WHY HOLD-UPS OCCUR: THE SELF-ENFORCING RANGE OF CONTRACTUAL RELATIONSHIPS
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
WHY HOLD-UPS OCCUR: THE SELF-ENFORCING RANGE OF
CONTRACTUAL RELATIONSHIPS
BENJAMIN KLEIN*
Court enforcement and private enfbrcement are not alternative contract enJorcement
mechanisms, but are used jointly by transactors to d@ne the se!f-enfbrcing range 4
a contractual relationship. Within this framework contract terms economize on the
limited amounts I$ private @rcement capital possessed by transactors, either by
directly controlling transactor behavior or by sh@ing private enfbrcement capital be-
tween transactors to coincide with likely fiture market conditions. Hold-ups occur
when market conditions change strfficiently to place the relationship outside the s e y
erfbrcing range. This probabilistic view I$ hold-ups is contrasted with opportunism
more generally and with moral hazard behavior.
One of my most enjoyable intellectual tionships knowing that a hold-up may
experiences was working with Armen Al- take place (but believing that the expected
chian on the Klein, Crawford and Alchian gains from trade outweigh the expected
[1978] hold-up paper. In this paper I ex- rent-dissipating costs associated with the
tend the basic framework presented in hold-up risk), is shown to have important
that paper, pointing out what I now con- implications for understanding the struc-
sider to be its shortcomings and providing ture of contracts adopted by transactors in
insights into the nature of hold-ups and the marketplace.
the form of contracts chosen by transac-
tors to avoid hold-ups. The major analyt- I. WHY DO HOLD-UPS OCCUR?
ical extension entails combining hold-up I begin with a simple example that
analysis with my work on private enforce- illustrates the basic economic forces in-
ment. Because private enforcement capital volved in a hold-up. Assume that a
is limited and written contract terms are builder constructs a house on a piece of
necessarily imperfect, transactors must land the builder does not own but, rather,
optimally combine court-enforced written only leases short-term. After the initial
terms together with privately enforced un- land lease expires, the landowner could
written terms to define what I call the hold up the builder by raising the land
self-enforcing range of their contractual rent to reflect the costs of moving the
relationship. Hold-ups occur when un- house to another lot. This example illus-
anticipated events place the contractual trates all the hold-up factors emphasized
relationship outside the self-enforcing in Klein, Crawford and Alchian-(a) the
range. This probabilistic framework, builder has made an investment that is
where transactors enter contractual rela- highly specific to a particular piece of land
and (b) the landowner has taken advan-
tage of the incompleteness of the contract
* Professor, University of California, Los Angeles. that governs the relationship (in particu-
I am grateful for comments from Armen Alchian, Har-
old Demsetz, Andrew Dick, Jon Karpoff, John Lott, lar, the fact that the lease does not cover
Kevin Murphy and an anonymous referee. future years) to (c) expropriate the quasi-
444
Economic Inquiry
(ISSN 00952583)
Vol. XXXIV, July 1996,444-463 @WesternEconomic Association InternationalKLEIN: SELF-ENFORCING RANGE 445
rents on the builder’s specific investment. unsatisfactory way to explain the inci-
The obvious question is why anything like dence of hold-ups. Simple examples of
this would ever occur; that is, why would deception, such as a builder constructing
someone be so naive as to build a house a house on land that is only rented short-
on land for which they had only a short- term, rarely, if ever, occur. More compli-
term lease? cated and less obvious examples of hold-
Our primary goal in Klein, Crawford ups may sometimes involve the deception
and Alchian was not to explain the exis- of an imperfectly informed transactor.
tence of hold-ups, but rather the institu- However, explanations of hold-up behav-
tions adopted by transactors to avoid ior based upon transactor deception are
hold-ups. For example, we would expect often either not refutable or clearly incon-
that builders, anticipating a potential sistent with the facts. For example, the
hold-up problem, would decide to pur- most extensively cited hold-up example
chase the land or at least to sign a long- presented in Klein, Crawford and Alchian
term ground lease before starting con- is the Fisher Body-General Motors case, a
struction. However, we do present some transaction between two large, sophisti-
examples in the paper of hold-ups that cated business firms with no evidence of
actually occurred. The implicit reason we any pre-contract deception on either
give for the occurrence of these hold-ups transactor’s part.
is transactor ignorance. Apparently, trans- The Fisher Body-General Motors case
actors are not always smart enough to concerned a contract signed by General
choose the contractual arrangement that Motors and Fisher Body in 1919 for the
would eliminate the hold-up problem. supply of automobile bodies by Fisher to
Oliver Williamson provides a similar, General Motors2 Fisher Body, in order to
but much more explicit answer to the produce the automobile bodies, had to
question of why hold-ups occur. When make an investment in stamping machines
defining ”opportunism” he states: and dies that was highly specific to Gen-
eral Motors. As a result, a significant po-
By opportunism I mean self-interest
seeking with guile. This includes but tential was created for General Motors to
is scarcely limited to more blatant forms, hold up Fisher. After Fisher Body made
such as lying, stealing and cheating. the specific investment, General Motors
Opportunism more often involves subtle could have threatened to reduce its de-
forms of deceit. ...More generally, op-
portunism refers to the incomplete or
mand for Fisher-produced bodies, or even
distorted disclosure of information, es- to terminate its relationship with Fisher
pecially to calculated efforts to mislead completely, unless Fisher reduced its
distort, obfuscate, or otherwise confuse.
f
prices.
For example, the hold-up may have oc- The Fisher Body-General Motors case
curred in our illustrative house construc- appears analogous to our naive house
tion example because the landowner de- construction qn rented land example.
