[Elster, Jon (1987), Comments on van der Veen and Van Parijs, Theory and Society 15:709-722]

[ Books | Articles | Reviews | Index | The main menu ]

Comment on van der Veen and Van Parijs

Jon Elster

[start of page 709]

I am a staunch disbeliever in the proposal put forward by Robert van der Veen and Philippe Van Parijs (hereafter VP). For one thing, it rests on a massive overestimation of our ability to predict and steer the course of economic development; for another, it completely lacks the potential for being rooted in a social movement and wedded to a conception of justice. I shall first sketch a general theory of the conditions that are required to carry out large-scale reforms in modern (complex, industrial, democratic) societies. I then apply the theory to the VP proposal, and find it wanting.

Arguments for reform are of basically two kinds.1 Some are consequentialist: the reform is desired on the grounds that it is believed to have good or desirable effects. Others are deontological: the reform is thought to be inherently valuable. An example of the first kind of reform is the introduction of the workhouse to replace poor relief; an example of the second could be universal suffrage. The nature of deontological reasoning is notoriously elusive. let me simply state, without much argument, that in my opinion the central deontological concept is that of autonomy, not the overly abstract notion of rights.2 I hope that the distinction between the two kinds of arguments will become clearer as I proceed.

I shall first state the negative side of my review, arguing against the very attempt to justify large-scale reforms and the VP proposal is certainly on a very large scale by consequentialist reasoning. On the one hand, I believe that the state of the social sciences is light-years away from allowing us to predict the global net long-term equilibrium effects of major institutional changes. On the other hand, I submit that piecemeal social engineering, through incremental planning or trial-and-error, is of little help as a remedy to this theoretical deficit, because this method only allows us to estimate local, partial, short-term, or transitional effects.

[end of page 709, start of 710]

Local versus global consequences. It is very difficult indeed to assess before the fact what the consequences will be of a reform that is both large and general. The direct and indirect ramifications will be endless and feed upon each other in a way that very soon becomes too complex to grasp. It may be possible to assess the consequences of the general implementation of a small reform, or of the partial implementation of a large one, but there is in general no reason to believe that the consequences will be the same or similar for a general large-scale reform. It is probably true that if all people became good it would be good for everybody, but "It may just be that a Sodom with ten righteous men is not better than a wicked Sodom where the war of all against all prevails."3 Similarly it may be better if all people are selfish than if all become moderately altruist, yet better still if they become strongly altruist.4

To illustrate the first point - the difference between implementing large changes in the small and in the large - we may consider two proposals for changing the ownership and incentive structure of the firm. On the one hand, there is the market socialist proposal that firms be owned and managed by workers. It is at least arguable that the very mixed success of small-scale market socialism, i.e., individual workers' cooperatives in a predominantly capitalist economy, tells us nothing about the prospects for large-scale market socialism in which all or most firms are run by workers.5 On the other hand, there is Martin Weitzman's "profit sharing" proposal, according to which employers and workers would bargain over the percentage of the net product rather than over the wage rate.6 Weitzman's proposal is motivated by various macroeconomic properties of profit-sharing that could never act as an incentive for the individual firm; hence the paucity of isolated experiments in profit-sharing does not tell against the viability of a profit-sharing economy.

Weitzman's proposal may also be used to illustrate the second point, viz., the difference between implementing small changes on a large scale and large changes on a large scale. Between a pure wage economy and a pure share economy there are many intermediate forms, each of which would have a base wage and a profit-sharing percentage. Assume that one attempted to implement the profit-sharing proposal by making it mandatory for all firms to have a small bonus - i.e., a share of the profit -on top of a fixed wage, and that the system worked better than the pure wage economy. One would not be justified in concluding that it would work even better with a larger bonus and a lower base wage. Conversely, if the small-bonus system performed worse than the pure wage system,

[end of page 710, start of page 711]

one could not exclude the possibility that a system with a larger part of the workers' income paid as a share of the profit might outperform the latter.

Partial effects versus net effects. The ceteris-paribus methodology of the social sciences tends to diminish their usefulness as a tool for planning and reform. Although within a given model, holding various things constant, one may predict that a proposed reform will have certain effects, it could well happen that the things held constant will in reality be affected by the reform. In particular, this is likely to happen in the case of major institutional and constitutional changes. For one thing, even with given motivations the new institutions may modify behavior in unexpected ways; for another, they may affect the motivations themselves.

