WPS4074
COOPERATIVE GAME THEORY AND
ITS APPLICATION TO NATURAL,
ENVIRONMENTAL AND WATER
RESOURCE ISSUES:
3. Application to Water Resources
Irene Parrachino, Ariel Dinar and Fioravante Patrone
Abstract
This paper provides a review of various applications of cooperative game theory (CGT) to issues of water
resources. With an increase in the competition over various water resources, the incidents of disputes
have been in the center of allocation agreements. The paper reviews the cases of various water uses, such
as multi-objective water projects, irrigation, groundwater, hydropower, urban water supply, wastewater,
and transboundary water disputes. In addition to providing examples of cooperative solutions to
allocation problems, the conclusion from this review suggests that cooperation over scarce water
resources is possible under a variety of physical conditions and institutional arrangements. In particular,
the various approaches for cost sharing and for allocation of physical water infrastructure and flow can
serve as a basis for stable and efficient agreement, such that long-term investments in water projects are
profitable and sustainable. The latter point is especially important, given recent developments in water
policy in various countries and regional institutions such as the European Union (Water Framework
Directive), calling for full cost recovery of investments and operation and maintenance in water projects.
The CGT approaches discussed and demonstrated in this paper can provide a solid basis for finding
possible and stable cost sharing arrangements.
World Bank Policy Research Working Paper 4074, November 2006
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas
about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully
polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and
conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World
Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at
http://econ.worldbank.org.
This Working paper is a product of the first phase of the study "Cooperative Arrangements for Allocation of Water
among Production and Environmental Uses under Stochastic Supply Conditions", funded by the Italian Trust Fund
and the Development Economics Research Group of the World Bank. Members of the study team include: Ariel
Dinar (Task Manager), Fioravante Patrone, Irene Parrachino, Stefano Zara of the University of Genoa, Carlo
Carraro, Carmen Marchiori, and Alessandra Sgobbi of the University of Venice. Comments by Rashid Sumaila,
University of British Columbia, Canada and Kim Hang Pham Do, Massey University, New Zealand, and
participants of the 6th meeting on Game Theory and Practice Dedicated to Development, Natural Resources and the
Environment, July 10-12, Zaragoza, Spain, are much appreciated. Two additional Working Papers are also devoted
to some special topics, which are chosen having in mind their use in applications to environmental issues: 1. Basic
Thoery (PRWP 4072), and 2. Application to Natural and Environmental Resources (PRWP 4073).
CONTENTS
INTRODUCTION 3
THE CASE OF MULTIPURPOSE WATER RESOURCE PROJECTS 3
URBAN WATER SUPPLY 9
IRRIGATION 12
HYDROELECTRIC POWER 17
WATER POLLUTION 18
WASTEWATER TREATMENT 21
GROUNDWATER 27
ALLOCATION OF WATER RESOURCES 29
INTERNATIONAL WATER BASINS 31
CONCLUDING REMARKS 40
REFERENCES 42
2
INTRODUCTION
Many economic activities have been subject to strategic behavior on the part of individuals or firms (from
hereafter--players or parties), especially when the groups involved are not too big. Game theory has
been providing important input to policy making in several sectors (e.g., communication, transportation,
aviation, energy). In the specific field of water resources, game theory has not been used to its full
capacity by social science experts and practitioners.1 While in other fields non-cooperative behavior is
common, we see much more focus on cooperation in the field of water. What makes water more
appropriate for cooperation among the parties involved? First, significant economies of scale make it
more attractive to build bigger rather than smaller water projects, thus, there is an incentive for inclusion;
and second, the level of negative externalities associated with water utilization is a big incentive to
cooperate and reduce such impacts.
Most of the cooperative game theory (CGT) applications are about sharing benefits. Benefits can be
created both from saving on investment and operational cost and from increasing welfare that is
associated with the utilization of the resource--e.g., new opportunities that are not available when no
cooperation is considered. What CGT offers in most cases is an allocation of the benefits from
cooperation that, depending on the solution (allocation scheme) concept, fulfill certain requirements.
The paper will review a set of works that have been identified in the literature that include applications to
allocations resulting from cooperation in dams, water supply and wastewater treatment facilities, water
exchange (water markets and water rights), groundwater, multipurpose water facilities, and transboundary
water conflicts. The list of CGT solution concepts will include: The Core, The Shapley Value; The
Nucleolus; The Generalized Shapley; Nash/Nash-Harsany; and others. We will also refer to some non
CGT allocation solutions, such as Egalitarian Non-Separable Cost (ENSC); Alternate Cost Avoided
(ACA); Separable Costs Remaining Benefits (SCRB); Marginal cost; and Proportional use. All the non
CGT allocation solutions have been applied in certain works and will be used for comparison purposes.
For the purpose of organization of the paper, we grouped the reviewed works by the sub-sectors to which
they have been applied. This structure allows us to provide sub-sector specific focus, but this also comes
at the expense of extrapolating among features of the sub-sectors that could be relevant to other sectors.
To fix this drawback, we added at the end of the paper a section that cross cuts among the sub-sectors and
attempts at deriving general lessons and recommendations.
THE CASE OF MULTIPURPOSE WATER RESOURCE PROJECTS
The Tennessee Valley Authority (TVA) case is probably both the earliest and the richest case that offered
game theorists an opportunity to examine, from a CGT point of view, a practical problem of cost
allocation in a water resources development project among various components that have been designed
as part of that project. Ransmeier, (1942) provides an authentic description of the TVA project, and
especially on the various attempts to allocate, using non CGT schemes, project costs among various
project components. Following Ransmeier, (1942), we find several attempts to apply CGT concepts to
the TVA case.
1This is a third in a series of three working papers on cooperative game theory principles and their application to
natural, environmental and water resources. The first paper "COOPERATIVE GAME THEORY AND ITS
APPLICATION TO NATURAL, ENVIRONMENTAL AND WATER RESOURCE ISSUES: 1. Basic Theory" by
Irene Parrachino, Stefano Zara and Fioravante Patrone. The second paper "COOPERATIVE GAME THEORY
AND ITS APPLICATION TO NATURAL, ENVIRONMENTAL AND WATER RESOURCE ISSUES: 2.
Application to Natural and Environmental Resources" by Stefano Zara, Ariel Dinar and Fioravante Patrone.
3
In a following paper, Parker, (1943) focuses on the various theories applicable to the allocation of costs
of a multiple-purpose water control project, and presents the procedure followed by the TVA in allocating
the cost of its first three dams to navigation, flood control and power. It is shown that a large difference
in the allocation of the common costs of a project affects to a much lesser degree the proportion of the
total cost charged to the individual purposes. Consequently the final allocations are approximately
correct, within reasonable limits, regardless of the exact methods adopted.
We report only the main methods because they capture implicit GT principles. (To remind ourselves, the
analysis is about the cost allocation for the first three dams built for navigation, flood control and electric
power production.)
The analysis of the TVA cost allocation (Ransmeier, 1942) distinguished between direct costs (due to a
particular sector of the multipurpose project) and the remaining joint costs. Although no cost sharing rule
was specifically chosen the Egalitarian, the Alternate Cost Avoided (ACA) method and the Alternative
Justifiable Expenditure method most influenced the final allocation. A modification of the ACA method,
called the Separable Cost Remaining Benefits (SCRB) method gained acceptance during the 1940s and
the Federal Inter-Agency River Basin Committee (1950) recommended its utilization for the allocation of
costs in a joint water facility project.
The rules used indicate that "Expenditures made solely for a single purpose are to be charges directly to
that purpose. The following methods of allocation were considered by the committee" (Parker 1943):
Vendibility Theory (but there isn't a market where the facilities dealt with can be sold), Benefit Theory
(but there are difficulties in estimating the benefits derived from the project), Use of Facilities Theory
(that is, a method allocating costs upon the basis of comparative use of the common facilities, but it is too
difficult to estimate comparative use due to lack of data), Equal Apportionment ("common-sense rule of
equity to be used when it is felt that no truly scientific basis of apportionment can be found. However
such a rule does not seem practicable where the respective uses for each function are not equal"), Special
Costs (apportionment of the cost proportional to the special expenditures for each use, but it can also be
used for the joint costs), Alternative Justifiable Expenditure (the joint costs can be divided in proportion
to the alternative justifiable expenditure minus the direct cost for each function, where for direct cost we
mean the marginal contribution to the joint cost of a project and the alternative justifiable expenditure is
the lowest cost, justified by the benefits obtainable, of realizing the project outside the joint facility to get
the same benefits). The author concludes that "There is no real justification for basing the allocation ...
upon any one mathematical formula". In the meantime the evolution of Cooperative Game Theory
provided many new tools for the analysis of that kind of problems, from the Imputation Set and the Core
concepts, anticipated by some TVA criteria in Ransmeier (1942), to more "sophisticated" solution
concepts.
Straffin and Heaney (1981), made the first relevant contribution to this literature. They outlined some
basic CGT principles embedded in the TVA analysis and "translated into GT language" the main cost
apportioning methods. Driessen (1988) applies several CGT cost allocation methods to the case of the
TVA, comparing the -value, the Nucleolus, and the Shapley value, and similarly to Tijs and Driessen
(1986) demonstrates, using some theorems, that for certain subclasses of games the -value coincides with
the ENSC and with the SCRB methods. In a similar way, Young (1994a, 1994b) proposes the TVA also
as a rich example for a range of issues and cost allocation methods.
Straffin and Heaney (1981) applied solution concepts, including the core, a special case of the nucleolus,
and the imputation which minimizes the maximum propensity to disrupt. A method equivalent to the
latter, but with a different rationale, is now in standard use among water resource professionals.
The authors describe the TVA multipurpose problem, in a CGT model, as follows. N is the set of n
purposes among which costs have to be allocated, C(S) is the cost function for each subset S of N (S N),
and thus C(N) is the total cost and C(i) is the cost of a single project.
4
A cost-allocation is a vector:
(c1,...,cn), such that n
i=1ci = C (N ) and ci 0
The "preliminary criterion of a satisfactory allocation" originally listed by Ransmeier (1942) in his
presentation of the problem corresponds to the Core of this cost game:
ci Ci and c i C(S) for all S N
iS
If we consider the savings-game (N, v) corresponding to this cost-game, we will have:
v(S) = C (i) -C (S)
iS
These savings can be obtained by the subset S of projects.
We have:
v(i) = 0 and v(S T ) v(S) + v(T ) in the TVA case study. This is equivalent to say that v is zero-
normalized and superadditive.
The allocation to be chosen will now be a vector of savings: x1,..., xn where:
( )
xi = C (i) - ci
and, if we look for an imputation lying in the Core, it must be:
x n
i = v(N), xi 0 and x i v(S)
i=1 iS
for all S N.
The Core of the TVA game is non-empty (the game is convex), but it consists of a set of infinite points,
thus there is the problem of choosing between them.
Almost all the methods considered by the TVA charge each purpose a minimum cost called separable
cost, that is, the change in the cost of a multipurpose project due to adding a project to the set of the
others:
SCi = C (N ) -C (N -{i})
If the savings game is convex it follows that:
SC = SCi C (N )
n
i=1
and the allocation problem turns to be the allocation of the non-separable costs:
NSC = C (N ) - SC
A possibility is the equal allocation of the NSC among the projects:
ci = SCi + NSC
n
In terms of the savings-game,
5
xi = C (i) - ci = v(N ) - v(N -{i}) - 1 n
n (v(n)-v(N-{j}))-v(N)
j=1
The authors propose three numerical examples (three-person games) to show that the vector
E = (x1,..., xn ) (obtained as explained above), can be an imputation or not, can be in the Core or not, and
can even coincide with the Nucleolus, depending on the game.
Another method proposed by a TVA consultant, Martin Glaeser, in 1938, assigns to each project this
charge:
ci = SCi + Ci - SCi
C(j)-SC
n
j
j=1
This method attributes to a project its separable cost and a rate of the NSC proportional to Ci - SCi . This
difference is the Alternate Cost Avoided.
In 1974 Gately proposed a new solution concept for n-person cooperative games defining player i's
Propensity to Disrupt an imputation:
di = v(N ) - v(N -{i}) -1
xi
This quantity represents the ratio of what the members of coalition N - i would loose if player i disrupted
the grand coalition, to what player i himself would lose. Gately's apportionment method, based on
minimizing the maximum propensity to disrupt, is exactly the TVA ACA method.
The final TVA cost allocation was "not based exactly on any one mathematical calculation or
formula, the final sums being fixed by judgment and not by computation" (Parker, 1943). However the
ACA method was the principal method used to guide that judgment, and a modification of this method is
the allocation method which has gained widest acceptance among water resource engineers and is still in
use.
In this modification C(i) is replaced by min [B(i), C(i)]:
ci = SCi + min B(i) -C (i) - SCi
minB(j)-C(j)-SC
n
j
j=1
This is called Separable Cost Remaining Benefits method.
The authors conclude: "No single allocation method, or no simple game theoretic solution concept, can
claim to be uniquely "best". In this domain, rationality seems to demand a multiplicity of viewpoints, and
narrow insistence on the virtues of one method is a vice rather than a virtue."
In the following papers the subset solution concepts are used as a guideline, and the one-point solutions
are compared and discussed in order to choose an allocation, whenever the Core boundaries leave the
possibility of such a choice, that is whenever the Core is nonempty. In the literature about multipurpose
water project the CGT framework always consists an n-person cooperative game with side payments, and
the cost-sharing rules that we found applied in the literature are: the ENSC, ACA, SCRB, MCRS
methods, the Shapley value, the Nucleolus, and the -value (NSCG method). Most of the papers compare
6
numerical outcomes derived from the different allocation rules, and some authors enter to the equity and
fairness fields also, requiring some properties for a cost allocation rule.
In a paper that focuses on efficiency and equity criteria of cost allocation methods in water projects,
Loughlin (1977) analyzes the SCRB method, which assigns to each purpose the separable cost of
including it in the multipurpose project, plus a share of the joint costs, and the Alternative Justifiable
Expenditure method, which substitutes the specific costs of the various functions to their separable costs.
Loughlin refers to three efficiency conditions: "(1) The separable cost of adding each purpose as the least
increment should not exceed the benefits deriving there from, (2) the sum of the total costs allocated to
each purpose should not exceed the sum of the total benefits allocated to each purpose, and (3) the total
costs allocated to each purpose should not exceed the cost of a single-purpose alternative providing
equivalent benefits." These three efficiency criteria are satisfied by the SCRB method.
Equity of a cost allocation refers to fairness in the distribution of total project costs among all the
purposes served by a multi-purpose development. Specifically, an equitable cost allocation is one which
permits all project purposes to share in a fair way the savings from multi-purpose rather than single-
purpose construction.
On efficiency grounds, only the SCRB formula satisfied all three efficiency criteria. From an equity
viewpoint, no one method allocates costs so that each purpose shares fairly in the savings from multiple-
purpose development. The source of inequity with the SCRB and alternative justifiable expenditure
methods arises because either separable costs or specific costs are subtracted from the justifiable costs on
a 1:1 basis. Conceptual problems in defining capacity preclude the use of facilities methods from
satisfying the equity objective. An adjustment in the SCRB method is proposed which will enable this
procedure to satisfy both efficiency and equity criteria.
The author introduces an "adjusted" SCRB method, to avoid inequity: he proposes to "apply a credit to
the separable costs so that separable costs are subtracted from justifiable costs on a grater than 1:1 basis.
The credit could be in the same ratio as that of the justifiable costs for a purpose plus justifiable costs for
all other purposes to the total project costs. This procedure would provide better results than the existing
SCRB method for meeting the equity criterion. (...) The effect of this change would be a tendency for the
savings allocated to each purpose to bear a more proportionate relationship to the savings from inclusion
of each purpose in the project."