ceived the builder with a low up-front However, contrary to our house construc-
land rental price and vague promises tion example, the transactors in the Fisher-
about the future. General Motors case clearly recognized
Relying on the ability of one transactor
to take advantage of the naivete or igno-
rance of another transactor is a highly 2. The description of the Fisher-General Motors
contract is taken from Klein et al. [1978,308-101. The
contractual agreement between Fisher Body and Gen-
eral Motors can be found in the minutes of the Board
1. Williamson [1985, 471. Also see Williamson of Directors of Fisher Body Corporation for Novem-
[1979, 234, n. 31. ber 7, 1919.446 ECONOMIC INQUIRY
the hold-up potential and attempted to was no economic reason for Fisher to
take account of it in their contract terms locate their plant close to the General
before any specific investments were Motors assembly plant when, according to
made. In particular, to prevent General the contract, they could profit by locating
Motors from appropriating the quasi-rents their plant far away from the General
from the Fisher investment by threatening Motors plant and put a 17.6 percent up-
to reduce its purchases from Fisher, the charge on their transportation costs. The
contract included a ten-year exclusive result was automobile bodies that were
dealing clause. This clause required Gen- very costly for General Motors to purchase
eral Motors to buy all of its closed metal and highly profitable for Fisher to pro-
automobile bodies from Fisher for a pe- d~ce.~
riod of ten years. The Fisher-General Motors case illus-
Obviously, such a contract had to set trates why transactors are concerned
the price at which Fisher would supply about hold-ups. When automobile bodies
bodies to General Motors. The transactors are produced and sold inefficiently, as
agreed upon a formula where the price they were by Fisher, the total gains from
was set equal to Fisher’s “variable cost” trade are reduced. We can expect in such
plus 17.6 percent. An upcharge over vari- cases that ex post renegotiation of the con-
able costs, rather than a formula based on tract will occur so that, after a lump sum
Fisher’s total cost, was probably used be- is paid to the transactor engaging in the
cause Fisher was selling automobile bod- hold-up, price and cost will return to the
ies to many different companies and it was efficient level. In the Fisher-General Mo-
difficult to isolate and measure the capital tors case the contract renegotiation took
and overhead costs associated with Gen- the form of a General Motors’ side pay-
eral Motors shipments. The 17.6 percent ment to the Fisher brothers along with
upcharge presumably was designed to purchase of the Fisher Body company.
cover Fisher’s anticipated capital and Since the probability of such ex post
overhead costs. lump-sum transfers will be taken into ac-
The Fisher-General Motors contract, count by transactors in their ex ante con-
therefore, was not totally unsophisticated, tract terms, these hold-up lump-sum
as was the short-term land-lease contract transfers may appear to be of no signifi-
in our hypothetical example. However, the cance if transactors are risk neutral. How-
Fisher-General Motors contract, as it ever, as the Fisher-General Motors case
turned out, was similarly inadequate in vividly illustrates, the transactor placed at
preventing a hold-up, albeit to the advan- a disadvantage during a hold-up does not
tage of Fisher rather than General Motors. immediately costlessly renegotiate the
After the contract was signed, the demand contract and make a lump-sum payment
for automobiles rose substantially. Fisher to the transactor engaging in the hold-up.
took advantage of the contract in the face Real resources are wasted during the hold-
of this large demand increase to adopt an up process, as transactors attempt to con-
inefficient, highly labor-intensive produc- vince their transacting partners that a
tion process and to locate its body-produc- hold-up potential does exist and of its
ing plants far away from the General Mo- magnitude. It is these dissipative, purely
tors assembly plant. From Fisher’s point redistributive costs associated with hold-
of view there was no economic reason to up behavior, not the lump-sum transfer
make capital investments when, according
to the contract, they could instead hire a
3. See deposition testimony of Alfred I? Sloan, Jr.
worker and put a 17.6 percent upcharge in United States v. Dupont 6 Co., 366 U.S. 316 (1961),
on the worker’s wage. In addition, there 186-90 (April 28, 1952) and 2908-14 (March 14, 1953).KLEIN: SELF-ENFORCING RANGE 447
itself, that are wasteful. Because of these The costs associated with contractual
costs it is efficient for transactors to design specification that lead transactors to use
contractual relationships that reduce the incomplete and imperfect contracts in-
likelihood of a hold-up occurring. volve much more than the narrow trans-
The obvious question in the Fisher-Gen- action costs of writing down responses to
eral Motors case is why a hold-up oc- additional contingencies. In addition to
curred-that is, why did General Motors these extra ”ink costs,” complete contrac-
use such an imperfect or incomplete con- tual specification entails wasteful search
tract which placed it in a position where and negotiation costs associated with dis-
it could be held-up by Fisher in the way covering and negotiating prespecified
it was? It is much too unlikely an expla- contractual responses to all potential ccn-
nation to rely on General Motors’ naivete tingencies. Because most future events can
or on Fisher’s deception. General Motors be accommodated at lower cost after the
and Fisher Body were aware of the hold- relevant information is revealed, much of
up problems inherent in their relationship, this activity involves largely redistributive
and both Fisher and General Motors had rent dissipation with little or no allocative
to have been aware that the contract they benefit. Transactors are merely attempting
adopted to solve their hold-up problem to obtain an informational advantage over
was ”defective” in the sense that it con- their transacting partners, hoping to place
tained obvious malincentives. Yet General themselves in a position where they will
Motors and Fisher adopted this incom- be more likely to collect on (and less likely
plete and imperfect contract because they to pay for) h o l d - u p ~ .Therefore,
~ rather
believed it would have been more costly than attempting to determine all of the
to write a more complete and perfect con- many events that might occur during the
tract. life of a contractual relationship and writ-
ing a prespecified response to each, the
gains from exchange are increased by the
It. THE USE OF INCOMPLETE CONTRACTS use of incomplete contracts.
General Motors and Fisher knowingly Transactors also use incomplete con-
entered into their incomplete contract be- tracts because writing something down to
cause they believed that this contract, be enforced by the court creates rigidity.
while imperfect, was optimally designed Since contract terms are necessarily imper-
to minimize the probability of a hold-up fect, once something is written down
occurring. Unfortunately, conditions de- transactors can engage in a hold-up by
veloped that permitted Fisher to use the rigidly enforcing these imperfect contract
contract to hold up General Motors. If terms, even if the literal terms are contrary
General Motors and Fisher had known to the intent of the contracting parties.
ahead of time what was to happen, no This is what occurred in the Fisher-Gen-
doubt they would have written their con- eral Motors case, where the written con-
tract to take account of the problems that tract terms that were meant to prevent
developed. In that sense the Fisher hold- General Motors from holding up Fisher
up of General Motors was unanticipated. were actually used by Fisher to create a
However, in an uncertain world where much greater hold-up of General Motors.
