Transcript Diapositiva 1

Sergio Beraldo
(University of Naples “Federico II” & ICER)
Lectures delivered at the University of Prague (VSE)
October 2011
III - IV
Outline of the course
 I: Institutions
 II: Coordination
 III: Cooperation 1
 IV: Cooperation 2
 V: Institutional structure and economic performance
III LECTURE
Cooperation
Cooperation
 For mutually beneficial exchange to take
place it is necessary that trading partners do
cooperate
 Institutions which ensure an acceptable
level of cooperation are necessary
The problem of cooperation
 In many social situations in which your opponent is
behaving in such a way as to make the achievement of
a cooperative outcome possible, it may seem contrary
to your self-interest to resist the temptation of
cheating
 There are gains to be made by resisting such a
temptation
 Indeed: a surplus may be generated if selfishness (or
lack of confidence) were put aside;
The problem of cooperation
 D.C. North, Journal of Economic Perspectives, 1991 →
Individuals will usually find it worthwhile to cooperate
with other players when:



the play is repeated
they possess complete information about the other player's
past performance
there is a small number of players
 [Think at earliest economies characterized by local
exchange within a village – kin or reputation ensure
compliance]
The problem of cooperation
 North (1991) → Cooperation is difficult to sustain when:



the game is not repeated (or there is an endgame);
information on the other players is lacking
there is a large number of players
 Notice : the productivity gains coming from specialization
and division of labour can only be reaped if there emerges
an institutional structure solving the problem of human
cooperation under the latter conditions
 Anonymous exchange requires cooperative individuals
Are human beings
cooperative beings?
 “Evolution is based on a fierce competition
between individuals and should therefore reward
only selfish behaviour...yet we observe cooperation
on many level of biological organization. Genes
cooperate in genomes. Chromosomes cooperate in
eukaryotic cells. Cells cooperate in multicellular
organisms. There are many example of cooperation
among animals. Humans are the champions of
cooperation...cooperation is the decisive
organizing principle of human society. The
question of how natural selection can lead to
cooperative behaviour has fascinated evolutionary
biologists for several decades” (Nowak, Science,
2006)
Are human beings
cooperative beings?
 “Human societies…are based on a detailed
division of labour and cooperation between
genetically unrelated individuals in large
groups”, (Fehr&Fischbacher, Nature, 2003)
 "The evolution of cooperation among nonrelated individuals is one of the
fundamental problems in biology and social
sciences.", (Hauert et al., Science, 2002)
The standard framework for the study of
cooperation: the Prisoner’s dilemma
Player 2
Player 1
Cooperate
Defect
Cooperate
1,1
9,0
Defect
0,9
5,5
How can cooperation emerge
in an evolutionary setting?
 M. Nowak (Science, 2006) discusses five possible
routes:
 Direct reciprocity
 Indirect reciprocity
 Network reciprocity
 kin selection
 Group selection
 Nowak (2006) shows that in any of these cases
cooperation can emerge if the cost to benefit
ration of the cooperative act is below a certain
threshold
Direct reciprocity
 Experimental evidence: individuals do cooperate even in one shot




interactions or in the last stage of a repeated game (Gintis & Bowles,
2003)
To explain how cooperation did evolve it is necessary to take into
account that for individuals living in small groups of hunter-gatherers
(humans until 10.000 years ago) it would have been relatively simple to
avoid punishment by joining a different group (Gintis and Bowles,
2003)
The evidence that cooperation among animals is actually based on
reciprocity is scarce (Silk, 2005; Fehr & Fischbacher, 2003;
Hammerstein, 2003; Clutton-Brock, 2009)
Hammerstein (2003) – Genetic and cultural evolution of cooperation,
MIT press: “After three decades of worldwide research on reciprocal
altruism and related phenomena, no more than a modest number of
animal examples have been identified”
How is it possible to extend the folk theorem from a group of two to a
group of n individuals? (public good game – if you do not cooperate
you are actually punishing also who cooperates)
Direct reciprocity: I scratch your back
and you’ll scratch mine
 Trivers, The Quarterly Review of Biology, 1971.
