Introduction

Science is cumulative. Once a discipline has reached a certain level of maturity, most research normally builds upon, extends, and corrects prior work, and thereby leaves us with better knowledge than we had before. In this cumulative process, any theoretical advance is bound to be extended and improved as time goes on. Thus I think we should try, in responding to tests and criticisms of our work, to avoid being defensive, and to strive to draw as much truth as can be found in the work of others, even when it is critical of our own efforts. Thus I intend these comments to be an appreciation of (and an addition to) the foregoing more than a refutation of them.

Yet I must not pretend to have more scientific virtue than I really have, and immediately concede that it is emotionally much easier to respond in an appreciative cumulative way in this instance than it sometimes is. The foregoing all take my last book very seriously indeed, and a majority of the tests and observations in the foregoing papers are, it seems to me, favorable to my argument. That argument is confronted with much the broadest array of evidence and the most impressive tests in Jan-Erik Lane's and Svante Ersson's paper, and these tests, as the authors point out, undoubtedly support my argument. are also many encouraging observations in other papers and many interesting factual observations that I believe are far more corroborative of my argument than is evident at first glance. It also turns out, as I claim to show later in this article, that the most serious of the reservations in the foregoing commentaries - which are in Agne Gustafsson's article on how my concept of 'encompassing organizations' relate« to Sweden - really apply more to the policy prescriptions of the neo-corporatist political scientists than they do to my views.

All of the foregoing review articles at least refer to neo-corporatism and all contain interesting facts and observations that seem to me useful in testing and developing neo-corporatist ideas. In view of this, and because most of the readers of Scandinavian Political Studies probably have an interest in the neocorporatismliterature

---

* lam grateful to the National Science Foundation and to the Thyssen Stiftung for support of my research and to Gudmund Hernes, Assar Lindbeck, Preben Munthe, and Fritz Scharpf for helpful

corporatismliteraturethat my audiences in economics may not share, I shall try
here to offer some ideas about this literature that will, I hope, be useful to neocorporatistsas
as to other political scientists.

A Clarifying Anecdote

The cumulative and constructive character of the four foregoing articles can be made clearer with the aid of a personal anecdote. I was once watching a television in which a distinguished astronomer was being interviewed. The superficial journalist who was interviewing the astronomer asked whether there was any truth in the astrological belief that people were influenced by the stars under which they born. Before going on to offer the expected rejection of astrology, the astronomer paid a kind of tribute to astrology by emphasizing that his own science had grown out of astrology - that is, out of that type of primitive mythological speculation and awe about the heavens. I was startled by this comment, for the science of astronomy seemed to me very nearly the opposite or enemy of the superstitions of astrology, and the notion that the one might have grown out of the other seemed unpleasantly incongruous. But the fault was with me rather than with the astronomer - there is, of course, no inconsistency between the idea that science is more or less the opposite of superstition and the hypothesis that the latter emerged out of the former. Indeed, there is considerable to believe that most sciences did emerge out of myth, religion, superstition, magic. The idea that astronomy grew out of astrology really ought to comfort us; it suggests that, over the very long run, efforts to understand the world can be very cumulative indeed, even when they start with myth and superstition.

The social sciences are, of course, no more able to claim an immaculate conception the physical and natural sciences. The histories of economic thought, for example, often begin with medieval superstitions about usury and the 'just price', and there is nothing terribly wrong in supposing that economics began with this primitive metaphysical argumentation. The social sciences have, in general, emerged from the pre-scientific stage of inquiry more recently than the natural and physical sciences, and, in my opinion, they still have not entirely escaped from it.

My point is not so much that there is still a lot of myth and ideological prejudice the social sciences (though there is), but rather that the almost subconscious among many social scientists about what theory is (or should be) grows out of an analogy with the role of religion and ideology.

Social science has emerged almost entirely in the Western world and the religious heritage of the West is, of course, monotheistic. Usually, when monotheisticreligions been in competition with polytheistic ones, the monotheisticreligions won, and they have always won in the West. I have long been interested in the myths of the Vikings and am struck by their beauty and

insight. Yet when Christian evangelism reached Scandinavia, a monotheistic religion quickly won the competition with the charming stories about the Norse gods. A similar account could be given of the defeat of the impressive mythology of the ancient Greeks.