complete contractual specification is
costly, transactors use incomplete con-
tracts that deliberately do not take account 4. These rent dissipating costs during the contract
of every contingency. As a result, transac- negotiation process are analogous to the costs associ-
ated with the purely redistributive oversearchina for
tors knowingly leave themselves open to an informatior$l advantage analyzed in Kenney-and
the possibility of hold-ups. Klein [19831.448 ECONOMIC INQUIRY It may appear that this type of hold-up, This does not mean that writing down where a transacting party uses the court contract terms is not beneficial to transac- and the threat of litigation to enforce an tors. Writing down binding contract terms imperfect contract term that is contrary to has the obvious benefit that the court can the intent of the contracting parties, is be used to enforce performance. The idea different from the type of hold-up that that court enforcement of explicit con- occurred in our house construction exam- tracts may be the mechanism by which a ple, where the landowner took advantage transactor engages in a hold-up merely of the absence of a contract to hold up the recognizes that contractual specification builder after the short-term land lease not only has benefits but also has associ- expired. We may wish to think of the court ated costs. For some elements of perfor- as unable to protect the builder in the mance there may be no trade-off in terms house construction case, whereas actually of added rigidity associated with writing the court is effectuating the hold-up by down contract terms. For example, con- strictly enforcing the written contract tractual specification is costless when de- terms in the Fisher-General Motors case. sired performance is measured accurately However, although this distinction may be by the contractually specified term and the important for contract law, the hold-ups term is costlessly observable by the court. are analytically similar. Both hold-ups are However, when transactors must use a caused by a transactor using the court to less than perfect proxy for performance in take advantage of an imperfection in the a contract there is a trade-off. Including contract that governs an economic rela- the proxy in the contract not only may tionship. In the Fisher Body-General Mo- help in enforcing the understanding but tors case, court enforcement of the imper- also may do harm by making the contrac- fect cost-plus contract sanctions Fisher’s tual arrangement more rigid. attempt to charge General Motors arbitrar- It is the very benefit of contract specifi- ily high prices. Similarly, in the house cation, i.e., that transactors’ hands can be construction case, court enforcement of tied with respect to certain variables that the obvious imperfection in the contract might otherwise be used to effectuate a (namely, that the contract only covers the hold-up, that creates the harm of contrac- short term) sanctions the landowner’s at- tual rigidity. As the Fisher-General Motors tempt to charge the builder an arbitrarily case illustrates, once an agreement is for- high price after the short-term land lease malized in a written contract, it cannot expires. cheaply be breached if unanticipated I am assuming in this discussion that changes occur in the market. The only the court only enforces written terms and limit on the cost to General Motors of not does not enforce unwritten terms. This is, performing to the literal terms of the im- of course, an oversimplification. Courts perfect contract when market conditions interpret both written and unwritten deviated substantially from ex ante expec- terms when enforcing contractual agree- tations was essentially General Motors’ ments. However, we can assume that the declaration of bankruptcy. amount of discretion exercised by the If, on the other hand, a contractual court with regard to unambiguous written understanding is not formalized in a writ- terms is limited, and that as transactors ten contract, transactors can more cheaply add additional things to their contracts the opt out of the agreement if subsequent likelihood that the court will effectuate a market conditions deviate substantially hold-up by rigidly enforcing these imper- from expectations. The understanding is fect contract terms increases. much more flexible because, without the
KLEIN: SELF-ENFORCING RANGE 449
court forcing transactors to perform to the other transactors in their dealings with
literal terms of the contract, transactors this transactor, the transactor engaging in
can renege and only lose the value of the hold-up will face increased costs of
whatever transactor-specific investments doing business in the future. Potential
are present in the relationship. Therefore, trading partners will become less willing
at some point transactors may decide to to rely upon the transactor’s promises and
avoid the rigidity associated with court demand more favorable and/or more ex-
enforcement of written contract terms by plicit contract terms. For example, if Gen-
intentionally leaving many elements of eral Motors had held-up Fisher and this
intended performance unspecified and en- was communicated in the marketplace,
forcing these terms instead by a private General Motors would have found it more
enforcement m e ~ h a n i s m . ~ expensive to purchase inputs in the future.
Each transacting party compares the
Ill. THE SELF-ENFORCING RANGE OF potential hold-up gain from breaching the
CONTRACTUAL RELATIONSHIPS contractual understanding with the capi-
The privately imposed sanction that tal loss from the private sanction. If the
permits transactors to enforce the unwrit- hold-up gain is less than the capital cost,
ten terms of their contracts can be thought then the transactor cannot credibly
of as consisting of two parts. One part is threaten breach of the contractual under-
the future loss that can be imposed di- standing. Therefore, although transactors
rectly on the transactor if the relationship could take advantage of the fact that all
is terminated. Given the presence of non- the elements of a contractual understand-
salvageable transactor-specific invest- ing are not perfectly specified in the writ-
ments, the threat of termination of the ten contract, they will not do so and will
relationship implies a potential capital instead perform in a manner that is con-
loss equal to the discounted value of the sistent with the mutually understood con-
quasi-rents from these investments. For tractual intent.
example, if General Motors had termi- The magnitude of the private sanctions
nated (or failed to renew) its relationship that can be imposed on each transactor
with Fisher, they could have imposed a who attempts a hold-up defines what can
capital cost on Fisher for non-performance be called the self-enforcing range of the
equal to the specific investments made by contractual relationship. The self-enforc-
Fisher in the General Motors specific tools ing range measures the extent to which
and dies. market conditions can change without
The other part of the private sanction precipitating a hold-up by either party.
that is imposed on a transactor who is Changes in market conditions may alter
engaging in the hold-up is the damage to the value of specific investments and,
the transactor’s reputation in the market- therefore, the hold-up potential, yet as
place. If the violation of the contractual long as the relationship remains within the
understanding is taken account of by self-enforcing range where each
transactor’s hold-up potential gain is less
than the private sanction, a hold-up will
not take place. Only when changes in
5. The private enforcement mechanism upon
which the following analysis is based is presented in market conditions move transactors out-
Klein and Leffler [1981]. Lott [1988]extends the Klein side the self-enforcing range so that the
and Leffler model in the spirit of the present paper
by introducing random changes in cost or demand one-time gain from breach exceeds the
which alter the incentive of transactors to perform. A private sanction will the hold-up threat,
firm‘s decision to cheat is also considered to be sto- i.e., the threat of breach of the contractual
chastic in Darby and Kami 119731 and Karpoff and
Lott 119931. understanding, become credible. When450 ECONOMIC INQUIRY
this occurs the transactor will not be de- era1 Motors contract would have been
terred from breaching even if the transac- self-enforcing and the malincentives asso-
tor expects to be terminated and knows ciated with the cost-plus contract terms
that everyone in the marketplace will would not have mattered. Fisher would
think he is a “cheat.” This is what oc- have known that they could not take ad-
curred in the General Motors-Fisher Body vantage of the literal terms of the contract
case. Fisher and General Motors found without being punished by General Mo-
themselves outside the self-enforcing tors and that the punishment would have
range because of a very large increase in been greater than their hold-up gain.
