Why Plea-Bargaining Fails to
Achieve Results in So Many
Criminal Justice Systems: A New
Framework for Assessment*
Nuno Garoupa** and Frank H. Stephen***

ABSTRACT
The myth of American exceptionalism in the matter of plea-bargaining is certainly by
now quite untrue. In addition to forming an important part of criminal procedure in the
United Kingdom, plea-bargaining has been transplanted to several civil law countries such
as France and Italy. Informal versions, based on non-trial settlement, have been observed
in Germany, Belgium, the Netherlands, and around the world. The Law and Economics
literature on plea-bargaining views it as an efficient instrument of criminal procedure
because it reduces enforcement costs (for both parties) and allows the prosecutor to
concentrate on more meritorious cases. Yet the success of transplants relies on the existence
of appropriate incentives, and the detailed study of the Italian experience provides a good
indication that the traditional inquisitorial system might not generate such incentives.
Instead, this article offers a new theory emphasizing the role of the prosecutor and that of
the defence counsel. We argue that the incentives of the prosecutor and those of the defence
counsellor are determinants of the success or failure of plea-bargaining. We are sceptical
that plea-bargaining can lead to or is consistent with the desirable outcome in many
circumstances. In particular, a major implication of our analysis is that the comparative

*

**
***

We are grateful to two anonymous referees, Oren Gazal-Ayal, Tom Ginsburg, Morag Goodwin, Jon
Klick, Larry Ribstein, Limor Riza, Justice Robert Steigmann, Cyrus Tata and seminar participants at
the law schools of Illinois, Chicago and Florida State Universities for helpful comments, and to Lydie
Ancelot for excellent notes on the French system. Yeny C. Estrada has provided excellent research
assistantship. The usual disclaimer applies.
Professor of Law, University of Illinois College of Law; Research Professor, IMDEA (Madrid) &; amp;
Research Professor of Law, School of Law, The University of Manchester; ngaroupa@law.uiuc.edu.
Professor of Regulation, School of Law, The University of Manchester; frank.stephen@manchester.
ac.uk.

15 MJ 3 (2008)

323

Nuno Garoupa and Frank H. Stephen

efficiency of plea-bargaining to a larger extent depends on the possibility of a legal system
to address the multiple principle-agent problems in criminal litigation.
Keywords: plea-bargaining; comparative criminal law; law and economics; prosecutor;
defence counsel; judge

§1.

Introduction

Plea-bargaining is very much a reality outside the United States. The myth of American
exceptionalism in this matter is certainly by now quite untrue. It is now an important
part of criminal procedure in the United Kingdom (with its three separate criminal law
jurisdictions of England & Wales, Scotland and Northern Ireland, all with somewhat
different versions of plea-bargaining). Plea-bargaining has also been transplanted to
several civil law countries (most notably France, Italy, Poland, and Argentina), despite
the severe criticism from traditional doctrinal approaches, but apparently without major
empirical success (either in widespread use or in cost reduction).1 Informal versions,
based on non-trial settlement, have been observed in Germany, Belgium, the Netherlands,
and around the world, again generating serious opposition.2
The reality is that 95% of all convictions in the US are secured with a guilty plea; guilty
pleas are widely observed in the UK (more than 90%); but only 8% of all convictions
in Italy are obtained by plea-bargaining.3 It seems to us that these figures confirm the
importance of plea-bargaining in criminal procedure. However, plea-bargaining is rarely
used outside common law countries, where criminal procedure is adversarial in nature.
Plea-bargaining is not frequently used in European civil law countries where criminal
procedures are inquisitorial.4
While the discussion that has emerged as a result of the comparative observations
presupposes the beneficial nature of transplants in criminal procedure,5 we prefer to see
1

2
3

4

5

324

See e.g., N. Boari, ‘On the Efficiency of Penal Systems: Several Lessons from the Italian Experience’, 17
Int. Rev. of L. and Ec. 115 (1997); N. Boari and G. Fiorentini, ‘An Economic Analysis of Plea Bargaining:
The Incentives of the Parties in a Mixed Penal System’, 21 Int. Rev. of L. and Ec. 213 (2001) (For the
Italian experience, patteggiamento). For the French case (plaider coupable), we are not aware of any
empirical assessment so far.
It includes summary procedures for those who do not contest guilt; unconditional and conditional
dismissals; confession of guilt for a reduction in the sentence, warnings and reprimands.
One should be careful in comparing numbers, since a conviction secured with a guilty plea does not
necessarily imply the formal use of plea-bargaining. These numbers are discussed in Boari, ‘On the
Efficiency of Penal Systems: Several Lessons from the Italian Experience’, and Boari and Fiorentini, ‘An
Economic Analysis of Plea Bargaining: The Incentives of the Parties in a Mixed Penal System’.
Obviously the lack of success in terms of use depends, in part, on how we define plea-bargaining. If we
take a broader definition of plea-bargaining as any form of negotiated sentence that avoids criminal
trial, then we might account for half or more of the convictions in many civil law countries.
See e.g., J. Herrmann, ‘Bargaining Justice: A Bargain for German Criminal Justice’, 53 University
of Pittsburgh L. Rev. 755 (1992); H. Jung, ‘Plea-Bargaining and its Repercussions on the Theory of

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

it as an indication that the endorsement of plea-bargaining requires a broader assessment.
The success of transplants relies on the existence of appropriate incentives,6 and the
detailed study of the Italian experience provides a good indication that the traditional
inquisitorial system might not generate such incentives.7 A similar conclusion can be
drawn from the French system.8
The success of plea-bargaining in adversarial systems is also not so clear cut, unless
we measure success by the rate of guilty pleas. In order to understand what we mean
by success in the widespread use of plea-bargaining in the United States (as generally
interpreted by legal economists), we first should realize that the American model of pleabargaining is quite complex and diverse. It is not a simple bargain between the accused
and the prosecutor in order to settle for a given sentence in return for a guilty plea.9 In

6
7

8

9

Criminal Procedure’, 5 Eur. J. of Crime, Criminal Law and Criminal Justice 112 (1997); E. Harnon,
‘Plea-Bargaining in Israel – The Proper Functions of the Prosecution and the Court and the Role of
Victims’, 31 Israel L. Rev. 245 (1997); E. Grande, ‘Italian Criminal Justice: Borrowing and Resistance’,
48 Am. J. of Comp. L. 227 (2000); Y. Ma, ‘Prosecutorial Discretion and Plea-Bargaining in the United
States, France, Germany, and Italy: A Comparative Perspective’, 12 Int. Criminal Justice Rev. 22 (2002);
M. Langer, ‘From Legal Transplants to Legal Translations: The Globalization of Plea-Bargaining and
the Americanization Thesis in Criminal Procedure’, 45 Harvard Int. L. J. 1 (2004); J. J. Jackson, ‘The
Effect of Human Rights on Criminal Evidentiary Processes: Towards Convergence, Divergence, or
Realignment?’, 68 Mod. L. Rev. 737 (2005).
See, N. Garoupa and A. Ogus, ‘A Strategic Interpretation of Legal Transplants’, 35 J. of Legal Studies 339
(2006) (for a general debate).
Boari, ‘On the Efficiency of Penal Systems’, and Boari and Fiorentini, ‘An Economic Analysis of
Plea Bargaining’ identify several problems with the Italian experience. To start with, the objective
of the transplant was to warrant a pre-trial settlement opportunity to reduce delays, not to enhance
prosecutorial discretion as economists usually think. On the side of the prosecutor, they identify
the following problems: lower concentration of resources because it is usually after the preliminary
investigations have taken place; applies only to minor offences which do not justify an intensive use
of resources; prosecutors advance their careers by seniority and not performance. On the side of the
defendant, it applies to a limited group of defendants (minor offences), hence is less attractive to accept;
there are many acquittals for loss of evidence or witnesses and amnesties granted regularly; hence
delays make it worthwhile to protract trials as long as possible to avoid jail.
The French plea-bargaining system has two stages: proposal and homologation. The prosecutor makes a
proposal to the accused that can be in the form of a fine or an imprisonment sentence. However the deal
must be reviewed by a judge, who can unilaterally reject it. The judge is intended to check the culpability
of the accused in relation to the evidence, the recognition of guilt by the accused, and the legality and
proportionality of the proposed sentence. If homologation fails, the prosecutor must continue the case.
Plaider culpable is regulated by the Code de Procédure Pénale. Most of the echoed criticism in French
legal writing against the introduction of this system addressed the procedural rights of the accused,
the negative consequences of empowering the prosecution (the Ministère Public), and the incentives to
avoid trial.
See Y. Kamisar et al., Modern Criminal Procedure: Cases–Comments–Questions (American Casebook
Series, 12th ed. 2008). Furthermore, it varies across states. Plea-bargaining is a negotiated agreement
between prosecution and defence in the federal system. In some states, such as New York and Illinois,
bargaining is mostly with the judge, who imposes due process concerns. For example, in Longley v.
State, 902 So. 2d 925 (Fla Dist. Ct. App. 5th Dist. 2005), the Court held that the defendant’s right to
due process was violated by the trial judge’s involvement with initiating the plea-bargaining process.
As a result, more and more states are hesitant to formally permit judges to become involved with the

15 MJ 3 (2008)

325

Nuno Garoupa and Frank H. Stephen

the United States, a plea bargain is an agreement between the state and the defendant,
which is approved by the court.10 Generally, the transaction consists of the prosecutor
making concessions, such as an agreement that the state will dismiss a second count
against the defendant, in exchange for the defendant’s concession to plead guilty or nolo
contendere.11 However, in the United States the validity of a plea bargain must meet
certain constitutional norms in addition to any local statutory requirements or rules of
the court.12 It is the court’s duty to insure that all the necessary elements of a valid guilty
plea have been met.13
Overall, plea-bargaining is a contract-like negotiation embedded in a set of
relationships involving other interests: the accused and his lawyer; the prosecutor
assigned to the case and the prosecutorial body in general;14 the victim and, potentially,
her lawyer; the judge presiding over the case and the judiciary in general; third parties
and society in general. Plea-bargaining is a particular contract within this complex
nexus. In the language of new institutional economics, we could say that plea-bargaining
is part of a complex nexus of contracts by very different players.15 It certainly requires a
better understanding of the institutional environment before jumping to conclusions.16
Plea-bargaining has been the object of very serious criticism. It is highly controversial
among legal scholars even in common law countries, including reputed law professors
in the US and the UK.17 In that respect, the Law and Economics literature contrasts
with many other legal approaches in its upbeat assessment. The most essential message
of the vast economic literature on plea-bargaining is that it is an efficient instrument
of criminal procedure because it reduces enforcement costs (for both parties) and
allows the prosecutor to concentrate on more meritorious cases.18 Plea bargains are

10
11
12
13
14
15
16

17
18

326

initiation of the plea bargaining process. Currently, statutes or rules in several states expressly prohibit
the trial judges from participating in plea negotiations, limiting their participation to the acceptance or
rejection of the agreement negotiated by the prosecutor and the defence lawyer.
State v. Kennedy, 698 So.2d 349 (Fla. App. 4th Dist. 1997).
Benjamin v. State, 874 S.W.2d 132 (Tex. App.14th Dist. 1994).
D. J. Newman, ‘Pleading Guilty for Consideration: A Study of Bargain Justice’, 46 J. of Criminal Law,
Criminology and Political Science 780 (1956).
Ibid.
For example, in England & Wales, the Crown Prosecution Service.
Economists use contracts not in the legal sense, but for every kind of explicit and implicit agreement.
Some economists have already highlighted the institutional context as a necessary element in the
evaluation of plea-bargaining. Compare F. Easterbrook, ‘Criminal Procedure as a Market System’,
12 J. of Legal Studies 289 (1983) (who goes in that direction but simply defends American criminal
procedure as superior to the German model), with R. Adelstein and T. J. Miceli, ‘Toward a Comparative
Economics of Plea Bargaining,’ 11 Eur. J. of L. and Ec. 47 (2001) (another comparative view); contra
M. Dubber, ‘American Plea Bargains, German Lay Judges, and the Crisis of Criminal Procedure’, 49
Stanford L. Rev. 547 (1996) (a contrasting view).
See the recent seminar on plea-bargaining (Fall 2007) published by 91 Marquette L. Rev. (2007).
Easterbrook, ‘Criminal Procedure as a Market System’ (the most optimistic paper).

