participation in organized and unorganized protests and rebellions

www.elsevier.com/locate/econbase

European Journal of Political Economy

Vol. 19 (2003) 861–874

Participation in organized and unorganized

protests and rebellions

Kevin Siqueira*

School of Business, Clarkson University, Potsdam, NY 13699, USA

Received 14 November 2000; received in revised form 19 April 2002; accepted 8 March 2003

Abstract

The paper investigates the individual decision of whether to join and participate in a protest or

rebellion. When the protest activity is unorganized and individuals join spontaneously without regard

to the effect on the probability of success and failure and without regard for potential benefits,

multiple equilibria can exist, which may in turn have consequences for government policies. I also

consider the case where a protest or rebellion is coordinated by a leader. The broad conclusion is that

institutional analysis is required to specify the environment within which a protest or rebellion takes

place.

D 2003 Elsevier B.V. All rights reserved.

JEL classification: D74

Keywords: Protest; Rebellion; Organization

1. Introduction

In the United States and other parts of the world, there have been high-profile

protests against international organizations. Because of the violence that has at times

accompanied the protests, governments hosting the meetings of international organiza-

tions have responded with increased planning, restrictions on movement, and, at times,

with shows of force.1 Under such conditions, peaceful protests can become disruptive

0176-2680/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/S0176-2680(03)00040-5

* Tel.: +1-315-268-6609; fax: +1-315-268-3810.

E-mail address: [email protected] (K. Siqueira).1 For example, this was evident in the Canadian government’s preparations and actions during the Free Trade

Area of the Americas meeting in Quebec City in March 2001. These preparations and actions included erecting

fences surrounding the site of the meetings and imposing tighter security at the Canadian/U.S. border prior to and

during the Summit of the Americas.

K. Siqueira / European Journal of Political Economy 19 (2003) 861–874862

and if the risk of violence exists, this should also figure in the calculus of the

participants themselves. Despite the seeming spontaneous nature of events as the

1992 Los Angeles Riot (LA Riot), media coverage enabled potential participants to

be aware of the gains and costs and the associated risks of becoming involved.2

Participants, even if they cannot predict the events themselves, are at least cognizant

of the consequences of participating and not participating. However, in order to describe

individual behavior under these circumstances, a framework must also appeal to the

literature concerning the broader context of movements that take shape in the form of

rebellion or terrorism.3

Tullock (1974) and Muller and Opp (1986) address the issue of individual

participation within a revolutionary movement. If the object of the movement is a

nonexcludable public good and individual efforts do not greatly affect the probability

of success, the problem of free riding arises. However, private benefits from

participating can provide sufficient incentives for collective action to take place (Olson,

1965). Such private benefits can take the form of private material rewards such as loot

captured or a monetary payoff obtained in the event of success. In the case of Islamic

terror, there are private benefits in Paradise through the belief that 72 virgins await the

martyr who dies in the jihad. There also may be psychological rewards in the form of

camaraderie that comes from serving a cause or belonging to a certain dissident

group.4

Kirk (1983) argues in a similar vein within the context of terrorism, but regards the

use of ideological motives and the application of psychological and sociological factors

to resolve the dilemma of participation as questionable because these are not observable.

Instead, emphasis is placed on material rewards as the motivating force. Thus, if a

government skims portions of rents it has created for certain segments of the populace

and there exist barriers to political influence, the resort to terrorism can be represented as

a struggle over the distribution of rents.5 One implication of the argument is that an

increase in government size insofar as it represents increased rent seeking, can lead to

increased terrorism. Control of rents or income can also be obtained through control of

territory or natural resources as well as through acts of extortion and kidnapping.

Although material motivation may lead the behavior to be described as ‘‘criminal’’, a

2 Extensive visual coverage of the 1992 LA Riot provided striking images of looting and violence along with

the other hazards involved throughout the areas and days of rioting. Nonetheless, in the end, the 1992 riot left

some 50 people dead and more than 2000 injured. Property damage was estimated to be near US$1 billion.3 Kuran (1989, 1995) explains the seemingly unpredictable nature of revolutions, rebellions, and events

including their spontaneous character and bandwagon effects. The main mechanism at work in his models is

preference falsification by individuals. Although interesting and compelling, the explanation of spontaneity of

unorganized rebellions is not the focus of the current paper. The focus is rather on why some protests take off

while some do not.4 In a different context, Schnytzer and Sustersic (1998) investigate motives that induce individuals to join or

leave the party in a one-party system.5 Somewhat similar in spirit to Kirk is Grossman (1999). Grossman develops a theory of revolutions based

upon rivalry over the extraction of income from producers between an incumbent kleptocrat and a potential

kleptocrat (the leader of the revolutionary movement).

