ESSAY

THE PAROL EVIDENCE RULE, THE PLAIN MEANING RULE,

AND THE PRINCIPLES OF CONTRACTUAL INTERPRETATION

ERIC A. POSNERt

INTRODUCTION

Most commentators believe that when a contract's terms specify

the obligations of the parties in the case of some contingency, and
that contingency occurs, a court should, barring procedural irregu-
larities, enforce the terms. Much scholarship has focused on what
courts should do when the terms are silent with respect to a contin-
gency. When such a "gap" in the contract exists, this scholarship ar-
gues that the court should enforce a "gap-filling rule" or "default
rule," which allocates obligations in such a way that either maximizes
the contract's ex ante value or provides appropriate ex ante incentives
to disclose information. Little work, however, deals with how the
court should decide whether a gap exists in the first place.
Analysis of this issue requires consideration of the ways that courts
can weigh different kinds of evidence, such as the final writing (if
any), earlier oral or written negotiations, trade custom, and course of
dealing after the contract is signed. In theory, courts could weigh
evidence in whichever way seems to produce the most plausible re-
sult, or general rules could direct courts to put more weight on some
kinds of evidence than on others. In fact, such general rules do exist.
This Essay focuses on these rules of contractual interpretation, and
the parol evidence rule in particular.
The parol evidence rule deals with a common contractual situa-
tion: where initial negotiations, in which preliminary oral or written
promises are exchanged, conclude with a writing that appears to em-

t Assistant Professor of Law, University of Pennsylvania. My thanks to Lisa

Bernstein, Rich Hynes, Bruce Johnson, Jason Johnston, Randy Picker, and an audi-
ence at the George Mason Law School, and particularly to Richard Craswell, whose
comments were extensive and very helpful. Also thanks to Andrew Gallo and Brett
Goldblatt for valuable research assistance.

(533)
534 UNIVERSITY OFPENSYLVANIA LAW REVIEW [Vol. 146: 533

body the entire agreement. The question is whether the court's in-
terpretation of the contract should rely at all on evidence related to
the earlier negotiations, known as "extrinsic evidence," or should rely
entirely on the writing.
Two stylized, polar positions can be distinguished. Under what I
will call the "hard-PER," the court generally excludes extrinsic evi-
dence and relies entirely on the writing. Under the "soft-PER," the
court gives weight both to the writing and to the extrinsic evidence.
There is a division among courts between those whose approaches re-
semble hard-PER and those whose approaches resemble soft-PER.
Part I of this Essay discusses these approaches. Part II provides an
analysis of the stylized hard- and soft-PERs, and discusses the ways in
which these approaches may be implemented as legal rules both
separately and in combination. Part II also discusses the plain mean-
ing rule and other rules of interpretation as they relate to recurrent
transactions, such as form contracts, consumer transactions, collective
bargaining agreements, and long-term contracts. Part III examines
Corbin's influential argument about the parol evidence rule.
The Essay lays out systematically the arguments in favor of the two
different conceptions of the parol evidence rule. These arguments
stress the costs to the parties of using a writing to inform courts of
their contract, and the danger ofjudicial error. The hard- and soft-
PERs respond to these problems in complex ways, and with varying
degrees of success.

I. THE PAROL EVIDENCE RULE

Most courts would subscribe to something close to the following
statement of the parol evidence rule: A court will refuse to use evi-
dence of the parties' prior negotiations in order to interpret a written
contract unless the writing is (1) incomplete, (2) ambiguous, or (3)
the product of fraud, mistake, or a similar bargaining defect.
The rule is susceptible to hard and soft interpretations in several
ways, each of which turns on the use of extrinsic evidence to deter-
mine whether any of the exceptions apply. First, consider the com-

'Purists will object that I conflate the plain meaning rule, which I treat as excep-
tion two, and the parol evidence rule. As far as I can tell, nothing turns on this distinc-
tion, and my version avoids needless complexities. Because both the parol evidence
rule and the plain meaning rule concern the same issue-under what circumstances
extrinsic evidence can be used to supplement a writing-they are best analyzed to-
gether. Courts and commentators often treat the two together, but sometimes do not.
1998] PAROL EVIDENCE RULE

pleteness exception. The harder courts declare a writing complete if

it looks complete "on its face." Writings generally look complete if
they are long and detailed, or at least contain unconditional lan-
guage, cover many contingencies, or at least the most important con-
tingencies, and contain a clause, such as a merger clause, which says
that the contract is complete.2 The softer courts declare a writing
complete only if the extrinsic evidence supports that determination.
In theory, this means that a court looks for extrinsic evidence that
suggests that the writing is not complete, such as an oral statement
that the writing is just a memorandum. If the court finds no such
evidence, no extrinsic evidence is admitted. In some cases, this pro-
cedure may result in the exclusion of relevant extrinsic evidence. In
practice, however, courts adopting this soft version of the complete-
ness exception generally admit all relevant extrinsic evidence, be-
cause any inconsistent extrinsic evidence suggests (however indi-
rectly) that the contract is incomplete, and any consistent extrinsic
evidence that might be excluded would not affect the interpretation
of the contract anyway.3
A similar point can be made about the ambiguity exception. The
harder courts consider a writing ambiguous if it looks ambiguous "on
its face." A contract is facially ambiguous when the writing has con-
flicting terms or no provision relating to the contingency under
which the dispute arises. In contrast, the softer courts decide whether
a writing is ambiguous by looking at extrinsic evidence. Even if the
writing has no apparently conflicting terms, extrinsic evidence may
indicate that the parties agreed on terms that conflict with the written
terms. This conflict creates an ambiguity that requires admission of
the extrinsic evidence. In contrast, if the extrinsic evidence does not
conflict with the writing, it is not admissible, but its admission would
make no difference, anyway. Again, the soft version of the ambiguity
exception for the most part allows courts to consider any relevant ex-
trinsic evidence.4

2 SeeE. ALLAN FARNSWORTH, CONTRACTS § 7.3, at 474 (2d ed. 1990) ("Many courts,
particularly in cases decided in the first half of this century, have agreed that .... [i]f,
on its face, the agreement appears to be completely integrated, the court should sim-
ply accept that this is so.").
' This is an old point. See 4 SAMUEL WILUSTON, WILLISTON ON CONTRACTS § 633,
at 1014 (3d ed. 1961); see a/soJohn E. Murray, Jr., The ParolEvidence Process and Stan-
dardized Agreements Under the Restatement (Second) of Contracts, 123 U. PA. L. REV. 1342,
1357 (1975).
4 Under another formulation, if "the language of a contract, in the light of all the
circumstances, is 'fairly susceptible of either one of the two interpretations contended
536 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146:533

The fraud exception presents a difficulty for the harder courts

because one cannot read fraud off a contract's face; one must look at
extrinsic evidence in response to an allegation of fraud. Once one
looks at extrinsic evidence for fraud, however, it is easy for a party to
gain admission of all extrinsic evidence. The party alleges that the
promisor fraudulently violated a pre-writing promise, which induced
him or her to sign the writing. Although in theory it is harder to
prove promissory fraud than to prove breach of contract,5 in practice
the distinctions between the two causes of action break down. The
fraud exception swallows the parol evidence rule.6 Harder courts re-
sist this outcome by distinguishing two sorts of promissory fraud,
"fraud in the inducement," which refers to the use of fraudulent
promises, and "fraud in the execution," which refers to the use of a
false writing such as a forgery, and admitting extrinsic evidence only
to show fraud in the execution. Hard courts also sometimes limit the
fraud exception by restricting its use to cases where the allegedly
fraudulent promise does not directly contradict a promise contained
in the writing.
An example will clarify the relationship between the different ver-
sions of the parol evidence rule. Suppose that Seller and Buyer sign a
written contract for sale of a used automobile, that is detailed and
looks complete, but does not contain a merger clause. Before the
writing is signed, Seller states in response to Buyer's query that the
carburetor is in fine shape. The writing contains many provisions
describing the parties' obligations, but does not contain a provision
that guarantees the quality of the carburetor. In fact, the carburetor
is in poor shape.
Under hard-PER, Buyer would have difficulty convincing a court
to take account of the oral representation for the following reasons:
First, the contract appears complete "on its face" because it is detailed

for .. .,' extrinsic evidence relevant to prove either of such meanings is admissible."
Pacific Gas & Elec. Co. v. G.W. Thomas Drayage & Rigging Co., 442 P.2d 641, 646 (Cal.
1968) (en banc) (omission in original) (citations omitted).
5 Fraud requires proof that the promisor intended to break the promise at
the
time that it was made; ordinary breach of contract requires no such showing. Fraud
claims allow the promisee to obtain reliance damages; contract claims allow expecta-
tion damages. See, e.g., Lipsit v. Leonard, 315 A.2d 25, 28-29 (N.J. 1974).
6 The problem is reproduced when courts allow promissory
estoppel claims to
circumvent the parol evidence rule. For a discussion, see Michael B. Metzger, The
ParolEvidence Rule: PromissoryEstoppel's Next Conquest?, 36 VAND. L. REV. 1383, 1454-66
(1983) (explaining the advantages and disadvantages of using promissory estoppel in
parol evidence cases).
19981 PAROL EVDENCE RULE

and refers to the most probable contingencies; second, the contract

appears unambiguous "on its face" because (we assume) it contains
no gaps or conflicting terms; third, the fraud claim could not plausi-
bly allege fraud in the execution, but only fraud in the inducement.
As no exceptions apply, the extrinsic evidence cannot be used to in-
terpret the contract.
Under soft-PER, the court would more likely admit evidence of
the oral representation, for any of three reasons. First, the oral repre-
sentation suggests that the contract's failure to stipulate or disclaim
the quality of the carburetor means that the parties failed to put a
crucial part of their contract in the writing. As the writing is incom-
plete, the oral representation can be used to interpret the contract.
Second, the oral representation conflicts with (let us suppose) a
rather vague written disclaimer in the contract. Was the disclaimer
intended to extend to the carburetor? The oral representation may
be considered in order to resolve this ambiguity. Third, the oral rep-
resentation may be considered as evidence of fraud in the induce-
ment. Under any of the three exceptions, the extrinsic evidence of
the oral representation
7
will be admitted for the purpose of interpret-
ing the contract.
As noted earlier, parties can, in principle, contract out of the pa-
rol evidence rule by including a "merger" or "integration" clause.
This clause typically states that courts should treat the writing as a
complete embodiment of the agreement. Soft-PER and hard-PER
courts approach the merger clause differently. Hard-PER directs
courts to defer to the clause except when the extrinsic evidence refers
to an entirely different contract-as courts sometimes say, a contract
that would not naturally be joined to the contract in dispute. Soft-
PER directs courts to admit extrinsic evidence for the purpose of de-
termining whether the merger clause was intended to cover the terms
at issue. In the example above, a soft court could evade a merger
clause by holding that the oral representation shows that the merger
clause was not intended to extend to the question of warranty, that
the merger clause is ambiguous (as to whether it covers only some

7 As an aside, note that a contract can be either fully integrated or partially inte-
grated. If it is fully integrated, courts may not use extrinsic evidence to vary any terms
within the scope of the written agreement. If it is partially integrated, courts may use
any extrinsic evidence that is consistent with the writing. See RESTATEiENT (SECOND)
OF CONTRACTS § 210 (1981). The concept of partial integration reflects the idea that
parties may intend their writing to cover some, but not all, terms of the contract. In
practice, courts have been unable to use it in a consistent way.
538 UNIVERS17Y OFPENNSYLVANIA LAWREVIEW [Vol. 146: 533

terms or all terms), or that the oral representation fraudulently in-

duced Buyer to enter the contract, including the integrationaspect of it.
The hard and soft versions of the parol evidence rule, and these
various qualifications, will be discussed in greater detail later in this
Essay. The reader should, for now, understand that the reality is
more complex than the stylized versions of the parol evidence rule
developed for the purpose of analysis. Although some jurisdictions
use something like the hard-PER,8 while other jurisdictions use some-
thing like the soft-PER, 9 many jurisdictions take different and often
conflicting approaches to the treatment of extrinsic evidence. In
some jurisdictions, for example, the courts adopt a hard attitude to-
ward the incompleteness exception, while taking a soft attitude to-
ward the ambiguity exception. ' ° In other jurisdictions, the courts do

