The Economics of Litigation - DecisionBoundaries

The Economics of Litigation

 

Download:The Economics of Litigation

 

[Equations are best formatted in the downloadable PDF Above]

All substantive areas of law have a common concern with the processes by which legaldisputes get resolved. The existing corpus of economic literature on courts is modest,
but understanding the litigation process has become important, as courts intrude more
forcefully upon resource allocation. The current cost of trials is unprecedented, although
difficult to quantify.This article consists of four sections. Section 1 focuses on the application of economictools to the study of courts and outlines the chronology of a legal dispute. In our framework,
legal disputes are resolved at various stages of a sequential decision-making process in which
parties have limited information and act in their own self-interest. Section 2 reviews the
predictions obtained from modeling these decisions, and Section 3 discusses their normative
significance. Section 4 concludes.

 

 

 

The Economics of Litigation

Carlos A. Abadi

April 22, 2019

All substantive areas of law have a common concern with the processes by which legal

disputes get resolved. The existing corpus of economic literature on courts is modest,
but understanding the litigation process has become important, as courts intrude more
forcefully upon resource allocation. The current cost of trials is unprecedented, although
difficult to quantify.

This article consists of four sections. Section 1 focuses on the application of economic

tools to the study of courts and outlines the chronology of a legal dispute. In our framework,
legal disputes are resolved at various stages of a sequential decision-making process in which
parties have limited information and act in their own self-interest. Section 2 reviews the
predictions obtained from modeling these decisions, and Section 3 discusses their normative
significance. Section 4 concludes.

1

Chronology of a Legal Dispute

Legal scholarship has long concerned itself with how the rules and practices controlling
adjudication affect the quality of court decisions. Much of the existing economic litera-
ture on courts concerns a variety of microeconomic models involving perfect competition,
bargaining, principal-agent relationships, and collective choice.There are special attributes
of legal disputes that must be taken into account when adapting any of these models to
the study of courts. We develop such a list by briefly describing the chronology of a legal
dispute.

Initially, in the first stage, there is an underlying event, such as an accident or crime,

in which one person (the injurer) allegedly harms another (the victim). The frequency of
harm is affected by decisions that people make concerning activities and precaution. To
illustrate, the probability of a car accident increases with the amount that a person drives
and decreases with the amount of precaution taken when driving. High levels of certain
types of activities and little precaution in performing them increase the frequency with
which one person harms another.

Curtailing the activity or taking greater precaution to lower the social cost of the harm

is costly in itself. As a result, economic efficiency requires balancing the cost of harm

1

 

 

 

against the cost of avoiding it. If the parties were able to bargain together, the balance
is struck by the market. This observation is the source of the best-known proposition in
the economic analysis of law, the Coase theorem (Coase 1960), which states that, in the
absence of impediments to exchange, legal entitlements will be allocated efficiently in the
market regardless of their initial allocation by law.

1

In many situations, however, bargaining is inhibited or blocked, and the social costs of

harm are externalized. For example, drivers and pedestrians do not negotiate agreements
in advance to allocate accident costs. For these accidents, the balance between harm and
the cost of avoiding it must be struck by law, not the market. The initial allocation of
legal entitlements is therefore essential to providing efficient incentives for activity levels
and precaution against external harm.

In the second stage of a dispute the party that allegedly suffered harm decides whether

or not to assert a legal claim. A rationally self-interested person makes those decision by
solving a sequential game that balances immediate costs (hiring a lawyer, filing a complaint)
against benefits expected in the future (the proceeds from settlement or victory at trial).

The third stage occurs after a legal claim is asserted, but before trial. During this

stage the parties reply to complaints, attend preliminary hearings with the judge, engage
in pretrial discovery, and set trial dates. The overall objective of the court at this stage is
to encourage plaintiffs (the victims) and defendants (the injurers) to bargain together and
settle their disputes. The attribute of litigation bargaining – rivalry, communication, side
payments, interdependency, and uncertainty – characterize bargaining games as analyzed
in microeconomics. The third stage of the litigation process can be viewed, then, as a
bargaining game whose cooperative solution corresponds to a settlement out of court, and
whose noncooperative solution corresponds to an adversarial trial.

In settlement negotiations, as in any bargaining game, the interests of the two parties

diverge with respect to division of the surplus, but converge with respect to an efficient
resolution of the dispute. A legal dispute is resolved efficiently when legal entitlements
are allocated to the parties who can bear them at least cost, and the transaction costs of
dispute resolution are minimized.

A complicating feature of litigation bargaining is that the parties in most legal disputes

are represented by lawyers, whose interests are not identical to their clients’. Designing
contracts between attorneys and their clients so that incentives favor good representation
is an agency (principal-agent) problem superimposed upon the basic bargaining game.

The law prods disputants to resolve their differences by private bargaining and, when

negotiations fail, the courts dictate a resolution in the fourth and final stage of a legal
dispute. From the perspective of settlement bargaining, the expected outcome of a trial
defines the threat points of the parties.

Unlike settlement bargaining, the adversarial

element dominates in trials, with each party trying to win as much of the stakes as possible.
Litigants, as represented by their counsel, view trials as negative-sum games.

1

Robert Cooter presents an analytical treatment of the Coase theorem.

2

 

 

 

TABLE 1

MATCHING STAGES OF A LEGAL DISPUTE TO ECONOMIC MODELS

Stage1:

Harm – market models or externality models

Stage 2:

Assertion of legal claim – decision under uncertainty to be solved recursively

Stage 3:

Bargaining – strategic bargaining model with principal-agent overlay

Stage 4:

Trial: negative-sum game for disputants, grafted onto collective choice by impartial court

Adjudication by the courts has two distinct outputs: dispute resolution and rule-

making. From the private viewpoint, trials are a method of resolving disputes between
rational self-interested plaintiffs and defendants. But, from a social viewpoint, trials are a
mechanism for interpreting and creating laws to regulate and govern society. The decision-
makers in appeals courts, where laws are made and interpreted, are judges whose interests
differ substantially from those of plaintiffs and defendants.

In our chronology of a legal dispute we distinguish among initial harm, the assertion of

a legal claim, settlement bargaining, and trial. The initial harm can be analyzed by market
models or externality models of the kind economists apply to conventional economic goods.
The decision to assert a claim is a decision under uncertainty to be solved recursively by
computing the expected values of subsequent stages in the dispute. Microeconomic models
of bargaining are applicable to settlement bargaining. The limitations of bargaining theory,
however, are not as severe as the absence of an economic theory of disinterested behavior
that is needed to explain how judges interpret statutes and make laws.

The match-up between stages in a legal dispute and the economic modeling of them is

summarized in the table above;

A chart depicting the frequency with which disputes go from prior to subsequent stages,

with ”harm” at the bottom and ”appeals court trial” at the top, looks like a broad-based
pyramid. A typical finding is that ten disputes settle out of court for every one that is
tried, although this figure varies widely by type of dispute. Generally speaking, the farther
along the litigation process the dispute has gone, the better the empirical evidence. The
steep slope of the ”dispute pyramid” and the relative superiority of data describing the top
as opposed to the bottom make the empirical study of litigation especially difficult.

2

Resolving Disputes Through the Litigation Process

What incentives do litigants face as they proceed through the litigation process? We answer
this question in this section in the context of a formal model that is a hybrid of the models
of suit, settlement, and trial that have been developed by William Landes, Richard Posner,
Steven Shavell, and others. In Section 3 we go on to treat a number of related normative
questions.

Our hybrid model of the litigation process stylizes facts to direct the reader’s attention

to fundamental causal relations. We assume that all accidents occur between strangers

3

 

 

 

Table 2

DEFINITION OF VARIABLES

c

cp

= cost to plaintiff of asserting a legal complaint

c

sp

(c

sd

)

= cost to plaintiff (defendant) if the case is settled

c

tp

(c

td

)

= cost to plaintiff (defendant) if the case is tried

D

p

(D

d

)

= plaintiff’s (defendant’s) estimate of the compensatory damage to be awarded
if the plaintiff wins at trial

H

p

= victim’s subjective value of the harm he suffers

L

p

(L

d

)

= potential plaintiff’s (defendant’s) subjective expected value of a legal claim

p

tp

(p

td

)

= plaintiff’s (defendant’s) subjective probability that a complaint that is asserted
will be tried rather than settled

p

vp

(p

vd

)

= plaintiff’s (defendant’s) subjective probability of a plaintiff victory at trial

q

p

(q

d

)

= victim’s (injurer’s) subjective probability that an accident will occur and the
victim will assert a claim

S

p

(S

d

)

= subjective value to plaintiff (defendant) of settling the case rather than going to trial

T

p

(T

d

)

= subjective value to plaintiff (defendant) of possible damage award by the court

x

p

(x

d

)

= plaintiff’s (defendant’s) precaution against harm that gives rise to legal disputes

and, therefore, outside a market context. This rules out ”Coasian” bargaining and the
possibility that prices convey information to the parties. There is a single injurer, who
becomes a defendant, and a single victim, who becomes a plaintiff, when a suit is filed.

2

Both parties can affect the probability of an accident occurring. Initially, it is assumed
that each party bears its own litigation costs. This assumption will be relaxed when we
analyze alternative rules for allocating litigation costs.

We will forgo chronology and discuss the four stages of the litigation process in reverse

order.This allows us to emphasize the their interdependence and, in particular, the fact
that a decision at each point in the process depends crucially on the parties’ expectations
about the future. The variables in Table 2 above will be used in the analysis.