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

the criminal counterpart of civil out-of-court settlements that Law and Economics has
largely supported.19
The present paper takes the view that there is a socially desirable (efficient) outcome
for each criminal case. We are sceptical that plea-bargaining can lead to or is consistent
with the desirable outcome in many circumstances. Therefore, our point is not to reject
an efficiency assessment of plea-bargaining, or criminal procedure in general, but rather
mitigate any general optimism about the result of such assessment.
In Section 2, we summarize the current Law and Economics of plea-bargaining.
We review the multiple arguments that legal economists have developed to endorse
the widespread use of plea-bargaining (including reduction of costs, reduction of
uncertainty, and the use of a market-approach to the criminal justice system). Most of
these arguments do not rely on any specific criminal justice system, and therefore fail to
explain why plea-bargaining has largely failed in civil law jurisdictions.
In Section 3, we look at the usual arguments against plea-bargaining and identify
two strands of the literature. First, we summarize arguments that in our view are not
compelling from a comparative perspective. We then analyze those arguments that reflect
the issues encountered with plea-bargaining in both common and civil law jurisdictions.
Such powerful arguments have not convincingly been addressed by legal economists in
our view.
In the fourth section, a more comprehensive framework to assess plea-bargaining
from a comparative perspective is considered. Our new theory emphasizes the role
of the prosecutor and that of the defence counsel. We argue that the incentives of the
prosecutor and those of the defence counsellor are determinants of the success or failure
of plea-bargaining. For example, on the defendant’s side, the way a criminal legal aid
system is implemented, or the way public defender offices are organized, explain many of
the difficulties encountered by plea-bargaining in England & Wales as well as in France
or in Italy. On the prosecutor’s side, the literature has been inspired by the American
empowered prosecutor, yet the model is more complicated when one takes into account
the serious institutional differences with other prosecutorial bodies, seen as weaker in
England & Wales and more bureaucratic in most civil law jurisdictions.
Our approach also encompasses two other important dimensions: third party effects
and judicial scrutiny. Although the interests of the victims could play an important role in
designing plea-bargaining rules, we are not convinced that empowering the victim with
veto rights is a wise solution to the alleged gap between the goals of the prosecutor and
the victim’s welfare. Further, judicial scrutiny is a major cause of the observed differences
between civil and common law jurisdictions. Our approach looks at the advantages
and disadvantages of empowering judges with a more active role (such as they have in
France or in Italy), and conclude with a cautions endorsement for intervention after plea19

See e.g., S. Shavell, Foundations of Economic Analysis of Law (Harvard University Press, 2004), chapters
17–19; T. Miceli, The Economic Approach to Law (Stanford University Press, 2004), chapter 8.

15 MJ 3 (2008)

327

Nuno Garoupa and Frank H. Stephen

bargaining. Quite clearly we reject the view endorsed by legal economists that judges
should not be part of the plea-bargaining procedure, but we also find serious problems
with models of judicial intervention before plea-bargaining is concluded.
The final section concludes our argument by stating that the failure of plea-bargaining
in civil law jurisdictions shows that legal economists have neglected some important
arguments with respect to the framework in which plea-bargaining operates.

§2.	The Economics of Plea-Bargaining
The field of law and economics is quite optimistic about the widespread use of pleabargaining. There are three fundamental lines of reasoning to justify the efficiency of
plea-bargaining. First, entering a plea-bargain reduces costs and allows the prosecutor
to allocate resources more effectively.20 Secondly, it maximizes social welfare because it
reduces uncertainty (since the outcome of a criminal trial is not certain; notice, however,
that when sentence benchmarking is applied this argument is qualitatively weak).
Thirdly, it operates just like discretion in regulatory proceedings,21 hence the presence of
a market-approach – which is largely believed to be a good idea – enhances the quality
of the prosecution.22 Plea-bargaining thus enhances the quality of prosecution because
it provides consistent signals that can improve prosecutorial decisions; it creates an
efficient screening of cases; and therefore provides the adequate incentives for an efficient
decision to stand trial. Finally, the use of plea-bargaining is consistent with the lowprobability high-penalty enforcement strategy proposed by economists (by imposing
punishment infrequently).23
20

21

22

23

328

See Easterbrook, ‘Criminal Procedure as a Market System’; G. M. Grossman and M. L. Katz, ‘Plea
Bargaining and Social Welfare’, 73 Am. Ec. Rev, 749 (1983); C. Y. C. Chui, ‘Plea Bargaining with the
IRS,’ 41 J. of Public Ec. 319 (1990).
See S. J. Schulhofer, ‘Criminal Justice Discretion as a Regulatory System,’ 17 J. of Legal Studies 43 (1988)
(making the point that the criminal process is a regulatory process while plea-bargaining bypasses it
and makes it less regulated).
See e.g., J. F. Reinganum, ‘Plea Bargaining and Prosecutorial Discretion’, 78 Am. Ec. Rev. 713 (1988); S.
Baker and C. Mezzeti, ‘Prosecutorial Resources, Plea-Bargaining, and the Decision to Go to Trial’, 17 J. of
L., Ec. and Org. 149 (2001); R. Wright and M. Miller, ‘The Screening/Bargaining Trade-Off’, 55 Stanford
L. Rev. 29 (2002); G. Lynch, ‘Screening versus Plea Bargaining: Exactly what are We Trading Off?’, 55
Stanford L. Rev. 1399 (2003); S. Bibas, ‘Plea Bargaining Outside the Shadow of Trial,’ 117 Harvard L.
Rev. 2463 (2004); W. Stuntz, ‘Plea Bargaining and Criminal Law’s Disappearing Shadow’, 117 Harvard
L. Rev. 2548 (2004); S. Mongrain and J. Roberts, ‘Plea Bargaining with Budgetary Constraints’, U.
Toronto mimeograph and SSRN, Working Paper (2004); D. Bjerk, ‘On the Role of Plea Bargaining and
the Distribution of Sentences in the Absence of Judicial System Frictions’, 28 Int. Rev. of L. and Ec. 1
(2008) (although Easterbrook, ‘Criminal Procedure as a Market System’ already makes this point, there
is an extensive literature on plea-bargaining and the quality of prosecution).
See A. M. Polinsky and S. Shavell, ‘The Economic Theory of the Public Enforcement of the Law’, 38 J. of
Ec. Lit. 42 (2000) (for the general theory); See B. Kobayashi and J. Lott, ‘Low-Probability-High-Penalty
Enforcement Strategies and the Efficient Operation of the Plea-Bargaining System’, 12 Int. Rev. of L. and
Ec. 69 (1992) (for the precise relationship between plea-bargaining and efficient enforcement).

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

We argue that the economic optimism for plea-bargaining arises from it being
viewed as a counterpart of out-of-court settlements in civil procedure. Plea-bargaining
is thus perceived as a contract between two parties,24 and, as such, should enhance social
welfare if voluntarily performed. Due to the existence of an asymmetry of information
concerning guilt and evidence, an appropriate design must be used to make the most
of such a contract. If well designed, plea-bargaining helps prosecutors to acquire
information and evidence.
The general reference to plea-bargaining by legal economists can be misleading and
confusing since there are several types of bargains that can be loosely placed under
such a designation (in fact, plea-bargaining is more of an umbrella term for different
procedures). Instead, they can differ with respect to the content of the bargain (the
scope of the contract-like negotiation), the point at which the bargaining occurs, and the
mechanism by which it is achieved.
With respect to the scope of the bargaining, we should consider charge bargaining
(multiple charges: drop some in return for a plea of guilty to one of them; unique charge:
drop a serious charge in exchange for a plea of guilty to a less serious charge); fact
bargaining (agreement for a selective presentation of facts in return for a plea of guilty);
specific types of bargaining (nolo contendere pleas: accepts a sanction without pleading
guilty; Alford pleas: accepts a sanction but defendant asserts innocence; parties negotiate
and agree on charges and facts but leave the sanction to be determined by the court);
and sentence bargaining (there is a specific discount for pleading guilty).25 Although
the different types of bargaining result in different problems,26 we will also discuss pleabargaining in the general sense since the exact scope of the bargaining does not affect
our approach. This is not the same as lumping together all the types of bargaining and
arguing they are all the same. Whatever scope it has, legal economists see plea-bargaining
as a contract-like negotiation that reduces expected costs for the accused. In that respect,
the focus of our analysis is on the institutional set-up of the contract-like negotiation and
not the exact scope of the bargaining. Obviously the problems we identify in this paper
have different degrees of importance for each type of plea-bargaining.
In relation to the mechanism by which plea-bargaining is achieved, it can be explicit
(a prosecutor makes a sentence recommendation or judges indicate the sentence they
are minded to impose), implicit (a defendant pleads guilty, considering the existence of
sentence discounts or the possibility of achieving a particular sentence, and without the
need for real negotiation), or a simple negotiated diversion (warnings and reprimands
in return for some restitution, transactions and conditional dismissals). It is quite
24
25

26

See R. E. Scott and W. J. Stuntz, ‘Plea Bargaining as Contract,’ 101 Yale L. J. 1909 (1992).
See A. Malani, ‘Habeas Bargaining’, 92 Virginia L. Rev. 1 (2006) (there is also habeas bargaining in the
sense that the accused exchanges his habeas rights (i.e., the right to seek relief from unlawful detention)
in return for shorter sentences much as they do with their trial rights in plea bargains).
See S. Bibas, ‘Harmonizing Substantive-Criminal-Law Values and Criminal Procedure: The Case of
Alford and Nolo Contendere Pleas’, 88 Cornell L. Rev. 1425 (2003).

