A GAME THEORY ANALYSIS OF TAXATION AND REVOLUTION: A PROGRESS REPORT


CONFLICT RESEARCH CONSORTIUM

Working Paper 89-17, September 1989.
By Thomas F. Mayer
Department of Sociology and Institute of Behavorial Science
University of Colorado, Boulder

This paper was written with a small grant from the Conflict Research Consortium, University of Colorado. Funding for the Consortium and its Small Grants Program was provided by the William and Flora Hewlett Foundation. The statements and ideas presented in this paper are those of the author and do not necessarily represent the views of the Conflict Research Consortium, the University of Colorado, or the William and Flora Hewlett Foundation. For more information, contact the Conflict Research Consortium, Campus Box 580, University of Colorado, Boulder, Colorado 80309. Phone: (303) 492-1635, e-mail: burgess@colorado.edu.

 

This is a first draft and contains results which may require further revision. Please do not cite or quote this paper.

Copyright (C) 1989. Thomas F. Mayer. Do not reprint without permission.

Funded by the William and Flora Hewlett Foundation and the University of Colorado, the Conflict Research Consortium is a coordinated program of research, education and application on three of the University's four campuses. The program unites researchers, educators, and practitioners from many fields for the purposes of theory-building, testing, and application in the field of conflict resolution. Current focus areas include international conflict; environmental and natural resource conflict; urban, rural, and inter-jurisdictional conflicts; and the evaluation of alternative dispute resolution mechanisms.

WORKING PAPERS

The Conflict Research Consortium working paper series includes a variety of papers written by our members as a part of their research. Usually these papers are in preliminary draft stage and are being prepared for eventual publication in professional journals or books. Other papers record discussions from Conflict Research Consortium seminars and plenary presentations.

The purpose of the working paper series is to generate a dialogue about the work presented. Readers are encouraged to respond to the papers either by contacting the author directly or by contacting the Consortium office.


1. A Theoretical Problem

Although social revolution is an extremely important form of conflict, until very recently modern conflict theory has seldom attempted a systematic explanation of revolution. This is partly because conflict theory has focused upon problems of conflict resolution, and partly because revolutions seem to be historically specific events not amenable to any sort of general explanation. Two of the most influential recent works on revolution, Jeffrey Paige's Agrarian Revolution (1975) and Theda Skocpol's States and Social Revolution (1979), both take a strict structuralist position minimizing the relevance of intentional action for explaining either the occurrence or the outcome of revolution.

Such a position discourages the application of modern conflict theory -- much of which is based upon concepts of intentional action -- to the analysis of revolution. Recently, however, theoretical thinking on social revolution has started to change. Social scientists increasingly recognize that structuralism and intentional action are not irreconcilable; on the contrary, each requires the other. A structural analysis which does not incorporate concepts of intentional action lacks motivational foundations and thus cannot give a plausible account of how individuals behave in the context of a social structure. On the other hand, a theory of intentional action which avoids structural analysis cannot be historically grounded and thus cannot differentiate between various revolutionary situations. An early step towards the integration of structure and intention in the study of revolutionary conflict was taken by Samuel Popkin who proposed a rational choice theory of peasant insurgency in Vietnam.1 This same integration project was carried forward by the Norwegian political scientist Jon Elster who criticized functionalist thinking in Marxist theories of revolution and proposed the use of game theory as a means of understanding the relationship between conflict and change.2 One of the people who pursued Elster's suggestions was Adam Przeworski, a distinguished political conflict theorist from the University of Chicago. Przeworski used a rational choice model of class conflict to explain why the working class in advanced capitalist societies is seldom revolutionary.3

However, the really decisive step in combining structural analysis and rational choice to explain social revolution is due to the University of California economist John Roemer. Whereas Elster only proposed the use of game theory, and whereas Przeworski's class actors are optimizers but not really strategizers, Roemer actually constructed a game theory model of strategic interaction between an agent and an opponent of revolutionary change.4 He derived many mathematical results from this model, showing that important aspects of revolutionary ideology -- such as the revolutionary injunction to take from the rich and give to the poor, and the tendency of status quo defenders to treat lower class insurgents more harshly than upper class rebels -- could be understood as forms of rational action given a class structure of the sort which might foment revolution. Roemer's game theory model goes a long way towards bringing revolution within the purview of a systematic social conflict theory.