8 Relatively clear examples are Missouri,

seeJake C. Byers, Inc. v. J.B.C. Invs., 834
S.W.2d 806, 814 (Mo. Ct. App. 1992) (stating that"[o]ur Supreme Court prohibits the
use of extrinsic evidence to interpret an otherwise unambiguous contract"), Ohio, see
Trinova Corp. v. Pilkington Bros., P.L.C., 638 N.E.2d 572, 575 (Ohio 1994) (stating
that where an "agreement is complete and unambiguous on its face, parol evidence is
inadmissible to show a contrary intent of the parties"), Pennsylvania, see infra note 14,
and Virginia, see Pulaski Nat'l Bank v. Harrell, 123 S.E.2d 382, 387 (Va. 1962) (stating
that where a written contract is "in clear and explicit terms the writing shall be the sole
memorial of that contract, and it is conclusively concluded that the writing contains
the whole contract").
9 Relatively clear examples of states in which courts say they follow soft-PER are
Alabama, see Hibbett Sporting Goods, Inc. v. Biernbaum, 375 So. 2d 431, 434 (Ala.
1979) ("Where there exists doubt that the written agreement was ever intended to
reflect the full agreement of the parties, the courts of this State have not hesitated to
admit contradictory parol evidence .... "), Arizona, see Darner Motor Sales, Inc. v.
Universal Underwriters Ins. Co., 682 P.2d 388, 398 (Ariz. 1984) (en banc) (stating that
"the interpretation of a negotiated agreement is not limited to the words set forth in
the document"), California, see infra note 15, Oregon, see Abercrombie v. Hayden
Corp., 883 P.2d 845, 853 (Or. 1994) ("The trial court may consider parol and other
extrinsic evidence to determine whether the terms of an agreement are ambiguous."),
Texas, see Johnson v. Cherry, 726 S.W.2d 4, 7 (Tex. 1987) (stating that"courts must
look beyond the face of the deed to ascertain the parties' intent"), and Washington, see
Emrich v. Connell, 716 P.2d 863, 866 (Wash. 1986) (en banc) ("In making
this ... determination of whether the parties intended the written document to be an
integration of their agreement.., the trial court must hear all relevant extrinsic evi-
dence, oral or written.").
'0 Two examples are Oklahoma, see Mercury Inv. Co. v. F.W. Woolworth Co., 706
P.2d 523, 529 (Okla. 1985) ("While parol testimony cannot vary, modify, or contradict
the terms of the instrument, it is admissible to explain the meaning of words when
there is a latent ambiguity in the written text of the agreement."), andVermont, com-
pare Big G Corp. v. Henry, 536 A.2d 559, 561 (Vt. 1987) (stating that a negotiable
instrument with an absolute promise to pay on its face cannot be varied by extrinsic
evidence), with Isbrandtsen v. North Branch Corp., 556 A.2d 81, 84 (Vt. 1988)
(holding that courts may "consider the circumstances surrounding the making of the
agreement" in order to determine whether an ambiguity exists).
1998] PAROL EVIDENCE RULE

the opposite, adopting a soft attitude toward the incompleteness ex-

ception, while adopting a hard attitude toward the ambiguity ex-
ception." In addition, within a single jurisdiction, the parol evidence
rule may vary considerably over time. In Pennsylvania, for example,
hard-PER prevailed from the 1920s to the 1950s,12 confusion that
gradually resolved itself into soft-PER prevailed from the late 1950s
through the 1970s,'3 and hard-PER has made a comeback in the late
1980s and 1990S.14 In California, hard-PER prevailed until the 1960s,

" Arguably, Connecticut is an example. Compare Heyman Assocs. No. 1 v. Insur-

ance Co., 653 A.2d 122, 133 (Conn. 1995) (forbidding admission of extrinsic evidence
when the meaning of "pollutant" was plain), with Shelton Yacht & Cabana Club, Inc. v.
Suto, 188 A.2d 493, 496 (Conn. 1963) (allowing consideration of extrinsic evidence to
determine whether parties intended for a writing to be complete). Goetz and Scott
claim that the majority of modem courts adopt this view. See Charles J. Goetz &
Robert E. Scott, The Limits of Expanded Choice: An Analysis of the InteractionsBetween
Express and Implied Contract Terms, 73 CAL. L. REV. 261, 309 & n.127 (1985) (citing
Mellon Bank, NA v. Aetna Bus. Credit, Inc., 619 F.2d 1001 (3d Cir. 1980), and four
other cases). My research does not confirm their claim.
" See, e.g., Gianni v. R. Russell & Co., 126 A. 791, 792 (Pa. 1924) ("[W]e propose to
stand for the integrity of written contracts."); Russell v. Sickles, 160 A. 610, 611 (Pa.
1932) ("[W]here parties have deliberately and formally committed their engagements
to writing, they cannot be permitted, by parol evidence, to vary or modify the written
contract.. .. "); Walker v. Saricks, 63 A.2d 9, 11 (Pa. 1949) ("'[U]nless fraud, accident,
or mistake be averred, the writing constitutes the agreement between the parties ....-
(citation omitted)); Bardwell v. Willis Co., 100 A.2d 102, 104 (Pa. 1953) ("[P]arol
evidence to vary, modify or supercede the written contract is inadmissible in evi-
dence.").
"3There were several important soft-PER cases. See Berger v. Pittsburgh Auto
Equip. Co., 127 A.2d 334, 337 (Pa. 1956) (admitting evidence of pre-contractual repre-
sentations that were not included in the writing); Highmont Music Corp. v. J.M.
Hoffmann Co., 155 A.2d 363, 366 (Pa. 1959) (noting that a provision in a lease that
tenant had inspected premises and accepted the property in its present condition did
not nullify the oral misrepresentation of the landlord with respect to latent defects);
Myers v. Rubin, 160 A.2d 559, 561 (Pa. 1959) ("[I]t is established law that misrepresen-
tation of a material fact, if relied upon by the deceived party, inducing him to enter
into the contract, confers upon that party the right to nullify the agreement ...
Greenwood v. Kadoich, 357 A.2d 604, 608 (Pa. Super. Ct. 1976) (admitting evidence of
appellee's pre-contractual misrepresentation); National Bldg. Leasing v. Byler, 381
A.2d 963, 965 (Pa. Super. Ct 1977) ("[E]vidence that one of the parties was induced
to enter the contract through fraud or misrepresentation is not barred by the parol
evidence rule."). A hard-PER case was Nicolella v. Palmer,248 A.2d 20, 22-23 (Pa. 1968)
(stating that, when a subject has been dealt with specifically in the written contract,
oral agreements cannot be introduced to vary terms of the written agreement in the
absence of fraud, accident, or mistake).
" See HCB Contractors v. Liberty Place Hotel Assocs., 652 A.2d 1278, 1279 (Pa.
1995) (holding that the parol evidence rule barred claim alleging false representation
where that issue was specifically addressed by the written instrument).
540 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146:533

and soft-PER thereafter.' 5 In contrast, Virginia has maintained hard-

PER fairly consistently during the last century, with only minor varia-
tions. 16 In virtually every jurisdiction, one finds irreconcilable7
cases,
in doctrine, confusion, and cries of despair.
frequent changes

II. ANALYSIS

A. Contract Drafting and Judicial Error

When parties strike a deal, they usually make some effort to for-
mulate it in a way that publicizes the bargain. Some formality, such as
a handshake or a writing, makes clear to the parties involved that
obligations have been exchanged. Part of the function of the formal-
ity is to signal to possible future adjudicators that the parties intend to
be legally bound. To the extent that this signaling function is distinct
from the value-producing exchange at the heart of the contract, for-
malities are costly. But the adjudicatory function served by formalities

" Compare Gardiner v. McDonogh, 81 P. 964 (Cal. 1905) (hard), and Heffner v.
Gross, 178 P. 860 (Cal. 1919) (hard), with Masterson v. Sine, 436 P.2d 561 (Cal. 1968)
(en banc) (soft), Pacific Gas & Elec. Co. v. G.W. Thomas Drayage & Rigging Co., 442
P.2d 641 (Cal. 1968) (en banc) (soft), and Delta Dynamics, Inc. v. Arioto, 446 P.2d 785
(Cal. 1968) (en banc) (soft). For a more recent soft-PER case, see Garcia v. Truck
Insurance Exchange 682 P.2d 1100 (Cal. 1984) (en banc). But see Susan J. Martin-
Davidson, Yes, Judge Kozinski, There Is a ParolEvidence Rule in California-theLessons of a
Pyrrhic Victory, 25 Sw. U. L. REV. 1, 9 (1995-1996) (arguing that California courts use
the "many standard but incompatible formulations of the parol evidence rule... as
needed").
16 See Pulaski Nat'l Bank v. Harrell, 123 S.E.2d 382, 387 (Va. 1962) (commenting

on Virginia's consistent adherence to hard-PER); Amos v. Coffey, 320 S.E.2d 335, 337
(Va. 1984) (stating that "[t]he parol evidence rule is a time-honored fixture in the law
of this Commonwealth").
17 To answer these two questions [raised by the parol evidence rule], we, in
Missouri, no different than the courts in most other jurisdictions, have used a
variety of principles, chosen randomly with no consistency, from the common
law, the treatises of Professor[s] Williston and Corbin, and the First and Sec-
ond Restatement of the Law of Contracts.... [T] he random selection of prin-
ciples.., has made the parol evidence rule in Missouri, no different than in
most otherjurisdictions, a deceptive maze rather than a workable rule.
Jake C. Byers, Inc. v.J.B.C. Invs., 834 S.W.2d 806, 811 (Mo. Ct. App. 1992); see also Betz
Lab., Inc. v. Hines, 647 F.2d 402, 405 (3d Cir. 1981) ("Pennsylvania decisions on the
parol evidence rule are neither clear nor consistent .... "); Arthur L. Corbin, The
Interpretation of Words and the ParolEvidence Rule, 50 CoRNELL L.Q. 161, 164 (1964-
1965) (discussing the confusion of the courts); Charles T. McCormick, The ParolEvi-
dence Rule as a ProceduralDevice for Control of the Jury, 41 YALE L.J. 365, 383 (1932)
(referring to courts' handling of the parol evidence rule as "haphazard," "oblique,"
and "unsatisfactory").
1998] PAROL EVMDENCE RULE

justifies their cost. The question addressed in this section is whether

and how legal rules can minimize this cost.
Suppose that Seller and Buyer conclude a contract after several
days of negotiations. 8 During the negotiations, Seller made many
statements, including promises and disclaimers, which we shall refer
to as the set S. The contract consists of a subset of those statements,
which we shall call C. The statements in C are all the statements in S
that the parties would want enforced if the enforcement mechanism
were perfect. C is smaller than S, because during the negotiations
Seller makes all sorts of proposals, some of which Buyer accepts, some
of which Buyer rejects, and some of which Seller withdraws before
Buyer has a chance to accept or before formal conclusion of the ne-
gotiations. C' consists of the statements in S that are not in C-the
statements that were made but not put in the contract, or what I will
call the "extra-contractual statements." The parties put into a writing
a subset Wof C; and we call the statements that are in C, but not in W,
W-namely the statements in the contract that are not put in the
writing, or the "unwritten statements."' 9 Wis smaller than C because
it is costly to write down every statement in the contract and because,
at some point, the marginal cost of writing down an additional state-
ment exceeds whatever marginal benefit results from putting it in
writing.20 Figure 1 illustrates the analysis. 21

'8Nothing in this example or any other turns on the identification of one party as
Buyer and the other as Seller.
'91 use the term "unwritten" for convenience, since the parol evidence rule bars
admission of even written extrinsic evidence. By "unwritten statements" I mean state-
ments, written or not, that are not included in the writing that represents the final
contract.
20 For now, we assume there is no merger clause in W, but see infraPart H.B.
2' Discussion of some complexities about the content of Wis deferred to Section
I. In particular, the reader should note that the analysis does not deal with implied-
in-fact terms of the contract-those terms the parties would want enforced, ex ante,
but do not stipulate to because of transaction costs.
542 UNIVERSITY OFPENNSYLVANIA LAW REVEW [Vol. 146:533

FIGURE 1

Now we can define hard-PER as a rule that tells the court to en-
force all statements in Wand no statement in W' or C' and soft-PER
as a rule that tells the court to enforce all statements in Wand W -
namely, C-but none in C. In words: Hard-PER tells the court to en-
force written contractual statements, and not to enforce unwritten
contractual or (written or unwritten) extra-contractual statements;
soft-PER tells the court to enforce written and unwritten contractual
statements, and not to enforce any extra-contractual statements.
If courts never made errors, soft-PER would clearly dominate
hard-PER, because only soft-PER allows courts to evaluate all relevant
evidence. The parol evidence rule makes sense only on the assump-
tion that courts make errors, an assumption we make. Specifically, a
court errs by enforcing an extra-contractual term (C'), which means
failing to enforce a written or unwritten contractual term (Wor W).22
Courts make characteristic errors under each version of the parol
evidence rule. Under soft-PER, courts err by enforcing extra-
contractual statements with some probability greater than zero. The