15 MJ 3 (2008)

329

Nuno Garoupa and Frank H. Stephen

important what effect these different mechanisms have on uncertainty and therefore the
extent to which they generate different incentives for parties. Plea-bargaining is perceived
by legal economists as a mechanism to reduce uncertainty for both sides; the source of
uncertainty is usually the fact that the outcome of a trial is unknown. There are however
two quite different interpretations. One possibility is that there is uncertainty concerning
sentencing or with respect to the court finding the accused guilty and both parties have
the same prior distribution of probabilities. Where this is the case, plea-bargaining is
essentially assurance for both sides. Another possibility is that the parties have different
perceptions. Then plea-bargaining is also a mechanism to reveal information concerning
guilt and evidence.27
The timing of plea-bargaining is also important. Generally speaking, there is ‘plea
before venue’ (the defendant has to decide whether or not to engage in plea-bargaining
in early stages); in pre-trial hearings (the prosecutor and the defendant exchange
information before bringing a plea bargain); during the trial hearings (including the
possibility of a so-called cracked trial where the case is concluded without the last stage
of trial due to a late or delayed guilty plea). As the possibility of a cracked trial increases,
one of the economic arguments for plea-bargaining (reduction of costs to both sides)
loses significance. On the other hand, plea-bargaining in the early stages of prosecution
suffers from severe uncertainty and asymmetry of information that create quite serious
agency problems. The balance is, thus, quite complicated. As times goes by, more
information will become available (hence there will be a lower probability of mistakes)
and cost savings go down. Therefore, there is an optimal timing for plea-bargaining.
The problem is that the prosecutor and the accused may not have the same perception
of timing. It is possible that costs have different profiles. Also, the availability of new
information might depend on evidence rules and discovery.28 Certainly it seems likely
that different crimes have different optimal timings, hence imposing a fixed framework
for plea-bargaining across crimes could be counter-productive. It seems to us that the
issue of timing (because of what type of information is available to the parties) has been
underestimated by legal economists.29
Part of the explanation for the current system of plea-bargaining and its institutional
framework can eventually be provided by legal historians. Unfortunately historical
27

28
29

330

Plea-bargaining can be seen as signalling or a screening game. When the prosecutor proposes an
agreement to the accused, we have a screening game in that the prosecutor is attempting to separate out
different types of accused. When the accused proposes an agreement to the prosecutor, it is a signalling
game where the accused wants to signal to the prosecutor that he or she is of a certain type (e.g., guilty
or innocent).
It is also clear that any discussion concerning plea-bargaining cannot be isolated from the rules of
evidence and discovery in a given jurisdiction.
Contra L. Franzoni, ‘Negotiated Enforcement and Credible Deterrence’, 109 Economic Journal 509
(1999) (he develops a model of law enforcement in which indicted offenders and the prosecutor can
negotiate the penalty prior to the completion of the investigation. The analysis focuses on the credibility
of the conviction threat).

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

justifications of plea-bargaining in the United States are unclear.30 Contrary to what is
a widespread belief among legal economists, it might be that plea-bargaining had less
to do with empowering the prosecution and bypassing jury trials (a similar argument
has been made for the case of England & Wales31), and more to do with finding a simple
way to reduce the workload of courts in general and judges in particular; that is, it was
an adaptation to caseload pressure (the difference with the reforms in Italy and France
quite recently is that they were imposed by the legislator rather than being adaptation
by the courts).32 Therefore, the empowerment of prosecutors could have been a nonintended consequence of plea-bargaining that legal economists have rationalized as a
market mechanism to improve the quality of prosecution (which would be mitigated in
inquisitorial systems anyway and possibly diluted by mandatory prosecution).33

30

31

32

33

See G. Fisher, Plea Bargaining’s Triumph: A History of Plea Bargaining in America (Stanford University
Press, 2003); M. Feeley, ‘Legal Complexity and the Transformation of the Criminal Process: the Origins
of Plea Bargaining’, 31 Israel L. Rev. 183 (1997); G. Fisher, ‘Plea Bargaining’s Triumph’, 109 Yale L. J. 857
(2000); B. P. Smith, ‘Plea Bargaining and the Eclipse of the Jury’, 1 An. Rev. of L. and Social Science 131
(2005). See also, M. Vogel, ‘The Social Origins of Plea Bargaining: Conflict and the Law in the Process
of State Formation, 1830–1860’, 33 L. and Soc. Rev. 161 (1999) (arguing that the development of pleabargaining in the nineteenth century United States was as part of the nation building process).
In England & Wales, an indication of a benchmark sentence by a judge is now possible by the Criminal
Justice Act 2003; until recently there were some restrictive rules concerning the exchange of points
of view between judge and defendant with respect to sentencing, the so-called Turner Rules, which
delimited the judge’s role in pre-trial procedures. See M. McConville and G. Wilson, The Handbook
of the Criminal Justice Process (OUP, 2002), 353–377. Under R v Turner [1970], the accused must only
be advised to plead guilty if he has committed the crime in question; the judge should never indicate
the sentence he has in mind if the accused were convicted after entering a plea of not guilty (but could
say what sentence he would impose on a guilty plea). These rules were embodied in a Court of Appeal
Practice Direction (1976). The problem of discussion in the judge’s private room was addressed by
R v Pitman [1991]. The Runciman Royal Commission made important recommendations on more
formalized plea-bargaining. After R v Goodyear [2005], a judge dealing with a case may be asked by
the accused to give, and if asked, is entitled to give an indication of the likely sentence, which should
ordinarily be the maximum sentence he would impose on the accused if a plea of guilty were to be
tendered at the stage at which the indication is sought. Obviously there are pre-trial hearings to facilitate
exchange of information and bring about pleas of guilty, but it seems clear that judicial influence and
control over criminal procedure is relatively reduced. See also, M. Zander, ‘Please M’lud: How Long will
I Get?’, 155 New L. J. 677 (2005).
In England & Wales, it was formally introduced under the ‘plea before venue system’ by the Criminal
Procedure and Investigations Act 1996; although judicial precedent had created the one-third discount
for guilty-pleas, this was incorporated into criminal law later by the Criminal Justice and Public Order
Act 1994. The discount for pleading guilty was given legislative effect by taking into account (a) the stage
of the proceedings at which the offender indicated his intention to plead guilty and (b) circumstances
in which this indication was given. Subsequent controversy led to a set of Guidelines on the Acceptance
Pleas by the Attorney-General in December 2000; now sentencing guidelines deal with pleas in detail
(the recent Criminal Justice Act 2003).
The main point by Smith, ‘Plea Bargaining and the Eclipse of the Jury’, is that if plea-bargaining was one
of several methods employed by common law administrators to dispose of criminal cases without juries,
then when compared with other modes to bypass trial by jury (e.g., bench trial), it is less distinctive
and less American.

15 MJ 3 (2008)

331

Nuno Garoupa and Frank H. Stephen

A second line of reasoning would indicate that the choice of plea-bargaining
(bypassing or avoiding but not eliminating jury trials) over bench trials in the United
States in the nineteenth century and in England & Wales quite recently was pushed by the
need to reduce costs or delays (that is, secure benefits and avoid costs of criminal trials),
hence raising the question of more effective alternative reforms in criminal procedure.34
Notice that such an observation would explain an apparently puzzling question, namely,
why judges would agree with an instrument that clearly reduces their influence over the
criminal justice system.35
Legal history might actually support the thesis that the introduction of plea-bargaining
across jurisdictions might have happened for reasons that are not so dramatically
different as is usually assumed. Also, viewing plea-bargaining as part of American
criminal procedure without similarities across the world is simply misleading. Hence the
nexus of relationships in which it is applied emerges as more important than relying on
a specific historical explanation. For example, the standard explanation that the United
States has legal proceedings dominated by lawyers whereas continental Europe is more
dominated by bureaucrats seems to us to be pointing to what we have designated as the
nexus of relationships rather than some precise historical distinction with respect to the
influence of the legal profession.
Looking at the nexus of relationships in which plea-bargaining might emerge, legal
economists have identified potential implementation problems. First, it is argued that pleabargaining reduces expected sanctions and hence it dilutes deterrence; to keep deterrence
at current levels, plea-bargaining should be coupled with harsher sentencing.36 Secondly,
it has been argued that increased sanctions and a differentiated use of plea-bargaining
across crimes can lead to a substitution effect between different offences, resulting in
more severe crime.37 Thirdly, one possible outcome from plea-bargaining is that of
giving the lowest penalty to the most culpable defendant if it increases the probability of

34
35

36

37

332

P. Darbyshire, ‘The Mischief of Plea-Bargaining and Sentencing Rewards’, Criminal L. Rev. 895 (2000).
See also, S. J. Schulhofer, ‘Is Plea-Bargaining Inevitable?’, 97 Harvard L. Rev. 1037 (1984).
The influence of judges is more adversely affected where the bargaining is exclusively conducted by
the prosecutor and the defence. It is less of a problem when it is conducted, even if only partially or in
limited circumstances, with judges, like the English sentence bargaining or plea-bargaining in many
US jurisdictions. See L. Baum, ‘What Judges Want: Judges’ Goals and Judicial Behavior’, 47 Political
Research Quarterly 749 (1994), for why judicial support to plea-bargaining might reflect multiple
goals.
See T. J. Miceli, ‘Plea Bargaining and Deterrence: An Institutional Approach’, 3 Eur. J. of L. and Ec. 249
(1996) (although such changes might not have the expected result). See also C. La Casse and A. Payne,
‘Federal Sentencing Guidelines and Mandatory Minimum Sentences: Do Defendants Bargain in the
Shadow of the Judge?’, 42 J. of L. and Ec. 245 (1999); I. Kuziemko, ‘Does the Threat of the Death Penalty
Affect Plea Bargaining in Murder Cases? Evidence from New York’s 1995 Reinstatement of Capital
Punishment’, 8 Am. L. and Ec. Rev. 116 (2006) (for an empirical assessment).
See Mongrain and Roberts, ‘Plea Bargaining with Budgetary Constraints’.

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

conviction of conspirators; hence with multiple defendants, plea-bargaining should be
subjected to special design.38
Another important argument that has been made calls attention to the fact that pleabargaining is determined by increasing trial costs and binding budgets, not social welfare.
Hence, the efficiency depends on the objectives of prosecutors; clearly ex post objectives
of prosecutors should determine ex ante criminal policies of legislator. Some scholars
conclude that due to different prosecutorial objectives, plea-bargaining increases social
welfare in adversarial but not in inquisitorial systems.39
The role of judicial scrutiny in plea-bargaining has also received attention in the
literature. Initially the argument was against allowing judges to interfere with pleabargaining since prosecutors pursue the most meritorious cases and the quality of their
work is improved.40 Current literature has discovered some advantages of allowing
judges a more active role, namely improving the screening of cases between guilty and
not guilty41 and enhancing the introduction of sentencing guidelines.42 Furthermore,
judicial intervention avoids exceedingly lenient plea bargains. By hindering the
prosecutor’s ability to agree to exceedingly lenient sentences, courts increase the cost
of handling weak cases without obstructing the prosecutor’s ability to settle stronger
cases.
As a summary, we can say that law and economics is quite positive about pleabargaining; it relies on a model of a mutually beneficial contract; prosecutorial discretion
is perceived as being efficient (hence broadly more adequate to common law jurisdictions
than continental Europe); it tends to downplay the role of judges. However, as we discuss
below, this view neglects potential agency problems on both sides of the contract-like
negotiation as well as third party effects.