Impressive though it is, Roemer's model does not solve all the important problems connected with a conflict theory analysis of revolution. One of the problems it does not solve is the so-called "free rider" or "collective action" problem first identified by Mancur Olson.5 The free-rider problem, in its revolutionary incarnation, may be stated as follows. The contribution of any single individual to the success of a revolution is almost certainly small. If so, why should a rational person participate in a dangerous and personally costly revolutionary movement when she could get all the benefits of victory without enduring the risks of participation? Roemer recognizes the free-rider problem clearly enough and briefly suggests how a revolutionary movement solves it: charismatic leadership and changing preference structures from the Prisoner's Dilemma form to the Assurance Game form.6

These off-hand suggestions notwithstanding, Roemer's game theoretic model of revolution does not overcome the free-rider problem. Yet a satisfactory account of social revolution based upon conflict theory principles cannot leave this issue in abeyance. [??] be premature and perhaps disappointing. Subsequent papers will present the model in mathematical form and will give exact mathematical proofs for all the results I have been able to derive. An important clue about how the free rider problem could be overcome appears in a recent article by a Dutch sociologist Siegwart Lindenberg who urges we consider individual collective action in addition to group collective action when trying to explain social revolution.7 Individual collective action is action carried out individually by a large number of people, not through organizational coordination of any kind, but simply because they have similar motivations, similar resources, and face similar social circumstances. Individual collective action can weaken an existing state and strengthen a revolutionary movement. But individual collective action can also yield rewards quite apart from whether a revolution succeeds or fails. These independent rewards of individual collective action are available only to people who really undertake the action in question; they cannot be obtained through free-riding. Thus individual collective action is not obstructed by free-rider type thinking. Lindenberg presents evidence that this kind of process contributed to the success of the French and Russian Revolutions. Specifically he shows that the pre-revolutionary states in both France and Russia acquired huge budgetary deficits which they tried to lay upon their own citizens, but were prevented from doing so by widespread though uncoordinated tax resistance (i.e. individual collective action).

The model proposed in this paper builds upon John Roemer's game theoretic framework and incorporates some of Lindenberg's insights about how the free rider problem might be overcome in the context of social revolution. The model deals with the economic structure of revolutionary situations and focuses upon the strategic dynamics of revolutionary action . So as to obtain the sharpest possible results about the economic dimensions of revolution, I assume people in revolutionary situations are not exclusively motivated by material concerns, and that they choose between alternative courses of action on the basis of economic considerations. The work presented in this paper is not yet completed. The paper is in fact a progress report intended for the general social scientist and any other educated person interested in the topic. I have omitted all mathematical symbolism in hopes of making the main ideas of the proposed game theory model as plain as possible to all readers. Although I have obtained some interesting results, I am still very much in the process of exploring the model, and thus a formal presentation, while certainly feasible, would [??]


2. A Game Theory Model of Revolution: Basic Concepts

Following John Roemer's approach, the basic conception of the model I propose involves two strategic actors: a revolutionary agent (whom Roemer calls "Lenin") and a defender of the status quo (whom Roemer calls "Tsar") assumed to control the state. The revolutionary agent tries to maximize the probability of revolution, while the status quo defender tries to minimize this probability. Since any gain for the revolutionary is a loss of corresponding magnitude for the defender (and vice versa), the resulting process of interaction is a two-person zero-sum game. Of course other people are involved in the revolutionary situation, but they are not conceived as strategic actors. They simply react to the situations jointly created by the initiatives of the revolutionary agent and the status quo defender. More specifically, the other people (i.e. the ordinary members of society) evaluate their economic prospects under exogenously determined circumstances and on this basis choose between a small set of pre-defined alternative actions. The model assumes ordinary people are short-term economic maximizers and will choose the combination of actions which yields the highest expected income.

