Upload
trinhnhu
View
215
Download
0
Embed Size (px)
Citation preview
Tilburg University
Are moral objections to free riding evolutionarily stable?
Güth, W.; Nitzan, S.
Publication date:1993
Link to publication
Citation for published version (APA):Güth, W., & Nitzan, S. (1993). Are moral objections to free riding evolutionarily stable? (CentER DiscussionPaper; Vol. 1993-2). CentER.
General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
- Users may download and print one copy of any publication from the public portal for the purpose of private study or research - You may not further distribute the material or use it for any profit-making activity or commercial gain - You may freely distribute the URL identifying the publication in the public portal
Take down policyIf you believe that this document breaches copyright, please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
Download date: 07. Sep. 2018
CBMR
84141993
2
~i i~ - DISCUSSIOri~;o~ paper~ ' Research
i!n~i Niii Nim ni n~ nm um ni~;u~ ~i
. ~.-~
No. 9302
ARE MORAL OHTECITONS TO FREERIDING EVOLiTITONARILY STABLE?
try Werner Guthand Shmuel Nitzan
January 1993
ISSN 0924-7815
ARE MORAL OBJECTIONS TO FREE
RIDING EVOLUTIONARILY STABLE?~
Werner Giith
Department of Economics
L'niversity of Frankfurtt
and
Shmuel Nitzan
Department of Economics
Bar Ilan Universityt
November 1992
'This paper waa written while both authore were viaiting the CentER for Economic Research at
Tilburg University. We are indebted to CentER Cor its aupport and excellent reaearch environment.
14'erner Guth gratefully acknowledgea the generous Alexander von Humboldt-award via the Dutch Sci-
ence Foundation (N14'O) which made his stay at CentER poesible.
tP.O. Box 11 13 32 (Fach 81), D-6000 Franlcfurt~M., Federal Republic of Germany
tRamat Gan, 52900, Israel
Abstract
In this study basic preferences, namely the utiGty impGcation of ftee riding, are
determined endogenously. By assuming individual rationality with respect to given
preferences, the motivational structure results from an evolutionary analysis which
compares the relative reproductive success of all possible preference types. The
tastes that emerge are usually different from those assumed in models resorting to
altruism, social sense of guilt, or moral obligations. In general, an effective social
conscience preventing (ree riding is not evolutionarily stable. Thus, voluntazy
conttibutions to the provision of public goods are not to be expected. In a more
general framework there are, however, two notable exceptions, namely unanimity
games and two person games where a contribution by a single player is sufficient
to produce the public good.
1
One of the most prominent social dilemmas arises when economic agents are asked to
voluntarily provide a public good. In such games the possibility of nobody contributing
(or contributing sufficiently) is a rather stable and viable one, e.g., in the sense of being
a strict equilibrium, that is, a combination of strategies where every player suffers from a
unilateral deviation. The dilemma is due to the existence of other strategy configurations
that can makeeverybody better off. Unlike in the so called prisoners' dilemma situations,
in binary games the players' equilibrium strategies need not be dominant strategies.
Typically, both contributing and not-contributing can be the unique best reply to the
others' behavior.
A rather unsatisfactory way to overcome the dilemma associated with private provi-
sion of public goods is to assume that people have moral commitment to social cooper-
ation or moral objections against free riding. In other words, to achieve better (more
efficient) allocation of public goods, one introduces a utility determinant which repre-
sents social conscience: utility is derived from acts of social cooperation or contribution,
or conversely, disutility is associated with non-cooperative behavior such as free riding.
James Andreoni (1989), for example, studies charity giving assuming impure altruism
where people derive some utility ("warm glow" ) from the act of giving. Thomas Palfrey
and Howard Rosenthal (1988) assume that the altruistic components of utility are drawn
from a distribution that is identical for all individuals and common knowledge. In their
game of incomplete information they analyze the effect of altruism on behavior. More
recently, Heinz Hollander (1990) offers an axiomatic model of social exchange in which
cooperative behavior is motivated by the expectation of emotionally prompted social
approval.
Although we do not deny that such motivational factors exist, there is a lot of evidence
that they are often insufficient to resolve the social dilemma. More importantly, such an
approach simply transforms the problem of resolving the social dilemma to the problem
of rationalizing or explaining the emergence of such preferences that alleviate or resolve
the social dilemma.
There have been several attempts to determine endogenously the players' motivational
structure or incentives to cooperate. For example, Robert Frank (1987) focuses on the
2
problem of identifying the kind of preferences that maximize the attainment of selfish
objectives. Guttman, Nitzan and Spiegel (1992) and Mark Gradstein (1992) study the
possibility of altruistic preferences determination by the individuals themselves in a
sequential strategic game with complete information. Preferences are determined in the
earlier stage of the game, e.g., by investment in education, and the public good provision
is determined in the later stage of the game. An alternative approach is pursued by
Ingmar Hansson and Charles Stuart (1990). Since social biology tells us (see, for example,
Richard Dawkins (1976) or Edward Wilson (1980)) that preferences and, in particular,
benevolence or altruism are mainly adoptive, they study natural selection of preferences
using a golden-age model with endogenous population in which in a long-run equilibrium
all agents have preferences with maximum biological fitness. Our study can be viewed
as another attempt of this latter approach. However, the endogenous determination of
the motivational structure in the current paper is based on arguments of evolutionary
stability in strategic settings.