2.1

Behavior at Trial

2.1.1

Trial Effort of Plaintiffs and Defendants

The plaintiff goes forward with the trial because he expects to win something from the
defendant. The value of this transfer depends on the intrinsic merits of the case, which is
determined in part by the relevant laws and in part by the particulars of the case. The
pertinent laws may describe the burden of proof, the legal standard of evidence, the scope of
damages (including a possible augmentation of compensatory damages reflecting punitive

2

When the suit is filed by a regulatory agency, or by lawyers on behalf of a class of victims, the analysis

becomes more complex because the objectives of the active plaintiff may differ from the victims’.

4

 

 

 

damages in tort cases or treble damages in antitrust cases), and the rules of procedure. The
relevant facts might describe the past actions of plaintiff and defendant, and the particular
circumstances that determine the application of legal rules. These actions would include
the levels of precaution in a tort suit, promises in a contract suit, facts of ownership in a
trespass suit, etc.

3

The amount that the plaintiff expects to win is determined not by the merits of the case

alone, but also by the efforts the parties devote to winning. The efforts of the parties can
be measured by expenditures on the trial, denoted c

tp

and c

td

for plaintiff and defendant,

respectively. The subjective expected trial payoff to the plaintiff is this given by the function

T

p

(c

tp

, c

td

)

The plaintiff’s cost of proceeding with the trial must be set against his expected win-

nings. The plaintiff’s expected gain from bringing suit, prior to trial and net of trial costs,
is thus given by

T

p

(c

tp,c

td

)−c

tp

(1)

Similarly, defendant’s subjective expected loss, including trial costs, is given by

T

d

(c

tp

, c

td

) + c

td

(2)

The partial derivative [Equation in PDF]

can be thought of as the marginal productivity of

plaintiff’s effort at trial. If effort is productive, then [Equation in PDF]

[Equation in PDF]

similarly for defendant.

The expenditure of effort at trial can serve an important signaling function. The court

must decide cases in which the defendant’s liability in civil suits or guilt in criminal ones
is uncertain, because information about the law or the facts is incomplete. The parties
to the dispute usually know more than the court about crucial facts and transmitting the
information to the court is costly. Thus the effort that a party puts into trial provides a
signal to the court. A stronger signal increases the probability that the judge or jury will
favor the facts as represented by its sender.

4

While effort is typically productive for both parties, the relative productivity depends

on the merits of the case in a complex way. For example, if the defendant is negligent, effort
by the plaintiff to discover and prove the facts can be very productive. But, on the other
hand, effort by the defendant to represent the facts differently could also be productive.

The variables c

tp

and c

td

are chosen by the litigants as part of their trial strategy.

The plaintiff chooses c

tp

to maximize his expected gain, while the defendant chooses c

td

to

minimize his expected loss. The first-order condition for the plaintiff is:

∂T

p

∂c

tp

+

∂T

p

∂c

td

i

p

= 1

(3)

3

The parties’ perception about the law and the facts may differ from each other and may be inaccurate

as well.

4

This model is developed in Rubinfeld and David Sappington in the criminal context in which there is

a single defendant.

5

 

 

 

where i

p

= dc

td

/dc

tp

is the plaintiff’s conjectural variation – a measure of how the defen-

dant’s costs will change in response to the change in plaintiff’s cost. Equation (3) tells us
that the plaintiff will expend money at trial so that the marginal benefit from more effort
is equal to the marginal cost. The decision about how much money to spend during trial
thus depends on strategic considerations related to i

p

. A similar condition applies to the

defendant.

A number of legal variables that can influence trial effort have been explored in the

literature. Consider, for example, the effect of an upward adjustment to compensatory
damages on the effort of both parties. To make such an evaluation consider a game in which
plaintiff and defendant take each others’ effort as fixed (as in a Nash game). Then i

p

= 0,

so that equation (3) reduces to ∂T

p

/∂c

tp

= −1. An upward adjustment to compensatory

damages should increase the marginal productivity of plaintiff’s effort (∂T

p

/∂c

tp

), It follows

that c

tp

will increase (as will c

td

, under comparable assumptions). Thus, as the stakes

increase, both parties will increase their trial expenditure.

If strategic behavior occurs, so that i

p

6= 0 and/or i

d

6= 0, the analysis is more com-

plicated. Even when conjectural variations are not zero, we would expect a positive rela-
tionship between the upward adjustment to compensatory damages and the effort of both
parties. But, without further structure, we cannot rule out the possibility that strategic
behavior will lead to a contrary effect. Suppose, for example, that the game is sequential
and only three levels of effort are possible – high, medium, and low. Initially, when only
compensatory damages are given by the court, both parties choose a medium level of effort.
Subsequently, when damages are increased, the plaintiff chooses first and opts for a high
level of effort. Now, it may be in the defendant’s interest to opt for a low effort level, real-
izing that he’s unlikely to win the case whatever his choice. In this example, augmenting
compensatory damages led one party to make more effort and the other party to make less.

The rule for allocating legal costs is another variable whose effect upon trial effort has

been explored. In the American legal system, the parties to a dispute usually bear their
own legal costs. In the UK, however, the loser in a trial bears the legal costs of the winner.
Theoreticians have compared the effects of these two rules. John Hause uses a model in
which the probability that plaintiff prevails at trial is a function of the legal expenditures
of both parties. He concludes that a switch from the system in which both parties pay
their own legal fees to one in which the loser pays the winner’s costs would increase trial
expenditures for those suits that go to trial. Avery Katz also takes interactions and strategic
considerations into account when he models trial effort. He reaches essentially the same
conclusion as Hause, arguing that higher stakes raise the marginal value of additional
expenditures to both parties. At the same time, the possibility that the losing side will
pay part of the winner’s litigation expenses lowers the expected marginal cost of litigating
the case.

5

5

Other trial effort papers include Ronald Braeutigarn, Bruce Owen, and John Panzer.

6

 

 

 

2.1.2

The Outcome of the Trial

The outcome at trial (a win for plaintiff or defendant) is the result of a complex interaction
between the efforts that both parties put into the trial and the facts and law of the case.
If both parties are only interested in winning the stakes in this trial, rather than being
interested in the law or reputation on future disputes, then the levels of effort chosen and
trial outcomes will depend on the relative productivities of both parties.

In many cases, however, parties are likely to engage in similar litigation in the future –

so that a repeated game framework becomes more appropriate. When one or both parties
is concerned about the future , the probability that the plaintiff will win may increase or
decrease. Typically, the probability of winning will increase for the party with a future
interest in victory. To see why, consider a Nash game in which the parties initially have a
50 percent chance of winning. Now suppose that the defendant acquires a future interest
in victory, so that the cost of a trial judgment increases by a multiple m, where m > 1. It
follows from equation (3) that the defendant’s expenditures on trial , which were formerly
determined by ∂T

p

/∂c

td

= −1, are mow given by ∂T

p

/∂c

td

= −1/m. The defendant’s trial

effort will consequently grow, and the probability of defendant’s victory will increase to a
level above 50 percent.

6

Jeffrey Perloff and Rubinfeld have suggested that defendants typically have more at

stake than plaintiffs because defendants are likely to be involved in future litigation of the
same type.

7

In this situation, the loss to the defendant is greater than the plaintiff’s gain.

The defendant will, consequently, choose to spend more on trial than the plaintiff and will,
therefore, have a greater than 50 percent chance of winning. Using an antitrust dataset,
Perloff and Rubinfeld find support for this view, because approximately 70 percent of all
antitrust cases in their data set are won by defendants. This percentage is substantially
higher than the rate of defendant victories in the cases studied by Priest and Benjamin
Klein.

Settlement bargaining is a filter and the small percentage of cases that pass through it

and go on to trial are not a random selection of all suits. Consequently, the frequency with
which plaintiffs win at trial depends on the nature of the selection process. Hypotheses
about the selection process and the frequency of plaintiff victory have been advanced and
studied empirically. These hypotheses all build on the view that cases fail to settle as
a consequence of a mistaken prediction about the outcome of a trial made by one of the
parties. If, for example, the predictions of defendants and plaintiffs are normally distributed
around the true mean, each party is equally likely to make a mistake. Priest and Klein use
such an argument, supported by data, to conclude that cases go to trial in which defendants

6

See Rubinfeld. This is not an equilibrium argument, because the plaintiff will respond to the defendant’s

increase in effort, but the direction of change should be the same in equilibrium as in the first round of
responses, provided that the reaction functions have the expected shape.

7

Econometric evidence also confirms that juries will award greater damages when there are corporate

defendants, all else equal (James Hammitt, Steve Carroll, and Daniel Relles . Jury awards are discussed
more generally by Mark Peterson, Peterson and George Priest, and Mark Shanley and Peterson.

7

 

 

 

and plaintiffs each have a 50 percent probability of winning (see also Priest).

Donald Wittman replies that when the parties disagree about expected trial award the

50 percent rule of Priest and Klein can be seriously biased. Disagreement concerns in part
the meaning of ”winning”, which is ambiguous in the context of trials. The plaintiff ”wins”
a civil suit, in one sense of the word, if a court awards damages or provides injunctive relief.
Many civil suits, however, concern not the fact of defendant’s liability but its extent. From
this perspective, the plaintiff ”wins” at trial only if the damage award is larger than the
defendant’s settlement offer.

A further ambiguity arises when one of the parties to a dispute has a future interest

in the trials’s outcome.

An interest in reputation or precedent by one of the parties

makes the stakes asymmetrical. Even if the 50 percent rule were true when the parties
are symmetrically situated, it will not be when there are asymmetries. To illustrate, a
defendant who wants to cultivate a reputation for tough bargaining will contest cases that
he has little chance of winning. Conversely, a defendant who wants to avoid the publicity
of a trial will settle cases that he has a high probability of winning.

