non-cooperative game theory: three fisheries games marko lindroos jss
Post on 21-Dec-2015
244 views
TRANSCRIPT
Non-cooperative game theory: Three fisheries games
Marko Lindroos
JSS
This lecture is about
Non-cooperative games classification Nash equilibrium
Applications in fisheries economics basic game (Mesterton-Gibbons NRM 1993) stage games (Ruseski JEEM 1998) repeated games (Hannesson JEEM 1997)
Non-cooperative games
Individual strategies for the players Reaction functions, best reply Nash equilibrium definition
Stages games at different levels
Repeated games, folk theorems, sustaining cooperative
behaviour as equilibria
Dynamic games
Why non-cooperative
Classification: strategic (static), extensive (dynamic),
coalition
Important in fisheries non-cooperation (competition) vs
cooperation
Division not clear, almost all games have both non-
cooperative and cooperative elements
Typically in economics non-cooperative game theory
dominates
What are non-cooperative games about
How fisher’s decisions interact with other fishers’
decisions
What is the best strategy for the fishers
What is exected to happen is the fishery? Depends on
rules of the game, number of players, biological factors
Why fishers behave as they do?
Assume rational choice
International fisheries negotiations
Nature of negotiations Countries attempt to sign and ratify agreements to
maximise their own economic benefits
Negotiations typically time-consuming
Agreements not binding self-enforcing or voluntary
agreements
Explaining the tragedy of the commons
Can we explain the seemingly irrational behaviour in the
world’s fisheries, overexploitation, overcapitalisation,
bycatch…
Non-cooperative game theory explains this behaviour
Non-cooperative games vs open access (freedom of the
seas)
Nash equilibrium
Each player chooses the best available decision
It is not optimal for any single player to unilaterally change
his strategy
There can be a unique equilibrium, multiple equilibria or
no equilibria
Fisher’s dilemma
Modified prisoner’s dilemma
Non-cooperation vs cooperation
Example 1: Two countries exploiting a common fish stock
Country 2
Deplete Conserve
Country 1 Deplete 3, 2 40, -5
Conserve -5, 40 30, 20
Fisher’s dilemma explanation
Deplete: Corresponds to non-cooperation. The country is
only interested in short-run maximisation of economic
benefits. No regulation. Conserve: Optimal management of the fishery.
Cooperative case. The cooperative solution (Conserve, Conserve)
maximises the joint payoffs to the countries, equal to 50.
However, neither of the countries is satisfied with the
cooperative strategy. Both would gain by changing their
strategy to Deplete (free-riding). This is the game-
theoretic interpretation of tragedy of the commons. In the Nash equilibrium (Deplete, Deplete) unilateral
deviation is not optimal for the countries.
Reaction (best response) functions
Gives the best decisions a player can make as a function
of other players’ decisions If a decision is not a best response it can not be a Nash
equilibrium Typically best response functions are derived from a set
of optimisation problems for the players. In an n player
game there are n best response functions. Nash equilibrium is found at the intersection of the best
response functions (solution to the system of equations) Strategy is best response if it is not strictly dominated
Repeated games
deterring short-term advantages by a threat or
punishment in the fisher’s dilemma escaping the non-
cooperative Nash equilibrium
folk theorems (understood not published)
credibility of threats
Numerical repeated game
Assume that the game in example 1 is repeated infinite
number of times. If one player deviates from the
cooperative strategy Conserve to the non-cooperative
strategy Deplete, it will also trigger the other player to
choose Deplete forever after the deviation. This means
that both countries punish severely deviations from the
common agreement. Cooperation can be sustainable if the present value of
choosing Conserve is higher than deviating once from
cooperation. Present value of cooperation to player 1 when discount
rate is 5%:
Cooperation vs. deviation
This infinite sum of the geometric progression and can be
solved as follows: = 600
Next we calculate the present value of deviation. Country 1
first receives 40 and thereafter only 3 since country 2 uses its
trigger strategy, according to which it never again signs an
agreement. Hence, the present value of deviation is:
= 37 +3/(1-0.95) = 97
nCPV05.1
30...
05.1
30
05.1
3030
2
95.01
30
CPV
nDPV05.1
3...
05.1
3
05.1
340
2
n05.1
3...