Dickinson and Heaney (1982) suggest an alternative allocation method, the minimum costs, remaining
savings (MCRS), as a generalization of the presently used separable costs, remaining benefits method.
The purpose of the paper is to address some fairness criteria corresponding to the Core of an n-person
TU-game. These criteria reduce the set of admissible solutions, but additional criteria are required to get
a unique solution. The method is introduced by a numerical example, and later formalized.
It is shown that the SCRB method works fine if the Core is nonempty: in the case of an empty Core the
SCRB leads to a "not very reasonable compromise solution". A way to find a compromise solution is to
relax some Core bounds until a Core appears and one way is to relax the bounds on the intermediate
coalitions S.
A linear program problem can be formulated:
min, subject to :
x(i) c(i) i N
x(i)- c(S) c(S) S N
iS
7
x i= c(N), x(i) 0
iN
The meaning of these equations is, in the case of an empty Core, to find the minimum value which
yields a nonempty Core: the new bounds will provide a point Core. In the authors' opinion this method is
"a more general procedure for allocating costs which is felt to be theoretically sound, computationally
simple and easy to relate to existing practice".
Driessen and Tijs (1986) compare four different allocation methods, which allocate the joint costs of
water resource projects among its participants on the basis of separable and nonseparable costs: the
egalitarian nonseparable cost (ENSC) method, the separable cost remaining benefits (SCRB) method, the
minimum cost remaining savings method, and a new method, the so-called nonseparable cost gap method,
which is derived from the -value, a game theoretical concept (Tijs, 1981).
In the SCRB method the separable cost is attributed to each player, and the allocation of the remaining
non-separable costs is based on the alternate cost of the one-person coalition. However there is no reason
why the remaining alternate costs of other coalitions should not be taken into account in the allocation of
the non separable cost.
The authors introduce the cost gap function gc , assigning to any coalition S the remaining alternate cost
of S given by:
gc (S) = c(S) - SCj (c)
jS
Note that:
gc(N) = N SC
Let gc = 0 .
( )
In general, for the games dealing with multipurpose situations we have:
gc (S) 0 S N .
If we consider a player i T , we can imagine that the player will reject an allocation that charges him
more than SCi + gc T : here the other players in the coalition T are charged only by their SC, while
( )
player i is charged the remaining cost (worst situation for i). This is true for each coalition he belongs to,
thus he won't be paying more than
minT:iT SCi (c) + gc (T ) = SCi (c)+ minT:iT gc (T )
If we define the concession amount for player I as
i(c) = minT:i gc(T)
T
then the non separable cost gap method allocates the NS costs among the players proportionally to i . (If
the one-person coalitions determine the minimum gap, then the NSCG method coincides with the SCRB.)
Without entering into details the authors note that the Cost Gap method has "nice" properties such as
efficiency, individual rationality, symmetry, continuity, the dummy player property, the dummy out
property, the strategic equivalence property and the complementar monotonicity.
The application of this method was demonstrated by Young et al. (1982), using the three neighboring
municipalities dealing with a water supply problem.
8
After this description of the Non Separable Cost Gap (NSCG) method, it is compared to the SCRB
method (the two allocations coincide if the cost game is semi-convex) and to the MCRS method. For
convex cost games the three rules use the same bounds for the Core and consequently their outcomes
coincide, but there is no guarantee for these allocations to be inside the Core of the game.
The NSCG is also compared with the ENSC. This allocation rule, which divides the non-separable
costs equally among the players/sectors, should be preferred in a special subclass of games called one-
convex cost games. For these games, in fact, the resulting cost allocation lies in the centre of gravity of the
Core. In this situation the ENSC coincide with the SCCG and MCRS methods, but does not with the
SCRB.
In a separate attempt, Owen (1993) provides two characterizations of the alternate cost avoided method,
one on a class of cost games with a fixed player set, and the other on a class of cost games with a variable
player earlier two of the methods examined by the TVA in the 30's, while facing their cost allocation
problem for the multipurpose water resource project, were the Egalitarian Non-Separable Cost (ENSC)
method, and the Alternate Cost Avoided (ACA) method. A modification of the ACA method is the
Separable Cost Remaining Benefit (SCRB) method.
A game theoretical basis for the ACA method was first proposed by Gately (1974), with his new
concept of players' "propensity to disrupt" a coalition. Owen (1993), after an overview in a GT
framework of some of the TVA cost allocation methods, provides an axiomatic characterization of the
ACA method for a certain class of cost games with a fixed player set (CGN), and for a class with a
variable player set, using a reduced game property. The author demonstrates through a theorem, in a
similar way to the characterization of the -value given by Tijs (1987), that the ACA method is the unique
cost allocation method satisfying invariance w. r. t., strategic equivalence and weak proportionality
properties.
In a recent paper, Serghini (2003) Introduces the FDC (Fully Distributed Costs) allocation rules, and
applies it with the Shapley value and the Nucleolus to multipurpose dam projects in Morocco. More than
the half of dams in Morocco are multipurpose (hydroelectric power, irrigation, urban supply and,
sometimes flood prevention). The entire infrastructure is owned by the ministry of public works, which
manages also the dams. The irrigation and urban water supply facilities are managed by the ministry of
rural development and water. As can be imagined, disputes over timing and tariffs are common. The
author describes two methods for apportioning the costs: the Fully Distributed Costs (FDC) methods,
assigning the cost of the service plus a fraction of the joint costs and the main apportioning methods
based on applied and interpreted in a CGT framework modeling for a multipurpose water resource
project.
A case study (the Dchar El Oued dam for hydroelectric power, irrigation, urban and industrial supply)
is analyzed. The author explains the allocation mechanism that has been chosen for the cost of the dam.
According to the World Bank report (1995) concerning the water sector in Morocco, the urban fees have
to cover all the costs, including those due to mobilizations. This will imply an increased cost charged to
the users. Choosing a cost allocation method in a multipurpose project is not easy, but it allows the
application of an acceptable and equitable method for all the users. Serghini underlines that the choice of
a rule has different consequences on "small" and "large" users: for example the "small" services take
advantage of non separable cost allocation proportional to the rate of consumption and Nucleolus, while
the "large" ones "prefer" the Shapley value, the ACA and the SCRB methods. Among the ten methods
applied to the project, the Shapley value seemed to be the most suitable.
URBAN WATER SUPPLY
The following two papers deal with three distinct cost allocation steps for an urban water supply
system. The problem consists of three steps, namely (1) aggregating the players into groups to keep the
9
problem manageable; (2) finding an allocation among the players in each group; and (3) finding an
allocation scheme for each type of use (e.g., large homes, small homes) within each municipality.
Hashimoto et al. (1982) using a case from Sweden, demonstrate two of these steps. Eighteen Swedish
municipalities wishing to develop an urban water supply system are divided into 6 groups, each
considered as single agents; the second step is only mentioned in that paper, and it consists of a cost
allocation among the single municipalities within each group. The third step is demonstrated by Lippai
and Heaney (2000), and concerns urban fees for different kinds of users. The peculiar aspect in these
applications of CGT to water supply systems is just the presence of `natural' groups considered as single
agents in the models.
At the highest level (groups of municipalities), the cities are naturally bent together because of
logistical considerations (geography, hydrology, existing networks...), at the lowest (single-family users)
the charges are drown for different categories of users on the basis of the quantity and quality measures of
water consumption.
Hashimoto et al. (1982) address the notion of `fairness' within the comparison of the `classical'
allocation methods to the CGT ones: fairness in this case is associated with the concept of the Core of a
TU-game (individual rationality, group rationality, efficiency). The authors' conclusion is that even if
CGT methods need more information and data, they are less arbitrary than the non-CGT ones, and thus
they can be more suitable: for example in the SCRB method the boundaries between direct and indirect
costs are not so well defined. The application of `classical' cost apportioning methods does not guarantee
individual and group rationality, while for CGT methods some problems may arise because of the non-
monotonicity of many solution concepts.
The authors' analysis consists of two parts. First, they discuss certain established principles of
`fairness' through which the different methods can be judged. Then, they compare the solutions given by
each method in the case study. The focus is on the behavior of these apportioning methods in practice,
rather than on a strict and theoretical axiomatic characterization. The CGT concepts applied are the Core,
the Least Core, the Nucleolus, the weak Least core and the weak Nucleolus, the proportional Least Core
and the proportional Nucleolus, and they are compared with some `classical' apportioning methods, such
as the SCRB method and other methods allocating costs in proportion to some single numerical criterion
(e.g. use, population, benefit etc.).
While defining the cost function, the first problem in this case study was to identify the independent
actors in the system: 18 municipalities were involved, but only some of the possible 218 -1 groupings
being realistic. Some `natural' groups could be identified, on the basis of past associations, geographical
location, existing water supply network and hydrological and geographical features: the cities have been
consequently grouped into six independent units, according to these considerations.
The apportionment of costs can be done in two steps: at first among the six groups and later among the
single municipalities. (Note that considering the single municipalities since the beginning might not give
the same results.)
Some uncertainties in the definition of the cost function arise, since "the definition of the cost function is
always somewhat arbitrary in practice".
Using the developed cost function, the cost allocations given by the different methods described in the
previous section are compared. The results of the proportional allocation methods (proportional to water
demand and to population) and those given by the SCRB don't satisfy the requirement of individual
rationality and/or group rationality. Furthermore, "in practice the borderline between direct and indirect
costs is not always clear".
In the numerical comparison of the different outputs, the authors note "the `proportional' allocations
differ markedly from the others".
10
"Since group rationality and marginal cost coverage seem to be essential from the standpoint of equity,
that is, to provide sufficient incentives for cooperation, the remaining three methods, the nucleolus, the
weak nucleolus, and proportional nucleolus, are potentially more desirable than those discussed above, as
they always produce a Core imputation if one exists."
A method should also be adaptable to changing conditions (since Ransmeier, 1942), that is, for example,
if total costs increase then no participant should be charged less. This property is called monotonicity and
several of the methods considered do not satisfy it. This may cause problems because total project costs
are not well defined until the project is completed, and the first agreement will be based on an estimate:
thus it becomes fundamental for the method to `follow the changes' in a reasonable way. `Proportional'
methods are obviously monotonic, and the same holds for the Shapley value, the weak and the
proportional Least Core, the weak and the proportional Nucleolus.
Among the `classical' methods the proportional one is considered preferable by the authors, while
among the CGT ones the proportional Nucleolus seems to be the best: first of all it is monotonic, second
the authors believe that perhaps "A reasonable scheme would be to divide any unforeseen costs in
proportion to the benefits enjoyed. This is precisely the way in which the proportional Nucleolus work."
Appendix A, at the end of their paper carefully describes the procedure used to determine the capital
costs of pipes and pumps necessary to supply the demand of water; Appendix B demonstrates that some
of the methods satisfy the monotonicity property.
A similar assessment is made by Lippai and Heaney (2000) where CGT allocation solutions are
considered more useful than the `classical' ones in getting fair charges for the users. The purpose of their
paper was to present a method for the determination of efficient and equitable impact fees for urban water
systems for each user, based on the type of demand on the system. The paper deals with the importance
of assuring a fair assessment of impact fees for an urban water supply system: the cost of the project
needs to be equitably allocated among all existing and new users.
Nelson (1995) described various ways to determine System Development Charges (SDCs) for water,
wastewater, and stormwater systems: Market Capacity method, Prototyp System method, Growth-related
Cost Allocation method, Recoupment Value method, Replacement Cost method, Marginal Cost method,
Average Cost method, Total Cost Attribution method. Applying these methods to a case study Nelson
showed that the deriving charges are distributes on a wide range. None of the mentioned methods takes
anyway into account the presence of different classes of users or demand types.
Crossmit and Green (1982) identify an ordering for water uses and thus constructed the charges with an
incremental cost allocation scheme: the costs are first calculated for a project serving the primary
purpose, then the second purpose is added and the difference in the costs for the two different projects is
allocated to the second one, and so on. However the definition of a purpose as primary or secondary
seems very arbitrary.
Crossmit and Green (1982) propose an allocation of costs among users based on certain CGT principles,
and first of all enunciate the properties of the Core of an n-person game.
The general charge for the i-th user is:
x(i) = x(i)min + (i)c(N ) - x(i)min ,
n
i=1
where i ( ) represents the rate of the remaining costs c N -
n
( ) x(i) that the user should pay,
min
i=1
i N and N = {1,...,n}. We have (i)=1.
iN
11
The minimum cost x i ( )min can be considered equal to 0, to the direct or specific costs or to the separable
costs (James and Lee, 1971), but can also be identified with the minimum charge for i defined by the
Core bounds.
(i) can be (James and Lee, 1971): (1) equal for all the users, (2) proportional to a measure of use of the
water, e.g. volume, (3) entirely attributed to the highest priority participant, (4) proportional to the benefit
in excess of assigned minimum cost, (5) proportional to the excess cost to provide the service by some
alternative means, (6) proportional to the minimum between (4) and (5).
Considering each user independently would require the evaluation of a very large number of different
water supply systems, but the number of combinations can be reduced by replacing individual users with
(1) zones of service areas within the boundaries of water district; (2) user classes with similar demand
characteristics; and (3) demand types. The cost function can be consequently derived, even if data are not
so easy to collect.
The authors consider a hypothetical example and apply their method for equitable and efficient impact
fees for urban water systems in Almos. The players are grouped into medium (M) and low (L) density
single-family users, and a warehouse (W), each of them considered as a single agent. Some CGT concepts
are used in order to allocate costs among these users. The Core of the three-person game in question is
nonempty, and the means between the minimum and maximum Core bounds for each player are selected
to assign the costs. To construct fair charges based on the true cost of providing service some
considerations on the use of water have to be added to the model (for example indoor use, outdoor use
and fire use for each group of users).
IRRIGATION
In Nakayama and Suzuki's paper (1976) we find a typical cost allocation problem for a joint enterprise
(agricultural urban water supply) and the proposed model is a TU-cooperative game where the
Nucleolus seems to offer a reasonable solution as a cost-sharing rule. This work has a certain relevance,
considering it is one of the first applications of the CGT tools to a water resources management project.
In more recent literature we find some NTU models also.
In the proposed model, the N players of the game are partitioned into two subsets: A, consisting of
agricultural associations, and B, consisting of city services. Each city i in B requires an additional annual
quantity of water i (i = 0 if iA), and can choose among: (1) constructing a dam with or without the
cooperation of other cities and (2) making an agreement with the agricultural associations for the direct
diversion of water from them to the city or (3) a combination of (1) and (2).
A cooperative game is constructed and the concept of Nucleolus is introduced as a `fair' division of costs
among participants in a joint project.
The proposed case study is about Kanagawa prefecture in Japan, where a dam will be built and costs will
be allocated among two agricultural associations and three city services:
N = {1,2,3,4,5}
A = {1,2}
B = {3,4,5}
The cost function is determined through a linear programming problem, and the question turns to how to
determine a fair apportionment of v(N) among the different players.
The authors require for a solution the properties of (1) efficiency, (2) individual rationality and (3) group
rationality, and introduce the Nucleolus: "However it might be that no allocations satisfying (1) and (2)
12
also satisfy (3). This being so, any `fair' allocation will simply have to come as `close' to meeting (3) as
possible. One notion of `close' leads to the Nucleolus of a game (...)".
The case study is considered and the Nucleolus of the Kanagawa Prefecture game is calculated solving a
linear programming problem. The authors don't comment on the numerical results of their paper.
Ratner and Yaron (1990) examine the consequences of the choice of a TU- or an NTU- framework: the
possibility of side payments has a relevant political, economic and social impact on the players. They
apply their analytical framework to a regional cooperation among farmers dealing with water quality.