Roemer's model assumes the status quo defender tries to prevent revolution by specifying economic penalties (i.e. fines) which people will suffer if they join the revolutionary coalition and if the revolution is unsuccessful. The penalties specified by the defender are individualized: each member of society faces her own particular fine should she support an abortive revolution. The status quo defender acts first by posting the list of economic penalties. She is constrained from posting too severe penalties because, other things being equal, harsher penalties make the regime seem more tyrannical thereby increasing the probability of revolution. The revolutionary agent, knowing the penalties each person faces, proposes a redistribution of national income. The total redistributed income cannot exceed the present national income which prevents the revolutionary agent from gaining adherents through pie-in-the-sky promises. The probability of revolution is assumed to be a function of the nature of the revolutionary coalition, the existing income distribution, and the set of penalties imposed for revolutionary participation. Roemer does not specify the exact form of this function, but he makes various plausible assumptions about it. One such assumption is "coalition monotonicity" which means that adding people to the revolutionary coalition increases the probability of successful revolution. Another assumption is the one mentioned above usually called "penalty monotonicity": increasing the penalties increases the probability a fixed coalition will make a revolution. Roemer also makes a number of other assumptions about the probability of revolution, but they need not concern us.

The strategic problem facing the defender of the status quo (who acts first) is to choose a set of penalties which minimize the maximum probability of revolution that can be obtained by the revolutionary agent. The strategic problem facing the revolutionary agent (who acts in full knowledge of what the defender has done) is to propose a redistribution of income which maximizes the probability of revolution given the specific penalty schedule posted by the defender. Given these assumptions Roemer is able to obtain a number of remarkable results about the strategies of revolutionaries and counter-revolutionaries. He shows, among other things, that an equilibrium pair of strategies exists, that the status quo defender never selects the maximum feasible penalties, that the penalties selected by the defender decrease monotonically with income, that the poorest members of society will always be in the revolutionary coalition, and that the redistribution of income proposed by the revolutionary agent favors the poor. The final section of this report compares some of Roemer's results with those either derived or anticipated for the model I propose.

Despite its mathematical virtuosity, however, Roemer's model never addresses the important free rider problem. In deciding whether or not to support the revolutionary coalition, people compare the income they can expect if they join the coalition with the income they now have, but not with the income they could expect if they did not join the coalition. Thus Roemer simply ignores the free rider option. It is easily seen, moreover, that unless a person has an amazingly strong individual effect on the probability of revolution or risks an extremely small penalty, she can always expect more income by not joining the revolutionary coalition.


3. Taxation and Revolution

The model I propose is specifically crafted to deal with the free rider problem neglected by Roemer.* It accepts the framework outlined above, but makes several important modifications and additions. The most far reaching of these is the addition of a taxation process to Roemer's model. Besides levying penalties for joining revolutionary coalitions, the state also collects taxes. The more taxes the state collects the stronger it will be and the smaller the probability of revolution. Citizens, however, dislike taxation because it reduces personal income, and I assume they have an opportunity to resist paying taxes. To forestall the inclination towards tax resistance, the status quo defender imposes a penalty upon such behavior. If a person resists paying taxes, she has a certain probability of getting away with it; i.e. there is a certain probability that the state will not be able to collect taxes from resisters. This probability decreases with the volume of taxes the state collects without resistance (because the state becomes stronger) and increases with the severity of the penalties imposed for tax resistance (because, as in the case of penalties for revolution, the state seems more tyrannical and less legitimate). Although collected taxes strengthen the state, the status quo defender is constrained from imposing excessively high taxation schedules since these would inspire more individuals to undertake tax resistance. The defender is likewise constrained against mounting exorbitant resistance penalties because they diminish the chances of collecting taxes from resisters. In contrast to the benefits of revolution, tax relief is not a public good. One cannot obtain tax relief by free riding on the efforts of others. Unless a person individually resists paying, the state will indeed collect her taxes (excepting the possibility of revolution in which case taxes are off but so is the existing income distribution as a whole). For this reason tax resistance movements are not at all foiled by the temptation to free ride.