Whereas in evolutionary biology, John Maynard Smith (1982), a genotype is assumed
to determine behavior, we rely on the modified evolutionary approach originally proposed
by ~~'erner Guth and Menachem Yaari (1992). According to this approach genetical
evolution determines preferences or motivational forces and players behave rationally
with respect to any given preferences. Equipped with this methodology, Guth and Yaari
(I992) are able to justify the existence of bilateral reciprocal incentives, i.e., a desire
to exchange in kind. Werner Guth (1990) generalizes this result to situations where
the other's type is not known. That is, one dces not know whether the other player is
eager to exchange in kind or not when interacting with him. Using a similar approach
Werner Guth and Klaus Ritzberger (1992) are able to rationalize resistance against mass
immigration and Werner Guth and Hartmut Kliemt (1992) explain cooperation based
on trust. Given these "positive" results, it is tempting to investigate also in the context
of voluntary provision of public goods whether a moral objection against free riding is
evolutionarily stable and whether its existence is sufficient to overcome the dilemma
inherent in such situations.
In the next section we present the basic model of voluntary contributions to the
3
provision of a discrete public good in the presence of two types of individual social
conscience. The unique solution of this multi-player basic game is derived in Section II.
Our evolutionary approach is then described in Section III. Our multi-agent adaptation
of the evolutionary approach (proposed by Guth and Yaari (1992) in the context of
two interacting agents) involves a number of problems that need to be addressed. In
particular, the definition of evolutionarily stable strategies must be modified. There
exist alternative specifications that we consider and compare to each other. Sections IV
and V contain, respectively, our results in the general case and in four special boundary
cases. It turns out that, in general, the assumption that individuals develop an effective
social conscience which is sufficient to prevent them from free riding is not justified. The
emergence of such a conscience is not evolutionarily stable.
The unique evolutionarily stable conscience is ineffective, namely with it the free-
riding problem cannot be resolved as in the resulting equilibrium no player contributes
to the provision of the public good. The exceptions to these conclusions are obtained in
two special boundary cases. In a unanimity game where the public good is produced only
when everybody contributes, or in two-player games where the public good is produced
if at least one player contributes. Section VI contains some brief concluding remarks.
I The basic model
Consider a group N-{ 1, ..., n} of individuals, e.g., a local community or tribe, who
may or may not voluntarily provide a discrete public good, e.g., a public library or
protection against invasion by foreign enemies. Each individual i has two choices, ó; - 1,
i.e., to contribute to the provision of the public good, or ê; - 0, i.e., not to contribute.
If the numbern
N(ó) - ~ b;i-1
exceeds the threshold m with n~ m~ 1, then the public good is provided. If the group
falls short of providing enough, that is, N(ó) c m, then the public good is not provided.
It is assumed that the cost of contributing cannot be avoided even if N(ó) c m. The
4
utility of the public good is normalized to 1 so that the payoff U;(ó) of all players i E N
for all strategy vectors b- ( ó~, ..., 6„) with 6; E{0,1 } can be written as follows:
1-c if N(ó)~m and ó;-1
-c if N(ë) G m and À; - 1U;(ó) -
1 if N(ó) ~ m and ó; - 0
0 if N(ó) c m and 6; - 0
Here c, 0 G c C 1, is the cost of the individual contribution which can be avoided only
by choosing Á; - 0. This game was studied by Thomas Palfrey and Howard Rosenthal
(1984). A number of alternative extended versions of it have been more recently studied
by Anat Admati and Motty Perry (1991), Mark Bagnoli and Barton Lipman (1989) and
Shmuel Nitzan and Richard Romano (1990).
Observe that all strategy vectors ó' -(êi,...,ó;,) with N(ó') - m are pure strategy
equilibria of this game. (ê~,...,ê„) with N(ó) - 0 is also a strict equilibrium point in
the sense that every player loses by unilaterally deviating from his equilibrium strategy.
Here we are not interested in the model described above but in a slightly modified version
where the payoff function U;(ó) is substituted by U;(ë) which differs from U;(ó) only for
strategy vectors b with N(ó) ? m and ó; - 0 for which
U,`(ó) - 1- e instead of U;(b) - 1.
Palfrey and Rosenthal ( 1988) consider a related situation using a game of incomplete
information about the players' altruism.
The preference parameter e is not an exogenous parameter. As will become clear in
the next section, its endogenous determination is the distinctive and more interesting
characteristic of our approach. Assume, for instance, that e is positive. An individual i
facing a situation N(ó) ? m and 6; - 0 would thus have bad feelings when ( s)he free rides.
If e, however, is negative, such an individual would even enjoy free riding in addition to
getting the public good costlessly. Our main concern is with the determination of the
value of this parameter representing the individual social conscience.
5
Note that unlike in the otiginal model with e; - 0, for all i E N, now strategy
combinations ó with N(ó) 1 m can also be strict equilibria. Such a strategy vector 6 is
a strict equilibrium if
c~ e; for all i E N with 6; - 0
and
c G e; for all i E N with ó; - 1.
Henceforth, it is assumed that there are only two possible mutants e- and et which,
furthermore, constitute the option set for all n individuals. For the values e- and e} it
is only assumed that
e-GcGe}.
Thus the mutants e- and e} can be described as an e--type who prefers free riding if
he knows that the public good is provided even without his contribution and an et-type
who in such a situation would morally object free riding. For all individuals i E N the
preference parameter e; is therefore an element of the mutant set
E - {e-,et}.
Since individuals i and j can only differ with respect to their preference characteristics
e; and e„ group .V is completely described by the number .bI of individuals i E N with
e; - et where, of course, ~41 E N and 0 G Nf G n.