The conclusion is that even if regional cooperation with side payments could provide a better increase in
the income, in practice it is preferable to aspire to the NTU gain, which is less than the amount under the
transferable utility hypothesis. The reason is that the farmers would not accept the side payments policy,
where a share of the income generated by the most efficient farm should be transferred to another one;
beside this concern, another question is which part of this gain has o be transferred? It is a hard work to
find an allocation of the surplus of the gains that are able to get a certain acceptance among the players.
The Nash-Harsanyi approach seems to be the most appropriate for the conditions studied.
Water quotas, whose allocation was often stated long ago, may now prove to be inefficient.
An interesting problem arises when the regional authority gives a region the opportunity of increasing the
salinity of the water supply to the region and to be compensated by an increased quantity: the
determination of the optimal mix, when different regions have different desired Quantity-Salinity (Q-S)
mix, is the aim of this paper.
The water scarcity problem, together with inefficiency issues emphasized the need for inter-farm and
inter-regional cooperation for water mobility.
Once determined the maximum income for the region, through a linear programming model, two possible
income distribution policies are considered: with or without side payments.
In the model without side payments the cooperation is restricted to the exchange of water quotas; in the
other one direct income transfers are allowed.
The models discussed are applied to the analysis of a potential cooperation among three farms in the
Negev area of Israel. The allocation of the gain, under the TU-cooperative model, is computed through
the Nucleolus, the Shapley value and the shadow cost pricing, while the situation without side payments
is modelled with a Nash-Harsanyi approach.
Whereas a priori the shadow pricing method can seem reasonable, some problems occur because the price
of water for the farmers is often subsidized and even lower than the shadow price.
In the NTU-framework the problem is to identify a point on the efficiency frontier and the authors
follow the approach proposed by Nash (1950) for two players and generalized by Harsanyi (1959) for n
players with n > 2. This approach consists in the maximization of Z (where yi denotes income to farm i
and yi is the income for the same firm, without the cooperative agreement):
0
n
Z = (y )
i- yi .
0
i=1
Some assumptions (corresponding to the axiomatic characterization of the Nash-Harsanyi solution)
underlie this approach: joint efficiency, invariance with respect to linear utility transformations, symmetry
and independence of irrelevant alternatives.
A quasi-empirical application of the different models is proposed for three farms in the Negev Area of
Israel: the authors explain that all the data are real, except for the `boundaries' of the region.
13
The results under the different assumptions are presented and compared: the authors conclude that an
increase of 17% could occur to the regional income though cooperation with side payments; on the
contrary the region can get an increase of 8% if cooperation does not involve side-payments.
On the other hand some difficulties may arise from the side payment policy (see introduction), while the
model without side payments seems to be more reasonable: its assumptions are simple and easy to
understand, and for this case study it is computationally simple.
If compared with most CGT application literature Aadland and Koplin follow a mirror-like approach
in their two following papers (1998, 2004). Their analysis deals with some real-world cost-sharing rules
for irrigation (they consider the case study of 25 irrigation ditches in Montana) and they try to see these
rules under a game-theoretical point of view, providing in the first paper an ex-post axiomatic
characterization. The mathematical tool does not prevail over reality: it is at its complete disposal. The
problem in the use of axiomatic approaches and of the deriving solution concepts is that a reasonable
axiom can be too restrictive or even cause unreasonable outputs for certain real situations. `Fairness' also,
on the authors' opinion, has to be intended as long-term acceptance by people judgment. In their second
paper Aadland and Koplin, while reexamining the same case study ask to themselves if there is a
correspondence between some environmental aspects and the choice of a certain allocation rule by the
farmers; they try to answer with the aid of an econometric model.
In his paper Dinar (2001) presents a cooperative bargaining model with NTU, and uses the Nash
solution because "The allocation is feasible, fulfills Individual Rationality, is Pareto optimal, symmetric,
non related with nonrelevant alternatives and to a specific presentation of the VNM utility function." The
peculiarity of Dinar's model is that it allows an optimization of the capacity of the joint facility
simultaneously with the cost allocation procedure.
The article by Norde et al. (2003) deals with another branch of CGT--fixed tree games. In some
irrigation systems it can be helpful to model the situation as a tree where the farmers are located on one
(or more in the case of FRP games) vertex, and it is possible to construct a corresponding cooperative
game. The paper is almost theoretical but it is presented to give the reader an idea of a different CGT tool,
which can be employed in modeling some irrigation water supply networks.
A work by Aadland and Koplin (1998) examines cost-sharing arrangements in a sample of twenty-five
irrigation ditches in south-central Montana. These arrangements--variations of serial and average cost
sharing--often date back to the corresponding ditch's construction at the turn of the century. The authors
present an empirical and axiomatic analysis of the allocation mechanisms employed by the ranchers in the
region, using a list of static and dynamic equity principles as was revealed by direct phone surveys. The
analysis shows that the observed cost-sharing arrangements can be characterized as unique solutions to
equity-constrained welfare maximization.
The cost-sharing rules dominating the case study, the serial/Shapley and average cost-share rules, have
been kept surprisingly stable, suggesting that they are really perceived as `fair' by the players. This
motivates the authors' efforts to construct characterizations of the sampled mechanisms based exclusively
on principles of equity.
While collecting the data, the authors directly made phone calls to one or more irrigators along each ditch,
asking about (a) details regarding the cost-sharing arrangements currently in use, and (b) general feeling
about what can be considered to be fair methods of sharing irrigation costs.
Two classes of sharing rules dominated: average cost sharing (egalitarian) and serial cost sharing. The
average cost-sharing rule simply takes the total cost of ditch maintenance and divides it equally amongst
all users. The serial cost-share rule divides the cost of maintaining each ditch segment equally among all
agents that require the use of this segment to meet their irrigation needs. As in the `airport games' of
Littlechild and Owen (1973), the allocation offered by the serial rule is identical to that prescribed by the
Shapley value of an appropriately defined cooperative game. Per acre and per water-share weighted
14
variations of restricted average and serial rules were also found in the case study, and axiomatically
characterized.
Dinar (2001) introduces into an allocation game of a joint groundwater development both the size and the
equity considerations, in a stylized example. The paper presents an application of the Nash model to a
combined problem of optimal development of a joint facility (determination of the optimal capacity) and
related cost-benefits allocation considerations. Two alternative modifications and a diagrammatic
procedure are developed. The conditions for an optimal Nash solution are derived. The paper proposes a
procedure allowing to deal simultaneously with capacity optimization and cost allocation. The Nash
solution to a cooperative bargaining problem can fit this kind of situations, giving an axiomatic unique
allocation to a cooperative problem between two players. (The Nash model has been generalized for n
players by Harsany (1959, 1986). Two alternative Nash models are developed: Model A is the classical
Nash model for a two-person bargaining game; Model B follows the modification proposed by Harsanyi
and, even if in the case study there are only two players, it can provide an efficient solution when more
than two participants are involved also.
The paper shows that the two models produce similar solutions, under a similar set of choice variables.
The hypothetical case study deals with two agricultural producers wishing to cooperate for a new water
source, e.g. a well; each farmer has his own water supply limit, demand, and technology. By including in
the algorithm a negative service fee, the model allows the transfer of income from one player to the other;
in this case the degree of economies of scale is such that it provides the other player with the appropriate
level of incentives for cooperation.
Hamers, Miquel, Norde, and van Velzen (2003) apply CGT to a problem of allocation of maintenance of
an irrigation system. They introduce fixed tree games with repeated players (FRP games), which are a
generalization of standard fixed tree games. This generalization consists in allowing players to be located
in more than one vertex. As a consequence, these players can choose among several ways of connection
with the root. The authors show that FRP games are balanced. They prove that the core of an FRP game
coincides with the core of a related concave fixed tree game, and show how to find the nucleolus and we
characterize the orders which provide marginal vectors in the core of an FRP game.
The FRP game in this paper is a generalization of the fixed tree problem, introduced by Megiddo (1978),
where the players may be located in more than one vertex of the given tree. A fixed tree problem consists
in a rooted tree and a set of agents N, each agent being located exactly at one vertex of and each
vertex containing only one agent. Megiddo (1978) associated this problem with a cooperative cost game
(N, c), called a fixed tree game, where, for every coalition S N, c(S) denotes the minimal cost needed to
connect all members of S to the root via a subtree of .
As a motivation for this generalized model one can consider for example the following irrigation problem.
Consider a set of parcels in a desert environment, which need to be irrigated from a well. For that reason,
a network has been designed which allows the conveyance of water from the well to at least one of the
corners of each parcel. Consider the situation with eight parcels in figure 1.
15
Figure 1: The Irrigation network in a Perimeter
Bold lines indicate the network that has already been made up. The players are facing the problem of
dividing the maintenance costs of this network, so they are facing a fixed tree problem where the tree is as
depicted in figure 2 below.
Figure 2: The fixed tree irrigation problem
The concept of Core of a cost game is recalled, and constructed below.
A fixed tree problem with repeated players (FRP), in the game described above, is a 5-tuple:
= N,(V,E),0, S v
( ( ( )) , a e
( ( )) ), where:
vV eE
N = {1,2,...,n} is a finite set of players.
(V,E) is a tree with vertex V and edge set E.
0 is a special element of V called the root of the tree.
S (v) N is for every vV a (possibly empty) subset of players occupying vertex v.
a(e) > 0 is for every edge e E the cost associated with edge e, which satisfies the following
assumptions:
(A1) for every iN there is a vV with iS(v)
(A2) for each leaf tV there is an iN such that iS(t)
and iS (v) for every vV \{t}
The authors show that FRP games are balanced and they focus on the structure of the Core. (Standard
fixed tree games are known to be concave, consequently these games are balanced. The Core of these
games is not empty.)
While the papers discussed so far assume that the cost allocation schemes operate in neutral
environments, one knows that this is not always the case and that the environment affects the
16
effectiveness and stability of the allocation schemes. Aadland and Koplin (2004) use shared irrigation
costs as a context for examining the extent to which the structural environment explains the selection of a
cost sharing rule. They find that environmental factors that - induce greater dependence on the
cooperation than others, influence majority interests, create difficulties for interpersonal utility
comparisons, or impact notions of "fairness" - all have impressive explanatory power. In this work, as in
Aadland and Koplin (1998) examine some `real-world' cost-sharing rules, trying to understand how the
environmental context can influence them and make them more or less stable.
The case study and thus the examined cost-sharing rules are the one of the previous paper, but here the
authors better enter into details of the data and collecting data instruments and try to answer some further
questions.
Multiple sharing mechanisms can often be observed coexisting harmoniously in seemingly identical
environments. Does such behavior suggest that several sharing mechanisms are equally appropriate and
the choice of which to adopt is essentially arbitrary? Or do environmental cues exist that reliably reveal
the sharing mechanism of choice? If such explanatory power can be empirically established, does it
support or run counter to the central spirit of axioms employed in the theoretical cost-sharing literature?
The average class rules distribute total costs equally per capita, per acre, or per shares of stock owned.
The serial class rules share total costs either serially per capita, per acre, or per shares of stock owned or
agents pay only for costs incurred on their property. The serial class contains those rules that protect
agents from the burden of costs incurred downstream from their property, a principle the authors refer to
as excess demand protection (EDP).
An econometric analysis reveals that the choice of a rule is not arbitrary and can be explained through a
reduced form of the structural environment. Some peculiar environmental aspect can make EDP more
(less) compelling, and the rules in the serial (average) class become more likely to the players.
The developed econometric model assumes that the farmers choose between the two classes of rules
depending on structural environmental characteristics, under this formula:
Yi = Xi +i
where:
Yi = latent variable measuring the likelihood of choosing either the serial or average cost-sharing class of
rule for ditch i,
Xi = row vector of explanatory variables for ditch i,
= column vector of coefficients
i = normally distributed error term with mean equal to 0
The authors conclude that the estimated coefficients are nearly all statistically significant and the signs are
generally in agreement with expectations. Moreover, the overall fit of the model is excellent.
HYDROELECTRIC POWER
While hydroelectric plants are a very important endavor in the water sector among political
jurisdictions (states, provinces) sharing the same water source, and sectors using the services of thes
plants, the literature is suprisingly poor in CGT applications. Gately (1974) provides a very
comprehensive example of the allocation of both the electricity and its cost of production among Indian
states that can potentially share hydroelectric plants. Water is quite marginal in this article, but the paper
17
introduces the new concept of players' propensity to disrupt a coalition, recalled by many other authors in
CGT literature.
The paper focuses on the problem of determining a mutually acceptable division of the gains from
regional cooperation, with reference to planning investment in hydro-electric power in the Southern
Electricity Region of India. It calculates the costs of providing power to the region of the case study under
different assumptions about the degree of cooperation and considers some mutually accepted cost
allocations. The region has been divided into three areas, and the five possible coalition structures
(corresponding to different degrees of cooperation) are considered: complete cooperation among all
states, self-sufficiency for all states and partial cooperation among the states.
The difference between an area's costs under self-sufficiency and its costs under a given coalition
structure states its payoff from that coalition structure and provides the characteristic function. When this
difference is negative it would be irrational for that area to cooperate. The GT concepts of imputation and
Core are introduced to characterize a vector of final payments.
The Core of the game is non-empty, and the idea of "propensity to disrupt" is introduced to select an
imputation inside the Core. For any proposed imputation, area i's propensity to disrupt the three-area
agreement is the ratio of how much the other two areas would lose if area i refused to cooperate to how
much it would lose if refused to cooperate. Even after employing the disruptive propensity concept to
eliminate those core imputations where only one area's final payment is "too small" there still remains a
wide range of mutually acceptable ways of sharing the gains of cooperation.
The paper examines the possibility of (1) Equal shares, (2) Equal ratio of total final costs to costs under
self-sufficiency, (3) Kernel, (4) Demand weighted shares, (5) Shapley value and (6) Equal propensity to
disrupt. The first two cost allocations are not in the Core; the third, defined by the author "naively
`equitable'" does not take into account the different bargaining power of the areas; (3) and (4) can provide
a Core allocation, but some area's propensity to disrupt might result too high; (5) for the case study is in
the Core and gives small propensities to disrupt; (6) leaves each area equally satisfied, in terms of
propensity to disrupt. Among the considered cost allocation schemes the Shapley value and the one with
equal propensities to disrupt can be considered mutually acceptable.
WATER POLLUTION
In this section we include some CGT application on pollution, and specifically pollution control and
prevention. Cost allocation for polluted water treatment will be dealt separately. We made this distinction
because it can be useful to see how different approaches to the problem of water pollution, can be
managed with cooperative game theoretic models.
In the paper by Kilgour et al. (1988) a water management authority fixes a maximum allowed load for
COD (Chemical Oxygen Demand) for a certain water body, and later distributes "COD quotas" among
the dischargers. Thus, the allocation is of the pollution allowance agreed among the polluters (cities), and
they have an economic incentive for cooperation in COD reduction (by joint management and projects).
A new regulatory system for water quality is described, analyzed and illustrated with its main properties.
BOD (Biochemical Oxygen Demand) and COD standards, like pH, suspended solids and phosphorate
phosphorus can be relevant quality criteria. However, for other toxic pollutants separate and often
specific methodologies are required. Furthermore, empirical evidence suggested that a criterion based on
the total amount of pollutants should be more useful than conventional concentration limits: this solution
can be called "load control" to distinguish it from "concentration control". (This is reasonable, for
example, if we deal with `closed' water systems where the load effect is more relevant.)
Finding the right quality criteria is not sufficient. The `fairness' of the division of total load among
polluters can become a difficult social and political point. Cooperation may provide a reasonable
18
solution. The load function is approximated by a formula that fits most of the COD data with a
reasonable accuracy, and it is shown that cities can very often reduce load more efficiently by sharing the
effort.
A general system for load control is described: the central authority can reach its goal on a certain COD
load by selecting, for each city i, a proportional quota pi satisfying:
n
pi 0 for all i N, and p =1. Further, the central authority must require that each city spend
i
i=1
an amount xi, so that v S = -min x :LS x ApS
( ) { ( ) } (where A is the maximum load accepted and L is
the load function).
