A COOPERATIVE GAME THEORY SOLUTION

IN AN UPSTREAM-DOWNSTREAM RELATIONSHIP
MARCHI, Ezio
Instituto de Mte!"ti# A$%i#d Sn Luis A&'entin, Uni(e&sidd
N#ion% de Sn Luis
COHEN, Pu% And&e
Uni(e&sitt Aut)no! de *&#e%on
GARCIA-CESTONA, Mi'ue%

Uni(e&sitt Aut)no! de *&#e%on

A+st&#t
What happens when a firm does not want to be vertically integrated? Within a
buyer-supplier relationship, and in a context of specific investments, we analyze
firm interactions when they decide to cooperate instead of becoming vertically
integrated. We develop a model that presents an alternative to vertical
integration with firms that cooperate with each other under incomplete
contracts. The contribution of our approach is twofold: first, a mathematical
component, where we show how to apply the maximin to compute the
characteristic functions of the model and second, an economic element, where
we reconsider the upstream-downstream relationship under a cooperative
framewor! and no integration. We find that the total generated value is "areto
optimum if firms cooperate and distribute the benefits following the #hapley
value solution. $inally, the #hapley value measures the power that each firm
has in the bargaining process.
,EYWORDS- cooperative game, #hapley value, core, incomplete contracts,
specific investments, upstream-downstream relations, characteristic functions.
AC,NOWLEDGMENTS
%. &archi is indebted to "rofessor 'enaro (. 'utierrez who invited him and encouraged him to wor! in
applications of game theory into economic fields. )e also appreciates some material provided by
"rofessor 'utierrez. The authors gratefully ac!nowledge financial support from the #panish &inistry of
%ducation and #cience *#%(+,,--.-/01-234-,+5.