§3.	Traditional Arguments against Plea-Bargaining
The arguments against plea-bargaining have been made systematically by scholars in
criminal law in the US and in Europe.43 The simplest summary would be that plea38
39
40
41
42
43

See B. Kobayashi, ‘Deterrence with Multiple Defendants: An Explanation for Unfair Plea Bargains,’ 23
Rand J. of Economics 507 (1992).
See Adelstein and Miceli, ‘Toward a Comparative Economics of Plea Bargaining’.
See Easterbrook, ‘Criminal Procedure as a Market System’.
See O. Gazal-Ayal, ‘Screening, Plea Bargains and the Innocent Problem’, University of Michigan Legal
Working Paper Series (2005).
See J. F. Reinganum, ‘Sentencing Guidelines, Judicial Discretion and Plea Bargaining’, 31 Rand J. of
Economics 62 (2000).
See e.g. A. W. Alschuler, ‘The Prosecutor’s Role in Plea Bargaining’, 36 Uni. of Chicago L. Rev. 50
(1968); A. W. Alschuler, ‘The Defense Lawyer’s Role in Plea-Bargaining’, 84 Yale L. J. 1179 (1975); A. W.
Alschuler, ‘The Trial Judge’s Role in Plea Bargaining’, 76 Columbia L. Rev. 1059 (1976); A. W. Alschuler,
‘The Changing Plea Bargaining Debate’, 69 California L. Rev. 652 (1981); A. W. Alschuler, ‘Implementing
the Criminal Defendant’s Right to Trial’, 50 Uni. of Chicago L. Rev. 93 (1983). See also D. D. Guidorizzi,

15 MJ 3 (2008)

333

Nuno Garoupa and Frank H. Stephen

bargaining is a disaster in criminal procedure.44 Some of the arguments are somewhat
philosophical, such as criminals receive undeserved leniency (this implies that the social
costs of imposing sanctions should be neglected and assumes that the probability of
conviction at trial is one)45 or that plea-bargaining undermines procedural guarantees
for the accused, thus becoming an important constitutional issue46 (hence requiring that
all these rights and privileges cannot be derogated from even when the defendant wishes
it so).
A discount for pleading guilty, under whatever process the specific bargain is made,
imposes a penalty or a price on the right to be tried (the so-called trial tax). Scholars argue
that it undermines the presumption of innocence by penalizing those who exercise their
right to a trial and induces self-incrimination.47 However, as long as plea-bargaining is a
voluntary contract it should not be a problem since those who go to court expect a higher
payoff. After all that is why they reject the sugared guilty plea. A completely different
issue is if the accused is coerced in one way or the other, thus taking a decision with a
lower expected payoff. In such a situation, they are penalized because plea-bargaining
is no longer mutually beneficial. This is an issue of unfair bargains which we discuss
below.
Other arguments are more significant in our view; plea-bargaining hurts the innocent
and gives rise to unfair bargains in content and in time. Hurting the wrongly accused
has been the subject of attention by economists. Usually the argument means that the
strategic selection of cases to go to trial has two implications: (i) those who are innocent
(of the specific charge brought) stand trial more frequently than the guilty; (ii) there is
pressure for the innocent to plead guilty. Nevertheless, from an economic perspective,
those are not the most relevant questions. The issue is whether the introduction of pleabargaining affects the innocent in a disproportional way (in the statistical sense) from its
effect on the guilty. The conviction of an innocent at trial is possible (there are wrongful
convictions). In other words, unless plea-bargaining is disproportionately biased against
the innocent than against the guilty, what harms those who are innocent is a set of rules

44

45
46

47

334

‘Should We Really Ban Plea Bargaining? The Core Concerns of Plea Bargaining Critics’, 47 Emory L. J.
753 (1998) (a good summary).
See S. J. Schulhofer, ‘Plea Bargaining as Disaster’, 101 Yale L. J. 1979 (1992). See also R. E. Scott and W.
J. Stuntz, ‘A Reply: Imperfect Bargains, Imperfect Trials, and Innocent Defendants’, 101 Yale L. J. 2011
(1992).
See R. Christopher, ‘The Prosecutor’s Dilemma: Bargains and Punishments’, 72 Fordham L. Rev. 93
(2003) (this is obviously a retributivist argument whereas an economic approach is consequentalist).
See A. Ashworth and M. Redmayne, The Criminal Process (OUP, 3rd ed. 2005), chapter 12 (The
availability of the not guilty plea is regarded as a guarantee of certain defendants’ rights, namely the
presumption of innocence; the right to examine witnesses against the accused; the right to a fair and
public hearing. It is therefore controversial as to whether plea-bargaining can be compatible with the
European Convention on Human Rights).
See R. Henham, ‘Bargain Justice or Justice Denied? Sentence Discounts and the Criminal Process’, 62
Mod. L. Rev. 515 (1999).

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

of evidence that unfortunately do not allow a perfect separating equilibrium between
guilty and innocent.48
As long as the probability of conviction is positively correlated with the probability
of guilt, the selection-of-cases effect must imply a reduced proportion of innocent
defendants.49 The reason is that prosecutorial discretion is not unconstrained. Hence,
even if there is an incentive to offer a plea to low probability convictions, prosecutors have
to (i) select cases under a fixed budget; (ii) cannot offer unlimited sentence discounts;
and (iii) cannot control extra legal sanctions triggered by guilty pleas. Therefore, pleabargaining is more likely to prevail where the accused are guilty.
Since the economic approach is based on a contractual view of plea-bargaining, we find
the argument against plea-bargaining based on unfair contracts to be of importance.50 It
is difficult to have a precise notion of unfair contract, but we interpret the point raised by
several scholars as meaning that sometimes one of the parties – normally the defendant
– will accept a contract that is not in his own best interests51 (that is, the expected payoff
from acceptance is less than that from going to trial). Within a rational theory framework,
we need a complex model to justify such a possibility52 and assess the implications.53
Some scholars have argued that plea bargains are not civil settlements because civil
litigants care only about (i) strength of evidence and (ii) expected punishment after
trial. These scholars argue that there are important factors such as psychological bias and
48

49

50
51
52

53

See e.g. O. Bar-Gill and O. Gazal-Ayal, ‘Plea Bargains Only for the Guilty’, 49 J. of L. and Ec. 353 (2006);
D. Bjerk, ‘Guilt Shall Not Escape or Innocence Suffer? The Limits of Plea-Bargain when Defendant
Guilt is Uncertain’, 9 Am. L. and Ec. Rev. 305 (2007). Contra B. Kobayashi and J. Lott, ‘In Defense
of Criminal Defense Expenditures and Plea Bargaining’, 16 Int. Rev. of L. and Ec. 397 (1996) (on the
other hand, criminal defence expenditures result in lower than expected penalties for the innocent at
trial in comparison to those found guilty, and they help ensure that plea-bargaining efficiently screens
defendants).
Plea bargains allow prosecutors to charge more defendants and hence the number of defendants
(guilty and innocent) is likely to increase. The proportion of innocents might decrease under certain
conditions.
See Bibas, ‘Plea Bargaining Outside the Shadow of Trial,’ (for a general overview). See T. W. Church, ‘In
Defence of Bargain Justice’, 13 L. and Soc. Rev. 509 (1979) (for a defence).
Whatever the definition one might have of a defendant’s interests in this context, including emotional
attrition, personal and family circumstances, aversion to being held in custody, etc.
Notice that it is quite distinct from a harsh contract. See A. Stein and U. Segal, ‘Ambiguity Aversion and
the Criminal Process’, 81 Notre Dame L. Rev. 101 (2006) (making the point that defendants are in a very
weak position because they face more ambiguity (i.e., more uncertainty concerning the probability of
conviction) than the prosecutor, who is an experienced player. Although the argument neglects the role
of the defence lawyer, who is also an experienced player, it points to harsh but not unfair agreements
since the accused have more to lose than the prosecutor from a negotiation failure). See also A. Tor, O.
Gazal-Ayal, and S. Garcia, ‘Substantive Fairness and Comparative Evaluation in Plea Bargain Decision
Making’, Haifa University Working Paper (2006) (tries to incorporate notions on fairness in assessing
plea-bargaining outcomes).
Yet another source of a harsh contract could be the fact that whereas the accused has a non-exclusive
right over the time of the prosecutor, the prosecutor has a monopoly right over the future of the accused.
Therefore, one party has more power than the other in the bargaining process. Judicial scrutiny reduces
that monopoly power.

15 MJ 3 (2008)

335

Nuno Garoupa and Frank H. Stephen

structural forces (the quality of the defence lawyer, hence the wealthy will be favoured),
agency costs, bail and detention, legal rules, sentencing guidelines and statutes) that
produce skewed bargains and hence unfair contracts.54
A second line of reasoning points out that plea-bargaining cannot be seen as the
formation of voluntary contracts because the affected parties lack the incentive to assure
public interest in effective law enforcement (third-party effects) and deny defendants a
bargaining position to guarantee fair exchange. In particular, prosecutors face pressure
for convictions and plea-bargaining is a mechanism to get them at the expense of
procedural guarantees. There might be overcharging in order to increase pressure for a
guilty plea.
These criticisms go in the direction of underlining the weak position of the defendant
and the strong role of the prosecutor.55 We think that an overcharging effect cannot
be so dramatic because there are two signals concerning guilt (one for the prosecutor
and another for the jury and the judge), and both are correlated. Hence, the result must
depend very much on the rules of discovery and evidence as well as on the ability of the
defendant’s lawyer.
The view of the present authors is that most of the criticisms described hint at the idea
that plea-bargaining cannot be perceived as involving a simple contract in the way that
an out-of-court settlement is treated in civil law. The weak position of the defendant is
in many cases the result of an inefficient contract between the defendant and his lawyer,
a point neglected in most of the literature. The role of the prosecutor is determined
by institutional aspects that determine the nature of the relationship between the
prosecutor and the rest of society. Aspects such as the adversarial versus inquisitorial
nature of criminal procedure; mandatory versus selective prosecution; judicial
scrutiny of prosecutorial decisions; the role of prosecutorial guidelines; the existence
of prosecutorial proposed sentencing and judicial pre-commitment to sentencing, all
necessarily tailor the way prosecutors engage in plea-bargaining, and these aspects vary
widely across jurisdictions.56

54
55

56

336

See Bibas, ‘Plea Bargaining Outside the Shadow of Trial’.
See e.g. A. Hessick and R. Saujani, ‘Plea Bargaining and Convicting the Innocent: The Role of the
Prosecutor, the Defense Counsel, and the Judge,’ 16 Brigham Young Uni. J. of Public L. 189 (2002);
M. Langer, ‘Rethinking Plea Bargaining: The Practice and Reform of Prosecutorial Adjudication in
American Criminal Procedure,’ 33 Am. J. of Criminal L. 223 (2006).
On the role of the prosecutor there is an extensive literature. See e.g. D. G. Gifford, ‘Meaningful Reform
of Plea-Bargaining: The Control of Prosecutorial Discretion,’ Uni. of Illinois L. Rev. 37 (1983); R. Misner,
‘Recasting Prosecutorial Discretion,’ 86 J. of Criminal L. and Criminology 717 (1996); R. Heller, ‘Selective
Prosecution and the Federalization of Criminal Law: The Need for Meaningful Judicial Review of
Prosecutorial Discretion’, 145 Uni. of Penn. L. Rev. 1309 (1997); S. J. Schulhofer and I. Nagel, ‘Plea
Negotiations under the Federal Sentencing Guidelines: Guideline Circumvention and its Dynamics in
the Post-Mistretta Period’, 91 Northwestern Uni. L. Rev. 1284 (1997).