Because C' consists of all extra-contractual terms, enforcement of any of those

terms constitutes failure to enforce terms in C
1998] PAROL EYTDEWCE RULE

reason is that courts have trouble distinguishing unwritten statements

that are part of the contract from unwritten statements that are not
part of the contract. For the moment, we assume that courts under
soft-PER enforce unwritten and written promises with an equal
amount of error-proneness. Under hard-PER, courts "err" by refusing
to enforce unwritten contractual statements with a probability of one.
So the characteristic error under soft-PER is the court's erroneous
enforcement of statements in C: The characteristic error under
hard-PER is the court's refusal to enforce statements in W, which are
part of the contract, even if not part of the writing. The expected loss
under each rule is the product of its characteristic error and the con-
tractual gains that are not captured as the result of error.
One might object that because soft-PER allows courts to use more
information than hard-PER does, soft-PER should enable courts to
produce better decisions on average even if the decisions of courts
applying soft-PER will exhibit greater variance. This alternative charac-
terization would not affect the analysis for parties that are risk-averse,
since such parties are willing to pay for a reduction in variance. Even
risk-neutral parties would prefer the reduction in variance because
the uncertainty of the legal decision in case of a dispute would cause
parties to incur greater litigation costs than they would if the legal de-
cision could be accurately predicted.
Another relevant cost is the cost to the parties of writing down
promises, including the cost of remembering to put promises in writ-
ing and checking the writing to make sure that it does not contain
errors. This cost will be called the "transaction cost." Consider the
case where transaction costs are zero. Under hard-PER, the parties
have the choice between writing down a promise or not writing it
down. If they do not write it down, it will not be enforced, and they
will lose the gains from trade. If they do write it down, it will be en-
forced with a probability of one, supposing that the contingency it
governs occurs. (Many contractual terms, of course, concern low-
probability contingencies, so the benefit of including them is prob-
abilistic.) Because it costs the parties nothing to write down the
promise, the parties include all promises in the writing, and all prom-
ises are enforced with perfect accuracy. Under soft-PER, the parties
have the same choice, but now because they gain nothing from writ-
ing down the promises, they do not write down any promises. But
because under soft-PER the courts erroneously enforce some extra-
contractual promises, and therefore fail to enforce some of the un-
written contractual promises, the full gains of trade are not realized.
544 UNIVERSITY OFPENNSYLVANIA LAWREVEW [Vol. 146:533

Accordingly, when transaction costs are zero, hard-PER dominates

soft-PER.
When transaction costs exceed zero, the analysis becomes more
complex. One must distinguish the initial decision whether to use a
writing from the decision to incorporate additional promises into the
writing. If the parties do not use a writing, the parol evidence rule
does not apply to a dispute. Initially, we consider only cases where
parties use a writing. Once they decide to use a writing, they must
decide how many terms to place in the writing. It is reasonable to
assume that parties will first put the most important promise (a term
is just a subsidiary promise) in the writing, then the next most impor-
tant, and so on, where importance is measured by how much value
the promise contributes to the contract. Thus, the marginal value of
adding a promise to the writing declines.
At the same time, adding a promise to the writing is costly. The
reason is that the drafter must recall every promise that was made,
and then must write it down in a way that is satisfactory to both par-
ties. Because one can divide possible future states of the world into
indefinitely fine partitions, even the simplest contracts would gener-
ate huge transaction costs if the parties tried to put in the writing
obligations for all possible future states of the world.
Under hard-PER, the addition of a promise to the writing pro-
duces costs and gains. The cost is the transaction cost of the addition
to the writing. The gain is the value of enforcement of the promise
should the promisor breach. When the transaction cost of the mar-
ginal promise exceeds the marginal expected value of enforcement of
that promise, the parties do not incorporate that promise in the writ-
ing.
Under soft-PER, the addition of a promise to the writing produces
costs but does not produce gains. The addition of the promise to the
writing is costly because the drafter incurs costs in writing down the
promise. But the addition of the promise does not generate any
gains, because under the extreme version of soft-PER the court does
not put more weight on the written promise than on the oral prom-
ise. The court enforces either promise with the same degree of error-
proneness. Because the parties gain nothing by putting the marginal
promise in the writing, they do not use writings.
We can avoid this counterintuitive result by making two addi-
tional assumptions. First, the parties use writings for their own pur-
poses as well as for legal purposes, namely, so that they can remember
and refer to obligations after time has passed and memories have
1998] PAROL EVIDENCE RULE

dimmed. Second, even the softest courts put more weight on written
promises than on oral promises. Accordingly, it is reasonable to as-
sume that the probability of failure to enforce an unwritten promise is
greater than the probability of failure to enforce a written promise.
Under these additional assumptions, one would expect parties in a
soft-PER regime to put some promises in a writing, although fewer
than similar parties in a hard-PER regime.
The advantage of hard-PER is that when the transaction costs are
less than the value of the promise, parties will write it down and be
assured of accurate enforcement. The disadvantage is that when
transaction costs are greater than the value of the promise, parties will
not write down the promise and the resulting failure to enforce de-
prives them of the promise's entire value. The advantage of soft-PER
is that when transaction costs are greater than the value of the prom-
ise, so that parties do not write down the promise, the parties do not
lose the full value because courts will enforce unwritten promises,
albeit with some chance of error. The disadvantage of soft-PER is that
when transaction costs are less than the value of the promise, parties
cannot obtain the value of certain judicial enforcement by writing
down the promises, because courts' willingness to enforce unwritten
promises creates a risk ofjudicial error.
An example will clarify the discussion so far. Suppose that Buyer
offers to pay Seller $X for a widget. In this contract, there are two
promises, Seller's promise to deliver the widget when Buyer transfers
$ and Seller's warranty that the widget performs certain functions.
Buyer values the widget at $20 and the warranty at $10. The parties
put the promise to deliver in the writing, and now must decide
whether to put the warranty in the writing. The decision turns on
whether putting the warranty in the writing reduces the chance of
judicial error should a dispute arise. Under hard-PER, incorporation
of the warranty into the writing increases the value of the contract to
Buyer by $10 ((1)$10 - (0)$10). Suppose that a soft-PER court would
correctly enforce a written promise with probability of 0.6 and would
correctly enforce an oral promise with probability of 0.2. Incorpora-
tion of the warranty into the writing under soft-PER increases the
value of the contract to Buyer by $4 ((0.6)$10 - (0.2)$10).
Now consider three levels of transaction costs. If transaction costs
are $1, the parties under hard-PER would add the warranty to the
writing, resulting in a net gain of $9. The parties under soft-PER
would add the warranty to the writing, resulting in a net gain of $3.
546 UNIVERSITY OFPENNSYLVANIA LAW REViEW [Vol. 146: 533

If transaction costs are $6, the parties under hard-PER would add
the warranty to the writing, resulting in a net gain of $4. The parties
under soft-PER would not add the warranty to the writing, and so
would gain only $2 from the oral warranty.
If transaction costs are $20, the parties under hard-PER would not
add the warranty to the writing, resulting in a gain of $0. The parties
under soft-PER also would not add the warranty to the writing, but
would still gain $2.
This example shows that the optimal treatment of extrinsic evi-
dence of a promise depends on (1) the amount it would have cost the
parties to add that promise to the writing (call this amount "B"); and
(2) the value that the promise contributes to the contract discounted
by the increased probability of correct judicial enforcement should a
dispute arise (call this amount "pL"). When B is high and pL is low,
the advantage belongs to soft-PER because it permits enforcement of
the promise, even if not with perfect accuracy. When B is low and pL
is high, the advantage belongs to hard-PER because it allows the par-
ties to ensure accurate enforcement by allocating promises to the
writing. In short, if B > pL, the court should admit the extrinsic evi-
dence and enforce the unwritten promise; otherwise, it should ex-
clude the evidence and decline to enforce the unwritten promise.
Having stated the general thrust of the argument, several qualifi-
cations are in order. First, it is unrealistic to assume that a court ac-
curately estimates the parties' transaction costs and the value of the
promise in question but still enforces oral promises with a high de-
gree of error. The assumption will be dropped in a later section,
where it will be shown that-in the absence of this assumption-the
court's decision should be based on a general rule rather than on
direct evaluation of these variables.
Second, the assumption that courts accurately enforce written
promises with a probability of one under hard-PER is also false. But
the analysis survives as long as one assumes that courts under hard-
PER enforce written promises more accurately than courts under soft-
PER enforce written and oral promises, an assumption that is based
on the premise that hard courts do not try to enforce oral promises
that might be extra-contractual.2
Third, some readers might find troublesome the assumption that
the parties use a writing and that the choice of whether to write down
an additionalpromise is the subject of analysis. However, because the

This premise is examined in more detail in Part III.

1998] PAROL EVIDENCERULE

parol evidence rule applies only if the parties use a writing, this as-
sumption is justified. A later section of this Essay will address the
decision to use a writing in the first place by considering why the pa-
rol evidence rule does not apply to oral contracts.

B. The ParolEvidenceRule as a Default Rule

When contracts do not specify the use of extrinsic evidence, there

is a "gap" in the contract. Even when contracts do specify the use of
extrinsic evidence through a merger clause, there is a gap in the
sense that the scope of the merger clause is not self-interpreting.
Courts fill in gaps through the use of default rules.
A default rule can be more or less "tailored."2 4 A more tailored
rule (or "standard") directs a court to fill in the gap with the terms
that the actual parties would have chosen had transaction costs been
zero. A less tailored rule directs a court to fill in the gap with terms
that the average party would have chosen if transaction costs had
been zero. There has been much discussion about whether default
rules should be more or less tailored, but little progress has been
made at an abstract level.
For parol evidence, a highly tailored default rule makes little
sense. A highly tailored rule would direct a court to admit extrinsic
evidence only if the parties involved would have wanted the court to
do so. The best evidence of the parties' hypothetical desires is (1) the
transaction cost of adding the promise in question to the writing; and
(2) the expected value that the promise would have contributed to
the contract. Although it seems possible that courts could estimate
the transaction costs in a rough way, two barriers stand in the way of
accurate estimation of the promise's expected value.
First, there is an epistemic barrier. Suppose the parties accurately
believed that the court that would decide a dispute is highly error-
prone, and therefore they put most of their promises in the writing in
the hope that the court would exclude extrinsic evidence. The court
should then be asked to determine whether the parties had this be-
lief. But if the parties' belief was accurate, and the court is error-
prone, it is unlikely that this court could determine that the parties
would have wanted, and in fact did want, the court to exclude the
extrinsic evidence. As a result, the court might erroneously admit the

24 See Ian Ayres & Robert Gertner, Milling Gaps in Incomplete Contracts: An Economic

Theory of Default Rules, 99 YALE L.J. 87, 91-92 (1989) (discussing default rules for con-
tract interpretation).
548 UNIVERSITY OFPEWNSYLVANIA LAWREVIEW [Vol. 146: 533

evidence. In a phrase, it does not make sense to ask courts to deter-

mine whether parties expected them to be error-prone, because their
error-proneness will prevent them from discovering the answer.
Second, there is a practicalbarrier. Even if an error-prone court
correctly realizes that the parties would not have wanted it to admit
extrinsic evidence, an untailored rule requires a court to exercise a
self-discipline that is likely to be beyond it. Although juries decide
many questions of fact in contract cases, judges generally have the
power and obligation to interpret the contract. In interpreting the
contract, judges rely on all admitted evidence. To determine whether
to admit evidence under the parol evidence rule, judges first must
evaluate the evidence. They cannot estimate the transaction costs the
parties would have incurred to add a promise to a writing, and the
expected added value of that promise, without evaluating the extrin-
sic evidence of that promise. But having done so and having decided
to exclude the extrinsic evidence, it may be difficult for the judge to
purge the extrinsic evidence's influence on his evaluation of the dis-
pute. In contrast to procedures involving the exclusionary rules of
criminal law-where the jury, as decisionmaker, does not know about
evidence that the judge has excluded-a judge in a contract case
must confront and evaluate evidence before excluding it.
In sum, if a court is, in fact, incompetent, then it will not be able
to realize, both epistemically and practically, that the parties recog-
nize its incompetence and would not want it to evaluate the evidence
because of its incompetence. If a court is so competent that it can
accurately evaluate the extrinsic evidence for the purpose of determin-
ing its admissibility, it might as well use it for the purpose of interpret-
ing the contract.
These problems are avoided if the parol evidence default rule is
untailored. By this I mean that a single parol evidence rule is applied
in all contract cases. The untailored rule avoids the paradox of courts
admitting or excluding evidence according to whether they think
they are competent to evaluate it. Instead, courts automatically con-
sider all extrinsic evidence if the rule is soft-PER; or 5they automati-
cally exclude extrinsic evidence if the rule is hard-PER2

2 It is true that a court would have to consider extrinsic evidence before excluding
it. But under untailored hard-PER, a court would refuse to consider any evidence
about a promise after eliciting an admission from a party that the promise is not writ-
ten. Thus, even if a court learns from the pleadings that the party claims that an ex-
trinsic promise occurred, it would not have a chance to evaluate the plausibility of that
claim.
1998] PAROL EVWDENCE RULE

