Anda di halaman 1dari 11

Provision of Collective Goods As a Function of Group Size

Author(s): John Chamberlin

Source: The American Political Science Review, Vol. 68, No. 2 (Jun., 1974), pp. 707-716
Published by: American Political Science Association
Stable URL:
Accessed: 13-07-2017 00:08 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

American Political Science Association is collaborating with JSTOR to digitize, preserve and extend
access to The American Political Science Review

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
Provision of Collective Goods
As a Function of Group Size*

University of Michigan

Introduction fied. The conclusions cited above assert that

Interest in the theory of collective goods has everything else being equal, small groups will fare
been increasing among political scientists since better than large groups in both actual and relative
the publication of Mancur Olson's The Logic of performance in providing collective goods. The
Collective Action.' Drawing on developments in analysis that follows is concerned with the first
the theory of public goods by Samuelson and of these questions-the relationship between
others,2 Olson shows how this theory can be group size and the amount of the collective good
actually provided by the group. The closest that
used to analyze the formation of voluntary associ-
ations including, among others, such important Olson comes to specifying the form of this rela-
political groups as international alliances and tionship is the following paragraph:
special interest groups. Among Olson's major In a small group in which a member gets such a large
conclusions are the following: fraction of the total benefit that he would be better
off if he paid the entire cost himself, rather than go
(1) "unless the number of individuals in a group is
without the good, there is some presumption that the
quite small, or unless there is coercion or some
collective good will be provided. In a group in which
other special device to make individuals act in
no one member got such a large benefit from the col-
their common interest, rational, self-interested
lective good that he had an interest in providing it even
individuals will not act to achieve their common or
if he had to pay all of the cost, but in which the indi-
group iiterests."3
vidual was still so important in terms of the whole
(2) "The larger a group is, the farther it will fall short
group that his contribution or lack of contribution to
of providing an optimal supply of any collective
the group objective had a noticeable effect on the costs
good, and the less likely that it will act to obtain
or benefits of others in the group, the result is inde-
even a minimal amount of such a good. In short,
terminate. By contrast, in a large group in which no
the larger the group, the less it will further its
common interests."4 single individual's contribution makes a perceptible
difference to the group as a whole, or the burden or
Two separate questions are addressed in the benefit of any single member of the group, it is certain
conclusions cited above. The first question con- that a collective good will not be provided unless there
cerns the amounts of a collective good that will is coercion or some outside inducements that will lead
the members of the large group to act in their common
actually be provided by groups whose only differ-
ence is in size. The second question is a relative
one, involving a comparison between the amount While this paragraph is not explicit about the
of the collective good actually provided and that form of the relationship in question, it seems not
amount which would be provided if the optimal- unreasonable to infer from it and the second of
ity conditions of the economic theory were satis- Olson's conclusions cited above that he considers
* The author wishes to thank Phillip Gregg and the relationship to be a decreasing one. It remains
Mancur Olson for many helpful comments on earlier open to question whether the relationship is a
drafts of this paper. Research support was provided gradually decreasing one, or a step function, with
by the Institute of Public Policy Studies, The Univer- the step (down) occurring at the point where any
sity of Michigan.
one individual's actions become imperceptible
I Mancur Olson, Jr., The Logic of Collective Action:
Public Goods and the Theory of Groups (Cambridge, others. It is shown below that the relationship
Mass.: Harvard University Press, 1965). A somewhatbetween group size and the provision of a colt
different and more general formulation of Olson's tive good is more complex than Olson asserts
analysis subsequently appeared in an article co-au-
and that, in many cases, it is the opposite of that
thored by Olson and Richard Zeckhauser, "An Eco-
nomic Theory of Alliances," Review of Economics and suggested by Olson.
Statistics, 48 (November, 1966), 266-279. The next section discusses collective goods and
2See Paul Samuelson, "The Pure Theory of Public the framework in which group activity will be
Expenditure," Review of Economics and Statistics, 36 analyzed. The paper then analyzes the relation-
(November. 1954), 387-389; "Diagrammatic Exposi-
tion of a Theory of Public Expenditure," Review of ship between group size and the provision of
Economics and Statistics, 37 (November, 1955), 350- collective benefits. The last section discusses the
356; and John Head, "Public Goods and Public Pol- results, compares them with those of Olson, and
icy," Public Finance, 17 (November 3, 1962), 197-219. comments on some additional factors which
3Olson, p. 2.
4Ibid., p. 36. 5 Ibid., p. 44.