The first problem is that of strategic adaptation to reform by those affected by it. Moral hazard is the most important problem of this kind with applications to social as well as private insurance. A reform that creates a security net under the competitive market will also lead to more people needing the' net, by reducing the incentive to survive without it. A deeper and more difficult problem is that of the endogeneity of preferences, illustrated by the following example.7 A reform movement to increase the scope for self-realization in industry, in the work process itself, or through industrial democracy, might fail because there is no widespread desire for self-realization, as opposed to a life oriented toward consumption, broadly understood. Yet one might speculate that the lack of desire for self-realization is mainly due to the lack of opportunities for self-realization in capitalist societies. By the mechanism of adaptive preferences,8 one often dismisses as undesirable what is unattainable anyway. If more opportunities for self-realization become available, they may liberate the latent desire for this mode of life. Hence one could argue that an institutional reform in this direction would create its own need, so that the full set of consequences would go much beyond that of satisfying the desires of those who already want it.

While plausible, this argument is not compelling. Adaptive preference formation is not the only mechanism that shapes our wants and desires. Even if more opportunities for self-realization were created, the desire might lag behind. Or - to take the converse possibility - the desire for self-realization might outrun the opportunities, with a net loss in welfare. All of these effects are possible, none of them is certain. After the fact it may be possible to tell which of them is operating and in what combination, but the social sciences are in no state to predict ex ante what the outcome will be.9

[end of page 711, start of page 712]

Short-term versus long-term consequences. The effects of an intitutional framework can be assessed within different time perspectives. In particular, one must distinguish between the static and the dynamic efficiency of different economic systems. Static efficiency is defined as the optimal use of existing resources, dynamic efficiency as the optimal creation of new resources. The latter includes an optimal investment policy and, even more fundamentally, an optimal rate of technical innovation. A system that performs well in the first respect may fare badly in the other, and vice versa. Again we shall have to look at the net effect. A very dynamic system may be so wasteful that the gains from innovation leak away; conversely, a very careful husbandry of resources may paralyze the growth of system.

We may compare, from this perspective, the three forms of market production: wage capitalism, profit-sharing capitalism, and market socialism. By far the most important aspect of static efficiency in this context is the ability to maintain full employment.10 More precisely, the system has to be judged (a) by its ability to absorb new workers and (b) by its ability to retain old workers in the face of a recession. Real-world capitalism, i.e., wage capitalism, performs badly on both counts, whereas on theoretical grounds one could expect profit-sharing capitalism to do well on both, because the system is set up in such a way that employers always have an incentive to hire more workers. Market socialism would, again on theoretical grounds, be expected to do well on the second count, because workers would not have an incentive to fire themselves, but badly on the first, since they would not want to share the revenue with more workers. On the other hand, we know that wage capitalism has performed well as regards dynamic efficiency. It has been argued on theoretical grounds that self-managed firms will have less incentive to invest and innovate, unless the individual worker plans to remain in the firm for a very long time. The prospects for a profit-sharing economy are more unclear in this respect.

It is perhaps possible to have some confidence in purely theoretical or ex ante arguments about static efficiency, although I have doubts. We may note that Schumpeter argued in 1942 that "whatever the economic goals desired by whoever is in the position to give effect to his desires, socialist management could attain them with less disturbance and loss without necessarily incurring the disadvantages that would attend attempts at planning progress within the framework of capitalist institutions."11It is hard to believe that he would write these words today. (Remember that the form of socialism he had in mind was centrally

[end of page 712, start of page 713]

planned state socialism.) In any case, arguments about dynamic efficiency must remain highly speculative. The determinants of innovative activity, in particular, are as elusive and unpredictable as the "animal spirits" in which Keynes found the explanation of investment behavior. The problem of predicting innovative activity involves that of predicting how firms in (what is now) the future will predict (what will then be) the future, when they face uncertainty both with respect to what the activity will yield and with respect to what other firms are doing. There are very strong arguments against the possibility of anticipating the outcome of what will essentially be a non-rational decision.12

Transitional effects versus steady-state effects. When assessing a reform on consequentialist grounds, one must take care not to perform the assessment before the dust raised by the change has settled, and it is possible to see what the new equilibrium looks like. It would be pointless, for instance, to evaluate the success of a revolution by considering the state of the economy immediately after it has taken place. It is not simply that capital that has been destroyed must be replaced, and that opponents of the change must come to accept it. Expectations and motivations must adjust to the new opportunities and constraints, and plans made on that new basis must have time to come to fruition. We must be able to assess dynamic as well as static efficiency.