2.2

Settlement Versus Trial

The economic issues surrounding whether suits are settled or brought to trial have a long
history in the law and economics literature. The early literature, including work by Landes,
John Gould, Posner, and William Baxter, treated the private incentives of the parties,
while Shavell went further by distinguishing private from social incentives. Most of the
more recent literature on the economics of settlements has moved toward a game-theoretic
framework in which there are information asymmetries and a variety of sequences by which
settlement offers are made by one or both parties. In this subsection we treat the parties’
incentives, and then briefly survey the theoretical results concerning the effect of changes
in policy instruments when the parties behave strategically.

2.2.1

The Incentives of Plaintiffs and Defendants

In some legal disputes there is scope for settlement, whereas in others trials may be in-
evitable. To distinguish between them, consider a civil dispute in which the parties have
no future interest, so the bottom line is how much defendant pays plaintiff. The parties
have expectations about the size of the transfer that would result from a trial and its cost.
Plaintiff’s expected gain from going to trial, net of trial costs, is given in equation (1) above,
while defendants’s expected loss, including trial costs, is given in equation (2) above. These
expected gains and losses represent the subjective threat values of the parties. Any change
that strengthens one player’s threat value should increase his gains from the bargain. For
example, Hugh Gravelle shows that plaintiffs with smaller risk aversion will receive larger
settlements in a model in which courts have imperfect information.

8

 

 

 

The sum of the subjective threat values equals the players’ assessment of the games’s

noncooperative value:

Noncooperative value

= (T

p

− c

tp

) − (T

d

+ c

td

)

= (T

p

− T

d

) − (c

tp

+ c

td

)

If a trial can be avoided, the parties must still bear the transaction costs associated with
settlement, which are denoted c

sp

and c

sd

for plaintiff and defendant, respectively. In a

settlement, the net transfer necessarily equals zero. The cooperative value of the game
thus equals the actual net transfer (zero) less the transactions costs incurred:

Cooperative value = −(c

sp

+ c

sd

)

The difference between the cooperative and noncooperative values of the game equals the
surplus;

Surplus = (c

tp

+ c

td

) − (c

sp

+ c

sd

) + [T

d

− T

p

]

(4)

The surplus from cooperation equals the sum of the term in braces, representing the dif-
ference in the costs of trial and settlement, and the term in brackets, representing the
difference in subjective expectations about damages awarded at trial.

Transactions costs are less when a case is settled than tried:

(c

tp

+ c

td

) − (c

sp

+ c

sd

) > 0

Indeed, trial costs are so much greater than settlement costs that many authors choose the
simplifying assumption that settlement costs are nil, that is, c

sp

= c

sd

= 0. In this case,

the surplus reduces to the gap of expectations of the parties concerning the value of trial:

Surplus = (c

td

+ c

sp

) + (T

d

− T

p

) = (T

d

+ c

td

− (T

p

− c

tp

)

For a risk-neutral plaintiff, the subjective value of the possible damage award at trial,

denoted T

p

, equals the money value of expected damages, D

p

, times the subjective prob-

ability of their award, denoted p

vp

, that is, T

p

= p

vp

D

p

.

similarly, for a risk-neutral

defendant, T

d

= p

vd

D

d

. When plaintiff and defendant have the same expectation about

trial (p

vp

= p

vd

and D

p

= D

d

), they concur about its expected value , so that T

p

= T

d

. If

the parties are relatively pessimistic about the prospects at trial (p

vp

< p

vd

and D

p

< D

d

),

plaintiff will expect to win less than defendant expects to lose, so that T

p

> T

d

.

If the surplus is negative, the disputants prefer a trial to any possible settlement, so

trial is inevitable. If the surplus is positive, however, there is scope for settlement out
of court. The frequency of settlements presumably increases with the magnitude of the
surplus. There is more scope for settlement when litigation is costly (c

tp

and c

td

are large),

negotiations are inexpensive (c

tp

and c

sd

are small), and the disputants are pessimistic

about trial outcomes (p

vp

< p

vd

, D

p

< D

d

). As a result, any policy that increases litigation

9

 

 

 

costs. lowers settlement costs. or makes disputants pessimistic about their trial prospects,
will increase settlements.

Now consider the effect of risk aversion upon litigants with the same information about

possible outcomes of trials. A trial represents a gamble, so the subjective value to risk-
averse disputants will diverge from its expected value. For example, when the parties are
both risk-averse and they have the same expectations about trial, their subjective values
of trial diverge:

T

p

< p

vp

D

p

= p

vd

D

d

< T

d

Risk aversion thus increases the surplus as given in equation (4), which presumably in-
creases the probability of a settlement. Notice that risk aversion increases the surplus in
the same way as pessimism – by increasing the difference between the subjective values of
plaintiff’s trail gains and defendant’s trial losses.

2.2.2

The Effects of Legal Rules

Most models have assumed that settlement occurs automatically whenever the surplus in
equation (4) is positive. This assumption has the effect of ruling out strategic behavior.
Its main justification is pragmatic – predictions can be derived readily from nonstrategic
bargaining models, whereas strategic models are often intractable. Given the fact that the
term in braces (c

tp

+ c

td

) − (c

sp

+ c

sd

) is positive, and assuming nonstrategic bargaining,

the trial/settlement split falls into two zones determined by the sign of the surplus, with
one intermediate point:

[T

d

− T

p

] > −(c

tp

+ c

td

) − (c

sp

+ c

sd

) ⇒ settlement

[T

d

− T

p

] > −(c

tp

+ c

td

) − (c

sp

+ c

sd

) ⇒ tipping point

[T

d

− T

p

] > −(c

tp

+ c

td

) − (c

sp

+ c

sd

) ⇒ trial

These relationships help to generate a prediction about the effect of treble damages and

punitive damages upon the frequency of trials. Consider how augmenting damages affects
a case at the tipping point between settlement and trial. The fact that the term in braces is
positive implies that [T

d

− T

p

] < 0 at the tipping point. Augmenting damages increases the

absolute value of this negative magnitude, which tips the case into the trial zone, so there
are more trials and fewer settlements. Augmenting damages in a nonstrategic bargaining
model thus strengthens the tendency of optimism to cause trials.

This conclusion must be modified once account is taken of the resulting change in

trial effort. Augmenting damages increases the stakes of the trial, which typically elicits
more effort at trial by the parties, as explained above. With more effort, the term in
braces (c

tp

+ c

td

) − (c

sp

+ c

sd

) increases in value. The resulting increase in the surplus

from cooperation presumably makes settlement more likely.

Risk aversion also affects the comparison. Augmenting damages, by increasing the

stakes at trial, makes trial more risky, which makes trial less attractive to risk-averse

10

 

 

 

disputants. Risk-averse disputants at the tipping point under a regime of compensatory
damages may be nudged into settlement by a change to a regime of augmented damages
because trial has become too risky.

In sum, augmenting damages increases the stakes at trial, which has the opposing

effects upon the ability to settle out of court in a nonstrategic model. On one hand, more
weight is given to the parties’ optimism, which tends to increase the frequency of trials.
On the other hand, trials become more costly and more risky, which tends to decrease their
frequency.

Similarly, changing the rule for distributing trial costs has the opposing effects upon

the ability to settle out of court. Under the American rule, the parties know with certainty
that they will pay their own costs, and under the UK rule the loser pays all the plaintiff
is expected to bear trial costs, c

tp

+ c

td

, only if he loses, which occurs with probability

(1 − p

vp

). Assuming risk neutrality, we can modify equation (4) to contrast the two rules:

U S = (c

tp

+ c

td

) − c

sp

+ c

sd

) + [p

vd

D

d

− p

vp

D

p

]

(5)

U K = (1 − p

vp

+ p

vd

)(c

tp

+ c

td

) − (c

sp

+ c

sd

) + [p

vd

D

d

− p

vp

D

p

]

(6)

Equation (6) reduces to (5) when the parties have the same subjective beliefs about the
probability of plaintiff’s victory, p

cp

= p

vd

, but not otherwise.

Consider the effect of the change in rules on a case at the tipping point between set-

tlement and trial under the American rule. The surplus is zero at the tipping point, so
(5) = 0 by assumption. The term in braces in equation (5) is positive by assumption.
Hence, p

vp

D

p

> p

vd

D

d

, which indicates that the parties are optimistic about trial. Assume

that this optimism extends to the expectations about the probability of plaintiff’s victory,
so that p

vp

> p

vd

p

vp

> p

vd

implies (5) > (6). This fact and the fact that (5) = 0 imply

that (6) < 0. Thus, a switch to the UK rule causes the surplus at the tipping point to turn
negative, resulting in more trials.

This conclusion must be modified when trial effort and risk aversion are considered. The

switch in cost distribution rules from US to UK increases the stakes at trial by including
trial costs in the gamble. The effect of higher stakes upon trial effort and risk aversion
has already been discussed – the effect of a switch in the distribution rule for trial costs
parallels the effects of augmenting damages.

8

Theory tells us, therefore, that a switch from

the rule of each pays his own (American) to the rule of loser pays all (British), tends to
increase the frequency of trials by giving more weight to the parties’ optimism, and to
decrease it by making them more costly and more risky.

The direction the overall effect upon the frequency of trials from changing the cost

distribution can, therefore, not be determined by theory alone. The common belief among
lawyers that fewer suits will occur when the loser pays more of the legal costs enjoys

8

Katz compares trial efforts of the parties under the British and American rules. See also Braeutigam,

Owen, and Panzar, Shavell, and Posner

11

 

 

 

some support from empirical economics.

9

This belief has motivated a modification of

the American rule to more closely resemble the British rule. Geoffrey Miller considers
one prominent example. Suppose plaintiff rejects the defendants’ final offer to settle for
a specified sum of money and that, after a trial, plaintiff is awarded less than the final
settlement offer. Under these circumstances the plaintiff can be said to have lost in court
relative to the settlement offer. Rule 68 of the code governing procedure in Federal courts,
which is similar to procedural rules in a variety of states, specifies that a plaintiff who loses
in court relative to the defendant’s settlement offer must pay some of the winner’s court
costs, including such items as the cost of depositions and filings, and excluding attorneys’
fees.