This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
708 The American Political Science Review Vol. 68

influence a group's ability to provide its members but the nonexelusion property still holds. In this
with collective benefits. case, the benefit any single individual receives
from a unit of the good decreases as the size of
Derivation of the Model the group increases. An instance of such a good
A collective good is generally characterized is Olson's example of a price increase brought
by one or both of the following two properties:' about in an industry through output restriction.
The benefits of this price increase (increased
(1) Consumption of the good by one individual does
not subtract from others' consumption of the profits) are available to any producer (or potential
good. This property is often as non- producer) who supplies some of the demand for
rivalness of consumption.' the industry's product. The total amount of
(2) Individuals who do not share in paying for the increased profit is limited, however, and the con-
good cannot be excluded from enjoying the sumption of these profits is subject to perfect
benefits of the good. The benefits accrue auto- rivalness. In contrast, an "inclusive" good, such
matically to all individuals. This is referred to as
as a decrease in the corporate income tax, is not
the nonexclusion property.
subject to rivalness at all. An unlimited number
These two properties result in individuals having of firms can receive the benefits of the lower tax
incentives to misrepresent their preferences for without detracting from the benefits received by
collective goods and to engage in strategic be- the other firms.
havior to avoid paying for the benefits of these In arriving at the conclusions cited earlier,
goods. The ensuing problems make the provision Olson assumes that there is no cooperation among
of an optimal amount of a collective good via a the parties involved, that is, no cost sharing ar-
market or the political process extremely unlikely. rangements are permitted, and each individual
Of primary interest in this analysis are those goods must bear the entire cost of any increase in the
for which the nonexclusion property holds. The amount of the collective good which he initiates.
activities of interest groups clearly fulfill this con-The amount of the collective good provided is
dition, for if political action is undertaken by thus determined through a process in which the
others and an individual benefits from govern- individual reacts independently to the behavior
ment policy as a result of this action, the indi- of others in deciding how much of the collective
vidual cannot be excluded from these benefits. Thegood to provide by himself. The noncooperative
nonrivalness property plays an important part in nature of this process is also crucial to Olsen's
the analysis as well. Using this property, Olson "exploitation" hypothesis, for this hypothesis
classifies groups as "inclusive" or "exclusive" is based upon an examination of the prop-
according to the type of collective benefits their erties of the equilibrium of this noncooperative
members receive.8 These categories correspond to process.9 This process may be illustrated as fol-
the endpoints of the continuum measuring the
degree to which the nonrivalness of consumption 9It is important that the nature of this process be
understood. In particular, it is important to note how
property is met. A pure collective good, which
the "imperceptibility" of an individual's actions is
exhibits perfect nonrivalness, is an "inclusive" treated. The process is one of dynamic adjustment, in
good, and the group which receives the benefits which each individual reacts to the behavior of others
of the good may also be labeled "inclusive." An on the assumption that they will not perceive his
action (more correctly, will not change their behavior
individual's evaluation of an "inclusive" collective
as a result of his action), even though in reality this
good is not affected by the number of persons in is not the case. The model is equivalent to a Cournot
the group who receive the benefits. On the other duopoly model. See Kalman Cohen and Richard Cyert,
hand, an "exclusive" good exhibits the same Theory of the Firm: Resource Allocation in a Market
rivalness of consumption as does a private good' Economy (Englewood Cliffs, New Jersey: Prentice-
Hall, Inc., 1965), chapter 12, for one discussion of
See Head, "Public Goods and Public Policy," for this model. It is important to distinguish this treatment
a full discussion of the characteristics of collective of "imperceptible" actions from the cases in which the
goods. effects of an individual's actions are either impercep-
7Samuielson makes the following distinction between tible to himself because of uncertainty, lack of infor-
the polar cases of private and pure collective goods. mation, etc., or are imperceptible to others for the
For a private good, same reasons. The analysis in Olson's book and in the
article by Olson and Zeckhauser assumes that each
individual has perfect knowledge of the (past) actions
LXi = A } of all individuals and of the production function for
the good in question. For an examination of the prob-
while for a pure collective good x' = x2 . xi = lem of an actor's behavior under uncertainty, see Nor-
. . .- =?n = X, where xi is the consumption of indi- man Frolich and Joe Oppenheimer, "I Get By With a
vidual i, and X is the amount of the good produced. Little Help From My Friends," World Politics, 23
See Samuelson, "The Pure Theory of Public Expendi- (October, 1970), 104-120. For a discussion of Olson's
ture," p. 387. "exploitation" hypothesis (that the small exploit the
8 Olson, pp. 36-42. great), see Olson, p. 35.