K. Siqueira / European Journal of Political Economy 19 (2003) 861–874 863

group’s motives, even if ideological or political, are financially driven when there is a

need to feed, cloth, arm, and sustain members in order to remain viable over the long

run.6

Such aspects are taken into account in the model I present. Less emphasis is placed

on the public good/free-riding aspect of the collective action problem and more

attention is given to the presence of private benefits. Such an approach is consistent

with the contention of the centrality of incentives in problems of collective dissent

(Lichbach, 1995). In addition, the formulation of the decision to participate or not is

made explicit. However, a complication arises. If the movement tends to be sponta-

neous and unorganized in the sense of not controlling membership, the possibility of

multiple equilibria exists. The movement then can be either smaller or larger than

would be the case if it were more organized and membership strictly coordinated by a

guiding leader.7

In characterizing the case where the movement is unorganized and nonexclusive, I have

in mind protests against globalization at IMF, World Bank, and WTO meetings or events

such as the 1992 LA Riot. In the latter case, although civil unrest existed prior to the riot, it

was the jury’s acquittal of four police officers—charged with the Rodney King beating—

that led to the riot. One of the more salient highlights of the rioting involved the behavior

of individuals who used the occasion to loot stores. Such behavior is worthy of attention in

that it demonstrates that the activity of various participants was motivated by personal

interest.

When characterizing the case where the protest or rebellion is organized, membership is

assumed to be controlled by exclusion and the objective is to promote the collective well

being of a group by maximizing the expected utility of a representative participant. This

ensures that the per-capita distribution of rent among members remains high and preserves

the psychological benefit of uniqueness of belonging to a movement.8 I also investigate

the case of a leader who maximizes his or her share of benefits while leaving the remainder

to be distributed among the remaining members.

To formalize the issues, I use and reinterpret the model of Jimenez (1985)

describing land invasions. In that paper, the author models individual housing choice

under uncertainty and investigates the consequences of community size and control.

One interesting conclusion is that if a community of squatters is successful in

becoming viable, efforts by government to counter the community and reduce its size

7 For the purposes here, I distinguish between the two cases: the former one being more decentralized, o

nonexclusive, and possibly leaderless, and the latter, as being centralized, exclusive, and having leaders. A

organization, such as al Qeada, contains elements of both central leadership and decentralized cells that enabl

parts of the organization to act spontaneously and independently of one another.8 Although consisting of privately motivated individuals, a more tightly controlled movement that i

violently antistate or antisystem is consistent with most definitions of terrorist organizations; see Hoffman (1998

and Lichbach (1995, p. 61). On choice-theoretic models of terrorism and possible governmental responses, se

Sandler et al. (1991).

6 Although subtle distinctions may be made, e.g., between Columbia’s narco-terrorist groups and the more

traditional leftist guerilla groups, the Revolutionary Armed Forces of Columbia (FARC) and the Nationa

Liberation Army (ELN), I do not pursue these differences in the paper. On the view that rebellion is essentially

based on greed, see Collier (2000).

l

r

n

e

s

)

e


may have the opposite result. I show that in certain circumstances, this same result

holds for protest activities.

2. The basic model

Proceeding with an adaptation of Jimenez (1985), I assume that actual and potential

participants have separable utility:

U ¼ uðxÞ þ vða; bÞ

where x is the private numeraire good and a is the activity of protesting (or under

nonparticipation, some other activity). If an individual chooses to protest, then a represents

a private benefit directly obtained from the time spent engaged in the activity. b represents

the benefit enjoyed by all members of the movement and is a function of both the number

of participants, n, and the level of the excludable collective good, z, that may or may not be

provided depending on whether the movement is successful or not. The functions u(.) and

v(.) are strictly increasing and strictly concave in their arguments. In addition, let b be zero

when z is not provided (the unsuccessful state) and positive in the successful state.

Specifically, if the movement is successful, the benefits are given by the function b(z,n)

where bz>0, bzz< 0, bn < 0, bnn>0. That is, benefits increase at a decreasing rate with

respect to z, but decline as more join the movement. Although there may be some initial

benefits due to camaraderie, benefits fall if rents are fixed and are divided among a

growing membership. The same applies if belonging to a movement gives one a sense of

uniqueness or if success provides access to political influence and increased voice; all of

which can be diluted if the level of participation moves beyond some point as n becomes

large. Because rents can decline at an increasing or decreasing rate, my assumption must

be arbitrary.9 In addition, assume for all a that va(a,b)>va(a,0); that is, the marginal benefit

from protest activity is higher when the movement is successful than when it is not.