10

Whatever effect this rule has on the frequency of trials, it strengthens the bargaining

position of defendants.

2.2.3

Strategic Aspects of Settlement Behavior

The nonstrategic bargaining model in the preceding section assumes that disputes will
always settle out of court when the cooperative surplus, as perceived by the players, is
positive, whereas trials will occur when it is perceived as negative. There is, however,
another case of trials – the distribution problem itself. The problem of dividing the surplus
created by settlement is a source of instability that can lead to bargaining breakdowns.

Attempts by theorists to model the distribution problem in bargaining games in general

have produced not a consensus among economists, but a variety of predictive and normative
theories that rival each other. The unsatisfactory state of bargaining theory is reflected in
strategic models of the litigation process. One approach to settlement bargaining generates
definitive predictions by making restrictive assumptions about the scope of bargaining, the
timing of offers, and the ability of the parties to transmit information. Thus, in Janusz
Ordover and Ariel Rubinstein and Ivan P’ng, the settlement amount is fixed and not open to
bargaining. In P’ng the defendant knows whether he is negligent and uses this information
to decide whether to make a settlement offer, whereas plaintiff responds without knowing
for certain whether the defendant was negligent. In Lucian Bebchuk, the settlement amount
is endogenous, but the plaintiff knows the actual harm and the defendant knows only the
probability distribution of possible harms. In Bebchuk’s model, the relatively uninformed
plaintiff makes the first and only settlement demand, which the defendant must either
accept or reject in favor of a trial. The response of defendant to plaintiff’s offer conveys
some information about the defendant, but uncertainty persists, so cases go to trial.

The models discussed so far do not face the distribution problem squarely.

When

bargaining over distribution of the cooperative surplus, the players are uncertain about the
extent to which other parties will concede. A rational player will gauge his demands such

9

Donald Coursey and Linda Stanley found a higher settlement rate when their experimental subjects

decided disputes under Rule 68 than under the British Rule

10

The Supreme Court extended Rule 68 to cover attorneys’ fees in cases where the statute under which

the action is brought allows recovery of attorneys’ fees. See Merrick v. Chesney, 473 U.S. 1, 105 Ct. 3012

12

 

 

 

that the gain from settling on slightly more favorable terms is offset by the increased risk of
negotiations breaking down. Thus, the optimal strategy in settlement bargaining balances
a larger share of the stakes against a higher probability of trial. When these considerations
are balanced at the margin, expected utility is maximized relative to the distribution of an
adversary’s possible strategies. A bargaining equilibrium can thus be characterized as a
situation in which everyone maximizes expected utility given complete knowledge about the
distribution of strategies followed by others.

11

This equilibrium concept has the advantage

of permitting strategic behavior to cause trials.

While the optimal strategy, as characterized above, is best relative to the distribution of

the other party’s strategy, it is not necessarily best against the actual strategy that will be
chosen. A party may overestimate a particular opponent’s willingness to make concessions,
which can cause a breakdown in settlement negotiations and a trial. To illustrate, suppose
that the parties must choose between a hard strategy (make no concessions) and a soft
strategy (concede). Each party knows the frequency with which these strategies are chosen
by others, but no one finds out his particular adversary’s strategy in a specific dispute until
after it is resolved. Trials occur under these circumstances when both parties commit to
hard strategies, and settlements occur otherwise. To illustrate, suppose that in equilibrium
30 percent of plaintiffs and 30 percent of defendants are pursuing hard strategies. Then 9
percent of disputes end in trial and 91 percent settle out of court, and no one is surprised
by these proportions.

This equilibrium concept presupposes some means by which the parties generate their

expectations about the probability that other players will concede. A full account of the
genesis of concessionary expectations would go beyond the legal process into psychology
and sociology. For the purposes of economic analysis, however, it is usually sufficient to
assume that disputants have expectations prior to beginning a legal dispute, and then
to predict how the legal process modifies them. An earlier example of this approach by
Cooter, Stephen Marks, and Mnookin sought for conditions under which the predictions
of the nonstrategic models could be extended to strategic bargaining. The specification
of the information structure of the game is not adequate in this early work, but has been
corrected in subsequent work. In William Samuelson both parties make settlement offers
simultaneously, so uncertainty persists and bargaining can fail. Stephen Salant assumes
that plaintiffs come in two types, slightly injured and badly injured, and defendants cannot
tell them apart in pretrial bargaining.

Several papers have applied the concept of sequential equilibria (David Kreps and

Robert Wilson) to settlement bargaining, notably Urs Schweizer, Jennifer Reinganum and
Louis Wilde, and Barry Nalebuff for civil suits, and Reinganum for criminal cases. At
each node or state of the game, each party chooses the strategy that is optimal for the
remainder of the game, given uncertainties about other players and their future actions.
Parties update their beliefs in light of information provided at each stage of the game.

11

For an analysis of this Bayesian-Nash equilibrium, see John Harsanyi

13

 

 

 

To generate definite predictions, these approaches must exploit facts about the litigation
process that prescribe sequences of moves and generate asymmetric information. Thus,
plaintiff must make the first move to assert a legal claim. In settlement bargaining for civil
disputes, the defendant often has more information concerning the existence of liability
(e.g., whether negligence can be proved), and the plaintiff has more information about the
extent of liability (e.g., how severe was the injury).

If enough structure is imposed to generate sequential moves with asymmetric informa-

tion, some predictions can be derived that may contradict the nonstrategic models. To
illustrate, an important topic in strategic bargaining is the information transmitted by the
exchange of offers. Signaling in settlement bargaining was studied by Nalebuff, who re-
lied upon information asymmetries to generate predictions about equilibria in a sequential
subgame. In the first step, plaintiff makes a single demand; next the defendant either re-
jects the demand or settles the case; finally, if the demand is rejected, the plaintiff decides
whether to proceed to trial. Plaintiff’s demand in the first stage conveys information to
defendant about the probability that plaintiff is prepared to proceed to trial. Defendant’s
rejection of plaintiff’s demand in the second stage conveys information to plaintiff about
the strength of defendant’s case. In equilibrium, plaintiffs know the distribution over the
strength of the case of defendants who settle, and defendants know the proportion of cases
that plaintiffs litigate.

A comparison of Bebchuck, Nalebuff, and our hybrid model illustrate that different

specifications of the game affect important predictions about the litigation process. Con-
sider the effect of an increase in plaintiff’s trial costs on the terms of settlement. An
increase in plaintiff’s trial costs weaken plaintiff’s threat position, which leads to lower set-
tlement offers in Bebchuk’s model and our hybrid model. While agreeing with this general
argument, Nalebuff points to an alternative possibility. He argues that when trial costs
increase , plaintiff willnot be prepared to go to trial unless he expects to win a larger judg-
ment. To make the threat of going to trial credible, he will demand a larger settlement.
A full specification of the information structure in settlement bargaining , including the
signal contained in the offers they make, may thus lead to predictions that contradict the
nonstrategic model.

2.2.4

Empirical Studies of Trial/Settlement Split

Courts have been studied by sociologists and other social scientists from both a longitudinal
and a cross-sectional point of view. However, it is relatively recently that economists have
begun the task of specifying and estimating structural models of the behavior of the parties
during the dispute resolution process. The greatest attention has focused on the settlement
decision. A satisfactory model f settlement must take account of uncertainty in settlement
bargaining, which results in specification errors, and the possibility that bargaining breaks
down due to strategic behavior.

A structural model for empirical research on the trial/settlement split can be developed

14

 

 

 

from our hybrid model. Suppose we posit that the plaintiff’s expected gain from trial
consists of the systematic component T

p

and a randomly-distributed error. Similarly, a

defendant’s expected loss from trial consists of a systematic component T

d

and a randomly-

distributed error. Trial costs may be random as well. Combining all these terms, the
cooperative surplus from settlement becomes

Surplus = (T

d

− T

p

) + (c

td

+ c

tp

) + 

(7)

where G is the systematic component and  is a random disturbance term. In this frame-
work, a dispute may fail to settle even though the systematic component of the cooperative
surplus is positive, provided that the error is large and negative.

A reduced-form model can be obtained from equation (7) in which the probability of

trial, denoted p

t

, is determined by evaluating the probability distribution function of the

systematic component of the settlement surplus:

p

t

= p

t

(G)

(8)

A question investigated empirically by using equation (8) is whether augmenting com-

pensatory damages will result in more or fewer trials. Recall our previous discussion in
which we concluded that, when damages are augmented, the tendency of optimism to cause
trial is strengthened, the tendency of risk aversion to discourage trials is strengthened, and
more costly effort is elicited to win trials that occur, which further discourages trials. Perloff
and Rubinfeld found evidence suggesting that in antitrust cases, where reputational effects
are important, and where the parties tend to be pessimistic (T

d

− T

p

> 0), treble damages

lead to a decrease in the proportion of cases resolved by trial and an increase in the number
of settlements.

Other studies, however, have suggested a contrary result, including Danzon and Lee

Lillard . They applied a model of the settlement process to medical malpractice claims.
Their model consists of two trial equations that explain the probability of plaintiff winning
(p

vp

= p

vd

= p) and the amount of verdict (T

p

= T

d

= T ), and two settlement equations

that explain the minimum demand of the plaintiff (T

p

− c

tp

) and the maximum offer (T

d

+

C

td

) of the defendant. Both the minimum demand and maximum offer depend positively

on the perceived probability of winning and the perceived verdict. As in our hybrid model,
the authors assume that cases will settle when the minimum asking price of the plaintiff
is greater than zero , but less than the defendant’s maximum offer. When the minimum
asking price is greater than the maximum offer, the case will go to trial.