But, supposing one is committed to an untailored rule, should

soft-PER or hard-PER be that rule? We saw before that under an un-
tailored rule (or standard) a court should admit extrinsic evidence of
an unwritten promise when the transaction cost of adding that prom-
ise to a writing exceeds the expected value of that promise. Because
this standard is applied at the margin, it cannot be translated easily
into a rule that applies to all promises. To see why, imagine that in a
jurisdiction there are n identical transactions, in which the first prom-
ise generates an expected gain of $20, the second promise generates
an expected gain of $10, and the third promise generates an ex-
pected gain of $5, and in each case it costs $8 to add the promise to a
writing.
Suppose further that courts under hard-PER enforce written
promises with a probability of one, and unwritten promises with a
probability of zero; courts under soft-PER enforce written promises
with a probability of 0.7 and unwritten promises with a probability of
0.2. Under hard-PER, the parties would add the first and second
promises to the writing, but not the third. The gain is $14 (20 + 10 - 8
- 8). Under soft-PER, the parties would add the first promise to the
contract (because (0.7)20 - (0.2)20 = 10 > 8), but not the second
(because (0.7) 10 - (0.2) 10 = 5 < 8) or the third. The gain under soft-
PER is thus $9 ((0.7)20 + (0.2)10 + (0.2)5 - 8). To determine
whether hard-PER is globally superior to soft-PER, one would have to
look at all the promises made in the average contract, take for each
promise the greater of (a) the probability of enforcement if written,
multiplied by the value of that promise, minus the transaction cost of
adding it to a writing; or (b) the probability of enforcement if unwrit-
ten, multiplied by the value of that promise-and then sum up the
relevant amount for each promise. If the sum is higher under hard-
PER than under soft-PER, hard-PER is globally superior.
Because of difficulties in measurement, this formula is useless for
determining whether a particular jurisdiction should choose soft-PER
or hard-PER as its untailored parol evidence rule. However, the for-
mula highlights important factors. If our continuing assumption is
correct, that hard-PER results in high-probability enforcement of
written promises and low-probability enforcement of unwritten prom-
ises, while soft-PER results in medium-probability enforcement of
both promises, then (a) declining transaction costs favor hard-PER,
and (b) declining judicial competence favors hard-PER. Actually, the
latter conclusion depends on an additional assumption that less com-
petent courts enforce written promises with a high probability when
550 UNIVERSITY OFPENNSYLVANIA LAWREVEW [Vol. 146:533

directed by hard-PER to enforce only written promises, while more

competent courts enforce unwritten promises with a higher probabil-
ity than do less competent courts when permitted by soft-PER.
The importance of factors (a) and (b) becomes clear as we move
to consideration of a final kind of rule. This kind of rule is neither a
highly tailored nor an untailored rule, but a semi-tailored rule. Such
a rule would divide contract cases into different classes, and courts
would apply soft-PER to cases in some of the classes and hard-PER to
cases in other classes. The classes could be based on the kind of
transaction, such as sales of goods, real estate transactions, and bills
and notes; or on the kind of parties, such as sophisticated parties,
consumers, and lawyer-assisted parties. For a particular class, the
parol evidence rule should be hard-PER when the transaction costs of
writing down the typical promise in that class are low and judicial
competence is low. This follows from the analysis of untailored rules,
except we apply the analysis to a class of cases rather than to all the
cases in ajurisdiction.
The semi-tailored rule comes at a cost, however. Before applying
hard- or soft-PER, the court must classify the case before it, and may
err in classification. The court might, for example, classify a con-
sumer case as a business case and apply hard-PER when soft-PER is
appropriate. This cost must be balanced against the cost of using a
less tailored rule, such as global hard-PER, under which courts will
not miscategorize cases but will apply hard-PER to cases for which
soft-PER would be more appropriate; and against the cost of using a
more tailored rule, under which courts do not miscategorize cases but
will often err in their attempts to evaluate whether parties would have
wanted them to exclude or admit extrinsic evidence. Because courts
err in classifying cases under the semi-tailored rule, parties will err in
predicting how courts will classify their case and thus will lose the
benefits that otherwise would result from the enforcement of that
parol evidence rule which is optimal for a given class of cases.26
Measuring these tradeoffs is beyond the competence of the observer.
As discussed shortly, however, it is clear that the parol evidence re-
gime in American law is one of semi-tailored rules.
A Note on Penalty Defaults. We have assumed that the desirable de-
fault rule is majoritarian, and that the only question is the extent of

For a general discussion of the effect of complex rules on individual decision-

making, see Louis Kaplow, A Model of the Optimal Complexity of Legal Rules, 11J.L. ECON.
& ORG. 150 (1995).
1998] PAROL EVIDENCE RULE

tailoring. Some commentators have argued influentially in recent

years that default rules sometimes 27 should stipulate terms that the
majority of parties would not want.
One version of this argument is based on the fear that parties ex-
ternalize their costs onto a subsidized court system. Suppose parties
decline to put terms in a writing even though the cost of doing so is
less than the court's cost of interpreting the contract ex post. Al-
though most parties prefer soft-PER because it saves them transaction
costs, courts should choose hard-PER This argument is discussed at
greater length subsequently.
Another argument in favor of penalty defaults is that they im-
prove parties' incentives to reveal value-maximizing information to
each other prior to entering a contract. This argument, which lies
behind the support among many commentators for the Hadley rule, is
that penalty defaults, by forcing buyers with private information to
bargain for favorable terms, eliminate inefficient cross-subsidization
that would otherwise result from the seller's inability to distinguish
types of buyers. This kind of argument, already problematic in the
Hadley case for reasons beyond the scope of this Essay, 28 does not have
a plausible application to the treatment of extrinsic evidence.
A Note on Merger Clauses. Contracts often contain merger clauses
that direct courts not to use extrinsic evidence. Interestingly, con-
tracts never, as far as I know, contain clauses that direct courts to rely
on extrinsic evidence. The lack of such "anti-merger" clauses might
be a clue that most parties would reject soft-PER-a point to which I
will return later.
The problem is that it is conceivable, even likely, that the parties
intend a merger clause to govern some, but not all, elements of their
contractual relationship. In the classic case, the parties enter an inte-
grated, written contract that is unrelated to a prior oral contract.
Most agree that the merger clause in the later contract does not pre-
clude enforcement of the prior contract, because parties would not
discharge a contract in such an indirect way. But suppose that the
prior and subsequent contracts have some elements in common.

See Ayres & Gertner, supra note 24 (popularizing the term "penalty default,"
though the basic idea goes back many years).
28See Eric A. Posner, Contract Remedies: Foreseeability, Precaution, Causation, and
Mitigation, in ENCYCLOPEDIA OF LAW AND ECONOMICS (Boudewijn Bouckaert & Gerrit
De Geest eds., forthcoming 1998) (examining the weaknesses in the argument that
the Hadley rule gives incentives to promisees to reveal information which would give
promisors incentive to take care).
552 UN!VERSITY OFPENNSYLVANIA LAWREVJEW [Vol. 146: 533

Should the merger clause be read broadly, to govern the earlier con-
tract, or narrowly, to govern only the later contract?2 As the contracts
become more and more closely related, the question becomes harder
to resolve.
Courts should approach this problem like any other contractual
gap. Our previous analysis suggests, for example, that when transac-
tion costs are low and judicial error is likely, courts should interpret
merger clauses broadly. The reason is that the parties can at low cost
move terms from W' to W, and their use of a merger clause suggests
that their failure to do so was not inadvertent. If transaction costs are
high and judicial error unlikely, courts should interpret the merger
clause narrowly so as to include extrinsic evidence of a prior oral con-
tract. 0 As a doctrinal matter, courts that adopt hard-PER hold that a
merger clause creates a strong, sometimes conclusive, presumption
that the parties intend courts not to rely on extrinsic evidence.3 '
Courts in soft-PER jurisdictions generally hold that a merger clause
• ° • presumption
creates a rebuttable 32 that the parties intend courts not to
rely on extrinsic evidence.
Another rationale for discounting merger clauses is procedural ir-
regularity. Sometimes merger clauses are buried in fine print. If it
appears that the consumer did not consent to the merger clause, a
court might reasonably use the unconscionability doctrine to avoid
the merger clause and use the default analysis discussed above to fill
in the resulting gap.3

Cf Mitchill v. Lath, 160 N.E. 646, 647 (N.Y. 1928) (holding that evidence of a
prior agreement was precluded because the promise was of the sort which naturally
would have been included in the final contract).
'0 This gives some substance to the often-invoked but vacuous principle that a
court should refuse to consider extrinsic evidence of an earlier agreement because of
a merger clause in a later written contract, if "the parties would ordinarily or naturally
be expected to embody the [earlier] agreement.., in the [later] written agreement."
Id. at 649 (Lehman,J, dissenting).
S" See FARNSWORTH, supra note 2, § 7.3, at 476 (noting that courts have traditionally
given effect to merger clauses as showing an intention that the contract be completely
integrated).
12 See, e.g., L.S. Heath & Son, Inc. v. AT&T Info.
Sys., Inc., 9 F.Sd 561, 569 (7th Cir.
1993) (explaining that a merger clause is strong, but not dispositive, evidence that the
parties intended the writing to be complete); see also FARNSVORTH, supra note 2, § 7.3,
at 476 & n.36 (listing cases).
2 See Richard Craswell, Property Rules and Liability Rules in Unconscionability and
Related Doctrines, 60 U. CHI. L. REv. 1, 57-60 (1993) (discussing the proper way for
courts to address the problem of contracts which are misleading because of unfavor-
able terms buried in the fine print).
1998] PAROL EVLDENCE RULE

C. DoctrinalImplications of the Analysis

No jurisdiction has a bright-line hard-PER or soft-PER. Courts
might state one or the other as a general rule, but all sorts of subsidi-
ary doctrines provide exceptions. The analysis sheds light on these
exceptions by relating them to a unified conception of contractual
interpretation.
By making some assumptions about the relationship among the
variables identified by the model and business practices, we can make
some useful, though rough, generalizations. The first assumption is
that transaction costs (B) are likely to be high when the parties are
unsophisticated or the contract is complex. Unsophisticated parties
face high transaction costs because they cannot draw upon experi-
ence in order to allocate terms among writings and because they may
not know about the law. Parties to complex contracts face high trans-
action costs because they must remember to put a large number of
terms into a writing. The second assumption is that the probability of
judicial error (p) is likely to be high when the contract is unconven-
tional, because courts do a better job of enforcing terms they have
seen before than terms they have not.3 These two assumptions gen-
erate the matrix displayed in Figure 2.

FIGURE 2

conventional unconventional
simple/sophisticated ? hard-PER
complex/unsophisticated soft-PER ?

Figure 2 shows that when contracts are conventional and com-

plex, soft-PER is optimal. The reason is that complexity or lack of
sophistication prevents the parties from moving terms into a writing,
while the conventionality of the contract suggests that judicial error
will be low. When contracts are unconventional and simple, hard-
PER is optimal. The reason is that simplicity or sophistication sug-

Again, these generalizations are crude, and there are objections to each. If a
contract is complex, courts might have difficulty interpreting it, and thus judicial error
is likely to be high. Judicial error is therefore a function of both factors, not just con-
ventionality. Moreover, if a contract is unconventional, the transaction costs of reduc-
ing it to a writing might be high because of the lack of models or forms. Thus,
transaction costs are a function of both factors, notjust complexity. These complica-
tions are dealtwith below.
554 UNIVERSITY OFPENNSYLVANIA LAWREV!EW [Vol. 146:533

gests that the parties can easily move terms into a writing in anticipa-
tion of the likelihood ofjudicial error caused by the unconventional-
ity of the deal. In the other two boxes, it is not clear which of the
factors dominates, so one cannot say with much confidence whether
hard-PER or soft-PER is likely to be superior. We can further compli-
cate the matrix by taking into account L (the value that the promise
contributes to the contract), which suggests that hard-PER is likely to
be superior for high-value contracts because the greater amounts at
stake justify higher transaction costs. Using these considerations, we
can make some general comments about recurring fact situations and
doctrines relating to contractual interpretation.

1. Ordinary Consumer Contracts

Because these contracts are "complex" from the perspective of
unsophisticated consumers, but routine and easy to understand for
courts, ordinary consumer contracts are good candidates for soft-PER.
That such contracts are low-value strengthens this conclusion, since it
is not worth the costs for consumers to read the contracts carefully or
to invest in legal advice and other sources of information in order to
learn whether they should read the contracts carefully.3
That soft-PER should apply when extrinsic evidence benefits con-
sumers, however, does not mean that soft-PER should apply when
extrinsic evidence benefits sellers. Because sellers are sophisticated
and because they obtain economies by formalizing their contracts,
sellers have lower transaction costs than buyers do. Hard-PER, there-
fore, is plausibly applied against sellers who seek to have extrinsic
evidence admitted for their benefit, although it may also be the case
that the expected loss from judicial error is low enough that sellers,
too, would not find it worthwhile to use a writing.
The larger point is that the court's approach to the buyer and
seller need not be symmetrical. The court could sensibly admit ex-
trinsic evidence that benefits buyers while excluding extrinsic evi-
dence that benefits sellers, on the ground that sellers can put relevant
promises in the writing more cheaply than can buyers. The analysis
resembles the view in tort law that contributory negligence promotes

"' Courts are particularly solicitous of consumers in insurance cases. See, e.g.,
Darner Motor Sales, Inc. v. Universal Underwriters Ins. Co., 682 P.2d 388, 401 (Ariz.
1984) (reversing lower court's summary judgment award for insurer because the na-
ture of the transaction suggested that insured acted reasonably in not reading a par-
ticular policy provision).
1998] PAROL EV1DENCE RULE

efficiency by giving the victim an incentive to take precautions when

the victim is the cheaper loss-avoider.