An individual can choose to join the movement or not. If the choice is not to join, time

is divided between working, t, for a wage, w, and engaging in some other activity ao. If

consumption of the private numeraire good is not time-intensive, the time constraint can be

written as T= t + ao, where T represents the total time available. Given exogenous income,

I, and combining the budget and time constraints, the individual’s problem can be

characterized as choosing x and ao in order to maximize U = u(x) + v(ao) subject to

wao + x =wT + I. The solution to this problem gives the indirect utility function,

V=V(w,T,I). Using F to represent full income, F =wT + I, the indirect utility function is

V=V(F).

If an individual joins the protest, the choice is a lottery, where the probability of success

is p. I assume that the number of protesters n and the level of government repression or

efforts to quell the protest r have the following effects on the probability of success: pn>0,

9 Nevertheless, the sign bnn>0 is consistent with available rents being fixed, divisible, and equally shared

among participants in case of a successful outcome (i.e., b= z/n). Noh (1999) shows that such a egalitarian

intragroup sharing rule can be an optimal equilibrium outcome in a conflict between two rival groups.


pnn < 0, pr < 0, prr< 0. Expected utility for a participant engaging in this type of activity

can then be written as

EU ¼ pUS þ ð1� pÞUU

or

EU ¼ p½uSðwT þ I � wa� AÞ þ vSða; bÞ� þ ð1� pÞ½uUðwT þ I � wa� A� PÞþ vUða; 0Þ� ð1Þ

where utility is now indexed according to whether the movement is either successful (S) or

unsuccessful (U). Also note that these two possible outcomes have two (respective) budget

constraints: wa + x +A=wT + I and wa + x +A +P=wT + I. The variable A under both

states represents the personal costs associated with being part of the movement. These

costs are fixed and derive from costs of being informed and socially conscious or from

organizing and maintaining ties to the movement and the cause it serves. The variable P

indicates the monetary value of a fine or penalty if the movement is unsuccessful.10

From Eq. (1), optimizing with respect to the time spent on protest activity gives:

pvSa þ ð1� pÞvUa � puSxw� ð1� pÞuUx w ¼ 0: ð2Þ

That is, the expected benefit must be set equal to the expected cost of engaging in the

activity as measured by the weighted benefits foregone in terms of consuming an

additional unit of x in each state. Utilizing the implicit function theorem and Eq. (2),

one can determine how the time spent in the movement responds to a change in p, b, A,and P. The results are such that: Ba/Bp>0, Ba/Bb>0, Ba/BA < 0, and Ba/BP < 0. That is, an

increase in the probability of success or in the collective benefits of participants increases

the time devoted toward that activity. An increase in the fixed costs of participating or in

the penalty reduces the individual’s optimizing level of protest activity.11

11 See also Tullock (1974), Mueller (1989), and Sandler and Hartley (1995). In Tullock’s model, the expected

benefits and costs of revolutionary activity are weighed against each other, and although the overthrow of the

government gives collective benefits to all, participation is assumed to yield some private benefit to rebels.

However, the pure public good aspect of the problem implies that the total expected marginal benefit—the

marginal benefit that results from an individual’s contribution to the movement’s probability of success plus the

private marginal benefit—is likely to be relatively small when compared to the expected marginal costs of

participating. Consequently, there are incentives to free ride. In the case depicted here, however, it is assumed that

individuals ignore the more public aspects of the problem (such as their effect on the probability of success) and

are motivated by the presence of selective incentives.

10 Because in the next section I will apply the model used by Jimenez (1985) to explain land invasions, it is

important to see how the expression in Eq. (1) differs from the expected utility function of a squatter household.

Although the present model takes into account participant activity in terms of time spent in the movement, both

models have an income term, an organizing costs term, and a punishment term in the event of an unsuccessful

event. Each model also has a public aspect that can be associated with crowding. In the present model, this is

represented by a benefit function in the event of a successful outcome, whereas in the squatter model, the price of

informal housing (whether a squatter is evicted or not) is positively related to the number of squatters. The

squatter household thus cares about housing services that it must prepay before finding out whether it is evicted or

not, or whether the invasion is successful or not. If evicted, the household loses its housing investment in the

informal sector and is forced to pay for housing services in the formal sector.