II The solution
One can easily see that there exists only one symmetry invariant - the solution does not
depend on the labelling of players and~or strategies nor on afíine utility transformations
- and strict equilibrium in all games with M G m, namely ó with N(b) - 0. For M 1 m
there exist two such solution candidates, namely 6 with N(ó) - 0 and b' with 6; - 1 for
all M et-type players i and ó; - 0 for all e--type players i. Now one has U;(b') - 1- c
6
for all et-type players i and U;(6') - l- e- for all e--type players i, wheteas U,`(6) - 0
for all players i in N. Since e- G c G 1,
U;(6') ~ U,`(6) for all i E N
which implies that in the range ;tif ~ m the strict, symmetry-invariant equilibrium 6'
payoff dominates 6 with N(b) - 0. To sum up,
Uniqueness: For e- G c G et, there exists a unique symmetry-invariant strict equilib-
rium which is not payoff dominated by another symmetry-invariant strict equilibrium,
namely b with N(6) - 0 for M G m, and 6' for M~ m.
One could justify the unique solution described above by the theory of equilibrium se-
lection developed by John C. Harsanyi and Reinhard Selten (1988). This theory gives
priority to stcict over non-strict equilibria and also to payoff dominance relationships
among equilibria. Furthermore, it is based on invariance with respect to isomorphisms
which implies symmetry invariance. The solution for the special case M- m can be
easily justified by adjusting the solution for market entry games as determined by Rein-
hard Selten and Werner Giith (1982), who apply the Harsanyi and Selten-theory of
equilibrium selection.
We prefer not to apply the Harsanyi and Selten-theory of equilibrium selection. In-
stead we rely on the concept of symmetry-invariant strict equilibria which are not payoff
dominated by other such equilibria as an ad hoc-selection theory for the class of games
which we envisage with two mutants e' and et satisfying e- G c G et. Fortunately, this
intuitive ad hoc-criterion ensures the existence of a unique solution in our basic game.
7
III The evolutionary approach
In evolutionary game theory (see, for instance, John Maynard Smith (1982), Reinhard
Selten (1988), Eric van Damme (1991)) one usually considers strategic interactions of
two genotypes which determine, respectively, the behavior of two strategically interacting
players in symmetric situations. More specifically, an evolutionary game G-(S, R) is
determined by its strategy~mutant set S and its reproductive success function R(s, s)
which assigns player 1's reproductive success to every strategy combination (s, s) with s
and s being, respectively, player 1's and player 2's strategy or genotype.
In an evolutionary game strategies are genetical programs which compete against each
other. Clearly, two players of the same species must rely on the same mutant set S, which
justifies the symmetry with respect to the strategy set S. Reproductive success measures
the expected number of offsprings. Thus, R(s,s) 1 R(s,s) means that a mutant s will
spread more rapidly in a population than a mutant s if both are confronted with a
genotype s. Clearly, player 2's reproductive success for (s,s) must be R(s,s), which
explains why it suffices to define reproductive success only for player 1.
A strategy~mutant~genotype s' E S is evolutionarily stable if
(i) R(s',s') ~ R(s,s') for all s E S,
(ii) for all s E S with R(s',s') - R(s,s'),R(s',s) ~ R(s,s).
Condition (i) requires that an evolutionarily stable strategy (ESS) is optimal in an s'-
monomorphic population. If in such a population another mutant s would also be op-
timally adjusted, then condition (ii) prevents the possibility that s spreads out in the
population. Since R(s', s) ~ R(s, s), the ESS s' earns a higher reproductive success in
every population containing s-genotypes in addition to s'-ones.
This approach has been recently applied by Guth and Yaari (1992) to encouters
of human players whose basic motivational forces are genetically determined. Our basic
model, described in Section II, assumes more than two players. To apply the evolutionary
approach and, in particular, the one proposed by Guth and Yaari (1992) in our multi-
player setting, we therefore have to adjust it to strategic encounters of more than two
players.
8
One possibility would be to assume the mutant set S- E,
E-{e-, et} with e' C c G et.
This would essentially mean that we restrict ourselves to analyzing whether the two
monomorphic populations, namely the one with e--, respectively, et-individuals only,
are evolutionarily stable. More specifically, an e-monomorphic population would be
viewed as evolutionarily stable if R(e, e) ) R(ê, e) for all é E E with é~ e. The more
general approach that we adopt will allow us to assess the evolutionary stability of such
e-monomorphic populations.
Another more general possibility is to let S- E where
~ - [0,1].
Here a mutant x E~ would denote the probability of developing as an et-individual, i.e.,
genotypes are genetical stochastic programs specifying the probability for developing as
an et- or e--phenotype. Clearly, by identifying e' E E with x - 0 and et E E withx- 1, the former mutant set E can be embedded into E.
Suppose now that x E E is confronted with i E E' as assumed in evolutionary game
theory. The difficulty of S- P is that we cannot really judge the probability of certain
numbers M, M E N and 0 C M G n, of e}-type in society without specifying howmany individuals in society are of the et- and e--type. The obvious adaptation ofthe evolutionary approach would therefore be to define x' E~ as evolutionarily stable if
R(x', x') ~ R(i, x') for all i E E with i~ x'. Here R(i, x') would measure reproductive
success if only individual 1 is of the i-type whereas all other n- 1 individuals in society
are of type x'. We will analyze evolutionary stability also from the viewpoint of S- E.
The mutant set S- S allows only for probability distributions P(A1) on
~1-{MEN:OGMcn}
which implies either P(0) - 1, or P(n) - 1 or P(tL1) ~ 0 for all M E rl~l. Thus, there
9
is no possibility for a number M with 0 C M C n to describe an evolutionarily stable
constellation of a society with n interacting individuals. In our view, this, however, is an
interesting constellation which we do not want to neglect. An M-society with 0 G M C n
can be described as one which can be partitioned into a group of 11~f individuals who are
strictly against free riding and a group consisting of the remaining individuals who would
take advantage of the opportunity to free ride. Of course, also S-~ can imply such
a partitioning. But it necessarily implies a positive probability for all M E~l. Thus,
there is no way to have with certainty a constant M with 0 G M G n when relying on
S- E. This explains why we are also interested in the mutant set S- ~N.