Co&&es$ondin' ut.o&/
1
INTRODUCTION
The theory of the firm deals with the boundaries of a firm and it also focuses on
the coordination and motivation problems faced by the different participants.
The first definitions of a firm were economic models that considered the firm as
a blac! box where physical inputs and labor came out in an output, at minimum
cost and maximum profit *)art, 60015. This definition had important limitations
that gave place to new theories of the firm. 2oase *607-5 is recognized as the
first researcher to provide an explanation about the existence of firms. )e
suggested that there were mar!et imperfections that generated transaction
costs and, in this context, the presence of firms may help to alleviate these
mar!et failures. 8n particular, this author argued that, under some
circumstances, transactions could be done through an organization at a lower
cost rather than through mar!ets. 9ater, in 60-+, :lchian and ;emsetz defined
the firm <as a nexus of contracts= and they also provided a description of the
<classical capitalist firm=.
8n 60-1 Williamson too! the concept of transaction costs given by 2oase, and
he elaborated it further. 8n fact, Williamson redefined the concept of transaction
costs and applied it to different forms of organizations, not >ust the mar!et,
ma!ing a substantial progress in the study of the firm. )is most important
contribution was to elucidate the roles of imperfect contracts and specific
investments *asset specificity, human specificity5. :nother important
contribution was the fact of comparing the transactional efficiency of alternative
governance structures, including vertical integration, non-standard contracts
and relational contracts *'ibbons, +,,,5
Williamson?s wor!s opened the door to new investigations based on the
ownership of the physical assets. Within this group, one can find the seminal
wor! of 'rossman and )art *60/.5, who defined the firm as <a collection of
physical assets that are >ointly owned= *@ingales, 600/5.
:lthough the ownership of physical assets has been used to explain the
boundaries of the firm and vertical integration serves as a mechanism to save
transaction costs, another important problem remains: the specificity of human
2
assets. The problem of specific human assets in a context of vertical
integration was already treated by Alein et al in 60-/, at the time of describing
the problems between 'eneral &otors *'&5 and $isher Body *$B5. )uman
assets cannot be acCuired because they are inherent to human beings. Alein
revisited the vertical integration between 'eneral &otors and $isher Body and
argued that the costs associated with vertical integration are generally incentive
costs that are unrelated to the degree of specific investments. Therefore,
vertical integration becomes the most plausible solution when the degree of
specific investments becomes high *Alein, 60//5. Dertical integration avoids
transaction costs when the costs are associated with physical assets, but it
does not explain what happens when vertical integration is lin!ed to specific
human assets.
Ea>an and @ingales *+,,,5 indicate that 'eneral &otors was, and still is, a
vertically integrated firm, which controls the physical assets through ownership.
)owever, they argue the nature of the firms is changing. 9arge conglomerates
have been bro!en up and their units have been spun-off as stand-alone
companies. 3ne example is Fucor, a steel manufacturer that abandoned the
tradition of bac!ward integration and out-sourced the entire supply chain of raw
material *)olmstrGm and Eoberts, 600/5.
:s a result, vertical integration does not seem to be a good solution any more
under certain circumstances. #o we wonder if there is another way, different
from vertical integration, to solve the problems. 8n particular, we analyze what
happens when some firms decide to cooperate instead of being vertically
integrated in a scenario with the presence of specific investments. &ore
specifically, what happens if a supplier does not accept to be vertically
integrated and it prefers, instead, to deal with several customers? To answer
these Cuestions, we develop a model that attempts to bring a new alternative
solution to the problem, a model where the firms negotiate and cooperate to
each other in line with the observed practices of many companies.
The contribution of this model is twofold: first, a mathematical contribution
where we show how to apply the maximin that is included in the &inimax
3
Theorem *see &archi, 60.-5 to compute the characteristic functions of the
model and second, an economic contribution to solve a supplier-manufacturer
relationship. ;oing this, we present an alternative way to loo! at the upstream-
downstream relationships under a cooperative framewor! and incomplete
contracts.
We find that, under cooperation, the total value generated is "areto optimum
when they cooperate and distribute the benefits following the #hapley value
solution. 8n fact, the #hapley value measures the power that the different
participants have in the bargaining process. The bargaining power under
cooperative game theory is determined by which player is needed the most
*Brandenburger, +,,-5. 8n our model, the bargaining power is given by the
specificity of the investment. The result of our model is that the supplier?s
bargaining power is what ma!es the cooperation between firms possible,
ma!ing such upstream-downstream relationship enforceable. :nd, as a
conseCuence, the first best that maximizes the total welfare can be achieved.
This paper proceeds as follows: in section 6, we present the literature review,
while in section +, the economic environment of the model is described. We
develop a model under an incomplete and cooperative framewor!, being the
core and the #hapley value the solutions to such model. $inally, in section 7, we
present the conclusions of the model, followed by a description of some
limitations and future research avenues.
0- LITERATURE REVIEW
The firm, as a blac! box where physical inputs and labor came out in an output,
at minimum cost or maximum profit, has been the basic model of a firm for two
hundred years. )owever, this theory presented important limitations because it
ta!es into account neither the incentives inside the firm, nor its internal
structure. 8n addition, it did not establish the boundaries of the firm. These
limitations gave place to the appearance of new theories of the firm. 2oase
*607-5 was the first researcher to give an explanation about the existence of the
4
firms. )e suggested that the mar!ets were imperfect and firms existed to solve
these mar!et failures. 2oase also dealt with the internal organization and
suggested that firms would exist only in an environment in which they could
perform better than mar!ets *'ibbons, +,,,5. $urthermore, he argued that,
under some circumstances, transactions can be made through the organization
at a lower cost rather than through mar!ets. This is so because the
organizations ameliorate some mar!et failures and they have the advantage of
using employment contracts as a way to save transaction costs.
9ater, in 60-+, :lchian and ;emsetz defined a firm <as a nexus of contracts=.
They said a firm is characterized by something more than legal status, namely
the technology of team production, which means production with an inseparable
production function. Thus, they introduced the free-riding problem in team
productions. Their solution was to propose the presence of a monitor, who had
the right to fire and hire the members of the team and, by doing that, they
described the <classical capitalist firm=
8n 60-1 Williamson too! the concept of transaction costs given by 2oase and
redefined it. 8n fact, Williamson introduced the transaction cost approach to the
study of organizations, suggesting a new way of thin!ing about the firm and
focusing on the analysis of transactions and contracts. 8n particular, he said that
the transaction costs approach relies on two assumptions concerning the
individuals: 65 human agents are sub>ect to bounded rationality, and +5 at least
some agents may behave in opportunistic way *Williamson, 60/65. $urthermore,
the characteristics of the transactions, such as their freCuency, the presence of
specific assets and the presence of information asymmetries may be important
sources of transaction costs. %conomic exchanges could be efficiently
organized by contracts but, due to bounded rationality, the nature of the
transactions and the opportunistic behavior of the individuals, contracts are
often incomplete.
Williamson made a substantial progress in understanding the nature of the firm,
and one !ey contribution was to elucidate the roles of imperfect contracts and
specific investments *asset specificity and human specificity5. :sset specificity
5
is critical because, once the investment is made for a determined transaction
between a buyer and a supplier, the value of this asset becomes lower if it used
in a different transaction. Then, the supplier is <loc!ed into= the transaction
*Williamson, 60/65. This also applies to a buyer when the buyer cannot ma!e
transactions with others suppliers.
:nother important contribution was to compare the transactional efficiency of
alternative governance structures, including vertical integration, non-standard
contracts and relational contracts *'ibbons, +,,,5. :n example of vertical
integration was treated by Alein et al in 60-/, explaining the relationship
between 'eneral &otors *'&5 and $isher Body *$B5. 'eneral &otors signed a
contract with $isher Body in which '& argued to purchase all its closed bodies
from $B for 6, years in 6060, and they agreed on a price for delivering based
on a cost plus a margin, that also included provisions that '& would not be
charged more than other rival automobile manufacturers. :s it is well-!nown,
'& faced a much higher demand than the forecasted one, increasing its
dependence on $B. :s a response to this situation, '& was unsatisfied with the
price settled and proposed $B to locate its body plants near the '& assembly
plants, an offer that $B refused. :s a conseCuence, '& started to acCuire $B
stoc! in 60+4 and, finally, it completed a merger agreement in 60+.. : '& car
needed specific investments, and site specificity implied transaction costs. This
vertical integration between '& and $B solved the problems associated with
the presence of uncertainty in demand and costs.
2handler *60--5 and "orter and 9ivesay *60-65 analyzed forward integration
into distribution activities for the case of manufacturing firms. 8n retailing, vertical
integration was made for those commodities that needed a considerable
number of sale information points. 8n this context, specific human assets were
needed to provide service and achieve larger sales, and the integration into
wholesaling occurred for those commodities that were perishable and branded.
The explanation for the existence of forward integration is that the
manufacturer?s reputation was at ris!: contracts to turn on inventories or to
destroy stoc!s were neither self-enforcing nor incentive compatible. :s
Williamson *60/65 explained <the commodities that had none of these
6
characteristics *perishable and branded5 were sold through mar!et distribution
channels where no special hazards were posed=.
Williamson?s analysis encouraged further research based on the ownership of
physical assets. The seminal wor! of 'rossman and )art *60/.5, who defined
the firm as <a collection of physical assets that are >ointly owned=, is based on
Williamson?s approach, and along with &oore *600,5 these authors developed a
theory of incomplete contracts and property rights. There, it is this notion of
property rights what determines the right to ta!e decisions concerning all the
issues that are not explicitly covered in the contract *@ingales, 600/5.
:fter this, when explaining the boundaries of the firm, most researchers have
assumed that the ownership of physical assets confers the right to ta!e
decisions. This means that the ownership of physical assets has been used as
an incentive mechanism or a control mechanism to induce the agent to ma!e
the optimal effort or to carry out the optimal investment.
3ur point is that even when ownership of physical assets is used to explain the
boundaries of the firm and vertical integration can save transaction costs,
another problem still remains: the specificity of human assets. :s Alein *60//5
pointed out, human assets cannot be acCuired because they are inherent to a
human being. 8n fact, this author revisited the 'eneral &otors and $isher Body
case and argued that the costs associated with vertical integration are generally
incentive costs unrelated with the degree of the specific investments. Therefore,
the vertical integration is the most possible solution while the degree of specific
investment is high *Alein, 60//5. #o, vertical integration avoids the transaction
costs when the costs are associated with physical assets but, what happens
when vertical integration occurs in a context with specific human assets? 8n this
case, vertical integration will not solve the transaction costs problem because
the specific human capital is inherent to those human beings and it cannot be
owned by a third party. Thus, the opportunistic behavior present in the original
organizational arrangement is not eliminated and the benefits of a vertical
integration process remain unclear.
7
8n fact, if the conflict between 'eneral &otors and $isher Body was only based
on a hold-up problem of the investments in physical assets and specific human
assets did not matter, the best solution would have been the acCuisition by '&
of the $isher Body?s physical assets. But vertical integration, meaning that the
$isher brothers become employees instead of independent contractors, does
not eliminate the potential hold-up. 3wnership of the specific human asset was
still belonging to the $isher brothers. #o, as Alein pointed out, the solution was
that 'eneral &otors did not buy the closed bodies from $isher Body, but
'eneral &otors had to produce the closed bodies with the assistance of $isher
Body. 'eneral &otors was the owner of the plants and could determine where
the plants were located, but the $isher brothers became managers with the
capacity to control or hold up '& with regard to specific investments in human
assets.
This vertical integration implied that 'eneral &otors transformed not only the
$isher brothers into employees but also all the employees of $isher Body into
'& employees. 'eneral &otor stopped buying to start producing, acCuiring in
this way the organizational capital of $isher Body 2orporation. '& became the
owner of all employment contracts and also the owner of the !nowledge
concerning the production of closed bodies. 8t is in this way that 'eneral &otors
owned the specific human asset *Alein, 60//5.
$urthermore, this solution was a successful one because the number of
employees was huge. 8f few employees are involved, they can leave the firm
and the organizational capital will be gone with them. But the threat that all the
employees will leave the firm is not credible when the number of employees is
so huge. :fter the vertical integration process, the $isher brothers could not
threat 'eneral &otors saying that all their former employees would leave the
firm *ta!ing their !nowledge with them5. 8n this sense, vertical integration in
organizations with large human teams implies the ownership of the human
asset as it happens with physical assets. 8t is in this sense that Alein mentions
that specific human capital can be acCuired and that the vertical integration
could be a solution under some circumstances.
8
Ea>an and @ingales *+,,,5 indicate that 'eneral &otors was, and still is, a
vertically integrated firm, that controls the physical assets through ownership.
)owever, as these authors point out, the nature of many firms is changing:
<9arge conglomerates have been bro!en up and their units have been spun-off
as stand-alone companies. Dertically integrated manufacturers have
relinCuished direct control of their suppliers and have moved toward looser
forms of collaboration= *Ea>an and @ingales, +,,,5. 3ne example of this is
Fucor, a steel manufacturer that abandoned the tradition of bac!ward
integration and out-sourced the entire supply chain of raw material *)olmstrGm
and Eoberts, 600/5. %ven 'eneral &otors is changing and it has facilitated the
spin off of its ma>or part supplier, ;elphi. Toyota, and the relationship with its
suppliers, provides another example of new forms of organization to deal with
specific assets. 8n fact, :o!i *600,5 tal!s of Cuasi-integration to refer to these
lin!s between Toyota and its suppliers which remain independent legal entities
but exchange information and closely cooperate in the production process and
the development of new products.
8n summary, vertical integration seems not to be a good solution under some
circumstances, as many researchers have already pointed out. This fact ma!es
us wonder if there is another way to solve these problems that vertical
integration does not address. 8n particular, we as! ourselves: what happens if
the firms decide to cooperate instead of being vertically integrated under the
existence of specific investments? Additionally, what happens if the supply firm
refuses to be vertically integrated? To answer these Cuestions, we develop a
model attempting to bring a new alternative solution to vertical integration. We
propose a model where the firms cooperate with each other and we find that the
total value generated, under cooperation, is "areto optimum if they distribute
the benefits following the #hapley value solution.
We proceed now to mention briefly some important features of the 2ooperative
'ame theory, before describing the model.
9
0/0- COOPERATIVE GAME THEORY
The game theory of an arbitrary set of players or agents is created by (ohn von
Feumann and 3s!ar &orgenstern in 6044
6
. The game theory is the study of
games, also called strategic situations *#errano, +,,-5. The game theory is
divided in two main approaches: the cooperative and the non-cooperative game
theory. The actors in non-cooperative game theory are individual players who
may reach agreements only if they are self-enforcing, while in cooperative game
theory, the actors are coalitions, group of players. :s #errano *+,,-5 points out,
the fact that a coalition has formed and that it has a feasible set of payoffs
available to its members is now ta!en as given.
These two approaches of game theory imply two different forms to loo! at the
same problem. :s :umann *60105 puts it, <the game is one ideal and the
cooperative and non cooperative approaches are two shadows=. 'ame theory
models imply situations in which players ma!e decisions to maximize their own
utility, while the rest of the players do the same. The decisions of the latter
affect each other utilities. 2ooperative game theory loo!s for the possible set of
outcomes, study what the players can achieve, which coalitions will be formed,
how the coalitions will distribute the outcomes and whether the outcomes are
robust and stable *#osic and Fagara>an, +,,.5.
The terms of non-cooperative game theory may, mista!enly, suggest that there
is no place for cooperation, and the term of cooperative game theory might
suggest that there is no room for conflict or competition. But as Brandenburger
*+,,-5 has already pointed out, neither is the case. "art of the non-cooperative
game theory studies the possibility of cooperation in ongoing relationships,
while many papers that use cooperative game theory also include the possibility
of competition among players.
We follow a cooperative game approach that consists of two elements: 65 a set
of players and +5 a characteristic function specifying the value created by
1
Morgenstern, O., Neumann, von, J., 1944. Theory o !ames an" #$onom%$ &ehav%or. 'r%n$eton
(n%vers%ty 'ress.
1)
different subsets of the players involved in the game. The cooperative game
theory attempts to answer how the total value is divided up among the players
and this answer will depend on their bargaining power. $urthermore, in
cooperative game theory, a player?s bargaining power depends on how much
other players need him to form coalitions, or his marginal contribution.
Brandenburger defines it as follows: =the marginal contribution of a particular
player is the amount by which the created overall value would shrin! if this
particular player leaves the game=. 8t is in this way that cooperative game theory
captures the idea of competition among different players in bargaining
situations.
There are different ways of solving cooperative games and we want to
emphasize two of them: the core and the #hapley value. The core was first
proposed by $rancis Hsidro %dgeworth in 6//6, and it was later reinvented and
defined in game theoretic terms by 'illies *60175. The core is a solution concept
that allocates to each cooperative game the set of payoffs that no coalition can
improve upon or bloc! *#errano, +,,-5. 2oncerning the #hapley value, it was
first proposed by 9loyd #hapley in his 6017 "h; dissertation. This solution
prescribes a single payoff for each player, which is the average of all marginal
contributions of that player to each coalition he or she is a member of *#errano,
+,,-5.
1- THE ECONOMIC ENVIRONMENT
This model is developed under a cooperative game framewor! and the
presence of incomplete contracts. "layers ta!e their decisions of buying and
producing, depending on their expected benefits. To compute those expected
benefits, we have chosen the #hapley value solution because it ta!es into
account what each player could reasonably get before the game starts.
&ore specifically, we consider a model with three agents, two downstream
parties and one upstream party. The upstream party is a supply firm which sells
its product to the downstream parties. The supplier has to ma!e specific
11
investments to obtain the product
q
, which has no alternative use out of these
relationships. We assume the existence of a parameter, , that measures the
degree of investment specificity, where
[ ] 1 , )
. The higher the value of , the
more specific the investment. 8f 1 , the investment is fully specific, while a
value ) indicates that the investment becomes totally general. $urthermore,
this product
q
reCuires a Cuality process that is not observable to third parties
and, therefore, it is not possible to write down an enforceable contract. The
supplier gets this
q
at a cost
s
c
.
The downstream parties buy the product from the supplier and they sell it in
different mar!ets. %ach firm is a monopoly in its mar!et. These downstream
firms have incentives to merge or to be vertically integrated with the supplier
because this would ensure them the reCuired Cuality. #o we can say the
downstream firms compete for the supplier giving him bargaining power, in the
sense that both downstream firms need the supplier. 8n this model, the
bargaining power responds to the existence of specific investments. :s a result,
the incentive problem is twofold: on one hand, the downstream firms want the
supplier to produce under a given Cuality process. 3n the other hand, the
supplier, after having made the specific investments, faces the possibility that
the downstream firms may argue that the product has not reached the reCuired
Cuality and, conseCuently, they will pay less for the product or, in the worst
case, they will refuse to buy it.
We consider a one-shot supply transaction. The downstream firm 6 buys
1
q
to
the supplier and the second firm buys
2
q
, such that
q q q +
2 1
and
) ,
2 1
I q q
.
The supplier has to ma!e specific investments to produce these Cuantities at a
cost 2 1
,
s s
c c
and sells them at prices
2 1
, w w
, being
) ,
2 1
w w
and
1
]
1

