any investigations, firms will
have no impediment to collude and will probably do so. A well-designed
leniency program will change the per-period payoffs through fines and fines
reductions to make reporting more profitable.
3.2 Competition Model with Leniency Programs
The following model is described in Spagnolo (2005). It is a dynamic model based
on an economy made up of a lot of oligopolistic industries that, at the same time,
consist of a number of risk neutral agents (firms). There is also a benevolent
legislator (competition authority) who controls the market and determines the
policies. A discounted infinitely repeated oligopolistic game between the agents
takes place in each of the industries. The legislator is in charge of setting the
parameters of the law enforcement policy. Any collusive behaviour that reduces
the social welfare is forbidden.
Firstly, the legislator sets the enforcement parameters and afterwards, having
observed the parameters, the agents begin their interaction. Agents live infinitely
and discount their future payoffs through a factor δ that is common to all of them.
Leniency Policies Jaime Plaza Aparicio
23
The market game played every period to the infinity can be interpreted as a
Bertrand or Cournot competition. Competing is the only Nash Equilibrium in the
one-shot game. Firms competing obtain a per-period payoff of 𝜋 𝑛
. However, a
higher payoff could be reached if players can make a collusive agreement. 𝜋 𝑑
and
𝜋 𝑐
denote the period payoffs of deviation and collusion, respectively. A collusive
agreement (e.g. a cartel between agents fixing prices) reduces the social welfare
and is, therefore, forbidden by the legislator.
Firstly, a world without law enforcement will be considered. Firms can sustain
collusive agreements in a subgame perfect Nash Equilibrium, if the value of
collude indefinitely (Vc) exceeds the value of a unilaterally defection (Vd) from
this agreement. The discounted sum of expected payoffs must be compared
𝑉 𝑐
=
𝜋 𝑐
1− 𝛿
> 𝜋 𝑑
+ 𝛿𝑉 𝑝
= 𝑉 𝑑
.
Or, normalized by (1 − 𝛿), 𝜋 𝑐
> (1 − 𝛿) 𝜋 𝑑
+ 𝛿𝑣 𝑝
,
where πd denotes the value from defection when the others stick to the
agreement, Vp denotes the value expected, i.e. the discounted sum of payoffs, at
the beginning of the punishment phase that follows the unilateral defection, and
𝑣 𝑝
denotes the time average payoff of the punishment phase with
𝑣 𝑝
= (1 − 𝛿) 𝑉 𝑝
.
Since to make collusion credible agents must penalize a deviation, it follows
𝜋 𝑑
> 𝜋 𝑐
> 𝑣 𝑝
. After a deviation, the punishment stage always come. With 𝜋 𝑏
denoting the payoff of sticking to the collusive agreement in a period when an
agent defects, it also follows 𝜋 𝑑
> 𝜋 𝑐
> 𝜋 𝑛
≥ 𝜋 𝑏
.
It can be argued that without a law enforcement policy agents will find the
collusive agreement more profitable. As described before, collusion requires
communication and this will produce some evidence every time agents meet.
With a law enforcement policy and a leniency program, the evidence generated
will become of greater importance. The law enforcement policy also affects the
communication among the cartel members and changes the timing of the game.
At a first stage, cartel members, having observed the policy parameters, exchange
information and update and confirm collusive strategies, generating evidence of
a collusive agreement. At a second stage, the same members set the market
variables and choose whether to report the collusive agreement or not. In the first
case a reporting member will apply for leniency through the leniency program.
The evidence produced disappears at the end of the period. However, as long as
Leniency Policies Jaime Plaza Aparicio
24
the evidence is valid, cartel members can transfer it to third parties at no cost. If
the legislator gets a report with evidence of a cartel, it is convicted with
probability one. The times the Legislator gets a report, it will go public at the end
of the period so that every agent in the industry will know who reported it.
The parameters of the law enforcement policy presented by Spagnolo are the
following:
 A monetary fine F that convicted firms have to pay. This fine F is
restricted to be in an interval with 𝐹 ∈ [ 𝐹, 𝐹 ].
 A reduced fine RF that a cartel member has to pay when he discloses
information to the authorities enabling the cartel to be convicted. The
reduced fine (positive transfer or reward if RF < 0) is also restricted to be
in an interval with 𝑅𝐹 ∈ [ 𝑅𝐹 , 𝐹].
 Beside the leniency program, the legislator can carry out investigations
and has the following (per-period) probabilities of conviction.
- α, denotes the probability of a cartel being convicted when no cartel
member reports the illegal activity.
– γ, denotes the probability of cartel members that unilaterally deviate
from the agreement being convicted.
– β = γ + η, with η ≥ 0, denotes the probability of a cartel member that
sticks to the collusive agreement in a period when at least another
member deviates being convicted.
To focus on realistic parameters Spagnolo assumes α, β, γ <
1
2
. Whereas
administering fines has no cost, the implementation of such probabilities has
some costs. Ck(k) denotes the social costs of implementing the probabilities k,
with k ∈ (𝛼, 𝛽, 𝛾). In addition, it is also assumed that 𝐶 𝑘(0) = 0, 𝐶 𝑘
΄ (𝑘) ≥ 0,
𝐶 𝑘
΄΄(. ) > 0. Costs rise with any new increase in the probabilities, but the increase
is higher when costs are at a low level.
3.3 Spagnolo (2005)
The paper from Spagnolo, on which the model from the previous section is based,
focuses on leniency programs and their ability to undermine trust, making the
deviation from a collusive agreement more profitable. Spagnolo differentiates
between courageous and moderate leniency programs. The first one gives a reward
to the reporting party financed by the fines from the rest of the cartel members.
Moderate programs only reduce or cancel sanctions and do not allow for rewards.
Leniency Policies Jaime Plaza Aparicio
25
In a world without leniency programs but with investigations from the
authorities, collusion can still be sustained. Namely, the value of colluding (VC
),
which is the discounted expected per-period payoff of collusion minus the
probability of paying the fine, should exceed the value of defecting (VD
), which
is the payoff of deviation minus the probability of getting convicted plus the
discounted sum of payoffs in the punishment stage, i.e.
𝑉 𝑐
=
𝜋 𝑐−𝛼𝐹
1−𝛿
> 𝜋 𝑑
− 𝛾𝐹 + 𝑉 𝑝
= 𝑉 𝑑
.25
Observing this condition, it follows26
that for γ below a certain threshold (think
of a very low γ) is optimal for cartel deterrence to set the maximal possible fine
(𝐹). For a higher γ (even higher than α), increasing fines does not increase cartel
deterrence, but deviation deterrence. Members will stick to collusive agreements
because deviation has a high probability of conviction (γ). Since increasing γ is
costly, the optimal policy leads to γ = 0. For further analysis, Spagnolo assumes
𝛼 > 𝛾 and the optimal fine set at its maximal possible level.
Leniency effects
Spagnolo also names two possible characteristics of leniency programs; namely,
exploitable and effective leniency programs. An exploitable leniency program makes
members of a collusive agreement benefit from reporting it.27 If the program is
too generous (low RF), cartel members could consensually report the cartel to the
authorities but may keep colluding and reporting in the following periods.
An effective leniency program makes deviating firms increase their payoffs
through a report.28 When a firm deviates from a collusive agreement, it can report
the collusion or remain silent. Within an effective program, firms will always
report the cartel after a defection. For deviating firms this means that the
probability of being convicted and paying the fine should exceed the reduced
sanction (𝑅𝐹 < 𝛾𝐹). Spagnolo calls this the protection from fines effect.
Courageous Leniency programs
As explained before, Spagnolo classifies leniency programs into courageous and
moderate leniency programs. Courageous leniency programs are self-financing, i.e.
the reward is paid by the fines imposed on the other cartel members. Observing
the possible effects of leniency programs, it turns out that restricting eligibility to
the first reporting party reduces the set of exploitable programs. A leniency
25
Spagnolo (2005) p.14 Condition 2
26
Spagnolo (2005) p.14 Proposition 1
27
Spagnolo (2005) p.15 Definition 2
28
Spagnolo (2005) p.16 Definition 3
Leniency Policies Jaime Plaza Aparicio
26
program looking for optimal cartel deterrence will restrict leniency to the first
party only29
.
The introduction of leniency programs creates an incentive compatibility
condition (IC)30
to colluding agents. The value of any kind of collusion (both the
regular one (VC
) and the consensual reporting one (VC/
)) should exceed the value
of deviation (both deviation with (VD
) and without (VD/
) reporting after
deviation). A collusive agreement is only possible if this condition is met.
max { 𝑉 𝐶
, 𝑉 𝐶/
} > max{ 𝑉 𝐷
, 𝑉 𝐷/
} (𝐼𝐶)
Spagnolo concludes that for every possible fine level, the reduced sanction must
be negative, i.e. the legislator must reward the reporting party.31
Rewards make
programs more efficient by increasing the value of “deviate and report” (VD
) and
this narrows the (IC). A deviating firm facing the decision whether to report the
cartel or not will be more likely to report, if the leniency to be obtained has the
form of a reward.
On the other hand, too high rewards can make the program exploitable. Spagnolo
finds that when fines are high enough and the detection probability (α) increases,
the optimal reward (RF*) decreases. This implies that in optimum, investigations
are a substitute for rewards as law enforcement instruments. When α is large, the
optimal reward must be small. In that case, the high probability of detection (α)
even allows a fewer reward level. At the same time, the optimal reward (RF*)
increases in fines, making fines and rewards complementary instruments.
Regardless of any other policy parameter, increasing fines allows for a higher
reward without making the program exploitable.
As described, investigations and rewards are substitutes. Whereas investigations
are costly, rewards are self-financing, making it much more profitable for an
optimal law enforcement policy to rely, whenever possible, on self-financing
rewards. Spagnolo finds that there is a finite level of fines at which complete and
costless deterrence (α = β = γ = 0) is achieved.32
This can be done by setting the
reward for the first reporting agent at an amount equal to the sum of all the fines
paid by the cartel and, at the same time, interrupting any investigation. The
authority can reject every possible investigation and wait for reports. The high
level of fines empowers the leniency program making investigations redundant
and suboptimal. This is the first time that the first best is achieved in such an
environment.
29
Spagnolo (2005) p.17 Proposition 2
30
Spagnolo (2005) p.17 (IC) Condition
31
Spagnolo (2005) p.17 Proposition 3
32
Spagnolo (2005) p.19 Proposition 4
Leniency Policies Jaime Plaza Aparicio
27
Moderate Leniency programs
On the other hand, moderate leniency programs put some constraints on fines and
rewards. Moderate leniency programs also have the protection from fines effect and
enable a protection from punishment effect. If the optimal first best fine cannot be
achieved because of the constraints on fines, the second best law enforcement
policy will imply positive investigation costs and also a non-maximal reward.33
Since the constraint on rewards does not allow for any reward, i.e. a positive
transfer, the optimal reward would be a complete reduction of the fine (RF = 0).
Spending resources on investigations will also be optimal considering that fine
reductions are not as attractive as rewards to cartel members.
When complete reduction of fines (RF = 0) is implemented with a probability of
conviction when deviating (γ) of zero, moderate leniency programs do not have
deterrence effects. It is clear that protection from fines effect (RF < γF) disappears
making the program ineffective. The program needs γ > 0 to enable this effect
and make the program effective. This means that even deviating firms could be
convicted for past collusive agreements. Firms only care about the value of γ if
they are willing to report.
Even in a scenario with γ = 0 and firms using a “stick-and-carrot” strategy, a
moderate leniency program may disrupt cartel activities. Firms in a “stick-and-
carrot” strategy incur a cost in the “stick” phase but they do so because this
allows them to enter the “carrot” phase in the next period. If the authority
punishes repeated wrongdoers more firmly, imposing higher fines and looking
more closely into their activities, making the probability of detection (α) higher
for them, incurring the “stick” phase cost will not be profitable anymore. The per-
period payoff of collusion would be the same but the fines will increase in each
new “stick” period making collusion unworkable. This is called the protection
from punishment effect.
Through both of these effects, moderate leniency programs, if well implemented,
can deter and detect cartel activity.
Risk of cheating
Spagnolo states that an additional reason explaining the deterrence effect of
moderate leniency programs is the perceived risk of being involved in collusive
agreements. In previous analysis, collusion was sustainable if the (IC) was
satisfied and cartel deterrence was achieved with an (IC) violation. Now it is
assumed that trust is also required for a collusive agreement. Without knowing
33
Spagnolo(2005) p.20 Proposition 5
Leniency Policies Jaime Plaza Aparicio
28
for sure that cartel members will stick to the agreement, no agreement can be
made.
Spagnolo presents a simple model of two firms that entered into a collusive
agreement and in each period must decide whether to stick to the agreement or
to deviate from it and expect punishment from its partners. Spagnolo applies
Harsanyi and Selten’s definition of risk dominance and calculates the Nash
products of the two pure strategy equilibria; namely, both agents colluding and
both agents deviating. He does so to measure the riskiness of both equilibria for
situations with and without a leniency program. Spagnolo concludes showing
that the colluding equilibrium with a restricted leniency program is the most
risky.34 This means that agents colluding in an environment with a restricted
leniency program will gain more from deviation than deviating in any other
environment (unrestricted leniency program or no leniency program situations).
Conclusions
A leniency policy that rewards reporting firms as well as a lack of active
investigations by competition authorities is fully against the reality. Active
investigations inducing fear in cartel members, as stated in Hammond (2000 &
2004), are crucial in cartel deterrence for the Antitrust Division.
Restricting the leniency program to the first reporter only is in line with the policy
applied by most of the competition authorities. Spagnolo shows how a restricted
program makes both the program more efficient and gives greater incentives to
deviate from a collusive agreement. This view is shared by the Antitrust Division
that considers the fear of arriving second as key to the efficiency of the program.
The European Commission goes in the opposite direction with its perception that
a collaboration has an intrinsic value to be awarded. In my opinion, a policy as
designed in the U.S. will induce much more fear in cartel members since “wait
and see” strategies lose all of their value. Awarding the second, third or
subsequent reporting parties with a fine reduction seems too generous to me and
goes against the spirit of what I suppose was the intended purpose of the leniency
program.
The European Commission has been accused of bureaucracy since its inception.
Acknowledging that the collaboration of cartel members will reduce the time and
resources spent in such processes. I consider that announcing a reduction for
every useful collaboration is too generous. Probably, these cartels have operated
in a lot of countries damaging consumers from different nationalities.
34
Spagnolo (2005) p.26 Proposition 7
Leniency Policies Jaime Plaza Aparicio
29
To measure exploitability the following reasoning can be used. Assuming that
with the adoption of a leniency program cartel fines increased, as stated in
Spagnolo (2006), it could follow that strategies of reporting in turns decrease their
value, since programs only accept the first party. Cartel members accept this kind
of “stick-and-carrot” strategies, especially in the stick-phase, because of the carrot
following in the next periods. If the authority toughens the fines but maintains
the awarded leniency constant, at a certain point this strategy will turn out
unprofitable.
Rewarding without investigating is the most important point of Spangolo’s
work. A self-financing reward that is fully paid by the fines of the cartel members
can become problematic. The authority only changes the money from hands.
Under a legislation as the E.U. this could be a problem. Firms could make an
agreement under which every cartel member pays an amount of the total fines
and after the reporting member gets the reward, the cartel members distribute it
among them. Firms could participate in a collusive agreement that is reported in
every period, paying and receiving the same amount without any loss. Of course
this is a very simplistic view and the authority will react to such behavior.
Nevertheless, Spagnolo does not count on such agreements. He models
consensual reports in which every cartel member reports the cartel every period.
In some periods a cartel member will have to pay the fine and in others, when he
manages to be the first one to report, he will receive the fines from the other
members. Since no good can be expected from a multinational operating cartel,
any kind of collusive agreement could be used by the cartel members, no matter
how “diabolic” in terms of taking advantage of the authority. This reasoning
could be reinforce with the fact that there are no investigations (in Spagnolo’s
framework) which makes the fear of being discovered disappear. Such
consensual reporting strategies will work in legislations without prison
sentences, but only if monetary fines are imposed as penalty, as the E.U.
A lack of investigation tools to spare resources is completely contrary to any
competition authority in real life. What induced the fear in the directors of
collusive firms were ultimately the tough sanctions, especially the possibility of
imprisonment. A fear of high sanctions could result in more reports but, as
explained in section 2.2., after the reports comes the learning. Knowing how firms
fix prices, exchange information or limit bids will help following investigations
of collusive agreements.
Leniency Policies Jaime Plaza Aparicio
30
3.4 Chen & Rey (2013)
Chen and Rey study the optimality of offering leniency before and after an
investigation is launched. Characteristics such as restricting the program to the
first informant only or offering leniency to repeated offenders are also studied.
The model suggested by Chen and Rey is much more technical than the model
from Spagnolo. However, the infinitely repeated interaction between firms is
repeated. An environment with multiple industries is presented. Collusion can
be sustained without a leniency program in some industries. This means that
firms will want to collude without breaking the agreement even without antitrust
enforcement. The discount factor of future payoffs (δ) is the same across
industries, whereas the potential win from collusion differs across them. This
sounds realistic considering that in the real world not every industry has the
same sales volume or “collusive” potential.
The parameters of the antitrust enforcement policy are exogenously given. Given
a collusive agreement and no-one reports, there are two probabilities: the
probability of carrying out an investigation (α) and the probability of success (p)
of this investigation making the probability of success of any investigation (αp).
With probability α an investigation is launched in a collusive industry. If an
investigation is launched, the collusive agreement is discovered with probability
p. Therefore, a firm involved in a collusive agreement that has no intention to
report, must expect discovery of the cartel probability αp.
The level of fines (F) is also exogenously given and is set to its maximal possible
level, which the authors think is the level where optimal deterrence is achieved.
Investigations are public and can be observed by any firm. Collusion, as in the
previous model, leaves some evidence that only remains valid for one period.
The authority implements a leniency program with two different leniency levels,
depending on the state of the investigation at where the report arrives. The
leniency offered can vary depending on the firm reporting before or after an
investigation has started. The leniency offered is a reduction (q) on the fine so
that a convicted firm will have to pay (1 − q)F. With q > 1 the leniency would be
offered in form of a reward; otherwise, the firm would only obtain a fine
reduction.
The timing of the game increases and so does the number of strategies firms can
use. Firms agree on collusive agreements and, at the same time, the competition
authority launches investigations. Firms can report before the investigation is
launched as well as with the investigation ongoing. A deviation from the
collusive agreement, i.e. in the Bertrand Competition underbidding the collusive
Leniency Policies Jaime Plaza Aparicio
31
price and winning the whole market, is not observed until the end of the period.
A deviating firm can also report the cartel and benefit from the leniency.
The Benchmark without a leniency program sets a critical value for the payoff
needed for collusion.35
Firms in an industry achieving, through collusion, a per-
period payoff higher than the critical value will always collude without a
leniency program. In industries where the collusion payoff is below the critical
value, collusion is not sustainable.
Before determining the optimal leniency program, the various collusive
strategies must be considered. `Normal collusion’ (N), `collude and report
systematically´ (R) or `collude and report in case of an investigation´ (I) are the
possible collusive strategies. Firms can also deviate from any collusive
agreement, which leads to multiple possible scenarios. The objective of Chen and
Rey is to focus on the deviation from these agreements and to look for the optimal
level of leniency where deviation becomes profitable.
Leniency before the start of an investigation
When no leniency after the start of an investigation is considered, only the
`normal collusion´ (N), stick to the agreement and no report, and the `collude and
report systematically´ (R) agreements are available. The optimal value of fine