3. Nonexclusive membership and participation

I now portray a loosely knit protest or movement where decisions are independently

made by participants. Coordination in such loosely formed protests can be very limited

and may simply be accomplished by the broadcast of events by the news media or by word

of mouth in the street.12 Organization can, however, be more systematic while remaining

essentially leaderless and decentralized. For example, during antiglobalization protests,

formal and informal coordination among various groups have kept potential protesters

informed of upcoming events and as to the sites they would focus or converge upon during

the specific events (The Economist, 2000). Such communication took place via numerous

channels. For example, information was conveyed through links and ties among union and

trade groups, university professors, students, and other autonomous groups that maintain

websites and utilize emailing lists.13

Given the assumption of independent behavior, I assume that individuals ignore the

effect of their decision to join or not to join on both the probability of success and on the

available collective benefits if successful. This imposes both positive and negative external

effects on others in the movement in terms of probability of success and benefits although

individuals continue to join as long as the expected benefits outweigh the opportunity

costs of not joining (taken as given and measured by the utility received from the next best

alternative). That is, individuals continue to join as long as EVP>V(F), where EVP

represents the expected value of participation when a is chosen optimally. As more

people join, eventually, the probability of success changes as well as the benefits available

in case of a successful outcome. In equilibrium, EVP�V(F) = 0, or

pfuS½F�wað:Þ � A� þ vS½að:Þ; b�g þ ð1� pÞfuU½F � wað:Þ � A� P�þvU½að:Þ; 0�g� V ðFÞ ¼ 0; ð3Þ

where the arguments for the function denoting the optimal choice of a are suppressed.

Using Eq. (3), one can define the locus of different values of p and b, where the typical

participant is indifferent between joining and not joining. The membership or M-locus is

downward sloping, signifying the tradeoff between benefits and the probability of success

when expected utility is constant. It can be shown under reasonable assumptions that the

M-locus is convex.14

12 During the LA Riot, the broadcast media gave extensive coverage of the events, informing those interested

in joining (or wanting to avoid) the riot as to where looting was taking place, where police enforcement was

nonexistent, and where the most violence-prone spots lay.13 Hoffman (1998) offers further examples of terrorist groups using information technology to convey

messages. Examples include the IRA and the Michigan Militia. The latter organization is noted for its link to

Timothy McVeigh, the one held responsible for the bombing of the Alfred P. Murrah Federal Office Building in

Oklahoma City. However, these examples involve groups where membership tends to be more tightly controlled

and, therefore, more pertinent to the analysis in Section 4. Nonetheless, Hoffman also cites efforts of groups to

utilize internet technology and media opportunities to decentralize and become essentially ‘‘leaderless’’.14 The term B

2p/Bb2/M can be shown to be unequivocally positive if the term � prbaS (Ba/Bb) is small in

magnitude. Given the comparative static results and the assumption that the cross-partial term is positive, the term

is the only one of the overall expression that is negative. Although not exactly comparable, interested readers can

refer to Jimenez (1985). Otherwise, detailed notes are available from the author upon request.

Fig. 1. Equilibria-decentralized membership.


Individuals who join the movement affect the probability of success and the benefits or

rents if the movement succeeds. Although the participants do not take this into account,

the tradeoff between p and b must be factored in when defining the equilibrium. Using

both functions, p = p(n,r) and b = b(z,n), I define the L-locus as representing the

technological aspects of this tradeoff as n varies. Under the assumptions made, the locus

is downward sloping. As more individuals join the protest, the probability of success

increases, but at a cost of diminished available benefits to participants (e.g., in terms of a

declining sense of uniqueness or in terms of available loot). The curvature of the L-locus

depends on the relative magnitudes of two terms in the numerator of the expression: B2p/Bb2/L-locus.

15 If the positive term dominates, the L-locus is convex.16

To ensure an interior maximum for the case when membership is explicitly controlled, I

assume that the M-locus is more convex than the L-locus. Equilibrium can then be

described as the intersection or tangency of the two loci. Two equilibria are shown in Fig.

1. At both these points, the locus defining the preference-based tradeoff between p and b

(as derived from Eq. (3)), intersects and is equal to the L-locus defining the feasible

tradeoff as implicitly determined by n. Note that the equilibrium at A, which has a high

probability of success and a low payoff, is associated with a higher number of participants

than B, where there is a low probability of success and a high payoff. Also note that the

slope of M-locus is steeper relative to L-locus at point A, while the opposite holds at B;

therefore, point A is a stable equilibrium, while B is not.

If the size of movement is to the left of point A, at point C, participants will tend to

leave the movement. Thus, the level of membership at A represents an upper bound on the

15 See Jimenez (1985) for the derivation of this expression: B2p/Bb2/L=(1/bn)2[pnn� (pn/bn)bnn].

16 If, on the other hand, bnn< 0, the sign of the expression is negative and the locus is concave. However, I

retain the assumption that bnn>0 because this is consistent with the special case of b= z/n, mentioned in footnote

9. Later in the paper, I discuss the implications when the locus is neither globally convex nor concave.


size of the movement. However, for a given probability of success, if the number of

individuals participating is consistent with being at point D, to the right of A, potential

participants will tend to join the protest movement because they will do better than their

next best alternative. Either way, whether to the right or left of point A, these tendencies

will continue until equilibrium is restored. On the other hand, any movement away from B

results in further displacement. If the move is to the right, there is a tendency for the

protest to dissipate because individuals are able to do better in their next best alternative.