1
1
, ) w w
and
1
]
1

2
2
, ) w w
The downstream parties buy the Cuantities at prices
2 1
, w w
and sell them at
2 1
, p p
, respectively. The prices
2 1
, p p
are positive and
1
]
1

1 1
, ) p p
,
1
]
1

2 2
, ) p p

12
and they are decided by each firm. The downstream firm 6 faces a demand
+
function given by ( )
1 1 1 1 1 1
p p f q and the downstream firm + faces the
demand function ( )
2 2 2 2 2 2
p p f q . $or simplicity, we assume linear
demand functions. :dditionally, to guarantee that the inverse demand functions
exist and they are well defined, we further assume that:
6. the demand functions have negative slopes
+.
2 1
, f f
are differentiable functions whose first derivatives are strictly
negative and finite for any
1
]
1

1 1
, ) p p
and
1
]
1

2 2
, ) p p
such that
) ,
2 1
I f f
.
We also assume that
1
1
1 1
)

p w
and
2
2
2 2
)

p w
The players of the model have transferable utilities and they are rational, so
they ta!e decisions that maximize their expected utilities. The respective utility
functions are:
( )
s s s
U U +
for the supplier, ( )
1 1 1
+ U U for the
downstream firm 6 and ( )
2 2 2
+ U U for the downstream firm +. $urthermore,
2 , 1
,
s are random variables that follow a normal distribution with
( ) ( ) ( ) )
2 1
E E E
s
and variances
2
2
2
1
2
, ,
s
, respectively. $or simplicity,
we also assume the different players are ris! neutral.
$inally, the players? expected benefits are given by the following eCuations.
( )
1 1 1 1 1 1
, q w p q R , for the downstream firm 6, where the revenue is eCual to
( )
1 1 1 1 1
* , q p p q R
( )
2 1 2 2 2 2
, q w p q R , for the downstream firm +, where ( )
2 2 2 2 2
* , q p p q R ,
and
2 2 1 1 2 2 1 1 3
q c q c w q w q
s s
+
, for the supplier.
1/0 -THE TIMING