05.1
3
05.1
33340
2
Tragedy of the commons solved
We see that the present value of deviating is clearly
smaller and thus, cooperation (Conserve, Conserve) is
now the equilibrium of the repeated game.
Note that the discount rate is critical in repeated games.
As discount rate approaches infinity the present value of
cooperation approaches 30 and the present value of
deviation approaches 40. The critical discount rate, over
which deviation is profitable, is therefore finite.
The first non-cooperative fisheries game
Assume there are n players (fishers, fishing firms,
countries, groups of countries) harvesting a common fish
resource x Each player maximises her own economic gains from the
resource by choosing a fishing effort Ei
This means that each player chooses her optimal e.g.
number of fishing vessels taking into account how many
the other players choose As a result this game will end up in a Nash equilibrium
where all individual fishing efforts are optimal
Building objective functions of the players
Assume a steady state:
By assuming logistic growth
the steady state stock is then
0)(
1
n
i
ihxFdt
dx
)1( 1
R
Eq
Kx
n
i
i
Stock biomass depends on all fishing efforts
hi=qEix
Objective function
Players maximise their net revenues (revenues – costs)
from the fishery max phi –ciEi
Here p is the price per kg, hi is harvest of player i, ci is unit
cost of effort of player i
ii
n
ii
ii EcR
EqKpqE
)1(max 1
Deriving reaction curves of the players
The first order condition for
player i is
The reaction curve of player i
is then
0
21
22
i
n
ijji
i
i cR
KEpqKEpq
pqKE
)1(22
1
i
n
ij
ji b
q
REE
bi=ci/pqK
Equilibrium fishing efforts
Derive by using the n reaction curves
The equilibrium fishing efforts depend on the efficiency of
all players and the number of players
1
)1()1(
)1()1(
n
ij
jii bqn
Rb
qn
nRE
Illustration
Nash-Cournot equilibrium
Symmetric case
Schäfer-Gordon model
Exercises
Compute the symmetric 2-player and n player equilibrium.
First solve 2-player game, then extend to n players.
A two-stage game (Ruseski JEEM 1998)
Assume two countries with a fishing fleet of size n1 and n2
In the first stage countries choose their optimal fleet
licensing policy, i.e., the number of fishing vessels. In the second stage the fishermen compete, knowing how
many fishermen to compete against
The model is solved backwards, first solving the second
stage equilibrium fishing efforts Second, the equilibrium fleet licensing policies are solved
Objective function of the fishermen
The previous steady state stock is then
The individual domestic fishing firm v maximises
))(
1( 21
R
EEqKx
vvv cexpqe 111max
1
111
1
wheren
vw
wv eeE
Reaction functions
In this model the domestic fishermen compete against
domestic vessels and foreign vessels
The reaction between the two fleets is derived from the first-
order condition by applying symmetry of the vessels
02
1
22
12
12
1
1
1
cR
KEpqKepqKepqpqK
e
n
vwwv
v
v
venEbq
R
n
nE 112
1
11 ])1([
1
Equilibrium fishing efforts
Analogously in the other country
By solving the system of two equations yields the equilibrium
])1([1 1
2
22 Eb
q
R
n
nE
)1
1(
21
22 nn
b
q
RnE
)1
1(
21
11 nn
b
q
RnE
Equilibrium stock
Insert equilibrium efforts into
steady state stock
expression
The stock now depends
explicitly on the number of
the total fishing fleet
21
21
1
])(1[
nn
bnnKx
Equilibrium rent
Insert equilibrium efforts and
stock into objective function
to yield
221
21
1)1(
)1(
nn
bnRpKP
First stage
The countries maximise their welfare, that is, fishing fleet
rents less management costs
The optimal fleet size can be calculated from the FOC
(implicit reaction function)
FnPW 111max
0)1(
)1)(1(3
21
221
1
1
Fnn
bnnRpK
n
W
Results
Aplying symmetry and changing variable m = 1+2n
With F=0 open access
0)21(
)1(3
2
F
n
bRpK
1
)1(
2
13/12
1 F
bRpKn
Discussion
Subsidies
Quinn & Ruseski: asymmetric fishermen
entry deterring strategies: Choose large enough fleet so
that the rival fleet is not able make profits from the fishery