To define evolutionary stability of numbers M E ~:~1, we go back to the basic idea of
evolutionary stability which is the stability of dynamic evolutionary processes. The basic
justification of an ESS is that in many circumstances (but not always, see, for instance,
Weissing, 1991) it defines dynamically stable situations of evolutionary processes, e.g.,
the so-called replicator equation (see Weissing, 1991). For the situation at hand with
S- ~t the dynamics are obvious: If for a given M E M an e}-type earns a higher
reproductive success than an e'-type, i.e., if
R~,(M) ~ R~-(,1f),
then M should increase if this is still possible. Conversely,
R~t(M) ~ R~-(~Lf)
should imply a decrease of itf if this is still possible. Thus a nmutant" bf' E Nl is
evolutionarily stable if for all ~4f E.til where M G M',
R~.(M) ~ R~-(M)
and for all M E~1 with M 1 M',
R~f(.tif) ~ R~-(~f).
lo
As in Guth and Yaari (1992), we assume for simplicity that reproductive success is a
linear function of the resources earned by the player. There remains the question whether
and how the preference parameter e is related to reproductive success. Here we do not
wish to impose any restrictive assumption. We therefore allow for close relationship or
for complete independence between e and reproductive success (the latter assumption is
made in Guth and Yaari (1992)). Precisely, let s, s E[0,1], be the share of e which relates
to reproductive success. That is, when the public good is provided a non-contributing
individual of type e has a reproductive success which is equal to 1- se. In all other cases
the reproductive success is equal to the payoff U; .
Due to our equilibrium selection criterion, introduced in Section II, the reproductive
success of an e-type in general depends on M as follows:
1- c for M~ m and e- et
R~(M) - 1- se for M? m and e- e-
0 otherwise
The mutant set ~1~t together with the reproductive success function R~(M) define
the evolutionary model for which one can study the existence of evolutionarily stabfe
mutants M' E ~1.
In the following section we will analyze evolutionary stability for S-~l as well as
for S- E and thereby also for S- E. Only for the special case of n- 2, which will be
discussed in Section VI, we will also apply the original ESS-concept for S- E since for
n- 2 and m- 1 this provides new possibilities of evolutionarily stable constellations.
Of course, an even more general mutant set than M would be the set
P- {P-(P1,...,P )E [0,1)": OC~P, c 1};-i
of probability distributions P over M. Here P, is the probability for M- i, i.e., Ibf - 0
results with probability 1- Pl -...- P,,. By identifying z E E with P E ~ where
nP; - z'(1 - z)"-' for i- 1,...,n,
i
11
the mutant set E' can easily be embedded into P. Of course, also JN can be embedded
into P by identifying M E A~1 with P-(P~,...,P„)where P~y - 1. The basic flaw of
S- P is that, in general, a mutant P E P can only be justified as a population strategy,
i.e., a societal genotype and not as an individual genetical program. For this reason we
prefer to neglect the possibility of S- P.
IV The evolutionarily stable social conscience and
number of contributors
Assume S-,M and the appropriate definition of evolutionary stability given in Section
III. Then,
Theorem 1 : There exists no M' E A'1 which is evofutionari(y staóle.
Proof: For M~ m since e- G c and s E~0,1], one obtains
R~.(.tif)-1-cGR~-(.4f)-1-se-
This proves that no .bf' ~ m is evolutionarily stable. For ~f G m the result is
R~i (M) - 0 - R~- (M)
which also proves that no M' E ~1~1 with M' G m is evolutionarily stable. Q.E.D.
The non-existence of an evolutionarily stable M in the range M G m causes no
surprise. By adopting the concept of a limit evolutionarily stable strategy (LESS) de-
veloped by Reinhard Selten (i 988), we can easily ensure existence of an evolutionarily
stable M' E~1. Let e be a positive minimum choice probability for ê, - 1 and all i E.N,
12
i.e., a player i E N can only use mixed strategies by which he chooses á; - 1 at least
with probability e. Let R~(M) denote the reproductive success of an e-type in a game
with e) 0. In the range M G m the possibility of the public good being provided can
no longer be disregarded. More specifically, in this case we obtain that for e- et and
e- e- and for all M G m,
~(M) - (1-E)(1-SC) ~ ~ n - I ~ ek(1-e)n-~-k~e(I-C) ~ ~ n - 1 ~ ek(1-e)n-~-k.k-m 1` 1C k-m-I 1C
Sincen-1
R~-(M) - R~t(M) - (1 - E)s(et - e-) ~n-1 lJ ek(1 - e)"-'-k ~ O ,
for e~ 0 and s) 0, one obtains that
~-(M) ~ ~.(M)
for all M with 0 C M G n. We have thus established that,
Theorem 2: Assume a positive minimum choice probability e Jor contributing and
s~ 0, i.e., e is not totally unrelated to reproductive success. Then the e-monomorphic
society, i.e., 1Lf - 0, is the only evolutionarily stable situation.
We call a mutant M' E Nl limit-evolutionarily staóle if one can find some arbitrarily
small minimum choice probability c for one of the possible moves in the game such that
M' E ~1~t is evolutionarily stable.
k
Theorem 3: For s ~ 0, i.e., if the preference parameter e is not totally unrelated to
reproductive success, there eaists a unique limit-evolutionarily staóle society, namely the
e-monomorphic society with M- 0.