 Folk Theorem;
 Nowak (2006) → prob. of another encounter between the
same two individuals> cost/Benefit ratio of the
cooperative act
 As n increases it becomes more and more difficult for
cooperation to evolve
 Stated in different terms: the probability that a proper
number of individuals be sufficiently forward looking
shrinks as n increases (Bowles & Gintis, 2003; Beraldo &
Turati, 2011)
Indirect reciprocity: I make a point of going to other people’s
funeral otherwise they won’t come to mine (Yogi Berra,
baseball player)
 Remember yogi bear, the Hanna-Barbera’s cartoon, (however cartoonists
denied they were inspired by Berra). Berra is well known for his Yogiism,
often taking the form of either obvious tautology or a paradoxical
contradiction.
 However, the sentence above is less paradoxical of what may seem
 Indirect reciprocity – based on status or reputation
 IR has recently obtained considerable attention
 Social sciences → cooperation when interaction is not face to
face
 Evolutionary biology → Nowak and Sigmund, Nature, 2005:
→ “humans not only feel strongly about interactions involving
them directly, they also judge the actions between third parties
(e.g. gossip), indirect reciprocity is therefore likely to be
connected with the origins of moral norms and has probably
played a pivotal role in the evolution of collaboration and
communication (language)”
Open problems
Actual debate: does indirect
reciprocity work in sustaining
cooperation in an evolutionary
environment?
How does it actually work?
What is a reliable model of
cooperation based on indirect
reciprocity?
Two competing models
 The image-scoring model (Nowak and Sigmund, Science,
1998);
 The good standing model (Sugden, 1986)
 The difference is that the latter distinguishes between
justified and unjustified defection, while the former does
not;
 Although computer simulations point out that the standing
model would perform better under a wide set of
circumstances, it is often maintained that it requires
individuals with an implausibly large capacity of processing
recursive information (e.g. Engelmann and Fischbacher,
Games and Economic Behaviour, 2009) .
Open Problems
 Good standing model → B’s standing does matter
t=m
t=m+1
A
B
Help, Not Help
Help, Not Help
b, -c
b, -c
C
C
A
Open Problems – Good standing
 Recursive information: a potential donor k should
not only know the behaviour of j in his last interaction
as a donor, i.e. whether j cooperated or defected; as k’s
behaviour has to be contingent on j’s standing, k
should be aware of the standing of j', the potential
recipient with whom j last interacted as a potential
donor. As the standing of j' depended on the choice
made when j' acted as potential donor of j'', k should
be aware of j''’s standing at the time, and so on.
Is recursive information necessary
for the standing model to work?
It is possible to prove that the
standing model does not require
recursive information (see Beraldo S.,
International Review of Economics, 2011)
Does indirect reciprocity work?
 Adam Smith (1763/1978) was the first to suggest
that cooperative practices emerge whenever the
cost of acquiring a good reputation is more than
offset by the material gains accruing from it. Any
economist would still agree that the need of
displaying a good image is (at least partially)
driven by the desire of reaping the fruits accruing
from market interaction. A good standing or
image, in other words, pays.
 Greif, A., 1989. Reputation and coalitions in
medieval trade: evidence on the Maghribi traders.
Journal of Economic History, 49, 857-882.
(overseas commerce made possible by indirect
reciprocity)
Does indirect reciprocity work?
 As the development of a far than
rudimentary language is a necessary
condition for large scale cooperation based
on it, indirect reciprocity is certainly not
suitable to provide a convincing explanation
for the widespread level of cooperation
observed in nature (e.g. de Waal., 2006)
Does indirect reciprocity work?
 In a forthcoming book, Sam Bowles and Herbert Gintis
(2010) argue that indirect reciprocity is also unable to give
account of the evolution of the biological traits which make
humans a cooperative species, since, as soon as human
communities get larger, high quality information is no
more available and the rate of errors in perception becomes
excessively high for indirect reciprocity to work
Does indirect reciprocity work?
 Cooperation is not a single phenomenon with a unified
causal explanation. Even if a form of reciprocity based on
status or reputation is inadequate to provide a consistent
account of the biological bases of our attitude to cooperate,
there are reasons to believe that it helps in drawing a
reliable picture of the forces sustaining cooperative
practices in many social and economic environments [In
this respect, the standing model, starting from more
realistic hypotheses about human behaviour, seems to be
the preferred candidate to catch the basic aspects of how
reputation works and is affected by one’s conduct]
Kin Selection: I will jump into the river to save
two brothers or eight cousins (J.B.S.Haldane, biologist)!