In the same way, one must be impressed by the success the monistic ideologies of the last century have had in winning converts. Marxist governments rule a large part of the world's population and Marxist ideas claim the allegiance of a substantial part of the intelligentsia in non-Marxist countries as well. Marxism also offers an explicitly monocausal explanation of all historical change - even intellectual and religious beliefs are explained in terms of class struggles growing out of the material conditions of production. There are some schools of classical liberal and anarcho-capitalist thought that are nearly as monocausal, and which also have a considerable capacity to attract converts.

I hypothesize that most human beings have only a limited ability to tolerate ambivalence, unanswered questions, and unresolved dilemmas. Thus a religion or ideology that offers a complete and unified explanation appears to have an evangelical advantage over incomplete and polytheistic or polycausal explanations.

Probably the religions and ideologies out of which the social sciences have grown have left us with a visceral sense of what theory is or should be. We sometimes a theory, at least when it is a very general theory as opposed to a special-purpose model, to offer a complete and unified explanation of social and economic reality. When a very general theory is found to be persuasive, it is taken to explain more than it does. In some circles, Keynesian economic theory, for example, is taken to be an overall approach to all of economic policy and choice of economic system; it is taken to be a justification for an economic system that is 'mid-way' between laissez faire and a Soviet-type economy, which tolerates a market system yet encourages übiquitous government intervention. In fact, Keynes' contribution dealt only with much narrower questions about macroeconomic Similarly, to the extent that a general theory leaves out crucial aspects of social or economic reality, it is sometimes thought to be faulty; it is often expected to offer a world view that offers a comprehensive basis for political and social choices. But 'world view' is a literal translation of a German word for 'ideology', and ideology and religion surely ought to be distinguished very sharply from theory, even if theories have emerged from them.

The Function of Theory

No fully satisfactory account of the function or purpose of theory in the social sciences has come to my attention. Because of this, and because my response to the comments in the foregoing papers on encompassing organizations and also my reaction to neo-corporatism hinge on the question of how theory ought to be understood and used, I believe it will be useful to consider here why we need

theory. Epitomizing the philosopher Bergson, we can say that 'time is a wonderfulinvention, it keeps everything from happening at once!' Less whimsically,we say that 'theory is a wonderful invention, because it spares us from thinking of everything at once!

Reality, and especially economic and social reality, is so incredibly complex and multi-faceted that it is, of course, impossible to understand it all at once. To understand everything at once would require more knowledge than we could acquire keep in our heads. A theory that took account of everything at once would also be logically so complicated that it would be mathematically intractable. try to understand everything at once is as foolish as trying to jump into one's trousers two legs at a time. The only way to understand the incredible complexity economic and social reality is to analyze relationships one at a time, or (when there is simultaneity) a few at a time. A theory is, above all, a simplification, if it is a good theory it is a fruitful simplification. The French poet Valery put it very tersely: What is simple is not true, what is not simple is not useful. simplifying, we understand a relationship that we did not understand before. the aid of the insight attained, we can obtain insights into other relationships, this often takes a long time. It took economists about a hundred years (from Adam Smith to, say, Alfred Marshall) to get a satisfactory understanding of the process by which price is determined in even a single, simple, purely competitive market; general equilibrium models that incorporate collective goods, imperfect information, technical change, and monopoly all at once are still beyond us.

In testing a theory empirically, there will often be misspecification if some of the variables a theory leaves out are not taken into account. Lane & Ersson's tests are so interesting partly because of the variety of the variables they have taken into account. Yet even in statistical tests there must be simplification, at least in the assumption that what is left out can be taken to be randomly distributed.

It is only when there is a logical relationship or complexity that we need a theory. A tentative generalization about the facts, such as the interesting conjecturein paper that ideological motives will sometimes help a broadly based political party limit the depredations of special-interest groups, is a hypothesis,not theory. This is not said to belittle hypotheses; they are sometimes far more useful than theories and they are also indispensable for the construction of theories. With good hypotheses and correct logical deductions about what other, non-obvious things must also be true if the hypotheses at issue are true, we have a useful theory - a hypothetico-deductive theory. It often turns out that the logical deductions in our theories, once they are properly understood and explained,are seen to be trivially simple. The simple supply and demand price determination model that it took economists of the past (some of them acknowledgedto men of genius) a century to get right is now sometimes taught even in secondary schools. But if a proposition has no logical structure, even of the

simplest kind, we are merely being pretentious if we call it a theory.