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
1974 Provision of Collective Goods As a Function of Group Size 709

lows. Consider an individual faced with a choice vertical axis now indicates the amount of the col-
of allocating his income to purchases of two lective good provided by other individuals and
goods, one a private good (y) and the other a the horizontal axis measures the total amount of
collective good (x). His preferences for these the collective good provided. A reaction curve for
good are shown in Figure 1 by a set of indiffer- the individual can be derived by considering the
ence curves. His budget constraint is given by the horizontal difference between the 45' line and the
line AB, and he maximizes his satisfaction by locus of equilibria in Figure 2. The reaction curve
purchasing a commodity bundle on the highest is shown below in Figure 3, indicating that the
indifference curve consistent with the budget con- individual will provide an amount x0 of the good
straint. In Figure 1, his optimum allocation is at if no one else provides any, and that he will pro-
point E(with amounts x0 andy0 of the two goods), vide none if the amount provided by others
the point of tangency between indifference curve exceeds an amount x8. The form of the reaction
12 and the budget line AB. Suppose another indi- curve depends upon the individual's income
vidual provides an amount x of the collective elasticity of demand for the collective good.
good. Because of the non-exclusion property, this Goods may be placed in three classes on the basis
amount of the collective good is automatically of their income elasticities-inferior goods, nor-
available to the first individual. This has the effect
mal goods, and superior goods.10 In the case of an
inferior good, an extra unit of the collective good
of shifting the vertical axis in Figure 1 to the right
by the amount Y. The individual's budget line is provided by others induces the individual to cut
also shifted to the right by the amount x7, since he back his own provision of the good by at least
now consumes this amount of the collective good one unit. For a normal good, an extra unit pro-
without having to pay anything for it. If the private
vided by others induces a reduction in the pro-
good has a price of one, this action by the other vision by the individual of less than one unit, and
individual has the effect of increasing the first for a superior good, an extra unit provided by
individual's income by an amount AD, and he is others induces no reduction, and perhaps an
free to spend his income as he chooses so long increase, in the amount provided by the indi-
as he consumes at least an amount Y of the col- vidual. These categories of goods have cor-
lective good. The individual now finds his equi- responding reaction curves with the following
librium at E, and he pays for an amount (xl --x)
of the collective good. If all possible amounts of
type of good slope of reaction curve
the collective good provided by others were con-
inferior negative and > -1
sidered, the locus of the equilibria of the individ-
normal <-1, but finite
ual would be the line EE', known as the "income-
superior infinite or positive
consumption" curve. Note that if others provide
an amount x8 of the collective good, the individual Three representative reaction curves are shown in
will devote his entire budget to the purchase of the '? For a discussion of this classification, see Cohen
private good. The locus of equilibria is shown in a and Cyert, pp. 75-76 as well as Olson and Zeckhauser,
slightly different manner in Figure 2, where the p. 270, footnote 15.