2. High-Value Consumer Contracts

The case for soft-PER becomes weaker when the consumer pur-
chases something of great value, such as a house. In such cases the
expected error costs rise significantly, because (1) the goods are
worth more than in ordinary consumer contracts, and (2) the con-
tracts are less routine, thus raising the likelihood of judicial error.
The second point is more likely to be true for some high-value sales,
such as real estate transactions, for which markets are thin, than for
thick.3 6
others, such as automobile transactions, for which markets are
Hard-PER may appropriately encourage the consumer to hire a law-
yer before making a large purchase. Alternatively, if consumers natu-
rally hire lawyers before making a large purchase, it encourages the
lawyer to put terms in the writing.

3. Unusual but Simple Contracts

The strongest case for hard-PER is a contract that is frequently

created by sophisticated parties, simple, unconventional, and of high
value. One possibility for this category is the complex financial in-
strument, such as the derivative. The terms of derivatives are so un-
usual that courts are likely to misunderstand them; yet, to
sophisticated parties, the contracts are fairly simple, in the sense that
there are not many contingencies to worry about and the contract can
be written in a way that accounts for them. In addition, these con-
tracts are typically of high value.

4. Notes and Instruments

A promissory note is a simple, highly formalized device for trans-

ferring value, but it is usually exchanged as part of an underlying
contract which may be more complex. The use of a note is one way to
contract around the parol evidence rule: it restricts the promisor's
ability to resist paying by arguing that the contract is invalid. The
parties have an incentive to use notes when they want to limit judicial
involvement and judicial error. Courts, however, often admit extrin-

Cf Lee v. Joseph E. Seagram & Sons, Inc., 552 F.2d 447, 451-52 (2d Cir. 1977)
("[I] ntegration is most easily inferred in the case of real estate contracts for the sale of
land, .. . or leases." (citations omitted)).
556 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146: 533

sic evidence for the limited purpose of showing that the promissory
note was intended to be valid only for the purpose of disbursement
from a particular fund37 This limited soft-PER exception to the hard-
PER rule is justified by the ease with which the existence of a fund
can be proved, compared to the difficulty of showing that perform-
ance of the contract was not completed in the proper manner.

5. Complex Business Contracts

Another candidate for hard-PER is the complex business deal,
such as the sale of a firm. Because of the large number of statements
made during preliminary negotiations, the number of statements that
are outside the contract (C') is likely to be high. As a result of the
idiosyncrasy of the terms, erroneous judicial enforcement of some of
those statements is likely. And because of the high value of the trans-
action, errors in enforcement are costly. Yet, the complexity of the
contract typically means that the parties face high transaction costs.
Therefore, the benefits of hard-PER may not be obtainable3

6. Formss9

Some courts are more willing to apply soft-PER to printed form

contracts that do not permit amendment than to other kinds of con-
tracts. This approach is consistent with our analysis. Because the
forms forbid amendment, the transaction costs of putting additional
terms in the writing are high. Because these form contracts are often
routine, the likelihood ofjudicial error is low. Accordingly, soft-PER
is appropriate.

37 See, e.g., Mozingo v. North Carolina Nat'l Bank, 229 S.E.2d

57, 61-62 (N.C. Ct.
App. 1976) (admitting extrinsic evidence to show the agreed upon method of pay-
ment on a note).
""Inmore complex situations, in which customary business practice may be more
varied, an oral agreement can be treated as separate and independent of the written
agreement even though the written contract contains a strong integration clause."
Lee, 552 F.2d at 451-52. The point is that, because they are simpler, real estate transac-
tions have lower transaction costs than complex business transactions. See also
Abercrombie v. Hayden Corp., 883 P.2d 845, 852-53 (Or. 1994) (holding that extrinsic
evidence of a promise to extend a closing date was improperly admitted to vary the
terms of a quitclaim deed). The court in Abercrombiepurported to find no ambiguity in
the deed, even in light of the extrinsic evidence, but one suspects that it feared that
the contrary holding would destroy the utility of the quitclaim deed for quieting title.
s9 This and the next two categories were suggested by Richard Craswell.
1998] PAROL EVIDENCE RULE

7. Terms That Are Central to the Contract

Courts are often reluctant to admit extrinsic evidence to vary writ-
ten terms that are central to the contract. If the term is central, then
the expected cost of judicial error is likely to be high. In addition,
the transaction cost of moving it is likely to be low, because the parties
are unlikely to overlook it. Thus, a harder version of the parol evi-
dence rule is justified.

8. Failure to Use a Writing

Does the analysis have implications for oral contracts? The parol
evidence rule could apply to the oral contract by barring admission of
evidence of promises that precede the final (oral) agreement. But
the parties' failure to use a writing suggests (1) that the parties do not
believe that the expected costs of judicial error are high, because
otherwise they would use a writing to constrain judicial interpretation,
and (2) that the transaction costs of using a writing are high relative
to the gains from more accurate enforcement. Most of the transac-
tion costs involved in writing contracts do not result from the physical
act of the writing but from the difficulty of formalizing a series of
negotiations without creating errors. These costs remain present
when parties seek to come to a final oral agreement that concludes
earlier negotiations. If the costs of using a writing are high relative to
the gains, the costs of formally repeating earlier promises in the final
oral agreement are also likely to be high relative to any gains. There-
fore, much of the value of the contract would be contained in earlier
promises, and hard-PER would prevent gains obtainable under soft-
PER.

9. Long-Term Contracts

Long-term or relational contracts are contracts whose terms are

underspecified because the parties cannot anticipate contingencies
with much accuracy and so prefer to leave the trading relationship
open to continuous renegotiation and revision. Because the terms of
the relationship are expected to develop over time, the parties cannot
at low cost commit the entire agreement to a writing. Transaction
costs are high, but at the same time the cost of judicial error is high
558 UNIVERSITY OFPENNSYLVANIA LAWREVEW [Vol. 146:533

because of the large number of statements in C' and their idiosyn-. 40

crasy. Thus, it is not clear whether hard-PER or soft-PER is superior.

10. Collective Bargaining Agreements

Collective bargaining agreements are complex, because they
cover every aspect of the employment relationship. However, both
employers and unions are usually sophisticated. In addition, judicial
error is likely to be common, because collective bargaining agree-
ments reflect the idiosyncrasies of the local workplace. The cost of
error is likely to be high also, because collective bargaining agree-
ments affect many people and last years or decades. Stressing these
last two factors, Judge Easterbrook argues that courts should adopt a
restrictive attitude toward admission of parol evidence for the pur-
pose of interpreting collective bargaining agreements. 4' Whether
these costs outweigh the transaction costs of putting the full contract
into the writing, however, is difficult to determine.

11. The Interpretation of Contracts Against the Drafter

This principle reflects an assumption that the drafter can more
cheaply ensure that the contract is reflected in the writing than the
other party can. When the drafter seeks admission of extrinsic evi-
dence that benefits it, this principle acts as hard-PER; when the
drafter seeks to resist admission of extrinsic evidence that benefits the
other party, the principle acts as soft-PER.

12. Course of Dealing, Course of Performance, and Trade Usage

Courts generally admit evidence of the parties' prior dealings, of
their dealings after the contract has been entered, and of trade us-
age. Commentators often argue that the distinction between these
kinds of evidence and the extrinsic evidence barred by the parol evi-
dence rule is arbitrary. But there is a reason for this distinction:
Course of dealing and course of performance leave a trail of receipts

'0 Cf. Goetz & Scott, supra note 11, at 317-20 (discussing the interpretation of
relational contracts).
41 See Bidlack v. Wheelabrator Corp., 993 F.2d 603, 618-19 (7th Cir. 1993)
(en
banc) (Easterbrook, J., dissenting) (arguing that parol evidence should rarely be
allowed in labor contract cases because of the desire for flexibility in these sweeping,
long-term contracts, the low reliability of parol evidence that may be decades old, and
the chilling effect that potential parol evidence surprises may have on negotiations).
42 This rule is codified in U.C.C. § 2-202(a) (1990).
1998] PAROL EWDENCE RULE

and other documents, and events to which disinterested witnesses can

testify. Trade usage can be determined through the publications of
trade associations and the testimony of uninvolved industry actors. In
contrast, promises made during negotiations are asserted by and fre-
quently known only by interested parties. Because admission of ex-
trinsic evidence regarding course of dealing, course of performance,
and trade usage is less likely than admission of other forms of extrin-
sic evidence to result in judicial error, a more permissive stance to-
ward the former isjustified.4

13. Ambiguity Versus Completeness

As noted earlier, courts sometimes admit extrinsic evidence for
the purpose of interpreting ambiguous contracts but not for the pur-
pose of interpreting incomplete contracts, and vice versa. The differ-
ent treatment of the different circumstances assumes that transaction
costs and error costs differ systematically when ambiguity arises and
when incompleteness arises. Courts never justify this assumption, and
the cases give the impression that courts veer to one extreme to avoid
the unpleasant consequences that would result from rigorous adher-
ence to the other.
A rationale can, however, be found in the analysis of Goetz and
Scott They argue that courts should take a soft approach toward
incompleteness and a hard approach toward ambiguity." The reason
for the soft approach toward incompleteness is that parties frequently
leave out customary terms when they draft contracts. If courts rou-
tinely excluded evidence of customary terms, either they would en-
force contracts inaccurately or they would force parties to incur
substantial costs in producing more complete contracts. The reason

Recentjudicial approval of this distinction can be found in AMInternationa Inc.

v. GraphicManagement Associates, Inc., 44 F.3d 572, 575 (7th Cir. 1995), which distin-
guishes between objective and subjective evidence of contractual ambiguity. Bernstein
criticizes admission of evidence regarding trade usage, course of performance, and
course of dealing, arguing that basing legal obligations on such evidence interferes
with contracting by (1) preventing parties from relying on the extra-legal promises
that are essential to value-maximizing business relations, and (2) forcing them to incur
high transaction costs to contract out of the rule (presumably by stating in their con-
tracts that trade usage, course of performance, and course of dealing are not legally
relevant). See Lisa Bernstein, Merchant Law in a Merchant Court: Rethinking the Code's
Search for Immanent Business Norms, 144 U. PA. L. REV. 1765, 1796-815 (1996)
(discussing the U.C.C.'s treatment of trade usage, course of performance, and course
of dealing). Bernstein's formalism relies on empirical assumptions on which I take no
position.
'4 See Goetz & Scott, supranote 11, at 313-16.
560 UNIVERSITY OFPENNSYLVANIA LAWREVEW [Vol. 146: 533

for the hard approach toward ambiguity, which is a version of the

penalty default argument, is that parties should be encouraged to
contract out of custom, because externalities that arise in connection
with the evolution of custom cause custom to lag behind technologi-
cal change. If courts routinely admitted evidence of customary terms,
parties would have no incentive to evaluate and modify custom in
response to technological change.
The problem with Goetz and Scott's analysis is that the assump-
tions of each argument conflict. The completeness argument as-
sumes that courts accurately discover customary terms and that high
transaction costs prevent parties from incorporating these terms di-
rectly into the writing. The ambiguity argument assumes that courts
fail to recognize the inadequacies of customary terms and that low
transaction costs enable parties to contract around customary terms.
Because the ambiguity rules and completeness rules overlap so much,
with both of them referring essentially to the use of extrinsic evidence
to interpret writings that do not allocate obligations in sufficient de-
tail, the use of different assumptions for each argument is not justi-
fled. This can be seen by exchanging "incompleteness" and
"ambiguity" wherever they appear in the description of their argu-
ment above. The substitution does not affect the plausibility of either
argument.

14. Involvement of a Lawyer

Some courts apply hard-PER to a contract against a party that
used a lawyer. 5 This practice follows straightforwardly from the as-
sumption that transaction costs are lower for sophisticated parties.

15. Contemporaneous Negotiations

Courts disagree over whether the parol evidence rule should ap-
ply to negotiations that occur at the same time that the contract is
signed. Corbin argues that this problem is illusory since every nego-
tiation is necessarily before or after the point in time at which a con-
tract is entered.46 But this surprising bit of formalism overlooks the
real problem. The question raised by the parol evidence rule is
whether courts should treat extrinsic evidence of statements made

4' See FARNSWORTH, supra note 2, § 7.16, at 548 (noting that courts place greater
burdens on parties with greater bargaining power and drafting skill).
41 SeeArthur L. Corbin, The ParolEvidence Rule, 53YALE L.J. 603,
618 (1944).
1998] PAROL EVIDENCERULE

immediately or a short time before the creation of the contract dif-

ferently from extrinsic evidence of statements made a long time be-
fore the creation of the contract. A court could reasonably make a
soft-PER exception to a general hard-PER rule for "contemporane-
ous" negotiations-that is, negotiations that occur shortly before the
signing of the contract-on the ground that such evidence on aver-
age sheds more light on the meaning of the contract than evidence of
earlier negotiations. Moreover, admission of "contemporaneous"
evidence is less likely to cause judicial error than admission of stale
evidence.

16. Scrivener's Error

Even the strictest adherents to hard-PER make exceptions for
scrivener's errors. Critics of hard-PER sometimes argue that this ex-
ception shows the weakness of the argument for hard-PER. if extrin-
sic evidence reliably discloses scrivener's errors, it can reliably disclose
that a term is unintended. However, the use of extrinsic evidence for
the limited purpose of identifying scrivener's errors is not likely to
create high expected error costs, and it eliminates the high transac-
tion costs that would be necessary to proofread a writing with great
care.