Thus, the movement size at point B represents the minimum number of participants for the

movement to be viable. Alternatively, a move away from B, toward the left, results in the

movement operating at an increasingly larger scale until settling at equilibrium point A.

We can now consider comparative statics around equilibrium point A.17 The signs of

which are given by:

Bn=Br < 0 Bn=BA < 0 Bn=BP < 0 Bn=Bz > 0

Bp=Br < 0 Bp=BA < 0 Bp=BP < 0 Bp=Bz > 0

Bb=Br > 0 Bb=BA > 0 Bb=BP > 0 Bb=Bz < 0:

We can also consider changes in the location of point B. The location of this point is

affected in the following way:

Bn=Br > 0 Bn=BA > 0 Bn=BP > 0 Bn=Bz < 0

Bp=Br > 0 Bp=BA > 0 Bp=BP > 0 Bp=Bz < 0

Bb=Br < 0 Bb=BA < 0 Bb=BP < 0 Bb=Bz > 0:

Thus, if a government increases r through a show of force, the effect is fewer protestors

if the movement is initially at point A. On the other hand, if the movement was relatively

small (at B), the policy has the opposite effect insofar as it changes the location of point B,

increasing the minimum viable number of participants. Although the movement would

enjoy a greater probability of success, its viability appears less likely. Because the

government’s action reduces the probability of success for a given level of benefit b, this

requires smaller movements to tradeoff lower benefits (which are already relatively high)

for increased probability of success (via an increase in the minimum threshold level of n).

Note that the opposite holds when the government allows increased dissent. Movements

that are already quite large will tend to become larger, while the minimum threshold size of

smaller movements becomes even smaller.18

17 Given p� p(n,r) = 0, b� b(z,n) = 0, and Eq. (3), comparative statics with respect to the variables r, A, P,

and z are performed with respect to this system of equations. Note that according to Samuelson’s (1947)

Correspondence Principle, comparative statics around stable equilibria can be informed by its stable properties

and will tend to have predictive content. Comparative statics around unstable ones, however, have to be

cautiously interpreted and will have limited predictive content (Quirk and Saposnik, 1968).18 Kuran (1989) cites the example of the Iranian Revolution where the easing of opposition to protests by the

Shah had the effect of helping the movement grow larger.


There are similar consequences when the fine or the costs of organizing increase.

Thus, if the protest were at equilibrium A, the policies reduce the equilibrium number

of participants and decrease the movement’s likelihood of success. At B, the minimum

viable amount of participants needed to sustain the movement increases. With respect

to z, if the government can lower benefits when the number of participants is large,

there will be less participation. For smaller-sized movements, the same actions

increase the equilibrium level of participation needed in order to make it a viable

protest movement.

These results underscore the possible tradeoffs faced by authorities when deciding on

the response to take. With multiple equilibria, the answer to the question is not always

unequivocal. However, despite the problem, policy may have effectively similar impacts.

A government response of seeking to establish law and order, or repression can, for

example, lower membership if the equilibrium is stable and a higher minimum threshold

level of membership for one that is unstable. In the latter case, the drawback for a

government seeking to lower the size of the protest movement is that if the movement does

becomes viable, it is larger and has a relatively higher probability of success than before.

Furthermore, because point B is unstable, a small change can send a movement currently

at B to a new equilibrium around point A.19

The same implications hold for informal leaders of the movement. Although

membership can be relatively uncontrolled by the organizers, it can be influenced by

other means. If the movement were already quite large and if the costs of organizing and

informing participants are lowered by using alternative methods or technologies (such as

the Internet), more people are encouraged to join, increasing the probability of success.

If the movement happened to be at B, the effect is a reduction in threshold size.20

Furthermore, although a fine or punishment might be explicitly set by a government,

the movement can offer a support network or legal defense fund for those who are

apprehended. Such a policy lowers the overall penalties faced by individuals. Again,

how the implementation of these policies plays out depends on where the movement

finds itself. At this point, one might ask what would be the outcome and implications if

it were possible for the movement’s leaders to directly control membership and, thus, be

able to determine the scale of the protest itself. This is the question to which I turn to

next.