private " \
good \\


X X? X1 BxX0 x = collective good

Figure 1. Income consumption curve

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
710 The American Political Science Review Vol. 68

ZXj =
good provided
by others

Exj ~ x? x x =ttlamuto

j= collective good
Fxigr total amount of
Figure 2. Locus of equilibria

Figure 4.11 It is to be expected that most collective will reach an equilibrium E where the reaction
goods will have normal income elasticities." curves intersect.13 The equilibrium results in A
The case in which two such individuals are in- purchasing xA of the good and B purchasing 4 of
dependently reacting to one another is shown in the good. Both individuals consume an amount of
Figure 5. The independent adjustment process
11 In Figure 4, the curves are drawn in such a way 13 Under the assumptions of the model, an individual
that it appears that a good is always inferior (or nor- will change his behavior if the point (XA, X8) corre-
mal, or superior). This is generally not the case. The sponding to the amounts of the collective good pro-
good may fall in different classes for different individ- vided by the individuals does not lie on the individ-
uals, and as the level of consumption of the good rises ual's reaction curve. Since the intersection of the two
for an individual, the good may move from one class reaction curves is the only point at which neither in-
to another. dividual will change his behavior, this is the only
12 In The Logic of Collective Action, Olson ignores possible equilibrium. If both reaction curves are con-
income effects, considering only the case in which the tinuous, an equilibrium will exist. Sufficient assumptions
slope of the reaction curve is - 1 (the case of no for a continuous reaction curve are that the individual
income effects). Olson and Zeckhauser consider in- have a continuous utility function and that the indif-
come effects more fully, particularly on p. 270, foot- ference curves derived from this function obey the
note 15. If collective good is a superior good in the condition of strong convexity (i.e., if two possible
framework under consideration, groups will be led allocations are indifferent, then their weighted average
either to invest all of their resources in the provision with positive weights is preferred to them). The par-
of the good (if the slope of the reaction curve is ticular configuration shown in Figure 5 is not guar-
positive) or to provide an amount of the good in antecd, but is the most likely case. Other possibilities
direct proportion to group size (if the slope of the are shown and discussed in the Appendix. It should
reaction curve is infinite). Since such behavior is be noted that this equilibrium also results if the prob-
highly unlikely, superior goods will be ignored in the lem is viewed as a noncooperative game rather than
analysis to follow. as a Cournot duopoly problem.

EjX. XsX

xi Xi = collective good provided by

individual i

Figure 3. Reaction curve

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
1974 Provision of Collective Goods As a Function of Group Size 711

X xI normal good ZXJ slope = (n-1)


s \ |
xI L

inferior \

X? Xi

Figure 4. Types of reaction curves

E 0 x
the collective good equal to the sum of these two
purchases. Figure 6. Equilibrium for an inclusive good

Attention may now be turned to examining the of size n, then the amount provided by others
relationship between group size and the amount of must be (n- 1)xE. Thus the equilibrium for a
a collective good provided by the uncoordinated group of size n is found where a straight line
activities of the group members. For any group, through the origin with slope (n-1) intersects
this amount is found by determining the equili- the reaction curve. As the size of the group in-
brium of the process described above. One further creases, the equilibrium moves along the reaction
assumption will be made at this point. In order to curve from x? (when n=1) to xi (as n-*oo), as
facilitate the analysis, it will be assumed that all shown in Figure 7. Two characteristics of the
individuals in a group are identical in terms of provision of an "inclusive" collective good are
preferences and resources.'4 evident at this point:
Consider first the case of an inclusive collective
good. The reaction curve for the ith individual (a) x4->O as n-*>oo
is shown in Figure 6. Since the good is "inclusive," (b) nxE (the total amount provided)-4x as
the reaction curve is unaffected by group size and
thus this one curve will serve as a basis to analyze
groups of different sizes. The assumption that all The relationship between nxE and n may now be
individuals are identical makes solution for the established. The total amount of the good pro-
equilibrium quite easy, for if the individuals are vided is equal to the sum of the horizontal and
identical, each will provide the same amount of vertical components of the equilibrium (nx =xE
the collective good at the equilibrium. In Figure +(n-1)xj'). If the reaction curve is stated as a
6, if xE is the amount of the good provided by function of xi, sayf(xi), then nxY = xi'+f(xY). The
the ith individual at the equilibrium for a group relationship between nxE and n is established by
14 This assumption greatly simplifies the analysis to examining the derivative of (x- +f(x!)) with re-
follow without affecting the substance of the results. spect to xE,