Perhaps the key assumption of the entire model -- or at least of my reformulation of Roemer's original model -- is that the state has difficulty distinguishing between tax resistance and rebellion. Consequently the status quo defender (whom I treat as the persona of the state) punishes tax resisters the same as rebels when she is able to apprehend them. There is, in other words, no extra penalty for rebelling in addition to tax resisting; although the probability of being apprehended if one rebels is greater (since apprehension for rebelling occurs whenever revolution is unsuccessful, while apprehension for resistance alone occurs only if revolution is unsuccessful and the state is also able to collect taxes). This clarifies part of the process by which the free rider problem is circumvented. The calculus of individual self interest may cause a person to resist paying taxes. Not only can this weaken the state and diminish its capacity to repress revolution, it also exposes the tax resister to almost as much jeopardy as if she also joined a revolutionary coalition. Since the jeopardy is almost as great, many of the barriers to rebellion are eliminated at least for those tax resisters whom the revolution promises substantial economic gains.


4. The Two-Stage Process of Coalition Formation

The blindness of the state about the distinction between tax resistance and rebellion moves us towards a solution of the free rider problem, but does not quite complete the job. If the adherence of a tax resister to a revolutionary coalition only very slightly increases the probability of revolution, then the resister remains better off by not becoming a rebel. To cope with some of the remaining aspects of the free rider problem, I introduce a two stage conception of how a revolutionary coalition comes into being. I will call the first stage of the process atomistic coalition formation and the second stage collective coalition formation.

In terms of the model, a coalition -- symbolized (R,T) -- consists of two sets of people: a set of people R who rebel against the state, and a set of people T who resist paying taxes. Assuming that people maximize their expected income, it can be shown that everyone who rebels also (tax) resists but, as already indicated, there may be people who resist but do not rebel. A coalition is atomistic-formable if and only if, when the coalition exists, the specified choices are optimal for each person who is a member of the coalition (i.e. for each person who either rebels or resists). For example, if Sally resists but does not rebel under an atomistic-formable coalition (R,T) then her expected income would be diminished by either joining the revolution, or abandoning tax resistance, or doing both. Whether a coalition is atomistic-formable depends upon the existing income distribution, the income distribution proposed by the revolutionary agent, the reason for the occasional plural usage. I hope the context will clarify what I am talking about. Given a certain existing income distribution and certain probability functions it is conceivable that the status quo defender could select a taxation schedule and penalty list so that no coalition would be atomistic formable. 

Investigating the conditions under which atomistic coalition formation can take place is an important object of model exploration. To further clarify the meaning of atomistic coalition formation, consider the nature of the probability of revolution function mentioned above. One plausible assumption would be that this function exhibits declining marginal probability growth with respect to coalition members. That is, although every new coalition member increases the probability of revolution somewhat, (other things being equal) the first members of the coalition increase the probability more than do members added after the coalition is already large. Since the atomistic-formability of a coalition depends heavily upon the individual impact of its members on the probability of revolution, the declining marginal growth assumption suggests that small revolutionary coalitions may be atomistic-formable while large coalitions may not be. Moreover, even if two coalitions are each atomistic-formable the coalition resulting from their union may not be. This unfortunate property greatly confuses the issue of which coalition will actually come into being. It also hints the existence of a second stage in the coalition formation process. Collective coalition formation is based upon atomistic coalition formation, but it does not depend upon the individual impact of coalition members upon the probability of revolution. A coalition is collective-formable if (a) it is a union of atomistic-formable coalitions, and (b) each person in the coalition has higher expected income under it than under any of the constituent atomistic-formable coalitions of which she is a member.** Note that a collective-formable coalition may not be atomistic-formable; indeed this is the main reason we need the concept of collective-formability. The idea underlying this notion is that people who belong to at least one atomistic-formable coalition have already made a certain commitment to rebellion and/or resistance. Having once done so, they subsequently compare their welfare under various possible coalitions rather than agonizing over whether they should join a coalition in the first place.

In developing the concept of collective formability I had hoped to show that any set of atomistic-formable coalitions would be collective formable. Since I assume (a) the probability of revolution increases monotonically with respect to coalition formation, and (b) the revolutionary agent is a rational actor seeking to maximize the probability of revolution, this would imply the formation of a unique coalition: the union of all atomistic-formable coalitions. Unfortunately this is not always the case, and it is not hard to see why. A wealthy person might find it prudent to resist paying taxes if her tax burden were high, her chances of successfully evading the tax collector good, and if the probability of revolution (which would drastically lower her income) were low. A union of atomistic-formable coalitions could sharply increase the probability of revolution thus being disadvantageous for the wealthy tax resister.