13
Proof: If there is a positive minimum choice probability for contributing, it follows from
Theorem 2 that M - 0 is limit-evolutionarily stable and that M - 0 is the only such
mutant in J~l. Obviously, a positive minimum choice probability E for non-contributing
cannot change the result for M~ m if E(1 0) is small enough. For M C m it dces not
change the result since 6- 0 is chosen anyhow with maximal ptobability. If ó; - 1 and
ó; - 0 both have 0 minimum choice probabilities, no !1f E A'l is evolutionarily stable
since R~. - R~- and m~ 1. Q.E.D.
According to Theorem 2 there can be no society N with M 1 0, i.e., with a positive
number of et-type members who would voluntarily provide the public good given that
their total number M meets the threshold requirement m in the sense of M 1 m.
Assume now S- i- [0,1]. To rely on the adaptation of the LESS-concept, we first
have to assume a more restricted mutant set
S-ï'`-[E,1]
We conclude this section by showíng that,
Theorem 4: The only evolutionarily staóle r E~ is x' - E provided that E is small but
positive.
Proof: According to the definition of evolutionary stability for S- E, described in
Section IV, one has to show that
R(E, E) - E ~ R~1 ~ n- 1 ~ Ek(1 - E)"-1-k - CI }(1-E) I n~' ~ n- 1 ~ Ek(1 - E~"-1-k - 3E-
kvml-1 k J IIILkL`cm 1C ~
n-1 n - 1 1i R(2,E) - 2 ~ ~ ~ Ek(1 - E)"-1-k - CJk-m-1 1C
"~1 (n-1t(1 - x) ~ I` Ek(1 - E)"-1-k - se-k-m k
14
for all x E ï with x~ c. By writing x - e~- ~ with 0~ 0 one obtains
R(E, E) - R(x, E) - ~ ~k ~ 1 ~ n k 1~ Ek( 1- E)~-1-k - C~
"-1 n - 1-~
~~ ~
Ek(1 - E)~-1-k - SC-k-m k
- Q ~ Em-~(1 - E)~-m - ~(C - 3E-).
Since c~ e- and s E[0,1~,
R(c, c) - R(x, e) G 0
if e(~ 0) is sufficiently small. Q.E.D.
By Theorem 4, the only evolutionarily stable mutant x E ï in the sense of the LESS-
concept developed by Selten (1988), is therefore (im~.-,ox' - Iim~-,ac - 0. This justifies
the claim we made in Section III that our evolutionary analysis for S-~1 will imply
the same conclusion as S - E and thereby S- E.
V Border cases
Except for the assumption regarding the nature of the production function of the public
good and the obvious restrictions 1) c 1 0 and e- G c G et which define our stylized
setting, the main restriction of our model is the requirement
m-1
1GmGn.
15
In this section we consider the border cases
(i) m - n,
(ii) m - 1,
(iii) n ~ oo,
(iv) n - 2.
For m- n our basic game is a unanimity game. Notice that
R~. (n) - 1- c~ R~-(n) - 0.
andR~t(M) - 0- R~-(M) for all M C n.
Introducing a positive minimum choice probability c for b; - 1 as in Theorem 2 there-
fore implies that the only limit-evolutionarily stable M' E~1 is t~1' - n.
Remark 5 : For aff m - n~ 2 there is a unique limit-evolutionarily stabfe 111' E.1~1,
namely M' - n(the et-monomorphic populationJ.
Notice that the same result can also be obtained by the direct application of the LESS-
concept to the mutant set E' which justifies z' - 1 as the only LESS-stable genotype in
~ - ~o, l~.
For m- I one obtains
R~t(:b1)-l-cCR~-(:Lf)-1-e- for M11
and
R~.(0) - 0 - R~-(0)
This proves
16
Remark 8 : For all n~ m- 1, the only evolutionarily staóle M' E~l is M' - 0(the
e--monomorphic population~.
The results of Section IV apply to all cases where 1 G m G n. Thus the validity of
Theorem 3 is retained also for the limit case where n-. oo and m remains finite.
The remaining border case is n- 2. Since the case n- 2- m is covered by Remark
5, we only need to study the game with n- 2 and m- 1. For our original approach
based on S-~l, the general conclusion for all games with n~ m- 1, Remark 6,
applies. We therefore want to demonstrate that for the special case of n- 2 a direct
application of the ESS-concept can yield a different result.
For the case n- 2 and m- 1 the reproductive success function R(e~, ez) is described
in Table 1.
e~el et e-
et 1-c 1-c
e- 1 - se' 0
Table 1.
Since e' G c and s E[0, 1],
R(et, e}) - 1- c G R(e-, et) - 1- se-,
i.e., et E E is not evolutionarily stable. Similarly,
R(e-, e-) - 0 G R(et, e-) - 1- c
implies that e- E E is also not evolutionarily stable.
Remark 7: For n- 2 and m- 1 there is no evolutionarily staóle strategy e E E-
{e-,et}.
17
Let us consider the more general mutant set ï-[0,1] for n- 2 and m- 1 where
x E (0, 1] is the probability for developing as an et-type. By identifying et with x- 1
and e- with x- 0 the former mutant set E-{e', et} can be embedded into ~-[0,1],
i.e., E C ï. Clearly, the only candidate x' E E for an evolutionarily stable x E ir is the
mixed strategy equilibrium x' of the game presented in Table 1 which is given by:
x' -1-c
1 - se- ~
e' G c and s E[0,1] ensure that 0 G x' G 1.
Obviously, every other mixed strategy x E E is an alternative best reply against
x'. Thus, we must check whether condition ( ii) of an evolutionarily stable strategy is
satisfied for all x E E with x~ x'. Nows
R(x',x) - (1 - c) t x(1 - se-),1 - se-
whereas
R(x, x) - x(1 - c) t(1 - x) . x(1 - se-).