Danzon and Lillard assume that an increase in the stakes involved in the case (brought

about when damages are augmented) will increase random errors proportionally. The costs
of litigation, however, increase less than proportionally. So G in equation (7) is negative
in more cases, and more cases will be litigated.

12

12

Posner comes to the same conclusion as Danzon and Lillard but makes the assumptions that the costs

of litigation are fixed, and that the parties disagree only about the probability that the plaintiff will win at
trial.

15

 

 

 

Among the interesting results of this study are the following: (i) the higher the award

at trial, the greater the probability that the case will go to trial; (ii) the higher the plain-
tiff’s probability of winning at trial (as perceived equally by both parties), the lower the
probability that the case will go to trial; and (iii) plaintiffs win only 28 percent of the cases
that go to trial.

2.3

Assertion of a Legal Claim

A dispute is initiated when a party with a complaint asserts it, either formally by filing
the required legal document, or informally by private communications between the parties.
Some legal disputes are settled privately and never come to the court’s attention.

In

other cases, such as tortious injuries to minors, the resolution of the dispute is not legally
binding until approved by the court. In the best-documented disputes, however, an official
complaint is filed by the plaintiff against the defendant.

The decision to assert a legal claim is difficult to investigate empirically because cases

that do not come to the attention of judicial authorities never enter official records. Danzon
and Lillard partly avoided this problem by studying insurance records. In two data sets of
medical malpractice, they found that 50 percent of cases were resolved before a suit was
filed, 40 percent were settled before a verdict, and 10 percent were tried to a verdict.

Asserting a complaint, whether informal or official, uses plaintiff’s time and/or money.

The expected benefit of asserting a legal claim consists of the possibility of settlement or a
favorable court judgment. Shavell and Posner, among others, have assumed that rational
decision makers assert a complaint because the cost of doing so is less than the expected
benefit. Let c

cp

denote the cost to plaintiff of asserting a legal claim, and let L

p

denote the

plaintiff’s subjective expected benefit.

The expected benefit of asserting a legal claim can be determined explicitly from pre-

ceding sections of this article. The plaintiff’s subjective value of a possible court judgment,
conditional upon a trial occurring, has ben written as T

P

(c

tp

, c

td

). Let p

tp

denote the plain-

tiff’s subjective probability that a complaint will eventually lead to a trial. The plaintiff’s
(unconditional) subjective value of a court judgment that could result from asserting a
complaint, net of litigation costs, is thus p

tp

[T

p

(c

tp

, c

td

) − c

tp

]. Similarly, let S

p

denote the

subjective expected value of settlement for plaintiff, conditional upon a settlement being
reached, which occurs with probability 1 − p

tp

. The plaintiff’s subjective expected value of

the legal claim, L

p

, is thus

L

p

= p

tp

[T

p

(c

tp

, c

td

) − c

tp

] + (1 − p

tp

)(S

p

− c

sp

)

(9)

Equation (9) implies that claims are more valuable to a victim when litigation costs and

bargaining are inexpensive (low c

tp

and c

sp

), and plaintiff is optimistic about his prospects

at trial or settlement (high T

p

and S

p

). Further, the plaintiff knows that a settlement will

occur only if it makes him better off than going to trial. As a result, L

p

must be dreasing

in P

tp

.

16

 

 

 

The decision rule for the rational plintiff balances the subjective value of a legal claim

against the cost of assering it:

c

cp

< L

p

→ Assert a legal claim

c

cp

= L

p

→ Tipping point

c

cp

> L

p

→ Do not assert a legal claim

The literature is divided, however, on the appropriate measure of L

p

for the rational

plaintiff. Suppose both parties have complete information about trial costs and outcomes
and certain other conditions in dispute are met. Then suits will be brought only when
the plaintiff’s expected benefit from trial net of trial costs is positive. Settlement costs,
probabilities, and amounts are irrelevant under these conditions. This can most easily be
seen in a model of repeated litigation in which both parties know the plaintiff’s expected
net benefits from trial to be negative. Then it will be in defendant’s interest not to agree
to settle such a case and, consequently, plaintiff will not choose to bring the case in the
first place. Thus, when both parties have complete information, the settlement probability
is zero, and L

p

is equal to the plaintiff’s expected net benefit from trial.

Bebchuck has shown, however, that under a different assumption involving asymmet-

ric information, the victim’s decision to sue may depend on the likelihood and/or the
magnitude of a settlement. In this framework the more general definition of L

p

applies.

To analyze the relationship between legal costs and legal disputes, consider a person

at the tipping point of asserting a legal claim, where c

cp

= L

p

. A change in the law that

increases trial costs will immediately lower the value of l

p

. The equation will thus tip in

the direction c

cp

> L

p

, where the claim will not be asserted. More generally, laws that

increase the costs of resolving disputes are likely to decrease the frequency with which legal
claims are asserted and increease the cost of settling those that are asserted.

These conclusions have applied to the explanation of nuisance suits. A nuisance suit

can be defined as a suit that both parties recognize as having no merit, in which case the
expected damage award is nil: T

p

= T

d

= 0. thus the plaintiff’s benefit from asserting a

nuisance complaint from equation (9) reduces to:

L

p

= (1 − p

tp

)(S

p

− c

sp

) − p

tp

c

tp

(10)

The value of equation (10) obviously cannot be positive unless S

p

is positive. It is

irrational to file a nuisance suit unless the expected value of a possible settlement is positive.
In general, the decision rule allows for the possibility of asserting claims whose expected
trial value is nil only if their settlement value is positive.

13

Why would a defendant pay damages to a plaintiff to settle a suit without merit? The

answer offered in several models turns upon asymmetric costs. The central role of cost

13

See also Bradford Cornell, who uses an option pricing model to suggest that victims will file some suits

for which the net present value is negative.

17

 

 

 

asymmetries can be illustrated by applying the Nash bargaining solution to our hybrid
model. The Nash bargaining solution gives each player his threat value plus half of the
surplus from cooperation. Assuming risk neutrality, the general solution for plaintiff (for
all suits) can be written:

[Equation in PDF at Top of Document]

Consider the effect of asymmetric trial costs on equation (11). Assume that defendant’s

trial costs are greater than plaintiff’s, c

td

> c

tp

, and assume that the players are symmetric

with respect to settlement costs and information about trial, so that c

s

= c

sp

= c

sd

, P

v

=

p

vp

= p

vd

, and D = D

p

= D

d

. Under these assumptions, equation (11) reduces to

p

v

D + (1/2)(c

td

− c

tp

) − c

s

/2

(12)

The Nash bargaining solution under these assumptions requires that plaintiff’s payoff, net
of all costs, equal expression (12). The plaintiff pays his own settlement costs, so (12)
will be satisfied if the defendant pays to the plaintiff a settlement amount, S, equal to the
expected value of the trial judgment plus half the difference in trial costs:

S = p

v

D + (1/2)(c

td

− c

tp

)

(13)

An important conclusion follows from equation (13): Assuming strict symmetry in

information and costs (including c

td

= c

tp

), the Nash solution to settlement bargaining

requires the defendant to pay the plaintiff the expected judgment from trial. Furthermore,
assuming asymmetry in costs, the Nash solution to settlement bargaining requires the party
who saves relatively more from avoiding trial to share these gains with the party who saves
relatively less.

A precise prediction about the nuisance suits follows from equation (13). For nuisance

suits, P

v

D = 0 by definition, so S = (1/2)(c

td

−c

tp

). Thus, the bargaining solution between

risk-neutral players requires that defendant refuse to settle nuisance suits in which trial
costs are symmetric (S = 0 when c

td

> c

tp

).

The preceding model explains nuisance suits by asymmetries in the costs of defendant

and plaintiff. An alternative explanation rests upon asymmetries in the timing of costs.
For example, David Rosenberg and Shavell propose a sequential game in which plaintiff
files a suit at a negligible cost. Following this, the defendant must either settle or incur
litigation costs. Only after the defendant’s action must the plaintiff either withdraw or
incur costly litigation. So long as the defendant must expend effort on litigation prior to
the plaintiff, the defendant might find a small settlement cheaper than litigation.

Bebchuk uses a slightly different model that focuses on the settlement process itself.

He shows that nuisance suits can lead to a settlement when the defendant cannot be
sure whether the plaintiff will go to trial or withdraw if there is no settlement. Finally,

18

 

 

 

Thomas develops a sophisticated model of strategic bargaining and shows that asymmetric
information, not asymmetric costs, can lead to settlement of nuisance suits for a positive
sum of money.

The condition under which the victim will asset a claim is also sensitive to fee arrange-

ments that the victim makes with his lawyer. Under a contingent fee arrangement, the
incentive to assert a claim is different from what it would be under an hourly fee arrange-
ment. Under the former, the lawyer bears some of the client’s risk in exchange for a portion
of the proceeds if the victim receives an award at settlement or trial. Under the latter,
the fee paid to the lawyer is independent of the victim’s recovery. Danzon analyzes the
effect of these two fee arrangements on the assertion of claims. She shows, for example,
that a risk-preferring contingent fee attorney will accept a case that a risk-neutral hourly
fee attorney would not take. However, it is also true that some claims that would be filed
by a risk-neutral client using an hourly fee attorney would not be taken by a contingent
fee lawyer.

14

2.4

Precaution Against Harm

Our analysis has proceeded through three stages in a legal dispute in reverse chronological
order, beginning with trial, followed by settlement bargaining, and then turning to assertion
of legal claims. The fourth and final stage to consider is the harm that one person does
to another. Harm can take many forms, such as tortious injury, breach of contract, and
trespass upon property. Injurers and victims usually have access to forms of precaution
that reduce the probability and severity of harm.