2
+or s%m,-%$%ty, %t %s assume" the "eman" un$t%on %s .no/n 0y the "o/nstream ,art%es. 1e .no/ th%s %s
a strong assum,t%on 0ut % /e assume e2,e$te" "eman" un$t%ons, our resu-ts /%-- rema%n un$hange".
13
:s we mentioned earlier, the players decide to buy and produce in terms of their
expected benefits. The timing of the model becomes as follows:
Figure 1: Timing of the Model
:t time ,, the downstream firms decide the Cuantity of product they will buy and
the prices they are going to sell it. The supplier decides to produce this level of
product, along with the Cuality and the price it will charge to the downstream
parties. :t time 6, the three firms will determine the payoffs and the distribution
of the benefits according to the negotiation process. ;ue to the existence of
incomplete information and the framewor! of incomplete contracts, the players
decide output level and their prices as a function of their expected future
benefits and their distribution
1/ 1- *ENCHMAR,
Jnder perfect information and complete contracts, we compute the first best. To
compute the first best it is assumed there is one player who has perfect
information and is the owner of the assets. The firms act as one player who
maximizes the total expected benefits *
T

5. The eCuation of the total expected

benefit is:
2 2 2 2 1 1 1 1
q c q p q c q p
s s T
+
, being the demand functions
( )
1 1 1 1 1 1
p p f q and ( )
2 2 2 2 2 2
p p f q .
The players decide the
Cuantity to buy and
produce, and the prices
)
T
1
T
;etermination of the
benefits and distribution
of them as a result of the
negotiation process
14
To assure the presence of a maximum, it must be satisfied that
)
) 2
)
2
2 2 2
<
>

xx
xy yy xx
y x
f
f f f
f f
#o, the maximum is achieved.
) 2
1 1 1 1 1
1
+

s
T
c p
p
) 2
2 2 2 2 2
2
+

s
T
c p
p
$rom these derivatives, we get the prices and Cuantities that maximize the total
net profit.
2
2 2 2
2
2

s
c
p
+

and
2
2 2 2
2

s
c
q

1
1 1 1
1
2

s
c
p
+

and
2
1 1 1
1

s
c
q

The total benefit in the benchmar! case becomes:

( ) ( )
2
2
2 2 2
1
2
1 1 1
4 4

s s
T
c c
+

Fext, we will compare this benchmar! solution with the results obtained under
incomplete contracts and a cooperative game framewor!.
1/ 2- THE NON-COOPERATIVE OR NASH SOLUTION
Before showing the model under a cooperative game theory framewor!, we
compute the solution under a non-cooperative game framewor!. To do this, we
use the Fash eCuilibrium definition. 8n our model, we need:
( ) ( )
1 2 1 2 1 1 2 1 2 1 1
, , , , , , p all for w w p p w w p p *:5
( ) ( )
2 2 1 2 1 2 2 1 2 1 2
, , , , , , p all for w w p p w w p p
*B5
15
( ) ( )
2 1 2 1 2 1 3 2 1 2 1 3
, , , , , , w and w all for w w p p w w p p
*25
The eCuations of the model are the following:
( ) ( ) ( )
1 1 1 1 1 2 1 2 1 1
, , , p w p w w p p *65
for the downstream firm 6, where
1
]
1

1
1
1
, )

p
and
1
]
1

1
1
1
, )

w

( ) ( ) ( )
2 2 2 2 2 2 1 2 1 2
, , , p w p w w p p *+5
for the downstream firm +, where
1
]
1

2
2
2
, )

p
and
1
]
1

2
2
2
, )

w
( ) ( )( ) ( )( )
2 2 2 2 2 1 1 1 1 1 2 1 2 1 3
, , , p c w p c w w w p p
s s
+
*75
for the supplier and
1 1
w c
s

and
2 2
w c
s

.
To compute the Fash solution we ta!e eCuation *65 and apply *:5 loo!ing for the
value
1
p
that maximizes eCuation *65, ceteris paribus.
The Fash solution to the downstream firm 6 is
1
1 1 1
1
2
w
p
+

We do the same for the downstream firm +, and the value

2
p
that maximizes
the function
2

is
2
2 2 2
2
2
w
p
+

#imilarly, for the supplier, we must calculate the values

1
w
and
2
w
that
maximize eCuation *75 given the rest of variables ( )
2 1
, p p . The results are:
1
1 1 1
1
2

s
c
w
+

2
2 2 2
2
2

s
c
w
+

Thus, once the prices

1
w
and
2
w
have been obtained, we replace them in
1
p

and
2
p
, and we get:
1
1 1 1
1
4
3

s
c
p
+

2
2 2 2
2
4
3

s
c
p
+

16
Eeplacing this solution in the eCuations *65, *+5 and *75, the players? net benefits
become:
( )
1
2
1 1 1
1
16

s
c

( )
2
2
2 2 2
2
16

s
c

( ) ( )
2
2
2 2 2
1
2
1 1 1
8 8

s s
s
c c
+

:s it can be seen, the Fash #olution shows us the effects of the double-
marginalization. This is not an efficient result and, obviously, we do not reach
the first best.
1/ 3 - INCOMPLETE CONTRACTS AND COOPERATIVE GAME MODEL
$ollowing the timing of the model, at time ,, the three players decide the
Cuantity to sell and produce but these decisions will ta!e into account the future
benefits they expect to achieve at time 6. #o, the players have to estimate their
future benefits. To do this, we are going to use a cooperative game approach
based on the sub>ect of the Theorem of the &inimax created by von Feumann
*60+/5. Jnder cooperative games, the players negotiate and compute their
payoffs under different coalitions. The fact of using the maximin to compute the
payoffs implies that, in the negotiation process, the actors estimate the least
amount they can get if the other players play against them in other words, they
compute the payoffs in the worst scenarios. 8n order to compute the payoffs or
the characteristic functions of the game, we must define first what a
characteristic function is. The definition of a #.&#te&isti# 4un#tion
2
of an n-
person game assigns to each coalition, which is a subset S of the players, the
best payoff that each one can achieve without the help of other players. 8n other
words, that is the value
( ) S v
that coalition S can guarantee for itself by
coordinating the strategies of its members, no matter what the other players do.
3
Thomas, 3.4., 1984. !ames, Theory an" 5,,-%$at%ons. #--%s 6or/oo" 3%m%te"
17
8t is standard to define the characteristic value of the empty coalition, , as ,
so
( ) ) v
. The characteristic function implies that if s
X
is the set of
strategies available to the player in S , and
S N
Y

is the set of strategies
available to the players in S N , then
( ) ( )

y x e S v
i
S i
Y y X x
S N s
, m%n ma2
,
where
( ) y x e
i
,
is the payoff to player i when
x
and
y
are the strategies
played by the players of the parties S and S N . This is for the mixed
extension of finite games. 8t follows from this definition that if S and T are
dis>oint coalitions for finite n-person games, we get:
( ) ( ) ( ) { } + T S if T v S v T S v , J
.
That is, superadditivity.
3nce the characteristic function is defined, we present the eCuations of the
model:
( ) ( ) ( )
1 1 1 1 1 2 1 2 1 1
, , , p w p w w p p *65
for downstream firm 6, where
1
]
1

1
1
1
, )

p
and
1
]
1

1
1
1
, )

w

( ) ( ) ( )
2 2 2 2 2 2 1 2 1 2
, , , p w p w w p p *+5
for downstream firm +, where
1
]
1