reduction (𝑞̂) will be the level that makes a firm indifferent between the per-
period payoffs of the two deviations from the two possible agreements (N) and
(R).36
The optimal value (𝑞̂) decreases when the probability of success of an
investigation (αp) increases. There would also be an interval where this optimal
value can remain making the leniency policy effective. Outside this interval the
program is either too unattractive to cartel members or increases the value of the
strategies involving consensual reporting.
With the optimal leniency level (𝑞̂) a per-period payoff threshold (𝐵̂) will be
defined37
. In industries where firms can achieve a per-period payoff of collusion
higher than the threshold (𝐵̂), collusion is sustainable. Firms that cannot achieve
a collusion profit as high as the threshold will not enter into any collusive
agreement. The works looks for the optimal leniency level without any
restriction. Depending on the parameters, leniency can be a fine reduction or a
reward.
Therefore, Chen and Rey find that leniency is required before an investigation is
launched and also that the amount of leniency decreases when the probability of
success of an investigation (αp) increases. They consider that the best way to fight
35
Chen & Rey (2013) p.925 Condition (1)
36
Chen & Rey (2013) p.931 Condition (4)
37
Chen & Rey (2013) p.931 Condition (5)
Leniency Policies Jaime Plaza Aparicio
32
collusion is to crack cartels from inside. Giving incentives to deviate, such as
offering some leniency before the launch of an investigation, will help eliminate
the cartel.
This work also studies the first informant rule. Awarding leniency to more
informants decreases the per-period payoff threshold. This threshold marks the
level above which collusion is sustainable even with a leniency program. A lower
threshold will results in more “collusive” industries.
Leniency before and after the start of an investigation
When leniency is offered both before and during investigations, more collusive
strategies are available. The authors show that a program offering also post-
investigation leniency will perform strictly better when investigations are
unlikely to succeed.
As in the previous situation, the work focuses on finding the optimal leniency
levels to make deviation from collusive agreements more profitable. Two
strategies can destabilize cartel collusion: i) deviate and report in case of
investigation and ii) report in case of investigation.
Looking at the parameters that make firms want to destabilize cartels through
the first option, a critical value of success in random investigations (p) is found.38
If the probability of success of an investigation (p), exogenously given, is below
this critical value, offering some leniency during investigations will destabilize
normal collusion.
Indeed, the optimal levels of leniency, before and during investigations, will
make firms feel indifferent to any deviation from the three possible collusive
agreements ((N), (R) and (I)).
Applying this same procedure to make firms destabilize cartel collusion through
the second option, i.e. by reporting in case of an investigation (without
deviating), leaves the following conclusions depending on the level of the
discount factor (δ).
The idea is to differentiate between firms being patient (high δ) and firms being
rather impatient (low δ). In each of these two cases, depending on the level of the
investigation probabilities α and p, a specific leniency policy would be optimal.
There are three candidates for an optimal leniency policy: a leniency policy
available only before the investigation start, a leniency policy available both
before and during an investigation focused on cartel destabilization by firms
38
Chen & Rey p.934 Condition (9)
Leniency Policies Jaime Plaza Aparicio
33
deviating and reporting, and a leniency policy available also at both stages
focused on cartel destabilization through reports without deviation.
When firms are patient, but both probabilities α and p are low, the authority will
set a leniency policy designed so that firms will have incentives to report without
an early deviation. If instead one probability is high and the other low, another
leniency policy would be optimal for implementation. When there is a situation
with high p and low α, a leniency policy based only on pre-investigation reports
would be optimal. The small number of investigations launched will end
successfully most of the times and the best option would be a leniency program
that enables cartel destabilization through reports before the investigation starts.
In the opposite situation, high α and low p, a leniency policy that provides firms
with incentives to deviate from the collusive agreement and to report, either
before or after the investigation start, will be the optimal program.
When firms are less patient, the authority cannot count on a cartel destabilization
through reporting only without deviating. In such cases, firms value more
payoffs in earlier periods than in later ones. Deviating from the cartel agreement,
i.e. underbidding the cartel price, is too attractive for firms in situations with a
low discount factor (δ). Again, the authority chooses to implement one leniency
policy or the other based on the policy probabilities α and p. A high p and a low
α will promote a leniency policy, as explained in the previous paragraph, based
on reports before the investigation phase. The opposite situation will call for a
leniency policy allowing reports any time focused on cartel destabilization
through cartel deviation and report.
The conclusion of the authors is that offering some leniency before the start of an
investigation is always optimal. Offering leniency during investigations will be
optimal only under certain conditions.
The work focuses briefly on repeated offenders. The analysis does not focus on
the firms that apply for leniency more than once, but on the application of the
leniency program in the whole industry more than once. The analysis defends
the existence of an unlimited leniency program in terms of firms applying for
leniency in different periods, but only one firm obtaining it per period. In other
words, the restriction of leniency to the first party that reports in the industry will
not be effective in cartel destabilization.
Conclusion
Considering internal destabilization as the best form of cartel detection is a very
important remark in the work. The approach of other economists was to look for
an effective leniency program, since the authorities have scarce resources and
could not launch an infinite number of investigations. The approach, from the
Leniency Policies Jaime Plaza Aparicio
34
very beginning, of focusing on cartel destabilization through an effective leniency
program is a very important feature of this paper. The Antitrust Division could
have shared this view when implementing its leniency policy. As explained
before, leniency programs discover most cartels from all deterrence tools. This
work shares this idea with the Antitrust Division.
Chen and Rey defend that, under certain conditions, offering leniency after the
beginning of an investigation will increase the efficiency of the leniency policy.
Leniency programs have evolved in that direction. Being less strict in the
conditions required from firms applying during investigations was a common
characteristic of the leniency programs revisions.
They distinguish between the launch and the success of an investigation. A
competition authority in an investigation will consider the verdict as fair and
correct, whereas in some cases it would have not made the right decision.
Whether the authority makes the right decision or not is not observable. Thinking
of “real” probabilities to see which leniency policy would be the right one to be
implemented is a hard task but, in my opinion, the reasoning behind the leniency
policy decision is consistent with the view of the competition authorities.
Even so the authorities are more reluctant to guarantee leniency during
investigations. The fear of turning the leniency program exploitable could be a
reason. The truth is that leniency programs are a recent innovation in most of the
countries. The lifetime of cartels exceeds the lifetime of leniency programs. When
collusion began, leniency programs were not an issue. This work shows a very
good interpretation of the decisions firms must make nowadays whether to enter
or not into collusive agreements.
Restricting the program to the first informant only is an essential characteristic
for both competition authorities and economists. Leniency programs in the U.S.
have restricted leniency to the first party only. Awarding leniency to every
reporting member will make collusive agreements an unpunished activity.
Despite the different frameworks presented in different economic papers,
restriction to the first party is a common conclusion. As explained in previous
sections, the position of the European Commission is different. Collaborations
that help investigations should be awarded leniency regardless the entry order.
Chen and Rey also study the optimality of leniency to repeated offenders. They
could have thought collusive agreements are not a one-time decision of firms.
Entering into a collusive agreement is an illegal activity and must be carefully
considered. Multinationals can enter into and maintain multiple collusive
agreements in different markets and countries. The Antitrust Division has
offered mechanisms such as the Amnesty Plus to help firms leave not one but
Leniency Policies Jaime Plaza Aparicio
35
every collusive agreements in which they participate. The Division also punishes
any lie, if it is discovered, when a firm is asked if it is involved in any other
collusive agreement.
The model is a very good approximation of firm decisions nowadays. Having the
possibility to report at multiple stages and operating in a lot of markets, each of
them with different competition authorities, the number of available collusive
strategies increases. Competition authorities should be careful and design a
leniency policy which is as effective as possible.
3.5 Aubert, Kovacic & Rey (2005)
The authors compare leniency programs with reduced fines to programs with
positive rewards. They also study the case of awarding leniency to individuals
and how leniency programs can affect business decisions. In addition, reasons
for the puzzle of firms keeping hard evidence are given.
Firms play an infinitely repeated game. In each period firms can decide
communicate first (it is a choice). After that, firms can choose whether to collude,
if communication took place, or to compete. Firms discount payoffs using the
same factor 𝛿 ∈ (0,1). The model has the same assumption of evidence only
lasting one period. Communication (and thus collusion) leaves some evidence
that cannot be manipulated. The maximal possible fine (𝐹) does not succeed in
deterring collusion. The competition authority launches investigations with a
probability p. Moreover, it is assumed that collusive agreements will be
discovered in case of an investigation, as opposed to Chen and Rey (2005), where
investigations in collusive industries would not necessarily end successfully.
Leniency programs only reducing sanctions
Without a leniency program, collusion will be more profitable than deviating
from a collusive agreement as long as the gain from the collusive agreement is
higher than the gain from the deviation from a collusive agreement, when the
rest sticks to it. If this condition (A)39
holds, collusion is sustainable.
In the situation with a leniency program, the model assumes that a firm choosing
to report will also deviate from the collusive agreement since competition is
going to prevail in the market afterwards anyway. The reduced fine is lower than
the fine times the probability of investigation (𝑅𝐹 < 𝑝𝐹), as in any previous
model. With a leniency program, collusion will be sustainable only if the gain
39
Aubert, Kovacic, Rey (2005) p.11 Condition (1)
Leniency Policies Jaime Plaza Aparicio
36
through deviation is higher than the gain from collusion. Without this condition
(B)40
, collusion would not be sustainable.
The difference between two conditions with and without leniency program is
that the first one does not include the leniency obtained through deviation (thus
reporting).
Leniency programs restricted to fine reductions/cancellations can only deter
collusion by setting the policy parameters in a way that (A) holds and (B) not.41
The authors conclude that “leniency programs (only reducing/cancelling fines)
have no impact on the profitability of collusion and affect its sustainability only
by giving the deviators the opportunity to avoid fines from random audits”. The
probability of investigation (p) or the fine (F) have to be very large in order to
make the leniency program effective. However, in these two cases, collusion
would not be sustainable. Fear of being or fined would outweigh the profit from
collusion.
Leniency programs rewarding reports
The authors only focus on reports before the investigation phase, considering
reports during investigation as more costly to implement.
To make the analysis the easier, only deviating firms can report. The analysis
rejects the possibility of strategies based on consensual reports. The reward must
still be higher than the costs of conviction (− 𝑅𝐹 > − 𝑝𝐹) and must also
compensate the payoffs of the punishment phase that follows a report. If the
reward needed is positive (fine reduction), a full reduction of the fine would be
enough to deter collusion. The reward will tend to minus infinity if the discount
factor δ goes to 1.
Leniency programs oriented to individuals
The authors defend individual leniency programs that reward individual
informant as a complement of the corporate leniency programs. They mark
potential agency problems as an issue. The individuals who will have greater
incentives to report are the employees. Directors or Officers will have more direct
profits from collusion than employees.
Firstly, the authors suppose employees only live for one period. The idea is that
when leniency programs reward individuals firms will have to bribe these
individuals. By doing so, firms “buy” the silence of these individuals. This
additional cost decreases the benefits from collusion. Therefore, an increase in
40
Aubert, Kovacic, Rey (2005) p.11
41
Aubert, Kovacic, Rey (2005) p.11 Proposition1
Leniency Policies Jaime Plaza Aparicio
37
the reward by the competition authority decreases the value of collusion and
makes it “less” sustainable.
Even if the leniency program restricts the reward to the first individual only,
firms will have to pay the reward to every employee in the firm. If they do not
bribe every individual, the non-rewarded employees could have incentives to
report the agreement and cash the reward. Using the same reasoning, collusion
is more fragile when the number of employees increases.
When employees are “long-lived”, corporate leniency programs also play a role.
An individual leniency policy will not affect the sustainability of the collusion,
when there is no corporate leniency program. By adopting a corporate program,
the individual program can destroy collusion. This effect, as the previous one,
increases as the number of involved individuals rises. The authors point out that
the level of the rewards should be finite and not too high since rewards are not
supposed to be cashed in the equilibrium. Firms will react to the reward program
and bribe their employees with this amount so that no employee has incentives
to report (in the equilibrium).
Side effects of leniency programs
Although cartel deterrence can be achieved through reward programs42, the
authors think such programs may have some side effects. The nature of the
communication between competitors does not have to be necessarily bad. Firms
could exchange information making their competition more efficient. Also, a
reward program can have a huge impact not only on collusive decisions but on
internal firm decisions too. The authors present both the impact of the programs
on some of these decisions and the adjustments needed in the programs to limit
the impacts on the decision.
Thinking that not every communication is necessarily bad, firms can
communicate only to increase the efficiency of their competition, i.e. compete but
at a lower level of social costs. A model similar to the model first addressed in
this papers is presented. Firms can choose whether to communicate (with good
or bad intentions) or not. Communication with good intentions will lead to
higher profits in competition made by the firms and will also increase social
welfare with respect to competition without communication. Once again, firms
can deviate from a collusive agreement. If firms choose to communicate with
good intentions, the market outcome will be a more efficient competition.
As in previous models, any kind of communication leaves evidence. Firms can
report the evidence of communication to the competition authority. Since not
42
Leniency programs awarding rewards to informants.
Leniency Policies Jaime Plaza Aparicio
38
every piece of evidence is a signal of a collusive agreement, the competition
authority will have problems in classifying it. The competition authority
interprets “good” evidence as “bad”43
with a certain probability 𝜇̂ and, the other
way around, “bad” evidence as “good” with a certain probability 𝜇. 44 If the
authority does not receive any report, it will launch an investigation with
probability p, the same one as in the previous models.
By offering positive rewards to reporting firms, firms could make a false
denounce to benefit from the reward. Such firms will present “good” evidence to
the authority and cash the reward with a probability 𝜇̂. Firms will only cash the
reward if the authority misinterprets the “good” evidence and this happens with
probability 𝜇̂. An additional fine to false reports could be implemented. On the
one hand the additional fine can deter false reporting, but it can also deter the
good intention reports. This last scenario happens whenever the authority
misinterprets “bad” evidence imposing the additional fine on a firm that truly
reports “bad” evidence.
The authority should anticipate that and adjust the reward program accordingly.
It could do so by setting the reward and the additional fine so that two conditions
are met; namely, deviating firms would prefer to report any collusive agreements
and firms communicating with good intentions would find false reports
unprofitable. Such program avoids false reports and incentives the true ones.
Individual reward programs can also have some effects on internal firm
decisions. As stated before, collusion is more sustainable if employees are long
lived. Firms will therefore want employees to stay in the firm by bribing them
rather than allowing them to report. This will result in a rather fix employment
structure where informed employees will not leave the firm. Having a certain
group of informed employees staying at the same position with no internal
restructuring could result in loss of productivity. Informed employees that the
firm would want to replace will remain at their jobs. Otherwise they will report
collusive agreements and cash the rewards.
If firms must choose between entering into collusive agreement and restructuring
to compete, a high reward, as well as a high competition outcome after
restructuring, make the collusive agreement less profitable. By restructuring an
employment restructuration to achieve a higher competition payoff is meant.
Firms, at the time of the choice, are not bribing any employees and will not incur
in any cost if they leave. Restructuring will be the best choice, if the reward is
high enough. In this case a highly enough reward will deter collusion.
43
By “good” evidence, evidence of well-meaning communication is meant. “Bad” evidence refers to
evidence from a collusive agreement.
44
Aubert, Kovacic, Rey (2005) p.20
Leniency Policies Jaime Plaza Aparicio
39
When colluding firms have to choose between sticking to collusive agreements
and restructuring to compete, the level of the reward will not affect the choice. In
both cases, firms do not need to bribe any new employee. The expected fine and
the expected competition outcome after restructuring affect the decision. Firms
will stick to the collusive agreement (worst outcome for society) and will not
restructure to compete (best outcome) whenever the fine is high or the
competition gain through restructuring is moderate. The reward offered to
individuals by the competition authority will not affect these choices of already
colluding firms.
Keeping hard evidence in the firm
If firms agree to collude in the future, they, in theory, do not have any reason to
keep evidence of collusion, otherwise, the competition authority could find this
evidence when investigating. It is known that firms keep such evidence, since
leniency programs have been widely used and firms need this evidence to report.
The authors give several reasons to the puzzle of firms keeping evidence of
collusive agreements within the firm.
A reason given is the expectations of breakdown in the collusive agreement.
Firms keep evidence because they expect the agreement to be short-lived and
want to benefit from rewards in the future. This will only apply for high rewards.
Rather low rewards will not make firms want to keep evidence.
Threatening the cartel members with a report of the illegal activity to the
competition authority based on the evidence of collusion is another reason.
Designing leniency programs in a way that makes collusion unprofitable will
make firms destroy evidence. When collusion is no longer sustainable, firms do
not want to keep evidence of it because of the potential risks of being discovered.
As throughout the entire paper, the authors also focused on how the leniency
program and its consequences affect individuals. Firm employees could keep
evidence of collusion because of agency problems. Leniency programs can make
these agency problems bigger. The evidence of collusion is a very good card for
employees to play in the future when bargaining with the manager. On the other
hand, the firm can delay bonus payments, if an employee does not promise that
he will not report the collusive agreement.
Conclusions
`A leniency program that does not allow rewards to informants will not be as
effective as one that allows them´. This work shares this view with Spagnolo
(2005). Aubert, Kovacic and Rey conclude that moderate leniency programs can
truly deter cartel formation when fines or the probability of conviction are large,
Leniency Policies: Game Theory Approach
Leniency Policies: Game Theory Approach
Leniency Policies: Game Theory Approach
Leniency Policies: Game Theory Approach
Leniency Policies: Game Theory Approach
Leniency Policies: Game Theory Approach
Leniency Policies: Game Theory Approach