Condition (ii) for x~ x' requires therefore that
z j
11-se)~xll-c-x(1-se-)].
Now both the left and right hand sides of the above inequality are positive.
To check whether this inequality is satisfied we can therefore investigate the maximum
of the function
f(x) - x[1 - c- x(I - se")]
From
f'(x) - 1- c- 2x(1 - se")
la
one obtains that
Since se- G 1
1-cx- -x'.
1 - se-
J"(x) - -2(1 - se-) G 0
Hence, J(x) achieves its maximum at x- x' and f(x) G J(x') for all x~ x'. Now
J(x') - 0 whicó proves
Theorem 8 There exists an evolutionarily stable strategy x' E E, namely the mixed
equilébrium strategy
x' - 1-c1 - se-
oj the symmetric game in Table 1.
In an environment with two-player strategic encounters, one may expect voluntary
cooperation in the provision of a public good which requires the contribution of at least
one player since in such an environment the positive probability x' of developing the
e}-type is evolutionarily stable. The most important example seems to be monogamy,
i.e., the long term relationship between just one female and one male. Observe, however,
that x' justifies all possible distributions of the burden of providing the public good,
i.e., there is a positive probability that it is not provided at all, that only one of the two
carries the burden, and that both provide it (although this is inefficient).
VI Concluding remarks
1n this study we have applied the stability concept of evolutionary biology to the problem
of voluntary provision of a discrete public good. The motivational structure, i.e., the
pattern of social conscience emerges endogenously in the evolutionary game where the
payoff (reproductive success) functions are closely related to the payoffs in the public
19
good provision game. Our analysis illustrates that the original notion of an E5S and its
adaptation by Guth and Yaari (1992) can also be applied in a multi-person setting. We
have carried out the analysis for a particular model of public good provision. It is our
hope that futher studies of alternative frameworks will be inspired by our analysis.
The results obtained in our model imply that, in general, the evolution of individual
social conscience enhancing the resolution of the free-riding problem is not to be expected.
Within our framework evolutionarily stable mutants either do not exist at all or generate
a monomorphic society characterized by ineffective social conscience which, in turn,
results in universal non-contributing. The well observed fact that the bulk of voluntary
cooperation occurs in small social units is also supported by the border case ofour model,
namely, the case of two players. Our analysis predicts that in bilateral encounters there
exists an evolutionarily stable pattern of social conscience that alleviates the free-rider
problem. The same conclusion is obtained for unanimity games in which the production
of the public good depends on universal cooperation.
References
Admati, AAat and Perry, hlotty, "Joint Projects Without Commitments,~
Review of Economic Studies, 1991, 58, 2ï ï-298.
Andreoni, James, "Giving With Impure Altruism; Applications to Char-
ity and Ricardian Equivalence," Journal of Political Economy, 1989, 98,
1447-1459.
Bagnoli, blark and Lipman, Barton L., "Provision of Public Goods: Fully
Implementing the Core Through Private Contributions," Review of Eco-
nomic Studies, 1989, 56, 583-602.
Dawkins, Richard, The Selfish Cene, New York: Oxford University Press,
1976.
Frank, Robert H., "If Komo Economicus Could Choose His Own Utility
20
Function, Would He Choose One with a Conscience?," American Eco-
nomic Review, 1987, 77, 593-604.
Gradstein, Mark, "Voluntary Cooperation With an Endogenous Determina-
tion of Preferences: A Note," mimeo, 1992.
Guth, Werner, "Incomplete Information About Reciprocal Incentives - An
Evolutionary Approach to Explaining Cooperative Behavior," Frank-
furt~M., 1990.
Guth, Werner and Kliemt, Hartmut, "Menschliche Kooperation basierend auf
Vorleistungen und Vertrauen - Eine evolutionstheoretische Betrachtung,"
working paper, 1992.
Guth, Werner and Ritzberger, Klaus, "Resistance Against Mass Immigration
- An Evolutionary Approach", working paper, 1992.
Guth, Werner and Yaari, Menachem, "An Evolutionary Approach to Explain
Reciprocal Behavior in a Simple Strategic Game," in: Ulricht Witt, ed.,
Ezplaining Process and Change - Approaches To Evolutionary Economics,
Ann Arbor: The University of Michigan Press, 1992, 23-34.
Guttman, Joel, Nitzan, Shmuel and Spiegel, Uriel, " Rent Seeking and 5ocial
Investment in Taste Change," Economics and Politics, 1992, 4, 31-42.
Hansson, Ingemar and Stuart, Charles, "Malthusian Selection of Prefer-
ences," American Economic Review, 1990, 80, 529-544.
Harsanyi, John C. and Selten, Reinhard, A Geneml Theory of Equilibrium
Selection ín Cames, Cambridge, MA: Harvard University Press, 1988.
Hollander, Heinz, "A Social Exchange Approach to Voluntary Cooperation,"
American Economic Review, 1990, 80, 1157-116T.
Maynard Smith, John, Evolution and the Theory of Cames, Cambridge:
Cambridge University Press, 1982.
Nitzan, Shmuel and Romano, Richard E., "Private Provison of a Discrete
Public Good With Uncertain Cost," Journal of Public Economics, 1990,
21
42, 357-370.
Palfrey, Thomas R. and Rosenthal, Howard, "Participation and the Provi-
sion of Discrete Public Goods: A Strategic Analysis," Journal oj Public
Economics, 1984, 24, 171-193.
