unambiguously amplified by the endogenous spending decisions of the plaintiff and defen- dant under certain circumstances. For example, when the facts favor the plaintiff p . .5,
and g . p1 2 p, so that there is a strong jury bias in favor of the plaintiff, we have p . p
B
. p. Conversely, we have pˆ , p
B
, p when p , .5 and g , p1 2 p i.e., the facts favor the defendant, and there is a strong jury bias in the defendant’s favor. In both
cases, the endogenous spending decisions of the plaintiff and defendant reinforce the initial jury bias by pushing pˆ further away from p. On the other hand, when p . .5 and g , 1 or
p , .5 and g . 1, the initial jury bias is moderated by the endogenous spending decisions of the two parties.
These results show that there is no necessary relationship between the initial level of jury bias and the final bias reflected in the estimator pˆ. In fact, from Eq. 12 and depending on
the value of g, pˆ can take on any value between 0 and 1 regardless of the true value of p. As in Section 2, jury bias will affect the participation constraints of both players. It is
straightforward to show that if g . 1, the quality of cases that the plaintiff is willing to file will fall, while the quality of cases the defendant is willing to defend rises. Similarly, if g ,
1, the quality of cases the plaintiff is willing to bring rises, while the quality of the cases the defendant is willing to defend will fall.
4. The English rule
Under the so-called English rule, lawyers’ fees are shifted to the party who loses at trial.
10
There are many variants of fee shifting that have either been proposed, or are actually in use. We will present a stylized model in which both parties are subject to fee shifting, where all
costs other than fixed costs are subject to shifting. In order to conduct a sensible analysis of the English rule, we will need to reinterpret p and pˆ, which we do in the subsection that
follows. This is necessary, because under the comparative negligence interpretation, unless pˆ 5 0, there is always a finding for the plaintiff.
10
There are some subtle issues involved when considering fee shifting in the context of this model. In particular, there is the issue of whether “discarded” evidence will be discovered in the process determining fees
to be shifted. See Froeb and Kobayashi 1996 note 36. We do not attempt to address this issue here and assume all costs are shifted to the losing party.
Table 2 g 5
1 g .
1 g ,
1 p . .5
pˆ 5 p . p
B
5 .5 pˆ . p
B
, pˆ . p and p
B
, pˆ , p pˆ . p
B
. p when g .
p1 2 p p , .5
pˆ 5 p , p
B
5 .5 p
B
. pˆ . p pˆ , p
B
, pˆ , p and pˆ , p
B
, p when g ,
p1 2p 324
A. Farmer, P. Pecorino International Review of Law and Economics 20 2000 315–328
4.1. A reinterpretation of p and pˆ The model needs to be reinterpreted if we are to consider certain legal rules other than
comparative negligence. Consider a negligence rule under which the actions of the defendant in meeting the due care standard are viewed with some error. Suppose that the due care
standard is represented by a level of care X to be taken by the defendant. If the jury finds that the defendant failed to meet this standard, then it finds for the plaintiff and awards the full
amount S. Define the defendant’s true level of care to be X. The jury will form an estimate of X which we denote X
ˆ . This estimator reflects an error term, jury bias, and the legal arguments made by both parties to the dispute. If X
ˆ 2 X . 0, the jury finds for the defendant and if X
ˆ 2 X , 0, it finds for the plaintiff. For the purpose of our discussion we just consider the role of the error term.
Specifically, we assume the jury to be unbiased, and we ignore the adversarial presen- tation of evidence by the plaintiff and defendant. Imagine instead that the jury is
presented with an unbiased sample of the evidence regarding X, where some combina- tion of sampling error and cognitive limitation on the part of the jury introduce an error
term into the process. We present this as a base case which will allow us to make a translation between the defendant’s actions under the negligence rule and p, which we
have taken to represent the objective merits of the case. Assume that the error e is symmetric about zero so that the jury observes X 1 e. The jury’s best point estimate of
X is X
ˆ 5 X 1 «. When X 5 X, i.e., the defendant just meets the due care standard then in half of all cases we will have X
ˆ 2 X , 0; the defendant is found liable by 50 of all juries. We associate this case with p 5 .5. Defendant actions X . X are associated with
p , .5, while X, X is associated with p . .5. Thus, there is a mapping between the defendant’s actions under the negligence standard and the parameter p used in this paper.