To my mind, the literature of neo-corporatism, as erudite, informative, and insightful much of it is, does not yet contain any hypothetico-deductive theory or arguments with a non-trivial logical structure. To say this is not to disparage the value of this literature; it is a literature of relatively recent origin and it would be much too severe to demand that it should already have reached the hypothetico-deductive But the history of science teaches us that schools of thought do not usually survive unless they do eventually develop genuine theories that not only include testable hypotheses, but also deduce further implications of these hypotheses that were not obvious before. Thus neo-corporatists should give a high priority to taking their researches to the hypothetico-deductive stage. I shall accordingly offer later in this paper some arguments that will, it is hoped, help neo-corporatists do this. These arguments of mine draw, in turn, on some of the observations about my concept of 'encompassing organizations' in the Rasch-Sørensen paper and especially upon the criticisms of this concept in the Gustafsson paper.

What RADON Left Out

My comments on the purpose of theory also relate directly to some of the points made about my book in the foregoing papers. One of the main contentions in Bjørn Erik Rasch's and Rune Jørgen Sørensen's paper assessing how The Rise and Decline of Nations (hereafter, RADON) relates to Norway is that the book lacks the needed institutional detail. They say 'Olson has developed a parsimonious precise theory. When it comes to realism and accuracy of implications, the scores seem to be much lower ... Olson's discussion is... almost devoid of institutional nuances ... are left out! Similarly, David Willis says that 'the application of a logic of utility-optimising behaviour in political systems finds a particular powerful formulation in the work of Mancur Olson (1965, 1982). On the other hand, the analysis seems incomplete in its neglect of the particular qualities and criterion for resources re-distribution ... the theory is in danger of focusing on the methods of group action, whilst ignoring the content of goals'. Gustafsson says that ' The Rise and Decline of Nations is a most stimulating book', but offers, among other criticisms, the view that 'it is necessary to take other factors into account'.

It is not appropriate for me to offer a judgment on the elements of praise for RADON in the foregoing comments. But I agree totally that my theory is incomplete,that am guilty of leaving out the institutional details about individual countries, that I often do not distinguish the different contents of the goals of different distributional coalitions, and that I leave out any number of nuances that would be needed for a full and realistic description of reality in each country. In saying that I have left out a very great deal, and that much of what I have left out is very important, my critics are entirely correct. Though I emphasizedin

phasizedinRADON that the theory was not monocausal and repeatedly said that it was essential to remember how many things had been left out of account, I should probably have said this even more often. In analyzing significant matters that I did not take into account, or relate to particular countries, the foregoing articles are making valuable contributions. This is one of the reasons why I hope they will be widely read.

Given that my theory involves drastic simplifications, what should we make of it? Does the truth of the assertions that I have left out much that is important mean that we should start over again and strive to create a new theory that takes everything into account? If what I have said above, about simplification being the very purpose of theory, is correct (and I believe most people who have any experience constructing theories or models would agree with me), nothing whatever can be gained by trying to develop a theory that includes everything. If my book makes any contribution, it does so because of its simplifications, not in spite of them.

One of the things we need to do is more research of the kind that is embodied in the foregoing review articles. We need to study what has been left out of account a theoretical framework. The main purpose of such studies should not be to determine whether any of the omitted considerations are important - some of the things left out are almost certainly going to be of great importance at least for certain purposes and certain situations. It is more important that we should ask whether the simplifications in a theory are appropriate for the purposes and the situations for which they are used. The physicist's finding that falling bodies, irrespective of their shape or weight, will in a perfect vacuum accelerate at a rate of nearly ten meters per second/per second has proven to be of immense practical even though it is based on the most unrealistic assumption that the bodies in question are in a perfect vacuum. But it would be absurd to restrict oneself the assumption that falling bodies are in a vacuum when designing a parachute! we must, of course, add information about such matters as the density the atmosphere and the shape and size of the parachute to predict how fast the parachutist will fall. I submit that in the same way economists and other social scientists need to consider when it is necessary to add more details about institutions to a model and when it is not. I believe the foregoing articles provide valuable information and analyses that will greatly help us in making decisions about when and how we should extend the theory.