4. Organized membership and participation

When participation is determined independently, individuals ignore the impacts they

have on others and on the movement. However, in the presence of a leader and

organizer, there can be incentives to control for these impacts on rents or benefits and

the probability of success in order to reap gains in efficiency. Assume, therefore, the

19 Samuelson’s Correspondence Principle may therefore still hold ‘‘in the large’’.20 Note that because B is unstable, being at point B is an event that occurs with very small probability.


presence of a leader who can directly control the size of the movement.21 In addition,

the leadership might also seek to coordinate the number of participants with that of

their actions or with regard to the time devoted to the movement. For there to be

sufficient individual motivation to join, the movement will need to promulgate rules

and conditions about membership and actions so as to ensure that an individual’s

expected utility is taken into consideration or that participants at least achieve their

reservation level of utility outside the movement. This is especially so if many types of

movements and activities vie for the attention of potential participants. If so, the

leader’s problem is to choose a and n in order to maximize the expected utility of a

representative member (see also Jimenez, 1985):

pðn; rÞ½uSðF�wa�AÞþvSða; bðz; nÞÞ�þð1�pðn; rÞÞ½uUðF�wa�A�PÞþvUða; 0Þ�:

First-order conditions for an interior solution are:

pvSa þ ð1� pÞvUa � puSxw� ð1� pÞuUx w ¼ 0 ð4Þ

pn½US � UU� þ pvSbbn ¼ 0: ð5Þ

Eq. (4) is Eq. (2) again. Eq. (5) requires that the benefit of an additional member (in

terms of an increase in the probability of success weighted by the difference in utility

between the two possible states) be set equal to the added cost (in terms of diminished

rent) of another member. Rearranging condition (5), we can see that the optimum is at

the tangency of the M- and L-locus, i.e., � pvbS/[US�UU] = pn/bn.

22 Finally, holding

all else constant, participation will be higher or lower than it would be when it is

determined independently (as in Section 3) and the movement happens to be at B or

the equilibrium is at point A, respectively.

Despite previous assumptions, the sign of the comparative statics with regard to

changes in r are ambiguous. However, if repression decreases the marginal benefit in

terms of probability of success when adding another member more than it reduces the cost

of adding the member in terms of diminished rent, then Ba/Br < 0 and Bn/Br < 0. An

increase in government efforts reduces protest activity and decreases the number of

participants.

Because changes in the fixed costs of organization (A) and in the fine (P) are similar,

these can be treated together although the comparative static results cannot be unequiv-

21 In doing so within the present model, the leader is implicitly controlling the benefits available to the

members of the movement. This is consistent with Lichbach’s argument that a leader is responsible for

administering selective incentives among members of a rebel movement (Lichbach, 1995). Later, when

investigating the case of a leader maximizing his or her share of expected rent, we incorporate the explicit sharing

of rent and loot into the analysis.22 Note, however, that if the L-locus is neither convex nor concave, it is possible that the tangency, rather

than representing a maximum, could actually represent an interior minimum. In the case where membership is

nonexclusive and determined in a decentralized fashion as per Section 3, another possible implication would be

that there might be more than two intersections between the two loci and, therefore, more than just two equilibria.


ocally signed. However, if an additional protester has a relatively insignificant effect on the

probability of success, then Ba/A < 0, Bn/BA < 0, Ba/P < 0, and Bn/BP < 0. If so, authorities

can reduce protest activity and the number of members of the movement by increasing the

fine or by implementing policies that raise the cost of organization.

Comparative statics with respect to z yield Ba/Bz>0 and Bn/Bz>0. Thus, increases in the

level of the collective good in the event of a successful outcome result in an increase in

activity and more demonstrators. However, if the government is successful in reducing the

availability of perceived benefits, the government can reduce the level of protest activity

and the size of the movement.

As in Section 3, we can reinterpret the comparative static results from the viewpoint of

a protest leader. As a case in point, by trying to increase membership size through changes

in z, leaders might emphasize and focus on the ideological elements of the movement by

convincing potential members of gains from a successful outcome. Leaders might try and

convince members that the rewards are higher than what they are (Lichbach, 1995).23

Leaders may also try lowering the costs of organizing through the use of web pages on

the Internet or by seeking outside sources of funding for their organization.