demand by General Motors for Fisher-pro-
duced bodies. This increase in demand IV. AN ILLUSTRATION: THE ALCOA-ESSEX
CASE
increased the Fisher hold-up potential so
much that it became larger than the pri- The concept of the self-enforcing range
vate sanction that could be imposed on of a contractual relationship can be further
Fisher by General Motors and Fisher illustrated by the Alcoa-Essex case.7 Essex,
found it profitable to violate the intent of an aluminum cable manufacturer, located
the contractual understanding by taking its cable fabrication plant adjacent to an
advantage of imperfect terms of the agree- Alcoa aluminum production facility,
ment. thereby permitting shipments of pro-
The change in market conditions that cessed aluminum from Alcoa to Essex in
permitted Fisher to take advantage of molten form. While the Essex plant loca-
General Motors in this way was presum- tion lowered costs, it also created an Alcoa
ably unanticipated. When the contract was hold-up potential. Alcoa could threaten to
entered into in 1919 the dominant produc- hold up Essex by increasing the price of
tion process for automobiles consisted of delivered aluminum, thereby expropriat-
individually constructed, largely wooden ing the value of the Alcoa-specific element
open bodies; the closed metal bodies sup- of Essex’s investment, namely, the added
plied by Fisher were essentially a novelty. transportation cost of receiving aluminum
After 1919, demand for closed metal bod- from a more distant supplier and the in-
ies grew dramatically, and by 1924 they creased cost of reheating cold ingots.
accounted for about two-thirds of General To protect against such behavior Essex
Motors’ automobile sales.6 entered into a long-term contract with
This unanticipated shift in demand in- Alcoa, in which Alcoa agreed to process
creased the extent by which the contract alumina into aluminum for Essex at spec-
forced General Motors to rely on Fisher ified output rates and to be paid in accord
and made it profitable for Fisher to take with a predetermined price formula. The
advantage of the contract to hold up Gen- long-term pricing formula chosen by
eral Motors. The large increase in demand Alcoa and Essex tied the price Essex
increased Fisher’s hold-up potential of would pay over time to the increase in the
General Motors so that it became greater wholesale price index for industrial com-
than the private sanction that could be modities.8 This prevented Alcoa from tak-
imposed on Fisher by the loss of new and
future sales to General Motors and to 7. Aluminum Co. @America v. Essex Group, Inc., 499
others in the marketplace that learned F.Supp. 53 (W.D. Pa. 1980). This case is discussed in
Speidel [1981] and Goldberg [1985].
about its behavior. If this large change in
8. The wholesale price index was chosen for this
demand had not occurred, the Fisher-Gen- contract by Townsend-Greenspan (Federal Reserve
Chairman Alan Greenspan‘sold consulting firm). The
same index was also chosen for the contract litigated
6 . Sixteenth Annual Report, General Motors Corpo- in Missouri Pub. Ser. Co. v. Peabody Coal Co., 583 S.W.
ration, year ended December 31, 1924. 2d 721 (Mo. App.), cert. denied, 444 US.865 (1979).KLEIN: SELF-ENFORCING RANGE 451
ing advantage of Essex by arbitrarily in- extent of unanticipated changes in market
creasing the price after Essex had made its events is measured along the horizontal
highly Alcoa-specific plant investment. axis by the deviation of market prices,
Unfortunately, the wholesale price P,, from contracted prices, Pc. For any
index which the parties agreed to use in deviation of market prices from contracted
their contract turned out to be a very poor prices, the resulting associated potential
measure to rely upon. Although the hold-up (by the transactor who has gained
wholesale price index had historically by enforcing the literal terms of the agree-
tracked Alcoa’s costs, electricity costs (the ment) is measured along the vertical axis.
principal non-labor cost in aluminum pro- Let us assume for expositional simplicity
duction) began to rise much more rapidly that the contractually specified flow of
than the wholesale price index after the goods implies that each $1 price deviation
1973 crude oil supply crisis. By June 1973 from the contract price along the horizon-
Essex was receiving aluminum from Alcoa tal axis creates a potential hold-up gain
at a net cost of less than one-half the with a present value of $1 million. For
contemporary market price of aluminum, example, if the market price rises above
resulting in what the judge asserted was the contract price by $1, the potential
an estimated windfall gain to Essex of hold-up gain to the buyer, Essex, of enforc-
more than $75 million over the life of the ing the literal terms of the agreement,
c~ntract.~ which we denote by HE, is $1 million; if
The enforcement by Essex of the literal the market price falls, say, $2 below the
terms of this imperfect contract can be contract price, the potential hold-up gain
considered a hold-up since it can be as- to the seller, Alcoa, of enforcing the con-
sumed to be contrary to the original intent tract agreement, denoted by HA, is $2 mil-
of the contractual understanding. Like the
lion. The potential hold-up gain as market
Fisher Body-General Motors contract, the
price deviates from contract price, there-
long-term contract designed to protect
fore, is represented in Figure 1 by the 45
Essex against a threatened expropriation
degree line HH.’O
of rents by Alcoa resulted, because of
The self-enforcing range of contractual
unanticipated changes in market condi-
performance is determined by considering
tions, in a much greater threatened shift of
the transacting parties’ private enforce-
rents to Essex from Alcoa. ment capital. Assume, for example, that
Figure 1 graphically illustrates the con-
Essex’s private enforcement capital, which
cept of the self-enforcing range of the we denote by K,, is $5 million (say $4
Alcoa-Essex contractual agreement. The
million from the capital depreciation of
Essex’s Alcoa-specific investments and
9. The actual cost to Alcoa was substantially $1 million from the future income loss to
higher. The judge calculated the loss to Alcoa by con-
sidering Alcoa’s accounting costs, including the cost Essex of operating in the marketplace with
to Alcoa of constructing the additional plant necessary a poorer reputation). Therefore, Alcoa
to fulfill the Essex contract, over the period 7977-87.