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

§4.

A New Theory of Plea-Bargaining with
Implications for Comparative Law

In this paper we provide a more comprehensive approach to plea-bargaining that
addresses the problems we have identified above. We take plea-bargaining to be a
voluntary contract between two parties (the prosecutor and usually the defendant’s
lawyer) inserted in a more general nexus of relationships that influence criminal
procedure. It is a contract that takes place in a context of uncertainty and asymmetry of
information; hence problems of moral hazard and adverse selection are likely to occur.
It is also a contract between only two of the parties affected by it. Even if the terms of
such a contract satisfied the parties directly involved, it may not maximize social welfare
because of the negative consequences on others. It does not therefore guarantee a socially
optimal bargain.
A usual argument against plea-bargaining is that it reduces the opportunity for
improvement of the criminal justice system in terms of identifying the shortcomings
of the law and of procedure, assessing the work of the police, exposing the ability of
prosecutors, and lack of quality of defence lawyers. It decreases pressure for legal reform
and undermines information to assure general deterrence due to it usually not being
disclosed to the public. Although we recognize and discuss these third-party effects (in the
sense that a settlement produces a negative externality on the quality of the legal system),
they also apply to out-of-court settlements in civil law, and therefore do not constitute a
distinct element of the discussion of plea-bargaining (although it is conceivable that the
magnitude of the negative externality is more significant in criminal than in civil law,
the reasoning is not substantially different and has been addressed in the literature).57
In our view, the striking difference between out-of-court settlements in civil and
in criminal procedure is that the magnitude of agency costs is much higher in the
latter than in the former. For reasons we discuss below, incentives between each side
and respective legal representative are likely to be better aligned in civil rather than in
criminal litigation. Moreover, third party effects are more important in criminal than
in civil litigation because an effect on third parties differentiates an offence as criminal
and not a tort.58

57

58

Compare the classical O. Fiss, ‘Against Settlements’, 93 Yale L. J. 1073 (1984), with S. Shavell, ‘The
Fundamental Divergence between the Private and the Social Motive to Use the Legal System,’ 26 J. of
Legal Studies 575 (1997).
See R. Bowles, M. Faure and N. Garoupa, ‘The Scope of Criminal Law and Criminal Sanctions: An
Economic Analysis and Policy Implications’, 35 J. of L. and Society 389 (2008).

15 MJ 3 (2008)

337

Nuno Garoupa and Frank H. Stephen

A.	Scope of Plea-Bargaining
Asymmetry of information is important in assessing the efficiency of plea-bargaining. If
information were symmetric and perfect, we would expect plea-bargaining to generate
contract-like negotiations that would be efficient since agency costs would be zero. It is
important to realize that it is not the scope of plea-bargaining (how much parties can
negotiate) that increases the likelihood of inefficiency, but the existence of asymmetric
information that makes an efficient contract harder to obtain. Therefore, the reasoning
should be that we require instruments to reduce asymmetry of information rather than
the scope of plea-bargaining.
There are three important elements to the reduction of asymmetry of information:
sentencing benchmarks, rules of evidence and discovery, and timing. Sentencing
benchmarks reduce divergent expectations concerning the outcome of a trial. Penal
codes and sentencing guidelines favour this process, as do rules that impose on the
prosecution a duty to announce a target sentence or a general principle of sentencing
proportionality. Judicial discretion in this particular case has a twofold role. By making
trials more uncertain, it induces parties to be more willing to engage in plea-bargaining.59
Therefore, by reducing the possibility of knowing sentencing benchmarks, it increases
the possibility of inefficient contracts.
Rules of evidence are also relevant to reducing asymmetry of information. Mandatory
disclosure for both parties before plea-bargaining and failure to disclose as justification
for breach should be understood as mechanisms that favour efficient plea-bargaining.60
The strategic use of information and evidence will only increase the possibility that one
of the parties will agree to plea-bargaining when they should not.61
59

60

61

338

We implicitly assume here that risk aversion predominates. However, the effects of uncertainty might
be more complex. For example, self-serving bias and over-optimism might prevail when sentences are
less certain, and reduce the willingness to engage in plea-bargaining. Also, when the trial result is
uncertain, the prosecutor and the defence might have too different estimations over the sentence to
make a settlement easy.
We do not take the view that forced full disclosure of evidence to reduce asymmetry of information is
an absolute value; rather it must be balanced against constitutional principles and should not lead to
self-incrimination. Our point is simply that many disclosure rules can be understood as promoting a
reduction of asymmetry of information.
In the United States, evidence obtained through the plea-bargain negotiation cannot be used against
the defendant in trial. Usually, any information gathered through the plea bargain remains off the
record unless the prosecutors are able to open independent evidence to prove such information, which
requires more time and money for the prosecution to investigate. The following cases support this
argument: Wilson v State, 484 S.W.2d 82 (1972); Hineman v State, 292 N.E.2d 618 (1973); Robinson v
State, 644 P.2d 514 (1982); Dykes v State, 372 S.W.2d 184 (1963) (various states hold that communications
made in the course of plea bargaining are not admissible into evidence). Also Federal Rules of Evidence,
Rule 410 & Federal Rules of Criminal Procedure, Rule 11(e)(6) (the Federal Rules of Evidence and the
Federal Rules of Criminal Procedure state that, subject to certain limited exceptions, any statement
made in the course of plea discussions with an lawyer for the prosecuting authority which do not result
in a plea of guilty or which result in a plea of guilty later withdrawn is not admissible as evidence against
a participant in the plea discussions).

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

The relevant rules of evidence and sentencing benchmarks (which are only possible
after evidence is known) should determine the optimal timing of plea-bargaining. The
earlier it takes place, the more enforcement costs are saved, but the higher are the chances
that an inefficient contract is accepted. The later it takes place (cracked trials), fewer costs
are actually saved, but given that almost all information has been revealed, the chances
of an inefficient contract are quite low. There is a trade-off between cost saving and the
quality of the plea-bargaining contract. Obviously the appropriate timing is determined
by the nature of the crime and enforcement technology. When most of the evidence is
actually gathered by the police and there are few actions taken by the prosecutors (which
is loosely speaking the system in England & Wales where the main goal seems to be the
review of the police evidence to avoid hindsight bias rather than a more detailed or goaloriented investigation as in the American system), plea-bargaining should take place
earlier since not much evidence will be produced after the police investigation. When
most of the evidence is produced during trial, then plea-bargaining should take place at
later stages, an argument with significant implications for France and Italy.

B.

Informal versus Formal Plea-Bargaining

Taking the view of plea-bargaining as any kind of out-of-court settlement between the
prosecution and the defence that avoids a trial raises immediate questions concerning
the informal versus formal nature of these procedures. The advantages of a formal
system of plea-bargaining are transparency and the availability of information (hence
the less serious asymmetries of information usually associated with secrecy) not only
for the accused, but also for the victims and for the community in general, offering
better accountability of prosecutors and defence counsellors. Part of this improved
accountability can be achieved by an appropriate design of procedural rules that can only
be implemented in a formal plea-bargain. Finally, if we are to subject plea-bargaining to
judicial review we require a formal mechanism.62
There are three main disadvantages with formal mechanisms. One is that they are
usually more costly than informal negotiations. A second disadvantage is that the parties
are under the pressure of third parties (victims, judges, public opinion) that might limit
the ability of prosecutors to strike an effective deal in some circumstances. The last
disadvantage has to do with the need formally to enter a guilty-plea, which means that
if the bargain is later rejected by the judge, the accused could be in trouble if a change of

62

Our understanding is that any agreement subject to judicial review is a formal plea-bargain. In the
United States, no plea-bargain, either formally or informally negotiated, can be entered without a judge
signing off on it, so in that respect we consider both forms of ‘formal plea-bargaining’. Third party
pressure will be present no matter if the negotiation was done informally or formally since they will be
made public after the judge approves of it. So third party scrutiny will always be present in formal or
informal negotiations since the ‘people’ are a party to the case in criminal law.

15 MJ 3 (2008)

339

Nuno Garoupa and Frank H. Stephen

plea is not allowed.63 In an informal plea-bargain, if the judge rejects the deal, the losses
for the accused are less serious since no formal guilty-plea has been entered.
In our view, an informal mechanism can only be superior if agency costs on the defence
side, as well as on the prosecutor’s side, are relatively unimportant. The reasoning is that
an informal mechanism is essentially based on the interaction between the prosecutor
and the defence side without the interference of third parties. However, if agency costs
are substantial and third party interference is required, a formal mechanism seems more
appropriate.
The main conclusion from these considerations must be that one cannot separate
the use of plea-bargaining from the rules of criminal procedure and criminal law
in general. The balance between the benefits of saving trial costs and the costs of an
inefficient contract is crucially determined by criminal law. Naturally, different rules
and laws can but only generate a different optimal response to the balance of costs and
benefits. However, criminal law might also shape, in a determinant way, the incentives
for prosecutors and defendants, as we discuss in the following subsections.