More Related Content

Similar to Leniency Policies: Game Theory Approach

Stiglitz
StiglitzStiglitz
StiglitzM R
 
For this Unit 4 assignment, you will prepare an APA-formatted essa.docx
For this Unit 4 assignment, you will prepare an APA-formatted essa.docxFor this Unit 4 assignment, you will prepare an APA-formatted essa.docx
For this Unit 4 assignment, you will prepare an APA-formatted essa.docxhanneloremccaffery
 
Student 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docx
Student 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docxStudent 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docx
Student 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docxdeanmtaylor1545
 
MGT 498_Ethics_Auge_Final
MGT 498_Ethics_Auge_FinalMGT 498_Ethics_Auge_Final
MGT 498_Ethics_Auge_FinalIan Auge, BSB/M
 
Response one PADM--04The authors feel that welfare should be ba.docx
Response  one PADM--04The authors feel that welfare should be ba.docxResponse  one PADM--04The authors feel that welfare should be ba.docx
Response one PADM--04The authors feel that welfare should be ba.docxronak56
 
Mayur barochiya's corporate governance and the public interest
Mayur barochiya's corporate governance and the public interestMayur barochiya's corporate governance and the public interest
Mayur barochiya's corporate governance and the public interestMayur Patel
 
Generally, In A Political Science, The Notion Of...
Generally, In A Political Science, The Notion Of...Generally, In A Political Science, The Notion Of...
Generally, In A Political Science, The Notion Of...Crystal Alvarez
 
The ‘Reality’ Of Work And How This Is Similar To Or...
The ‘Reality’ Of Work And How This Is Similar To Or...The ‘Reality’ Of Work And How This Is Similar To Or...
The ‘Reality’ Of Work And How This Is Similar To Or...Beth Hall
 
Group report prepared by Diana – Student No. 19019635Goraya –.docx
Group report prepared by Diana – Student No. 19019635Goraya –.docxGroup report prepared by Diana – Student No. 19019635Goraya –.docx
Group report prepared by Diana – Student No. 19019635Goraya –.docxbenjaminjames21681
 
Small Family Happy Family Essay In Hindi
Small Family Happy Family Essay In HindiSmall Family Happy Family Essay In Hindi
Small Family Happy Family Essay In HindiLisa Young
 
RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...
RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...
RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...GlobalCompact
 
Wassim Zhani Management Final Project Part II.pdf
Wassim Zhani Management Final Project Part II.pdfWassim Zhani Management Final Project Part II.pdf
Wassim Zhani Management Final Project Part II.pdfWassim Zhani
 
Summary Of Bachelor Thesis
Summary Of Bachelor ThesisSummary Of Bachelor Thesis
Summary Of Bachelor ThesisAgne Valeckaite
 
DBA 7035, Business, Government and Society 1 Course L.docx
 DBA 7035, Business, Government and Society 1 Course L.docx DBA 7035, Business, Government and Society 1 Course L.docx
DBA 7035, Business, Government and Society 1 Course L.docxShiraPrater50
 
DBA 7035, Business, Government and Society 1 Course L.docx
DBA 7035, Business, Government and Society 1 Course L.docxDBA 7035, Business, Government and Society 1 Course L.docx
DBA 7035, Business, Government and Society 1 Course L.docxgertrudebellgrove
 
CHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docx
CHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docxCHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docx
CHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docxchristinemaritza
 
A New Approach to Tax Compliance.pdf
A New Approach to Tax Compliance.pdfA New Approach to Tax Compliance.pdf
A New Approach to Tax Compliance.pdfJessica Navarro
 

Similar to Leniency Policies: Game Theory Approach (19)

Stiglitz
StiglitzStiglitz
Stiglitz
 
For this Unit 4 assignment, you will prepare an APA-formatted essa.docx
For this Unit 4 assignment, you will prepare an APA-formatted essa.docxFor this Unit 4 assignment, you will prepare an APA-formatted essa.docx
For this Unit 4 assignment, you will prepare an APA-formatted essa.docx
 
Student 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docx
Student 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docxStudent 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docx
Student 1 Discussionby Jyothi Adi Kamisetty - Wednesday, 21 Au.docx
 
MGT 498_Ethics_Auge_Final
MGT 498_Ethics_Auge_FinalMGT 498_Ethics_Auge_Final
MGT 498_Ethics_Auge_Final
 
Response one PADM--04The authors feel that welfare should be ba.docx
Response  one PADM--04The authors feel that welfare should be ba.docxResponse  one PADM--04The authors feel that welfare should be ba.docx
Response one PADM--04The authors feel that welfare should be ba.docx
 
Mayur barochiya's corporate governance and the public interest
Mayur barochiya's corporate governance and the public interestMayur barochiya's corporate governance and the public interest
Mayur barochiya's corporate governance and the public interest
 
Generally, In A Political Science, The Notion Of...
Generally, In A Political Science, The Notion Of...Generally, In A Political Science, The Notion Of...
Generally, In A Political Science, The Notion Of...
 
The ‘Reality’ Of Work And How This Is Similar To Or...
The ‘Reality’ Of Work And How This Is Similar To Or...The ‘Reality’ Of Work And How This Is Similar To Or...
The ‘Reality’ Of Work And How This Is Similar To Or...
 
Sample Of Comparative Essay
Sample Of Comparative EssaySample Of Comparative Essay
Sample Of Comparative Essay
 
Group report prepared by Diana – Student No. 19019635Goraya –.docx
Group report prepared by Diana – Student No. 19019635Goraya –.docxGroup report prepared by Diana – Student No. 19019635Goraya –.docx
Group report prepared by Diana – Student No. 19019635Goraya –.docx
 
Small Family Happy Family Essay In Hindi
Small Family Happy Family Essay In HindiSmall Family Happy Family Essay In Hindi
Small Family Happy Family Essay In Hindi
 
RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...
RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...
RAISING THE BAR THROUGH COLLECTIVE ACTION Anti-Corruption Efforts in Action i...
 
Wassim Zhani Management Final Project Part II.pdf
Wassim Zhani Management Final Project Part II.pdfWassim Zhani Management Final Project Part II.pdf
Wassim Zhani Management Final Project Part II.pdf
 
Corruption and fdi: a theorical analysis
Corruption and fdi: a theorical analysisCorruption and fdi: a theorical analysis
Corruption and fdi: a theorical analysis
 
Summary Of Bachelor Thesis
Summary Of Bachelor ThesisSummary Of Bachelor Thesis
Summary Of Bachelor Thesis
 
DBA 7035, Business, Government and Society 1 Course L.docx
 DBA 7035, Business, Government and Society 1 Course L.docx DBA 7035, Business, Government and Society 1 Course L.docx
DBA 7035, Business, Government and Society 1 Course L.docx
 
DBA 7035, Business, Government and Society 1 Course L.docx
DBA 7035, Business, Government and Society 1 Course L.docxDBA 7035, Business, Government and Society 1 Course L.docx
DBA 7035, Business, Government and Society 1 Course L.docx
 
CHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docx
CHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docxCHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docx
CHAPTER 4The HR Role in Policy, Budget, Performance Management, and .docx
 
A New Approach to Tax Compliance.pdf
A New Approach to Tax Compliance.pdfA New Approach to Tax Compliance.pdf
A New Approach to Tax Compliance.pdf
 

Recently uploaded

The Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh KumarThe Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh KumarHarsh Kumar
 
Call Girls Near Delhi Pride Hotel, New Delhi|9873777170
Call Girls Near Delhi Pride Hotel, New Delhi|9873777170Call Girls Near Delhi Pride Hotel, New Delhi|9873777170
Call Girls Near Delhi Pride Hotel, New Delhi|9873777170Sonam Pathan
 
Unveiling Business Expansion Trends in 2024
Unveiling Business Expansion Trends in 2024Unveiling Business Expansion Trends in 2024
Unveiling Business Expansion Trends in 2024Champak Jhagmag
 
Governor Olli Rehn: Dialling back monetary restraint
Governor Olli Rehn: Dialling back monetary restraintGovernor Olli Rehn: Dialling back monetary restraint
Governor Olli Rehn: Dialling back monetary restraintSuomen Pankki
 
SBP-Market-Operations and market managment
SBP-Market-Operations and market managmentSBP-Market-Operations and market managment
SBP-Market-Operations and market managmentfactical
 
GOODSANDSERVICETAX IN INDIAN ECONOMY IMPACT
GOODSANDSERVICETAX IN INDIAN ECONOMY IMPACTGOODSANDSERVICETAX IN INDIAN ECONOMY IMPACT
GOODSANDSERVICETAX IN INDIAN ECONOMY IMPACTharshitverma1762
 
Overview of Inkel Unlisted Shares Price.
Overview of Inkel Unlisted Shares Price.Overview of Inkel Unlisted Shares Price.
Overview of Inkel Unlisted Shares Price.Precize Formely Leadoff
 
Stock Market Brief Deck for "this does not happen often".pdf
Stock Market Brief Deck for "this does not happen often".pdfStock Market Brief Deck for "this does not happen often".pdf
Stock Market Brief Deck for "this does not happen often".pdfMichael Silva
 
magnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdf
magnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdfmagnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdf
magnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdfHenry Tapper
 
Role of Information and technology in banking and finance .pptx
Role of Information and technology in banking and finance .pptxRole of Information and technology in banking and finance .pptx
Role of Information and technology in banking and finance .pptxNarayaniTripathi2
 
fca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdffca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdfHenry Tapper
 
Economic Risk Factor Update: April 2024 [SlideShare]
Economic Risk Factor Update: April 2024 [SlideShare]Economic Risk Factor Update: April 2024 [SlideShare]
Economic Risk Factor Update: April 2024 [SlideShare]Commonwealth
 
government_intervention_in_business_ownership[1].pdf
government_intervention_in_business_ownership[1].pdfgovernment_intervention_in_business_ownership[1].pdf
government_intervention_in_business_ownership[1].pdfshaunmashale756
 
Financial analysis on Risk and Return.ppt
Financial analysis on Risk and Return.pptFinancial analysis on Risk and Return.ppt
Financial analysis on Risk and Return.ppttadegebreyesus
 
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一S SDS
 
letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...
letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...
letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...Henry Tapper
 
NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...
NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...
NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...Amil baba
 
NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...
NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...
NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...Amil baba
 

Recently uploaded (20)

The Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh KumarThe Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh Kumar
 
Call Girls Near Delhi Pride Hotel, New Delhi|9873777170
Call Girls Near Delhi Pride Hotel, New Delhi|9873777170Call Girls Near Delhi Pride Hotel, New Delhi|9873777170
Call Girls Near Delhi Pride Hotel, New Delhi|9873777170
 
Unveiling Business Expansion Trends in 2024
Unveiling Business Expansion Trends in 2024Unveiling Business Expansion Trends in 2024
Unveiling Business Expansion Trends in 2024
 
Governor Olli Rehn: Dialling back monetary restraint
Governor Olli Rehn: Dialling back monetary restraintGovernor Olli Rehn: Dialling back monetary restraint
Governor Olli Rehn: Dialling back monetary restraint
 
SBP-Market-Operations and market managment
SBP-Market-Operations and market managmentSBP-Market-Operations and market managment
SBP-Market-Operations and market managment
 
Q1 2024 Newsletter | Financial Synergies Wealth Advisors
Q1 2024 Newsletter | Financial Synergies Wealth AdvisorsQ1 2024 Newsletter | Financial Synergies Wealth Advisors
Q1 2024 Newsletter | Financial Synergies Wealth Advisors
 
GOODSANDSERVICETAX IN INDIAN ECONOMY IMPACT
GOODSANDSERVICETAX IN INDIAN ECONOMY IMPACTGOODSANDSERVICETAX IN INDIAN ECONOMY IMPACT
GOODSANDSERVICETAX IN INDIAN ECONOMY IMPACT
 
Overview of Inkel Unlisted Shares Price.
Overview of Inkel Unlisted Shares Price.Overview of Inkel Unlisted Shares Price.
Overview of Inkel Unlisted Shares Price.
 