17. A Note on the Canons of Interpretation

Courts often rely on canons of interpretation, such as the injunc-

tion to interpret writings in a way that does not leave elements unex-
plained or superfluous, and the injunction to interpret the absence of
an item as intended when the writing contains a list of similar items.
Scholars scoff at these canons, and they are banal and do not affect
the treatment of extrinsic evidence, but they do cohere with the dis-
cussion of extrinsic evidence so far. For example, drafters can more
easily spot and correct glaring contradictions and omissions than
more subtle defects in the writing. So a court pressed with an inter-
pretation that creates a glaring contradiction or omission should hold
that the interpretation is erroneous. Because the cost of spotting and
correcting the contradiction is low, the court should assume that the
parties did not consider it a contradiction and had another interpre-
tation in mind. The relevant cost is not that of adding a promise to
the writing, but that of revising the writing to correct mistakes. Oth-
erwise the analysis is the same.
562 UNIVERS1TY OFPENNSYLVANIA LA W REVEW [Vol. 146:533

D. The Cost ofJudicialError

In prior sections we assumed that the parties do not want courts

to commit errors. This assumption perhaps seems obvious, but it is
useful to explore in more detail how judicial error produces costs for
parties.
Courts recognize that the parol evidence rule is related to the
value of "commercial certainty." Courts that support hard-PER argue
that this rule increases commercial certainty by enabling parties to
predict the promises that courts will enforce. Courts that support
soft-PER argue that soft-PER increases commercial certainty by allow-
ing judges and juries to consider all relevant evidence. 4 Few courts
discuss the way each rule reflects a different tradeoff,49 and none that
I know of has observed that the advantages of each rule relate to the
tradeoff of transaction costs and judicial error, or to contractual com-
plexity, conventionality, and value. However, by looking hard
enough, it is possible to find cases in which, for example, the simplic-
ity of the contract seems to cause a court to find an exception to the
soft-PER that prevails in its jurisdiction. 5 One can also find cases in

47 See, e.g., Wilson Arlington Co. v. Prudential Ins. Co., 912 F.2d 366, 369-70 (9th
Cir. 1990) (barring parol evidence); Trident Ctr. v. Connecticut Gen. Life, 847 F.2d
564, 568-69 (9th Cir. 1988) (criticizing California law); Delta Dynamics, Inc. v. Arioto,
446 P.2d 785, 789 (Cal. 1968) (Mosk, J., dissenting) (criticizing the California rule);
Masterson v. Sine, 436 P.2d 561, 570-71 (Cal. 1968) (Mosk, J., dissenting) (criticizing
the majority's liberal interpretation of the parol evidence rule); Jake C. Byers, Inc. v.
J.B.C. Investments, 834 S.W.2d 806, 811 (Mo. Ct. App. 1992) (emphasizing importance
of enhancing predictability and discouraging perjury); Hatley v. Stafford, 588 P.2d
603, 606-07 (Or. 1978) (emphasizing importance of commercial certainty); id. at 610
(Lent,J, dissenting) (same).
"' See the trio of soft-PER cases cited supranote 15.
'9One example is Cole Taylor Bank v. Truck Insurance Exchange, 51 F.3d 736 (7th
Cir. 1995). For scholarly discussion of these issues, see Robert Childres & Stephen J.
Spitz, Status in the Law of Contract, 47 N.Y.U. L. REV. 1, 8 (1972) (arguing for literal
contract interpretation); Goetz & Scott, supra note 11; Murray, supra note 3, at 1345-47
(arguing for a soft parol evidence rule); Edwin W. Patterson, The Interpretation and
Constructionof Contracts,64 COLUM. L. REV. 833, 846 (1964) (explaining how ambigu-
ous contract terms may justify admitting parol evidence). However, with the limited
exception of Goetz & Scott, whose focus is elsewhere, none of the commentators
analyzes the tradeoff in error costs produced by hard-PER and soft-PER, and little
emphasis is put on contractual complexity.
soThe California Supreme Court adopted a soft version of the parol evidence rule
in 1968 in the trio of soft-PER cases cited supra note 15. Two years later, in Salyer Grain
& Milling Co. v. Henson, 91 Cal. Rptr. 847 (Cal. Ct. App. 1970), which involved a con-
tract to haul grain, the court excluded testimony that the shipper bore the risk of loss
arising from an accident. Although the contract dealt with some other risk-allocation
issues, it was silent on this one. Under soft-PER, a court would admit the extrinsic
evidence since the contract appeared incomplete. However, the court argued that
1998] PAROL EVIDENCE RULE

which a court finds an exception to hard-PER because of the conven-

tionality of a contract 5 ' These cases are consistent with the model.
Courts rarely discuss why "commercial certainty"--that is, the re-
liable enforcement of promises-is valuable. The general answer to
this question is that the prospect of judicial error encourages the
promisor to engage in opportunism that reduces the gains from
trade. This problem is discussed extensively elsewhere,5 2 but it is use-
ful to show how it manifests itself in disputes involving the parol evi-
dence rule.
Return to the example of the automobile sale. Seller's state-
ments, S,provide information to Buyer about the product, such as its
price, function, and quality. Through the process of negotiation, the
parties refine their understandings of each other's interests. Later
statements made by the Seller confirm or disclaim earlier statements.
The contract contains those stipulations and disclaimers (set C); and
some subset of these stipulations and disclaimers is put in writing (set
M.
As suggested earlier, a court that follows hard-PER "errs" by ex-
cluding all of the statements in W',whereas the court that follows soft-
PER errs by including some of the statements in C . These errors
create two kinds of risks of opportunism. First, under hard-PER,
Seller (in our example) can bias the court's interpretation of the con-

when an otherwise detailed contract, which covers closely related issues, does not
allocate the risk of a particular contingency, and when that contract is negotiated by
sophisticated parties, one should assume that the parties intended the loss to lie where
it falls. See id. at 852. The implication is that the absence of a term must be taken to
be intended, so allowing in extrinsic evidence, ex post, creates the chance ofjudicial
error, and, ex ante, requires the parties to undergo the costly process of predicting
and disclaiming every possible judicial misinterpretation that could arise in the case of
a dispute.
-" In the early 1950s, Pennsylvania had an unambiguously hard-PER. See, e.g.,
Bardwell v. Willis, 100 A.2d 102 (Pa. 1953) (holding that if plaintiffs relied on any oral
understanding or representations made prior to the execution of a written lease, they
should have protected themselves by incorporating those promises into the written
contract). But in Berger v. PittsburghAuto Equipment, 127 A.2d 334, 337 (Pa. 1956), the
state supreme court admitted extrinsic evidence to show that an apparently uncondi-
tional lease of storage space in a building was (orally) conditioned on the floor being
strong enough to hold the lessee's equipment. In this straightforward transaction, it
must have seemed to the court that soft-PER's erroneous inclusion of extra-contractual
statements was considerably less dangerous than hard-PER's erroneous exclusion of
oral contractual statements. When transactions, such as this one, are routine, the
danger of erroneous inclusion of terms is slight, so requiring parties to spell out every
contingency serves no purpose.
52 See, e.g., TimothyJ. Muris, OpportunisticBehaviorand the Law of Contract4 65 MINN.
L. REV. 521 (1981).
564 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146:533

tract in its favor if it can make statements in W' that the court prop-
erly excludes (i.e., unwritten statements), but on which Buyer mistak-
enly relies. Second, under soft-PER, Seller can bias the court's
interpretation of the contract in its favor, if it can make statements in
C' that the court erroneously includes (i.e., extra-contractual state-
ments), on which Buyer does not rely. In both cases, Seller gains at
Buyer's expense. The risk of such opportunism causes buyers to take
costly precautions.
The first kind of opportunism is a standard concern in the litera-
ture on the parol evidence rules3 and is occasionally mentioned by
courts. The usual case occurs when a merchant orally tells an unso-
phisticated consumer that the object for sale has desirable attributes
that it in fact lacks. The writing, which is usually a complicated form,
states that the merchant makes no representations as to the quality of
the goods. Whether or not the writing also contains a merger clause,
hard-PER would generally bar the court from admitting the evidence
of the oral statement to vary the terms of the writing, leading con-
sumers to take costly self-protective measures. Ex ante, merchants
would like to commit themselves not to engage in this opportunism,
but because consumers refuse to read writings, they cannot. Soft-PER
is necessary to solve this problem and, indeed, courts in hard-PER
jurisdictions sometimes depart from the rule when a consumer is
involved.5'
The second kind of opportunism has not been discussed in the
literature or by the courts. Imagine that two sophisticated parties,
following negotiations, reduce their deal to a writing. The writing
contains a merger clause, but under soft-PER, courts might admit
statements from C' to supplement or vary the terms of the writing. As
a result, each party has an incentive to propose self-serving terms dur-
ing the negotiations, even though each knows that the other party will
reject the terms. The record of the self-serving terms creates a chance

53See, e.g., Murray, supra note 3, at 1372-89 (discussing standardized agreements

and the parol evidence rule); Justin Sweet, Promissoiy Fraud and the ParolEvidence Rule;
49 CAL. L. REV. 877, 896 (1961) ("Promises made without the intention on the part of
the promisor that they will be performed are unfortunately a facile and effective
means of deception."); W. Richard West, Jr., Note, ChiefJustice Traynor and the Parol
Evidence Rule; 22 STAN. L. REV. 547, 563 (1970) ("[T]he results of a mechanistic
protection of the written instrument are often harsh and unjust.").
54 See, e.g., Rempel v. Nationwide Life Ins. Co., 370 A.2d 366, 370 (Pa. 1977)
(stressing the vulnerability of consumers in insurance contracts); Mancini v. Morrow,
458 A.2d 580, 583 (Pa. Super. Ct. 1983) (applying a relaxed interpretation of the parol
evidence rule to the sale of a house).
1998] PAROL EVLDENCE RULE

that a court under soft-PER will erroneously enforce those terms

should a dispute arise. 5
This kind of opportunism not only gives each party an incentive
to state self-serving terms during the negotiations, but also gives each
an incentive to insist on a more ambiguous writing, as well as an in-
centive to take costly self-protective measures, such as carefully moni-
toring and recording the course of the negotiations. The parties
would prefer ex ante a regime in which they could commit themselves
to relying only on the writing. The problem with soft-PER is that it
hinders such a commitment.-
A related form of opportunism occurs when a party in a breach of
contract suit seeks the enforcement of a promise that was intended as
an extra-legal promise. An extra-legal promise is a promise made in
conjunction with the explicitly contractual promises but that was not
intended to be legally enforceable by the parties. Such promises oc-
cur because they are more effectively enforced through nonlegal
sanctions, such as injury to reputation, than through judicial sanc-
tions. Sometimes parties include extra-legal promises in the writing
along with disclaimers-a practice that is reflected in the cases involv-
ing employment manuals5 7 But parties also frequently use writings as

5 Sometimes, a court suggests that hard-PER prevents fraud. If the court is not
referring to fraudulent testimony (i.e., perjury), it might be referring to the opportun-
ism problem. See, e.g., Big G Corp. v. Henry, 536 A.2d 559, 562 (Vt. 1987) (noting that
the parol evidence rule forces contracting parties to understand that any contract
terms and conditions not in the final writing are null and void).
56As a possible illustration, consider the Alabama case of Hurst v. Nichols Research
Corp., 621 So. 2d 964 (Ala. 1993), in which a managerial employee sued the employer
for failing to allow him to purchase as much stock (through stock options) as he had
been promised. The written offer, to which the employee assented, specified a stock
option less than the size discussed during negotiations. Despite a strong precedent
supporting soft-PER, see Hibbett Sporting Goods, Inc. v. Biernbaum, 375 So. 2d 431,
436 (Ala. 1979) ("The question whether the parties have assented to a writing as a
complete and accurate integration of their contract is ultimately one of intent and
must be determined from the conduct and language of the parties and the surround-
ing circumstances."), the court affirmed the trial judge's refusal to admit the extrinsic
evidence. One explanation for this result is that the court was seeking to penalize
sophisticated employees for failing to disambiguate contractual negotiations. If the
employee considered the more valuable stock option to be a decisive element of the
contract, he should have made sure that the employer understood this. His failure to
do so could at least be considered negligent, and, more germane, could be considered
an attempt to obtain employment while maintaining a potential legal right to acquire
higher pay than the employer believed he had offered.
57 See, e.g., McDonald v. Mobil Coal Producing, Inc., 820 P.2d 986, 988 (Wyo. 1991)
(finding that whether an employment manual altered employment-at-will contract
terms was a genuine issue of material fact).
566 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146:533

a means of distinguishing legal promises, which are included in the

writing, from extra-legal promises, which are excluded from the writ-
ing. Allocating promises between the oral and the written is a cheap
and convenient way for parties to distinguish extra-legal and legal
promises.58
It is tempting to argue that soft-PER interferes with the parties'
ability to maintain the extra-legal component of their relationship.
By blurring the correspondence between oral representation and
extra-legality, on the one hand, and the correspondence between
written representation and enforceability, on the other, soft-PER in-
terferes with the use of nonlegal enforcement mechanisms to maxi-
mize the value of trading relationships.5 9 Although soft-PER may
cause parties to substitute to nonlegal enforcement by increasing the
cost of using writings to identify legal promises, this substitution may
either increase or decrease joint welfare. As the details of this argu-
ment are complex, they are placed in an appendix.