XB =
collective xAi
provided RCA = A's reaction curve
byB A

\ RCB = B's reaction curve

XA X. xB XA, = collective good

provided by A

Figure 5. Equilibrium with two individuals

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
712 The American Political Science Review Vol. 68

slope = (n-1 ) EX)

j& i XS 4 !\ / slope = 2

slope I RC

RC4 B e\ \ RC,
slope = 0 Xi

Xi Xi Figure 9. Locus of equilibria for an exclusive good

Figure 7. Locus
an individual changes as group of equi
size changes. A
possible set of reaction curves for an individual
are shown in Figure 8, where RC,, is the individ-
--(xi +f(X )) = ual's reaction curve for a group of size n. This set
of curves indicates that as group size becomes
where f'(x1f) is the slope of the reaction curve at infinite, the reaction curves converge to the origin.
Xi=XE. If f'(xE) -1, nx` increases as xE in- This is an extreme case of an "exclusive" good,
creases, meaning that nx-j decreases as 11 increases. and it is possible that the reaction curves, al-
If f'(xE) -1, then nxh is unaffected by group though shifting toward the origin, will not con-
size, and if f'(xE) <-1, nx4 increases as n in- verge to the origin. This case will be considered
creases. This result indicates that if the collective later. The equilibrium for a group of size n is
good is not an inferior good (i.e., the slope of the found in this case where a straight line through
reaction curve is less than -1), then the amount the origin with slope (n-1) intersects the reaction
of the good actually provided by a group in- curve for that size group. The path of these
creases with group size (approaching xs as n equilibria is shown by the broken line in Figure 9.
becomes infinite). Note that while nxE is increas-From the figure it is clear that, for an "exclusive"
ing with n, xE is decreasing. The decrease in the collective good:
contribution of each individual (x4) is more than
(a) xF--O as n- oo
offset by the increase in group size. Thus, in the
(b) nxf-*O as n-* oo
case of "inclusive" collective goods which are
(c) nxF decreases as n-* oo.
not inferior goods, the relationship between group
size and l he amount of the good actually providedThese conclusions correspond to those of Olson.
is the opposite of that asserted by Olson. For a good in the interior of the "inclusive-
The case of an "exclusive" collective good may exclusive" continuum, the reaction curves bear
be analyzed in the same way, with one exception.some similarity to those for an "exclusive" col-
If the good is "exclusive," the reaction curve forlective good. For these "intermediate" goods the



RC x
RC00 Xi

Figure 8. Reaction curves for an exclusive good Figure 10. Reaction curves for an intermediate good

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
1974 Provision of Collective Goods As a Function of Group Size 713

74Xi iii

N1 'l'