 Hamilton, Journal of Theoretical Biology, 1964;
natural selection favours cooperation when the
donor and the recipient are genetically related
 Hamilton’s rule: coefficient of genetic relatedness
(probabiliy of sharing a gene) > C/B
 The selfish gene (Richard Dawkins, 1976)
 Kin selection is unable to explain why individuals
do cooperate with those who are not genetically
related to them
 Es: why do individuals donate resources under
conditions of anonimity? (e.g., blood donations)
Network reciprocity –
Group selection
 Network reciprocity: cooperators give
rise to clusters
 Group selection: groups of cooperators
fare better than groups of defectors
To sum up
 No one of the five routes analyzed by Nowak can
explain the emergence of cooperation without
dropping one of the following two hypotheses:
 well-mixedness;
 Anonimity
 This problem is fundamental as anonymity and
well-mixedness are typical of many economic,
social and biological environments
Two further hypotheses for the evolution of
cooperation
 The green beard hypothesis (Dawkins, 1976), and the
hypothesis of voluntary participation (e.g. Tullock,
1985; Hauert et al. 2002, 2007).
 The green beard hypothesis → the evidence is very
limited (Keller and Ross, 1998); its general
applicability, has been deeply questioned (Fehr and
Fischbacher, 2005)
 The voluntary participation hypothesis → if the
Prisoner’s Dilemma payoffs are retained, cooperation
remains a (weakly) dominated strategy; in pairwise
interaction the only effect is that of replacing the
defection equilibrium with one of non-participation
No more prisoners of the dilemma
 Theorists have generally used the Prisoner’s Dilemma
as the paradigm model of cooperation problems. In
doing so, they may have set themselves an
unnecessarily difficult challenge. Furthermore, by
neglecting the existence of all those situations in
which the problem of cooperation is less intractable
than in the simple Prisoner’s Dilemma, the analysis
risks to be both incomplete theoretically and
dangerous socially (Worden and Levin, 2007)
No more prisoners of the dilemma
 Among the restrictive features of the Prisoner’s
Dilemma, a prominent one is that, in any given
interaction, an individual must act either prosocially or anti-socially; there is no opportunity to
be simply asocial
 Clutton-Brock (2009) → “social animals are
seldom constrained to cooperate with particular
partners and can develop profitable relationships
and terminate unproductive ones”
 This is even more true for humans, given their
higher cognitive capacities
No more prisoners of the dilemma
 As emphasized before Adam Smith’s pointed out
that honest behaviour is profitable for the
individuals performing it whenever the
opportunity is given of terminating trading
relationships with untrustworthy partners
 The voluntary participation hypothesis adds an
asocial strategy, that of opting out of the
interaction. However, in anonymous settings with
pairwise interaction, if the Prisoner’s Dilemma
payoffs are retained, the only effect is to replace the
defection equilibrium with one of nonparticipation
No more prisoners of the dilemma
 Beraldo and Sugden (2010) propose a voluntary
participation model which retains the dyadic form of
the Prisoner’s Dilemma;
 In their model the benefit that each player derives
from the cooperative activity (given the other’s
cooperation) is an independent realisation of a
random variable, known to the relevant player before
the game is played.
 Modelling the payoffs from the cooperative outcome as
subject to random variation is consistent with the
evidence that both in humans and among animals, the
outcome of many strategic situations implying gains
through joint activity, depends on subject to subject
variation (e.g. Johnson et al., 2002)
Results
 Beraldo and Sugden (2010) show that - provided the upper
bound of the distribution of cooperative benefit is not too
low - there is an equilibrium in which beneficial
cooperation occurs. The non-participation option plays an
essential part in this result, as it holds down the
equilibrium frequency of cheating and this allows
cooperation to persist
 In their model, cooperative behaviour does not depend on
benevolence, reciprocity or fear of punishment; it occurs
because the benefit that an individual would derive from
mutual cooperation is sometimes great enough to make it
worthwhile to run the risk that the opponent will cheat
The model
 Large number of individuals, interacting anonymously
in an indefinitely long sequence of periods.