The studies of matters that I have not included in the theory are also useful in other ways. As such studies accumulate, we may hope for additional simplificationsthat it possible for us to obtain new insights. These additional simplificationsmay be useful in the creation of new models and theories that would, for various purposes and settings, be more appropriate than the theory in RADON. But the new theories, if valid, would not refute the theory in RADON, if it is valid, since no two truths can be inconsistent. Though it doesn't

make any sense to accept both a monotheistic religion and a polytheistic religion as true, or to try to be a pious Marxist and a devout classical liberal at the same: time, there is nothing wrong with the idea that two different sets of theoretical simplifications can each be useful for certain purposes.

Indeed, we should also strive to integrate both of the valid and useful theories into a more general yet still simple theory. In successful sciences, the truth of today is not the error of tomorrow; it is more often the special case of the truth of tomorrow. Inquiries into matters that my theory leaves out of account will, I hope, eventually lead to new simplifications that will be integrated with my own theory, and thereby add to its explanatory power without excessive loss of parsimony.

Incentives for Narrow and Encompassing Organizations

The foregoing account of the need for logical structure in theory is, I believe, necessary deal properly with Gustafsson's distinction between what he calls my 'main theory' and my 'special theory! It will also be necessary hurriedly to review some elementary deductions in Chapter 3 of RADON. There I looked at the incentives face different kinds of coalitions for collective action, and showed that these incentives differ considerably according to whether the firms or individuals the decision-making unit receive a minuscule or a large percentage of the national income. (Gustafsson mistakenly says that I use the number of members to determine whether a coalition is narrow or encompassing, but this is emphatically the case; I use the 'proportion of the income-earning capacity' [page 48] of a country. It is also evident from the logic of my argument that I could not properly use the number of members to determine whether a coalition was narrow or encompassing.)

Consider first an organization for collective action that, like most such coalitionsin countries, includes only a minuscule share of the income-earning capacity of a country. Suppose, for easy figuring, that the firms or individuals in the coalition normally receive in the aggregate one percent of the national income.If coalition should use its resources to strive to make the country in which its members are located more efficient and prosperous, it will bear the whole cost of whatever it does to make the country more prosperous, but will normally receive, on average, only about one percent of the increase in the nationalincome. the organization's clients will normally gain from using their resources to make a country more prosperous only if the social or national costbenefitratio a hundred to one or greater, so such organizations cannot usually have any incentive to try to make a society more prosperous. The organization will normally serve its members better by using its capacity for collective action to obtain a larger share of the social output for its members by lobbying for special-interest legislation or by collusion or cartelization that obtains monopoly prices or wages for what its clients sell. The redistribution of the national income

in favor of the members of the organization will normally reduce economic efficiencyand of delays and complexities that are normally part of the process) reduce the rate of innovation and economic growth as well. Because the members of the coalition by stipulation receive only about one percent of the nationalincome, have an incentive to engage in redistributional struggle up to the point where the national income falls by a hundred or more times as much as the group wins in distributional struggle.

Now suppose, by contrast, that an organization for collective action encompasses of the income-earning capacity of a country. If that organization uses some of its resources to make the country in which its constituents live more efficient and prosperous, the constituents of the organization will get, on average, one-half of the increase in the national income. Thus such an organization some incentive to strive to make the country more prosperous. The organization will also have some incentive to seek to redistribute income from the other half of the society to its own constituents (Rasch & Sørensen usefully emphasize this). But since these constituents will, on average, absorb half of the loss in the national income, they will not gain from any redistribution to their own members that reduces the national income by two or more times as much as the amount they win in distributional struggle. The encompassing organization moreover, have a strong incentive to make the social losses or 'excess burden' of any redistributions to its own constituents as small as it possibly can, since its own constituents will bear half of that social loss.