Despite this disappointment, analysis of the process of collective formability does yield some interesting results. The process is structure preserving. That is, if a person is not both a rebel and a resister in any of the atomistic-formable coalitions unified into a collective-formable coalition, then that person is only a resister in the new collective-formable coalition. Similarly, if a person is both a rebel and a resister in at least one of the constituent atomistic-formable coalitions, then she is also both a rebel and a resister in the new unified coalition. Moreover, and more importantly, atomistic formable coalitions consisting entirely of people who are both rebels and resisters are always collectively formable. The main constraint to atomistic formability is the small effect any individual has on the probability of revolution: in other words the free rider problem. If the probability of revolution exhibits any significant discontinuity, however, there may well exist coalitions which are atomistic-formable.*** The collective formability process parlays these possibly fragmented coalitions into a much more formidable revolutionary coalition (or more accurately a coalition of rebels and resisters) because the free rider limitation has been overcome during the first stage of the process.

I can show that a solution to the interaction defined discontinuity with respect to the revolutionary coalition. by this two stage coalition formation process (i.e. a solution to the "game" of revolution) always exists. That is, I can show, the existence of an optimal redistribution of income to be proposed by the revolutionary agent, as well as optimal taxation and penalty schedules to be stipulated by the status quo defender. If both actors adopt their optimal strategies we can (in principle) specify the revolutionary coalition which emerges and the exact probability of revolution. I do not yet know how the optimal revolutionary and counter-revolutionary strategies can be determined accept in trivially simple cases.


5. Block Structure

The two stage process outlined above is still not a completely satisfactory solution to the free rider problem as it occurs in the context of revolution. If people act as isolated individuals there may not be any atomistic-formable coalitions. Hence the next phase of model development assumes either the existence or the creation of social structure. In the present context I shall mean by social structure blocks of people who retain their individual identity but make decisions on a collective basis. Here is how it works.

If a set of people constitutes a block, then they will decide whether to rebel and/or resist collectively: either everyone in the block rebels or resists, or no one does. However, a block of people can rebel or resist only if this is advantageous for every person in the block. If the contemplated action is disadvantageous for even a single person in the block then the block cannot undertake it. The default option, in other words, is no action at all. Block structure changes the process of atomistic formability. The comparison is no longer between an individual's expected income if she does or does not rebel and/or revolt, but rather between the individual's expected income if the block to which she belongs does or does not rebel and/or revolt. This changes things a lot because blocks, if large enough and endowed with sufficient resources, can have substantial impacts upon both the probability of revolution and the probability of successful tax resistance. According to this formulation, the temptation to free ride exists for entire blocks, but not for individuals within these blocks who are no longer conceived as autonomous decision makers even though the collective decision making process does respect their individual interests. But the temptation to free ride is much less for a block than it would be for an individual simply because the block cannot assume that its actions have little bearing on the outcome of the revolutionary movement. This greatly facilitates the atomistic coalition formation process (indeed the term "atomistic" is no longer appropriate), but does not really change the collective coalition formation process. Block structure can treated in two different ways: (a) as having a prior existence and thus as part of the framework within which the interaction agent and defender unfolds, (b) as part of revolutionary strategy and thus subject to formation through optimizing behavior by the revolutionary agent. I will briefly examine both of these alternatives. Suppose block structure has a prior existence. It might reflect any combination of family ties, neighborhood loyalties, ethnic or gender identities, political ideologies, organizational linkages, or whatever. The particular origins of block structure are extraneous to our game theory model of revolutionary strategy, but both the revolutionary agent and the status quo defender must take the structure into account.

Although I have not yet obtained any specific mathematical results regarding block structure, it seems likely that the defender would distribute taxation and penalty threats so as to disrupt the possibility of unified action by blocks, while the agent would propose an income redistribution so as to facilitate block unity. A strategically more interesting situation arises if we suppose no prior block structure, but rather endow the revolutionary agent with certain (but not unlimited) capacities to induce block structure. For example, the agent might be able to put a certain number of individuals in groups of the agent's own choice. This could be seen as a manifestation of the agent's organizational ability. Let us assume that the induction of block structure occurs before any action is taken by the status quo defender. Thus we have the following choice sequence: first the revolutionary agent induces a block structure, then the status quo defender announces taxation and penalty schedules, and finally the agent proposes a redistribution of national income. We have here a somewhat different sequential game than that originally formulated by Roemer. Treating this as a game of perfect information (a game in which all choices become known immediately), I think the existence of a solution can be demonstrated, but I have not yet investigated the existence problem in any depth.