At this stage in the argument there is no need to insist on the massive difficulties of predicting ahead of time what the equilibrium outcome will be. let me, instead, mention four special problems that arise. First, any serious reform proposal must take account of the time needed to reach equilibrium, not just of the properties it is expected to have when we get there. The goal of reform ought not to be the best possible society, but the best possible development of society from now on. Hence if per impossibile we discover a reliable theory affirming the existence, accessibility, and optimality of a general social equilibrium, we still ought not to embrace it unless we know what will happen during the traverse. The legacy of Marx includes not only the intellectual hubris of believing with total certainty that the advent of communism is inevitable, but also the moral error of thinking it justifies any sacrifice imposed on the intermediate generations.

Secondly, one must consider the possibility that the reform may set in motion a process that has no equilibrium, or at most a cyclical one ("limit cycles"). If the tendency to counteradaptive preference formation ("the grass is always greener on the other side of the fence") is widespread, this

[end of page 713, start of page 714]

is what we may expect to observe. The ugly aspects of capitalism may lead workers in capitalism to prefer socialism, and vice versa. The goal of social planning, to match desires with opportunities, may prove to be a will-o'-the-wisp. The lack of an equilibrium may also be due to problems inherent in the mechanism for aggregating preferences. Even assuming that individual preferences remain the same, cyclical majorities may undermine the existence of equilibrium.

Thirdly, it may be the case that the process has an equilibrium that would be reached within a reasonably short time in a constant environment, which changes so rapidly and unpredictably that the process of approaching the equilibrium is constantly thrown off course and has to start all over again. Social planning is somewhat like trying to hit an unpredictably moving target. When the solution is finally produced, the problem may no longer exist.

Lastly, there is a specifically political difficulty involved in predicting the outcome of a political reform movement. Because the success of a reform and the belief that it will last depend on each other, the ultimate result can be very sensitive to small and unpredictable differences in popular confidence during the early stages. To the extent that instrumental success depends on confidence and legitimacy, which must in turn be generated by instrumental success, reformist politics becomes an art rather than a science. To the extent that the legitimacy derives from other sources, this particular problem becomes less acute.

One upshot of this discussion is that major social reform is clouded in uncertainty, in the sense that we cannot attach any probabilities to the various possible outcomes of the reform. Although there is no unique rule of rational decision making for such contexts, many (perhaps most) people would choose the maximin principle of acting as if the worst were the case. They would choose the constitution the worst possible consequence of which would be better than the worst consequence of any alternative, thus justifying Hume's "maxim, that in contriving any system of government, and fixing the several checks and controls of the constitution, every man ought to be supposed to be a knave, and to have no other end, in all his actions, than private interest. "13 Hence "a constitution is only so far good, as it provides a remedy against maladministration."14 The last statement, surely, is too strong, but for our purposes a weaker version suffices: a constitution can be rationally expected to be good only insofar as it provides a remedy against maladministration.

[end of page 714, start of page 715]

This is a distressing conclusion, but there is worse to come. The obvious retort to the Humean skeptic is that we can, after all, learn about the effects of constitutions, by trial and error or piecemeal social engineering. Edmund Burke was no great believer in the power of reason to legislate a priori, but he was not without faith in stepwise reform: "By a slow but well-sustained progress, tile effect of each step is watched; the good or ill success of the first, gives light to us in the second, and so, from light to light, we are conducted with safety through the whole series. We see that the parts of the system do not clash. The evils latent in the promising contrivances are provided for as they arise. One advantage is as little as possible sacrificed to another. We compensate, we reconcile, we balance."15

We have seen, however, that this proposal will not do. On the one hand, it is clear that by requiring local and initial viability of institutional reform, the set of possible changes is severely restricted. By piecemeal social change we can at most achieve local maxima, because many institutions that are viable in the large and in the long term may not be so in the small or in the short term. True, we may stumble upon the global maximum by trial and error, but we cannot rationally count on finding it. On the other hand, it is no less obvious that one cannot conduct experiments that involve the whole society and whose results will not emerge for decades, even centuries. Uncertainty and transition costs will block the adoption of any such proposal. People will have no motivation to participate in massive and protracted experiments of uncertain efficacy, unless the reform is perceived to be inherently fair and just. I now turn to this consideration, which will provide the positive side of my argument.