The cities, as the authors show, will tend to cooperate to reduce load efficiently. We can define the quota
for a cooperating group of cities, or coalition pS = p , and thus, the requirement seen above can
i
iS
become LS xS ApS .
( )
Now the questions are: which coalitions will form? How should costs be split within these coalitions?
The CGT model may answer these questions.
The characteristic function of a COD allocation game is v S = -min x :LS x ApS
( ) { ( ) } and the game
is superadditive. An equivalent model, considering the savings game , has the following form
characteristic function: w S = v S - v i
( ) ( ) ( ) for S . For this game the Shapley value and the
iS
Nucleolus are considered as possible solutions.
In a different paper, Dinar and Howitt (1997) conduct a pollution control analysis focusing on drainage
water treatment. Different allocation schemes (proportional allocation, Nash-Harsanyi, allocation
according to marginal cost, Shapley value, the nucleolus and the Separable Cost, Remaining Benefit
principle) are used to allocate regional joint environmental control costs under two extreme states of
nature scenarios, resulting in different pollution flows. Two different states of nature are included in the
model: high flow drainage and low flow drainage in the Central Valley of California. They are been
separately taken into account in order to understand what happens in terms of stability and acceptability
of a solution if some natural changes occur. The results provide clear empirical evidence that regional
arrangements may vary with state of nature. Another important result is that the different allocation
schemes have different outcomes in terms of their acceptability of the players, and in terms of their
derived stability, as measured by the Shapley-Shubik Power Index, and by the Propensity to Disrupt.
The case study is about the joint treatment of irrigation drainage water by several water districts in
California and the allocation of the joint costs. The authors introduce a cooperative game with side-
payments in characteristic function form. A necessary condition for an allocation to be accepted by the
players is the fulfillment of the Core conditions.
Additional conditions are the stability of the coalition One measure used to assess the stability of an
allocation scheme is i , comparing the gains to a player with the gains to the coalition.
i = xi -v(i)
(x ), i N, =1
i
j-v( j) iN
jN
19
Acceptability and stability are tested in two states of nature: high drainage flow and low drainage flow. In
fact, the changes in water quantity to be treated have to be incorporated in the allocation scheme
otherwise the solution might not encourage cooperation.
Another set of papers, (Jorgensen and Zaccour, 2001; Fernandez, 2002) deal with cooperation in
differential games for a problem of pollution. In the first paper, roughly speaking, some properties are
required to the characteristic function and to a side payments scheme on a time interval 0,T . In the
[ ]
second one Fernandez models a transboundary pollution situation between USA and Mexico and analyzes
the consequences of cooperation and non-cooperation in different scenarios of trade liberalization.
Jorgensen and Zaccour (2001) determine a rule to share a surplus gained when two countries or regions
agree to coordinate their policies to reduce downstream pollution. An intertemporal decomposition
scheme for the total side payment is proposed. This scheme has the following individual rationality
property: in each subgame that starts along the cooperative trajectory, one country is guaranteed to
receive a higher payoff in the cooperative solution than in the disagreement solution. For this country
another notion of individual rationality holds. It will at any time during the play of the game receive a
higher payoff in the cooperative solution than in the disagreement solution.
In a case of downstream pollution a country/region suffers from the pollution generated by its upstream
neighbor(s). The authors suppose that the two countries/regions wish to make an agreement in order to
maximize the sum of their respective payoffs. The surplus given by cooperation is the difference between
the optimal joint payoff and the sum of the disagreement payoffs.
To distribute the surplus among the regions, side payments will be necessary and the authors require a
scheme with some desirable properties in terms of individual rationality. (The assumption is that the side
payment for the entire duration of the agreement is not paid as a lump sum: rather it is distributed as a
continuous stream over time.)
Three different notions of individual rationality are considered: (1) overall individual rationality, (2)
individual rationality in every subgame along the cooperative trajectory (time consistency), and (3)
individual rationality at any instant of time.
If we denote 0,T
[ ] the time interval of the agreement, (1) means that "for the entire duration of the
agreement, in the cooperative solution each region receives a payoff which, after side payments, is no less
than the disagreement payoff of the region"; (2) is individual rationality over ,T
[ ] for any [0,T],
that is, time consistency of a cooperative solution (Petrosjan, 1993,1997); and (3) is individual rationality
at any t 0,T : at any instant of time the instantaneous cooperative payoff, after side payments, of each
[ ]
region is not less than the instantaneous disagreement payoff of the region.
Considering the case of two regions, Region 1 and Region 2, called respectively vulnerable and non-
vulnerable region, according to Kaitala and Pohjola (1995), where Region 1 is only affected by its own
pollution, while Region 2 is affected by pollution of both countries, whereas property (1) is easy to satisfy
for both regions, they could only guarantee, in general, the satisfaction of (2) and (3) for the non-
vulnerable Region 2. (Under specific circumstances, however, (2) and (3) can be satisfied for the
vulnerable region, too.)
Starting from a non-cooperative scenario, Jorgensen and Zaccour, 2001 draw an intertemporal side
payments scheme based on the egalitarian principle (to divide the surplus equally among the regions).
Fernandez (2002) develops a differential game to examine the effects of trade liberalization on
transboundary water pollution. Water pollution is due to wastewater emissions from countries in a shared
waterway. The case in question is the U.S.-Mexico border. Noncooperative and cooperative games are
examined with changes in trade policy and public health damages. Results show trade liberalization leads
20
Mexico to curtail pollution in both games. Cooperation and trade liberalization limit emissions from both
countries and curtail strategic behavior of the United States from Mexico's pollution control efforts in the
noncooperative game.
In the case study application, the resource management agency chooses different types of emissions in
order to maximize its net benefits for the United States and for Mexico on an infinite time horizon. The
conclusion is that trade liberalization provides economic incentives to reduce pollution and reuse treated
wastewater for irrigation.
WASTEWATER TREATMENT
Giglio and Wrightington open, with their 1972 paper, an important literature thread about the application
of CGT to the field of wastewater treatment. A joint plant, developed by more than one agent (city,
factory, ...) can reduce costs without a loss of efficiency, but it is immediate to find a problem of cost
allocation. Loehman et al. (1979) applying the Generalized Shapley value (GSV; introduced few years
before by two of the authors of this paper) to a similar project for water treatment in Missouri, conclude
that a cost sharing method has to be simple to gain acceptance by the involved agents and that the GSV,
interesting from a theoretical point of view could be "unnecessarily complex in the light of the
community needs".
We find a CGT model of a cost-sharing problem in a joint project development also in Kogiku and
Sheenan (1980), Loehman (1995), Holler and Li (1996). Lejano and Davos, in the same context, (1995,
and 1999) introduce the concept of Normalized Nucleolus, taking into account the rate of return of joining
a coalition and being monotonic in a proportional manner. In the paper by Dinar et al. (1986) the gain to
be shared is due to reuse of treated wastewater in irrigation.
In Dinar at al. (2003) we find the first application of Stochastic Cooperative Game Theory to water
resource management, and specifically to an hypothetical example for development of a wastewater
treatment plant: "stochastic considerations may affect the nature of the solution, depending on the attitude
of the players towards risk and the nature of the (possibly not concave) cost function of the water
project".
While the applications of the other reported papers have a local or regional scale, Caswell and Frisvold's
presents an international case of water pollution and treatment (US-Mexico border). The framework is
more complicated because of the international scale of the problem, but the authors believe that the use of
a cooperative bargaining model can address the apportionment of costs of a waste collection and
treatment systems, once minimized the costs.
Several cost sharing methods are examined and compared by Giglio and Wrightington (1972): (1) based
on the measure of pollution, (2) based on single plant cost with a rebate proportional to the measure of
pollution, (3) SCRB method, (4) free market bargaining, (5) bargaining including the regional authority as
a participant.
In (1) the polluters are equally treated, because they pay the same amount for a pollution unit. In the
numerical example one of the polluters would pay more than he should if treating his waste alone. In (2)
each user has to pay his single cost to the authority, and the authority can return the difference between
the funds collected and its joint costs, and the rebate is assumed to be proportional to the measure of
pollution. According to this assumption, the largest polluter will receive the largest rebate, but he still
pays more, in total. With the example the authors show that, probably, a bargaining process will start,
making it more difficult to reach an agreement on the regional plan. Using the SCRB method (3) in the
example further bargaining is no longer needed. In the section about method (4), we find an introduction
to cooperative game theory, and the imputation set is described through its inequalities: the authors
analyse different kinds of solution these inequalities may lead to, and conclude that "although it clarifies
the situation, game theory approach does not usually provide a treatable solution to the problem". (Even if
21
the authors use the term bargaining, the framework is a TU-game.) In (5) the regional authority is
included in the game and the problem can be solved with the help of linear programming. The problem to
be solved is the following:
n
min X i = Pmin
j=0
subject to:
A11X1 +...+ A1 Xn P1
n
...
Am X1 +...+ AmnXn Pm
1
Xi 0
Where:
j is an index of polluter; Xj is savings returned to user j; i is index representing a coalition, i = 1, 2, ..., m;
Pi is payoff of coalition i; Aij = 1 if j is a member of coalition i, and 0 otherwise. Here a solution always
exists, but it's probably not unique.
The authors conclude that no method can be universally preferred, but some of them show some
advantages. Methods (2) and (3) can be particularly useful in situations with few attractive alternatives.
Loehman, Orlando, Tschirhart and Whinston (1979) depart in their analysis of water pollution control in
the Meramec River Basin in Missouri USA that polluters have the option of treating their pollution in a
regional system or constructing separate smaller systems. In order to encourage participation in a regional
system so that the economies of scale are realized, the cost allocation method employed must offer
economic incentives.
The paper provides a description of the Meramec River Basin and of the main dischargers. Later a cost-
game is constructed and the presence of an economy of scale provides subadditivity for the game. The
primary, secondary and advanced wastewater treatment plants are examined from the point of view of
their costs, in order to meet the required quality standards choosing the minimum costs.
The Shapley value is introduced and explained but the authors believe that "in an application to real life
situations such as that undertaken here, use of this value has a few deficiencies". The main problem is
that it seems to be unrealistic to consider all the coalitions equally possible. The Generalized Shapley
value, by Loehman and Winston (1976) does not require this assumption, and will be used here as well:
xi = ( p S,S -{i} C S -C S -{i}
) ( ) ( )
SN
iS
The Generalized Shapley value is a weighted sum of incremental costs, and in Loehman and Winston
(1976) the suggestion is to get the weights from the probabilities of certain orders (p is interpreted as the
probability that i joins the coalition S - {i}), replacing the usual coefficients [(n - s)!(s -1)!]/(n!) .
The Generalized Shapley charges are calculated and the result is that the charge is always less than the
cost of self-treatment, but the allocation does not lie in the Core, so that some coalitions may have
interests to disrupt the grand coalition. Gately's concept of (minimizing) the propensity to disrupt (1974)
is later used to get other possible charges.
22
In a different paper, Kogiku and Sheehan (1980) are concerned with fair apportionment. The study of
these problems is of considerable potential importance as most of the large water resource development
projects are collective endeavors. Furthermore, a large proportion of waste treatment projects involve at
least two major actors. In addition, in the last several years many water resource development situations
have attracted the attention of a third party in the form of the Environmental Protection Agency (and/or its
cousins on the State/Regional level; e.g. the Porter-Cologne Water Quality Control Boards in California).
This intervention opens up a range of new bargaining outcomes wherein standard-setting government
agencies allot themselves a share of the benefits of cooperation which would normally have accrued
solely to the local participants.
Several ideas of fairness and several systems for apportioning costs and benefits are discussed with
the help of a hypothetical case study: two polluters sited on a river, with an interest in developing a joint
wastewater facility. Some rules for the cost-sharing method are discussed, using a hypothetical example
as a thread for the discussion: (1) an equal share of the gains from cooperation; (2) a division of costs
proportional to the costs they would have paid under self-sufficiency; (3) equalizes the propensity to
disrupt of the participants (Gately, 1974); (4) the Shapley value; (5) the Kernel.
Dinar, Yaron and Kannai (1986) demonstrate the use of CGT to allocation of costs and benefits from
regional cooperation, with respect to reuse of municipal wastewater for irrigation at the Ramla region of
Israel. A regional model provides the maximal regional income from various technical and institutional
options which has to be redistributed among a town that generates the pollution and several farms that
may reuse it. Different allocations based on marginal cost pricing and schemes from cooperative game
theory like the core, Shapley value, generalized Shapley value, and nucleolus are applied. The Core of
this game is a four dimensional polyhedron. It is not conclusive, because it provides only bounds on the
claims of the participants.
Marginal cost pricing is not applicable in this case since the number of participants in the wastewater-
effluent "market" is small, they are not anonymous, and agreements and coalitions among some of the
potential participants are possible; the treatment costs depend on increasing economies and decreasing
marginal costs. In such a case, marginal cost pricing leads to a deficit, which would need to be covered
either by the users or some external agency (e.g., government), or both.
The town and one farm of the three have the main additional gains according to all allocation schemes
presented. The authors conclude that game theoretical approaches do not provide an objective,
indisputable solution, to the income distribution problem, but they provide a bound for the players' claims
(Core) and a better understanding of the situation.
Loheman (1995) extends the application of cooperative game theory to include both water allocation and
cost sharing. Combining cooperative game theory and theory of market games provide answers to
questions of what is the efficient allocation of water, which water supply project should be undertaken,
and what should be the price of water. Cost allocation and market games are first described separately
below, and then are combined for this purpose.
The author introduces some solution concepts that are based on the characteristic function (Core, Shapley
value, Generalized Shapley value, Nucleolus); then reports two solutions not based on transferable utility
cooperative games: the Nash bargaining solution and the Generalized Nash solution by Harsanyi (1959).
A small example is used to illustrate that different solution concepts can differ significantly in terms of
the distribution of gains and implied power. The example deals with four dischargers having to share
costs for pollution treatment at a single location (the costs include piping plus treatment costs). A market
game (Shubick, 1984) is later introduced to allow modeling a water allocation problem with a fixed
supply source, and solved using the Nash solution and the Generalized Nash solution.
Lejano and Davos (1995) introduce the Normalized Nucleolus and apply it to a N-Person Cooperative
game of water reuse project in Southern California. The results obtained with the normalized nucleolus
23
are compared with those derived with more traditional solution concepts such as the nucleolus and the
Shapley value.
The Nucleolus, in fact, is based on the absolute savings represented by the excess function, not taking into
account the rate of return of joining a coalition:
e (S, x) = c(S) - x(S) / x(S) = c(S)/ x(S) -1
In the Normalized Nucleolus, the problem to be solved is:
min -
subject to:
{x(S)/c(S) +}1 S N,
x(i)=c(N) ,x(i) 0
iN
An interesting property of this solution concept is that it is monotonic in a proportional manner. For
example, if the total cost increases, then the charges will increase keeping the same relative proportions to
all players. A case study of water reclamation and reuse in Southern California is described and the two
mentioned solution concepts together with the Shapley value are applied to the data, and compared. In
particular, it is discussed the impact of the three different allocations on the margin of assurance that the
decision will be actually implemented. It is also stressed (using the figures of the Project) the
distributional proportionality of the normalized nucleolus.
Holler and Li (1996) apply N-person cooperative game theory to analyze the efficient cost allocation
problem for a group of three communities constructing a common sewage system. Important solution
concepts such as the core, the Shapley value, the kernel, and the nucleolus are all used to establish
efficient pricing rules. Treating the sewage system as a public good, the contribution that a community
should make is also measured by using the public good power index. The formation of subcoalitions and
the equilibrium coalition structure are investigated and discussed for the case that the splitting method is
accepted to be a fair distribution rule. A model of a cost allocation problem for a joint sewage processing
system is described. The paper compares the different solution concepts to show how a fixed distribution
rule can prevail throughout the bargaining process between the communities and to analyze the
equilibrium coalition formation resulting from this bargaining process.