Palfrey, Thomas R. and Rosenthal, Howard, "Private Incentives in Social
Dilemmas: The Effects of Incomplete Information and Altruism," Journal
of Public Economics, 1988, 95, 309-332.
Selten, Reinhard, "Evolutionary Stability in Extensive Two-Person Games
- Correction and Further Development," Mathematical Social Sciences,
1988, 16, 223-266.
Selten, Reinhard and Guth, Werner, "Equilibrium Point Selection in a Class
of Market Entry Games," in: M. Deistler, E. Furst, and G. Schwódianer,
eds., Cames, Economic Dynamics, and Time Series Analysis - A Sym-
posium in Memonam Oskar Morgenstern, Vienna, 1982, 101-116.
Van Damme, Eric, Stability and Perjection oj Nash Equiliórig Berlin:
Springer Verlag, 1991.
Weissing, Franz-Josef, "Evolutionary Stability and Dynamic Stability in a
Class of Evolutionary Normal Form Games,r in: R. Selten, ed., Game
Equilibrium hlodels I, ELOlution and Came Dynamics, Berlin: Springer
Verlag, 1991, 29-97.
Wilson, Edward, Socioóiology, Cambridge, MA: Cambridge University Press,
1980.
DLgcussion Paper Series, CentER, Tilburg Unlverslty, Tbe Netherlands:
(For previous papers please consult previous discussion papers.)
No. Author(s)
9136 H. Bester and E. Petrakis
9137 L. Mirman,L. Samuelson andE. Schlee
9138 C. Dang
9139 A.de Zeeuw
9140 B. Lockwood
9141 C. Fershtman andA.de Zeeuw
9142 J.D. Angrist andG.W. Imbens
9143 A.K Bera andA. Ullah
9144 B. Melenberg andA. van Scest
9145 G. Imbens andT. Lancaster
9146 Th. van de Klundertand S. Smulders
9147 J. Greenberg
9148 S. van Wijnbergen
9149 S. van Wijnbergen
9150 G. Koop andM.FJ. Steel
9151 A.P. Barten
9152 R.T. Baillie,T. Bollerslev andM.R. Redfeatn
Title
The Incentives for t.,ost Reduction in a Differentiated Industry
Strategic Infotmation Manipulation in Duopolies
The D'; Triangulation for Continuous DefotTrtationAlgorithms to Compute Solutions of Nonlinear Equations
Comment on "Nash and StackelbergSolutions in a DifferentialGame Model of Capitalism"
Border Controls and Taz Competition in a Customs Union
Capital Accumulation and Entry Deterrence: A ClarifyingNote
Sources of Identifying Information in Evaluation Models
Rao's Score Test in Econometrics
Parametric and Semi-Parametric Modelling of VacationExpenditures
EfScient Estimation and Stratified Sampling
Reconstructing Growth Theory: A Survey
On the Sensitivity of Von Neuman and Morgenstern AbstractStable Sets: The Stable and the Individual Stable BargainingSet
Trade Reform, Poliry Uncertainty and the Current Account:A Non-Expected Utility Approach
Intertemporal Speculation, Shortages and the PoliticalEconomy of Price Reform
A Decision Theoretic Analysis of the Unit Root HypothesisUsing Mixtures of Elliptical Models
Consumer Allocation Models: Choice of Functional Fotm
Bear Squeezes, Volatility Spillovers and Speculative Attacksin the Hyperinflation 1920s Foreign Ezchange
No. Author(s) Title
9153 M.FJ. Steel Bayesian Inference in Time Series
9154 A.K Bera and Information Matrut Test, Parameter Heterogeneiry andS. Lee ARCH: A Synthesis
9155 F. de Jong A UnivatiateAnalysis of EMS Exchange Rates Using a Target
9156 B. le Blanc Economies in Transition
9157 AJJ. Talman Intersection Theorems on the Unit Simplex and theSimplotope
9158 H. Bester A Model of Price Advertising and Sales
9159 A. ~zcam, G. Judge, The Risk Properties of a Pre-Test Estimator for Zellner'sA. Bera and T. Yancey Seemingly Unrelated Regression Model
9160 R.M.WJ. Beetsma Bands and Statistical Properties of EMS Exchange Rates: AMonte Carlo Investigation of Three Target Zone ModelsZone Model
9161 A.M. Lejour and Centralized and Decentralized Decision Making on SocialH.A.A. Verbon Insurance in an Integrated Market
Multilateral Institutions
9162 S. Bhattacharya Sovereign Debt, Creditor-Country Governments, and
9163 H. Bester, A. de Palma, The Missing EquiLÓria in Hotelling's Location GameW. Leininger, E.-L. vonThadden and J. Thomas
9164 J. Greenberg The Stable Value
9165 Q.H. Vuong and W. Wang Selecting Estimated Models Using Chi-Square Statistics
9166 D.O. Stahl II Evolution of Smart, Players
9167 D.O. Stahl II Strategic Advertising and Pricing with Sequential Buyer Search
9168 T.E. Nijman and F.C. Palm Recent Developments in Modeling Volatíliry in Financial Data
9169 G. Asheim Individual and Collective Time Consistenry
9170 H. Carlsson and Equihbrium Seledion in Stag Hunt GamesE. van Damme
9201 M. Verbeek and Minimum MSE Estimation ofa Regression Model with FixedTh. Nijman Efíects from a Series of Cross Sections
9202 E. Bomhoff Monetary Poliry and Inflation
9203 J. Quiggin and P. Wakker The Axiomatic Basis of Anticipated Utility; A Ciarification
No. Author(s)
9204 Th. van de Klundertand S. Smulders
9205 E. Siandra
9206 W. Hárdle
9207 M. Verbeek andTh. Nijman
9208 W. Hkrdle andA.B. Tsybakov
9209 S. Albark andP.B. Overgaard
9210 M. Cripps andJ. Thomas
9211 S. Albaek
9212 TJ.A. Storcken andP.H.M. Ruys
9213 R.M.WJ. Beetsma andF. van der Ploeg
9214 A. van Soest
9215 W. Giith andK. Ritzberger
9216 A. Simonovits
9217 J.-L. Ferreua, I. Gilboaand M. Maschler
9218 P. Borm, H. Keiding,R. Mclean, S. Oortwijnand S. Tijs
9219 J.L. Horowitz andW. H~rdle
9220 A.L. Bovenberg
9221 S. Smulders andTh. van de Klundert
Title
Strategies for Growth in a Macroeconomic Setting
Money and Specialiration in Produdion
Applied Nonparametric Models
Incomplete Panels and Selection Bias: A Survey
How Sensitive Are Average Derivatives?