Now one of the reasons why I have placed so much emphasis in this article on the role of logical entailment in theory should be evident. Once it has been pointed out, the logic of the two preceding paragraphs is, of course, obvious and trivially simple. But it is equally obvious that the same logic was used to derive the conclusions about both narrow and encompassing organizations: exactly the same logic that shows that narrow coalitions for collective action have extraordinarily incentives to engage in distributional struggle with little or no concern about the cost to society, also shows that this incentive is considerably attenuated for highly encompassing organizations. So valuable as Gustafsson's article is, he is wrong in distinguishing the 'special' theory of encompassing organizations from the 'main' theory in RADON; it is all one indivisible theory. The inherent logic of the matter gave me no honest choice but to point out that the incentives are less perverse in encompassing than in narrow organizations for collective action. No special theory was elaborated to deal with countries with relatively encompassing organizations for collective action, such as Austria, Norway, and Sweden; the same theory was applied to these countries as to the other countries.

Political Parties

Exactly the same logic was also applied in RADON to political parties, and this means that Gustafsson's very interesting (and theoretically plausible) conclusions it is political parties that have been much the more important source of encompassing or socially-efficient policies in Sweden is also an application of the argument in RADON. Since the application of the argument in that book to political parties will be important in the rest of this paper, and since Gustafsson seems to have overlooked my argument about political parties, it will be necessary to quote extensively from the argument about political parties in RADON:

The logic of the distinction between narrow and encompassing interests is not limited to specialinterest This is evident, for example, from an inference many political scientists have drawn from the observation of American politics. This is the inference that the United State;; would gain from stronger and more responsible political parties. These political scientists observe that individual members of Congress are overwhelmingly influenced by the parochial interests of their particular districts and by special-interest lobbies, and that incoherent national policies are often the result. The leadership of whatever party is perceived to be in control usually is to some extent concerned about the aggregate national consequences of the policies chosen, since there is some connection between the state of the nation and the election prospects of the party deemed to be in control. Party discipline, however, is so weak that the influence of the party leadership and the concern about the party's fate in the next election exert only a marginal influence. The conclusion that if party discipline could be strengthened and each party be held responsible for^the policies it chose and their outcome, then national policies would tend to improve.

The logic set out in this book can help to explain and justify the political scientists' argument, if it is combined with an analysis of the electoral system. The United States does not use proportional or any other electoral system that gives candidates or parties that come in second, third, or worse in a general election some portion of the power; the winner in any given general election wins it all. Thus it does not make any sense to have a political party in the United States that would over the long run expect to get, say, a fourth of the votes in a presidential election; that expected to come in second or lower, however, could gain something by combining that gave them a chance of winning. Thus the electoral system in the United States encourages two-party system (as do some other factors that need not be discussed here). Given that, each of the parties will be quite encompassing; each will attempt to represent a majority of the electorate. A party whose clients comprise half or more of the society naturally is concerned abou_t the efficiency and welfare of the society as a whole, particularly in comparison with lobbies for special-interest groups and congressmen accountable only to small districts. It is accordingly not surprising that systematic observers should note that American political parties were, on balance, more concerned about the welfare of the nation than were special-interest groups or individual congressmen, and therefore would favor stronger political parties. (Pages 50-51.)

Neo-Corporatism and Encompassing Organizations

Careful readers of RADON may at this point object that I was exceptionally emphatic about the dangers of using the concept of encompassing interest organizationsas basis for policy recommendations, and that the qualifications in my argument about the perverse consequences of narrow distributional coalitionswere far less strongly. These careful readers will point out that I said in RADON that it 'would be a mistake ... to suppose that any increase in the extentto

tenttowhich a special-interest organization is encompassing in necessarily desirable' (p. 49) and that 'the idea of encompassing organizations is not necessarilyalways guide for reform, but it is essential to a complete understanding of many organizations and institutions' (pp. 52-53). They may further argue that, with respect to Norway and Sweden in particular, I pointed to some special dangers in encompassing organizational structures and concluded that 'there is no guarantee that encompassing organizations will always operate in ways consistentwith well-being of their societies, or that the societies with such organizationswill always prosper' (pp. 91-92). The careful reader may therefore ask why, if the same logic was used to derive the conclusions about both narrow and encompassing interest groups and for political parties as well, was the argument about encompassing interest organizations so much more heavily qualified?