Lejano and Davos (1999) suggest that many environmental management issues can be defined as
allocation problems, e.g., the allocation of rights to use common-pool resources or the allocation of the
cost of regional resource development projects. As such the allocation methods developed in the area of
cooperative N-person game theory are most appropriate for these problems because they focus on the
conditions for engendering and sustaining the necessary cooperation among the involved stakeholders.
These solution concepts seek to ensure that the allocation is based on some norm of equity and, most
often, also to minimize the incentive for any player to defect from the cooperative venture.
The authors' introduction to this paper is about Common Pool Resources (CPR) management and the
Tragedy of the Commons; they also introduce the payoff matrix of a two-player CPR non-cooperative
game. After a description of an example involving two towns whose wastewater and runoff impact a
nearby lake, they enter cooperative game theory and describe some of the main concepts, useful for a
cost-sharing problem. The solutions are applied to a case study of an inter-jurisdictional water reuse
project in Southern California, where a new supply of industrial quality water is being considered by
taking effluent from a city's wastewater treatment plant and treating it in order to reach the quality needed
for various industrial and commercial uses.
24
The different allocation methods are discussed, applied and commented, focusing also, at the end, on the
difficulties that can arise to get appropriate information, and on the issue of fairness.
In quite a different setting, Caswell and Frisvold (2000) introduce the issue of transboundary water
pollution between the US and Mexico in Southern California. In addition to the pollution problem, the
work also introduces the possible incentives associate with the NAFTA agreement. New institutions
created to address environmental concerns over NAFTA offer the promise of greater financial and
technical assistance for water management in border cities. The paper reviews US-Mexico border water
issues and institutions. Using insights from game theory, it draws policy lessons for institutions funding
border water projects. The paper examines how the design of assistance programs, technical support, and
pre-existing water rights and regulations affect project outcomes. The diversity and geographic dispersion
of water conflicts suggests potential for applying the interconnected game approach to US-Mexico water
negotiations.
The authors analyze the main features of trans-boundary management on the US-Mexico border and the
different agreements throughout the years. Then, they model the border water management as a
cooperative bargaining game. This approach is used to examine negotiated outcomes of pollution control
projects in three border metro areas: San Diego-Tijuana, Calexico-Medicali and Laredo-Nuevo Laredo.
The use of a cooperative bargaining model can address the apportionment of costs of a waste
collection and treatment systems, once minimized the costs, keeping in mind particular objectives.
The authors use this variant of the Nash product:
N = um (b -c) -um uu 1- )b -(1- )c -uu
(( )
where b and c are total project benefits and and (1 - ) are respectively Mexico's and USA's share of
benefits, and and (1 - ) the equivalent for costs.
If the project objective and least-cost means of meeting that objective are agreed upon, then b, c and are
exogenous and the sole bargaining parameter is , Mexico's cost share. If an equal cost sharing rule is a
binding constraint, however, the bargaining process takes the form of choosing b, c, and to maximize N
subject to = 0.5. Sometimes, such a condition may prevent a cooperative solution: for example, it is
biased against finding a cooperative solution for a joint project with high relative benefits for USA and
large absolute benefits (to the extent that b/c is closer to 1 where b c is large). Some projects were also
abandoned because Mexico's rate of expenses was over its budget possibilities. On the other hand, a
common facility for wastewater treatment across the border can significantly reduce costs.
Negotiations over terms of a proposal may be modeled as a sequential bargaining game with exogenous
risk of breakdown. One player makes an initial proposal that specifies how grant funds will be allocated.
The other player accepts or rejects the offer. Players make counter proposals until they reach an
agreement or negotiations end without agreement. Bargaining could end if the funding agency decides to
fund competing proposals. Delays in reaching an agreement increase the probability that funds will go to
other projects instead.
The bargaining game can be solved by:
N = vm (x, A) - vm (x, A = 0) × vu x, 1- ) A - vu x, 1- ) A = 0 1
( ( ) ( ( ) -
Where: N is a Nash product, vm and vu are the city's utilities if they receive the assistance, vm and vu the
city's utilities if no agreement is reached, A a measure of the value or size of the assistance package, the
city m's share of the assistance package, x the vector of bargaining parameters, x the values of parameters
in the event negotiations breakdown or the granting agency decides not to consider the project, and a
parameter measuring the bargaining power city m relative to city u.
25
At the end, interconnected games are introduced because a situation of unilateral externalities tends
toward a victim pays outcome (Bennett et al., 1997). Bennett et al. (1997) find victim pays regimes
unsatisfactory because they run counter to the polluter pays principle accepted in the international
community and because countries may wish to avoid appearing to be weak negotiators. A third criticism
is that with extreme income disparity, a downstream country may not be able to offer side payments to
discourage the upstream country from polluting or diverting trans-boundary waters. An alternative to side
payments or accepting externalities is to link negotiation issues.
Dinar, Ratner and Yaron (2002) point out some major difficulties, which arose in the authors' attempt to
apply CGT concepts to two empirical problems. These difficulties tend to be general rather than specific
to the problem studied. The first study (Dinar, 1984; Dinar, Kannai and Yaron, 1986) concerned a
regional project designed for wastewater treatment and reuse as an irrigation water supply. The game
consists of a transferable utility situation (side payments). At first the efficient solution (maximal regional
income) has to be found; the second stage considers the redistribution of income among the participants.
(The Core, in this case, is a four dimensional polyhedron and it may serve as a guideline for arbitration
also referring to other allocation criteria.)
The second study (Ratner, 1983; Yaron and Ratner, 1985, 1990) deals with regional inter-farm
cooperation in water use for irrigation and the determination of the optimal water quantity-quality
(salinity) mix for each of the region's water users. Salinity is due to an inefficient allocation of water
quotas. One possible way to increase the efficiency of water allocation among farms within regions is
through the voluntary establishment of regional water associations or farmers cooperatives that will
allocate water quotas in an efficient way, which is usually identified as non transferable utility case. (The
potential for additional income due to cooperation is higher when side payments are possible.)
In the first study a real problem and real players are presented; in the second study, only the farms
considered are real and the study attempts to explore the potential for inter-farm cooperation which does
not yet exists. In the two case studies reviewed, each addressing a real life problem of considerable
concern to farmers and policy makers, game theory approaches did not provide indisputable solutions to
the cost/income allocation problems.
In a NTU situation of a three-farm cooperative the Nash-Harsanyi solution is deduced. The Core is quite
large. Furthermore it is not a convex set. One may wonder if the Core is a practical concept applicable in
situations such as the one dealt with in this work.
Dinar, Moretti, Patrone and Zara (2003) introduce stochasticity into the analysis of water projects by
means of CGT. They depart from the observation that water resource projects are subject to stochastic
water supply patterns, which affect performance, sustainability and stability of any use arrangements
among participants. Traditionally, cooperative game theory has been applied to a variety of water
resource problems assuming a deterministic pattern of supply. This has lead to design and interpretation
of allocation schemes that are somehow limited in scope, suggesting that some solution concepts are not
possible. They have applied a stochastic game theory framework, to demonstrate how stochastic
considerations may change the solution, depending on the attitude toward risk aversion of the potential
users and the nature of the cost function of the water project. Specifically, they show that in a game with
some risk adverse players the core of the game is nonempty even if in the game with the same cost
function but different attitude of the players toward risk, the correspondent core is empty.
The stochastic nature of the problem not only affects the issues of optimal design, but also can concern
the allocation methods for costs and benefits. Players' attitude toward risk is considered in the model,
too. The authors use a hypothetical example to show that it is possible that in the case of a deterministic
game the Shapley value is not in the Core, while by including in the analysis stochastic considerations the
Core conditions do change.
26
GROUNDWATER
The three following CGT applications to groundwater management use bargaining approaches to model
possible cooperation in the considered aquifers. It's not a coincidence that an NTU situation has been
chosen by all of the authors of the reported works for an application to groundwater, as it is not a
coincidence that few cooperative models are present in the literature for groundwater. Even the works
presented here raise some doubts about the practical possibility of binding agreements, and a non-
cooperative model is presented alongside the cooperative one. The difficulties arising while dealing with
groundwater include asymmetric information, uncertainty of water rights, environmental degradation,
sustainability, high costs of control, etc.
Szidarovszky, Duckstein and Bogardi (1984) studied the management of a regional aquifer aiming at
achieving multiobjective management goals. The aquifer of the presented case study is a regional
groundwater in the Transdanubian Karstic region in Hungary and the conflicting objectives are with three
competing sectors, namely mining, water supply and environmental protection. The method used is a
generalization of Nash's cooperative game theoretical model, and may lean on Zeuther's bargaining
problem. The authors present the mathematical model and numerical results are compared with the one
reached for the same case study by compromise programming with an l1-norm (Duckstein, Bogardi and
Szidarovsky, 1979).
Most solution concepts are subjective in terms of weights, bounds, distances. In this paper the authors
provide a solution based on a certain set of axioms leading to a unique point: once accepted those axioms
and stated the status quo, the decision maker will have to accept the solution also.
A multiobjective programming problem is formulated by the authors as follows:
maxu (u)
U
where u is the decision vector,
=(1,...,n) is the vector valued objective function,
n is the number of objectives.
Here the set L = = 1 u ,...,n u ,u U
{ ( ( ) ( )) } denotes the feasible set and * = *,...,n
( *) is the
1
status quo point (disagreement point).
Two bargaining models are applied: the one by Zeuthen, based on the principle that the next concession
within the process always comes from the objective having least maximal risk in conflict, and the one by
Nash-Harsanyi (Harsanyi, 1959), satisfying Nash's axioms. (Both models require the acceptance of the
mentioned assumptions.)
The objective functions are described in detail and the problem is solved. Comparing the output by the
game theoretical models and the one by the compromise programming with an l1-norm (Young, 1994a)
the authors could not provide preference between the solutions since the two methods follow different
paths to account for the preferences of the decision maker, and consequently lead to different results. The
compromise programming solution incorporates preferences under the form of weights in an l1 distance,
while the cooperative game objective function is a geometric distance.
In another work, Netanyahu Just and Horowitz (1998) address the conflict over the Mountain Aquifer
between Israel and the Palestinians. Sharing of the Aquifer between Israel and the Palestinian Authority
is considered. Since 1967, the management of the Mountain Aquifer has been under the unilateral control
of Israel. The new peace accord, however, have begun a shift in control of parts of Judea and Samaria to
27
the Palestinian Authority. This may divide control of the Mountain Aquifer between Israel and the
Palestinian Authority. Without cooperation, overpumping and degradation of water quality are likely.
The authors give an overview on Israeli and Palestinians water problem: a significant increase in
water needs in the region can be faced through an improvement of water quality for existing water
resources, water desalination, wastewater recycling, water conservation, and cloud seeding.
For an efficient allocation of water, a bargaining approach is considered superior to pure market solution
not only because infrastructure is insufficient for markets to operate efficiently but also because common
property problems cannot be resolved efficiently by markets without public intervention.
The Nash bargaining solution maximizes the Nash product ui -ui up -up , where i and p stand for
[ ]
Israelis and Palestinians. The utility functions are drown by the authors and the problem is solved. The
Nash bargaining solution, in this case study, is highly robust to changes in the assumed parameters, but a
non-cooperative bargaining framework seems to better fit the real situation.
But binding agreements in this case may not be possible because (i) groundwater property rights are
ambiguous and (ii) enforcement is difficult or impossible due to lack of enforcement agencies, limitations
of technology, inadequate punishment for deviations, and high costs of monitoring.
The results show that the bargaining solutions implied by game theory are robust with respect to demand
elasticity for water, user cost, and cost of pumping. Further considerations of potential strategic behavior
and defection show that risk of breakdown in negotiations and impatience in realizing benefits of the
agreement are also important. Because of limited empirical information, the analysis at this point is
illustrative and demonstrates the rule of various factors and potential behaviors.
In a recent paper Just and Netanyahu (2004) apply an interconnected game theory to common pool
resources, using the Mountain Aquifer shared by the Israelis and Palestinians. One approach, the linkage
approach, has been proposed in real world negotiations. The paper demonstrates that by linking issues
with reciprocal importance (benefits) to the parties through interconnected games, one may avoid
outcomes that stem from the `victim pays' nature of certain situations. Because of the the imbalance of
power and access between the parties, Just and Netanyahu (2004) suggest that the linkage between
negotiated issues rather than water will actually eliminate a solution in the case of the Mountain aquifer
between the Israelis and Palestinians.
Another international aquifer case--the Jueco Bolson aquifer between Texas-Mexico--has been analyzed
by Montgomery, Nakao and Wichelns (2002). The aquifer provides municipal water supply for El Paso,
Texas and Ciudad Juarez, Mexico. The aquifer lies beneath the international border, and both cities
operate independently regarding pumping rates and withdrawals. Over-pumping causes an increase in the
salinity of water remaining in the aquifer from mud interbeds, degrading groundwater quality.
The potential gains from cooperation include reducing the negative impacts of externalities involving
pumping costs and depletion of a nonrenewable resource. The authors examine the potential gains from
cooperation in the withdrawal of water from the aquifer. Four dynamic scenarios are considered: 1) a
status quo scenario in which both cities continue extracting groundwater as they are at present, 2) a Nash
non-cooperative game scenario, 3) a Nash bargaining scenario, and 4) a scenario that involves
maximizing the sum of net benefits in both cities. All scenarios, including the non-cooperative game,
provide a longer useful life of the Hueco Bolson aquifer than does the status quo. In the Nash bargaining
scenario, both cities gain from cooperation and the sum of net benefits approaches the maximum that can
be obtained by maximizing that value explicitly.
The interaction between the two involved cities is described through per-unit pumping costs that increase
with pumping lift, as the volume of water remaining in the aquifer is reduced; the net benefits from
groundwater pumping in each city is derived.
28
Focusing on the cooperative models (scenario3 and 4), in (3) the Nash product to be maximized is
max (J )(J Th), ( Th Th)
1, J2
Th
2 - J2 where J1 , J2 is the net benefit values obtained in the non-cooperative
game scenario.
In (4) a social planner's goal is to allocate water between the two cities thus maximizing the sum of net
benefits obtained over the time. The results obtained when maximizing the sum of net benefits in this
problem are similar to those obtained in the Nash bargaining scenario.
ALLOCATION OF WATER RESOURCES
In the following papers water itself, instead of costs and benefits (e.g. money for a joint project), is
the medium to be allocated. In these works a common assumption is that one cannot rely on an efficient
market, thus the authors consider the consequences of a game theoretic based allocation.
In the paper by Braden et al. (1991) a (thin) spot market is analyzed and the focus is on the bargaining
rules that bring about an efficient solution, while in the other reported papers in this section we are in a
transferable utility framework and some CGT tools like the Core and the Shapley value are used in order
to draw an acceptable allocation.
Braden, Eheart and Saleth (1991) apply a multilateral Nash bargaining model to the Crane Creek
Watershed of Kentakee County, Illinois. The authors observe that markets for water rights, which have
long been advocated to resolve user conflicts and to improve efficiency are too rigid. Instead, rental or
spot markets for water can improve efficiency by allowing allocational adjustments in line with changing
conditions without disturbing the base right system.
They consider a small watershed, whose stream system is shared between eight farmers. Assumptions
about (1) certainty about the total water available for irrigation, (2) a strictly enforced water rights system,
specifying seasonal water withdrawal of each farmer, and (3) definition of water rights in terms of
consumptive use, suggest that binding flow constraints make some transfers infeasible.
According to the authors, neither a cooperative nor a competitive behavior is sufficient to realistically
describe water exchanges in a thin market. This is the reason why they prefer to use a multilateral
bargaining model, based on the contributions of Harsanyi (1986), Zeuthen (1930) and Nash (1950).