 In each period, individuals are randomly matched to
play a two-player game.
 In a representative game between players i and j, the
benefits from cooperation xi and xj are independent
realizations of a random variable X whose distribution
f(.) is continuous with support [xmin, xmax].
 Each player knows its own benefit but not that of the
other player. Given this knowledge, it chooses one of
three options – to cooperate (C), to cheat (D), or not to
participate (N).
Payoff matrix
[Beraldo-Sugden, 2010]
Player 2
N
Player 1
C
D
N
0, 0
0, 0
0, 0
C
0, 0
xi, xj
-b, a
D
0, 0
a, -b
-c, -c
xmax > a > xmin  0; b > a > c > 0.
Payoff matrix
N
C
D
xmax > a > xmin →C or D may be the better
response to C, depending on the realization of X.
b > c → D is better than C as a response to D.
N
0, 0
0, 0
0, 0
C
0, 0
xi, xj
-b, a
D
0, 0
a, -b
-c, -c
a > 0 → cheating gives a higher payoff than nonparticipation if the opponent cooperates;
c > 0 → the opposite is the case if the opponent
cheats; b > a → implies that the benefit from
cheating a cooperating co-player is less than the
cost inflicted on the latter.
Results
 If xmax > ab/c (more intuitively: provided that the
upper tail of the distribution of cooperative benefit is
not too short), there is at least one interior or
boundary equilibrium in which both C and D are
played with positive probability.
 These equilibria are ESS
Some Comments
 If participation in a potentially cooperative activity is voluntary, the
frequency of cheating can be held down to a level at which some
mutually beneficial cooperation can occur even in anonymous wellmixed populations
 Cooperative behaviour may not depend on benevolence, reciprocity or
fear of punishment; it may occur because the benefit that an individual
would derive from mutual cooperation is sometimes great enough to
make it worthwhile to run the risk that the opponent will cheat
 Cooperation is seen as a risky strategy, worthwhile only if the
probability that any occasional opponent will cheat is sufficiently low
 In the game, getting to the cooperative outcome may be in some
occasions so valuable for the players involved, that each of them would
risk cooperation rather than loosing the chance of getting such an
advantage, even if by so doing it is unavoidable to get exposed to the
risk of being cheated
Discussion
 There are differences between this framework,
some of the evidence gathered by zoologists so far,
and various strategic situations envisaged by
game-theorists.
 First note that: some apparently cooperative
behaviours are forms of mutualism, in which any
individual maximizes its own fitness and any effect
on the fitness of others is coincidental and does
not contribute to the selection pressures
maintaining the behaviour (e.g. Clutton-Brock,
2002, 2009).
Suricata Suricatta
.
suricata suricatta, small mammals
living in arid areas of southern
Africa (Clutton Brock et al., 1999).
Going on guard when no other
individual is guarding, may have
immediate, direct benefits. In cases
like these, variously termed as
mutualism, by-products or mutual
benefits, acts by one individual
confer immediate benefits to the
actor and (only) coincidentally to
others
Snowdrift game (R. Sugden, 1986)
V → Benefir to get
out of the
snowdriftC1 →
cost to dig alone
player 1;C2→ cost
to dig alone for
player 2
Player 1
even if the relevant player is
sure that the opponent defects,
it is in her interest to dig: it is
so important to get out of the
snowdrift that each player
would rather do all the digging
himself rather than remain
stuck.
Player 2
Dig
Not Dig
Dig
V- C1,
V-C2
V- C1, V
Not dig
V, V- C1
0,0
Each player may be
tempted to follow a
smaller prey instead
of obeying to a
concerted plan
suitably devised to
catch a deer - the
relevant player does
better by cooperating
only if the opponent
cooperates too.
Stug-hunt game
Player 2
Hare
Stag
Hare
2,2
2,0
Stag
0,2
3,3
Player 1
Cooperation in animal societies
 It is now clear that there are substantial differences
between humans and animals. In animal societies,
where some apparently cooperative behaviours are
indeed forms of mutualism, cooperation is mostly
based on kin, being therefore quite rare in groups
consisting of genetically unrelated individuals
(Clutton-Brock, Science, 2009)
A different view
based on empathy
The age of empaty