By far the greatest focus of economists who study common law rules has been on incen-

tives created for precaution by injurers and victims. Coverage ranges from the article by
John Brown (1973) in which alternative liability rules (e.g. strict liability, negligence, and
comparative negligence) are compared in a model in which accidents are treated as exter-
nalities, to the article by Coase (1960) in which similar rules are analyzed in a framework
in which injurers and victims bargain over the level of precaution that both parties take.
The level of precaution is determined by these models by a profit-maximizing or utility-
maximizing calculus in which the cost of precaution is traded off against its benefits, often
in the form of reduced liability.

Only in a few instances, however, has the analysis taken explicit account of the relation-

ship that is the subject of this section – incentives for precaution created by the litigation
process itself (see, for example, Jerry Green and P’ng). Because litigation is expensive,
expenditures on precaution will be made to reduce the probability and extent of litigation.

In extending our hybrid model to cover the relationship between litigation costs and

incentives for precaution against harm, we proceed on the assumption that harm is an
externality that cannot be cured in the market. In an externality model, harm done by
the injurer affects the victim, and the victim’s assertion of a legal claim affects the injurer.

14

Other contingent fee studies include Herbert Kritzer et al. and Danzon.

19

 

 

 

However, there is no bargaining between the parties to allocate the costs of harm before it
occurs. As a consequence, levels of precaution by the parties are determined by the legal
assignment of liability.

In an externality model the injurer trades off the cost of additional precaution against

the resulting reduction in legal claims. To formalize this optimization problem, the injurer’s
subjective probability that the victim will assert a legal claim, q

d

, is assumed to be a

decreasing, concave function of the injurer’s precaution, denoted x

d

, and other variables

not made explicit: q

d

= q

d

(x

d

), where dq

d

/dx

d

< 0, d

2

q

d

/dx

2
d

> 0.

Let L

d

denote the subjective expected cost to defefendant of plaintiff’s assertion of a

legal claim. Analogous to equation (9), this expected cost is given by:

L

d

= p

td

[T

d

(c

tp

, c

td

) + c

td

] + (1 − p

td

)(S

d

+ c

sd

)

(14)

From equation (14) it follows that more precaution by the injurer typically decreases
the expected cost of legal claims.

15

This relationship is assumed to be concave: L

d

=

L

d

(x

d

), where dL

d

/dx

x

< 0, d

2

L

d

/x

2
d

> 0. The injurer thus chooses precaution to mini-

mize the sum of the costs of precaution and legal claims:

min[x

d

+ q

d

(x

d

)l

d

(x

d

)]

(15)

Turning from injuries to victims, let x

p

denote the victim’s expenditure on precaution

against harm caused by injurer, and let q

p

and L

p

indicate the probability and value,

respectively, of the victim’s subjective expectation concerning the assertion of a legal claim.
To keep the analysis simple, assume that the victim asserts a claim if an accident occurs,
but not otherwise.

16

Thus q

p

can be interpreted as victim’s subjective probability of an

accident, which is assumed to be a decreasing, concave function of victim’s precaution (and
other implicit variables): q

p

= q

p

(x

p

), where dq

p

/dx

p

< 0, d

2

q

p

/dx

2

p

> 0.

The potential victim who suffers harm equal to H

p

receives a legal claim whose value

is denoted L

p

. The loss H

p

is a concave, nonincreasing function of the victim’s precaution

(and other variables): H

p

= H

p

(x

p

), where dH

p

/dx

p

≤ 0, d

2

H

p

/dx

2

p

≥ 0. The expected

value of the legal claim L

p

is also a function of x

p

: L

p

= L

p

(x

p

). The sign of the derivative

15

Under every rule of law known to us, additional precaution by the injurer (weakly) decreases injurer’s

liability. To illustrate, under a negligence rule, more precaution by injurer increases the probability that
the court will find the injurer not liable because he satisfied the legal standard of care. Similarly, under
a rule of strict liability, more precaution by injurer reduces the magnitude of the expected damage award.
Additional precaution by the injurer thus tends to reduce his expected cost of trial, T

d

. Furthermore,

lowering his expected cost of trial, T

d

, strengthens his threat position in settlement bargaining, so expected

settlement, S

d

, tends to fall as well. More precautions may even reduce the probability, p

t

, that disputes

will end up in trial. (More precaution tends to reduce the severity of accidents, which reduces the stakes,
and our previous analysis reached the tentative conclusion that larger stakes cause more trials.) Thus L

d

is a nonincreasing function of x. (Strategic effects of larger x

d

on plaintiff’s choice of variables are not

discussed here.)

16

Thus, we are assuming that c

cp

< L

p

for all accidents under consideration – the cost of asserting a claim

is less than plaintiff’s expected benefit L

p

.

20

 

 

 

is not generally determinate, as can be seen from equation (9).

17

The difference in value

between the harm H

p

and the lagal claim L

p

measures the victim’s net loss. The potential

victim thus chooses x

p

to minimize the sum of his precaution costs and his net loss from

harm:

min[x

p

+ q

p

(x

p

)[H

p

(x

p

) − L

p

(x

p

)]](16)

A change in law that shifts the function L

d

(x

d

) up, or increases its marginal value

dL

d

/dx

p

, will, according with equation (15), induce more precaution from injurers. Con-

versely, a change in law that shifts L

p

(x

p

) up, or increases its marginal value, dL

p

/dx

p

,

will, according to equation (16), induce less precaution from victims.

As an example, consider the effect of augmenting compensatory damages. This will

simultaneously shift L

d

(x

d

)andL

p

(x

p

) up. Assuming independence, this will lead to more

precaution by injurers and less by victims. This illustrates the contrast between distribution
and efficiency in courts. In general, therefore, compensation rules that effectuate transfers
increase incentives for precaution by one party, but they reduce incentives for the other
party.

Alternatively, consider a change in the law that increases the defendant’s subjective

probability that a given case will go to trial, that is, s

d

+ c

sd

< T

d

+ c

td

. It follows from

equation (14) that L

d

is an increasing function of the settlement probability. Consequently,

as A. Mitchell Polinsky and Rubinfeld suggest, the change in the law will shift L

d

(x

d

) up,

thereby generating more injurer precaution, and greater deterrence.

Finally, suppose another change in law increases the plaintiff’s costs of litigation. The

value of legal claims, L

p

, to plaintiffs, will decrease, thereby causing potential victims to

take more precaution. In addition, higher costs of trials will cause plaintiffs to assert fewer
claims. The cost of legal claims, L

d

, to defendants can either increase or decrease as a

consequence, depending upon whether the effect of fewer claims or more costly claims
dominates. In general, legal rules that increase the cost of resolving disputes increase
incentives for precaution by victims and may either increase or decrease incentives for
precaution by injurers.

3

Normative Issues

Legal policy has traditionally been evaluated by standards of fairness, whereas the norma-
tive standard in most economic models is efficiency. Although efficiency is more controver-
sial as a goal for law as opposed to markets, claims about efficiency have had significant
impact on legal scholarship, teaching, and, possibly, on courts (see Jerome Culp). In this

17

The plaintiff’s subjective expected value of asserting a legal claim, L

p

, equals the expected benefit,

given by equation (9), less the cost of asserting the claim. The sign of the derivative of L

p

with respect to

x

p

is indeterminate because more precaution by the victim generally reduces the magnitude of harm and

increases the proportion of the costs of harm borne by the injurer. The magnitude of these effects, which
go in opposite directions, depends upon the particulars of law and fact.

21

 

 

 

part of the article we will discuss some normative concerns that can be treated within
an efficiency framework. The first issue deals with the fourth stage of litigation, but the
remaining issues combine several stages. The last section goes further by examining the
behavior of judges as lawmakers.

3.1

Trial Effort of Litigants – A Normative Analysis

Courts and other lawmakers have several policy instruments to affect trial effort, including
(i) the legal standard of care,

18

(ii) the magnitude of damages (Rubinfeld and Sappington),

(iii) the burden of evidence production (Joel Sobel), (iv) the standard of proof (Rubinfeld),
and v) court costs (Rubinfeld and Sappington). Each of these policy instruments directly
affects the expenditure of parties at trial, and indirectly influences decisions at each stage
of a legal dispute – trial, settlement, assertion of claims, and precaution. This section
considers the direct effect of policy instruments on effort at trial and the outcomes of
trials.

First consider the effect of trial effort. Define a court decision as correct if it would

be reached by applying the law under conditions of full information.

19

Instead of having

full information, however, courts must make their decisions based on information provided
largely by the disputants themselves. Effort by plaintiffs and defendants, and the rules
governing evidence and procedure, determine a probability distribution of errors of Type
I (finding violations where conformity occurred), and Type II (finding conformity where
violations occurred).

It can be argued that more effort by both parties will disclose more information to the

court, so its decision will come closer to the full-information decision. More information
can thus reduce errors of both types. Furthermore, cases with large stakes induce more
effort by both parties. This view leads to the conclusion that bigger cases are more likely
to be decided correctly (see Posner).

One policy tool that typically induces greater trial effort by both parties is augmenting

compensatory damages. The question of whether the increase in litigation effort from
increasing damages has social value comparable to its cost has been investigated in several
studies of treble damages in antitrust law (William Breit and Kenneth Elzinga; Steven
Salop and Lawrence White). Polinsky and Rubinfeld show how the optimal damage level
changes when costly litigation is taken into account.

A different perspective is provided by considering incentives that cause more effort to

one party and less by the other. If effort is productive, then more effort by the party that
deserves to win increases the accuracy of court decisions, whereas more effort by the party

18

To our knowledge, the complex relationship between the standard of care and trial effort by plaintiffs

and defendants has not been formally studied.