Stock Market Brief Deck for "this does not happen often".pdf
Stock Market Brief Deck for "this does not happen often".pdfStock Market Brief Deck for "this does not happen often".pdf
Stock Market Brief Deck for "this does not happen often".pdf
 
magnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdf
magnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdfmagnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdf
magnetic-pensions-a-new-blueprint-for-the-dc-landscape.pdf
 
Role of Information and technology in banking and finance .pptx
Role of Information and technology in banking and finance .pptxRole of Information and technology in banking and finance .pptx
Role of Information and technology in banking and finance .pptx
 
fca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdffca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdf
 
Economic Risk Factor Update: April 2024 [SlideShare]
Economic Risk Factor Update: April 2024 [SlideShare]Economic Risk Factor Update: April 2024 [SlideShare]
Economic Risk Factor Update: April 2024 [SlideShare]
 
🔝+919953056974 🔝young Delhi Escort service Pusa Road
🔝+919953056974 🔝young Delhi Escort service Pusa Road🔝+919953056974 🔝young Delhi Escort service Pusa Road
🔝+919953056974 🔝young Delhi Escort service Pusa Road
 
government_intervention_in_business_ownership[1].pdf
government_intervention_in_business_ownership[1].pdfgovernment_intervention_in_business_ownership[1].pdf
government_intervention_in_business_ownership[1].pdf
 
Financial analysis on Risk and Return.ppt
Financial analysis on Risk and Return.pptFinancial analysis on Risk and Return.ppt
Financial analysis on Risk and Return.ppt
 
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
 
letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...
letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...
letter-from-the-chair-to-the-fca-relating-to-british-steel-pensions-scheme-15...
 
NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...
NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...
NO1 Certified kala jadu karne wale ka contact number kala jadu karne wale bab...
 
NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...
NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...
NO1 WorldWide Love marriage specialist baba ji Amil Baba Kala ilam powerful v...
 