In modern democratic societies, the perceived justice of social institutions and policies is a condition for their long-term viability. This proposition, while intuitively plausible, is not easy to document empirically. The importance of the "the justice motive in social behavior" equals its elusiveness. Like all norms, those of justice and fairness are extremely context-dependent in the way they are interpreted and applied.16 They are, in particular, highly sensitive to such phenomena as anchoring and framing. The important norm of equality, for instance, can be implemented in very different ways, depending on the reference group - equality with whom? - and the dimension along which people are required to be equal - equality of what? Almost any policy can be both criticized and justified on grounds of equality, by suitable choice of reference group and dimension. These ambiguities easily lead to self-

[end of page 715, start of page 716]

deception, manipulation, and, after a while, to cynicism. It is easy to argue that ideals of fairness, equality, and justice are mere window dressings for self-interest.

I believe, nevertheless, that some norms of justice are more robust and stable than others. Specifically, I shall argue that there are negative norms of justice that are constraints or necessary conditions for consequentialist reforms in democratic societies, and positive norms that are sufficient, non-consequentialist conditions for reform. Both sets of norms are defined in terms of the central democratic value of equality: equality of treatment in the negative case, equality of influence in the positive case.

Negative justice. This could also be called "justice as fairness," although not quite in the Rawlsian sense. Assume that we are dealing with a reform that is proposed and justified on consequentialist grounds. For each citizen, the reform will imply a temporal stream of burdens and benefits; for simplicity, I assume that the net discounted value is positive for everybody, thus excluding purely redistributive measures. I further assume that the reform is only one among a set of proposed measures that would provide a Pareto-improvement over the status quo. Typically each proposal would confer special advantages on some groups of citizens, in addition to the general benefits provided to everybody. A paradigm example could be property laws: it is in the interest of everybody that there be some stable set of property rights, and in the interest of each group to have one particular set chosen.

In such situations we may observe the paradoxical outcome that the very plurality of cooperative arrangements prevents any one of them from being chosen. For each proposal there is some group that will think it unfair, compared to some other proposal under which it would be better off. We have, in other words, a bargaining problem, in which the outcome may very well be the breakdown of negotiations and the perpetuation of the inefficient status quo, which has, at least, the advantage of existing. The formal branch of bargaining theory, stemming from the work of John Nash, is not very helpful in dealing with this issue. It does not try to explain whether an efficient outcome will be reached, but tries only to determine on axiomatic foundations which efficient outcome ought to be chosen. This class of theories also rests on assumptions that are not observed in real-life bargaining situations. For one thing, the assumption that the outcome of bargaining does not depend on interpersonal comparisons of utility is clearly unrealistic. For another, the as-

[end of page 716, start of page 717]

sumption that the set of outcomes can be rendered continuous by admitting lotteries between physically discrete outcomes does not in general apply to real-life situations.

A less general but more useful theory of bargaining is that pioneered by Thomas C. Schelling.17 lt turns on the immensely fruitful idea that in inherently conflictual situations the only way to reach agreement may be to hit upon a naturally salient outcome, a "focal point" as Schelling calls it, or a "Schelling point" as it is usually called. What is naturally salient depends on perceptual psychology, social conventions, and shared history, not on formal features that can be stated in terms of rational-choice models. Typical focal points in bargaining are "allow no exceptions," "divide equally," "do as we did last year," or "do nothing." A more complex example is a price and income freeze as a tool for economic stabilization.