2
2
2
, )

p
and
1
]
1

2
2
2
, )

w
and
( ) ( )( ) ( )( )
2 2 2 2 2 1 1 1 1 1 2 1 2 1 3
, , , p c w p c w w w p p
s s
+
*75
for the supplier and
1 1
w c
s

and
2 2
w c
s

. The degree of investment
specificity, , is high and it could be close to one in this model.
The possible coalitions of the model are:
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 3 , 2 , 1 , 3 , 2 , 3 , 1 , 2 , 1 , 3 , 2 , 1 , v v v v v v v v
.
The computation for the characteristic function for the downstream firm 6
becomes as follows:
18
$irst, we ta!e eCuation *65 and apply the maximin
( ) ( ) ( ) ( ) [ ]
1 1 1 1 1 1 1 1
, ,
2 1 2 1 1
, ,
m%n ma2 m%n ma2
2 1 2 1 2 1 2 1
, , , 1 p w p p w w p p v
w w p p w w p p


*45
We must find the minimum values of
2 1 2
, , w w p
that minimize this function for a
given
1
p
, and then we will maximize that function with respect to
1
p
. 8n our
context, the minimum value of
1
w
that minimizes this function given
1
p
is
1
1
1

w
, so we replace this value in the function and loo! for the
1
p
that
maximizes the function.
( ) ( ) ( ) ( )
2 1 2 1 1 1 1
1
1
1 1 1 1
, , ,
1
1
ma2
ma2
1
1
w w p p p p p v
p
p

1
]
1

) 2
1 1 1 1
1
1
+

p
p
To ensure the existence of a maximum, we chec! the second derivative
1
2
1
1
2
2

p
. 8t is negative and, therefore, we have got a maximum.
Thus,
1
1
1

p
and
)
1
q
, and replacing these values in the function *45 we
get that the characteristic function for the downstream firm 6 becomes:
{ } ( ) ( ) ) 1 1 v v
. *15
This is the payoff that the player 6 will have in his worst scenario. 8t means that
if the supplier charges him the highest price
1
1
1

w
, which is the highest cost
that the downstream firm 6 can face, the downstream firm 6Ks best response is
to set the highest price of
1
1
1

p
where the demand Cuantity is eCual to ,. 8n
that case, he gets zero profits.
19
The steps to compute the characteristic function to the downstream firm + are
similar to the ones already computed for downstream firm 6. We proceed to
simply write down the results and present the eCuation *.5.
( ) ( ) ( ) ( ) [ ]
2 2 2 2 2 2 2 2
, ,
2 1 2 1 2
, ,
m%n ma2 m%n ma2
2 1 1 2 2 1 2
, , , 2 p w p p w w p p v
w w p p w w p p

2
2
2

p
and
)
2
q
The characteristic function or the payoff that downstream firm + will receive in
case that the other players >oin themselves against him becomes
{ } ( ) ( ) ) 2 2
2
v v *-5
%Cuations *15 and *-5 or the characteristic functions for the downstream firm 6
and the downstream firm +, indicate the lowest value that they are able to get
under the worst scenario.
#imilarly, for the supplier, the characteristic function becomes:
( ) ( )
( ) ( ) ( ) ( ) [ ]
2 2 2 2 1 1 1 1 2 2 2 2 1 1 1 1
, ,
2 1 2 1 3
, ,
m%n ma2
m%n ma2
2 1 2 1
2 1 2 1
, , , 3
p c p c p w p w
w w p p v
s s
p p w w
p p w w
+

We have to loo! for the minimum values of
2 1
, p p
that minimize the function.
These values are:

( ) ( ) ( )
1
]
1

,
_

,
_

2
2
2 2 2 2
1
1
1 1 1 1
,
ma2
2 1
3

s s
w w
c w c w v
( ) [ ] ) 3
ma2
2 1
,w w
v
The characteristic function for the supplier is:
{ } ( ) ( ) ) 3 3 v v
*/5
2)
The eCuation */5 is telling us the value that, in the worst case, the supplier gets.
We can thin! that this is the case that the downstream firms >oin together and
argue that the Cuality is not the one reCuired and they do not want the product.
The supplierKs payoff is zero because the supplier cannot force the downstream
parties to buy the product.
The characteristic function for the coalition
( ) 2 , 1 v
is obtained considering the
sum of
2 1
+
because the coalition
{ } 2 , 1
gets the payoff of both players
together.
( ) ( ) ( ) [ ]
( ) ( ) ( ) ( ) [ ]
2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1
, ,
2 1 2 1 2 2 1 2 1 1
, ,
m%n ma2
m%n ma2
2 1 2 1
2 1 2 1
, , , , , , 2 , 1
p w p p p w p p
w w p p w w p p v
w w p p
w w p p
+
+
9oo!ing for the minimum values of
2 1
, w w
that minimize the function and
replacing them, we obtain *05:
( ) ( ) ( ) ( ) ( ) [ ]
2 1
,
2 2 2
2
2 2 2 2 1 1 1
1
1
1 1 1 1
,
ma2 ma2
2 1 2 1
2 , 1 +
1
]
1

+
p p p p
p p p p p p v

We have to loo! for the values of
2 1
, p p
that ma!e the function maximum.
Thus, we derivate with respect to
2 1
p and p
. The first derivatives must be
eCual to , and the second ones must be negative to ensure the presence of a
maximum.
[ ]
) 2 2
1 1 1
1
2 1

+
p
p

[ ]
1
2
1
2 1
2
2

+
p
[ ]
) 2 2
2 2 2
2
2 1

+
p
p


[ ]
2
2
2
2 1
2
2

+
p

These values are:
1
1
1

p )
1
q
21
2
2
2

p

)
2
q
Eeplacing in eCuation *05 we get the characteristic function
{ } ( ) ( ) ) 2 , 1 2 , 1 v v
*6,5
This is the payoff that will get the coalition between the downstream parties in
the worst condition that the supplier is against them. The supplier will charge
them the highest price and their response will be to put the highest prices and
they will sell ,. The benefits they get are eCual to zero in the worst condition.
This eCuation points out the fact that if the downstream firms cooperate with
each other, without any consideration of the third player *the supplier5, they will
be able to get ,. They cannot force the supplier to invest and sell them the
product.
The characteristic function for the coalition
( ) 3 , 1 v
is:
( ) ( ) ( ) [ ]
( ) ( ) ( ) ( ) [ ]
2 2 2 2 1 1 1 1 2 2 2 2 1 1 1 1
, ,
2 1 2 1 3 2 1 2 1 1
, ,
m%n ma2
m%n ma2
2 2 1 1
2 2 1 1
, , , , , , 3 , 1
p c p c p w p p
w w p p w w p p v
s s
p w w p
p w w p
+
+
9oo!ing for the minimum value of
2
p
that minimizes the function, we get:
( ) ( ) ( )
1
]
1

,
_

,
_

+
2
2
2 2 2 1 1 1 1
2
2
2 2 2 1 1 1 1
, ,
ma2
2 1 1
3 , 1

s s
w w p
c p c w p p v

( ) ( ) ( ) [ ] [ ]
3 1
, ,
1 1 1 1 1 1 1 1
, ,
ma2 ma2
2 1 1 2 1 1
3 , 1 +
w w p
s
w w p
p c p p v

*665
Fow, we have to find the value of
1
p
that maximizes *665
[ ]
) 2
1 1 1 1 1
1
3 1
+