Upstream Pricing and Advertising Signal DownstreamDemand
Reputation and Commitment in T~vo-Pera)n Repeated Games
Endogenous Timing in a Game with Incomplete Information
Extensions of Choice Behaviour
Ezchange Rate Bands and Optimal Monetary Aooommodationunder a Dirty Float
Discrete Choice Models of Family Labour Supply
On Durable Goods Monopolies and the (Anti-) Coase-Conjecture
Indezation of Pensions in Hungary: A Simple Cohort Model
Credible EquiL'bria in Games with Utilities Changing duringthe Play
The Compromise Value for NTU-Games
Testing a Parametric Model against a SemiparametricAlternative
Investment-Promoting Policies in Open Economies: TheImportance of Intergenerational and InternationalDistributional Effects
Monopolistic Competition, Product Variety and Growth:Chamberlin vs. Schumpeter
9222 H. Bester and E. Petrakis Price Competition and Advertising in Oligopoly
No. Author(s)
9223 A. van den Nouweland,M. Maschler and S. Tijs
9224 H. Suehiro
9225 H. Suehiro
9226 D. Friedman
9227 E. Bomhoff
9228 P. Borm, G.-J. Ottenand H. Peters
9229 H.G. Blcemen andA. Kapteyn
9230 R. Beetsma andF. van der Ploeg
9231 G. Almekinders andS. Eijffinger
9232 F. Vella and M. Verbeek
9233 P. de Bijl and S. Goyal
9234 J. Angrist and G. Imbens
9235 L. Meijdam,M. van de Venand H. Verbon
9236 H. Houba andA. de Zeeuw
Title
Monotonic Games are Spannirtg Network Games
A 'Mistaken Theories' Refinement
Robust Selection of EquiLbria
Economically Applicable Evolutionary Games
Four Emnometric Fashions and the Kalman FilterAlternative - A Simulation Study
Core Implementadon in Modified Strong and Coalition ProofNash Equilíbria
TheJoint Estimation of a Non-LinearLabour SupplyFunctionand a Wage Equation UsingSimulated Response Probabilities
Does Inequality Cause Inflation? - The Political Economy ofInflation, Taxation and Government Debt
Daily Bundesbank and Federal Resetve Interventions - Dothey Affect the Level and Unexpected Volatility of theDM~S-Rate?
Estimating the Impact of Endogenous Union Choice onWages Using Panel Data
Technological Change in Markets with Network Externalities
Average Causal Response with Variable Treatment Intensity
Strategic Decision Making and the Dynamics of GovernmentDebt
Strategic Bargaining for the Control of a Dynamic System inState-Space Form
9237 A. Cameron and P. Trivedi Tests of Independence in Parametric Models: WithApplications and IDustrations
9238 J.-S. Pischke Individual Income, Inoomplete Infonmation, and AggregateConsumption
4239 H. Blcemen A Model of Labour Supply with Job Offer Restrictions
9240 F. Drost and Th. Nijman Temporal Aggregation of GARCH Processes
9241 R Gilles, P. Ruys and Coalition Formation in Large Network EconomiesJ. Shou
9242 P. Kort The Effects of Marketable Pollution Permits on the Firm'sOptimal Investment Policies
No. Author(s)
9243 A.L. Bovenberg andF. van der Plceg
9244 W.G. Gale and J.K Scholz
9245 A. Bera and P. Ng
9246 R.T. Baillie, C.F. Chungand MA. Tieslau
9247 M.A. Tieslau, P. Schmidtand R.T. Baillie
9248 K Wïrneryd
9249 H. Huizinga
9250 H.G. Blcemen
9251 S. Eijffmger andE. Schaling
9252 A.L. Bovenberg andR.A. de Mooij
9253 A. Lusardi
9254 R. Beetsma
9301 N. Kahana andS. Nitzan
9302 W. Giith andS. Nitzan
Title
Environmental Poliry, Public Finance and the Labour Marketin a Second-Best World
IRAs and Household Saving
Robust Tests for Heteroakedasticity and AutocorrelationUsing Soore Function
The Long Memory and Variability of Inflation: A Reappraisalof the Friedman Hypothwis
A Generalized Method of Moments Estimator for Long-Memory Processes
Partisanship as Information
The Welfare Effects of Individual Retirement Accounts
Job Search Theory, Labour Supply and UnemploymentDuratíon
Central Bank Independence: Searching for the Philosophers'Stone
Environmental Tazation and Labor-Market Distortions
Perntanent Income, Current Income and Consumption:Evidence from Panel Data
Imperfect Credbility of the Band and Risk Premia in theEuropean Monetary System
Credibility and Duration of Political Contests and the Extentof Rent Dissipation
Are Moral Objections to Free Riding Evolutionarily Stable?