Leniency Policies: Game Theory Approach

  • 1. Leniency Policies Jaime Plaza Aparicio July 13, 2015 Bachelor Thesis in Economics Department of Economics Chair for Microeconomics and Game Theory Prof. María Sáez Martí Jaime Plaza Aparicio Matriculation-number: 11-732-005 jotaplaza93@gmail.com Phone: +41786923581 Friesenbergstrasse 16 8055 Zürich Switzerland
  • 2.
  • 3. Abstract This thesis compares leniency programs according to economic theory with real leniency programs .The analysis focuses mainly on offering leniency in form of a reward as well as on offering leniency before or/and during investigations. The Folk Theorem and a competition model with leniency policy (Spagnolo 2005) are described as well as three different leniency policy papers based on that model. This thesis also reviews the intentions that lead to the appearance of leniency programs as well as the main goals of them. The two main leniency programs (U.S. and E.U.) and the differences between one program and the other are explained to show how actual leniency programs work. Conclusions lead to real leniency programs been designed similar to the ones proposed by economic theory, except for the leniency in form of a reward.
  • 4.
  • 5. Table of Contents 1. Introduction.............................................................................................1 2. Leniency Programs.................................................................................2 2.1 What is a Leniency Program? .........................................................2 2.2 Why implement Leniency Programs? ...........................................4 2.3 Evolution of Leniency Programs....................................................8 2.3.1 U.S. Leniency Program..............................................................9 2.3.2 E.U. Leniency Program............................................................13 3. Economic Theory..................................................................................19 3.1 Folk Theorem...................................................................................20 3.2 Competition Model with Leniency Programs............................22 3.3 Spagnolo (2005) ...............................................................................24 3.4 Chen & Rey (2013) ..........................................................................30 3.5 Aubert, Kovacic & Rey (2005).......................................................35 4. Conclusion.............................................................................................41 References..................................................................................................44
  • 6. Leniency Policies Jaime Plaza Aparicio 1 1. Introduction Leniency programs have changed dramatically the way cartel problems are tackled. Cartels and similar anti-competitive practices are condemn by competition authorities all around the world. The Sherman Antitrust Act in the U.S. and the Article 81 of the European Commission Treaty prohibit activities that can distort market competition. Nowadays, a competition authority without a leniency program is rarely perceived as something feasible. If properly designed, leniency programs can change the incentive structure of cartel members and provide the means for a deviation to be profitable. In addition, if programs are effectively planned, they can also prevent cartel formation. It is important to highlight that economists started studying leniency programs and focused not only on the effectiveness of them, but also on their design. Hence, this thesis focuses on the history and the design of leniency programs. Directors of the competition authorities designed the programs in a way in which they do not always harmonize with economic theory. The differences between both lines of thought are remarkably high in certain aspects, for instance, in defining the form leniency adopts. The aim of the thesis is not to decide whether a leniency program is efficient in itself or not; this work is only trying to assess the differences between the leniency programs that prevail in “the two worlds”: the real world and the one described in the economic theory. Nonetheless, it is worth mentioning that the latter partly agrees with the programs implemented in the real world, and thus this thesis will explain this “partial” agreement. Economist defend leniency programs and have some well-established views on certain aspects related to them that have not been implemented yet in real world programs. This work is divided in four sections. Section 2 focuses on the history and the development of leniency programs, and the historical events and the intentions that motivated politicians to implement such programs are also explained. This section 2 also explains the two most representative leniency programs (U.S. and E.U) and the differences between them. The economic theory is explained in Section 3, where the folk theorem and a basis model are presented. Here, three major papers based on that model and the most important concepts of them are exposed. In addition, each of the papers has some differences to real leniency programs and the work tries to capture them too.
  • 7. Leniency Policies Jaime Plaza Aparicio 2 Section 4 divides leniency programs into three parts and concludes with explaining the differences between real leniency programs and the economic theory through these three parts. 2. Leniency Programs 2.1 What is a Leniency Program? A leniency program is a tool used by competition authorities to reduce sanctions or fines for a firm involved in a collusive agreement that self-reports and provides useful information and evidence. Leniency policies are one of the tools used by competition authorities in combination with active investigations or fines in order to deter the collusion among firms in various sectors of the market. The authority offers a degree of leniency in exchange for active participation in uncovering wrongdoing. A leniency program is an automatic agreement between the competition authority and a wrongdoer conditional on the latter meeting a series of requirements. These requirements are publicly known, and by fulfilling them, the agreement is guaranteed. The agreement can be seen as an exchange between two parties. The authority grants leniency in fines, if wrongdoers report the collusive agreement and meet certain requirements. One feature of leniency programs is the clear trade-off between its benefits and costs. The benefit from inducing a “death threat” to the cartel by giving the members an escape not as harmful as the expected sanctions (fines or prison) and, at the same time, by stimulating cartels, members can profit from ending the illegal activity whilst incurring in less damage than they would without a leniency program. This underlines the importance of the proper design of the leniency programs, tipping the balance in the proper direction. This means neither being too strict nor too generous in terms of the leniency awarded. The idea behind this concept is much older than the first rudimentary leniency program from the U.S. (1978). This program was the first step to nowadays corporate leniency programs in U.S. and the rest of the world. The main concept of inducing mistrust in a group (in this case a cartel) has been very common throughout human history, especially in such situations as wartime conflicts and the illegal activities of mafias.
  • 8. Leniency Policies Jaime Plaza Aparicio 3 Since trust is required in cartels or criminal organizations to build a durable and profitable relationship, wrongdoers must be certain that individual parties will not deviate in any way from the agreement. If members do not trust one another, they cannot be confident that the agreement will be sustained in the future. Spagnolo (2005 & 2006) highlights the famous “Divide et Impera” strategy employed by Julio Caesar to defeat his enemies by bringing a few of the opposing groups across to his side, in order to make it impossible for the enemies to unite and work together, thereby breaking the opponent’s front. His model from the 2005 work can be interpreted both as an optimal leniency program intended to deter collusion among firms and to reduce criminal misbehavior among criminal organizations. Treating cartels as criminal organizations is a very good counterpoint in the analysis of leniency policies. Differences in the national and international characteristics of the Antitrust Authority (US) and European Commission (European Union) make the possible punishments entirely different. Whereas in both cases colluding firms face monetary sanctions, within US territory they also face prison sentences. Spagnolo (2006) names three essential features that cartels and organized crime share. The first entails cooperation between members of the illegal activity whereby the classic economic problems of cooperation such as free riding or moral hazards are of greater importance. These activities usually take part in several periods of time making it an ongoing relationship, whereby future benefits and costs deemed important and therefore taken into account. As members cooperate among themselves in an ongoing relationship, they end up obtaining evidence of one another’s wrongdoing. The last feature is a result of the first two. Aside from their tasks in the illegal activity, members must monitor each other. The extensive information typically obtained from this ongoing relationship, regarding wrongdoing, is crucial for leniency programs. The aim of competition authorities when focusing on leniency policies is to give members an incentive to hand in evidence. When focusing on cartels and other collusive agreements, it should be noted that not all information is incriminatory. As in González (2007), a difference can be made between circumstantial and hard evidence. Whereas circumstantial evidence is open to interpretation, hard evidence proves without any doubt the wrongdoing. To this latter group belong, e-mails/letters, records, oral statements or documents alongside other form of correspondence. Circumstantial evidence is ambiguous and cannot be taken as incriminatory. Aubert, Rey and Kovacic
  • 9. Leniency Policies Jaime Plaza Aparicio 4 (2005) cite the example of the woodpulp case. The European Commission classified the parallel evolution of prices in the woodpulp industry as incriminatory, and as an evidence of a collusive agreement, but the European Court of Justice annulled the decision, asserting that price parallelism is compatible with competitive behavior. In general, observable data without any hard evidence makes it difficult to prove collusion. That which is meant by leniency is a full or partial reduction in the fine; however, there are some regulations that go further and reward agents who report an illegal activity. The authority rewards agents that provide them with hard evidence of a collusive agreement. Although the economic analysis defends the concept that a system of rewards, instead of fine reductions, is more effective, in reality, such systems are rare. Examples of this include South Korea, which in 2002 implemented a system that grants rewards for individual informants of collusive agreements, and the US Civil False Claim Act are some examples. The Act rewards individuals who inform the U.S. government of fraudulent activity. The frauds reported directly involve the U.S. government as the purchaser of goods and services or as insurer.1 2.2 Why implement Leniency Programs? In words of the Director of Criminal Enforcement from the Antitrust Division, Scott D. Hammond2 : “[…] an effective leniency program will lead cartel members to confess their conduct to authorities even before an investigation is opened”. Before the implementation of leniency programs, reductions in sanctions or fines were in use but they were neither publicly nor general knowledge. Agents colluding or committing a crime could not take leniency into consideration since there was no leniency program in place, and trying to negotiate more lenient terms was fraught with uncertainty. Antitrust authorities had, and continue to have, other techniques to detect cartels apart from the leniency policies. These techniques are not available in every jurisdiction, and sanctions also differ among jurisdictions. Different sanctions can also increase the fear of detection in stricter regions of the world, making 1 Aubert, Kovacic, and Rey (2005) 2 Hammond (2004)
  • 10. Leniency Policies Jaime Plaza Aparicio 5 collaboration in those regions more advantageous. The Antitrust Division3 listed the cooperation of “insiders”, the assistance of skilled FBI agents in interviews, the collaboration from immigration authorities to track cartel members, and the execution of search warrants on the offices of the suspected firms as some of the tools used in their investigations. Despite this, the new version of the Corporate Leniency Program (1993) has helped to detect a greater number of international cartels than the combination of aforementioned methods available to the Antitrust Division. Spagnolo (2006) remarks that the characteristic of being “ex ante”, general, and public is what differentiates leniency programs. This makes them different in terms of their effectiveness and form from previous leniency agreements. It is “ex-ante” because the program acts a stage prior to other mechanisms such as plea bargains, which are similar but come into play at the prosecution stage. The aim of these programs is to encourage the wrongdoers not yet detected to come forward. It is “general” in so far as any agent facing such situations can adopt this program and apply for leniency. It is “public” because agents must know the existence of both the program and the situations in which they can adopt it. Spagnolo also highlights codification also as a useful instrument of these new programs. Codification helps to reduce some of the main problems of these pre- program situations; namely, discretionality and uncertainty. If firms want to apply for leniency but do not know the amount of leniency offered by the authority, or are afraid of possible changes in jurisdiction, calculating the value of reporting will be difficult. Spagnolo (2006) also summarizes the goals of well-designed leniency programs. The first goal is the prevention of cartel formation, deterring the formation of cartels by inducing mistrust in the cartel structure from the onset, so that the probability of members losing confidence increases. The second goal is oriented to cartels that were formed before, or in spite of, the adoption of the program. The authority achieves its detection of cartels by reaching an agreement with one of the members and obtains the information from this source. These two concepts are very important in terms of the analysis. For this framework cartel deterrence, and only cartel deterrence, means to prevent a cartel from being formed because the leniency program makes the illegal activity unprofitable. Cartel detection means that the authority discovers a cartel as a result of one of the members reporting it through the leniency program. 3 Hammond (2004)
  • 11. Leniency Policies Jaime Plaza Aparicio 6 With the implementation of leniency programs, applying for leniency becomes more predictable. Transparency, publicity, anonymity or generality are some of the features that new well-designed programs fulfil that helps their predictability. To assist its implementation, firms should know the existence and workings of these programs. Speeches in workshops and conferences increase the program knowledge between its potential “users”; namely, the corporations and their directors, officers and employees. Scott D. Hammond, former Director of Criminal Enforcement of the Antitrust Division of the US, explains in one of his speeches the three fundamental cornerstones for the success of a leniency program: threat of severe sanctions, perceiving a high risk of detection, and high transparency and predictability of the program. If the program is well-designed, it should succeed in real world. Although competition authorities consider the leniency program their most powerful and successful deterrence weapon, and that with its implementation the number of convicted cartels and fines has increased, there is a need to be cautious when taking into considering the programs. The increase in leniency applications, cartel detection and fines, after they have been introduced is clear and undeniable, but to characterize the program as a complete cartel detector is perhaps an overly sweeping claim. Since the number of active cartels reigning in the market is unknown, the increase in cartel conviction cannot be solely attributed to the leniency programs. The rise could come from the existence of a greater number of cartels in the market or, as the authorities suggest, from leniency programs, but a clear statement cannot be formulated. When adopting a program with full deterrence and detection of cartels, the authority ought not to receive any leniency application. This is clearly not the case, since the leniency applications rates have increased with the implementation and updates of the programs. Even assuming this success comes directly from leniency programs, another issue arises. The design of the leniency program may have played a role, which meaning that a different design could have performed even better than the one implemented. It can also be argued that leniency programs have changed nothing in cartel deterrence and detection (same probability of detection), but simply that the number of active cartels have increased, and that it is this that has resulted in a notable increase in cartel convictions.
  • 12. Leniency Policies Jaime Plaza Aparicio 7 The introduction of leniency programs has increased the number of convicted cartels4 and has also radically changed the way in which the authorities discover cartels. The deterrence effects are difficult to quantify and qualify, but the analysis is easier for determining the effects of detection. Detection effects can be seen in the applications for leniency received by the competition authorities, as for example with HENKEL, a company that produces chemical products such as detergents or shower gels. HENKEL has applied for leniency three times in Spain, but also in Italy, in France, in Germany and at the European Commission. The program has proved very profitable for HENKEL’s interest, and it subsequently assisted several competing authorities with the discovery of cartels. In Spain alone HENKEL has saved over 15 million Euros. There are undoubtedly other companies, which, like HENKEL would be willing to apply for it or that have already applied. With the application comes the investigation in the offices of the reported firms; the emails, videos and other exchange of information between the wrongdoers; the knowledge of the practices used by them to fix the price or exchange information and these not only help the authority with regards to the reported cartel, but the authority also learns about those involved firms in the cartel and their bad practices. Multinational companies that collude in one particular market may also have other collusive agreements in other markets, which proved to be the case with Henkel and its cartel’s colleagues. If HENKEL and the other companies take turns in reporting and their agreement remains unbroken in spite of the reports, the competition authority will lose with regards to the leniency program and should change the incentive structure of it. A leniency program, when it is well designed so that firms do not systematically benefit from reporting, can prove greatly beneficial to the authority through the information and knowledge that it acquires about the running of cartels. This only happen in the worst case-scenario, whereas the program could actually deter and detect cartels at a very high rate. Despite the success and evolution of leniency programs, competition authorities did not dare to set leniency as a reward. High fine reductions or fine cancellations are the form leniency takes in the policies of the authorities. If the authority is willing to refrain from levying the full fine against a wrongdoer, why should it not go further by setting a reward? Through its leniency policies the authority 4 Spartling (1998), Delrahim (2004)
  • 13. Leniency Policies Jaime Plaza Aparicio 8 reveals that discovering an entire collusive agreement compensates the condonation of the sanction to a confessed wrongdoer. If the authority is willing to pay that price for cartel detection, why do not increase this effect by setting a reward instead of a reduction/cancellation? Aubert, Kovacic, Rey (2005) name certain possible implantation issues for rewards. A limited budget can impede the possible reward but the reward can be paid through the fines imposed on the other cartel members. There may be public condemnation of rewarding wrongdoers, especially at a time when a country is in recession. A third issue is the additional collusion incentives that rewards produce. If with a fine reduction firms can adopt collusive strategies to exploit the leniency authority, a reward can increase incentives to collude and loot the competition authority. 2.3 Evolution of Leniency Programs The aim of the previous sections is to explain what changes for the wrongdoers with the introduction of the programs, the characteristic of ideal programs and the trade-offs and mistrust induced by them. This section focus on how real leniency programs are designed, their evolution the years that have passed since the first programs were adopted and the differences between the main leniency programs. Figure 2 shows the year of adoption of a leniency program in each country and the percentage of countries from the total with leniency program in that year. In less than two decades most of the countries have adopted a leniency program. These programs can greatly differ between countries and can be more or less generous or restrictive. Since the revision of the Corporate Leniency Policy in 1993 more competition authorities in national and international regulations have adopted a leniency program for firms. Since, in this section, the approach is going to focus on how real programs are designed there must be clarity about the users of the program. When wrongdoer or agents was mentioned, it primarily referred to firms.5 Despite the existence of individual leniency policies, the majority of leniency programs are oriented towards corporations. This work focuses principally on leniency treatments to 5 For further analysis whenever wrongdoer or a similar term is mentioned is meant firms. Whereas the term individuals, and only it, will be used for persons.
  • 14. Leniency Policies Jaime Plaza Aparicio 9 firms; nevertheless, in some jurisdictions the leniency to corporations implies the leniency treatment extends to individuals. Leniency programs have changed with many useful revisions, but the pioneer is the U.S. The U.S. program has been the model for the rest of the programs. The main two competition authorities with regards to leniency regulation are the Antitrust Authority of the U.S. and the European Commission of the European Union (E.U). The following sections will focus on both leniency programs and their differences and similarities. 2.3.1 U.S. Leniency program The first version of the Corporate Leniency Policy dates from 1978 and was introduced by the U.S. Department of Justice (DoJ). Within the DoJ the Antitrust Division is the responsible for the policy. Wrongdoers reporting illegal activity before an investigation had been launched were eligible to obtain leniency. Amnesty6 was not granted automatically and the Division had great deal of prosecutorial discretion in the process. The program was not very appealing and this resulted in a very poor number of applications (only one application per 6 For this section, as in S. Hammond (2004), the terms amnesty, leniency and immunity will refer only to a full elimination of the fine and not to partial reductions. Figure 1: Borrell, García, Jiménez (2013)