The total gain from water trade can be shared among the n players in many ways, depending on the
bargaining rules and bargaining environments.
Harrison and Tisdell (1992) are concerned with the distributional consequences of alternative methods of
allocating transferable water licenses in Queensland, Australia. The equilibrium of a number of different
cooperative games is used to judge which method of initial allocation of water entitlements potentially
produces the most equitable distribution of the income derived from the regulated water.
Water entitlements, in Australia, are often tradable, the authorities believing that market forces can lead to
an efficient and equitable redistribution of the water entitlements. The market can provide individual
equity, but there is no guarantee that the market redistribution of income derived from the water will be
equitable according to any particular definition of social justice.
The study adopts a cooperative game model in characteristic form with transferable utility. A number of
different n-person cooperative games are used to estimate the distribution of income after trade among six
representative farms selected from the Border River region of Queensland. The potential distributive
consequences of four different allocation methods are assessed on the basis of trade between these farms.
The authors consider the allocations of the Nucleolus, the Shapley value, and a maximin allocation. A
Rawlsian criterion of social justice, consisting in maximizing the welfare of the worst off member, is
used to assess the relative distributive justice of the alternative allocation methods.
29
Suppose that individuals in a social state are ordered in terms of their welfare in a social state x; if the
welfare of the individuals in society is lexicographically ordered, then xi is the welfare of the individual in
position i. For two social state vectors x and y, the Rawlsian lexicographical rule would argue that x > y
(i.e., social state x is preferred to social state y) if and only if there is some individual j (1 j n) such
that xj > yj, and for all i < j, xi = yi.
The `optimal' payoff vector within the Core, which maximizes the payoff of the worst off player, can be
estimated by:
n
max log(x ),
i subject to Core condition constraints.
i=1
While such payoff vector comes unlikely from trade, as players would have to be completely altruistic for
such an outcome to occur, it may be the appropriate point of comparison among alternative market
outcomes.
Ambec, S.; Sprumont (2000) suggest a welfare allocation scheme among river riparians, who have quasi-
linear preferences over water and money. The paper addresses efficiency, stability (in the sense of the
core), and fairness (in a sense to be defined) considerations of the allocations of water and income. A
cooperative game associated with that problem is shown to be convex. Using the fairness requirement that
no group of agents (riparians) should enjoy a welfare higher than what it could achieve in the absence of
the remaining agents, the authors prove that only one welfare distribution in the core satisfies this
condition: it is the marginal contribution vector corresponding to the ordering of the agents along the
river.
The conditions of Stability (as Core concept), and Fairness (such that no coalition should enjoy more than
its aspiration welfare where, for aspiration welfare level of an arbitrary coalition of agents, the authors
mean the highest welfare these agents could achieve in the absence of the others, yield a unique allocation
of welfare. Under this allocation agents receive a payoff equal to their marginal contributions to the
coalition composed by their predecessors along the river.
In the model proposed in this article a river flows through some countries or regions or cities (agents)
N ={1,2,...,n} . The agents are identified with their location along the river and numbered from
upstream to downstream (figure 3).
Figure 3: Location of the agents along the river
30
In this schematic representation we can see that the river increases its volume along its course: the flow e1
increases from the source (1) and becomes e2 in (2). Side payments are possible in this model. It is
important to note that the water stream ei can only be consumed by i's followers. The authors prove that
the modeled game is convex, so the Core has a simple and well-known structure. They stress the concept
of fairness and add upper bounds to the agent's welfare, getting a unique solution in the Core, satisfying
their requirements for fairness. In the second section of the article, a non-cooperative analysis is
discussed.
Van den Brink, Van der Laan and Vasil'ev (2003) investigate a special class of cooperative gamesthe
line-graph games, being cooperative TU-games where the players are linearly ordered. Examples of
situations that can be modeled like this are sequencing situations, water distribution situation and political
majority voting. The main question in cooperative game models of economic situation is how to allocate
the earning of coalitions among the players. They apply the concept of Harsanyi solution to line-graph
games. The authors define four properties such that each selects a unique Harsanyi solution from the class
of all Harsanyi solutions. One of these solutions is the well-known Shapley value which is widely
applied in economic models.
Harsanyi solutions, for a given sharing system, assign to any player a payoff equal to the sum of its
corresponding shares in the Harsanyi dividends. That is, the concept of Harsanyi solution selects for any
game the payoff vector in the Harsanyi set corresponding to the chosen sharing system.
The line-graph games, considered in this work, are a special kind of TU-game in which the
communication (graph) structure is given by a linear ordering on the set of players. For this class of
games the authors define four different properties to be satisfied by a Harsanyi solution.
INTERNATIONAL WATER BASINS
International water resources are rivers, lakes and groundwater resources (some legal frameworks include
in the above definition atmospheric and frozen waters also) bordering two or more countries. We dealt
with groundwater separately, because of its peculiar features, reflected by the CGT applications in the
literature.
As in smaller scale water management situations, each actor can affect the quality and quantity of water
available for the other actors, with the difference that the involved players are represented by countries
and thus the political and legal context has to be taken into stronger account. Obviously there is still room
for a game theoretic approach.
Literature results show that a non-cooperative solution is not the inevitable outcome for the use of an
international (and often open-access) water resource, even if an international scale is more difficult to
handle.
"Conflict and Cooperation on Trans-boundary Water Resources", edited by Just and Netanyahu (1998) is
a very complete work about international water management. The book addresses both issues of conflict
and cooperation, covering a wide range of possible obstacles to water resource cooperation, such as lack
of transparency and asymmetric information, scientific and technological uncertainties, enforcement
limitations and upstream/downstream considerations. The analysis focuses on economics, with an
extensive use of game theory, but it covers a wide range of disciplines such as political science,
environmental sciences, international and national law. In this section we report the first chapter, by Just
and Netanyahu, which is a really exhaustive introduction to the problem.
Dufournaud and Harrington (1990, 1991) introduce in their analysis two interesting aspects: time (even if
they consider only two time periods) and admissible coalitions. They present a hypothetical example,
31
`inspired' by the cost-allocation problem of Young et al. (1982), and the aim is to implement that study
with some temporal aspects.
Becker and Easter wrote three papers about the Great Lakes management. In their 1991 work they
analyzed the maximization of benefits in order to get a characteristic function, assuming different
scenarios characterized by various degrees of cooperation among the states. In (1997), once a diversion
scheme for the Great Lakes' water has been chosen, the focus is on how to apportion the gains and some
typical CGT solution concepts are applied (Core, Nucleolus, Shapley value, Nash bargaining solution). In
the last paper (1999) they demonstrate that stable cooperative coalitions are possible when a subset of
users decides to cooperate: in the example proposed, once a minimum number of users decide to
cooperate, the remaining agents have an incentive (interest) to cooperate.
Rogers (1993, 1994), studying the Ganges-Brahmaputra Basin, outlines the applicability of cooperative
game theory and Pareto frontier analyses to water resources allocation problems, and places his GT
model in an accurate geographical and legal framework.
Barrett (1994) tries to answer this question: "Can negotiated treaties ensure that nations which share
bodies of water share gains form cooperation?" and concludes: "Often, but not always - and sometimes
only partly."
The necessity of taking into account the political realities surrounding trans-boundary waters is also in
Dinar and Wolf (1994). They illustrate the potential of water trading among Middle East countries
(Egypt, Israel, the Gaza Strip and the West Bank), accounting for political constraints: a PAS (Political
Accounting System) is included in the CGT model, in order to take into account the probabilities of
formation for the different coalitions.
Luterbacher and Wiegandt (2001) present a cooperative solution that is applicable to the Jordan river
basin region: cooperative and non-cooperative strategies are compared, and effects on water consumption
are evaluated.
Also in Yaron (2002) the problem of water scarcity is dealt and the starting point of the analysis is a non-
cooperative scenario. This paper was reviewed in the section on Groundwater.
Guldmann and Kucukmehmetoglu (2002), whose case study is about Tigris and Euphrates rivers, present
a linear programming model allocating water (and maximizing the net aggregate benefits) among
agricultural and urban uses of the three riparian countries (Turkey, Syria, and Iraq). Cooperative game
theory concepts (Core and Shapley value) are used to identify stable water allocations.
Dufournaud and Harrington (1990) extend cooperative game theory to temporal allocations The approach
is illustrated with an example problem involving three riparians and two time periods. Propensities to
disrupt are incorporated in the cooperative game in the form that permits the use of linear programming
techniques. The results suggest that a considerable range of choices for decisions can be generated by
systematic and random investigations near to optimal solutions. In a second paper Dufournaud and
Harrington (1991) extend their analysis of the previous model by including linear constraints which
satisfy the well known Shapley-value imputation scheme.
In these two papers Dufournaud and Harnington examine the difficulties of managing an international
basin, also from the temporal point of view: river basin engineering schemes are long term and, as a
consequence, the temporal pattern of costs and benefits must be considered. The framework presented in
these works is based on an example of a spatially variant cost-allocation problem presented by Young et
al. (1982). Their example is extended to make it a time-variant cost-allocation problem.
Once defined the Core of the above mentioned game and an overview on allocation methods found in the
literature, the example proposed uses the main concepts of cooperative game theory, the Propensity to
Disrupt, as defined by Gately (1974), and the notion of near optimality discussed in Harrington and
Gidely (1985).
32
In static games with n players the possible coalitions are 2n-1; if we consider j time periods the possible
coalitions will be 2jn-1. Not all of them make sense and the authors consider only those presumably
arising from the concerns specifically articulated by the riparians.
Some inequalities concerning equity and potential for cooperation and disruption are generated and
the Linear Programming Problem is solved. In the second paper the Shapley value is included in the
analysis, and it can be done in various ways: The first involves treating each riparian in each time period
as a separate player: according to this method, the grand coalition could form in 6! (720) ways. A second
way is to develop it for each time period: in this case there would be 5 different `3!' Shapley games in the
second period, if the level of cooperation assumed prevails in the first time period. A third way (chosen
in the example presented) develops the Shapley value over all time periods simultaneously.
Becker and Easter (1991) apply a game theoretic model to the multi dimensional conflict between the
USA and Canada over the Great Lakes conflict. In the face of increasing pressure for diversions of water
from the Great Lakes, this work considers alternative diversion restrictions and coalition structures for the
Great Lakes. The analysis presented here can be used by regulatory bodies and planning agencies to
determine which types and limits on diversion restrictions would be desirable. Results also have
implications for the desirability of cooperative rather than non-cooperative solutions: unless there is a
basin-wide cooperation, most states and provinces ultimately suffer large losses if unilateral and
unrestricted diversions occur.
In the past, American states neighboring the Great Lakes have used police power to prevent or regulate
interstate water diversions. This power was radically curtailed by the Supreme Court in the early 80's; its
ruling found economic purposes insufficient reason to prohibit interstate transfers of water. These rulings,
plus requests for additional diversion from the Grate Lakes during drought periods, were the main
impetus behind the Great Lakes Charter of 1985 and the Water Resources Development Act of 1986.
These two institutional changes created new mechanisms for regulating water diversions from the Great
Lakes Basin. The Water Resources Development Act granted the Great Lakes states the right of approval
for any federal interbasin diversion of water from the Great Lakes Basin. The Great Lake Charter also
established requirements for any state of province to consult and seek consent of all affected states and
provinces prior to approving any major new diversion or consumptive use of Great Lakes water (Frerichs
and Easter, 1988).
In the Great Lake basin, cooperation can occur on different levels: at the international level United States
and Canada may fully cooperate or act competitively; on the national level we can have either cooperation
within each country, or there may be full non-cooperation, with each state and province acting alone. The
authors consider these three possibilities and develop some models.
For the Region-Wide Cooperation (international level), they consider a maximization problem, subject to
hydrologic constraints, and solve it. Then they consider the cooperation at the national level, and the non-
cooperation between Canada and USA, and in this case we have a maximization problem for both
countries. As for the full non-cooperative situation, each state and province act separately and solve its
own maximization problem.
Results show the optimal diversion and net benefits for each coalition and diversion restriction. For the
fully cooperative case, the paper shows that the unrestricted diversion case brings the largest net benefits
among all the situations modeled. (The authors do not consider in this paper the problem of the potential
share of those benefits.)
Rogers (1993) reviews the phenomenon of international river basins, and concludes that the sharing of
river basins between and among countries is the rule rather than the exception for the major river systems
of the world. The growth of the nation state with rivers as agreed-upon boundaries, as cease-fire lines and
as natural defensive lines has led to the carving up of most river basins around the world. Two or more
countries share more than 200 river basins, accounting for more than 50% of the land area of the Earth.
33
When population densities were low there was plenty of water for all and major conflicts were avoided.
With the rapid population and economic growth experienced in the past few decades, conflicts over use of
water are becoming more important. It is expected that in the near future such conflicts will become much
more severe. The paper reviews the literature on attempts to analyze river basin conflicts and negotiate
solutions. Some rudimentary game models are examined, and some tentative conclusions, based upon
various game theory concepts of stability, are applied to the GangesBrahmaputra basin. The paper ends
with suggestions on how to plan Pareto-Admissible outcomes for international basins.
In a different work (Rogers, 1994) outlines some of the potential solutions to the conflict over the
Ganges-Brahmaputra basin water that are shared between India, Nepal and Bangladesh. In particular, the
paper emphasizes those Pareto solutions which can be obtained without making any of the participating
countries worse-off. While there are no simple solutions to this problem, economics and game theory
offer some concepts of what might be "good" solutions.
Rogers examines possible causes of the conflict rising from international water resource management that
is the unidirectional externalities in the use of a common resource. The general economic prescription to
deal with externalities is to internalize them, and the river basin itself is the ideal unit of analysis to
achieve this goal. The internationality makes it difficult to agree on these issues, and unless all the
riparian countries agree, the basin cannot be developed as a single unit. In this frequent case, in order to
internalize the externalities, taxes or fees can be applied but this method requires a strong supra-national
institution to impose and collect the taxes. The practical approach dealt with is called Paretian
Environmental Analysis, first formulated by Dorfman, Jacoby and Thomas (1972): they applied Paretian
analysis to upstream-downstream conflicts over the management of water quality.
If we let NBi(x) be the net benefits accruing to country i, should the resource allocation x be adopted, it is
assumed that each country wishes to maximize its net benefit. Formally an allocation x is Pareto-
Admissible if there isn't any feasible allocation y making NBi(y) > NBi(x) for some i with no worse for the
other involved players. Baumol's concept of superfairness (Baumol, 1986) rests above the Pareto
improvement criterion given above, and the concept of `fair division'.
Game Theory offers a variety of approaches coming from the analysis of the Core of an n-person
cooperative game, based on the idea that there is a set of imputations allowing no coalition to improve the
payoffs to its members.
Using the Ganges-Brahmaputra basin model the global maximum Max NB1 x + NB2 x + NB3 x
( ) ( ) ( )
xX
is equivalent to the grand coalition in game theory terms.
To generate Pareto-Admissible solutions the author followed Dorfman et al. (1972), chosing a set of
arbitrary weights W and computed Max W1NB1 x +W2NB2 x +W3NB3 x .
( ) ( ) ( )
xX
A wide range of admissible solutions is possible and, according to the way one weights the relative
net benefits, they could be used as starting point for arbitration.
The characteristic function is constructed and the Core of the corresponding game results to be non-
empty and "quite small": this should make the task of negotiators easier than in the case of a large range.
Rogers (1991) examines the Nucleolus of the game and the per-capita Nucleolus, where the per-capita
excess (proportional to the number of the players in the coalition) is substituted to the excess used to get
the Nucleolus. (Other proportionalities can be constructed on the values of the coalitions.) The Shapley
value of this game is also calculated and in this paper we find a table with the comparison of the
percentages of benefits allocated to the three states through these different allocation schemes.