19

Notice that a ”correct” decision gives everyone their due under law. This is Plato’s first definition of

justice in the Republic. Hobbes takes the view in Leviathan that there is no other concept of justice (no
justice in nature).

22

 

 

 

that deserves to lose has the opposite effect. In the context of criminal trials, Rubinfeld and
Sappington argue that more effort by innocent parties decreases the probability of Type I
errors (convicting the innocent), and more effort by guilty parties increases the probability
of Type II errors (not convicting the guilty).

Some legal rules, such as ”loser pays all”, may increase the accuracy of court decisions

by providing incentives for more effort by parties that deserve to win than parties that
deserve to lose. This point was already discussed in connection with nuisance suits, where
the British and American rules for allocating litigation costs were discussed. Thus, Mar-
ilyn Simon shows, assuming risk neutrality (but not otherwise), that a change from the
American to the British rule reduces the probability of court error.

Instead of reducing both types of errors, some policy variables decrease one type while

increasing the other. Thus, a shift in burden of proof from defendant to plaintiff might
reduce Type I errors while increasing Type II errors. Identifying the ”best” point in the
frontier between the two types of errors involves perplexing normative issues.

Insight into these normative issues can be obtained from a game-theoretic framework.

Suppose, as Sobel suggests, that both parties are bargaining with private information,
but cannot publicly misrepresent that information. The judge has a prior distribution
about the claims of the parties, but does not know their accuracy with certainty. In one
equilibrium, each party has a positive probability of winning even if he does not provide
evidence. In another equilibrium, a party wins only by presenting substantial evidence. In
general the rules of the game and, in particular, which party has the burden of production
of the evidence, will determine the equilibrium outcome.

Sobel shows, for example, with respect to this second equilibrium, that the overall cost

of obtaining evidence is lowest if the burden of production is placed on the party that has
the lowest cost of proving his claim. If the objective is to maximize the social value of
the trial process, it is often better to place the burden of production on the party with
the higher cost of providing evidence. If that party has a relatively weak claim, it will not
present a case. But, if it has a relatively strong claim, the presentation will be worthwhile,
despite the higher cost of evidence production.

20

A central normative issue in discussions of legal procedure is balancing the cost of

additional information against the benefit of reducing cost errors. There is reason to wonder
whether disputants value cumbersome procedural rules designed to produce accuracy as
highly as do courts. Private systems of dispute resolution in which the parties choose their
own rules, such as Visa’s system of arbitration among member banks, typically employ
much cheaper procedures than those adopted by public courts. Random inaccuracies are
not too serious when the stakes are small relative to the disputants’ wealth.

Unlike random inaccuracies in trial outcomes, which are unavoidable when information

is costly, systematic inaccuracies have the appearance of bias. An alleged source of bias

20

Sobel’s results are sensitive to the nature of the game and the assumptions concerning asymmetry of

information. Ordover and Rubinstein, and Samuelson describe some game-theoretic perspectives that could
lead to different conclusions.

23

 

 

 

is defendant’s identity. Econometric evidence has confirmed the belief among lawyers that
juries will award greater damages when defendants are corporations rather than individ-
uals, all else being equal (see Hammitt, Carroll, and Relles; and more generally Peterson;
Peterson and Priest; Shanley and Peterson). Perhaps courts impose rules that are too
cumbersome in an effort to reduce random inaccuracy, whereas bias is a serious concern.

3.2

Private Versus Social Incentives to Bring Suit

Many disputes involve claims for damages, which can be resolved by transfers of income.
In general, the rules for making these transfers affect the incentives of the parties subject
to current and future disoutes. It is not surprising, therefore, taht the private and social
value of suits may diverge.

Our hybrid model can be used to trace this divergence with respect to incentives to

assert legal claims. Recall that plaintiff decides whether to asset a legal claim by balancing
the subjective expected benefit from trial with the cost of filing the claim, which yields the
tipping point C

cp

= L

p

, where

L

p

= p

tp

[T

p

(C

tp

, c

td

− c

tp

] + (1 − p

tp

)(S

p

− c

sp

)

(17)

Note that the defendant’s litigation costs c

td

, which are triggered by the assertion of a

legal claim, are not borne by the plaintiff and do not figure directly in his decision to assert
a legal claim. Shavell suggested that the private costs of asserting a legal claim are less
than the social costs under the American system because the plaintiff does not bear the
defendant’s litigation costs. Thus a plaintiff who runs the gamble of asserting a legal claim
externalizes part of the cost of finding out whether it is worthless or valuable. Shavell also
noted, however, a consideration pointing in the opposite direction. When trial costs are
substantial, the private net benefit from trial may be negative even though the social gain
from deterring injurers is large.

Peter Menell countered Shavell’s argument that the private costs of suit are less than

the social costs by pointing out that when victims do not pay injurers’ costs of resolving
disputes, injurers may respond by taking additional precaution. The additional precaution
may or may not be socially efficient. Louis Kaplow refined this argument by distinguishing
between the effect of precaution on the extent of harm and its probability. Arguments
about the divergence of private and social incentives to sue were subsequently synthesized
by Susan Rose-Ackerman and Mark Geitsfeld.

3.3

Deterrence with a Costly, Uncertain, Litigation Process

Pioneering work on incentives for precaution, such as Brown (1973), compared the efficiency
of alternative rules such as strict liability versus negligence for allocating the cost of harm.

24

 

 

 

These studies assumed that all harm is pecuniary,

21

disputes can be resolved without cost,

and courts apply clear legal standards without error. Our hybrid model will be extended to
modify the conclusions when dispute resolution is costly and courts apply obscure standards
or make errors.

A full extension would compare incentives for precaution by injurers and victims under

alternative liability rules. For this article, however, the discussion will be restricted to the
incentives for precaution under the rule of strict liability and then under a negligence rule.
A basic conclusion of the early studies is that, assuming costless dispute resolution, the
injurer’s incentives for precaution under strict liability are efficient when the defendant
must fully compensate the plaintiff. Full compensation is achieved when the victim is
indifferent between avoiding the harm or suffering it and receiving compensation.

When the result is extended to the context of costly litigation, full compensation must

include the cost of resolving the dispute, not just the harm that gave rise to it. Suppose
disputes are resolved by trials. To achieve full compensation in our hybrid model under
this assumption, a victim who is certain to win at trial must be compensated, not just for
the harm H

p

caused by the accident, but for his trial costs c

tp

plus his costs of asserting

a claim c

cp

. If, in addition, the court sometimes makes errors, so that the injured plaintiff

wins at trial with probability p

vp

, full compensation requires setting the damage award D

p

so that

[Equation in PDF at Top of Document]

A rule requiring full compensation of victims by injurers causes the latter to internalize

costs, which induces efficient precaution by them. American law, however, typically re-
quires the injurer to compensate the victim for the harm that gave rise to the legal claim,
but not for the victim’s cost of resolving the dispute. The injurer who expects accidents to
result in trials will, consequently, choose a level of precaution knowing that he must pay
for his precaution, the expected harm, and his litigation costs.

The resulting externality will distort the injurer’s precaution. It might appear that this

element of externalized cost will always cost injurers to take too little precaution relative
to the socially efficient level. In fact, this will be true if litigation is relatively costly
and precaution is relatively expensive. (In this case the efficient level of precaution will
be higher with costly litigation than without, because additional precaution reduces the
expense of litigation substantially.)

However, Polinsky and Rubinfeld have shown that this level of externalized cost may

result in a greater than efficient level of injurer precaution in cases when litigation is
relatively inexpensive and precaution is quite costly. This surprising result occurs when
additional injurer precaution substantially reduces the number of suits that victims bring,
and thereby reduces the injurer’s liability and his litigation cost. (Recall that victims only

21

Nonpecuniary injuries, such as pain and suffering in tortious accidents, can affect the total utility

without affecting the marginal utility of money. In this situation, costly compensation is inefficient.

25

 

 

 

bring suit when their expected benefit exceeds their litigation cost. Additional precaution
by the injurer can tip many victims from the region in which they bring suit to the region
in which they do not.)

A further qualification of the efficient standard is required when the outcome of litiga-

tion is uncertain and the rule of strict liability is replaced by a negligence rule.’Early models
showed that if courts set the legal standard of care equal to the efficient level of precaution,
and if they apply this standard without error, injurers will exactly conform to the legal
standard, as required for efficient precaution. Suppose, however, that courts make errors
in applying a negligence rule, as a consequence of which some negligent defendants escape
liability and some non-negligent defendants are found liable. It may be advantageous for
injurers to depart from the legal standard under these circumstances. Whether they exceed
it or fall short is indeterminate in principle, although it seems likely that injurers will want
to exceed the legal standard to allow for a margin of error by the courts within which
liability is avoided.

22

Efficiency can be achieved by appropriate adjustment in the legal

standard to offset the departure of injurers from it.

Errors by the court need not be symmetric, and damage rules other than compensatory

damages are possible. P’ng has pursued this line of thought by focusing on the deterrent
effects of Type I and Type II errors. He points out (in the context of a rule of strict liability)
that Type I errors, in which penalties are mistakenly assessed against non-violators, lower
the relative cost of violating the law, rather than conforming to it. To insure that the
appropriate incentives are created, P’ng proposes a positive adjustment to compensatory
damages so that violators must pay more, along with a subsidy for those who engage in
the activity that runs the risk of being found in violation of the law by mistake.

3.4

Lawmaking By Courts

Most analyses relating to the courts have focused on the behavior of the parties prior to and
during the litigation process. An important area of study that has received less attention
is the role of courts in the lawmaking process. This section briefly summarizes the state of
the lawmaking literature and speculates about some fruitful avenues of research.