Kronbak and Lindroos ERE 2006 4 stage coalition game
Repeated games – a step towards cooperation
When cooperation is sustained as an equilibrium in the
game
The game is repeated many times (infinitely)
The players use trigger strategies as punishment if one of
the players defects from the cooperative strategy Trigger here means that defection triggers non-
cooperative behaviour for the rest of the game
Cooperation means higher fish stock than non-
cooperation, in the defection period the stock is between
cooperative and non-cooperative levels
Cooperative strategies
Cooperative effort from
SG-model
Cooperative fish stock
Cooperative benefits
)1(2
bnq
RECi
1
Ci
CCiC
icExpqE
)1(2
bK
xC
Optimal defection effort
Best response when all
others choose the
cooperative strategy
Optimal defection effort
)1(4
)1()1(
2b
nq
Rnb
q
REDi
)1
1(4
)1(
nq
bREDi
Non-cooperative strategies
Effort
Stock
)1()1(
bqn
RENi
)1
1(
n
nbKxN
Cooperation vs. cheating
Benefits from cheating
Condition for cooperative
equilibrium
Di
Di
Di cE
nq
bnRE
R
qKKpqE
)
2
)1)(1(((
1
)( Ni
NNiD
iDD
iCi
cExpqEcExpqE
Discussion
Hannesson (JEEM 1997) similar results
Higher costs and lower discount rate enable a higher
number of countries in the cooperative equilibrium
Self-enforcing agreements
On Species Preservation and Non-Cooperative Exploiters
Outline
Motivation
Model
Results
Conclusion
Discussion
Motivation
Combining two-species models with the game theory
What are the driving force for species extinction in a two-
species model with biological dependency?
Does ‘Comedy of the Commons’ occur in two-species
fisheries?
What are the ecosystem consequences of economic
competition?
Modelling approach
Two-species
n symmetric competitive exploiters with non-selective
harvesting technology
Fish stocks may be biologically independent or dependent
What is the critical number of exploiters?
Analytical independent species model
S-G model
Derive first E* as the optimal effort, it depends on the
relevant economic and biological parameters
An n-player equilibrium is then derived as a function of
E*and n.
Relate then the equilibrium to the weakest stock’s size to
compute critical n*, over which ecosystem is not
sustained.
Dependent vs independent species
Driving force of extinction:
Independent species Biotechnical productivity Economic parameters
Dependent species Biological parameters must be considered Gives rise to a complex set of conditions For example:
Natural equilibrium does not exist‘The Comedy of the Commons’
Numerical dependent species model
Cases illustrated: Biological competition, symbiosis and
predator-prey
Case 1: Both stocks having low intrinsic growth rate
Case 2: Both stocks having a high intrinsic growth rate
Case 3: Low valued stock has a low intrinsic growth rate,
high value stock has a high intrinsic growth rate.
Case 4: Low valued stock has a high intrinsic growth rate,
high value stock has a low intrinsic growth rate.
Parameter values applied for simulationp1 p2 Rlow Rhigh K1=
K2
c q OA MS θ1 θ2
1 2 0.3 0.9 50 7 0.5 60 60 [-0.2;0.2] [-0.2;0.2]
Case 1: low intrinsic growth rate
-0.2
-0.1
0
0.1
0.2 -0.2
-0.1
0
0.1
0.2
0
20
40
60
theta2(beta)theta1(alpha)
ncrit
Case 2: High growth
-0.2
-0.1
0
0.1
0.2 -0.2-0.1
00.1
0.2
0
10
20
30
40
50
60
theta2(beta)theta1(alpha)
ncrit
Case 4: Low valued stock has a high intrinsic growth rate, high value stock has a low intrinsic growth rate.
-0.2
-0.1
0
0.1
0.2 -0.2
-0.1
0
0.1
0.2
0
20
40
60
theta2(beta)theta1(alpha)
ncrit
Opposite case 3
Conclusion
‘Tragedy of the Commons’ does not always apply
A small change in the interdependency can lead to big
changes in the critical number of non-cooperative players
With competition among species a higher intrinsic growth
rate tend to extend the range of parameters for which
restricted open access is sustained
Discussion
From single-species models to ecosystem models
Ecosystem approach vs. socio-economic approach
Agreements and multi-species