However, if we consider the opportunity cost to Alcoa could impose a $5 million cost on Essex if
by comparing what they could have sold the alumi- it holds up Alcoa by insisting on delivery
num for in the marketplace with the price at which
they were contractually bound to sell the aluminum at contract terms when the value of alumi-
to Essex, the amount is much higher. For example, in num has risen above contract terms. As a
1979, when Essex received aluminum from Alcoa
under the contract at thirty-six cents a pound, Essex
resold some of their aluminum in the open market at 10. We are assuming that the court will always en-
seventy-three cents a pound, for a difference of thirty- force the written contract terms, not in the sense that
seven cents a pound. Multiplying this underpricing the court would require specific performance, but that
by the seventy-five million pounds Alcoa was com- the court would award money damages to Essex (if
mitted to deliver to Essex annually yields an oppor- market prices increased) or to Alcoa (if market prices
tunity cost to Alcoa of nearly $30 million in 1979 alone. fell) based on the written contract terms.452 ECONOMIC INQUIRY
FIGURE 1
The Self-Enforcing Range of the Alcoa-Essex Contract
Essex's hold-up gain (HE) and
private enforcement capital (KE)
(million $)
"E
/
H
10
9
8
7
6
(KE) 5
4
3
2
I
1
1
1
(KA)
9
10
H / HA
Alcoa's hold-up gain (HA) and
private enforcement capital (KA)
(million $)KLEIN: SELF-ENFORCING RANGE 453
result, Essex cannot credibly insist on re- More generally, the contractual under-
ceiving the goods at the contracted price standing is not likely to require contrac-
as long as the market price deviates from tual adjustments of all deviations of the
the contract price by less than $5. For market price from the contracted price.
example, if the market price increased, say, Given costly information about changing
$3 from the contract price, the $3 million market conditions, it will be wasteful for
gain to Essex of such enforcement is less transactors to devote resources to search
than the $5 million loss to Essex from for information and negotiate changes for
termination of the relationship with Alcoa every small deviation of contract terms
and communication to the marketplace of from market conditions. Consider, for ex-
Essex’s failure to adjust contract terms to ample, a case of a contractor who, after
market conditions.” agreeing to build an additional room on
The analysis is symmetrical for the case your home for $20,000, informs you at the
of the potential Alcoa hold-up of Essex start of construction that the contract price
when the market price falls below the has to be adjusted upward from the agreed
contracted price. If Alcoa’s private en- upon price of $20,000.00 to $20,010.00 be-
forcement capital loss, KA, is, say, $8 mil- cause of a change in his cost of nails which
lion (consisting of, for example, a $2 mil- occurred in the two weeks since the con-
lion loss of Essex-specific investments and tract was bid, negotiated and agreed upon.
a $6 million loss of market reputation cap- You would, of course, not be aware of this
ital), the market price can, in principle, fall cost change nor would it be practical for
up to $8 below the contract price and the you to verify his claim. More importantly,
contract adjustment in price still be made you would, with good reason, wonder
by Alcoa. This is because of the now what kind of contractor you were dealing
credible threat Essex can make to cut its with-apparently one that intended to en-
losses by terminating its dealings with gage in significant rent dissipating negoti-
Alcoa and communicating to the market- ating activities during the life of the con-
place the breach by Alcoa of the contrac- tract. Analogously, we should not assume
tual understanding. The entire self-enforc- that Essex is holding up Alcoa if it en-
ing range, therefore, covers all market forces its contract with Alcoa when the
price deviations from the contract price market price exceeds the contracted price
between minus $8 and plus $5. Within this by a small amount. It is more likely that
range of price deviations, represented in Alcoa would suffer reputational penalties
Figure 1 by the flat portion of the if it seeks release from its contract with
SS schedule between minus $8 and plus Essex unless a small upward adjustment
$5, contract terms will be “voluntarily” in price were made. If contractual adjust-
adjusted to the market price without any ment is part of the implicit contractual
side payments being made by the transac- understanding, the understanding gener-
tors to one another. ally is that adjustments are not made un-
less some sufficiently large minimum dis-
turbance occurs.12The contract terms that
11. I am assuming that Alcoa can credibly threaten
to terminate Essex in spite of the fact that it is costly define the self-enforcing range can be
for Alcoa to carry out such a termination threat be-
cause they have made Essex-specific investments.
Alcoa can credibly threaten to terminate Essex when 12. Some contracts explicitly formalize this by in-
Essex threatens a hold-up within what Alcoa believed cluding reopener provisions, where the contract is
was the self-enforcing range either because of what opened for renegotiationafter some market price index
Alcoa learns about Essex (that Essex has lower private moves more than a minimum amount. See Goldberg
enforcement capital) or because of what Essex and and Erickson [19871. Crocker and Masten [1991] pro-
other buyers would learn about Alcoa if Alcoa failed vide a general discussion of the contractual mecha-
to terminate Essex (that Alcoa has higher costs of im- nisms employed by transactors to flexibly adjust prices
posing the termination sanction). in long-term contracts.454 ECONOMIC INQUIRY
thought of as a ”contractual constitution” presence of Essex private enforcement
that is not anticipated to be frequently capital, Alcoa can impose a $5 million loss
arnended.I3 on Essex. In the case of ( P , - P,) equal to
Given the constitutional contractual un- $6, if Alcoa and the market consider pay-
derstanding of the parties and the time ment to Essex of any settlement greater
necessary to negotiate contractual than $1 million as a hold-up by Essex (i.e.,
changes, transactors can find themselves as a breach of the implicit contractual un-
outside of the self-enforcing range if sur- derstanding regarding settlements), then
prises take place-that is, if large and we can assume that Alcoa will be able to
sudden unanticipated changes occur in credibly impose the $5 million loss on
market conditions. When this occurs, Essex and Essex will willingly accept only
a transactor’s hold-up potential is greater $1 million to adjust the contract price up
than its private enforcement capital and $6.00 before continuing the business rela-
threats to breach the contractual under- tionship. The hold-up settlement payment
standing and enforce the literal terms of schedule is, therefore, represented in Fig-
the contract are credible. For example, if ure 1 by the schedule SS.15
the positive deviation between market This analysis suggests that when par-
and contract prices is greater than the $5 ties enter a contractual relationship they
given by Essex’s reputation capital, Essex can be thought of as buying what amounts
can credibly threaten Alcoa with litigation to an option representing the probability
to enforce the literal terms of the contract of a hold-up occurring. In particular, Essex
if it does not receive a side payment in has purchased a call and has written a put,
return for modifying the contractual ar- while Alcoa has purchased a put and has
rangement to coincide with market written a call. The defining points of the
prices.14 self-enforcing range can be thought of as
The magnitude of the necessary side the exercise prices of these put and call
payment settlement will be less than the options that Essex and Alcoa have written
potential transactor gain represented by and purchased along with their contract.