C.	The Defendant’s perspective
The economic theory of plea-bargaining treats the defendant’s side in a very simple way:
the defendant accepts the contract proposed by the prosecutor as long as the expected
payoff it provides is higher than the best alternative (e.g., go to trial). The most important
problem is that such a decision is usually taken by the defendant’s lawyer (even if this is
not formally the case, it will be very much influenced by her). Therefore, the defendant’s
side is affected also by a contract between these two parties, the accused and the lawyer.
If such a contract is good enough to perfectly align the interests of both sides, then
there are no agency costs (notice here that the principal is the defendant and the agent
is the defence lawyer). In tort litigation, the usual solution is to implement contingent
fees or some modified fee version (such as the British conditional fees).64 Such contracts
63

64

340

According to P. Hungerford-Welch, Criminal Litigation and Sentencing (Cavendish Publishing, 6th ed.
2004), in England and Wales, a change from not guilty to guilty is possible at any stage before the verdict
is given. However, a change from guilty to not guilty is at the discretion of the judge, and the defendant
must offer a reason for his change of mind (R v Sheikh [2004]). In the United States, it is discretionary
for the court to permit or deny the withdrawal of a plea of not guilty to allow a guilty plea to be entered
(see Douglas v. State, 663 N.E.2d 1153 (Ind. 1996)); for a withdrawal of a defendant’s guilty plea before
sentencing, the defendant has the initial burden of proving good cause or a fair and just reason for
withdrawing the guilty plea by a preponderance of the evidence. If the defendant establishes sufficient
grounds to warrant withdrawal, the burden shifts to the state to prove that it will be substantially
prejudiced by withdrawal of the plea. A defendant may also withdraw his guilty plea after sentencing if
he has shown ‘a manifest injustice’, i.e. if new material evidence was discovered after sentencing.
A. Daughety and J. Reinganum, ‘Economic Theories of Settlement Bargaining,’ 1 An. Rev. of L. and
Social Sciences 35 (2005) (an excellent survey on the economics of tort litigation). See W. Emons and
N. Garoupa, ‘US-style Contingent and UK-style Conditional Fees: Agency Problems and the Supply of
Legal Services’, 27 Managerial and Decision Economics 379 (2006) (with respect to the economics of

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

function to align the interests of the client and lawyer. However, such a practice is hardly
ever observed in criminal litigation, being in fact forbidden in many jurisdictions (not
least because a substantive transference of financial resources is usually not involved in
criminal litigation).65 Moreover, it is known that a very large proportion of defendants
are relatively poor (hence most are found to be judgment-proof). We expect agency costs
to be quite substantial, either because many defendants are on legal aid or defended by
a public defender or because they lack the necessary wealth to design a contract that
aligns incentives. When there are serious agency costs it is possible that contracts are
inefficient, that is, a certain contract should be rejected by the defendant, but is accepted
by the defendant’s lawyer (even when the settlement’s authority is formally assigned to the
defendant, the decision will nevertheless be based on the advice and on the information
provided by the counsellor who is supposed to act in the client’s best interest but does
not necessarily do so).66
Our reasoning thus moves the analysis in a new direction. In order to assess pleabargaining, we need to understand the role of the defendant’s lawyer in criminal cases.67

65

66

67

contingent fees). See also R. Iyengar, ‘Not Getting Their Due Process: An Evaluation of Federal Indigent
Defense Counsel’, ALEA 2006 conference paper, available at http://law.bepress.com/cgi/viewcontent.
cgi?article=1681&context=alea. But see M. McConville et al., ‘Standing Accused: The Organization and
Practices of Criminal Defence Lawyers in Britain,’ in Oxford Monographs on Criminal Law and Justice
(OUP, 1994) (with respect to the practice in the United Kingdom).
For example, even in the United States, contingent fee agreements between defendants and their lawyers
are proscribed as unethical in criminal cases, and expressly forbidden by the American Bar Association
Model Rules of Professional Conduct. See e.g., Lewin v. Law Offices of Godfrey G. Brown, 8 Misc. 3d
622, 798 N.Y.S2d 884 (N.Y. City Civ. Ct. 2005) (stating that a lawyer may not enter into an arrangement
for, charge, or collect a contingency fee for representing a defendant in a criminal case); American Bar
Association Model Rules of Professional Conduct Rule 1.5 (d)(2)).
See Easterbrook, ‘Criminal Procedure as a Market System’, 309 (argues that the agency problem, while
pervasive, is trivial because of reputation effects and is no different at any stage of criminal procedure).
We do not find this argument persuasive. In particular, as demonstrated below, a fixed payment system
increases the benefit to the lawyer of an early settlement. The confidential nature of the plea bargain
discussion and the lack of information to accused persons render the reputation effect implausible. As
we argue below different remuneration systems for defence lawyers have different implications for the
intensity of this agency problem. In any case, a plea bargain can be presented to the accused as being
a better result than going to trial, even when it is not. Ibid., 310 (Easterbrook also argues that those
benefiting from publicly financed defence benefit more from a negotiated plea than those paying for
their own defence). Again this will depend on the nature of the remuneration system of the publicly
funded defender. Easterbrook appeals to a model of plea bargaining contained in an Appendix to the
paper. However, this model abstracts from the very agency problem that is the focus of our discussion
because bargaining is not just between the prosecutor and the defendant as implied by the model, thus
ignoring the role of the defendant’s lawyer.
With the notable exception of the UK literature on legal aid, we have not found any previous work on
this matter in economics. Most economic models treat the defendant’s lawyer as the perfect agent of
the accused. See H. Gravelle and N. Garoupa, ‘Optimal Deterrence with Legal Defense Expenditure’, 40
Economic Inquiry 366 (2002). The role of the defence lawyer does feature in the literature on the legal aid
expenditure in the United Kingdom. See e.g. A. Gray, P. Fenn, and N. Rickman, ‘Controlling Lawyers’
Costs through Standard Fees: An Economic Analysis’, in Access to Criminal Justice (Blackstone Press,
1996); A. Gray, P. Fenn and N. Rickman, ‘An Empirical Analysis of Standard Fees in Magistrates’ Court

15 MJ 3 (2008)

341

Nuno Garoupa and Frank H. Stephen

In particular, the remuneration scheme has an important influence. If lawyers are
not paid for the effort exerted to represent and defend their clients, we might have a
problem of moral hazard (shirking) and adverse selection (bad lawyers, at least from a
market perspective, are matched with poor defendants). Both effects will in turn lead to
situations where the lawyer wants to get as many cases as possible in order to make more
money and spend as little time as possible on each case.68 Plea-bargaining is an obvious
way to achieve such objectives.
Therefore plea-bargaining could be highly inefficient not necessarily because out-ofcourt settlements are bad in criminal law, but because there are substantial agency costs
incurred by one of the two sides of the contract-like negotiation. It seems to us that
the introduction and development of plea-bargaining cannot be dissociated from the
reform of the defendant’s legal representation (unfortunately governments tend to leave
legal aid out of the picture when they are reforming criminal law and procedure). As the
judgment-proof nature of the defendant is the main origin of agency costs, legal aid is a
fundamental piece of the puzzle.
Pro-bono representation lacks the necessary incentives for lawyers to exert effort
(though we concede that intrinsic motivation could play an important role in mitigating
this problem) and, even if mandatory as in some jurisdictions, there are adverse selection
problems (experienced and high quality lawyers will subcontract to inexperienced and
young lawyers who end up doing most of the work, thus providing such defendants with
inexperienced, and sometimes low quality, legal representation).
We turn now to the different means by which publicly-funded representation may
be provided to accused persons facing criminal charges. Although certain financial
criteria may need to be met in order for the accused to qualify for legal aid, we abstract
from that issue to concentrate on the differential effects of the different mechanisms for
providing defendants with legal services, taking a closer look at the impact the different
mechanisms will have on the magnitude and nature of the agency problem between the
accused and his defence lawyer. Following on from this theoretical discussion, we draw
on a number of empirical studies for further enlightenment.

68

342

Criminal Cases’, Lord Chancellor’s Department Research Paper 5/99 (1999); F. H. Stephen, ‘Reform of
Legal Aid in Scotland’, 8 Hume Papers on Public Policy 23 (2001); F. H. Stephen and C. Tata, ‘Impact
of the Introduction of Fixed Payments into Summary Legal Aid: Report of an Independent Study’,
Scottish Executive Justice Department (2006); P. Fenn, N. Rickman and A. Gray, ‘Standard Fees for
Legal Aid: Empirical Analysis of Incentives’, 59 Oxford Economic Papers 662 (2007); F. H. Stephen, G.
Fazio and C. Tata, ‘Incentives, Criminal Defence Lawyers and Plea Bargaining,’ 28 Int. Rev. of L. and
Ec. 212 (2008).
The significance of this has been brought home by the findings of recent research on the impact of
changing the remuneration contract of defence lawyers in the Scottish legal aid system from one of
payment for work done to a fixed payment per case. See e.g. Stephen and Tata, ibid.; and Stephen, Fazio
and Tata, ibid.

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

1. Legally-Aided Private Defenders
In a number of jurisdictions publicly-funded defence services are provided through a
system of outsourcing69 to independent firms of lawyers. The agency problems that arise
in such a system depend on the nature of the remuneration system by which the lawyers
are paid for this work. It should be noted that such systems of legal aid involve multiple
agency relationships. In addition to the agency relationship between the accused person
and his lawyer, there is also one involving the lawyer and those providing payment. The
agency problems are in fact inter-related. Mechanisms designed to reduce the magnitude
of one set of agency costs may result in an increase of agency costs in another interrelated
agency relationship. In particular, remuneration systems for defence lawyers, designed
to reduce the level of inputs to the efficient level through a system of fixed payments or
block contracting, drive a wedge between the financial interests of the lawyer and the
desire of the accused person for his lawyer to exert the maximum effort on the case.
Consider a system of legal aid in which the defence lawyer is paid from public funds
for all time spent legitimately working on the case and for all expenses incurred to search
for evidence relevant to the defendant’s case. Under such a contract, the defence lawyer
will spend time on the case up to the point where the payment received is equal to the
lawyer’s opportunity cost of time, i.e. the next best use of his time. This will of course be
less effort than the accused would desire since the accused would wish the lawyer to exert
effort so long as his or her marginal valuation is not negative. Efficiency (and thus social
welfare) would require that effort be exerted up to the point where the payment equalled
the marginal benefit from expending more time on the case. This is likely to result in
the lawyer’s effort level being greater than the socially desirable level but less than that
desired by the accused.70 In the current context of plea-bargaining, this suggests that
the defence lawyer is likely to spend time and effort in gathering evidence against the
state’s case when it would have been socially optimal to settle with a plea bargain. Thus,
under such a remuneration system, we would expect more cases to go to trial or at least
to continue to the day of the trial (i.e. result in cracked trials).
Under these circumstances, criminal legal aid expenditure is likely to be higher than
socially desirable and largely determined by supplier driven decisions. This will lead
to a desire by governments to reduce such expenditure (as in the separate jurisdictions
69

70

See T. Goriely, ‘Legal Aid Delivery Systems: Which Value for Money in Mass Case Work? A Summary
of International Experience’, Lord Chancellor’s Department Research Series 10/97 (1997) (socalled judicare systems such as the Criminal Defence Service in England & Wales operated by the
Government’s Legal Services Commission and criminal legal aid scheme operated by the Scottish Legal
Aid Board).
This of course assumes that a socially desirable level of effort (whatever that level might be) exists
and that the lawyer’s opportunity cost is no higher than the rate being paid for legal aid work. This
assumption is plausible where specialisation in criminal work takes place since this is mostly legally
aided. Thus the market rate for civil work (particularly non-legally aided) is not attainable by criminal
law specialists.