E. FurtherEconomic Considerations

It is worth considering a few other arguments that have been

made in support of, or in opposition to, the parol evidence rule.
These arguments appear in opinions now and again, but either are
versions of the arguments I have been discussing or are not plausible.
The first of these arguments is that hard-PER is valuable for pro-
moting judicial efficiency. Courts can more easily evaluate written
contracts than oral contracts. Presumably, writings serve as more
reliable evidence than does testimony regarding oral contracts, be-
cause (1) the testimony depends to a greater extent on memory, and
memory is fallible; and (2) the testimony depends to a greater extent
on honesty, and witnesses have an incentive to be dishonest. By en-

Cf Jason ScottJohnston, The Statute ofFraud4 in THE NEW PALGRAVE DIcTIONARY

OF ECONOMICS AND LAW (Peter Newman ed., forthcoming 1998) (arguing that the
Statute of Frauds enables parties to signal whether their promises are to be legally
enforceable).
59 Compare the collective bargaining case discussed earlier, Bidlack v. Wheelabrator
Corp., in which Judge Easterbrook suggests that extra-contractual promises serve as a
flexible basis for post-contractual behavior in light of changed circumstances, a func-
tion that is defeated if courts enforce such promises as though they were part of the
contract. See 993 F.2d 603, 618 (7th Cir. 1993) (en banc) (Easterbrook,J., dissenting).
Standard works on extra-contractual promises include Benjamin Klein & Keith B.
Leffler, The Role of MarketForces in Assuring ContractualPerformanc489 J. POL ECON. 615
(1981), and Lester Telser, A Theory of Self-EnforcingAgreements, 53J.BUS. 27 (1980).
1998] PAROL EVIDENCE RULE

couraging parties to use writings, the parol evidence rule facilitates

judicial evaluation of the dispute60
There are two problems with this argument. First, there is no rea-
son to believe that the judicial savings are significant, or that they are
worth the disruptions in business practices that result from forcing
people to formalize their contracts. Second, the parol evidence rule
does not encourage parties to use writings so much as it encourages
them to use complete writings rather than partial writings. Hard-PER
may cause parties to revert to oral contracts rather than to more com-
plete written contracts. Most importantly, judicial accuracy is not
independently valuable; it is valuable only to the extent that it in-
creases the value of contracts.
A related argument is that hard-PER is valuable for protecting
third parties from fraud. The classic justification for the Statute of
Frauds is that it discourages wrongdoers from trying to convince
courts that some third party owes them a contractual obligation.'
The parol evidence rule extends this policy to relations between two
contracting parties. It discourages a promisee from claiming that a
promisor orally made promises that were in fact never made. This is a
valid argument, but it is difficult to evaluate its implications. Grant-
ing the problem, one can still quarrel about whether hard-PER pro-
tection is necessary or whether soft-PER is sufficient.
Another argument occasionally presented in support of hard-PER
is that it allows judges to control juries by preventing them from hear-
ing extrinsic evidence. The concern is that ifjuries considered all of
the extrinsic evidence, rather than just the writing, they would not
render good judgments. 62 This argument might have historical sup-
port, but as a normative argument it is unconvincing. If juries are
incompetent, why would limiting them to certain kinds of evidence
lead to a more accurate result? If the argument is that the writing is
more reliable, so that juries, despite their inadequacies, would be
more likely to render a correct result, then the argument is indistin-
guishable from the judicial efficiency argument, discussed above.
The jury-control argument derives its minimal plausibility from the

60 See Childres & Spitz, supra note 49, at 8 (arguing that strict adherence to the
parol evidence rule is supported by "considerations of administrative convenience").
6' See Eric A. Posner, Norms, Formalities,and the Statute of Frauds: A Comment, 144 U.
PA. L. REV. 1971, 1977 (1996) (discussingjustifications for the Statute of Frauds).
12 See McCormick, supranote 17, at 366 ("The average jury will,
other things being
equal, lean strongly in favor of the side which is threatened with possible injustice and
certain hardship by the enforcement of the writing.").
568 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146:533

unarticulated premise that extrinsic evidence biases juries toward

poorer, weaker consumers and away from crafty merchants, so that
juries would not resolve the dispute in a neutral way. This brings us
to a final argument.
The final argument suggests that hard-PER prevents juries from
biasing their decisions toward poor people or consumers, in violation
of contractual expectations.6 Hard-PER thus allows judges to exclude
any extrinsic evidence that juries could use to justify biasing contrac-
tual interpretation in favor of the consumer. This argument assumes
that the parol evidence rule is most commonly invoked in cases in
which customers sign writings without reading or understanding
them, after being swayed by earlier oral promises by the merchant.6
This argument hardly supplies an attractive normativejustification for
the parol evidence rule. If juries systematically refuse to enforce the
law, this suggests either that juries correctly believe that the law is bad,
and therefore the law should be changed, or that juries incorrectly
believe that the law is bad, and therefore the jury system should be
changed.
This argument in favor of hard-PER is regarded by some as a
stronger argument against using hard-PER. According to this theory,
the abolition of the parol evidence rule would benefit consumers
because the rule allows merchants to mislead consumers by making
oral representations that are inconsistent with the writings. This ar-
gument is discussed in Part II.C.1, above.

III. CORBIN'S ARGUMENT

We should spend a moment considering Corbin's argument in fa-

vor of soft-PER, because it has been influential. Corbin believed that
the supporters of the parol evidence rule overestimated the determi-
nateness of written words. In fact, Corbin argued, the meaning of a
word depends on its context,6 so the parol evidence rule should not

63 See id. at 368 (asserting that without a firm rule, ajury will make its decision out
of "sympathy for a party whom the shoe of the written contract pinches").
In Pennsylvania, however, the initial move toward soft-PER occurred before any
consumer parol evidence rule cases reached the state supreme court. See supra notes
12-14. Similarly, none of the three famous California cases decided byJustice Traynor
involved consumers. See supra note 15.
Philosophy of language distinguishes between semantic and pragmatic mean-
ings of a word; the former are invariant with respect to context, the latter are not. See
Michael S. Moore, A NaturalLaw Theory of Interpretation,58 S. CAL L. REV. 277, 288-91
1998] PAROL EVIDENCE RULE

be used to downgrade evidence of context.'

Williston, the chief defender of the strict version of the parol evi-
dence rule, never revealed his linguistic presuppositions. Judges oc-
casionally have dropped statements about the nature of language, but
more as flourishes on a conventional doctrinal argument than as part
of a mature theoretical argument. Corbin was thus not attacking any
particular existing theory when he wrote. He must have assumed that
a supporter of hard-PER is compelled to assume that the meaning of
words does not depend on context.
Corbin cannot have meant that language is indeterminate. The
parol evidence rule does not privilege language over something else;
it privileges some language (the final writing) over other language
(earlier written and oral statements).67 If language were completely
indeterminate, a rule that allowed a court to consider all relevant
messages (i.e., pieces of extrinsic evidence) would not produce any
more accurate or determinate a result than a rule that allowed a court
to consider only some messages.
When Corbin argues that meaning depends on context, he seems
to mean that the more context admitted as evidence, the better are
the court's chances of discovering the meaning of the contract. The
parol evidence rule permits the admission of extrinsic evidence only

(1985). Under a charitable reading of Corbin's argument, Corbin assumes (plausibly)

that the pragmatic aspect of meaning dominates when courts interpret contracts.
[N]o man can determine the meaning of written words by merely glueing
his eyes within the four corners of a square paper; ... that it is men who give
meanings to words and that words in themselves have no meaning;
and... that, when a judge refuses to consider relevant extrinsic evidence on
the ground that the meaning of written words is to him plain and clear, his
decision is formed by and wholly based upon the completely extrinsic evi-
dence of his own personal education and experience.
Corbin, supra note 17, at 164. Corbin's analysis draws on that of Wigmore. See 9JOHN
HENRY WIGMORE, EVIDENCE IN TRIALS AT COMMON LAW § 2400, at 6 (James H.
Chadbourn ed., 1981) ("[A] writing has no efficacy pers4 but only in consequence of and
dependence upon other circumstances external to itself."). For updated versions of
this argument, see STANLEY FISH, THERE'S No SUCH THING AS FREE SPEECH AND IT'S A
GOOD THING, Too 144-56 (1994); Walter Benn Michaels, Against Formalism: The
Autonomous Text in Legal and Literary Interpretation, 1 POETICS TODAY 23 (1979). This
also seems to be the thrust of Traynor's claim that the parol evidence rule is based on
a belief that "perfect verbal expression" is possible, a belief that "is a remnant of a
primitive faith in the inherent potency and inherent meaning of words." Pacific Gas &
Elec. Co. v. G.W. Thomas Drayage & Rigging Co., 442 P.2d 641, 643-44 (Cal. 1968) (en
banc) (footnote omitted).
67 See, e.g., Olivia W. Karlin & Louis W. Karlin, The CaliforniaParolEvidence
Rule, 21
SW. U. L. REV. 1361, 1378-79 (1992) (discussing the "inherently contradictory" nature
of Corbin's argument for allowing extrinsic evidence in contract interpretation).
570 UNIVERSITY OFPENNSYLVANIA LA WREVIEW [Vol. 146:533

after the judge determines that the writing is ambiguous or incom-

plete, but a judge can determine that a writing is ambiguous or in-
complete only if he or she knows the commercial context to which
the writing refers. The word "chicken," for example, is precise in a
contract between a seller who has only one chicken and a buyer who
has agreed to buy it; but it is ambiguous in a contract signed by a
buyer and a seller who are dealers in an industry that distinguishes
between kinds of chicken which are routinely bought and sold, in-
cluding fryers and roasters. A court cannot determine whether
"chicken" is ambiguous without first looking at parol evidence of
trade custom and the parties' dealings; having done so, it has already
violated the parol evidence rule.& One can imagine that if judges
strictly obeyed the parol evidence rule, their decisions in a case such
as this would depend on their personal knowledge. A judge who
happens to know something about the chicken industry-perhaps
because of his past experience as a commercial lawyer-will believe
that the word "chicken" is ambiguous; a judge who knows nothing
about the industry will believe that the word "chicken" is unambigu-
ous. The parol evidence rule excludes extrinsic evidence from con-
sideration, while allowing the judge to rely on his or her personal
knowledge, even though the former, more so than the latter, would
enable the court to determine the parties' intentions.
Corbin's mistake is that, in assuming that the purpose of contract
law is to enforce the intentions of the parties, he overlooks the fact
that the parties, in addition to their ordinary contractual intentions,
have intentions about how courts should evaluate their contract in
case of a dispute. To Corbin, the parol evidence rule looks like an
arbitrary barrier to the enforcement of these ordinary contractual
intentions, rather than a device for furthering them. 69 In fact, the

6s See Michaels, supra note 66, at 25-27 (discussing Frigaliment Importing Co. v.
B.N.S. Int'l Sales Corp., 190 F. Supp. 116 (S.D.N.Y. 1960)).
69 The cardinal rule with which all interpretation begins is that its purpose is

to ascertain the intention of the parties. The criticized [parol evidence] rule,
if actually applied, excludes proof of their actual intention. It is universally
agreed that it is the first duty of the court to put itself in the position of the
parties at the time the contract was made; it is wholly impossible to do this
without being informed by extrinsic evidence of the circumstances surround-
ing the making of the contract.
Corbin, supra note 17, at 162. Corbin mistakenly assumes that the parties' intentions
cannot also include a preference against judicial evaluation of extrinsic evidence, a
preference based, one supposes, on the parties' prediction of whether extrinsic evi-
dence is more likely to misleadjudges or inform them. Traynor repeats Corbin's error
in his influential opinion criticizing the plain.meaning rule. SeePacific Gas &Elec., 442
1998] PAROL EVIDENCE RULE

parties derive advantage from being able, in their contract, to limit

the evidence a court can use to decide a dispute should one arise, just
as they derive advantage from being able, in their contract, to choose
the forum in which disputes will be resolved. The attractiveness of
hard-PER for this purpose-and the evidence that it is not arbitrary-
is shown by the frequency with which parties use merger clauses."
Although parties do fail to include merger clauses from time to time,
they do not, as far as I know, include "anti-merger" clauses that ex-
plicitly ask courts to look at extrinsic evidence. Yet if Corbin were
right, at least some sophisticated parties would agree with him and
include "anti-merger" clauses in their contracts, particularly in juris-
dictions that have a strict parol evidence rule.
In legal disputes, the context is determined by the rules of evi-
dence. Thejudge and jury look at all evidence relevant to the facts in
dispute. Contractual disputes are unique because the parties can, in
advance, specify the relevant context. This is important, because the
parties know in advance that judges err, and can evaluate this risk in
light of their contractual objectives and structure the contract in a way
that minimizes this risk.7 ' The parol evidence rule can thus be under-
stood as a device for allowing parties to choose the appropriate con-
text. The parties estimate the likelihood that a court looking back at
their contract will be misled by extrinsic evidence reflecting the con-
text, and they, in effect, signal to the court the context which they
think would be ideal, whether it be writing alone or writing plus ex-
trinsic evidence. As an example, one might expect the parties to sig-
nal the courts to ignore the context when the writing has been
preceded by complex preliminary negotiations during which the par-
ties made a host of promises they no longer want enforced.