+

s
c p
p
[ ]
1
2
1
3 1
2
2

+
p
1
1 1 1
1
2

s
c
p
+

2
1 1 1
1

s
c
q

Eeplacing these values in the eCuation *665, we get the characteristic function
22
{ } ( ) ( )
( )
1
2
1 1 1
4
3 , 1 3 , 1

s
c
v v

*6+5
This is the payoff that the coalition between the downstream firm 6 and the
supplier can get if the downstream firm + is assumed to oppose them. The
supplier will produce
2
1 1 1
1

s
c
q

and the downstream will sell it at the price
1
1 1 1
1
2

s
c
p
+

. The characteristic function implies that, under any

circumstances, the downstream firm 6 and the supplier, together, are sure to
obtain the least amount given by this eCuation. This result is the same as if the
supplier and the downstream firm 6 were vertically integrated. 8t is necessary to
note that this coalition ma!es sense if the investments are specific because, if
the investments were general, the players neither need to form coalitions nor to
be vertically integrated.
;oing the same for the coalition between the downstream firm + and the
supplier, we get:
( ) ( ) ( ) [ ]
( ) ( ) ( ) ( ) [ ]
2 2 2 2 1 1 1 1 1 1 1 1 2 2 2 2
, ,
2 1 2 1 3 2 1 2 1 2
, ,
m%n ma2
m%n ma2
1 2 1 2
1 2 1 1
, , , , , , 3 , 2
p c p c p w p p
w w p p w w p p v
s s
p w w p
p w w p
+
+
9oo!ing for the minimum value of
1
p
that minimizes the function, we get:
( ) ( ) ( )
1
]
1

,
_

,
_

+
2 2 2 2
1
1
1 1 1
1
1
1 1 1 2 2 2 2
, ,
ma2
2 1 2
3 , 2 p c c w p p v
s s
w w p

( ) ( ) ( ) [ ] [ ]
3 2
, ,
2 2 2 2 2 2 2 2
, ,
ma2 ma2
2 1 2 2 1 2
3 , 2 +
w w p
s
w w p
p c p p v
*675
Fow, we have to find the value of
2
p
that maximizes *675
[ ]
) 2
2 2 2 2 2
2
3 2
+

+

s
c p
p
[ ]
2
2
2
3 2
2
2

+
p
23
2
2 2 2
2
2

s
c
p
+

2
2 2 2
2

s
c
q

Eeplacing these values in the eCuation *675, we get the characteristic function
{ } ( ) ( )
( )
2
2
2 2 2
4
3 , 2 3 , 2

s
c
v v

*645
This is the payoff that the coalition between the downstream firm + and the
supplier can get if the downstream firm 6 acts against them. The supplier will
produce
2
2 2 2
2

s
c
q

and the downstream will sell it at the price
2
2 2 2
2
2

s
c
p
+

. The characteristic function implies that, under any

circumstances, the downstream firm + and the supplier, together, are sure to
obtain the least amount given by this eCuation. :s in the previous case, this
result is the same as if the supplier and the downstream firm + were vertically
integrated.
The characteristic function for the coalition
( ) 3 , 2 , 1
is given by:
( ) ( ) ( ) ( ) [ ]
( ) ( ) ( ) ( ) [ ]
2 2 2 2 1 1 1 1 2 2 2 2 1 1 1 1
, , ,
2 1 2 1 3 2 1 2 1 2 2 1 2 1 1
, , ,
ma2
ma2
2 1 2 1
2 1 2 1
, , , , , , , , , 3 , 2 , 1
p c p c p p p p
w w p p w w p p w w p p v
s s
w w p p
w w p p
+
+ +
:nd now we must find the values
2 1
, p p
that maximize the function.
[ ]
) 2
1 1 1 1 1
1
3 2 1
+

+ +

s
c p
p
[ ]
1
2
1
3 2 1
2
2

+ +
p
1
1 1 1
1
2

s
c
p
+

2
1 1 1
1

s
c
q

) 2
2 2 2 2 2
2
+

s
c p
p
2
2
2
2
2

p
2
2 2 2
2
2

s
c
p
+

2
2 2 2
2

s
c
q

Eeplacing these values in the eCuation we get that:

24
{ } ( ) ( )
( ) ( )
2
2
2 2 2
1
2
1 1 1
4 4
3 , 2 , 1 3 , 2 , 1

s s
c c
v v

+

*615
This expression is the characteristic function of the grand coalition. This is the
maximum payoff that the grand coalition, or total coalition, can achieve if they
decide to cooperate with each other. 8f we compare this payoff with the
benchmar!, we can observe that we reach the first best if the players decide to
cooperate all together. 8f we also compare this result with the non-cooperative
solution, we can argue that the solution under cooperative framewor! is better
than the one under non-cooperative framewor! where the benefits for the
players are lower due to of the presence of double marginalization.
3nce we have computed the characteristic function for all the coalitions, we
also !now that, due to the definition of the characteristic function given by
Thomas *60/45, if S and T are dis>oint coalitions, we get:
( ) ( ) ( ) { } + T S if T v S v T S v ,
. Where

stands for the empty set.

#o, in our model, the following expressions are satisfied
( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( ) 2 , 1 3 , 2 2 3 , 2 , 1
3 , 1 2 , 1 1 3 , 2 , 1
3 , 2 3 , 1 3 3 , 2 , 1
1 3 , 2 3 , 2 , 1
2 3 , 1 3 , 2 , 1
3 2 , 1 3 , 2 , 1
3 2 3 , 2
3 1 3 , 1
2 1 2 , 1
3 2 1 3 , 2 , 1
v v v v
v v v v
v v v v
v v v
v v v
v v v
v v v
v v v
v v v
v v v v
+ +
+ +
+ +
+
+
+
+
+
+
+ +
This means that superadditivity holds for the characteristic function. We also
!now that the presence of superadditivity indicates that the introduction of one
player into the coalition adds value to it. Therefore, this means that the
marginal contribution of each player added to a coalition, is not null.
To solve this model, we use the core as the solution and then the #hapley
Dalue that is the baricenter of the core. $or a convex characteristic function, we
determine how the benefits of the model can be split between the players under
the negotiation process. 8n other words, we are going to solve the model ta!ing
25
into account the possible set of rewards that the players can reach. We will
consider the entire problem to find the solution, given that the game is essential.
That the game is essential implies that the characteristic function is
superadditive and therefore, for two dis>oint coalitions the value of the
characteristic function of the union is strictly greater that the sum of the value of
the individual characteristic functions. This is eCuivalent to say that
( ) ( ) ( ) ( ) 3 , 2 , 1 3 2 1 v v v v < + +
*&yerson, 60065.
1/5- THE CORE AS A SOLUTION O6 THE GAME
)aving the characteristic function of the game of the two downstream firms and
the supplier, it is important to find a suitable concept of solution. $or this reason,
first we choose the core as a possible solution. "revious to define the core, it is
necessary to define an imputation
4
. :n imputation in an n-person game with
characteristic function
v
is a vector
( )
n
x x x x ,...., ,
2 1

satisfying the following:

*i5
( )7
1

n
i
i
N v x
*ii5
( ) n i for i v x
i
,....., 2 , 1 ,
Where
i
x
is obviously the ith playerKs reward. The condition *i5 is a "areto
optimality condition or the rationality of the grand coalition.
( ) N v
is the most the
players can get out of the game when they all wor! together. #o, for any
possible set of individual rewards
i
x
we must have
( )

n
i
i
N v x
1
. 8f this was a
strict ineCuality, then, by wor!ing together, they could always share out the
rewards so that everyone got more. The condition *ii5 says that everyone must
get as much as they could get if they played by themselves.
3nce we had defined the imputation concept, we introduce the concept of the
core *'illes, 60175. The core of a game
v
, denoted by
( ) v C
, is the set of
imputations which are not dominated for any other coalition. This means that
you can not find another imputation
y
and a coalition S such that
4
Thomas, 3.4., 1984. !ames, Theory an" 5,,-%$at%ons. #--%s 6or/oo" 3%m%te"
26