Throughout the current section, I have assumed that the leader also chooses a and n to

maximize expected utility of a representative participant. The leader does not act

strategically vis-a-vis the government or possibly even in his or her own interest. The

only difference the leader makes compared to the decentralized case is to coordinate

membership and fully internalize the impact of joining. With the same underlying structure

of the problem in the two scenarios, the size of the movement is smaller under exclusive

membership when compared to a point such as A in Fig. 1. This means that the probability

of success is lower under the auspices of a leader, but each member obtains a higher

benefit or rent in the event of success. Strength through increased numbers therefore does

not always play a predominant role as one might expect under a strong leader. Rather,

other considerations may also be relevant such as the impact on rents or influence within

the movement.24

Given the importance of self-interest, selective incentives, and its control by a leader, I

also investigate the case of a leader maximizing personal utility or the expected share of

the rent. A self-interested leader cannot afford to completely ignore movement members

and needs to share a portion of the rent or loot with the membership so as to obtain their

participation in the face of competing alternatives. Even a powerful leader of a rebel

movement will be judged by his or her ability to obtain and redistribute resources among

his or her followers (Azam, 2001). Suppose then that a leader maximizes his or her share

of expected rents subject to ensuring that a representative participant at least receives their

reservation level of utility. Although the function b(z,n) can be viewed as having a general

implicit sharing rule, I investigate the special case where rent is fixed and divided up

between the leader and the rest of the movement. Therefore, let the leader’s portion of the

23 An example is the expected rewards awaiting martyrs in Paradise as mentioned earlier.24 Within the context of terrorism, an additional consideration and impetus for the movement to be smaller

(although not modeled here) may be due to the possibility that the larger conspiracies are, the more dangerous

they become (DeNardo, 1985). Larger clandestine movements are likely to be more prone to leaks, government

detection, and infiltration.


fixed rent be sz, where the share s is such that: 0V sV 1. The portion that gets distributed to

the rest of the movement is (1� s)z. Thus, the leader maximizes his or her expected portion

of the rents (psz) and distributes to each participant an amount equal to (1� s)z/n.25

Finally, let this final term, (1� s)z/n, be equal to b and, also, b = z/n.

The leader’s problem then can be specified as choosing s and n in order to

maximize pðn; rÞszsubject to pðn; rÞfuSðF � wa� AÞ þ vS½a; ð1� sÞðz=nÞ�g

þ ½1� pðn; rÞ�½uUðF � wa� A� PÞ þ vUða; 0Þ� � V ðFÞ ¼ 0:26

Using k for the Lagrange multiplier, the first-order conditions with respect to s and n are

respectively:

pz� kpvSbz

n¼ 0 ð6Þ

pnszþ k pn½US � UU� � pvSbð1� sÞ z

n2

n o¼ 0: ð7Þ

Solving for k in Eq. (6), we obtain k = n/vbS. Using this term, Eq. (7) can be written (after

slight manipulation) as:

pnszþnpn½US � UU�

vSb¼ pð1� sÞ z

n: ð8Þ

The terms on the left-hand side represent the sum of the marginal benefits of adding an

extra protestor to the movement from the point of view of the leader and fellow

representative members. This is set equal to the marginal cost of giving that person his

or her expected share. Rearranging and rewriting Eq. (8) using the expressions b = z/n and

bn =� z/n2, we obtain the following expression:

pn

bn¼ �

pvSb½US � UU� �

½p þ npn�sbvSb½US � UU�nbn

: ð9Þ

Note that if s = 0, Eq. (9) is similar to the tangency condition for the two loci. However,

when s is positive, the right-hand side of Eq. (9) has another positive term. Holding all else

constant, this implies an intersection of the L- and M-locus and an optimum that is similar

25 Such a simple sharing rule might work well in an ethnic rebel movement where members of the group take

on a government that consists of a majority of another ethnic group. See Azam (2001) and Clapman (1998) for

arguments that leaders of such movements mainly consist of educated elites discouraged by the existing

distribution of resources.26 In order to simplify the presentation of this alternative specification, the variable a is treated as a constant.

Also note that the notation with respect to the second argument of the function vS(.) will now be slightly different

from the one in the previously specified model although the change is inconsequential.


to a point such as A in Fig. 1. Thus, a leader who maximizes his or her share of expected

rents tends to expand the size of movement in order to increase the probability of

achieving a successful outcome. Although the leader does not ignore the impacts of

increased followers, the leader controls and (distorts) membership in order to serve his or

her interest. Of course, this comes at the implicit expense of participants, in that the larger

the leader’s share of the rent, the less a representative member stands to gain from a move

from a decentralized to a more centralized movement.

5. Final remarks

Changes in institutional detail and the rules of the game will alter some of the results

derived in this paper. For example, strategic interaction between the movement and

government could take place, and a more explicit division of activities among the various

actors could be specified as in Grossman (1995). Future research can also be more specific

in the modeling of members and leaders of the movement, as well as being more detailed

about the government. We also might ask how government institutions change in response

to protest. For instance, if the protest is unsuccessful, there might be some remaining

impetus for reform.