the H H schedule in Figure 1. For example, As in standard options pricing theory,16
if market prices move above contract the values of these options increase (1)as
prices so that (P, - P,) is $6, Alcoa need the value of the ratio of the underlying
not pay Essex $6 million to force Essex to asset price increases relative to the exer-
adjust the contract price up to the market cise price (in our case, as the value of the
price. In the real world we do not observe hold-up potential increases relative to the
discontinuous behavior such as no side private enforcement capital), and (2) as
payment being made when the price devi- the variance per period of the asset price
ation is $5.00, but a more than $5 million multiplied by the number of periods in-
side payment being made when the price creases (in our case, as the variance of
deviation is, say, $5.10. Because of the underlying market conditions and the
length of the contract increases). There-
fore, because hold-ups are costly, when the
13. Goldberg [ 1976,4261 has argued that it is useful variance of (P,-P,) is high, transactors
to think of transactors designing a contractual arrange-
ment as establishing a “constitution”to govern their
ongoing relationship.
14. A relationship outside the self-enforcing range
does not imply that litigation occurs. For litigation to 15. I am assuming for simplicity throughout the
take place it is necessary, in addition to the parties discussion a threshold model of private sanctions, that
being outside the self-enforcing range, for the parties is, any and all types of hold-ups trigger the same lump-
to have sufficient informational differences regarding sum private enforcement penalty.
what they have at stake and what their probabilities 16. See, for example, Brealey and Myers [1991],
of success in court are. chapter 20.KLEIN: SELF-ENFORCING RANGE 455
will require more reputational capital or ered as alternatives-firms will rely upon
reduce the length of their contract or the one or the other, but never both. Principal-
specific investments they make. The major agent models, for example, formulate the
difference between option analysis and contracting problem as if transactors do
our hold-up analysis is that by writing not possess any private enforcement cap-
particular contract terms transactors not ital. Therefore, it is not surprising that
only can vary the exercise price but, as we these models have limited predictive
shall see, they also can vary the underly- value in explaining real world contract
ing probability distribution that deter- terms. On the other hand, standard eco-
mines the value of their options. nomic models of reputational enforcement
provide no role for contractual specifica-
V. THE ROLE OF CONTRACT TERMS tion.**However, given the fact that private
The role of contract terms within this enforcement capital is limited, transactors
framework is very different from the stan- can be expected to use written contract
dard economic view of contract terms. The terms and, hence, the assistance of the
view of contract terms that underlies court, as a supplement to private enforce-
much of the principal-agent and mecha- ment.
nism design literature, for example, is that Unlike standard economic models, the
contract terms are used to create optimal probabilistic hold-up framework pre-
incentives on some court-enforceable sented here implies a fundamental com-
proxy for performance. Optimal but not plementarity between court enforcement
perfect incentives are created by contract and private enforcement. When employed
terms because the terms are only imperfect together the mechanisms are substitutes in
proxies for performance and ate assumed demand in the sense of a positive cross-
price effect, i.e., an increase in the price of
to represent the sole elements of perfor-
mance against which transactors maxi- one increases the demand for the other.
mize.I7 The problems that arose in the For example, an increase in the cost of
Fisher-General Motors case may be con- court enforcement increases investments
in private enforcement capital. However,
sidered as an example of the type of im-
the two enforcement mechanisms are com-
perfect incentives and associated ineffi-
ciencies created when imperfect contract plements in production in the sense of a
terms are used. However, it is a mystery positive cross effect in production, i.e., an
increase in the quantity of one increases
within this standard framework why
Fisher and General Motors would have the marginal product of the other. The two
enforcement mechanisms work better to-
considered it optimal to choose such
clearly imperfect contract terms to begin gether than either of them do separately.
Within this framework transactors use
with.
The problem with the standard eco- written contract terms not solely to create
nomic framework is that court enforce- an incentive to perform with regard to
ment and private enforcement are consid- some court-enforced, contractually speci-
fied proxy for performance. Rather, trans-
actors use written contract terms to define
optimally the self-enforcing range of their
17. A survey of the principal-agent literature is
provided in Hart and Holmstrom 119871. The com- contractual understanding. The goal of
plex, contractually specified incentive schemes that contractual specification is to economize
solve the agency problem in this literature are also
generally claimed to be only second-best because of on the amount of private enforcement
the presence of transactor risk aversion, which creates
a tension between the effect of a contract term in op-
timally rewarding productive work and in shifting un- 18. See Kreps [19901for a discussion of reputational
wanted risk to the agent. enforcement models in noncooperative game theory.456 ECONOMIC INQUIRY
capital necessary to make a contractual right of K G defines the probability of a
relationship self-enforcing by merely “get- General Motors hold-up, the area to the
ting close” to desired performance in a left of KF defines the probability of a Fisher
wide variety of circumstances (without Body hold-up, and the area between KF
creating undue rigidity) and to let the and K G is the probability that the General
threat of private enforcement move per-
formance the remainder of the way to the Motors-Fisher relationship remains within
desired level. the self-enforcing range.