( ) S i all for x y and S v y
i i
S i
i
>

Thus, if
x
is in the core, any coalition which forms, either says
x
is the best
imputation for it. Fotice there can be more than one imputation in the core. We
provide a result which can be found in Thomas *60/45.
Theorem:
x
is in the core if and only if
*i5
( )

n
i
i
N v x
1
,
and
*ii5
( )

S i
i
S v x
for all N S
3nce the concepts of an imputation and the core are given, we can apply the
core solution in our model. To apply the core, it is feasible to consider in a better
way its ,-6 reduction. 8t must satisfy:
( ) ( )

+
S i
i
w S kv S w
*6.5
#uch that:

( )
( ) 1 3 , 2 , 1
7 )

w
i w
Ta!ing the characteristic functions of the model
{ } ( ) ( ) ) 1 1 v v { } ( ) ( ) ) 2 2
2
v v { } ( ) ) 3 v
{ } ( ) ( ) ) 2 , 1 2 , 1 v v
{ } ( ) ( )
( )
1
2
1 1 1
4
3 , 1 3 , 1

s
c
v v


{ } ( ) ( )
( )
2
2
2 2 2
4
3 , 2 3 , 2

s
c
v v


{ } ( ) ( )
( ) ( )
2
2
2 2 2
1
2
1 1 1
4 4
3 , 2 , 1 3 , 2 , 1

s s
c c
v v

+

Fext we loo! for the value
( ) S w
. To simplify the mathematical calculation, we
denote
( )
1
2
1 1 1
4

s
c
and
( )
2
2
2 2 2
4

s
c
.
:pplying eCuation *6.5, we get:
27
( ) ( ) ) ) 1 1
1 1
+ + w k w kv w
( ) ) ) 2
2
+ w k w
( ) ( ) ) ) 3 3
3 3
+ + w k w kv w

( ) ( )
( ) 3 , 2 , 1
1
1 3 , 2 , 1 3 , 2 , 1
3 2 1
v
k
w w w kv w

+ + +
( ) ( ) ) 2 , 1 2 , 1
2 1
+ + w w kv w
( ) ( )
( )
( )

+
+ +
3 , 2 , 1
3 , 1
3 , 1 3 , 1
3 1
v
v
w w kv w
( ) ( )
( )
( )

+
+ +
3 , 2 , 1
3 , 2
3 , 2 3 , 2
3 2
v
v
w w kv w
These eCuations are the characteristic functions reduced to ,-6. :nd once we
have reduced the characteristic function, we can then apply the reduced core. 8t
must satisfy the following:
( )
( )
( )

+ +
+
+
+

1
3 , 2
3 , 1
) 2 , 1
)
3 2 1
3 2
3 1
2 1
x x x
w x x
w x x
w x x
x
i
Where
,
and

denote the characteristic functions in the ,-6 reduced core.

#o, we can get the feasible set of payoffs of the model as the figure + shows.
Figure : The reduced core solution
28
The painted area is the reduced core solution. Therefore all the points of the
reduced core are possible solutions for the game played by the three agents
under consideration. &oreover, under the condition that
* +
in the
reduced game, the game is convex in the sense
( ) ( ) ( ) ( ) T S v T S v T v S v + +

for all S and T
1
subsets of
{ } 3 , 2 , 1
. The last property is also valid for non-
reduced games.
8n the case that the characteristic function is convex, then both the reduced and
non-reduced core are non-empty, and have a regular structure. 8n particular,
there is the #hapley value which is the center of gravity of the extreme points of
the core. 8f the core is only a point, the #hapley value is uniCue and coincides
with the core *#hapley, 60.15.
1/7-THE SHAPLEY VALUE
: solution concept for the game under consideration is the #hapley value
*"eleg and #udhGlter, +,,75 and *Branzei and Ti>s, +,,15. #hapley *60175
loo!ed at what each player could reasonably get before the game has begun.
)e put three axioms, which he called
( ) v
i

, player iKs expectation in a game

with a characteristic function
v
, should satisfy the following:
#6:
( ) v
i

is independent of the labeling of the players. 8f

is a permutation of
6, +,...., n and
v
is the characteristic function of the game, with the players
numbers permuted by

, then ( )
( ) ( ) v v
i i

.
#+: The sum of the expectations should eCual the maximum available from the
game, so
( ) ( )

n
i
i
N v v
1

#7: 8f
v u,
are the characteristic functions of two games,
v u +
is the
characteristic function of the game playing both games together.

must satisfy
( ) ( ) ( ) v u v u
i i i
+ +
.
5
#hapley, 9loyd, 60-6. 2ores of convex games. 8nternational (ournal of 'ame Theory 6, 66-+.

29
'iven these assumptions, #hapley proved the following theorem:
Theorem. There is only one function which satisfies #6, #+ and #7, namely:
( )
( ) ( )
( ) { } ( ) ( ) i S v S v
n
S n S
v
S i S
i

8
9
9 : 9 1 :

Where the summation is over all coalitions S which contain player i and L S is
the number of players in the coalition S .
( ) v
i

is called the #hapley value. The

Cuantity
i

may be interpreted as the <eCuity value= associated with the

position of the i-th player in the game *#hapley, 60.15.
8n our model, the #hapley value is:
$or downstream firm 6,
( )
( )
1
2
1 1 1
1
4 2
1

s
c
v

$or downstream firm +,

( )
( )
2
2
2 2 2
2
4 2
1

s
c
v

$or the supply firm

( )
( ) ( )

,
_

2
2
2 2 2
1
2
1 1 1
3
4 4 2
1

s s
c c
v
:nd the total value created is eCual to
( ) ( )
2
2
2 2 2
1
2
1 1 1
4 4