Some economically-oriented scholars of the common law accept the positive and nor-

mative efficiency thesis, according to which judge-made law tends toward efficiency and
reinforcing this tendency is good public policy (Posner). The positive thesis is testable,
at least in principle, but there has been little quantitative research on how the common
law changes.

23

The normative thesis, while nota a statistically-demonstrable conclusion, is

22

Calfee and Craswell show that a reduced standard may be preferred, because increased precaution by

the injurer reduces the probability that a suit will be successful and that the victim will recover damages.

23

In a recent paper, Priest tried to test whether changes in doctrine by judges, which increase uncertainty,

cause an increase in the scope of disagreement among litigants. Priest’s data apparently shows that doctrinal
change and increased disagreement occur in the same year, but not which occurs first. This fact is consistent
with his hypothesis or with the rival hypothesis that changes in doctrine resolve uncertainties that cause
litigants to disagree (Priest; Cooter)

26

 

 

 

rather a conviction that some people reach by reading many cases.

How might lawmaking by courts lead to efficient outcomes? Two different explanations

correspond to two different conceptions of the common law. One conception, which regards
litigation as a market, views the common law process as driven by competition among
rationally self-interested actors. The other conception, which regards judging as an exercise
in public reason, views the common law process as driven by the theories of law embraced by
judges. We review the hypothesis that the law is market-driven here, but omit a discussion
that it is idea-driven because of the limited economic work that has been done.

Several ingenious attempts have been made to explain how competition among liti-

gants, like market competition among businesses, can produce efficiency without anyone
consciously aiming for it.

24

One such mechanism is selective litigation. Suppose that ineffi-

cient laws are more likely to be litigated than efficient laws. If inefficient laws are repeatedly
challenged in court, they may be overturned, whereas if efficient laws are less frequently
challenged, they are more likely to persist unchallenged. Selective litigation could work
like a strainer that catches inefficient laws while allowing efficient laws to slip past. The
product, being repeatedly sieved, becomes more efficient with the passage of time. Two
assumptions are enough to cause the law to evolve towards efficiency, at least weakly: (i)
a rule’s efficiency is negatively correlated to the probability that the litigants will test it in
court, and (ii) efficiency is not negatively correlated to the probability of a rule surviving
such a test before a judge.

25

For the process to operate, judges need not favor efficiency,

but they must not disfavor it.

Does litigation tend to select inefficient laws? Theory suggests a weak ”Yes”. The

more someone values a contested legal entitlement, the more that party will be prepared
to spend on litigation to obtain it. Larger litigation expenditures increase the frequency
of court challenges and improve their quality, which, in turn, increases the probability of
winning. Thus the value that a person places upon a legal entitlement should correlate
with the probability of winning it through litigation. By transferring legal entitlements
from parties who value them less to parties who value them more, the common law tends
toward efficiency.

This process can be redescribed as a contrast between distribution and efficiency. The

allocation of legal entitlements affects both the quantity of wealth and its distribution.
When legal entitlements are allocated inefficiently, the plaintiff who overturns the misal-
location stands to gain from both the increase in wealth and from its redistribution. In
contrast, when legal entitlements are already allocated efficiently, the plaintiff who over-
turns the allocation stands to gain from the redistribution of wealth and to lose from the
decrease in its quantity. Because the value of overturning inefficient laws exceeds the value
of overturning efficient laws, the frequency and quality of challenges to inefficient laws
should be higher than that for efficient laws.

24

This possibility was first raised by Paul Rubin. See also Priest; and John Goodman. For a discussion

of dynamic efficiency, see Landes and Posner; and Lawrence Blume and Rubinfeld.

25

A precise statement of the conditions for such evolution is found in Cooter and Lewis Kornhauser.

27

 

 

 

Selective litigation is similar to the ”invisible hand” in markets nut, unfortunately, the

grip of the invisible hand in courts is far weaker than on markets. A law is, by its nature,
general in the scope of its application, so challenging a law affects everyone who is, or will
be, subject of it. Most plaintiffs appropriate no more than a fraction of the value that new
precedent creates and redistributes.

The effects of a new, more efficient, precedent spill far beyond the litigants in the

case where it is set. Litigants, however, may have little regard for the social costs that
an inefficient rule imposes on others. The bias toward efficiency may be overwhelmed by
the inclination of plaintiffs to challenge laws when they can capture a large share of the
precedent’s value. Plaintiffs may thus bring suit when they expect the redistributive gains
of a successful challenge to be large, regardless of the law’s efficiency or inefficiency. The
problem with viewing litigation as a market is that redistributive gains are frequently more
important than inefficiencies in channeling litigation.

26

An exception to this pessimistic conclusion concerns laws that are vague. Bargaining

games are hard to settle when the parties do not know each others’ threat points (Elizabeth
Hoffman, Matthew Spitzer). An implication is that vague laws cause litigation. Laws whose
inefficiency derives from their vagueness will tend to be litigated until the courts achieve a
clear allocation of the underlying entitlements.

Our view is that, so far as common law tends toward efficiency, it must be driven by the

ideas of judges, not by competitive pressures in the market for litigation. There is some
evidence that the judiciary is giving a larger role to economic reasoning (Frank Easter-
brook), but there is also evidence that the judiciary tends to expand its own powers, much
as a bureaucracy engorges itself, without regard to benefits and costs. In addition, the fact
that important legal cases are decided by majority vote of panels of judges raises the pos-
sibility that courts are afflicted by the same voting paradoxes as legislatures (Eaterbrook;
Kornhauser and Lawrence Sager; Spitzer).

The ideal of an independent judiciary implies creating circumstances under which judges

decide cases that do not affect their private interests. The salary and tenure of federal
judges are independent of their performance, and their performance is apparently unrelated
to promotion to a higher court (Richard Higgins and Paul Rubin). These facts raise an issue
about whether disinteredness provides the best incentive structure, or whether competition
among adjudicators might improve the efficiency of dispute resolution.

Economists have compared the incentives of judges and arbitrators (Landes and Posner;

Robert Cooter; Christopher Bruce). Arbitrators maximize their own incomes by deciding
disputes so as to maximize the demand for their services. If an arbitrator’s decisions were
not on the Pareto frontier, a rival arbitrator could lure away the former’s customers by
offering decisions that both parties prefer. There is then, a strong incentive for arbitrators
to achieve Pareto efficiency with respect to the disputants. However, the parties to a

26

This problem is not solved by class action suits where the plaintiff represents a whole class of people

whose legal rights will be extinguished by resolution of the dispute.

28

 

 

 

dispute who hire a private adjudicator do not internalize all the benefits of changing the
law. Better rules will benefit future cases to which current disputants are not a party.
Thus the incentives of arbitrators for creating new laws may be deficient.

Besides making common law, judges interpret statutes. Interpreting statutes involves

supplying operational definitions for statutory language and applying these definitions to
decide cases.

Economists tend to conceive of legislation as the product of bargaining

among the representatives of various interests. This view suggests that statutes should
be interpreted according to the understandings and purposes of the underlying bargain,
much like the interpretation of business contracts (Easterbrook). The purpose of legislative
bargains, like business bargains, is to maximize the surplus from exchange. If this view
is persuasive, then efficiency considerations should enter directly into the interpretation of
statutes.

Unlike the collective choice literature, which is replete with impossibility theorems,

the efficiency theory sounds an optimistic note: Courts are efficient. This thesis, when
combined with the impossibility theorems, implies that courts are better than elected
officials at shaping efficient laws. This proposition, if true, has important implications for
judicial review: Instead of deferring to elected officials, courts should vigorously review
legislation and regulations (William Riker and Barry Weingast).

4

Conclusion

The economic models of legal disputes and their resolution by courts described in this article
represent a substantial improvement along some dimensions over traditional legal scholar-
ship. Explaining the process of dispute resolution as an equilibrium in the interaction of
self-interested decision makers draws upon a well-developed behavioral model that permits
a comparison of the efficiency of alternative legal rules. Indeed, the greatest strength of
this literature is its careful working out of the inexorable logic of self-interest. The models
consequently provide a point of reference that legal theory needs for an understanding of
courts and for deliberation over proposed changes in rules.

There are, however, significant obstacles and resistances that leave scope for devel-

opment and, possibly, breakthroughs in the future. First, the literature suffers from un-
satisfactory state of bargaining theory of legal disputes and courts. Indeed, the insights
needed to improve strategic bargaining theory may be inspired partly by law’s institutional
detail. Second, as long as disinterested decision making remains a mystery to economics,
the motives of judges cannot be endogenous in economic models. A better model of ju-
dicial decision making will force economists into the mainstream of jurisprudential debate
about the motivations of judges. Third, the law and economics literature have yet to pur-
sue adequately the modern economic theory of organizations, and to apply that theory to
the operation of courts and other institutions whose governance rules are primarily legal.
Finally, and most important, empirical research has lagged woefully behind theoretical ad-

29

 

 

 

vances. Improved data collection and additional econometric studies are needed to improve
the empirical grounding of the economic analysis of law.

We would be remiss if we failed to mention the gain to economics from the interaction

with law.

The courts, like the stock market, respond quickly to shocks in ways that

economists cannot seem to anticipate. In fact, the legal institutions that have evolved
to deal with the externalities created by injurers are more varied and subtle than the
traditional taxing institutions that are the focal point of many economists. For example,
a decade of effort by economists to develop theories of tort law succeeded on its own
scholarly terms, but economists all too often provided efficiency proofs for institutions that
most lawyers now view as inefficient. The proposals for reform that the ”tort crisis” has put
on the agenda of legislatures and courts raise issues of institutional design that economists
have just begun to consider. Economists can learn from lawyers how to make our policy
science more deft, flexible, and responsive to a living institution.

30

 


Leave a Reply

Your email address will not be published. Required fields are marked *