Figure 11. Loci of equilibria for two intermediate good

Xi Xi

Figure 11. Loci of equilibria for two intermediate goods

reaction curve shifts toward the origin as group size amount of a collective good is less likely in large
increases, but the set of curves converges to some groups than in small ones when behavior is rational
curve above the origin, an indication of the fact and self-interested.16
that, even as the group becomes infinitely large, Nothing in the above analysis would seem to
the nonrivalness property still holds to a certain support Olson's statement that the result ob-
degree. Such a set of curves is shown in Figure 10. tained for "inclusive" goods applies only to a
The path of equilibria is found in the same man- "special case." Only three major assumptions
ner as that of an "exclusive" good. Two possible were necessary for the analsysis, and none of these
cases arise, however, as shown in Figure 11. In is, in theory, less valid for large groups than small
both cases: groups. The behavioral assumption that an indi-
vidual will act as if his actions will not affect the
(a) x' --( as n- oo
(b) nx-*x->0 as n--4oo. actions of others is, if anything, more likely to be
valid for large groups than for small groups. The
But nx' may increase or decrease as n increases, assumptions implicit in the use of a budget line-
as indicated by Figures 1la and 1lb respectively. i.e., that there exist no fixed costs in the provision
In the first case, nxE increases with group size, of the collective good and that the price faced by
indicating that the income effect outweighs the an individual is unaffected by the actions of
crowding effect, while in the second case the op- others-affect all groups similarly, since they
posite is true. Based on this finding, it would seem present the same barrier to action to all individ-
that any collective good can be placed into one uals.17 Finally, the assumption that each indi-
of two broad categories, "inclusive" and "exclu- vidual has perfect information about the actions
sive," depending upon whether the amount of the of others and the effects of his own actions applies
good actually provided (nxE) increases or de- equally, in theory, to all groups regardless of size.
creases as group size increases. In reality, these information conditions no doubt
represent a greater obstacle to large groups than
to small groups, but differences in group per-
Olson recognizes the possibility of members of formance reflecting this fact cannot be derived
a large group providing some amount of a col- from a model based on certainty, such as-Olson's
lective good, but the tone of the following remarks or the one provided in the previous section.
indicate that he considers this possibility of little Olson's lack of clarity regarding the relationship
general interest or of little empirical significance
or both: 16 Mancur Olson, "Increasing the Incentives for In-
ternational Cooperation," International Organization,
There is one logically conceivable, but surely empiri- 25 (Autumn, 1971), 866-874, at p. 866, footnote 1.
cally trivial, case in which a large group could be 17Olson, (The Logic of Collective Action [p. 22])
provided with a very small amount of a collective points out the importance of fixed costs as a barrier
good without coercion or outside incentives."5 to collective action. He suggests that cooperative ac-
tion may be necessary to overcome this barrier. In
Moreover, apart from a special case. . . , it can be making use of a budget line, the present analysis ig-
demonstrated that even the provision of a minimal nores fixed costs. It should be emphasized that the
inability of a group to overcome these costs via non-
"6Olson, p. 48, footnote 68. cooperative action will not be a function of group size.

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
714 The American Political Science Review Vol. 68

XI) X33

Case 1 E Case 2 Case 3


E \E


I ~~E E



Case 5 E Case 6



I- a I ~~E
x,&, XA XA.

Xn XnI Xii

Case 7 E Case 8 Case 9

\RCA, RCir


Figure 12. Possible equilibrium configurations

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
1974 Provision of Collective Goods As a Function of Group Size 715