This way of posing the question of block structure raises a number of interesting questions. What kind of block structure should the revolutionary agent induce? A large block would be better equipped to overcome the free rider problem, but could encounter difficulty achieving unified action especially since the defender will know the structure and will allocate taxes and penalties so as to undermine block unity. Conversely, a more decentralized block structure would be harder to disunify but less effective in negating the free rider temptation. Should the members of a block be similar or diverse in terms of income, and how should block membership effect the taxes and penalties proposed by the status quo defender? Answers to at least some of these questions can be obtained through systematic analysis of our game theory model, but this task remains to be carried out.


6. Some Anticipated Results

As indicated earlier, Roemer derives an impressive array of analytical results from his model of interaction between the revolutionary agent and the status quo defender. I have reviewed his arguments to see which ones could be adapted for the models proposed above. The outcomes of my review are plausibility arguments rather than rigorous proofs. Although some of these arguments may prove false upon more meticulous analysis, it seems worthwhile to include a few of these anticipated results in this progress report. The anticipated results outlined below refer to a version of our general model of revolution incorporating taxation and a two stage coalition formation process, but not including block structure. As I mentioned earlier, the very existence of a game theory solution has not been demonstrated in the case of block structure models.

The main conceptual difference between Roemer's model and the ones suggested herein revolves around the free rider problem which Roemer brushes aside, but satisfactory resolution of which I make a necessary condition for the formation of a revolutionary coalition. According to my interpretation, it is not sufficient that the expected result of adhering to a revolutionary coalition is superior to the status quo; each individual's adherence to the coalition must also make some difference in the revolutionary process. With this distinction in mind let us survey some anticipated results. If an increase in penalties always increases the probability of revolution (a condition more formally stated in terms of partial derivatives), then the status quo defender in Roemer's model will never impose the maximum possible penalties. This result also seems to hold in the models I propose. Roemer shows that if a revolutionary coalition contains all the poorest people in society (is "poor connected" in his words), then the revolutionary agent has an optimal strategy systematically redistributing income away from the rich and towards the poor. This result also seems to hold for the models discussed above. If a collective-formable coalition includes all the poorest people, then optimal revolutionary strategy involves redistribution from the rich to the poor. Roemer proves a variety of theorems showing that society usually splits into three income classes -- poor, middle, and rich -- such that the poor class is entirely within the revolutionary coalition, the rich class is entirely outside of the revolutionary coalition, and the middle class can be either in or out. These theorems do not generalize to the models I propose. Under most plausible assumptions about the probability of revolution function, the richest people will not be included in the revolutionary coalition (although they may be tax resisters); but the distinction between poor and middle is not as sharp as Roemer's theorems have it. The dependence of the state on tax revenues is such that middle class people can sometimes be members of the revolutionary coalition while at least a few poor people are excluded. One way of preserving at least the form of Roemer's results would be to say that the poor become absorbed within the middle class (class structure bifurcates rather than splitting into three parts), but this usage does not seem very enlightening. Roemer also shows that the severity of penalties decreases monotonically with income, and that all people in the poor income class suffer the maximum possible penalty severity. It is not always possible to define a consistent severity function for the models I propose, and even when such a function can be specified penalty severity need not decrease monotonically with income nor be maximal for the poorest people. The essential reason for this is once again the state's dependence upon tax revenue and the consequent incentive to define penalties in a way that inhibits tax resistance.

This is by no means a complete rundown of the results I have derived or anticipate deriving. It reviews only the results which directly parallel theorems proved by Roemer. The introduction of taxation, two stage coalition formation, and block structure creates analytical possibilities which have no analog in Roemer's work. A few of these have been touched upon above, but the remainder must await discussion in a subsequent and more formal paper. The idea of "significant discontinuity" here is that there may exist special circumstances under which an individuals adherence to the revolutionary coalition does have a substantial effect on the probability of revolution.