15 MJ 3 (2008)

343

Nuno Garoupa and Frank H. Stephen

within the United Kingdom71). In order to control public expenditure, governments are
likely to prefer to move to a remuneration system which reduces the incentive for lawyers
to over-supply effort.72 This may be achieved by using fixed73 or standard payments74 per
case or a block contracting75 system. A problem with each of these systems is that, while
they substantially reduce the agency problem between the lawyer and the funding entity,
they exacerbate the agency problem between lawyer and accused. By, essentially, paying
a fixed fee for each case or block of cases, there is an incentive for the lawyer to reduce
effort on individual cases to the minimum necessary to satisfy any externally set quality
standard. Given the difficulty of externally monitoring the necessary lawyer effort, it is
likely that a degree of chiselling on the amount of effort used is possible. One form of
this which is pertinent to plea-bargaining is that the fixed payment may encourage the
defence lawyer to advise acceptance of a plea bargain at an earlier stage than might be
desirable from the accused’s perspective or than is socially optimal. Whilst this might,
ceteris paribus, reduce the number of cracked trials, thus reducing inconvenience to cited
witnesses, and making more efficient use of court and judicial time, it works against
the client’s interests and may result in adverse consequences in terms of verdicts and
sentences.
A recent study of the impact of changing the remuneration system for defence lawyers in
Scotland for summary cases provides evidence which is consistent with such behaviour.76
A system of payment for work done in defending clients in the Sheriff summary courts
71

72
73

74

75

76

344

For further economically-motivated discussion on this, see the works cited at footnote 67 above. It
should be noted that a significant difference between the policy discussions in Scotland on the one hand
and those in England & Wales on the other is that the former focussed on criminal legal aid while the
latter examined civil legal aid. In Scotland during the 1990’s over 60% of legal aid expenditure was for
criminal cases, while in England & Wales that proportion went on civil legal aid. See Stephen, ‘Reform
of Legal Aid in Scotland’. Indeed in Scotland during the 1990’s, 32% of all legal aid expenditure (civil
and criminal) went on cracked trials in the Sheriff Courts.
Again, the assumption is that there is an efficient level of effort socially desirable (whatever the level
might be).
Fixed payments for cases under summary jurisdiction were introduced in Scotland in 1999 where a fixed
payment was paid for all work up to and including the first 30 minutes of any trial. This payment is £500
in Sheriff summary courts and stipendiary magistrates’ courts and £300 in district court. Additional
payments are made for each subsequent day of the trial. See further Stephen and Tata, ‘Impact of the
Introduction of Fixed Payments into Summary Legal Aid’.
Standard payments were introduced for cases in the magistrate court in England & Wales in 1993. See
generally Gray, Fenn and Rickman, ‘An Empirical Analysis of Standard Fees in Magistrates’ Court
Criminal Cases’.
Block contracting systems have been introduced in England & Wales for many categories of legal aid.
See among others, E. Cape and R. Moorhead, ‘Demand Induced Supply? Identifying Cost Drivers in
Criminal Defence Work’, A Report to the Legal Services Commission (2005). See also A. Dnes and N.
Rickman, ‘Contracts for Legal Aid: A Critical Discussion of Government Proposals,’ 5 Eur. J. of L. and
Ec. 247 (1998) (on the incentive for lawyers to chisel on quality under such systems).
See e.g. Stephen and Tata, ‘Impact of the Introduction of Fixed Payments into Summary Legal Aid’;
C. Tata and F. H. Stephen, ‘‘Swings and Roundabouts’: Do Changes to the Structure of Legal Aid
Remuneration Make a Real Difference to Criminal Case Management and Outcomes?’, 52 Criminal L.
Rev. 722 (2006); Stephen, Fazio and Tata, ‘Incentives, Criminal Defence Lawyers and Plea Bargaining’.

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

and district courts (the lowest two jurisdictions in Scotland) was replaced by a system of
fixed payments from April 1999. The research compared the trajectories of cases in these
courts prior to the change in payment system with those after the introduction of fixed
payments. Two important results were identified: (i) a reduction in the proportion of
cases going to trial; and (ii) a reduction in the proportion of cases where a guilty plea was
entered.77 This had the effect of increasing both the proportion of cases terminating at
the pre-trial hearing78 and the proportion of cracked trials. Both of these effects increase
the likelihood that cases terminated after an attempt to plea bargain. Interviews with
defence lawyers also revealed that after the introduction of fixed payments it was less
common for defence lawyers to interview prosecution witness prior to the trial,79 since
the cost of doing so was now covered by the fixed payment whereas it had been a billable
cost under the previous system. Such an effect is consistent with either an increase or a
decrease in plea-bargaining but nevertheless shows that plea-bargaining is influenced
by the nature of the defence lawyer remuneration contract. The net increase in cracked
trials probably represents an increase in plea bargains. This research could not determine
which level of lawyer effort was optimal but nevertheless indicates that it changed with
the payment system suggesting that the effort put into the underlying plea-bargaining
process is sensitive to payment system.
Given that payments for legally-aided work are lower than rates for privately funded
work, there is adverse selection with respect to quality of lawyers.80 Usually remuneration
is fixed and not based on performance; therefore moral hazard is quite likely. Even when
contracting with lawyers is for block services rather than case-by-case, management of
individual cases will be likely to minimize effort disutility for the lawyer rather than the
defendant’s probability of acquittal. This problem can only worsen if, as some research
has concluded, most defence lawyers develop an expectation that the case will end in a
guilty plea, and become inclined to presume guilt.81
77

78
79
80

81

The change in remuneration system was found to be associated with an increase in the proportion of not
guilty pleas because legal aid is not given in Scotland where a guilty plea is entered at the pleading diet.
The inference drawn from the data is that the reduction in average income per case associated with the
change in remuneration system induced lawyers to undertake more cases. There was a corresponding
increase in the ratio of legally aided cases to total cases in these courts after the change in payment
system.
Known in Scottish summary criminal procedure as the Intermediate Diet.
A procedure known in Scotland as precognizing witnesses. During the period covered in this research
there was no system of pre-trial disclosure of evidence in Scotland.
The problem we refer to here is based on a comparison of rates paid by legal aid and by private clients.
This situation is mitigated, as in the case of the UK, when there is little or no privately funded criminal
defence work. Nevertheless, it is possible that some quality sorting takes place at the beginning of legal
careers, with the better graduates going into the more profitable legal areas.
In the United States, budgetary limitations have tremendously impacted upon the salaries for the
lawyers in public funded legal services. The problem most evident by these budgetary constraints is
the lawyer shortages found in public legal work, which leads directly to very high caseloads for those
lawyers working not-for-profit. There is evidence that the caseload for a lawyer in legal aid surpasses
the limit to the number of persons a lawyer can competently represent generally, thus affecting the

15 MJ 3 (2008)

345

Nuno Garoupa and Frank H. Stephen

2. Public Defenders
A system of public defenders has the usual agency costs of a state bureaucracy. Given
that salaries are lower than in law firms, but job security is higher, most public defenders
are quite likely to be more risk averse than lawyers in law firms. Therefore, they are
more likely to prefer safe strategies in criminal litigation, and plea-bargaining is one of
them.82
Public defenders, unlike private lawyers, are salaried lawyers whose income does not
depend upon the amount of time devoted to individual cases or the methods by which
cases are resolved.83 As a result, public defenders enter guilty pleas for their clients more
often than private lawyers.84 Therefore, low salaries and heavy caseloads make public
legal representation for indigent defendants less effective.
Moreover, it is not only remuneration that is fixed, but also promotions and annual
budget are usually based on seniority rather than performance.85 Moral hazard problems
therefore arise. There is yet another important problem: public defenders and prosecutors
play a repeated game that could easily generate behavioural norms that favour collusion
between the prosecutor and the public defender at the expense of the defendant.86
A review87 of legal aid delivery systems has discussed the experiences of the use of
public defenders in a number of English speaking jurisdictions. It reports research that
suggests that in Australia public defenders are cheaper than publicly funded independent

82

83
84
85

86

87

346

quality of their work. See S. E. Mounts, ‘Public Defender Programs, Professional Responsibility, and
Competent Representation’, Wisconsin L. Rev. 473 (1982).
There could be a countervailing factor to be considered. Public defenders have considerably more job
security than private lawyers and therefore they have less at stake. Private lawyers usually have their
reputation and finances at stake if they take risky paths.
See Alschuler, ‘The Defense Lawyer’s Role in Plea-Bargaining’.
Ibid.
Note that where performance is evaluated in such a bureaucratic system the number of cases dealt
with is likely to be a performance indicator. This is likely to encourage plea bargaining to settle cases
before trial. In other words the incentives become very similar to those of a private defence lawyer
remunerated through fixed payments or block contracts. See e.g. Stephen, ‘Reform of Legal Aid in
Scotland’, 38–39.
Clearly generating behavioral norms that favor collusion between prosecutor and public defender is
not always bad for the defendant. Certain established norms could increase efficiency (in both time
and money) in the plea bargaining process, which will ultimately benefit the defendant. Ethics plays an
important role in that respect. In the United States, the American Bar Association rules on professional
responsibility guard against collusion between prosecutor and defence lawyer at the expense of
defendant. See American Bar Association Model Rules of Professional Conduct, Rule 1.2 (a lawyer
shall abide by a client’s decisions concerning the objectives of representation and, as required by Rule
1.4, shall consult with the client as to the means by which they are to be pursued. A lawyer may take
such action on behalf of the client as is impliedly authorized to carry out the representation. A lawyer
shall abide by a client’s decision whether to settle a matter. In a criminal case, the lawyer shall abide by
the client’s decision, after consultation with the lawyer, as to a plea to be entered, whether to waive jury
trial and whether the client will testify).
See Goriely, ‘Legal Aid Delivery Systems: Which Value for Money in Mass Case Work? A Summary of
International Experience’.