Contract terms can accomplish this goal Panel A of Figure 2 represents the situ-
of economizing on private enforcement ation after Fisher has made its General
capital in two fundamental ways. First of Motors specific investment and a signifi-
all, contract terms can operate directly to cant General Motors hold-up potential of
control nonperformance. By defining per- Fisher has been created. Panel B repre-
formance with explicit court-enforceable sents the situation after Fisher and Gen-
contract terms, such as the quantity, qual- eral Motors have negotiated a contractual
ity and price of a product that must be arrangement which attempts to control
delivered, transactors control hold-up be- the hold-up potentials of the parties. As-
havior by legally “tying their hands” with suming that the rent-dissipating costs as-
regard to variables that can be manipu- sociated with hold-ups are proportion-
lated to expropriate rents from a transact- ately related to the magnitude of the hold-
ing partner. In the Fisher-General Motors up, transactors will attempt to minimize
case, for example, these contractual re- these costs when setting contract terms by
straints took the form of an exclusive minimizing the expected value of the
dealing clause with a specified price for- hold-up, i.e., the sum of the expected
mula. hold-up values associated with the tails of
This is illustrated in Figure 2, which the probability distribution. Because the
presents the probability distribution of the actual hold-up is assumed to be adjusted
hold-up potential in the General Motors downward by the private enforcement
and Fisher Body relationship,f (HI, which sanction that can be imposed on the trans-
is assumed for expositional simplicity to actor engaging in the hold-up, General
be related to some scalar measure of Motors and Fisher can be assumed to be
ex post market conditions that can be minimizing
measured along the horizontal axis, as in m KF
the Alcoa-Essex case.I9 Figure 2 measures
General Motors’ hold-up potential and
(1) I (H- KG)f(H)dH J (H- KF)f(H)dH*
-a
K,
private enforcement capital to the right of
zero and Fisher’s hold-up potential and
private enforcement capital to the left of Panel B illustrates that, although the
zero. (As illustrated, there is no reason contract terms substantially reduce the
that the probability distribution need be probability that General Motors will
centered on zero.) The shaded area to the hold up Fisher for its specific investment,
the contract terms also substantially in-
crease the probability of a very large
19. In reality there is unlikely to be market real-
izations that correspond to a single unique hold-up Fisher hold-up of General Motors if mar-
potential such that an increase in one transactor’s ket conditions change dramatically. The
hold-up potential necessarily implies a corresponding
decrease in the other transactor’s hold-up potential. contract decreases the probability of being
Therefore, one may have to model the situation with outside the self-enforcing range, but also
a separate probability distribution for each transactor. increases the far tails of the hold-up dis-
However, the formulation presented here does illus-
trate the fundamental economic forces at work. tribution. This corresponds to the rigidityKLEIN: SELF-ENFORCING RANGE 457
FIGURE 2
The Hold-Up Probability Distribution in the Fisher-General Motors Case
A. Fisher makes a General Motors specific investment
0
i
i!
B. Contract terms are set458 ECONOMIC INQUIRY costs associated with literal court enforce- transactors’ hold-up potentials under ment of imperfect contract terms dis- likely future conditions. By more closely cussed above. It is because of these rigid- relating actual private enforcement capital ity costs associated with contractual spec- to likely requirements, transactors widen ification that, after some point, each con- the ex post market conditions that are tract term which the transactors decide to likely to fall within the range where per- use involves a cost/benefit calculation. We formance remains assured. can expect, therefore, that the degree of This view of contract terms, as a way to contractual specification will be lower the increase the effectiveness of a self-enforce- greater the private enforcement capital ment mechanism by shifting private en- possessed by the transactors. forcement capital between transactors, has Where private enforcement capital is much greater predictive power in explain- larger, contracts will be ”thinner,” with ing the contract terms we observe in the transactors writing out only the essential world than the standard view of contract elements of the agreement, or perhaps terms. For example, consider the grant by even proceeding on the basis of a verbal a franchisor to a franchisee of an exclusive understanding and a handshake; where territory. The standard economic view of private enforcement capital is smaller, contract terms would emphasize the effect written contracts will be ”thicker,” with of the exclusive territory in creating the transactors attempting to specify more el- correct incentive on the franchisee to per- ements of performance and provide for form due to the increased customer repeat more contingencies. For example, con- sale created by the exclusive territory. The tracts between Japanese companies can be exclusive territory thereby reduces the expected to be much less completely spec- franchisee’s incentive to ”free ride” on ified than a similar contractual relation- other franchisees. However, the exclusive ship between U.S.companies. Japanese territory more importantly also creates a companies generally possess large franchisee premium stream and, therefore, amounts of private enforcement capital in gives the franchisee something valuable to transactions with one another because of lose if it is terminated by the franchisor for the significant sharing of information re- non-performance.20 garding performance among Japanese The exclusive territory and the associ- companies and the large potential ”loss of ated payment of a premium stream from face” in the Japanese marketplace if it is the franchisor to the franchisee can be perceived that one has engaged in a hold- thought of within our private enforcement up. When a large amount of private en- framework as a shift of private enforce- forcement capital is present, there is a ment capital from the franchisor to the lower likelihood of being outside the self- franchisee. The franchisee now has more enforcing range and, therefore, less need to lose if it is terminated, namely the to bear the costs associated with contrac- franchisee loses the discounted value of tual specification. the expected premium stream associated The second fundamental way in which with the exclusive territory; and the contract terms can reduce the expected franchisor now has less to lose if it un- hold-up probability is by shifting private fairly terminates the franchisee, namely enforcement capital between transactors. the franchisor saves the discounted value Rather than directly attempting to reduce of the larger expected premium stream the hold-up potential, contract terms can that it no longer has to pay the fran- shift the location of the self-enforcing range so that private enforcement capital coincides more accurately with the 20. See Klein and Murphy [1988] and Klein [1995].
KLEIN: SELF-ENFORCINGRANGE 459
chisee.21This shift in private enforcement (3) 1 - F(K, - X) = F ( K , + x).
capital, because it more accurately aligns
the transacting parties' enforcement capi-
tal with likely future franchisee hold-up Intuitively, if the probabilities are unequal,
possibilities, expands the self-enforcing the overall probability of a hold-up occur-
range. That is, the exclusive territory in- ring in the relationship could be reduced
creases the probability that ex post market by shifting private enforcement capital
conditions will fall within the range where from the transactor with the lower proba-
performance can be privately enforced. bility to the transactor with the higher
This example is illustrated in Figure 3 probability.22
where, once again, we assume there is a Another example of contract terms effi-
market realization of a single hold-up po- ciently shifting the self-enforcing range is
tential, with H , and H,. representing the the use of a contract to determine which
franchisee and franchisor hold-up poten- transactor makes the transaction-specific
tial and K , and K,. representing the fran- investment. In general, the transactor who
will make the specific investment is deter-
chisee and franchisor private enforcement mined by comparing the likely future pri-
capital, respectively. vate enforcement capital requirements
If we assume that the shifts in private
(i.e., the hold-up potentials) of each party
enforcement capital from the franchisor to
under alternative likely contingencies
the franchisee, represented by x in Fig-
with the amount of private enforcement
ure 3, are dollar for dollar, in the sense capital that each transactor has available.
that every dollar increase in the franchisor Therefore, it is usually the transactor with
hold-up (in every ex post state where a the smaller private enforcement capital,
franchisor hold-up occurs) implies a cor- such as Fisher Body, that will make the
responding saving of one dollar in the specific investment. The larger firm, be-
franchisee hold-up (in every ex post state cause of its increased repeat transaction
where a franchisee hold-up occurs), then frequency, generally has more private en-
this process of shifting private enforce- forcement capital and hence increased
ment capital will occur until credibility that it will fulfill the contract.