In a review paper, Barrett (1994) asks the question: Can negotiated treaties ensure that nations which
share bodies of water share gains form cooperation? With simplified examples from existing treaties from
34
the Columbia, Indus, and the Rhine Basins he answers that Often, but not always and sometimes only
partly.
Barrett, stresses the issue of interdependence in the management of international water resources
introducing a Prisoners' Dilemma Game to model the sharing of a common aquifer between two
countries, choosing to extract the water at a High or a Low rate (figure 4).
Figure 4: An aquifer game matrix
Low High
Low (5,5) (2,6)
High (6,2) (3,3)
If we suppose that the two parties negotiate an agreement specifying that if both choose High, one of
them should pay an amount equal to 2 to the other party, the agreement can change the Dominant
Strategies for the players, and of course, the agreement has to be binding.
In the examined case the externalities are from both parties, but in other situations the externalities are
unidirectional (upstream to downstream, but also vice versa).
The author presents three case studies to provide a perspective on the political economy of international
water resource management: the Columbia River Treaty, the Indus Water Treaty and the Convention on
the Protection of the Rhine Against Pollution by Chlorides. Then he introduces a theoretical framework
of unidirectional externalities.
The bargaining problem deals with the sharing of the gains resulting from cooperation (the difference in
the aggregate payoff between the non-cooperative and full cooperative outcomes). The full cooperative
outcome is the one maximizing the aggregate payoff (in the Prisoners' Dilemma Game presented, the gain
is (5 + 5) (3 + 3) = 4).
There is not a complete compelling theory predicting how the gains will be distributed, even if concerns
of equity and reasonableness are widely employed in the analysis of negotiated outcomes.
In a situation of reciprocal externalities, what prevents reciprocal externalities from being entirely
internalized is the requirement that such cooperative agreements be self-enforcing (they must include
mechanisms which, by themselves, can sustain a cooperative agreement). The point is that, in general,
such agreements may be better than the non-cooperative outcome, but may not be able to grant the full
cooperative outcome.
Becker and Easter (1997) consider the possibility of `economically desirable diversions' and how the
gains should be allocated among the Great Lakes states and provinces to foster cooperation. The concern
about other states diverting water from the Great Lakes has prompted the Great Lakes States and
provinces to adopt institutional arrangements that have effectively blocked any new diversions. Since the
current arrangements do not allow diversions, important opportunities may be lost in the future. They
show that in most cases, new institutional arrangements will be needed before agreements can be reached.
The Great Lakes system is surrounded by two countries, the United States and Canada but within the
USA there are eight states governing the lakes, while in Canada there are two provinces. Some states and
provinces govern more than one lake, while others border only one lake. In addition, the major
hydropower facilities are located downstream from Lake Erie which makes it difficult to develop
cooperative solutions. (Becker and Easter, 1997).
Game theory is used to determine how coalitions may be formed to reach cooperative agreements for
diversions. Five different lake diversion games are tried involving Lake Ontario, Lake Superior, Lake
Erie, Lake Michigan-Huron, and finally, all the lakes together. Diversions from Lake Ontario may offer
the best opportunity for cooperation since there are no inter-lake effects.
35
Once a preferable diversion scheme is selected, the problem of apportioning the gains between the
cooperating players still remains, and this is one of the main concerns of this paper.
With the help of game theory, the authors underline that in some cases cooperation may occur without the
need of new arrangements, in other cases it is impossible to reach a cooperative solution without new
institutional arrangements.
The cooperative game model is developed for Lake Ontario: the characteristic function is derived from
data and previous analysis conducted by the authors (and reviewed in a previous sections on water
allocation).
The game for the Upper Lakes is quite similar to the Lake Ontario's and the Core is always non-empty,
except for the Lake Michigan-Huron. It was found that if a diversion is to take place from Lake
Michigan-Huron, federal government invention will be required since there is no bargaining solution that
will benefit all parties. The Core of the game is non-empty. All the results are commented and the
authors conclude that in general, more reasonable and equitable solutions were found using the Nucleolus
rather than the Shapley value.
Dinar and Wolf (1997) address possible cooperation in the Western Middle East over the Nile water,
showing that there is a need to incorporate more politically related considerations on top of the economic
and game theory ones. Their thesis is that cooperation among players requires a realization of economic
benefits to all players and a meeting of efficiency requirements through economically driven allocations.
Cooperation among political (and sometimes hostile) players may not meet these requirements. Political
considerations, usually ignored in economic analyses, can hinder or even block possible arrangements.
The paper proposes a framework that includes both economic and political considerations for evaluating
transfers or trades of scarce resources. This method quantifies both the economic payoffs using N-person
game theory and the political likelihood of any of the coalitions actually forming, using the PRINCE
Political Accounting System. The economic-political approach is applied to a case of potential water
transfer in the Western Middle East. Results suggest that incorporating political considerations in the
analysis stabilizes the regional solution suggested by economic-related allocations.
A Political Accounting System (PAS) to quantitatively and qualitatively predict interactions among
players is also used in the political science literature. The PAS assumes that the political impact of players
on joint decisions be affected by three factors: issue position, power, and salience. The first element of
this system is the issue position, expressing the strength of each participant's position for or against each
of the issues. A second element is the power of a player to influence an issue, the third element is
salience, i.e. the importance each player attaches to a particular issue.
Two problems may affect the application of the PAS approach: first the measure of values to be included
in the PAS is, in many cases, subjective, ordinal, and possibly inconsistent, second this approach does not
consider coalitional arrangements among the players.
The authors consider a regional allocation game, where a water resource has to be shared, as well as
monetary compensations and exchange of technologies for water, under certain arrangements, among
different users in the region.
The Core, the Shapley value and the Generalized Shapley value are considered, and a PAS is included in
the model, to calculate political probabilities for the formation of different coalitions. (The PRINCE
Political Accounting System, developed by Coplin and O'Leary (1974, 1976, 1983) is used in this paper.)
The combined game theory and PAS approach has been applied to a case of water transfer in the Western
Middle East. The involved players are Egypt, Israel, the Gaza Strip and the West Bank. The values
assigned to the political components of the PAS in this paper are the best quantitative assessment of an
elusive quantity. The authors recognize the need to develop more objective measures for each component
of political viability.
36
The results of different allocation schemes for the regional gains are discussed and, to assess the stability
of the allocation solutions, the Shapley-Shubik (Shapley and Shubik, 1954) power index is applied:
i = xi -v(i)
x i -v(i)
jN
The power index compares the gains of a particular player to the gains of the coalition. A more
harmonized power distribution is more likely to yield a stable coalition. The coefficient of variation of the
power distribution S = is a measure of stability for each allocation, where is the variance and
is the mean value of . The allocation resulting from economic considerations only is the most
unstable; the game theoretic allocations are far more stable, with the GSVMP being the most stable and
the Shapley value being the least stable.
Just and Netanyahu (1998) work focuses on the difficulties in forming grand coalitions, including all the
riparian counties involved in the trans-boundary water basin's management. Their observation of the
multilateral agreements stock is that treaties including all riparian countries are an exception rather than a
rule (World Bank, 1993).
They examine some guidelines for water apportionment, considering some international treaties, and
underline some obstacles to trans-boundary water resource cooperation (asymmetric information,
scientific gaps and technical uncertainties, enforcement limitations, sovereignty, conflicting national and
international interests, asymmetric country characteristics, upstream-downstream considerations). Using
the literature on coalition formation they discuss difficulties in forming water agreements in water basins
with multiple (three or more) riparian states.
Individual countries form partial coalitions rather than a grand coalition if the game is not superadditive.
Aumann and Dreze (1974) specify these sources of non-superadditivity: inefficiency of the grand
coalition, moral hazard problems (for example a country cannot perfectly monitor the clean-up effort
invested by polluting riparian countries), and free rider problems.
In contrast to Aumann and Dreze (1974), Owen (1977) and Hart and Kruz (1983), the authors assume that
society operates efficiently and that a grand coalition will eventually be formed. The formation of smaller
coalitions is considered as a strategic step to bargaining for a larger share of the pie when the grand
coalition is formed.
The authors then introduce a hypothetical example of a game in which coalition formation takes place in
an environment with economic externalities. First, the example shows how the choice of partial
coalitions rather than grand coalitions may be optimal (both when the transaction costs of coordination
are considered and when water conservation encounters decreasing returns to scale). Second, the example
shows that once the coalition is formed, optimal behavior may allow to have different environmental
responsibilities or to receive different allocation opportunities, even when countries are identical, with
inequities compensated by side payments.
A similar finding is proposed in another paper on Lake Ontario by Becker and Easter (1999). They use
game theory to assess the potential for cooperative management. They demonstrate that a noncooperative
solution is not the inevitable outcome for the use of an open access water resource. Stable cooperative
coalitions are possible when a subset of users decides to cooperate. In their examples, once a minimum
number of users decides to cooperate, the remaining noncooperators find that it is in their best interests to
cooperate.
A Prisoners' Dilemma game is introduced in this paper, to model the case of reciprocal externalities with
full cooperation, non-cooperation and partial cooperation. The paper assumes to be the portion of users
signing an agreement, thus N is the number of users forming a coalition, while (1 - )N is the number of
37
free-riders. The equilibrium is calculated basing upon the Nash equilibrium for the (1 - )N players who
didn't sign the agreement, once fixed the amount of water diverted by the other players. A coalition is
stable if: (1) there is no user in the coalition that has an incentive to defect and (2) there is no user outside
the coalition who has an incentive to join the coalition.
Luterbacher and Weigandt (2001) describe how computer methods can be used to analyze environmental
resource problems and evaluate data bases about environmental trends in the Jordan River region.
Resource allocation and use depend on both physical and social processes. Conflicts may erupt over
overlapping claims to identical resource pools. But because of the complex interactions determining
availability and use, solutions to resource conflicts are difficult to elaborate. This study adopts a
formulation that focuses on the crucial linkages between the economic, sociocultural, political, and
demographic parts of a social system on the one hand, and an important resource sector, water, and its
interaction with climatic processes on the other. These linkages are expressed within a dynamic
simulation model which has been adapted to the particular case of the Middle East. Samples of the data
collection that was undertaken as well as model calibration calculations are given in appendices. The
paper explores competition, conflict, and possible cooperation between regions and nations in terms of
resource use and allocation, particularly water, and movements of people. The argument is made that
cooperative strategies and conflict resolution schemes can be effective in achieving sustainable water
management in this region. The underlying theoretical framework is that of game theory, which posits
that decision makers are engaged in a process in which their preferences, their possible choices and those
of other parties, as well as the mutual effects of choices on each participant, all affect its trajectory and
outcome. The advantage of the simulation approach is that it permits the investigation of basic conflict
and cooperation situations in terms of their game theoretical structures. The paper presents a cooperative
solution that is applicable to the Jordan river basin region.
"A differential game similar to a sequential Prisoners' Dilemma game which leads to a Nash solution in
which regions will minimize their water uses (a formal proof which shows that this bargaining
mechanism works in the case of two regions is provided). The analysis leads here to a result, which is
somewhat similar to the one of the Folk Theorem (Fudenberg and Tirole, 1992). However, if the regions
are too asymmetric in their access to water resources, as in an upstream downstream relationship, the
dynamic bargaining process mentioned above will not work.
Figure 4 below shows that cooperative limitation of water consumption will slow down water demand in
the industrial region. As a result, some reserves will be maintained longer and the decrease in water
availability in the rural region will be less steep despite increasing drought.
38
Figure 4: Evolution of water consumption in various sectors with and without cooperation.
Source: Luterbacher and Weigandt (2001)
Guldmann and Kucukmehmetoglu (2002) use a linear programming model that allocates the waters of the
Euphrates and Tigris rivers to agricultural and urban uses in the three riparian countries Turkey, Syria,
and Iraq while maximizing the net aggregate benefits from these activities while accounting for water
conveyance costs. Cooperative game theory concepts (core, Shapley value) are used to identify stable
water allocations, under which all three countries find it beneficial to cooperate.
Increasing water demand in the Mesopotamia region has brought attention to the transboundary water
of Euphrates and Tigris. The paper presents an optimization model, which takes into account for the two
rivers agricultural use, urban water supply and hydropower plants, and maximizes the aggregate net
benefits of the three involved countries. Then CGT is used to allocate water among the different uses and
countries in order to make cooperation feasible. Different scenarios are considered, related to future
energy prices, agricultural production efficiency, and total water availability. The model is carefully
described and different coalitional situations are analyzed, (figure 5).
The total benefit of the grand coalition is the maximum aggregate benefit achievable by the three
countries. The problem is to allocate this aggregate benefit among the three countries in a way that will
persuade them to accept this allocation. Individual and group rationality have to be met. For individual
rationality the authors consider the minimum benefit of a country, taking into account two possibilities:
39
(i) the country act independently as well as the others, and (ii) the other two countries form a coalition.
The Shapley value algorithm is described for Iraq.
The modeling approach is applied under each of 27 parameter scenarios, that is, combinations of
assumptions regarding energy prices, agricultural productivities, and total water resources. For each
scenario the authors find out whether the Core exists or not, and, if it does not, then what is the minimum
subsidy needed to create a Core. They also check whether the Shapley allocations are in the Core. Six
cases have no core, 12 have a single-allocation core, and 9 have a multiple-allocation core (i.e., there is an
infinite number of allocations in the core).
Most of the empty-Core cases happen when Iraq is more productive, and when resources are more
abundant: in fact these situations allow individual countries to achieve higher benefits on their own
(agriculture for Iraq, energy for Turkey), making it more difficult to achieve a sustainable allocation.
When energy is not a factor, all Shapley allocations are in the Core. When it is, none of the Shapley
allocations are in core. This occurs because under the assumption of high prices for energy, the Shapley
method assigns more benefits to Turkey and Syria, and less to Iraq, and the deriving allocation is out of
the Core.
Figure 5: Various coalitional formations and water flows
Source: Guldmann and Kucukmehmetoglu (2002)
CONCLUDING REMARKS
With increased competition over natural resources and environmental amenities, decision makers face
strategic decisions in various management and use aspects. Cooperation approaches have been used in
many cases, and were proven to be useful under certain conditions. In this review we provided
information on various Cooperative Game Theory (CGT) applications to issues of water resources. With
an increase in the competition over various water resources, the incidents of disputes have been in the
center of allocation agreements. Water conflicts are typical at various levels, including the local
40
(irrigation perimeter, canal, groundwater aquifer, city's treatment plant, etc...), or at regional or
transboundary levels (conflicts over allocation issues between sectors such as irrigation and urban, or
between states sharing a given river). The paper reviews the cases of various water uses, such as multi-
objective water projects, irrigation, groundwater, hydropower, urban water supply, wastewater, and
transboundary water disputes.
In addition to providing examples of cooperative solutions to allocation problems, the conclusion from
this review suggests that cooperation over scarce water resources is possible under a variety of physical
conditions and institutional arrangements. In particular, the various approaches for cost sharing and for
allocation of physical water infrastructure and flow can serve as a basis for stable and efficient agreement,
such that long-term investments in water projects are profitable and sustainable. The latter point is
especially important, given recent developments in water policy in various countries and regional
institutions such as the European Union (Water Framework Directive), calling for full cost recovery of
investments and operation and maintenance in water projects. The CGT approaches discussed and
demonstrated in this paper can provide a solid basis for finding possible and stable cost sharing
arrangements.