under consideration may stem partly from his use political process may be a sphere in which the
of the concept of "the fraction of the group gain" amount actually provided is the appropriate
(benefit) received by the individual.18 For an criterion, particularly if several groups are in-
"exclusive" good, the fraction of the group bene- volved and are in opposition to one another. The
fit received by an individual plays an important outcome in such a case is likely to be a function
part of the individual's decision, and its considera- of the relative actual strengths of the groups'
tion leads to the inverse relationship between efforts. For one interested in understanding and
group size and amount of the good provided that predicting the outcomes of such conflicts, the
was found for an "exclusive" good in the previous actual amount of the collective good provided
section. For an "inclusive" good, however, the (in this case, political action, defense, etc.) will be
fraction of the total benefit received by a single more useful than measures of suboptimal per-
individual is not so important, since there exists formance. This is particularly true since in such a
no rivalness of consumption. The addition of an case the optimal amounts are themselves relative,
individual to a group does not decrease the bene- being a function of the behavior of the opposing
fit any individual receives from a unit of the col- group. Most of the work making use of the theory
lective good, although it does decrease the frac- of collective goods has thus far dealt with the
tion of the total benefit the individual receives. actions of a single group. In light of this emphasis,
This decrease in the fraction of the total benefit an important next step in applying this theory to
results in the enlarged group's providing a more the study of political phenomena would be the
suboptimal amount of the collective good, but consideration of the behavior of several groups in
not necessarily a lesser absolute amount, as was competition. Such a consideration may yield a
shown. Olson's failure to differentiate in his for- better understanding of the relative powers of
mal analysis between "inclusive" and "exclusive" competing groups and of the appropriate criteria
collective goods, and the important difference by which group action should be judged.
this distinction makes in the use of the "fraction
of total benefit" concept may partially explain
his conclusions regarding the possibility of action
The configuration of reaction curves shown in
by large groups.
Another reason for Olson's conclusions may Figure 5 is only one of nine possible configura-
be his emphasis on the degree of suboptimality of tions. Figure 12 shows these configurations, with
group performance rather than on the amountCase of 1 being the one considered in the paper. The
the collective good actually provided. It is with equilibria are denoted by the letter E. While Case
respect to the former criterion that large groups 1 may be considered the most likely case, the
fare so poorly. Since in Olson's analysis smaller other cases deserve some comment. The analysis
to follow will consider only linear reaction curves
groups appear to fare better under both criteria,
in order to simplify the discussion.
the choice of which criterion to use does not
Two important characteristics of a configura-
appear to be particularly important. In light of
tion of reaction curves are the uniqueness and the
the analysis presented above, however, this choice
takes on greater importance. With respect to stability of the equilibrium. An equilibrium is
the second criterion, all groups are constrained stable if a change from the equilibrium sets in
by a rather low upper bound on the amount of themotion a pattern of behavior that results in a
good actually provided (x-). The fact that this is return to the equilibrium. Inspection of Figure 12
the amount at which an individual will provide will show that except in Cases 2 and 9 the equi-
none of the good himself makes the upper bound librium is unique and stable. In Cases 2 and 9,
multiple equilibria exist and all the equilibria are
seem particularly low. But there seems to be no
reason to suggest a priori, as Olson seems to, that unstable. If either of these cases were to occur
the actual amount provided by a large group will with any regularity, the usefulness of the theory
have little empirical importance, while the amount would be greatly impaired because of the in-
provided by a small group will be empirically ability to make precise predictions about behavior
important. Such a conclusion must be the result and because of the inherent instability of be-
of a fuller consideration of the appropriateness ofhavior in these cases.
the two measures of group performance."' The The nine configurations can arise from many
different situations, each of which is determined
"'Ibid., p. 23. by the income effects exhibited by the preferences
19 It may well be true, although no empirical evi- of the individuals and by the relationship be-
dence is available, that there are differences between
the types of collective goods desired by small and large
groups, and that these differences are such that an good may be "inadequate" for its group, a large one.
amount x: of one good may be an "adequate" amount Such a possibility seems of sufficient interest to merit
for the small group but an amount x: of another further thought.

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to
716 The American Political Science Review Vol. 68

Table 1. Possible Configurations for Various Combinations of Income Effects

Individual A

Normal Good No Income Effects Inferior Good

Normal Cases Cases Cases

n Good 1, 3,6,7,8 1,3,6,7,8 1-9

s No Income Cases Cases Cases

X Effects 1,3,6,7,8 7, 8,9 2,4, 5, 7,8

Inferior Cases Cases Cases

Good 1-9 2,4, 5, 7, 8 2,4, 5, 7, 8

tween the reaction curves (x'>, =, <x4 and of the individuals considers the collective good
>, =, <xs). These combinations give rise to to be an inferior good. Since it can reasonably be
51 different situations. Table 1 summarizes brieflyexpected that most collective goods will be nor-
the configurations which can arise for different mal goods, the uniqueness and stability of the
combinations of income effects. It will be noted equilibrium derived from the theory seem to be
that Cases 2 and 9 can occur only if at least one assured.

This content downloaded from on Thu, 13 Jul 2017 00:08:48 UTC
All use subject to