I must admit that at the time that RADON was written I had only an incomplete partly intuitive understanding of the dynamics and stability conditions encompassing interest organizations. In the book I raised questions about the stability and long run performance of these organizations, but unfortunately not develop an explicit model of the dynamics of societies with such organizations. Since the book was written, I have been able to make a little progress in the development of such a model, and thus can discuss this matter now. Interestingly, a more nearly explicit and complete account of the dynamics and stability conditions of encompassing interest organizations helps to explain some of the important observations in the articles by Gustafsson, Rasch-Sørensen, Willis. It also helps to show why I believe that the criticisms of aspects of my argument about encompassing organizations in the Gustafsson article, and the queries about it in the Rasch-Sørensen article, are better regarded as cumulative extensions of my argument than as refutations of it.

The facts emphasized by Gustafsson, at least, do, however, argue against some neo-corporatist policy presumptions. Neo-corporatists have not explained exactly why they believe neo-corporatist arrangements will work better than 'pluralistic' institutions. But they have described Austria, Norway, and Sweden as having more or less prototypically neo-corporatist institutions and these countries have relatively encompassing interest groups. The emphasis in most of the neo-corporatist literature on official favoritism to (if not legal monopoly status for) established interest groups also tends to favor encompassing interest groups. It is really only the logic of encompassing organizations, moreover, that would give neo-corporatist organizations more incentive to take account of the social interest than narrower or 'pluralistic⁵ interest groups would. Thus I regard the preference for neo-corporatist over pluralistic institutions, insofar as it has a logical basis, as resting on my concept of encompassing organizations. Notwithstandingthis, neo-corporatist literature is in other respects at odds with my argument; it certainly does not express the same concern about the depredations

of collusions, cartels, and special-interest lobbying that I do. The neo-corporatistsdo moreover, emphasize political parties and the significance of winner-take-all electoral systems or other condtitions that favor encompassing political parties. This is something that I discussed in the foregoing quotation, and in much more detail in my address to the International Political Science Association in July of 1985, which overlaps to some degree with the comments that follow.

The dynamics and stability conditions of a neo-corporatist system of encompassing groups can best be understood by comparing these interest groups to a political party of the kind discussed in the foregoing quotation where the electoral system favors large political parties over small ones. Suppose the stockholders or managers in, say, the automobile industry should want tariff protection, and decide to withdraw from the conservative party in a country wiih a two-party system to form a third political party designed to seek tariff protection the automobile industry. In some political systems, such a move could be rational. But in a winner-take-all two-party system, such a party could not win. Those who are major stockholders or managers in the automobile industries France or the United States cannot set up a political party designed to serve their special interests and expect to win the Presidency; only a coalition broad enough to get a plurality can win. Thus in certain electoral systems, large, encompassing parties can last indefinitely. When they do break apart it is likely to be because of a constitutional crisis that threatens the whole system, or because party's judgment becomes so deranged by ideological passion or other factors that it adopts politically suicidal positions. In quite a few countries, there are in fact large political parties that have been large for quite some time. One can easily imagine changes in constitutions and political rules that would give encompassing political parties even more durability.

Consider now an idealized neo-corporatist system with one encompassing interest group representing all employers and business and another representing all of organized labor. Suppose that for any reason, such as a recent devaluation that favors exporters or an exceptionally profitable innovation, there are extraordinary in the automobile industry. With exceptional profits, this industry will find it advantageous to continue in business even if it has to pay real wage rates that are above the normal wages for the relevant skill levels in the country as a whole. Higher real wages in the automobile industry, even when it is exceptionally profitable, will give the industry an incentive to produce in a lesslabor way and at a lower level of output than otherwise. With exceptional however, it may pay the industry to maintain its prior level of employment even at wages much above the level in the rest of the economy; it may simply decline to augment its work force the way exceptionally profitable industries would normally do.