Focal points in democratic politics do not always embody substantive equality, but they have to represent formal equality, in the sense of impartiality or fairness. The feature of price and income freezes by which they gain legitimacy is not that everybody suffers equally, but that the distribution of suffering is perceived to be blind, mechanical, and impersonal. The (relative) gainers gain by accident, not because someone intends them to. Although there is great reluctance to use lotteries, there is also great acceptance for any policy that can be justified by such notions as "the great lottery of life" or "you win some, you lose some." By contrast, fine-tuned policy measures are easily suspected of intentional bias, an outstanding example being "trickle-down" prescriptions for economic growth. Even simple measures may of course be suspected of bias if it is transparent that they will systematically benefit some groups more than others. Proposals for flat tax rates are an example. The ideal focal policy, in addition to being simple in its conception and efficient in its result, must be opaque in its operation; like grace, it must fall upon the deserving and the undeserving alike. In actual cases, efficiency must often be compromised to achieve fairness and legitimacy. Simple policies are rarely optimal in the unrestricted policy set, although they may come out on top when constraints of political feasibility are added. 18

To find focal policies that represent a reasonably good approximation to efficiency is the major task of political leadership in democratic societies with highly organized interest groups. Many of these groups have effective veto power if they feel that a proposal is substantially worse for their members than some alternative policy that could also be chosen. This

[end of page 717, start of page 718]

blocking power of organized minorities can only be overcome through proposals that (1) are better for everyone than the perpetuation of the status quo and (2) do not benefit people differentially in ways that are intrinsically correlated with membership in any particular interest group.

Positive justice. The argument from negative justice applies mainly to small or medium-sized reforms, which can be justified on consequentialist grounds provided that the side constraint of fairness is respected. For major institutional changes an argument from positive justice is needed, since the negative argument sketched above then blocks any purely consequentialist reasoning. It is my contention that the nonconsequentialist values of justice, liberty, and democracy have been the major proximate causes of social change over the last few centuries. To be sure, the efficacy of these values must in turn be explained by material conditions, but this is different from saying that they are themselves material in content.

If a reform is perceived as fundamentally just, people will be motivated to endure the costs of transition and the extensive trial-and-error procedures that may be required before a viable implementation is found. Historical experience provides some evidence for this view. Universal suffrage was widely believed to be a dangerous system, but once the injustice of restricted suffrage came to be felt, the system had to go - before anyone knew what the consequences would be. The contemporary crisis of authority at many levels of society - in the family, the enterprise, or in the organization - also arose because of the perceived injustice of the old patriarchy and hierarchy. There is no doubt that new forms of organization and decision making have created a lot of choas [sic], waste, and inefficiency. Yet people are prepared to put up with these difficulties, because they go together with a clear gain in justice. They perceive, more or less clearly, that the problems are temporary rather than permanent, and that the steady-state effect of a diffuse authority structure may look very different from the transitional effect of diffusing authority.

The two reforms compared above - profit-sharing and workers' self-management - fare very differently in this respect. The profit-sharing proposal does not pretend to embody values that differ from those of wage capitalism. The private ownership of the means of production is left in place, as is the system of bargaining between workers and owners. Moreover, there is no pretension that the long-term equilibrium outcome of the bargaining will be more just or fair than under wage capitalism.

[end of page 718, start of page 719]

The argument is strictly limited to the proposition that the change from bargaining over a wage to bargaining over the share of the net product will have desirable short-term consequences with respect to full employment. My question, then, is the following: if the profit-sharing proposal were implemented and turned out to have various perverse, unanticipated side effects, as it very likely would have, would there be a widespread consensus to persist in the experiment until one had got rid of the bugs? My claim is that there would not be any motivation to endure the difficulties. The gains would be perceived as too abstract, uncertain, and remote, compared to the immediate costs and obstacles. Indeed, the very anticipation that this will be the case is likely to block even the initial implementation.

By contrast, the case for workers' self-management rests on the intuitively appealing idea that any joint or cooperative enterprise ought to be governed in common, by the equal influence of all concerned and to the equal benefit of all concerned. Because the proposal rests on the nonconsequentialist value of autonomy, it is more resistant to practical difficulties of implementation. If one particular realization of self-management turns out to be too clumsy or inefficient, this will provide a motivation not to abandon the whole enterprise, but to search for another, more viable form. Thus if it proves difficult to find a form of direct democracy that is neither too demanding on the participants nor too detrimental for productivity, one can experiment instead with representative economic democracy. In particular, workers may prefer self-realization in the work process itself over self-realization through participation in decision making. It may turn out that they value the possession of the right to self-management more than its actual exercise, much as one can value the freedom of mobility without ever desiring to use it. The democratic right to choose what rights and what kind of democracy to have can lead in many directions, except that total abdication of the right to self-government would have to be proscribed.