  • 15. Leniency Policies Jaime Plaza Aparicio 10 year) and it was unable to discover an international cartel. Both theorists and directors of competition agreed the program had some errors in its design. As Spagnolo (2006) highlights, the program was neither transparent nor “automatic”. Given that the program was unable to bring firms to apply for leniency, the Division revised the Corporate Leniency Policy in 1993. The aim was to make it easier and make applying for leniency and cooperating more attractive to firms. Three aspects, as stated by several members of the Division7 , changed in the revision that exist and have made the program what it is today. i) Automatic amnesty if there is no investigation already launched. ii) Amnesty could still be awarded even if firms cooperate during an investigation. iii) Every individual director, officer and employee from the firm that obtains leniency is protected from criminal prosecution. The application rate quickly jumped to one application per month and reached rates near 20 applications per year8 . With this new face the Corporate Leniency Policy became the most effective tool in detecting international cartel cases. Branches from the same cartel came to the Antitrust Division within the same weeks, and even within the same day9 , but only the first to come was granted leniency. Spagnolo (2006) comments on the huge impact of the revision in the program. Together with the increase in the application rate, the magnitude of the penalties imposed increased dramatically. Supported by OECD reports that state that the increase in the use of leniency programs is powered by the substantial increase in penalties, Spagnolo asserts the two forces operate together. Spagnolo also defends the notion that the higher quality and quantity of evidence in hands of the Division was crucial for impose higher sanctions. Finally, higher sanctions make the risk of getting convicted greater and therefore the program proves more appealing. 7 Spratling (1998), Hammond (2000), Delrahim (2004). 8 Hammond (2004) 9 It did not happen in the U.S. but HENKEL and other company applied the same day for leniency; namely, the day the leniency program was first adopted in Spain.
  • 16. Leniency Policies Jaime Plaza Aparicio 11 This new version of the program, still currently in use, is divided into four parts10 . Leniency before an investigation has begun, alternative requirements for leniency, leniency for corporate directors & employees, and leniency procedure. A firm involved in an illegal activity11 , looking to be awarded with full immunity, that is not known by the Division ought to meet several requirements. The firm must i) be the first from an unknown cartel to report illegal activity; ii) take quick and effective action to end its participation in the illegal activity; iii) completely report to the Division and cooperate with it throughout the investigation; iv) confess as a truly corporate act; v) where possible make restitution to injured parties; and vi) have participated in the illegal activity without coercing other parties to take part in it, which means, among other things, being neither the leader or originator of the illegal activity. A firm reporting the illegal activity after the beginning of the investigation has a further opportunity to achieve immunity. The Division grants immunity to a firm that reports even if the investigation has already started, but it must not have incriminatory evidence at that point of the investigation. This firm must i) be the first to report and qualify for leniency from the members of the illegal activity reported; ii) report at a time when the Division has no evidence against the company that serves for the conviction of it; iii) take quick and effective action to end its participation in the illegal activity; iv) completely report to the Division and cooperate with it helping advance the investigation; v) confess as a truly corporate act; and vi) where possible, make restitution to injured parties. Other than that the Division must determine that awarding leniency to that firm, in consideration of the illegal activity, will not be unfair to others. Being the leader or having coerce others to participate as well as reporting after an investigation has started can jeopardize the granting of leniency by the Division. If a firm qualifies for leniency and meets all requirements its directors, officers and employees, having admitted their involvement in the illegal activity, will receive leniency, which means that they cannot be charged criminally. Qualifying for leniency can save the corporation millions of dollars and also eliminates the threat of prosecution that would have persisted by continuing with the illegal activity. 10 Corporate Leniency Policy (1993) 11 For sake of clarity, when in previous sections the term “cartel” was mentioned, it does not meant only cartels but every other form of collusive illegal activity disturbing competition that is condemned by the Antitrust Division.
  • 17. Leniency Policies Jaime Plaza Aparicio 12 Qualifying for leniency also entailed certain costs for the firms. Besides interrupting the typically profitable illegal activity, firms could be the target of federal and state-treble damage claims based on their involvement in the cartel. As mentioned by Spagnolo (2006), applicant firms could find themselves not only liable for three the damages suffered by customers but also thrice that suffered by the costumers of their conspirator. With a new legislation approved in 2004, Criminal Penalty Enhancement and Reform Act, that tries to solve this cost, the Division wanted to give firms a greater incentive to report. The new legislation reduces potential damages in private lawsuits to single damages based on their own role in the cartel for the receiver of leniency, whilst other participants remain fully liable for treble damages12 . As stated by Spagnolo (2006) this new legislation increases the potential liability for colluding firms that do not apply for leniency. In words of the Deputy Assistant Attorney General of the Antitrust Division M. Delrahim13: “the other cartel participants remain fully liable for treble damages based on harm caused by the entire conspiracy.” Within the leniency program there are some mechanisms that make reporting even more attractive to firms involved and which could result in more cartels being convicted. The Amnesty Plus Program encourages firms involved in ongoing investigations to seek out amnesty in other markets in which they compete. By qualifying for leniency (the Amnesty) in other uninvestigated market, and, precisely, by collaborating with the Division in the discovery of this new cartel firms will get a reduction (the Plus) on the fine imposed for the participation in the first cartel. The Penalty Plus Program works in the opposite direction to the Amnesty Plus Program. If a firm, which is been investigated and participates in cartels in other markets does not profit from the opportunity given by the Amnesty Plus and denies any other illegal activity, it will suffer harsh consequences if the Division discovers its implication in any other cartel. That which is meant by the Plus is that if the lie is discovered, the firm will not be condemned with the “regular” fine or jail sentence of each cartel, but the Division will pursue a harder level of sanction. 12 Delrahim (2004) 13 Delrahim (2004)
  • 18. Leniency Policies Jaime Plaza Aparicio 13 Through its directors, the Division has espoused the success of the program. Its work has changed a lot with the revision of the program in 1993 and such statements are very useful in illustrating how the revision has affected the behavior of the firms with respect to the leniency program. In 1994 the Division also published the Individual Leniency Policy.14 The program is oriented to individuals that approach the Division representing themselves, and not a corporation, and report an illegal antitrust activity. In the case of individuals, leniency means freedom from any criminal charges. As with the Corporate Leniency Policy, individuals must meet several requirements that run along similar lines to those required by the corporations. Currently, when the individual applies for leniency and reports an illegal antitrust activity i) the Antitrust Division must have not received any information about the illegal activity being reported; ii) the report must be complete and collaboration throughout the investigation must be offered to the Division; and iii) the participation in the illegal activity must be free from coercion to other parties. Firms usually run a race against the other cartel members to apply first for leniency. With the introduction of the Individual Leniency Policy, they also run a race versus their own employees. One of their employees could approach the Division, win the race and qualify for leniency leaving the firm defenceless. Directors of the Division15 have suggested that the utility of the program lies in the number of corporate applications it produces. Firms, afraid of losing the race against their employees, could apply for leniency as a corporation, securing leniency for both the corporation and all its employees. The Division values this more than the possibility of a great number of individual leniency applications. 2.3.2 E.U. Leniency Program The old continent took longer to implement a leniency program and revised it more than the U.S. The different versions of the E.U. programs take up more pages than those of the U.S., which are clearer and easier to read. 14 Individual Leniency Policy (1994) 15 Hammond (2004)
  • 19. Leniency Policies Jaime Plaza Aparicio 14 Notably, the E.U. Leniency program is oriented towards firms involved in cartels that operate in more than one country of the European Union. If the cartel only operates in one country then firms must seek leniency from the competing authority of that particular country. In 1996, three years after the revision of the U.S. program, the European Commission published the Commission Notice on the non-imposition or reduction of fines in cartel cases16 . This notice explained the conditions that firms applying for a reduction should fulfil. Applying for a fine reduction or leniency in this first version came with uncertainty. Despite some discussion, the notice set some basics for the following versions of the program. This first version only allowed for firms to obtain the total fine reduction providing they reported before an investigation had been launched and when the Commission did not have sufficient information to prove the existence of a cartel. Firms had to meet several conditions. Ending its participation in the illegal activity; being the first to report decisive evidence; providing the Commission with all relevant information available as well as extending the cooperation through all the investigation; having participated in the cartel without being the instigator or coercing other parties; and reporting prior to all other firms involved in that cartel were the important requirements that firms had to meet. If a firm met all this conditions, it could have benefited from a reduction of at least 75% of the fine that it could have received without the program. If firms come to report after the investigation has started, but meet all the aforementioned conditions, it still could have benefited from a reduction of between 50% and 75% of that fine. The Commission also gave reductions of fines between 10% and 50%, if firms that did not meet the named conditions cooperate in the investigation. Firms from 4317 different cartels applied for leniency between 1996 and 2014, with the years 2001 and 2002 being when most firms applied for a reduction. This first version lacked clarity, particularly with regards to the type of information that was deemed necessary to report and the precise nature of the process. Firms applying for a reduction were not sure about the amount discounted from the fine. This problem also appeared in the first version of the U.S. program in 1978. The Commission had a lot of discretion in the process resulted in lot of uncertainty regarding applying for it. 16 Commission Notice (1996) 17 Borrell, Jiménez, Ordóñez (2014)
  • 20. Leniency Policies Jaime Plaza Aparicio 15 For these reasons the Commission revised the Commission Notice of 1996, and in 2002 made the Commission Notice on immunity from fines and reduction of fines in cartel cases18 public. They admitted that, according to their judgement, the 1996 version was effective but could be improved by increasing its transparency and certainty. One desired feature of the program was to clarify the conditions (not for a fine reduction, but) for the immunity from fines. This new version modified some important aspects of the former one. The Commission granted immunity, as opposed to a high fine reduction, if the firm was first in reporting the illegal activity to the authority, and providing that they did not already have sufficient evidence to incriminate the cartel. Conditions like not having coerce any other party, ending the participation in the illegal activity from the point the cooperation starts or cooperating through the entire process were also in this second version. This second version introduces several novelties. It was open to both instigators and ringleaders. Only parties that coerced others were restricted from applying. It introduced the possibility to apply in hypothetical terms. In this kind of application, the reporting firm must present a list of evidence it proposes to reveal at a later agreed date. The documents shown are free from sensitive information and are used to illustrate the nature of the evidence. The Commission could withdraw the immunity, or fine reduction at any moment during the investigation if the firm fails to meet the conditions required. In this second version, firms that do not meet the mentioned conditions (a basic requirement, or that apply after an investigation has started) can apply for a fine reduction, but are rejected from the possibility of immunity. These firms must report the collusive agreement with evidence that adds significant value to the evidence already available to the Commission. The Commission Notice (2002) tries to clarify what is meant by significant value in order to reduce any uncertainty and to increase the transparency of the program. Firms that report after the first can still benefit from a reduction in fines. The first firm will get a reduction of between 30% - 50%, the second firm between 20% - 30% and, at best, any firm thereafter will obtain a maximum reduction of 20% or less. This version of the program has been used by firms in relation to 29 different cartels since 200219 . This second version still retained the basic elements introduced by the first version in 1996 but increases transparency and certainty by clarifying its conditions, the nature of the information required and by implementing new mechanisms, such as a proper understanding of the application in hypothetical terms. All this to make it easier for firms to end their participation in cartels since, 18 Commission Notice (2002) 19 Borrell, Jiménez, Ordoñez (2014)
  • 21. Leniency Policies Jaime Plaza Aparicio 16 as stated in their Notices, the Commission is aware of the existence of firms willing to finish its participation in the illegal activities. In 2006 the Commission published the Commission Notice on immunity from fines and reduction of fines in cartel cases (2006) going a step further in its work to increase the transparency and clarity of the program and implement new mechanisms to facilitate the report. The fact that the program does not lack transparency and certainty is very important. Reporting their involvement is a great risk for firms since it leaves them exposed. Firms declare to the competition authority and confirm their involvement in an illegal activity that has proved detrimental to consumers and had great social costs. It is evident that the greater that a firm perceives the risk of making a report backfiring, the more likely they are to back away and refrain from doing so. The 2006 version tries to eliminate uncertainty and provide applicants with greater guidance owing to the evident shortcomings in former versions of the program. The information handed to the Commission when seeking for leniency must be sufficiently substantial that the Commission can carry out a targeted inspection pertaining to that cartel, or can find an infringement of Article 81 EC20 . Article 81 EC states the behaviours that are incompatible with competition or common markets. Agreements of firms fixing prices (directly or indirectly), limiting the production, or sharing sources of supply among other practices are considered as conforming to such behaviour. With regards to the reductions of fines, the information handed to the authority will be more valid, if it does not need corroboration. Flexibility in the process, restringing evidence manipulation or obligating the continuous cooperation for both immunity and fine reduction applicants are some of the new features of the program. Whereas in the 2002 version the novelty lay in the application in hypothetical terms, in 2006 the Commission introduced the marker system. Firms can apply by sending only limited information that must be completed in the future, but they secure their approach position to the Commission. This means that a firm can arrive first and report with limited information that must be completed upon in the future and, in doing so, secure its place first for immunity. Since 2006 firms have applied for leniency with regards to 9 different cartels21. 20 European Commission Treaty 21 Borrell, Jiménez, Ordóñez (2014)
  • 22. Leniency Policies Jaime Plaza Aparicio 17 Each new version of the program has come with a fewer number of sentences sanctioning cartels by the authority. As stated in section 2.2, a reason for this could be that program is becoming more efficient with each consecutive revision. However, this might also be explained through firms that have fear of incurring even greater fines. This requires for further analysis and interpretation that goes beyond the parameters of this work. 2.3.3 Differences between programs The two main Leniency programs have some differences both in the prosecution and the design of the program. Spagnolo (2006) names some of the differences between the U.S. and the E.U. program. Ringleaders The leniency program in the E.U. is not restricted to ringleaders, whereas the U.S. the program does not allow them to apply for leniency. This fact can create differences in the formation of cartels. Firms might perceive taking leader positions as being overly risky. Furthermore, it raises the question of how the will define the role of the ringleader. With unclear criteria, firms might be unwilling to enter to cartel or, perhaps more clearly stated, one might posit that no firm is likely to want to take a leading role since that means rejecting any possible future leniency agreement. Milder leniency If the information reported by a firms adds additional value to the investigation, regardless of the order of the report, the European Commission awards the informant with some form of leniency. At the very least, a reduction of the imposed fine can be achieved by reporting significant information. The U.S. program denies this possibility. It only allows the first reporting firm to obtain full immunity. This is a very important point. The “winner takes it all” approach from the U.S. increase the incentives to report first. The position of the European Commission22 is that collaboration has an intrinsic value that must be awarded with immunity or a reduction in fines. In the E.U. reducing sanctions not only to the first firm makes that strategies like “wait and report, if somebody reports” increase its value against “report first”. If the difference in terms of leniency and fine reduction between reporting first and second increases, the incentive to report first will also increase. Spagnolo also comments that this approach from the E.U. will lead to lower fines paid by cartel. 22 Commission Notice (2002)
  • 23. Leniency Policies Jaime Plaza Aparicio 18 Immunity Whereas, in the U.S., the first reporting firm obtains automatic leniency, in the E.U. firms reporting after an investigation has started may obtain leniency, depending on the quality of the information. By quality is meant the novelty of the information provided which that adds to that already possessed by the Commission. In the E.U. there is a link between the value of the information reported and the awarding of leniency. Spagnolo argues that there is in fact no difference. The U.S. program sets conditions about the information reported. These conditions determine that the information reported must have high value to the Division either because it proves the existence of an unknown cartel or else because it helps the Division in the investigation of a cartel where the Division does not a have lot of evidence. Individual liability The Individual Leniency Policy complements the corporate program in the U.S. The possibility of individuals, employees, directors or officers, obtaining leniency in the U.S. can generate agency problems in the firms. The E.U. does not offer such possibility so that individuals cannot approach the Commission seeking leniency for themselves, but for the corporation as a whole. One more abstract reason, which complements itself with individual liability, is the difference in the perceived risk of the two programs because of the possible sanctions as stated in Hammond (2000 & 2004). The Antitrust Division is a national institution, whereas the European Commission comprehends a high number of countries. Cartel activity is treated as an administrative offense in the E.U. and the highest possible sanction is a very high fine. Nevertheless, cartel activity is treated as a criminal offense in the U.S. With this explanation, it is clear that, since the E.U. cannot incarcerate individuals, the fines imposed by the Commission have to be much larger than the ones imposed by the Antitrust Division. The Commission has to increase the level of fines. The other alternative would not prove attractive to the applicants. In the U.S., firms that do not obtain leniency often pay fines equal to 30% of more of the revenue generated by the sale of the “cartelled” product during the entire lifetime of the cartel23 . On the other hand firms unable to obtain any reduction in the E.U. have to pay a fine approximating 10% of the worldwide turnover for every product sold by that firm24 . It is clear that the difference between obtaining leniency and a reduction could save the company a lot of money. 23 Hammond (2004) 24 Hammond (2000)
  • 24. Leniency Policies Jaime Plaza Aparicio 19 Choosing between high fines and incarceration may be an easy decision, as appears to have been the case for a lot directors of corporations which, according to Hammond (2004), operated in cartel in Europe, Asia and other places around the world but avoided operating in a cartel in the U.S. In some of these cases the U.S. market was potentially the most profitable market for them but they refuse to operate there, and the reason given by the directors of the cartel firms for the avoidance of the U.S. on the cartel map was the possibility of getting caught and going to jail in the U.S. Hammond (2004) states that the leniency program of the E.U. is a very good one for a jurisdiction that does not have criminal sanctions. Even so, he considers that a jurisdiction with criminal sanctions and individual liability will always be more effective at inducing amnesty applications. The fear induced by the U.S. program thanks to the possible incarceration speaks for this statement. 3. Economic Theory After the Antitrust Division revised the Corporate Leniency Policy in 1993 and three years later with the adoption by the E.U. of its own leniency program, the programs became well-known. Firms started applying for the program and economists started studying the program. Economists tried to figure out whether the programs set the proper incentives to the applicants or not. The programs have to be designed in a way that they benefit the authority and not every cartel member, but the leniency receiver. Concepts like “winner takes it all” or “exploitable programs” were on the table. Economists have tried to produce models which approximate real world situations in order to see how much leniency has to be offered, who can apply for that leniency, when participants can apply for leniency and, even, if offering leniency was the right decision. Economists also have considered going a step further and not offering leniency but a reward, whereas competition authorities did not consider this possibility. This thesis chooses certain models and findings from an extensive bibliography of papers that deal with leniency programs. The models shown in the papers are based on an infinitely repeated interaction among firms. By proposing environments with and without leniency programs and defining fines applied and resources available to the competition authority, the authors take some conclusions on how well-designed leniency programs should be designed. Taking into account the idealized world described in the models, the aim of this