Similar reasoning explains why many
contracts may appear "one-sided" or "un-
K, + x
+5 --a
( H - K, + x)f(H)dH = 0. 22. This result assumes that the real costs associ-
ated with hold-ups are related to the magnitude of the
hold-up and that this relationship is the same for both
transactors. For example, the real costs may be pro-
This implies that in equilibrium the prob- portional to the magnitude of the expected hold-up
ability of engaging in a hold-up will be with both transactors having the same proportionality
constant. This will not be the case, however, if any
the same for the franchisor and the fran- franchisee hold-up gains entail primarily distribution
chisee, or effects with relatively little real costs compared to
franchisor hold-up gains of the same magnitude. For
example, a franchisee hold-up may entail some costs
associated with loss of product reputation on the part
of consumers and the cost of replacing the franchisee,
21. More completely, the franchisor's loss from ter- while a franchisor hold-up may entail much larger
minating franchisees is the discounted value of its cost costs associated with the efficiency of the franchising
advantage of running the operation as a franchise ar- arrangement compared to the next best alternative mar-
rangement compared to the next most efficient form keting arrangement. Since contract terms are set to
of operation from which is netted out the premium minimize real costs (and not hold-ups), this would
stream the franchisor must pay the franchisee to assure imply a higher probability of a franchisee hold-up than
franchisee performance. See Klein (19951. a franchisor hold-up in equilibrium.460 ECONOMIC INQUIRY
FIGURE 3
The Shift of Private Enforcement Capital from a Franchisor to a Franchisee
Y
fair.”23 For example, if General Motors additional costs on the firm), it may pay
possessed substantially more private en- both parties to use a termination-at-will
forcement capital than its suppliers, it contract. Although such a contract may
could avoid the rigidity associated with seem unfair, with the employee vulnerable
contractual specification by not promising to a potential hold-up by the employer,
anything in writing in its supply contracts. one cannot interpret such arrangements
The contracts would appear to be one- without recognizing that private enforce-
sided, but this would not substantially ment capital, in addition to the explicit
increase the probability that it would contract terms, also governs the relation-
hold up its suppliers. Another example is ship and that the employer may have
employee termination-at-will clauses. Be- sufficient private enforcement capital to
cause it is extremely difficult to specify in define a sufficiently broad self-enforcing
a court-measurable way all the conditions range.
of adequate employee performance, and
court enforcement of imperfect terms en- VI. A PROBABILISTIC VIEW OF HOLD-UPS
tails all the rigidity costs discussed above This analysis implies that contractual
in addition to the costs of artificial record arrangements can only be understood if
keeping, litigation expense and time delay we recognize that transactors optimally
(during which the employee may impose design their contracts to combine court-
enforced written contract terms with self-
23. See Klein [1980]. enforced unwritten terms. Given the par-KLEIN: SELF-ENFORCING RANGE 46 1
ticular contract terms they choose and the if employees sometimes take pencils home
private enforcement capital they possess, for their personal use, the pencils are part
transactors expect the relationship to re- of the wage and working conditions of the
main within the self-enforcing range, job. (And if the value of the pencils to the
where market conditions can change and employees exceeds their cost and employ-
the parties will perform as intended. How- ers do not expend resources to prevent the
ever, transactors also know at the time taking of them, there is nothing inefficient
they enter their contractual relationships about this form of compensation.)
and make their specific investments that More importantly, identifying moral
their private enforcement capital is lim- hazard behavior with a hold-up blurs a
ited, that their written contract is imper- fundamental analytical distinction. Con-
fect and incomplete and, therefore, that sider the example of the demand for med-
there is some probability of a hold-up oc- ical services by individuals with health
curring. In particular, transactors know insurance, a commonly cited case where
that there is some probability that market moral hazard behavior takes place. Health
conditions may change sufficiently (and insurers who, after writing the best con-
the value of the quasi-rents accruing to tracts they can, knowingly accept the fact
one of the parties increase sufficiently) so that their policyholders will take advan-
that one party will find it in its interest to tage of the low marginal price in these
engage in a hold-up. contracts to increase their demand for
For example, the unanticipated change health services. In spite of this behavior,
in market conditions that occurred in the the transacting parties still find it in their
Alcoa-Essex case was the 1973 crude oil interests to enter the relationship. It is true
“shock” which led to a quadrupling of the that the moral hazard behavior is ”non-
price of aluminum. This was a contin- performance,” in the abstract, ideal sense
gency that was not covered in the Alcoa- that if sufficient private enforcement cap-
Essex contract and once this unanticipated ital existed or perfect contract terms could
event occurred, the short-run gain to Essex be written, the behavior would not exist.
from the failure to adjust the price became However, the behavior is fully expected
greater than the depreciation of Essex’s and, I would maintain, has nothing to do
private enforcement capital. It became with a hold-up.
profitable for Essex to violate the intent of A hold-up is a particular kind of trans-
the contractual understanding by de- actor non-performance distinct from
manding enforcement of the contract as moral hazard behavior. Specifically, a
written. The contractual relationship had hold-up, as opposed to moral hazard be-
moved outside the self-enforcing range. havior, requires unanticipated events.
This probabilistic view of hold-ups Transactors may recognize that there is a
should be contrasted with Williamson’s positive probability of a hold-up occurring
view of opportunism which, he asserts, is and may know ahead of time that if par-
equivalent to moral hazard behavior.24 ticular conditions develop, a hold-up will
First of all, if moral hazard behavior is take place. However, hold-ups are always
fully anticipated, it should be considered surprises in the sense that the particular
as merely part of the price. For example, conditions that will lead to the hold-up are
considered unlikely and, because it is
costly to negotiate and specify contract
24. Williamson [1985,51, n. 81. Considering moral
hazard behavior as opportunism is inconsistent with terms, these unlikely conditions are not
Williamson’s definition of opportunism in terms of de- taken account of in the contract. If the
ception and guile (fn. 1 supra). Williamson attempts transactor being held up had expected
to reconcile this inconsistency by claiming that one
should consider deception broadly. that market conditions would develop toYou can also read