The implications of the above argument for the VP proposal are fairly obvious; hence I do not spell them out at length. The proposal would never gain acceptance on consequentialist grounds. The alleged effects are surrounded by massive uncertainty. They "could" come about, in some suitably abstract sense, but not in a sense that would motivate anyone to political action. Moreover, the proposal goes against a widely accepted notion of justice: it is unfair for able-bodied people to live off the labor of others. Most workers would, correctly in my opinion, see the proposal as a recipe for exploitation of the industrious by the lazy.

[end of page 719, start of page 720]

A counterargument might run as follows. Very little would be lost by implementing the proposal on a small scale, by a low-level guaranteed income. If it turns out that it has the predicted effect, one could then increase the guaranteed income up to the point, if ally, where further increases begin to have adverse effects. True, gradient-climbing cannot always get one to the global maximum, but sometimes it can. Also, even if we only reach a suboptimal local maximum, it might be an improvement over the present state. In other words, if there is nothing to lose and possibly something to gain, why not give it a chance?

The answer is simple. There is an abundance of actual or potential proposals of equal plausibility. In this "slum of possibles," why choose one rather than another? For reasons indicated above, implementing a major reform scheme requires commitment, belief, endurance; an uncommitted, hypothetical attitute [sic] would be a sure way to disaster. Experiments are useful, indeed necessary, if the underlying idea is widely held to be valid; they are pointless if the goal is simply to provide inputs to some social analogue of natural selection. Society cannot underwrite the pet ideas of each and every enthusiast who offers a panacea for our problems.


1. The following draws heavily on my "Arguments for Constitutional Choice," in Constitutionalism and Democracy, ed. J. Elster and R. Slagstad (forthcoming, Cambridge U.P.).

2. For an exploration of the distinction between utility and autonomy, see my Sour Grapes (Cambridge: Cambridge University Press, 1983), chap. 3.

3. A. Margalit, "Ideals and Second-Best," in Philosophy for Education, ed. S. Fox (Jerusalem: van Leer Foundation, 1983), 80.

4. S.-C. KoIm, La Bonne économie (Paris: Presses Universitaires de France, 1984), chap. II.

5. D. Miller, "Market Neutrality and the Failure of Cooperatives," British Journal of Political Science 11(1981): 309-329.

6. M. Weitzman, The Share Economy (Cambridge, Mass.: Harvard University Press, 1984).

7. For elaboration, see my "Self-realization in Work and Politics," Social Philosophy and Policy 3 (1985): 97 - 126.

8. See again my Sour Grapes, chap. 3.

9. For related arguments, see M. Granovetter, "'Advancement' in Theories of Social Evolution and Development," American Journal of Sociology 85 (1979): 489-515, and R. Boudon, La place du désordre (Paris: Presses Universitaires de France, 1984).

10. The following comments are indebted to Karl O. Meone. [sic]

11. J. Schumpeter, Capitalism, Socialism and Democracy (London: Allen and Unwin, 1961), 195.

[end of page 720, start of page 721]

12. For an argument to this effect, see my Explainiog Tehchnical Change Cambridge: Cambridge Unisersity Press, 1983), chap. 4.

13. D. Hume, Essays: Moralora, Political and Literary (Oxford: Oxford University Press, 1963), 40.

14. Ibid., 25.

15. F. Burke, Reflections on the Revolution in France (Garden City, N.Y.: Dolphin, 1961),185.

16. M. Yaari and M. Bar-Hillel, "On Dividing Justly," Social Choice and Welfare 1 (1984): 1-24.

17. T. C. Schelling, The Strategy of Conflict (Cambridge, Mass.: Harvard University Press, 1960).

18. I elaborate on this argument in my "Weakness of the Will and the Free-Rider ProbIem," Economics and Philosophy 96 (1985), 136-155.

[end of page 721]

[Elster, Jon (1987), Comment on van der Veen and Van Parijs, Theory and Society 15:709-722]

[ Books | Articles | Reviews | Index | The main menu ]