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

defence lawyers because they have an incentive to get through as many cases as possible.
The evidence from Canada suggests that public defenders enter more guilty pleas, spend
more time in plea and sentence bargaining with prosecutors but that their clients were
no more likely to be found guilty and were less likely to face prison sentences. The review
concludes that although Australian public defenders spend less time on cases, evidence
is required on whether this is because they handle the easy cases, provide worse service,
or are more efficient.
A very detailed research study88 on a pilot public defence solicitors’ office in Scotland
has produced comparative evidence on relative performance, including outcomes, with
private defence solicitors funded through legal aid. The study reveals that clients (i.e. the
accused) defended by the public defender were less likely to be satisfied and were less
likely to voluntarily use the same defence lawyer again than the legally-aided clients of
private lawyers.89 The study also found that clients of public defenders were more likely
to have their case concluded before the day of trial (65%) than those of private (legallyaided) lawyers (59%). However, when account is taken of case characteristics (including
the type of crime) the difference was even greater (71% vs. 59%) and significant at the
1% level of significance. The inference that public defenders’ cases were more likely to
conclude early probably explains why the research found that there was a small but
significantly higher probability of the accused being found guilty when defended by the
public defender.90 Overall, this evidence suggests that public defenders are more likely to
be involved in concluding a plea bargain that could be against their client’s interest.91
Public defenders, because of their bureaucratic context, may have less incentive to
innovate than private solicitors financed through fixed legal aid payments. At least in UK
and Commonwealth jurisdictions, a career pattern is likely to emerge of public defenders
alternating between the public defence and prosecution services. This is likely to further
exacerbate the lack of trust in public defenders displayed by clients in the Scottish study
discussed above.92
88
89
90
91
92

T. Goriely et al., The Public Defence Solicitors’ Office in Edinburgh: An Independent Evaluation (Scottish
Executive Central Research Unit, 2001).
Ibid., 194.
Ibid., 4, 105.
Taking the client’s primary interest to be non-conviction. Obviously a more lenient sentence is an
objective once a conviction is certain.
For example, in the United States, many indigent defendants are suspicious of, and even hostile
towards, their defenders as a result of the disparity in payment for public defenders versus private
lawyers. Indigent defendants often refer to their public defenders as ‘dump trucks’, which is a term that
derives from the defendant’s belief that defenders are not interested in giving a vigorous defence, but
rather seek only to ‘dump’ them as quickly as possible. Such sentiment sparked a series of courtroom
attacks on public defenders by their clients. When confronted with inadequate representation, and
with a judge unwilling to appoint a private lawyer, some indigent defendants believe that the best way
to get rid of their public defenders has been to physically assault them in open court. See S. O’Brien,
S. Pheterson, M. Wright and C. Hostica, ‘The Criminal Lawyer: The Defendant’s Perspective’, 5 Am. J.
of Criminal L. 283 (1977); P. B. Wice and P. Suwak, ‘Current Realities of Public Defender Programs: A

15 MJ 3 (2008)

347

Nuno Garoupa and Frank H. Stephen

3. Efficient Solutions
From our discussion it is clear that designing an efficient system for state-funded or
state-provided defence lawyers is not an easy task. The first binding constraint is that
performance measures are difficult to construct, and it is well-known from incentive
theory that bad measures might actually create more problems. Notably, if we increase
pay to those lawyers who are successful in defending their clients (to mitigate moral
hazard), we might induce a self-selection of cases based on the merit as perceived in early
stages (difficult or perceived lost causes would not find an eager or motivated defender
who would have an incentive to engage in early plea-bargaining to cut losses). The reason
for this problem is that measuring success in a system that has variable sentences for a
given crime and in which the charges can be varied by the process of plea bargaining is
conceptually difficult.
The second important constraint is that of how to introduce market-related
instruments (in order to mitigate or solve adverse selection) into a system that has serious
problems of asymmetric information, complexity and very little financial leverage on the
side of the client (the defendant). A system of fixed payments with freedom to choose
the particular supplier is a first approximation to an ideal voucher system but it still
suffers from moral hazard except for repeat players, but repeat players are likely to be
better informed on the nature of the plea bargain anyway (although experience is not
necessarily synonymous with learning in this context).
Although different countries have engaged in notably different solutions to solve these
two constraints (reforms of legal aid in criminal procedure have been quite common
in Europe recently), we can at most hope for second-best solutions. We should also be
quite careful in introducing objective performance measures and pay-for-performance
in defence counselling in the absence of a strong analytical understanding of complexity
in the criminal justice system. Given the institutional specificities of the criminal
justice system (where market incentives are not universal), it is also important to assess
comprehensively the transplantation of performance measures in defence counselling.

D.	The Prosecutor’s perspective
The role of the prosecutor in plea-bargaining depends very much on the understanding
of the nature of the accuser in criminal law. Tort litigation, for example, simply aims
at gaining compensation for some wrongdoing (thus assuring an efficient deterrence of
accidents). We could say that the prosecutor aims at obtaining a conviction in such a
way that the criminal compensates society for his crimes (hence achieving an efficient
deterrence of crime). Such an approach would make the prosecutor a kind of plaintiff’s
National Survey and Analysis’, 10 Criminal Law Bulletin 161 (1974); J. Kirsch, ‘Public Defenders’, New
West Magazine (7May 1979), 52.

348

15 MJ 3 (2008)

Why Plea-Bargaining Fails to Achieve Results in so Many Criminal Justice Systems

lawyer in criminal cases, an interpretation that several scholars seem to have of the
role of the prosecutor in adversarial systems.93 For reasons that are beyond the scope
of this article, but have been discussed within the Law and Economics literature,94 the
plaintiff in criminal cases is society and not the direct victim.95 Therefore, for the present
authors, the prosecutor is best characterized as a kind of lawyer who seeks the most
efficient punishment in the interests of society.96 That is, the economic model of the
‘efficient prosecutor’. The prosecutor is therefore the agent of society, with an aggregate
perspective of the criminal justice system, whereas lawyers usually have the perspective
of an individual case. How then are the interests of the agent to be aligned with those of
the principal (society)?
Economists have recognized that it is not obvious why prosecutors maximize social
welfare since there is no obvious high-powered contract between them and society. Some
scholars argue that if prosecutors do not maximize social welfare, they will be sacked or
will not be re-elected (where that is possible, i.e. in the U.S.). For example, in the United
States, the tremendous power exerted by criminal prosecutors is counterbalanced by
sentencing guidelines, the supervisory powers doctrine, the doctrine of separation of
powers, professional discipline, and the political process.97 These mechanisms help lower
agency costs between the prosecutors and society. In a sense, contrary to the typical
short-term relationship between defendant and his lawyer, the prosecutor has a longterm relationship with society (her entire career) and hence there is a repeated game.
Badly behaved prosecutors are likely to be punished at some point, and this should be
enough to align the general interests of society with those of the prosecutor. And so the
‘efficient prosecutor’ is born. Notice also that once one accepts the ‘efficient prosecutor’ it
only makes sense to allow her to have complete discretion.98
93

94
95
96
97
98

Boari and Fiorentini, ‘An Economic Analysis of Plea Bargaining’ (it is an immediate problem with such
interpretation in inquisitorial systems). See also B. Gershman, ‘The Prosecutor’s Dilemma: Bargains
and Punishments’, 14 Georgetown J. of Legal Ethics 309 (2001); L. Griffin, ‘The Prudent Prosecutor’, 14
Georgetown J. of Legal Ethics 259 (2001).
See Bowles, Faure and Garoupa, ‘The Scope of Criminal Law and Criminal Sanctions: An Economic
Analysis and Policy Implications’.
See R. Pinto, ‘The Public Interest and Private Financing of Criminal Prosecutions’, 77 Washington Uni.
L. Q. 1343 (1999).
By no means is the pursuit of punishment the only function of prosecution, but that only assists our
critique that the model of the ‘efficient prosecutor’ has serious problems.
A. W. Dunahoe, ‘Revisiting the Cost-Benefit Calculus of the Misbehaving Prosecutor: Deterrence
Economics and Transitory Prosecutors’, 61 N.Y.U. Annual Survey of Am. L. 45 (2005).
Easterbrook, ‘Criminal Procedure as a Market System’ (this is the essence of Easterbrook’s argument
for the ‘efficient prosecutor’. And yet the same line of reasoning is usually rejected by economists for
the ‘efficient legislator’ or the ‘efficient government’. Easterbrook argues that prosecutorial discretion is
efficient because, inter alia, the prosecutor faces incentives to behave in the public interest which other
‘administrators’ do not. Clearly he has in mind a US style prosecutor rather than a UK or continental
European career prosecutor in a civil service post. He argues that judges, because of security of tenure,
do not have the same incentive to behave efficiently. As the discussion in the text below suggests, we
challenge this logic). See also D. Bjerk, ‘Making the Crime Fit the Penalty: The Role of Prosecutorial

15 MJ 3 (2008)

349

Nuno Garoupa and Frank H. Stephen

We take the view that there are several reasons to think that agency costs between
society and prosecutor are usually quite high. Institutional arrangements in place, even
direct elections, do not guarantee that ‘inefficient prosecutors’ are punished, and hence
private interests might prevail over the maximization of social welfare. Those who
monitor prosecutorial behaviour – either the electorate that votes every four years or
the bureaucratic bodies that prevail in most countries – have their own agenda that is
usually not social welfare maximizing. Moreover, we are faced with the usual problem
of specialization versus capture. The larger and less specialized the monitoring body is
(e.g., the electorate), the less they are able to properly evaluate the decisions taken by the
prosecutors. The smaller and more specialized the monitoring body (e.g., a supervisory
body composed of senior and former prosecutors), the more likely it is to be captured.
Once we recognize that prosecutors are quite unconstrained by a powerful
monitoring body that assures efficient selection of cases and allocation of effort, two
important questions arise: the preferences or goals of the prosecutors; and how certain
constraints on prosecutorial decision-making (mandatory prosecution, prosecutorial
guidelines, judicial scrutiny of prosecutorial decisions) are actually weak substitutes for
effective monitoring. It is clear that any kind of constraint on prosecutors only makes
sense if we think that the ‘efficient prosecutor’ does not exist. We should emphasize that
the choice of cases by the prosecutor does not guarantee efficient deterrence of crime
and the maximization of external effects of going to trial (further development of the
law or jurisprudence and testing of doctrines) when the goals are different from social
welfare.99
The economic theory of prosecutors’ goals is divided in two branches. One branch,
the American-oriented branch, proposes that prosecutors want to maximize convictions
or severity of sentences (there is an ongoing debate here).100 The most common argument
is that the electorate, in general, and politicians, in particular, tend to evaluate success

99

100

350

Discretion under Mandatory Minimum Sentencing’, 48 J. of L. and Ec. 591 (2005); J. Bowers, ‘Grassroots
Plea Bargaining’, 91 Marquette L. Rev. 85 (2007) (looking at how interaction with the police constrains
the prosecutors); R. D. Covey, ‘Fixed Justice: Reforming Plea Bargaining with Plea-Based Ceilings’,
82 Tulane L. Rev. 1237 (2007) (urging abandoning constraints on prosecutors’ discretion and instead
preventing courts from imposing harsh sentences).
But see A. Burke, ‘Prosecutorial Passion, Cognitive Bias, and Plea Bargaining’, 91 Marquette L. Rev.183
(2007) (when prosecutors face significant biases that impede them from achieving social welfare
maximization).
See e.g. C. Albonetti, ‘Prosecutorial Discretion: The Effects of Uncertainty’, 21 L. and Society Rev.
291 (1987) (showing the decision to prosecute aims at avoiding uncertainty); E. Glaeser, D. Kessler,
and A. Piehl, ‘What do Prosecutors Maximize? An Analysis of the Federalization of Drug Crimes’, 2
Am. L. and Ec. Rev. 259 (2000); G. Huber and G. Sanford, ‘Information, Evaluation, and the Electoral
Incentives of Criminal Prosecutors’, 46 Am. J. of Political Science 334 (2002); R. Boylan and C. Long,
‘Salaries, Plea Rates, and the Career Objectives of Federal Prosecutors’, 48 J. of L. and Ec. 627 (2005); R.
Boylan, ‘What do Prosecutors Maximize? Evidence from the Careers of US Lawyers’, 7 Am. L. and Ec.
Rev. 379 (2005); M. Raghav, M. Ramseyer, and E. Rasmusen, ‘Convictions versus Conviction Rates: The
Prosecutor’s Choice,’ Harvard Law School mimeograph (2006); R.