  • 25. Leniency Policies Jaime Plaza Aparicio 20 Figure 2: Prisoner's Dilemma section is to compare the results obtained by the authors with the “real world” programs. Even some statements by directors of the Antitrust Division on how leniency should be are challenged by the economists. There is also a debate among economists regarding the design of leniency programs. Most of the models on leniency programs are based on the infinitely repeated interaction among firms; therefore, it is necessary to explain how this assumption “changes the game” dramatically. Large established corporations operating in multiple markets around the world meet each other and exchange information or agree on market shares. Even if one of these agreements is detected by a competition authority, the firm will probably pay the fine and make new decisions on further collusion or collaboration, but certainly will not disappear. These big corporations will make those decisions throughout all their lifetime. The Folk Theorem is an important concept to be emphasized when dealing with infinitely repeated interactions. 3.1 Folk Theorem A game with a unique Nash Equilibrium played a finite number of times by rational players will yield the same result in every period; namely, the Nash Equilibrium. Looking at the Prisoner’s Dilemma, (D,D) is the unique Nash Equilibrium. However, both players would be better off playing (C,C). Making credible that both will play C is not possible, unless there are mechanisms to punish a deviation (enforceable contracts) or the interaction lasts for a finite number of periods. If the time horizon changes to an infinitely repeated interaction, players will want to play (C,C) under certain conditions.
  • 26. Leniency Policies Jaime Plaza Aparicio 21 The Bertrand Competition will better illustrate this effect. There is a market with two firms that produce perfect substitutes. The firm that sets the lower price wins the whole market. Both have the same production costs per unit of c. They interact with each other T periods of time with T = ∞. In each period of time, the firms choose prices simultaneously. Choices may depend on former decisions, i.e., on the prices set in the previous periods. A firm i will want to maximize the presented discounted value of its payoffs, i.e. ∑ 𝛿 𝑡−1𝑇 𝑡=1 𝛱𝑖 (𝑝𝑖𝑡, 𝑝𝑗𝑡) . The payoff of a firm i depends on the payoff of both firms i and j. Furthermore, (p1, p2) = (c, c) is the only Nash Equilibrium in the one-shot game where the incentive to deviate outweighs every other strategy. The monopoly price pm can be sustained as an equilibrium price through the trigger strategy “set pm in the first period and continue with pm as long as both firms have set pm, otherwise set p = c”. With this strategy, each firm obtains half of the monopoly profit each period. The present discounted value of this strategy is 𝛱 𝑚 2 ( 1 + 𝛿 + 𝛿2 … ) = 𝛱 𝑚 2 1 1− 𝛿 . The alternative for a firm is to slightly underbid the monopoly price and to win the entire market. If a firm underbids pm in the first period, it can count on competition, and a payoff of 0, for the remaining periods. Even so, this firm will obtain the entire monopoly profit in the first period. Therefore the present discounted value of this strategy is 𝛱 𝑚 . The second strategy is a deviating strategy. For this section and the next ones, a deviating firm will be a firm that was initially involved in a collusive agreements, but deviates from the agreement setting a market variable which is not in line with collusion and making higher profits with it. In the Bertrand Competition a deviating firm will underbid the collusive price. A comparison of the value of both strategies shows that the trigger strategy is better off when 𝛿 ≥ 1 2 . Then 𝛱 𝑚 2 1 1− 𝛿 ≥ 𝛱 𝑚 → 𝛿 ≥ 1 2 . Applying this reasoning, any other price between c and pm can be an equilibrium price.
  • 27. Leniency Policies Jaime Plaza Aparicio 22 Figure 3: Folk Theorem in the Bertrand Competition The Folk Theorem states that every pair of profits Π1 > 0 and Π2 > 0 so that Π1 + Π2 ≤ Πm is an equilibrium payoff in each period for δ sufficiently close to 1. In other words the grey zone contains every sustainable per-period payoff for δ close to 1. This reasoning can be applied for the infinitely repeated interaction between firms under a leniency program. If the competition authority does not have a leniency program and does not carry out any investigations, firms will have no impediment to collude and will probably do so. A well-designed leniency program will change the per-period payoffs through fines and fines reductions to make reporting more profitable. 3.2 Competition Model with Leniency Programs The following model is described in Spagnolo (2005). It is a dynamic model based on an economy made up of a lot of oligopolistic industries that, at the same time, consist of a number of risk neutral agents (firms). There is also a benevolent legislator (competition authority) who controls the market and determines the policies. A discounted infinitely repeated oligopolistic game between the agents takes place in each of the industries. The legislator is in charge of setting the parameters of the law enforcement policy. Any collusive behaviour that reduces the social welfare is forbidden. Firstly, the legislator sets the enforcement parameters and afterwards, having observed the parameters, the agents begin their interaction. Agents live infinitely and discount their future payoffs through a factor δ that is common to all of them.
  • 28. Leniency Policies Jaime Plaza Aparicio 23 The market game played every period to the infinity can be interpreted as a Bertrand or Cournot competition. Competing is the only Nash Equilibrium in the one-shot game. Firms competing obtain a per-period payoff of 𝜋 𝑛 . However, a higher payoff could be reached if players can make a collusive agreement. 𝜋 𝑑 and 𝜋 𝑐 denote the period payoffs of deviation and collusion, respectively. A collusive agreement (e.g. a cartel between agents fixing prices) reduces the social welfare and is, therefore, forbidden by the legislator. Firstly, a world without law enforcement will be considered. Firms can sustain collusive agreements in a subgame perfect Nash Equilibrium, if the value of collude indefinitely (Vc) exceeds the value of a unilaterally defection (Vd) from this agreement. The discounted sum of expected payoffs must be compared 𝑉 𝑐 = 𝜋 𝑐 1− 𝛿 > 𝜋 𝑑 + 𝛿𝑉 𝑝 = 𝑉 𝑑 . Or, normalized by (1 − 𝛿), 𝜋 𝑐 > (1 − 𝛿) 𝜋 𝑑 + 𝛿𝑣 𝑝 , where πd denotes the value from defection when the others stick to the agreement, Vp denotes the value expected, i.e. the discounted sum of payoffs, at the beginning of the punishment phase that follows the unilateral defection, and 𝑣 𝑝 denotes the time average payoff of the punishment phase with 𝑣 𝑝 = (1 − 𝛿) 𝑉 𝑝 . Since to make collusion credible agents must penalize a deviation, it follows 𝜋 𝑑 > 𝜋 𝑐 > 𝑣 𝑝 . After a deviation, the punishment stage always come. With 𝜋 𝑏 denoting the payoff of sticking to the collusive agreement in a period when an agent defects, it also follows 𝜋 𝑑 > 𝜋 𝑐 > 𝜋 𝑛 ≥ 𝜋 𝑏 . It can be argued that without a law enforcement policy agents will find the collusive agreement more profitable. As described before, collusion requires communication and this will produce some evidence every time agents meet. With a law enforcement policy and a leniency program, the evidence generated will become of greater importance. The law enforcement policy also affects the communication among the cartel members and changes the timing of the game. At a first stage, cartel members, having observed the policy parameters, exchange information and update and confirm collusive strategies, generating evidence of a collusive agreement. At a second stage, the same members set the market variables and choose whether to report the collusive agreement or not. In the first case a reporting member will apply for leniency through the leniency program. The evidence produced disappears at the end of the period. However, as long as
  • 29. Leniency Policies Jaime Plaza Aparicio 24 the evidence is valid, cartel members can transfer it to third parties at no cost. If the legislator gets a report with evidence of a cartel, it is convicted with probability one. The times the Legislator gets a report, it will go public at the end of the period so that every agent in the industry will know who reported it. The parameters of the law enforcement policy presented by Spagnolo are the following:  A monetary fine F that convicted firms have to pay. This fine F is restricted to be in an interval with 𝐹 ∈ [ 𝐹, 𝐹 ].  A reduced fine RF that a cartel member has to pay when he discloses information to the authorities enabling the cartel to be convicted. The reduced fine (positive transfer or reward if RF < 0) is also restricted to be in an interval with 𝑅𝐹 ∈ [ 𝑅𝐹 , 𝐹].  Beside the leniency program, the legislator can carry out investigations and has the following (per-period) probabilities of conviction. - α, denotes the probability of a cartel being convicted when no cartel member reports the illegal activity. – γ, denotes the probability of cartel members that unilaterally deviate from the agreement being convicted. – β = γ + η, with η ≥ 0, denotes the probability of a cartel member that sticks to the collusive agreement in a period when at least another member deviates being convicted. To focus on realistic parameters Spagnolo assumes α, β, γ < 1 2 . Whereas administering fines has no cost, the implementation of such probabilities has some costs. Ck(k) denotes the social costs of implementing the probabilities k, with k ∈ (𝛼, 𝛽, 𝛾). In addition, it is also assumed that 𝐶 𝑘(0) = 0, 𝐶 𝑘 ΄ (𝑘) ≥ 0, 𝐶 𝑘 ΄΄(. ) > 0. Costs rise with any new increase in the probabilities, but the increase is higher when costs are at a low level. 3.3 Spagnolo (2005) The paper from Spagnolo, on which the model from the previous section is based, focuses on leniency programs and their ability to undermine trust, making the deviation from a collusive agreement more profitable. Spagnolo differentiates between courageous and moderate leniency programs. The first one gives a reward to the reporting party financed by the fines from the rest of the cartel members. Moderate programs only reduce or cancel sanctions and do not allow for rewards.
  • 30. Leniency Policies Jaime Plaza Aparicio 25 In a world without leniency programs but with investigations from the authorities, collusion can still be sustained. Namely, the value of colluding (VC ), which is the discounted expected per-period payoff of collusion minus the probability of paying the fine, should exceed the value of defecting (VD ), which is the payoff of deviation minus the probability of getting convicted plus the discounted sum of payoffs in the punishment stage, i.e. 𝑉 𝑐 = 𝜋 𝑐−𝛼𝐹 1−𝛿 > 𝜋 𝑑 − 𝛾𝐹 + 𝑉 𝑝 = 𝑉 𝑑 .25 Observing this condition, it follows26 that for γ below a certain threshold (think of a very low γ) is optimal for cartel deterrence to set the maximal possible fine (𝐹). For a higher γ (even higher than α), increasing fines does not increase cartel deterrence, but deviation deterrence. Members will stick to collusive agreements because deviation has a high probability of conviction (γ). Since increasing γ is costly, the optimal policy leads to γ = 0. For further analysis, Spagnolo assumes 𝛼 > 𝛾 and the optimal fine set at its maximal possible level. Leniency effects Spagnolo also names two possible characteristics of leniency programs; namely, exploitable and effective leniency programs. An exploitable leniency program makes members of a collusive agreement benefit from reporting it.27 If the program is too generous (low RF), cartel members could consensually report the cartel to the authorities but may keep colluding and reporting in the following periods. An effective leniency program makes deviating firms increase their payoffs through a report.28 When a firm deviates from a collusive agreement, it can report the collusion or remain silent. Within an effective program, firms will always report the cartel after a defection. For deviating firms this means that the probability of being convicted and paying the fine should exceed the reduced sanction (𝑅𝐹 < 𝛾𝐹). Spagnolo calls this the protection from fines effect. Courageous Leniency programs As explained before, Spagnolo classifies leniency programs into courageous and moderate leniency programs. Courageous leniency programs are self-financing, i.e. the reward is paid by the fines imposed on the other cartel members. Observing the possible effects of leniency programs, it turns out that restricting eligibility to the first reporting party reduces the set of exploitable programs. A leniency 25 Spagnolo (2005) p.14 Condition 2 26 Spagnolo (2005) p.14 Proposition 1 27 Spagnolo (2005) p.15 Definition 2 28 Spagnolo (2005) p.16 Definition 3
  • 31. Leniency Policies Jaime Plaza Aparicio 26 program looking for optimal cartel deterrence will restrict leniency to the first party only29 . The introduction of leniency programs creates an incentive compatibility condition (IC)30 to colluding agents. The value of any kind of collusion (both the regular one (VC ) and the consensual reporting one (VC/ )) should exceed the value of deviation (both deviation with (VD ) and without (VD/ ) reporting after deviation). A collusive agreement is only possible if this condition is met. max { 𝑉 𝐶 , 𝑉 𝐶/ } > max{ 𝑉 𝐷 , 𝑉 𝐷/ } (𝐼𝐶) Spagnolo concludes that for every possible fine level, the reduced sanction must be negative, i.e. the legislator must reward the reporting party.31 Rewards make programs more efficient by increasing the value of “deviate and report” (VD ) and this narrows the (IC). A deviating firm facing the decision whether to report the cartel or not will be more likely to report, if the leniency to be obtained has the form of a reward. On the other hand, too high rewards can make the program exploitable. Spagnolo finds that when fines are high enough and the detection probability (α) increases, the optimal reward (RF*) decreases. This implies that in optimum, investigations are a substitute for rewards as law enforcement instruments. When α is large, the optimal reward must be small. In that case, the high probability of detection (α) even allows a fewer reward level. At the same time, the optimal reward (RF*) increases in fines, making fines and rewards complementary instruments. Regardless of any other policy parameter, increasing fines allows for a higher reward without making the program exploitable. As described, investigations and rewards are substitutes. Whereas investigations are costly, rewards are self-financing, making it much more profitable for an optimal law enforcement policy to rely, whenever possible, on self-financing rewards. Spagnolo finds that there is a finite level of fines at which complete and costless deterrence (α = β = γ = 0) is achieved.32 This can be done by setting the reward for the first reporting agent at an amount equal to the sum of all the fines paid by the cartel and, at the same time, interrupting any investigation. The authority can reject every possible investigation and wait for reports. The high level of fines empowers the leniency program making investigations redundant and suboptimal. This is the first time that the first best is achieved in such an environment. 29 Spagnolo (2005) p.17 Proposition 2 30 Spagnolo (2005) p.17 (IC) Condition 31 Spagnolo (2005) p.17 Proposition 3 32 Spagnolo (2005) p.19 Proposition 4
  • 32. Leniency Policies Jaime Plaza Aparicio 27 Moderate Leniency programs On the other hand, moderate leniency programs put some constraints on fines and rewards. Moderate leniency programs also have the protection from fines effect and enable a protection from punishment effect. If the optimal first best fine cannot be achieved because of the constraints on fines, the second best law enforcement policy will imply positive investigation costs and also a non-maximal reward.33 Since the constraint on rewards does not allow for any reward, i.e. a positive transfer, the optimal reward would be a complete reduction of the fine (RF = 0). Spending resources on investigations will also be optimal considering that fine reductions are not as attractive as rewards to cartel members. When complete reduction of fines (RF = 0) is implemented with a probability of conviction when deviating (γ) of zero, moderate leniency programs do not have deterrence effects. It is clear that protection from fines effect (RF < γF) disappears making the program ineffective. The program needs γ > 0 to enable this effect and make the program effective. This means that even deviating firms could be convicted for past collusive agreements. Firms only care about the value of γ if they are willing to report. Even in a scenario with γ = 0 and firms using a “stick-and-carrot” strategy, a moderate leniency program may disrupt cartel activities. Firms in a “stick-and- carrot” strategy incur a cost in the “stick” phase but they do so because this allows them to enter the “carrot” phase in the next period. If the authority punishes repeated wrongdoers more firmly, imposing higher fines and looking more closely into their activities, making the probability of detection (α) higher for them, incurring the “stick” phase cost will not be profitable anymore. The per- period payoff of collusion would be the same but the fines will increase in each new “stick” period making collusion unworkable. This is called the protection from punishment effect. Through both of these effects, moderate leniency programs, if well implemented, can deter and detect cartel activity. Risk of cheating Spagnolo states that an additional reason explaining the deterrence effect of moderate leniency programs is the perceived risk of being involved in collusive agreements. In previous analysis, collusion was sustainable if the (IC) was satisfied and cartel deterrence was achieved with an (IC) violation. Now it is assumed that trust is also required for a collusive agreement. Without knowing 33 Spagnolo(2005) p.20 Proposition 5
  • 33. Leniency Policies Jaime Plaza Aparicio 28 for sure that cartel members will stick to the agreement, no agreement can be made. Spagnolo presents a simple model of two firms that entered into a collusive agreement and in each period must decide whether to stick to the agreement or to deviate from it and expect punishment from its partners. Spagnolo applies Harsanyi and Selten’s definition of risk dominance and calculates the Nash products of the two pure strategy equilibria; namely, both agents colluding and both agents deviating. He does so to measure the riskiness of both equilibria for situations with and without a leniency program. Spagnolo concludes showing that the colluding equilibrium with a restricted leniency program is the most risky.34 This means that agents colluding in an environment with a restricted leniency program will gain more from deviation than deviating in any other environment (unrestricted leniency program or no leniency program situations). Conclusions A leniency policy that rewards reporting firms as well as a lack of active investigations by competition authorities is fully against the reality. Active investigations inducing fear in cartel members, as stated in Hammond (2000 & 2004), are crucial in cartel deterrence for the Antitrust Division. Restricting the leniency program to the first reporter only is in line with the policy applied by most of the competition authorities. Spagnolo shows how a restricted program makes both the program more efficient and gives greater incentives to deviate from a collusive agreement. This view is shared by the Antitrust Division that considers the fear of arriving second as key to the efficiency of the program. The European Commission goes in the opposite direction with its perception that a collaboration has an intrinsic value to be awarded. In my opinion, a policy as designed in the U.S. will induce much more fear in cartel members since “wait and see” strategies lose all of their value. Awarding the second, third or subsequent reporting parties with a fine reduction seems too generous to me and goes against the spirit of what I suppose was the intended purpose of the leniency program. The European Commission has been accused of bureaucracy since its inception. Acknowledging that the collaboration of cartel members will reduce the time and resources spent in such processes. I consider that announcing a reduction for every useful collaboration is too generous. Probably, these cartels have operated in a lot of countries damaging consumers from different nationalities. 34 Spagnolo (2005) p.26 Proposition 7
  • 34. Leniency Policies Jaime Plaza Aparicio 29 To measure exploitability the following reasoning can be used. Assuming that with the adoption of a leniency program cartel fines increased, as stated in Spagnolo (2006), it could follow that strategies of reporting in turns decrease their value, since programs only accept the first party. Cartel members accept this kind of “stick-and-carrot” strategies, especially in the stick-phase, because of the carrot following in the next periods. If the authority toughens the fines but maintains the awarded leniency constant, at a certain point this strategy will turn out unprofitable. Rewarding without investigating is the most important point of Spangolo’s work. A self-financing reward that is fully paid by the fines of the cartel members can become problematic. The authority only changes the money from hands. Under a legislation as the E.U. this could be a problem. Firms could make an agreement under which every cartel member pays an amount of the total fines and after the reporting member gets the reward, the cartel members distribute it among them. Firms could participate in a collusive agreement that is reported in every period, paying and receiving the same amount without any loss. Of course this is a very simplistic view and the authority will react to such behavior. Nevertheless, Spagnolo does not count on such agreements. He models consensual reports in which every cartel member reports the cartel every period. In some periods a cartel member will have to pay the fine and in others, when he manages to be the first one to report, he will receive the fines from the other members. Since no good can be expected from a multinational operating cartel, any kind of collusive agreement could be used by the cartel members, no matter how “diabolic” in terms of taking advantage of the authority. This reasoning could be reinforce with the fact that there are no investigations (in Spagnolo’s framework) which makes the fear of being discovered disappear. Such consensual reporting strategies will work in legislations without prison sentences, but only if monetary fines are imposed as penalty, as the E.U. A lack of investigation tools to spare resources is completely contrary to any competition authority in real life. What induced the fear in the directors of collusive firms were ultimately the tough sanctions, especially the possibility of imprisonment. A fear of high sanctions could result in more reports but, as explained in section 2.2., after the reports comes the learning. Knowing how firms fix prices, exchange information or limit bids will help following investigations of collusive agreements.
  • 35. Leniency Policies Jaime Plaza Aparicio 30 3.4 Chen & Rey (2013) Chen and Rey study the optimality of offering leniency before and after an investigation is launched. Characteristics such as restricting the program to the first informant only or offering leniency to repeated offenders are also studied. The model suggested by Chen and Rey is much more technical than the model from Spagnolo. However, the infinitely repeated interaction between firms is repeated. An environment with multiple industries is presented. Collusion can be sustained without a leniency program in some industries. This means that firms will want to collude without breaking the agreement even without antitrust enforcement. The discount factor of future payoffs (δ) is the same across industries, whereas the potential win from collusion differs across them. This sounds realistic considering that in the real world not every industry has the same sales volume or “collusive” potential. The parameters of the antitrust enforcement policy are exogenously given. Given a collusive agreement and no-one reports, there are two probabilities: the probability of carrying out an investigation (α) and the probability of success (p) of this investigation making the probability of success of any investigation (αp). With probability α an investigation is launched in a collusive industry. If an investigation is launched, the collusive agreement is discovered with probability p. Therefore, a firm involved in a collusive agreement that has no intention to report, must expect discovery of the cartel probability αp. The level of fines (F) is also exogenously given and is set to its maximal possible level, which the authors think is the level where optimal deterrence is achieved. Investigations are public and can be observed by any firm. Collusion, as in the previous model, leaves some evidence that only remains valid for one period. The authority implements a leniency program with two different leniency levels, depending on the state of the investigation at where the report arrives. The leniency offered can vary depending on the firm reporting before or after an investigation has started. The leniency offered is a reduction (q) on the fine so that a convicted firm will have to pay (1 − q)F. With q > 1 the leniency would be offered in form of a reward; otherwise, the firm would only obtain a fine reduction. The timing of the game increases and so does the number of strategies firms can use. Firms agree on collusive agreements and, at the same time, the competition authority launches investigations. Firms can report before the investigation is launched as well as with the investigation ongoing. A deviation from the collusive agreement, i.e. in the Bertrand Competition underbidding the collusive