Another avenue for future research could allow for differences in movement compo-

sition. Although this complicates the analysis, a heterogeneous membership allows for the

possibility of interaction between various types of members within a protest group and

also among the groups themselves. A model similar to that developed by Schram and van

Winden (1991), explaining why people vote, might also help in explaining why some

movements are more organized and others are not. For example, some participants can

have incentives to create social pressure in the hope of inducing others to participate. If

they are successful, more will participate. Such a model allows comparisons to be made

among situations and outcomes where some groups form more tightly knit groups than

others for inciting social pressure as considered, e.g., by Mancur Olson (1965).27 We

might also inquire as to whether society is better or worse off with such groups. Falkinger

(1999), for example, investigates and illustrates the relationship between social stability

and instability, efficiency, and a state’s income redistribution policy. An analysis along

these and other lines will add to our understanding of why individuals join organizations

and movements in the first place.

Acknowledgements

The author would like to thank Frans van Winden, Arye Hillman, three anonymous

referees, Hiro Shimizu, and the seminar participants in the Public Choice Society meetings

27 In the context of a one-party state, Schnytzer and Sustersic (1998) find that party membership is positively

related to rents distributed by the regime and, to a certain extent, to the popularity of the regime’s economic

policies. Moreover, proxies for these variables seem to be more important in explaining participation than those

that proxy government repression.


in San Antonio in 2001 for helpful comments and suggestions. Errors that might remain

are the sole responsibility of the author.

References

Azam, J.P., 2001. The redistributive state and conflicts in Africa. Journal of Peace Research 38, 429–444.

Clapman, C. (Ed.), 1998. African Guerillas. James Currey, Oxford.

Collier, P., 2000. Rebellion as a quasi-criminal activity. Journal of Conflict Resolution 44, 839–853.

DeNardo, J., 1985. Power in Numbers: The Political Strategy of Protest and Rebellion. Princeton Univ. Press,

Princeton.

Falkinger, J., 1999. Social instability and redistribution of income. European Journal of Political Economy 15,

35–51.

Grossman, H.I., 1995. Insurrections. In: Hartley, K., Sandler, T. (Eds.), Handbook of Defense Economics. North-

Holland, Amsterdam, pp. 191–212.

Grossman, H.I., 1999. Kleptocracy and revolutions. Oxford Economic Papers 51, 267–283.

Hoffman, B., 1998. Inside Terrorism. Columbia Univ. Press, New York.

Jimenez, E., 1985. Urban squatting and community organization in developing countries. Journal of Public

Economics 27, 69–92.

Kirk, R.M., 1983. Political terrorism and the size of government: a positive analysis of violent political activity.

Public Choice 40, 41–52.

Kuran, T., 1989. Sparks and prairie fires: a theory of unanticipated political revolution. Public Choice 61, 41–74.

Kuran, T., 1995. Private Truths, Public Lies: The Social Consequences of Preference Falsification. Harvard Univ.

Press, Cambridge, MA.

Lichbach, M.I., 1995. The Rebel’s Dilemma. University of Michigan Press, Ann Arbor.

Mueller, D.C., 1989. Public Choice II. Cambridge Univ. Press, Cambridge.

Muller, E.N., Opp, K.D., 1986. Rational choice and rebellious collective action. American Political Science

Review 80, 471–487.

Noh, S.J., 1999. A general equilibrium model of two group conflict with endogenous intra-group sharing rules.


Olson, M., 1965. The Logic of Collective Action. Harvard Univ. Press, Cambridge, MA.

Quirk, J., Saposnik, R., 1968. Introduction to General Equilibrium Theory and Welfare Economics. McGraw-Hill,

New York.

Samuelson, P.A., 1947. Foundations of Economic Analysis. Harvard Univ. Press, Cambridge, MA.

Sandler, T., Hartley, K., 1995. The Economics of Defense. Cambridge Univ. Press, Cambridge.

Sandler, T., Enders, W., Lapan, H., 1991. Economic analysis can help fight international terrorism. Challenge 34,

10–17.

Schnytzer, A., Sustersic, J., 1998. Why join the party in a one-party system? Popularity versus political exchange.


Schram, A., van Winden, F., 1991. Why people vote: free riding and the production and consumption of social

pressure. Journal of Economic Psychology 12, 575–620.

The Economist, 2000. Anti-capitalist protests. September 23, 357, 85–87.

Tullock, G., 1974. The Social Dilemma: The Economics of War and Revolution. Center for Study of Public

Choice, Blacksburg.

participation in organized and unorganized protests and rebellions

Documents