s s
c c
+

. :s it can be
chec!ed, this total value is eCual to the first best and to the value obtained when
the three firms cooperate with each other. That is, the grand coalition?s
characteristic function.
The #hapley value indicates how the total benefits can be split in a fair way. We
can see that the supplier gets more benefits compared with the benefits that
each downstream firm gets. This indicates what we posed in the previous
section that the cooperative game theory attempts to answer how the total value
is divided up among various players. We remar! that this answer will depend on
the bargaining power of the players, and this is determined by which players are
3)
most needed. 8n this model, the supplier is the most needed because the firms
have incentives to merger or to be vertically integrated with the supply firm, for
the reason that a merger or the vertical integration ensures them that the
production will have the reCuired Cuality. 8n other words, the bargaining power
of the supplier relies on the specificity of the investments because it is this
specificity what ma!es the cooperation credible. 8f the investments were
general, non-specific, the cooperation framewor! would not be enforceable
because each player could sell or produce outside the relationship. #uch
bargaining power is measured by the #hapley value and it is what ma!es
possible the cooperation among the three players, achieving a "areto optimum
solution. Thus, this higher supplier?s bargaining power is what ma!es the
upstream-downstream relationship an enforceable one. :s a result of this, if the
the three firms commit to cooperate and they distribute the benefits as the
#hapley value indicates, at time ,, they will decide the Cuantity level and the
prices that maximize total welfare.
2- CONCLUSIONS
8n this paper we have developed a model within the upstream-downstream
relationship that attempts to find an alternative solution to vertical integration.
3ur contribution is twofold: first, a mathematical part where we show how to
apply the maximin to compute the characteristic functions of the model and
second, an economic contribution, where we explain how cooperation can lead
to organizational forms that generate more value. 8n fact, we set another way to
treat the upstream-downstream relationship under incomplete contracts and a
cooperative framewor!, providing an alternative to vertical integration.
:s a solution, vertical integration presents some problems and it seems to be a
poor solution under certain circumstances. 8n particular, the existence of these
failures made us wonder if there could be another way to solve those problems
that vertical integration did not solve. &ore specifically, what happens if the
supply firm does not want to be vertically integrated? What happens if the firms
decide to cooperate instead of being vertically integrated under the presence of
specific investments? To answer these Cuestions we have developed a model
31
where the firms cooperate with each other. We find that, under cooperation, the
total value generated is "areto optimum if they cooperate and distribute the
benefits through the #hapley value solution. The #hapley value measures the
power that firms en>oy in the bargaining process, where bargaining power under
cooperative game theory is defined in terms of how much each player is
needed. The result of our model is that the supplier?s bargaining power is
determined by the specificity of the investments and this ma!es the cooperation
among the firms possible, and the upstream-downstream relationship
enforceable. :s a conseCuence, the first best, that maximizes the total welfare,
can be achieved.
8n this paper, we have presented a one-shot supply transaction. We want to
consider further developments of this model, with n-periods and more
participants, in our future research. 8n such scenarios, one must include
relational contracts and the reputation of the firms, increasing the complexity of
the analysis. We are also interested in the use of biform games, which combine
both the non-cooperative game theory and the cooperative game approach in
the same analysis. We believe that more efforts should be implemented in this
direction to achieve a better understanding of organizational forms.
32
5-*I*LIOGRAPHY
:lchian, :., ;emsetz, )., 60-+. "roduction, 8nformation 2osts and %conomic
3rganization. :merican %conomic Eeview. .+, -----01.
:o!i, &., *60//5. 8nformation, 8ncentives and Bargaining in the (apanese
%conomy. 2ambridge Jniversity "ress, 2ambridge, JA *60//5.
:o!i, &., *600,5. Toward an %conomic &odel of the (apanese $irm. (ournal of
%conomic 9iterature +/, 6M+-.
:umann, E.(., 6010. :cceptable "oints in 'eneral 2ooperative n-"erson
'ames. 2ontribution to the Theory of 'ames 8D. "rinceton Jniversity "ress,
pp. +/--7+4.
Bondareva, 3.F., 60.+. 2ertain applications of the methods of linear
programing to the theory of cooperative games. "roblemy Aiberneti!i 6,, 660-
670 *in Eussian5.
Brandenburger, :dam, *+,,-5. 2ooperative 'ame Theory: 2haracteristic
$unctions, :llocations, &arginal 2ontribution. #tern #chool of Business. Few
Hor! Jniversity!

Branzei, E., ;imitrov, ;., Ti>s, #., +,,1. &odels in 2ooperative 'ame Theory.
2risp, $uzzy and &ulti-2hoice 'ames. #pringer-Derlag Berlin )eidelberg.
2achon, '., 9ariviere, &., +,,1. #upply 2hain 2oordination with Eeveneu-
#haring 2ontracts: #trengths and 9imitations. &anagement #cience 16, 7,-
44.
2antisani &., &archi %., +,,4. The Weighted 2ore with ;istinguished 2oalition/
8nternational 'ames Theory Eeview . *+5, +70-+4..
2handler, :., 60.+. #trategy and #tructure. Few Hor!: ;oubleday!
2oase, E., 607-. The Fature of the $irm. %conomica 4, 71--7-..
%dgeworth, $.H., 6//6. &athematical "hysics and $urther "apers. "olitical
%conomy. 3xford Jniversity "ress.
'ibbons, E., +,,,. Why 3rganizations are #uch a &ess *and What an
%conomist &ight ;o :bout 8t5. Eough ;raft. &8T.
'illies, ;. B., 6017. #olutions to general non-zero-sum games. 2ontributions to
the Theory of 'ames 8D. "rinceton Jniversity "ress, pp. 4--/1.
'rossman, #., )art. 3., 60/.. The 2osts and Benefits of 3wnership: : Theory
of 9ateral and Dertical 8ntegration. (ournal of "olitical %conomy 04, .06--60.
33
'rossman, 3., &oore, (., 600,. "roperty Eights and the Theory of the $irm.
(ournal of "olitical %conomy 0/, 6660-661/.
)art, 3., &oore, (., 600,. "roperty Eights and the Theory of the $irm. (ournal
of "olitical %conomy 0/, 6660-661/.
)art, 3., 6001. $irms, 2ontracts and $inancial #tructures. 3xford Jniversity
"ress.
)olmstrGm, B., Eoberts, (., 600/. The Boundaries of the $irm Eevisited. (ournal
of %conomic "erspectives 6+, -7-04.
Alein, B., 2rawford, E., :lchian, :., 60-/. Dertical 8ntegration, :ppropriable
Eents and the 2ompetitive 2ontracting "rocess. (ournal of 9aw and
%conomics +6, +0--7+..
Alein. B., 60//. Dertical 8ntegration as 3rganizational 3wnership: The $ischer
Body-'eneral &otors Eelationship Eevisited. (ournal of 9aw, %conomics and
3rganization 4, 600-+67.
&archi, %., 60.-. 3n the &inimax Theorem of the Theory of 'ames. :nn. &at.
"ura ed :ppl. #er. 8D, Dol --, +,--+/+ *&E 7/,60645
&archi, %., &atons, &., +,,-. Eeduced TJ-'ames. 8nternational (ournal of
:pplied &athematics, 'ame Theory and :lgebra 6-, 8ssue 1N..
&yerson, E., 6006. 'ame Theory. :nalysis of 2onflict. )arvard Jniversity
"ress.
Fagara>an, &., #oOiP, '., +,,.. 'ame-Theoretic :nalysis of 2ooperation
:mong #upply :gents: Eeview and %xtensions. %uropean (ournal of
3perational Eesearch
Feumann, von, (., 60+/. #ur Theorie der 'esellschaftsspielle. &ath. :nnals
6,,, +01-7+,
Feumann, von, (., &orgenstern, 3., 6044. Theory of 'ames and %conomic
Behavior. "rinceton Jniversity "ress, F(.
"eleg, B., #udhGlter, "., +,,7. 8ntroduction to The Theory of 2ooperative
'ames. Aluwer :cademic "ublishers.
"orter, '., 9ivesey, )., 60-6. &erchants and &anufacturers. Baltimore: (ohn
)op!ins Jniversity "ress.
Ea>an, E., @ingales, 9., +,,,. The 'ovenance of the Few %nterprise. %n Q.
Dives, edr., 2orporate 'overnance: Theoretical and %mpirical "erspectives.
2ambridge Jniversity "ress, 2ambridge.
34
#errano, E., +,,-. 2ooperative 'ames: 2ore and #hapley Dalue. %ncyclopedia
of 2omplexity and #ystems #cience. #pringer. Berlin.
#hapley, 9loyd., 6017. : value for n-person games. 2ontribution to the Theory
of 'ames 88. "rinceton Jniversity "ress, pp. 7,--76-.
#hapley, 9., 60.-. 3n balanced sets and cores. Faval Eesearch 9ogistics
Ruarterly 64, 417-4.,.
#hapley, 9loyd, 60-6. 2ores of convex games. 8nternational (ournal of 'ame
Theory 6, 66-+..
Thomas, 9.2., 60/4. 'ames, Theory and :pplications. %llis )orwood 9imited.
Williamson, 3., 60-1. &ar!ets and )ierarchies: :nalysis and :ntitrust
8mplication: : #tudy in the %conomics of 8nternal 3rganization, Few Hor!,
$ree "ress.
Williamson, 3., 60/6. The %conomics of 3rganization.The Transaction 2ost
:pproach. :merican (ournal of #ociology /-, 14/-1--.
@ingales, 9., 600/. 2orporate 'overnance. The Few "algrave ;ictionary of
%conomics and the 9aw, ". Fewman ed, &acmillan, Few Hor!, FH.
35