An encompassing neo-corporatist union representing all workers in the

country, if it is intelligent enough to act in the best interests of its clientele, will have policies to get the highest possible real incomes for workers as a whole. It will have a 'solidary' wage policy designed to raise real incomes of the whole working class. But the highest possible incomes for workers in this society will be attainable only if the economy is maximally efficient. An economy cannot maximize output if it has different prices and wages for any homogeneous resource. If workers of a given skill have wages that are half-again higher in the automobile industry than in the rest of the economy, then profit-seeking employers in the automobile industry will stop taking on more workers when the increment in output - the marginal productivity - of the extra worker equals the high wage. Employers in other sectors will maximize their profits by taking on more labor up to the point where the marginal product of additional workers is as low as the wage in these other sectors. In this case the society has obviously not maximized its productivity, for more output could be obtained from the same work force if workers shifted from the sectors with lower wages and marginal productivity to the automobile industry. Output will reach its maximum only if this movement continues until the wage rates and marginal productivities are the same, for workers of a given skill, everywhere in the economy. So an encompassing neo-corporatist union intelligently and sincerely serving the interests of the working class would seek the highest possible wages consistent with full employment throughout the economy as a whole. The exceptionally high wages that the exceptionally profitable automobile industry could afford to pay some workers could only be attained at the expense of a less efficient economy and lower real wages for the rest of the work force, so a rational neo-corporatist union representing the whole of the working class would not seek, or even allow, disproportionately high wages in exceptionally profitable sectors.

Those workers in the automobile industry who keep their jobs will of course gain from a wage that is above that appropriate to their skill levels. They will get the higher wage, whereas other workers will have the lower wages arising because this industry isn't taking on as many workers as it otherwise would and thus pressing up real wages in the economy as a whole. The economy will be less efficient because of the non-uniform wages for equally skilled workers and this will tend to reduce the real national income below what it would have been. If the workers in the automobile industry are a small part of the total economy, they will bear only a small part of this loss, but get the whole of the gains from the higher wage. Thus the leaders of the automobile industry branch of the encompassing union, if they intelligently seek to serve the best interests of their clients, will demand disproportionately high wages even though this is bad for the economy as a whole.

Indeed, if any branch of any union can control the work force in its industry,
it will gain from using this monopoly power to obtain monopoly wage levels. The

firms in any industry will also gain if they can collude to charge monopoly prices for their outputs, even though this will similarly reduce the efficiency and the national income of society. When those firms or workers that can control the supply in some market obtain a monopoly price or wage, a smaller quantity of the product they sell is demanded. In general this will mean that less capital or labor will be used in this market. The capital or labor that would otherwise have been used in the cartelized sector will add to the supply in uncartelized sectors, and this will lower returns in these unorganized sectors (except, as has been shown with two-sector general equilibrium models, in certain very special cases).

When many of the firms in an economy have colluded and in effect cartelized their industries, and when many of the labor markets are cartelized by professional or labor unions, the scope for employment in unorganized sectors shrinks. When this occurs, there is involuntary unemployment (which, as I have shown elsewhere (Olson 1985), does not occur in economies with no collusion cartelization). The process that has been described is not merely a theoretical but rather a process that has in fact taken place in a great many economies, both corporatist and non-corporatist. It mainly explains the increases the unemployment rates since the 1950's in a country such as West Germany, for example, and is also part of the explanation of the reduction in rates of economic growth.

Just as the firms and workers in particular branches of industry have an incentive cartelize the markets they control, so they also have an incentive to lobby for special-interest legislation. If the firms in an industry can get tax loopholes or subsidies from the government, they will normally gain from this even though the economy is made less efficient. If the firms and the union in an industry can cooperate to pressure the government to give them protection against imports, or to provide bailouts when the firms are losing money, they will both normally gain from this even though this means that less efficient industry is being preserved the expense of more efficient industry.

The general point is that organizations to cartelize markets or to lobby governments not need a majority or even a plurality of the society to obtain their objectives, thus are in a very different situation from the encompassing political party in an electoral system with some winner-take-all features. All a collusion or cartel needs is control over the supply in a single market. All a lobby need is enough resources to employ a lobbyist, or to make campaign contributions, to provide enough campaign workers to make a difference in crucial districts, and so on. A trade association of firms in a particular industry, or the union that represents the workers in a single craft or industry, will normally be large enough to have significant lobbying power, even if the members of the organization in the aggregate get, say, only one percent of the national income.