* The reader may become a little confused because I mention both "model" in the singular and "models" in the plural. In its full generality there is indeed a single model, but I sometimes work with simplified versions of the general model in which certain features have been eliminated. This gives rise to a variety of sub-models and is the taxation schedule, the list of penalties, the probability of revolution function, and the probability of successful tax resistance function.

** Since the population of any society is a discrete number, we know that the probability function will show some

***A person may be a member of several of the constituent atomistic-formable coalitions.

1 Samuel L. Popkin, The Rational Peasant (Berkeley: University of California Press, 1979).

2 See the following works by Jon Elster: "Marxism, Functionalism and Game Theory", Theory and Society, Vol. 11, 1982, pp. 453-482; Sour Grapes (Cambridge: Cambridge University Press, 1983); Making Sense of Marx (Cambridge: Cambridge University Press, 1985).

3 Adam Przeworski, Capitalism and Social Democracy (Cambridge: Cambridge University Press, 1985).

4 John E. Roemer, "Rationalizing Revolutionary Ideology", Econometrica, Vol. 53, January 1985, pp. 85-108. Also see Michael Taylor (ed.), Rationality and Revolution (Cambridge: Cambridge University Press, 1988).

5 Mancur Olson Jr., The Logic of Collective Action: Public Goods and the Theory of Groups (Cambridge, Mass.: Harvard University Press, 1965).

6 The Assurance Game is like Prisoners' Dilemma except that the actors prefer mutual cooperation to unilateral defection. Amartya Sen, "Isolation, Assurance and the Social Rate of Discount", Quarterly Journal of Economics, Vol. 80, 1967, 112-124.

7 Siegwart Lindenberg, "Social Production Functions, Deficits, and Social Revolutions", Rationality and Society, Vol. 1, n. 1, July 1989, 51-77.


Abstract

This progress report presents a reformulation of John Roemer's path-breaking model of rational interaction between a revolutionary agent and a defender of the status quo in which the former tries to maximize while the latter tries to minimize the probability of revolution. The purpose of the reformulation is to suggest how the free rider problem might be overcome in the context of revolution, a problem which Roemer chooses to ignore.

Three major revisions are made in Roemer's model. The first and most drastic revision involves the introduction of taxation and tax resistance. Successful tax resistance is not a collective good and thus is not subject to free riding. However, a state which is subject to revolutionary turmoil cannot distinguish between tax resisters and revolutionaries, and punishes both equally if it is able to apprehend either one. This process goes part of the way towards explaining how the free rider problem is overcome in revolutionary situations.

The second revision in Roemer's model postulates a two stage coalition formation process: atomistic coalition formation followed by collective coalition formation. The free rider problem only pertains to the earlier atomistic coalition formation process. The second part of the process assumes the free rider issue has been dealt with, and parlays the fragmented groups able to overcome the free rider problem into larger and more politically potent coalitions.

The third modification of Roemer's model complicates things even more by introducing a structure of social blocks which make decisions collectively even while respecting the interests of their individual members. Revolutionary coalitions are then constructed from these blocks. Individuals cannot separate themselves from the blocks of which they are members and thus cannot free ride on the revolution. Blocks of people can indeed free ride, but are unlikely to do so since the impact of each on the probability of revolution is hardly negligible.

This report avoids all mathematical symbolism in the interests of achieving the widest possible comprehensibility.

Copyright (C) September 1989 by Thomas F. Mayer. All rights reserved.

Single copies of this paper may be reproduced for personal use with the following conditions:

- All information concerning copyrights, authorship, acknowledgement of grant support, and publication must not be deleted from printed or electronic copies.

- Any use of this material must be fully cited and in compliance with all copyright statutes and ethical fair use principles. - The paper may be reproduced only in its entirety.

This paper may not be reposted on any other electronic bulletin board or retrieval system without formal permission from the Consortium or the author.

This paper is provided free of charge and may not be offered for sale by anyone other than the Consortium or the author(s).

Graphic images are not included in this file. For information on how to obtain graphics contact the CRC at the address below.

All correspondence related to this paper should be addressed to:

E-mail:

burgess@colorado.edu

Postal mail:

Conflict Research Consortium
Campus Box 580
University of Colorado
Boulder, CO 80309

Telephone:

(303) 492-1635