Note that all cases with p . .5 are cases which the plaintiff “should” win because they are associated with cases where the defendant failed to meet the negligence standard. If
a finding is for the plaintiff, then he receives the full award S, where pˆ reflects the equilibrium probability of him winning such an award.
Under the negligence standard it is not necessarily true that any deviation of pˆ from p should be considered undesirable. For example, all cases where p , 12 are ones in
which the defendant has met the negligence standard. If an unbiased jury could view X, the true level of care, it would find for the defendant in 100 of these cases; these are
all cases the defendant should win based on the objective merits. If endogenous legal expenditure moves pˆ towards 1, when p . 12 and towards 0 when p , 12, then it
improves the chances that the party favored by the objective merits wins at trial. To the extent it is desirable that the party favored by the objective merits actually wins the case
at trial, this effect should be considered a good thing. To preview the results below, we find in the absence of jury bias that the English rule behaves in this fashion in the
following sense. If p . .5, we have pˆ
E
. pˆ
A
; the jury’s estimate of p under the English rule exceeds its estimate under the American rule. Similarly, when p , .5, we have
pˆ
E
, pˆ
A
. However, in the presence of a strong jury bias, it is possible that a party who is disfavored by the objective facts but favored by the jury will spend more than their
325 A. Farmer, P. Pecorino International Review of Law and Economics 20 2000 315–328
opponent at trial. In this case, fee shifting will reinforce the initial jury bias, and this effect is clearly undesirable.
4.2. The model with fee shifting When fee shifting is added to the model from section 3, the plaintiff’s problem becomes
Max
H
pˆS 2 ~1 2 pˆ
S
cH p
1 cT
1 2 p
D
2 k
p
, while the defendant’s problem is to
Max
T
2 pˆ
S
S 1 cT
1 2 p 1
cH p
D
2 k
D
. Note in each case we assume that the fixed costs of litigation k
p
and k
D
are not subject to shifting. This is appropriate if they reflect psychic costs associated with proceeding to trial.
Substitute for pˆ from Eq. 7, take the first order conditions and solve to yield
H 5 g
rpS c~1 2 r
S
g 1
S
1 2 p g
p
D
r12r
D
, 13a
T 5 rS~1 2 p
S
1 2 p g
p
D
r12r
c~1 2 r
S
g 1
S
1 2 p g
p
D
r12r
D
. 13b
Substituting into Eq. 8, the solution for the jury’s estimate of p becomes pˆ 5
g~g p
r12r
g~g p
r12r
1 ~ 1 2 p
r12r
. 14
Compare Eq. 14 to Eq. 12 to see that in the absence of jury bias g 5 1, endogenous expenditure under the English rule pulls pˆ closer to 1 if p . 12 and closer to 0 if p ,
12. Specifically, for p . .5, we have pˆ
E
. pˆ
A
. .5 and for p , .5, we have pˆ
E
, pˆ
A
, .5. As noted above, this effect may be considered a good thing, since it implies that there
is a higher probability that the plaintiff wins when the defendant fails to meet the negligence standard, and a higher probability the defendant wins when she does meet the
negligence standard.
When there is jury bias under the English rule, an analysis of Eq. 14 reveals that H . T iff gp . 1 2 p. In other words, the “favored” party has the incentive to produce
more arguments, but being “favored” may reflect jury bias g Þ 1 rather than just the true merits of the case. Recall that in the absence of fee shifting, H . T iff p . 1 2 p
so that the party objectively favored by the facts always produces more arguments to support their position even if the jury is biased against them. By contrast, with fee
326 A. Farmer, P. Pecorino International Review of Law and Economics 20 2000 315–328
shifting, it is possible that the party disfavored by the facts of the case will produce more arguments if the jury bias is sufficiently large. Specifically, it is possible that gp . 1 2
p while p , 1 2 p or gp , 1 2 p while p . 1 2 p. In either of these cases, the jury bias causes the party with the objectively weaker case to produce more arguments. As a
result, the endogenous spending decisions of the plaintiff and defendant will accentuate the initial jury bias. For example, if the objective merits favor the defendant p , .5, but
g
p . 1 2 p, then we have pˆ . p
B
. .5 . p.
11
Under the American rule see Table 1, we see that whenever jury bias is in favor of the party disfavored by the objective merits
of the case p . .5 and g , 1 or p , .5 and g . 1, jury bias is muted by the endogenous spending decisions of the two parties. Thus, if strong jury bias is believed to be a
problem, fee shifting would appear to be a particularly unattractive remedy.
12
5. Conclusion