  • 36. Leniency Policies Jaime Plaza Aparicio 31 price and winning the whole market, is not observed until the end of the period. A deviating firm can also report the cartel and benefit from the leniency. The Benchmark without a leniency program sets a critical value for the payoff needed for collusion.35 Firms in an industry achieving, through collusion, a per- period payoff higher than the critical value will always collude without a leniency program. In industries where the collusion payoff is below the critical value, collusion is not sustainable. Before determining the optimal leniency program, the various collusive strategies must be considered. `Normal collusion’ (N), `collude and report systematically´ (R) or `collude and report in case of an investigation´ (I) are the possible collusive strategies. Firms can also deviate from any collusive agreement, which leads to multiple possible scenarios. The objective of Chen and Rey is to focus on the deviation from these agreements and to look for the optimal level of leniency where deviation becomes profitable. Leniency before the start of an investigation When no leniency after the start of an investigation is considered, only the `normal collusion´ (N), stick to the agreement and no report, and the `collude and report systematically´ (R) agreements are available. The optimal value of fine reduction (𝑞̂) will be the level that makes a firm indifferent between the per- period payoffs of the two deviations from the two possible agreements (N) and (R).36 The optimal value (𝑞̂) decreases when the probability of success of an investigation (αp) increases. There would also be an interval where this optimal value can remain making the leniency policy effective. Outside this interval the program is either too unattractive to cartel members or increases the value of the strategies involving consensual reporting. With the optimal leniency level (𝑞̂) a per-period payoff threshold (𝐵̂) will be defined37 . In industries where firms can achieve a per-period payoff of collusion higher than the threshold (𝐵̂), collusion is sustainable. Firms that cannot achieve a collusion profit as high as the threshold will not enter into any collusive agreement. The works looks for the optimal leniency level without any restriction. Depending on the parameters, leniency can be a fine reduction or a reward. Therefore, Chen and Rey find that leniency is required before an investigation is launched and also that the amount of leniency decreases when the probability of success of an investigation (αp) increases. They consider that the best way to fight 35 Chen & Rey (2013) p.925 Condition (1) 36 Chen & Rey (2013) p.931 Condition (4) 37 Chen & Rey (2013) p.931 Condition (5)
  • 37. Leniency Policies Jaime Plaza Aparicio 32 collusion is to crack cartels from inside. Giving incentives to deviate, such as offering some leniency before the launch of an investigation, will help eliminate the cartel. This work also studies the first informant rule. Awarding leniency to more informants decreases the per-period payoff threshold. This threshold marks the level above which collusion is sustainable even with a leniency program. A lower threshold will results in more “collusive” industries. Leniency before and after the start of an investigation When leniency is offered both before and during investigations, more collusive strategies are available. The authors show that a program offering also post- investigation leniency will perform strictly better when investigations are unlikely to succeed. As in the previous situation, the work focuses on finding the optimal leniency levels to make deviation from collusive agreements more profitable. Two strategies can destabilize cartel collusion: i) deviate and report in case of investigation and ii) report in case of investigation. Looking at the parameters that make firms want to destabilize cartels through the first option, a critical value of success in random investigations (p) is found.38 If the probability of success of an investigation (p), exogenously given, is below this critical value, offering some leniency during investigations will destabilize normal collusion. Indeed, the optimal levels of leniency, before and during investigations, will make firms feel indifferent to any deviation from the three possible collusive agreements ((N), (R) and (I)). Applying this same procedure to make firms destabilize cartel collusion through the second option, i.e. by reporting in case of an investigation (without deviating), leaves the following conclusions depending on the level of the discount factor (δ). The idea is to differentiate between firms being patient (high δ) and firms being rather impatient (low δ). In each of these two cases, depending on the level of the investigation probabilities α and p, a specific leniency policy would be optimal. There are three candidates for an optimal leniency policy: a leniency policy available only before the investigation start, a leniency policy available both before and during an investigation focused on cartel destabilization by firms 38 Chen & Rey p.934 Condition (9)
  • 38. Leniency Policies Jaime Plaza Aparicio 33 deviating and reporting, and a leniency policy available also at both stages focused on cartel destabilization through reports without deviation. When firms are patient, but both probabilities α and p are low, the authority will set a leniency policy designed so that firms will have incentives to report without an early deviation. If instead one probability is high and the other low, another leniency policy would be optimal for implementation. When there is a situation with high p and low α, a leniency policy based only on pre-investigation reports would be optimal. The small number of investigations launched will end successfully most of the times and the best option would be a leniency program that enables cartel destabilization through reports before the investigation starts. In the opposite situation, high α and low p, a leniency policy that provides firms with incentives to deviate from the collusive agreement and to report, either before or after the investigation start, will be the optimal program. When firms are less patient, the authority cannot count on a cartel destabilization through reporting only without deviating. In such cases, firms value more payoffs in earlier periods than in later ones. Deviating from the cartel agreement, i.e. underbidding the cartel price, is too attractive for firms in situations with a low discount factor (δ). Again, the authority chooses to implement one leniency policy or the other based on the policy probabilities α and p. A high p and a low α will promote a leniency policy, as explained in the previous paragraph, based on reports before the investigation phase. The opposite situation will call for a leniency policy allowing reports any time focused on cartel destabilization through cartel deviation and report. The conclusion of the authors is that offering some leniency before the start of an investigation is always optimal. Offering leniency during investigations will be optimal only under certain conditions. The work focuses briefly on repeated offenders. The analysis does not focus on the firms that apply for leniency more than once, but on the application of the leniency program in the whole industry more than once. The analysis defends the existence of an unlimited leniency program in terms of firms applying for leniency in different periods, but only one firm obtaining it per period. In other words, the restriction of leniency to the first party that reports in the industry will not be effective in cartel destabilization. Conclusion Considering internal destabilization as the best form of cartel detection is a very important remark in the work. The approach of other economists was to look for an effective leniency program, since the authorities have scarce resources and could not launch an infinite number of investigations. The approach, from the
  • 39. Leniency Policies Jaime Plaza Aparicio 34 very beginning, of focusing on cartel destabilization through an effective leniency program is a very important feature of this paper. The Antitrust Division could have shared this view when implementing its leniency policy. As explained before, leniency programs discover most cartels from all deterrence tools. This work shares this idea with the Antitrust Division. Chen and Rey defend that, under certain conditions, offering leniency after the beginning of an investigation will increase the efficiency of the leniency policy. Leniency programs have evolved in that direction. Being less strict in the conditions required from firms applying during investigations was a common characteristic of the leniency programs revisions. They distinguish between the launch and the success of an investigation. A competition authority in an investigation will consider the verdict as fair and correct, whereas in some cases it would have not made the right decision. Whether the authority makes the right decision or not is not observable. Thinking of “real” probabilities to see which leniency policy would be the right one to be implemented is a hard task but, in my opinion, the reasoning behind the leniency policy decision is consistent with the view of the competition authorities. Even so the authorities are more reluctant to guarantee leniency during investigations. The fear of turning the leniency program exploitable could be a reason. The truth is that leniency programs are a recent innovation in most of the countries. The lifetime of cartels exceeds the lifetime of leniency programs. When collusion began, leniency programs were not an issue. This work shows a very good interpretation of the decisions firms must make nowadays whether to enter or not into collusive agreements. Restricting the program to the first informant only is an essential characteristic for both competition authorities and economists. Leniency programs in the U.S. have restricted leniency to the first party only. Awarding leniency to every reporting member will make collusive agreements an unpunished activity. Despite the different frameworks presented in different economic papers, restriction to the first party is a common conclusion. As explained in previous sections, the position of the European Commission is different. Collaborations that help investigations should be awarded leniency regardless the entry order. Chen and Rey also study the optimality of leniency to repeated offenders. They could have thought collusive agreements are not a one-time decision of firms. Entering into a collusive agreement is an illegal activity and must be carefully considered. Multinationals can enter into and maintain multiple collusive agreements in different markets and countries. The Antitrust Division has offered mechanisms such as the Amnesty Plus to help firms leave not one but
  • 40. Leniency Policies Jaime Plaza Aparicio 35 every collusive agreements in which they participate. The Division also punishes any lie, if it is discovered, when a firm is asked if it is involved in any other collusive agreement. The model is a very good approximation of firm decisions nowadays. Having the possibility to report at multiple stages and operating in a lot of markets, each of them with different competition authorities, the number of available collusive strategies increases. Competition authorities should be careful and design a leniency policy which is as effective as possible. 3.5 Aubert, Kovacic & Rey (2005) The authors compare leniency programs with reduced fines to programs with positive rewards. They also study the case of awarding leniency to individuals and how leniency programs can affect business decisions. In addition, reasons for the puzzle of firms keeping hard evidence are given. Firms play an infinitely repeated game. In each period firms can decide communicate first (it is a choice). After that, firms can choose whether to collude, if communication took place, or to compete. Firms discount payoffs using the same factor 𝛿 ∈ (0,1). The model has the same assumption of evidence only lasting one period. Communication (and thus collusion) leaves some evidence that cannot be manipulated. The maximal possible fine (𝐹) does not succeed in deterring collusion. The competition authority launches investigations with a probability p. Moreover, it is assumed that collusive agreements will be discovered in case of an investigation, as opposed to Chen and Rey (2005), where investigations in collusive industries would not necessarily end successfully. Leniency programs only reducing sanctions Without a leniency program, collusion will be more profitable than deviating from a collusive agreement as long as the gain from the collusive agreement is higher than the gain from the deviation from a collusive agreement, when the rest sticks to it. If this condition (A)39 holds, collusion is sustainable. In the situation with a leniency program, the model assumes that a firm choosing to report will also deviate from the collusive agreement since competition is going to prevail in the market afterwards anyway. The reduced fine is lower than the fine times the probability of investigation (𝑅𝐹 < 𝑝𝐹), as in any previous model. With a leniency program, collusion will be sustainable only if the gain 39 Aubert, Kovacic, Rey (2005) p.11 Condition (1)
  • 41. Leniency Policies Jaime Plaza Aparicio 36 through deviation is higher than the gain from collusion. Without this condition (B)40 , collusion would not be sustainable. The difference between two conditions with and without leniency program is that the first one does not include the leniency obtained through deviation (thus reporting). Leniency programs restricted to fine reductions/cancellations can only deter collusion by setting the policy parameters in a way that (A) holds and (B) not.41 The authors conclude that “leniency programs (only reducing/cancelling fines) have no impact on the profitability of collusion and affect its sustainability only by giving the deviators the opportunity to avoid fines from random audits”. The probability of investigation (p) or the fine (F) have to be very large in order to make the leniency program effective. However, in these two cases, collusion would not be sustainable. Fear of being or fined would outweigh the profit from collusion. Leniency programs rewarding reports The authors only focus on reports before the investigation phase, considering reports during investigation as more costly to implement. To make the analysis the easier, only deviating firms can report. The analysis rejects the possibility of strategies based on consensual reports. The reward must still be higher than the costs of conviction (− 𝑅𝐹 > − 𝑝𝐹) and must also compensate the payoffs of the punishment phase that follows a report. If the reward needed is positive (fine reduction), a full reduction of the fine would be enough to deter collusion. The reward will tend to minus infinity if the discount factor δ goes to 1. Leniency programs oriented to individuals The authors defend individual leniency programs that reward individual informant as a complement of the corporate leniency programs. They mark potential agency problems as an issue. The individuals who will have greater incentives to report are the employees. Directors or Officers will have more direct profits from collusion than employees. Firstly, the authors suppose employees only live for one period. The idea is that when leniency programs reward individuals firms will have to bribe these individuals. By doing so, firms “buy” the silence of these individuals. This additional cost decreases the benefits from collusion. Therefore, an increase in 40 Aubert, Kovacic, Rey (2005) p.11 41 Aubert, Kovacic, Rey (2005) p.11 Proposition1
  • 42. Leniency Policies Jaime Plaza Aparicio 37 the reward by the competition authority decreases the value of collusion and makes it “less” sustainable. Even if the leniency program restricts the reward to the first individual only, firms will have to pay the reward to every employee in the firm. If they do not bribe every individual, the non-rewarded employees could have incentives to report the agreement and cash the reward. Using the same reasoning, collusion is more fragile when the number of employees increases. When employees are “long-lived”, corporate leniency programs also play a role. An individual leniency policy will not affect the sustainability of the collusion, when there is no corporate leniency program. By adopting a corporate program, the individual program can destroy collusion. This effect, as the previous one, increases as the number of involved individuals rises. The authors point out that the level of the rewards should be finite and not too high since rewards are not supposed to be cashed in the equilibrium. Firms will react to the reward program and bribe their employees with this amount so that no employee has incentives to report (in the equilibrium). Side effects of leniency programs Although cartel deterrence can be achieved through reward programs42, the authors think such programs may have some side effects. The nature of the communication between competitors does not have to be necessarily bad. Firms could exchange information making their competition more efficient. Also, a reward program can have a huge impact not only on collusive decisions but on internal firm decisions too. The authors present both the impact of the programs on some of these decisions and the adjustments needed in the programs to limit the impacts on the decision. Thinking that not every communication is necessarily bad, firms can communicate only to increase the efficiency of their competition, i.e. compete but at a lower level of social costs. A model similar to the model first addressed in this papers is presented. Firms can choose whether to communicate (with good or bad intentions) or not. Communication with good intentions will lead to higher profits in competition made by the firms and will also increase social welfare with respect to competition without communication. Once again, firms can deviate from a collusive agreement. If firms choose to communicate with good intentions, the market outcome will be a more efficient competition. As in previous models, any kind of communication leaves evidence. Firms can report the evidence of communication to the competition authority. Since not 42 Leniency programs awarding rewards to informants.
  • 43. Leniency Policies Jaime Plaza Aparicio 38 every piece of evidence is a signal of a collusive agreement, the competition authority will have problems in classifying it. The competition authority interprets “good” evidence as “bad”43 with a certain probability 𝜇̂ and, the other way around, “bad” evidence as “good” with a certain probability 𝜇. 44 If the authority does not receive any report, it will launch an investigation with probability p, the same one as in the previous models. By offering positive rewards to reporting firms, firms could make a false denounce to benefit from the reward. Such firms will present “good” evidence to the authority and cash the reward with a probability 𝜇̂. Firms will only cash the reward if the authority misinterprets the “good” evidence and this happens with probability 𝜇̂. An additional fine to false reports could be implemented. On the one hand the additional fine can deter false reporting, but it can also deter the good intention reports. This last scenario happens whenever the authority misinterprets “bad” evidence imposing the additional fine on a firm that truly reports “bad” evidence. The authority should anticipate that and adjust the reward program accordingly. It could do so by setting the reward and the additional fine so that two conditions are met; namely, deviating firms would prefer to report any collusive agreements and firms communicating with good intentions would find false reports unprofitable. Such program avoids false reports and incentives the true ones. Individual reward programs can also have some effects on internal firm decisions. As stated before, collusion is more sustainable if employees are long lived. Firms will therefore want employees to stay in the firm by bribing them rather than allowing them to report. This will result in a rather fix employment structure where informed employees will not leave the firm. Having a certain group of informed employees staying at the same position with no internal restructuring could result in loss of productivity. Informed employees that the firm would want to replace will remain at their jobs. Otherwise they will report collusive agreements and cash the rewards. If firms must choose between entering into collusive agreement and restructuring to compete, a high reward, as well as a high competition outcome after restructuring, make the collusive agreement less profitable. By restructuring an employment restructuration to achieve a higher competition payoff is meant. Firms, at the time of the choice, are not bribing any employees and will not incur in any cost if they leave. Restructuring will be the best choice, if the reward is high enough. In this case a highly enough reward will deter collusion. 43 By “good” evidence, evidence of well-meaning communication is meant. “Bad” evidence refers to evidence from a collusive agreement. 44 Aubert, Kovacic, Rey (2005) p.20
  • 44. Leniency Policies Jaime Plaza Aparicio 39 When colluding firms have to choose between sticking to collusive agreements and restructuring to compete, the level of the reward will not affect the choice. In both cases, firms do not need to bribe any new employee. The expected fine and the expected competition outcome after restructuring affect the decision. Firms will stick to the collusive agreement (worst outcome for society) and will not restructure to compete (best outcome) whenever the fine is high or the competition gain through restructuring is moderate. The reward offered to individuals by the competition authority will not affect these choices of already colluding firms. Keeping hard evidence in the firm If firms agree to collude in the future, they, in theory, do not have any reason to keep evidence of collusion, otherwise, the competition authority could find this evidence when investigating. It is known that firms keep such evidence, since leniency programs have been widely used and firms need this evidence to report. The authors give several reasons to the puzzle of firms keeping evidence of collusive agreements within the firm. A reason given is the expectations of breakdown in the collusive agreement. Firms keep evidence because they expect the agreement to be short-lived and want to benefit from rewards in the future. This will only apply for high rewards. Rather low rewards will not make firms want to keep evidence. Threatening the cartel members with a report of the illegal activity to the competition authority based on the evidence of collusion is another reason. Designing leniency programs in a way that makes collusion unprofitable will make firms destroy evidence. When collusion is no longer sustainable, firms do not want to keep evidence of it because of the potential risks of being discovered. As throughout the entire paper, the authors also focused on how the leniency program and its consequences affect individuals. Firm employees could keep evidence of collusion because of agency problems. Leniency programs can make these agency problems bigger. The evidence of collusion is a very good card for employees to play in the future when bargaining with the manager. On the other hand, the firm can delay bonus payments, if an employee does not promise that he will not report the collusive agreement. Conclusions `A leniency program that does not allow rewards to informants will not be as effective as one that allows them´. This work shares this view with Spagnolo (2005). Aubert, Kovacic and Rey conclude that moderate leniency programs can truly deter cartel formation when fines or the probability of conviction are large,