Several highlights from the sectoral reviews include: (i) The ability to allocate joint costs of multi-
purpose water projects (e.g., irrigation, hydropower, flood control components in one project) in such a
way that the total cost is covered and at the same time each component is financially and economically
justified is a big advantage for planners who need to demonstrate the viability of the entire project and its
components. (ii) The urban water supply cost allocation schemes support the viability of projects that
serve many customers and are in most countries in jeopardy. CGT demonstrates its ability to account for
both efficiency and equity, and as such provides valuable cost sharing rules for policy makers and urban
water developers. (iii) Applications to irrigation water cases, although similar to that of urban water,
demonstrate the usefulness of NTU and TU situations that are typical to games where players can transfer
resources, or the benefits from using the resources, as side payments. A similar situation exists in
transboundary water conflict games.
41
REFERENCES
Aadland, D., Koplin, V. (1998). "Shared irrigation costs: An empirical and axiomatic analysis",
Mathematical Social Sciences 849 (1998): 203-218.
Aadland, D., Koplin, V. (2004). "Environmental Determinants of Cost Sharing", Journal of Economic
Behavior and Organization 53(4): 495-511.
Ambec, S., Sprumont, Y. (2000). "Sharing a River", Recherche en Energie from Laval.
Aumann, J.R., Dréze, J. (1974). "Cooperative games with coalition structures", International journal of game
theory, pp. 217-237.
Barrett, S. (1994). "Conflict and Cooperation in Managing International Water Resources", Policy
Research Working Paper, The World Bank, Policy Research Department, 1303
Baumol, W.J. (1986). Superfairness. MIT Press, Cambridge
Becker, N., Easter, W. (1991). "Cooperative and Noncooperative Water Diversion in the Great Lakes
Basin", in Dinar, A., Loehman, E. (eds.) Water quantity/quality management and conflict resolution:
Institutions, processes, and economic analyses. Westport: Praeger Publisher, pp. 321 336
Becker, N., Easter, W. (1997). "Water diversion from the Great Lakes: is a cooperative approach
possible?", International Journal of Water Resource Development 13 (1): 53-65
Becker, N., Ester, K. W. (1999). "Conflict and Cooperation in Managing International Water Resources
such as the Great Lakes", Land Economics 75 (2): 233-24
Bennett et al., 1997. "Facilitating international agreements through an interconnected game approach: the case of
river basins", in: Parker, D., Tsur, Y. (eds.), Decentralization and Coordination of Water Resource Management.
Kluwer Academic Publishers, Boston, MA, pp. 6185
Braden, J., Eheart, J., Saleth, R. (1991). "Bargaining Rules for a Thin Spot Water Market", Land
Economics 67 (3): 326-339
Caswell, M., Frisvold, G. (2000). "Transboundary Water Management. Game Theoretic Lessons for
Projects on the US-Mexico Border", Agricultural Economics 24: 101-111
Coplin W., and O'Leary, M. (1983). "Political Analysis through the PRINCE System", Learning Packages in the
Political Sciences (PS23). Public Affairs Program. Syracuse University
Coplin, W. and O'Leary, M. (1974). "Activities of the PRINCE Project", Policy Studies Journal, Summer
Coplin, W. and O'Leary, M. (1976). Everyman's Prince: A Guide to Understanding Your Political Problems.
Duxbury Press, New York
Crossmit, C.W. and Green, D.D. (1982). ``Water utility costs and rate design for fire protection services'' J. AWRA
74 (6): 270277
Dickinson, R., Heaney, J. (1982). "Methods for Apportioning the Cost of a Water Resource Project",
Water Resources Research 18 (3): 476-482
Dinar, A. (2001). Scale and Equity in water Resource Development: a Nash Bargaining Model", Natural
Resource Modeling 14 (4): 477-494
Dinar, A., Howitt, R. (1997). "Mechanism for Allocation on Environmental Control Cost: Empirical Tests
of Acceptability and Stability", Journal of Environmental Management 49:183-203
Dinar, A., Kannai, Y., Yaron, D (1986). "Sharing Regional Cooperative Gains From Reusing Effluent for
Irrigation", Water Resources Research 22 (3): 339-344
42
Dinar, A., Moretti, S., Patrone, F., Zara, S. "Applications of Stochastic Cooperative Games in Water
Resources", XV Italian Meeting on Game Theory and Application (IMGTA), Urbino, Italy, July 12-16,
2003
Dinar, A., Ratner, A., Yaron, D. (2002) "Evaluating Cooperative Game Theory in Water Resources", in
Dinar, A., Zilberman, D. (eds.), Economics of Water Resources: the Contribution of Dan Yaron. Kluwer
Academic Publisher, pp. 165-181
Dinar, A., Wolf, A. (1997). "Economic and Potential Considerations in Regional Cooperation Models",
Agricultural and Resource Economics Review, April: 7-22
Dorfman, R., Jacoby H.D., and Thomas, H.A. (1972). Models for Managing Water Quality. Harvard University
Press
Driessen, T. (1988). Cooperative Games, solutions and applications. Kluwer Academic Publishers, pp. 1-
12; pp. 91-110
Driessen, T., Tijs, S.H. (1986). "The cost gap method and other cost allocation methods for multipurpose
water projects", Water Resources Research 21 (10): 1469-1475
Duckstein, L., Bogardi, I. And Szidarovsky, F. (1979). "Tradeoff between regional mining development and
environmental impact", Proc. 16th International Symposium on Applications of Computers and Operations Research
in the Mineral Industry, SME/AIME Fall Meeting, October 17-19, Tucson (1979), pp. 355-364
Dufournaud, C., Harrington, J. (1990). "Temporal and Spatial Distribution of Benefits and Costs in River
Basin Schemes: A Cooperative Game Approach", Environment and Planning A 22: 615-628
Dufournaud, C., Harrington, J. (1991). "A linear constraint formulation for spatial and temporal cost
imputations in accord with Shapley values", Environment and Planning A 23: 1517-1521
Federal Inter-Agency River Basin Committee (1950). Proposed practices for economic analysis of river basin
projects. Washington DC
Fernandez, L. (2002). "Trade's Dynamic Solutions to Transboundary Pollution", Journal of
Environmental Economics and Management 43: 386-411
Frerichs, S., Easter, K.W. (1988). "Managing the Great-Lakes Common: An Evaluation of Recent
Institutional Changes", Economic Report ER-88-3, Dept. of Agricult. And Applied Econ., Univ. of
Minnesota, St. Paul, Minnesota, USA.
Fudenberg, D. and Tirole, J. (1992). Game Theory. Cambridge, Mass: MIT Press
Gately, D. (1974). "Sharing the Gains from Regional Cooperation: a Game Theoretic Application to
Planning Investment in Electric Power", International Economic Review 15 (1): 195-208
Giglio, R., Wrightington, R. (1972). "Methods for Apportioning the Cost among Participants in Regional
Systems", Water Resources Research 18 (3): 476-482
Guldmann, J.M., Kucukmehmetoglu, M. (2002). "International Water Resources Allocation and Conflicts: The Case
of the Euphrates and the Tigris", 42nd European Congress of the Regional Science Association, August 27-31, 2002,
in Dortmund, Germany
Hamers, H. J. M., Miquel, S., Norde, H.W., van Velzen, S. (2003). "Fixed Tree Game with Repeated
Players", Discussion Paper Tilburg University 2003-87
Harrington, J.J., Gidely, J.S. (1985). "The variability of alternative decisions in a water resources planning
problem", Water Resources Research 21: 1831-1840
Harrison, S., Tisdell, J. (1992). "Estimating an Optimal Distribution of Water Entitlements", Water
Resources Research 28 (12): 3111-3117
43
Harsanyi, J.C. (1959). "A bargaining model for the cooperative n-person game"' in Tucker, A.W. and Luce, R.D.
(eds.), Contributions to the Theory of Games, 1-4. Princeton University Press, pp. 325-355
Harsanyi, J.C. (1986). Rational Behavior and Bargaining Equilibrium in Games and Social Situations. Cambridge
University Press, New York
Hart, S. and Kruz, M. (1983). "Endogenous Coalition Formation", Econometrica 51, pp. 1047-64
Hart, S. and Kruz, M. (1983). "Endogenous Coalition Formation", Econometrica 51, pp. 1047-64
Heaney, J., Lippai, I. (2000). "Efficient and Equitable Impact Fees for Urban Water Systems", Journal of
Water Resources Management and Planning, March April 2000: 75-84
Holler, M. J., Li, X. (1996). "Efficient Public Good Pricing: An Application of Cooperative Game
Theory", Working Paper Faechergruppe Volkswirtschaftlehre from University of Hamburg, Institute of
Economics
James, L.D., and Lee, R.R. (1971). Economics of Water Resources Planning. McGraw-Hill Book Co.,
New York
Jorgensen, S., Zaccour, G. (2001). "Time consistent side payments in a dynamic game of downstream
pollution", Journal of Economic Dynamics and Control 25 (2001): 1973-1987
Just, R., Netanyahu, S. (1998). "International Water Resource Conflicts: Experience and Potential" in
Just, R., Netanyahu, S. (eds.), Conflict and Cooperation on Trans-boundary Water Resources. Kluwer
Academic Publishers, pp. 1-26
Just, Richard E. and Sinaya Netanyahu, 2004. "Implications of `Victim Pays' Infeasibilities for
Interconnected Games with an Illustration for Aquifer Sharing under Unequal Access Costs, Water
Resources Research, 40(5):W05S01.
Kaitala, V., Pohjola, M. (1995). "Sustainable international agreements on green house warming: a game theory
study", Annals of the International Society of Dynamic Games 2: 67-88
Kilgour, D. M., Nishikori, A., Okada, N. (1988). "Control Regulation of Water Pollution: An Analysis
Using Game Theory", Journal of Environmental Management 27: 179-194
Kogiku, K., Sheehan, M. (1980). "Game Theory Analyses Applied to Water Resources Problems", Socio
Economic Planning Sciences 15 (3): 109-118
Lejano, R. P., Davos, C. (1999). "Cooperative solutions for sustainable resource management",
Environmental Management 24 (2): 167-175
Lejano, R. P., Davos, C. (1995). "Cost allocation of multiagency water resource projects: A game-
theoretic approach and case study", Water Resources Research 31: 1387-1393.
Littlechild, S.C., Owen, G. (1973). "A simple expression for the Shapley value in a special case",
Management Science 20 (1973): 370-372
Loehman, E. (1995). "Cooperative Solutions for Problems of Water Supply", in Dinar, A., Loheman, E.
(eds.), Water quantity/quality management and conflict resolution: Institutions, processes, and economic
analyses. Westport: Praeger Publisher, pp. 301-319
Loehman, E., Orlando, J., Tschirhart, J., Whinston, A. (1979). "Cost Allocation for a Regional
Wastewater Treatment System", Water Resources Research 15 (2): 193-202
Loehman, E., Winston, A. (1976). "A generalized cost allocation scheme", in Stevens, A., Lin, Y. (eds.), Theory and
Measurement of Economic Externalities. Academic, New York, pp. 87-101
Loughlin, J. (1977). "The Efficiency and Equity of Cost Allocation Methods for Multipurpose Water
Projects", Water Resources Research 13 (1): 8-14
44
Luterbacher, U., Weigandt, E. (2001). "Water Control and Property Rights: An Analysis of the Middle
Eastern Situation"
Megiddo, N. (1978). "Computational complexity of the game theory approach to cost allocation for a tree",
Mathematics of Operations Research 3, 189-196
Montgomery, J., Nakao, M., Wichelns, D. (2002). "Game Theory Analysis of Competition for
Groundwater Involving El Paso, Texas and Ciudad Juarez, Mexico", Annual Meeting of the American
Agricultural Economics Association, Long Beach, California, July 28-31, 2002
Nakayama, M., Suzuki, M. (1976). "The Cost Assignment of Cooperative Water Resource Development -
a Game Theoretic Approach", Management Science 22 (10): 1081-1086
Nash, J.F.Jr. (1950). "The Bargaining Problem", Econometrica 18: 155-162
Nelson, A.C. (1995). System development charges for water, wastewater, and stormwater facilities. CRC, Boca
Raton, Fla.
Netanyahu, S., Just, R., and Horowitz, J. (1998). "Bargaining over Shared Aquifers: the Case of Israel and
the Palestinians". in Just, R., Netanyahu, S. (eds.), Conflict and Cooperation on Trans-boundary Water
Resources. Kluwer Academic Publishers, pp. 41-60
Owen, G. (1977). "Values of Games with a Priori Unions", in Hein, R. and Moeschlin O. (eds.), "Essays in
Mathematical Economics and Game Theory". Springer-Verlag, New York
Owen, G.T. (1993). "Characterization of a Game Theoretical Cost Allocation Method" Tilburg Working
Paper, Tilburg-Centre of Economic Research
Parker, T. (1943). "Allocation of the Tennessee Valley Authority projects", Transactions of the American
Society of Civil Engineers 108: 174-187
Petrosjan, L., (1993). Differential Games of Pursuit. World Scientific, New York
Petrosjan, L., (1997). "Agreeable solutions in differential games", International Journal of Mathematics, Game
Theory and Algebra 7 (2/3): 165-177
Ransmeier, J.S. (1942). The Tennessee Valley Authority: A Case Study in the Economics of Multiple
Purpose Stream Planning. Vanderbilt University Press, Nashville (Tennessee)
Ratner, A., Yaron, D. (1990). "Regional cooperation in the use of irrigation water, efficiency and game
theory analysis of income distribution", Agricultural Economics 4 (1) 1990
Rogers, P. (1993). "The Value of Cooperation in Resolving International River Basin Disputes", Natural
Resources Forum, May: 117-131
Rogers, P. (1994). "A model to generate Pareto-Admissible Outcomes for International River Basin
Negotiations" Nota di Lavoro Fondazione Enrico Mattei, 48-1994
Serghini, M. (2003) "La Problématique de l'Allocation des Coûts de Mobilisation des Ressources en
Eau", Hommes Terre & Eaux 32 (124): 20-24
Shapley, L.S. and Shubik, M. (1954). "A Method for Evaluating the Distribution of Power in a Committee System",
American Political Science Review 48: 787-792
Shubik, M. (1984). A Game Theoretical Approach to Political Economy. MIT Press
Straffin, P. and Heaney, J., (1981) "Game Theory and the Tennessee Valley Authority", International
Journal of Game Theory 10 (1): 35-43
Szidarovszky, F., Duckstein, L., Bogardi, I. (1984). "Multiobjective management of mining under water hazard by
game theory", European Journal of Operational Research 15 (2): 251-8
45
Tijs, S.H. (1981). "Bounds for the core and the -value" in Moeschlin, O. and Pallaschke, D. (eds.), Game Theory
and Mathematical Economics, pp. 123-132. North Holland, Amsterdam
Tijs, S.H. (1987). "An axiomatization of the -value", Mathematical Social Sciences 13, pp.177-181
United Nations, 1975. "Management of International Water Resources: Institutional and Legal Aspects" Natural
Resources/Water Series, No. 1, New York
Van den Brink, R., Van der Laan, G., Vasil'ev, V. (2003). "Harsanyi Solutions in Line-graph Games",
Tinbergen Institute Discussion Paper
World Bank (1993). "Water Resources Management", Policy Paper
World Bank (1995). Le Secteur de l'Eau au Maroc, Rapport N°12649-MOR, p.23.
Yaron, D. (2002). "An Approach to the problem of Water Allocation to Israel and the Palestinian Entity"
in Dinar, A., Zilberman, D. (eds.) Economics of Water Resources: the Contribution of Dan Yaron.
Kluwer Academic Publisher, pp. 201-217
Young, H.P. (1994a). "Cost Allocation", in Aumann, R.J. and Hart, S. (eds.), Handbook of Game Theory, vol. 2, pp.
1215-1217. Elsevier Science
Young, H.P. (1994b). Equity: in theory and practice. Princeton University Press
Young, H.P., Okada, N., Hashimoto, N. (1982). "Cost Allocation in Water Resources Development",
Water Resources Research 18 (3): 463-475
Zeuthen, F. (1930). Problem of Monopoly and Economic Welfare. Routledge & Kegan Paul, London
46