P.2d at 644 (stating that the plain meaning rule "either den [ies] the relevance of the
intention of the parties or presuppose[s] a degree of verbal precision and stability our
language has not attained"). Fish and Michaels also repeat this error. See FISH, supra
note 66, at 146; Michaels, supra note 66, at 28-29.
70 I have not tried to do systematic research on the frequency with which
sophisti-
cated parties use merger clauses, but it is clear that they are extremely common.
Examples can be found among the form contracts reproduced in JOHN F. DOLAN,
UNIFORM COMMERCIAL CODE: TERMS AND TRANSAGIONS IN COMMERCIAL LAW (1991).
See, e.g., id. at 31 (K Mart's order contract); id. at 79 (Chrysler's sales and service
agreement with dealers); id. at 121 (Unisys equipment lease); id. at 198 (Michigan
National Bank business loan agreement); id. at 275 (Michigan National Bank purchase
of chattel paper); id. at 281 (Michigan National Bank aircraft security agreement).
71 Cf Gillian K. Hadfield, Judicial Competence and the Interpretationof Incomplete
Con-
tracts, 23J. LEGAL STUD. 159, 163 (1994) (discussing the concept ofjudicial error in
the context of contractual incompleteness).
572 UNTVERS17Y OFPENNSYLVANIA LAWREVIEW [Vol. 146: 533

Still, perhaps this argument does not represent Corbin's views

fairly. Corbin could be understood to say the following: The set Hof
promises enforced under hard-PER is a function both of the writing
and the general interpretive rules used by courts, including rules for
implying terms and the courts' prejudices about the form and pur-
pose of typical business transactions. The scope of every term in Wis
not self-defining, but depends on the linguistic practices of the par-
ties.7 In order to interpret these terms, courts must either rely on
their own linguistic practices or make judgments about the parties'
linguistic practices. The set G of promises enforced under soft-PER is
a function both of the writing and the extrinsic evidence as inter-
preted in light of the judges' commercial prejudices and linguistic
practices.
Thus, whereas the analysis in Part I assumes that H is more sensi-
tive to the contents of the writing than G is, it is clear that under both
regimes the courts' use of the writing will be mediated by elements
outside the control of the parties: the general interpretive rules, the
court's prejudices under hard-PER, and the extrinsic evidence as in-
terpreted in light of the courts' prejudices under soft-PER. It is pos-
sible that the influence of these mediating elements on the courts'
behavior dwarfs the impact of the writing, in which case the practical
distinction between hard-PER and soft-PER is nil.
All of this is possible. But because the principles governing the
implication of terms and the general interpretive principles are in-
variant with respect to the facts of contractual negotiation, parties
should be able to take account of these principles when negotiating
their contract and predicting judicial enforcement under hard-PER.
Because judges are appointed or elected from a homogenous group
of people, and because their interpretive prejudices are revealed in
their decisions and opinions, these prejudices should be relatively
predictable at the time of contracting. But parties cannot know in
advance the effect of extrinsic evidence on judicial decisions under
soft-PER, because each party cannot know in advance what the other
party might introduce as extrinsic evidence-which excerpts from the
general chit-chat or which pages of scrawled notes-should a dispute
arise.
Corbin's argument is, at its core, based on a skepticism about the
capacity of formalities to restrain courts and render their judgments

2 See David Charny, HypotheticalBargains: The Normative Structure of ContractInterpre-

tation, 89 MICH. L. REv. 1815, 1856 (1991) (discussing interpretive issues).
1998] PAROL EVDENCE RULE

predictable. If he is right, then talk of the differing effects of soft-PER

and hard-PER is idle, but so too is talk of the advantages and disad-
vantages of any doctrine of contract law.

CONCLUSION

This Essay has made an argument about the conditions under

which stricter and weaker parol evidence rules are consistent with a
value-maximizing contract law. The normative implications of the
argument are limited, however, because it is difficult to measure the
variables on which the value of the parol evidence rule depends. But
the argument does tell a unifying story, a story that, pace Corbin and
others, shows that the parol evidence rule serves straightforward func-
tions and coheres with a large number of contract law doctrines.
Although nothing in the argument compels a jurisdiction to choose
one parol evidence approach over the other, the argument supplies a
coherent way of thinking about the parol evidence rule that, if fol-
lowed, will promote doctrinal consistency regardless of which general
approach is taken.
The discussion has implications beyond the issue of the treatment
of extrinsic evidence. For contract law, the discussion sheds light on
old debates about the optimal tailoring of default rules, emphasizing
the decisive roles of judicial competence and transaction costs. The
discussion also has implications for statutory analysis. The use of ex-
trinsic evidence to interpret written contracts is analogous to the use
of legislative history in statutory interpretation. This analogy, of
course, has been pursued in earlier writings about statutory interpre-
tation, but those writings relied on now-dated theories of contract
interpretation.
The discussion suggests that the use of legislative history for the
purpose of interpreting statutes should depend, among other things,
on judicial competence and the "transaction costs" of reducing legis-
lative negotiations to a written statute. Judicial competence varies
from state to state and from country to country; it may be poor, for
example, in developing nations. Thus, the analysis suggests that the
optimal form of legislative interpretation will vary among jurisdic-
tions. Transaction costs vary with the complexity and novelty of legis-
lation. Legislative history may, for example, be less justified for
interpreting incremental modifications of existing legislation than for
interpreting statutes that generate an entirely new set of legal rights
and entitlements or administrative machinery.
574 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146:533

APPENDIX

One argument in favor of hard-PER is that it reduces the cost of

using a writing in order to obtain an enforceable contract. If, how-
ever, parties rely on both legal and nonlegal sanctions for the en-
forcement of contracts, reduction in the cost of a writing does not
necessarily increase the ex ante value of contracting. The reason is
that the reduction in the cost of legal enforcement may cause parties
to substitute from nonlegal enforcement to legal enforcement even
when nonlegal enforcement remains cheaper overall. This Appendix
sketches a model demonstrating this point.
Suppose there are two types of sellers, "high types" (S,) and "low
types" (SL), with high types constituting a fraction, q, of the popula-
tion of sellers. High types value future gains more than low types do.
Buyers cannot distinguish sellers by type. A seller makes an oral offer
to sell a widget for a certain price. A buyer (B) then chooses among:
(1) rejecting the offer, in which case payoffs are zero for both parties;
(2) accepting the offer orally; and (3) accepting the offer on the con-
dition that the seller supplies a writing that meets the requirements
for a legally enforceable contract. The oral contract is not legally
enforceable; the written contract is legally enforceable to the extent
that performance is verifiable by courts. Unless the buyer rejects, the
seller then either "cooperates" or "cheats." These terms are defined
in the following way. To cooperate is to perform the contract in a way
that exceeds legal requirements-for example, delivering before
deadline, forgiving a late payment, or providing unusually high qual-
ity goods, as sellers often do for their best customers. To cheat is to
do as little as is legally required, which means either (a) not perform-
ing if the contract is legally unenforceable, or (b) performing in a
crabbed and literalistic way if the contract is legally enforceable.
More precisely, suppose that the optimal contract contains some
terms whose performance can be verified by courts but other terms
whose performance can be observed only by the parties: then to
cheat on a written contract means to perform only verifiable value-
maximizing actions, whereas to cooperate means to perform both
verifiable and observable value-maximizing actions.
B receives v if S performs literalistically (that is, cheats on a writ-
ing); dv if S cooperates (d > 1); and -r if S fails to perform (that is,
cheats on an oral contract). It is assumed that B, rather than S, ab-
sorbs the cost.of the writing, w, but nothing turns on this assumption.
S receives p if B accepts and S cooperates. This amount is multiplied
1998] PAROL EVUDENCERULE

by a if S is a high type and B accepts orally (a > 1). This multiplier

represents the discounted long-term gain from future business. It is
assumed that if S cooperates, that is, treats B well during the transac-
tion-by performing at a level beyond the requirements of the writ-
ing-B will more likely take future business to S than if S performs
merely adequately. But only S gains from the expected future busi-
ness. SL does not benefit from the multiplier, because S does not
care about future payoffs. It is assumed, for reasons given below, that
if B demands a writing, SH cannot obtain long-term gains by cooperat-
ing. Either type of seller gains bp if she cheats B on a writing, b > 1,
but b < a. This reflects the fact that a seller does better by cheating
than by cooperating if there are no future consequences, but the high
type of seller gains even more by establishing a long-term relationship
with the buyer. Either type of seller gains cp if she cheats B on an oral
contract, where a > c > b. A seller saves money by performing in a
literalistic way, rather than in a value-maximizing way; she saves more
if she can refuse to perform than if she performs literalistically, but
the high seller would rather lose the short-term gains from cheating
than lose future business.
Return to the assumption that if B demands a writing, SH will not
obtain long-term gains from cooperating. This assumption is based
on the intuition, which is not modeled, that by demanding a writing,
B signals to S, that B does not trust S, in which case S, would be re-
luctant to incur the costs of entering a long-term relationship. To
model this intuition, one would have to assume that there are two
types of B, a high type and a low type, and only B, gains from a long-
term relationship with SH S ,, then, would want to initiate such a rela-
tionship, through cooperating, only if she believed that the buyer
belonged to the high type. B, would signal his type by declining to
demand a writing, thus exposing himself to opportunism, while BL
would demand a writing. Given that only a BL would demand a writ-
ing, SH would not cooperate with someone who demanded a writing.
This additional assumption is realistic, because the problem of dem-
onstrating trust is usually two-sided, but incorporating it in the model
would produce great complexities, and it is best left to future work.
There are three equilibria of interest. In the distrust equilibrium, B
demands a writing, and S cheats the buyer. The buyer does not devi-
ate, if v-w > -r. (Actually, v-w > 0; if v-w < 0, then, as described below,
a reject equilibrium will obtain.) Given that the seller will cheat, the
buyer should protect himself with a writing. The seller (either type)
does not deviate if bp > p, which is true by assumption. Given that the
576 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 146: 533

buyer will demand a writing and therefore cannot be trusted, S will

cheat.
In the trust equilibrium, B accepts orally; S, cooperates; SL cheats. B
does not deviate if (qdv+(1-q)(-r)) > v-w, that is, w > (v-qdv+(1-q)r).
Given that B gains more from cooperating with S, than he loses from
being cheated by SL, relative to his payoff in the distrust equilibrium,
B does better by accepting orally. SH does not deviate if ap > cp, which
is true by assumption. SL does not deviate if cp > p, which is true by
assumption. SH prefers the long-term relationship that can be
achieved through cooperation; SL prefers the gains from cheating.
In the reject equilibrium,B rejects S's offer. In the distrust equilib-
rium B would receive v-w; in the trust equilibrium B would receive
qdv+(1-q)(-r). If w > v and if ((1-q)(r)) > qdv, then B would be better off
receiving 0. So B rejects regardless of what S would do. When writ-
ings are expensive and the expected loss from dealing with SL is suffi-
ciently high, B rejects all offers.
To see the effect of the parol evidence rule, imagine that a trust
equilibrium exists. Notice that B accepts orally only as long as the
cost of the writing is sufficiently high. If the cost of the writing de-
clines sufficiently, to w' < w, as it might if soft-PER were converted to
hard-PER, then B would deviate. The new equilibrium would be a
distrust equilibrium. B's new payoff is v-w' and S;'s payoff is bp. This
is a gain to B of v-w'-qdv+(1-q)r. S's payoff falls from ap to bp; and SL's
payoff falls from cp to bp. So the reduction in the cost of the writing
causes an aggregate loss if (v-w'-qdv+(1-q)r) < ((a+c-2b)p).
This result shows that a shift from soft-PER to hard-PER can re-
duce aggregate wealth. As noted above, an important step lies out-
side the model: When BH demands a writing to protect himself
against exploitation by an SL, S, can no longer distinguish a BH from a
B, In the model, then, S, will cheat if B demands a writing. B ac-
cepts orally if the value of the trust relationship with SH exceeds the
cost of being exploited by SL, given the cost of using a writing to pro-
tect himself. As the cost of using a writing declines, however, protec-
tion against SL becomes more attractive, until B prefers it to the long-
term relationship with SR. In making this move, B does not take ac-
count of the costs to Si, in particular, the cost to S, from losing the
ability to enter long-term contracts.
There are many problems with this argument, which can be re-
solved only with a more complicated model. Therefore, the argu-
ment should be taken as conjecture only. If the argument is correct,
it has significance beyond the question of the optimal parol evidence
1998] PAROL EVWDENCE RULE 577

rule. If nonlegal enforcement of contracts is significant, then a re-

duction in the cost of using the legal system can make parties worse
off. This has importance for debates about rules and standards, for-
malities, and other issues of legal process, where the usual assumption
is that reducing the cost of the legal system makes all parties better
off.
* * * * * *