This means that the branch organizations of an encompassing neo-corporatistbusiness
labor organization have an incentive to push for the interests of

their own branch, even when this is not in the interests of the members of the encompassingorganization a whole. The constituent units in the 'hierarchically organized' interest groups that Phillippe Schmitter considers a defining feature of a neo-corporatist society, have strong incentives to be insubordinate and little or no incentive to take the view of the encompassing organization as a whole. If, as is sometimes the case, the encompassing organization is a federation of partly independent organizations, the organizations in separate sectors can break away with less difficulty than if they are simply branches.

Some very large constituent units in the encompassing organization may be to some degree encompassing themselves and have some incentive to participate in encompassing policies. But the logic of 'the exploitation of the great by the small' explained in The Logic of Collective Action indicates that in the long run the members of these relatively encompassing parts of the neo-corporatist organization suffer in comparison with the members of units that are less encompassing. same logic applies to the members of any narrow distributional coalitions that exist in an environment with a highly encompassing organization; the encompassing organization will have an incentive to make some sacrifices in the social interest, and thereby benefit the narrow distributional coalition members as well as itself, while the narrow distributional coalition will have an incentive to pillage the society for all it can get even as it gets a free ride from the encompassing organization. The members of the encompassing organization may eventually tire of the exploitation of the great by the small and break down into smaller units. On the other hand, exceptionally astute politicians may also occasionally be able, if they understand the logic of the situation, to organize agreements among all the distributional coalitions that increase social efficiency, giving each of these groups some share of the resulting social gains.

It might seem that the problem of insubordinate branches and independent coalitions could be controlled by appropriate by-laws for the encompassing neocorporatist group or by government legislation or policies that discriminate or even outlaw, independent action by branches or other nonencompassing The neo-corporatist literature does often speak of government licensing or other devices to give a monopoly to the corporatist pressure group.

Severe legislation might, albeit in ways that would have to put some strain on traditional democratic freedoms, prevent break-away organizations from controlling markets. But how could it prevent branch units or local units of the neo-corporatist organization with a legal monopoly from pressuring the leadership of the organization to allow the branch to have a monopoly price or wage? Or from lobbying the leadership of the encompassing organization to pressure the government on behalf of the special interest of the branch of industry in question?

The way to get elected to the leadership of a neo-corporatist interest group,

like any other organization, is normally to put together a coalition of sub-units that is large enough to obtain a plurality. The leaders of branch units therefor«; often have an incentive to insist that only candidates for central office that agree to allow them freedom to set prices or wages separately in their own market, or who promised to support special-interest legislation for those in this market, will get the branch's support.

Even when the neo-corporatist organization is given a legal monopoly, subsets of members that would gain from monopolizing a particular market, or from special-interest legislation for that market, may be able, if 'selective incentives' can be found, to organize a caucus or lobby within the neo-corporatist organization pressure it to serve the sectional interest at the expense of the encompassing This is a fundamental problem for the neo-corporatist solution that has not, to my knowledge, been examined. In the very long run, how could a society prevent subsets of members of the neo-corporatist organization with a legal monopoly of representation from being controlled in large part by internal lobbies working on behalf of internal sub-groups that found the selective incentives to organize internal caucuses or lobbies? A government can stipulate by law that only specified organizations are allowed to lobby the government. But how in practice would it prohibit the creation of internal caucuses organizations to lobby internally to change the policies of the legally established lobby, without abolishing freedom of organization itself? Perhaps it is no accident that so many of the early corporatists were also fascists.

Conclusion

The theoretical possibilities that have just been described may, perhaps, facilitate research into what has been happening lately in some neo-corporatist democracies. are observers of the Swedish scene, at least, who point to events that would appear to be consistent with the theoretical logic I have just set out. The economist Assar Lindbeck argued in conversation with me that recently the actual policies of encompassing organizations in Sweden have apparently often been dictated by relatively small subsets of the membership with special interest s; government bailouts of Swedish shipbuilding were not in the interest of Swedish labor as a whole, but the LO, presumably at the behest of its constituent units in the shipbuilding industry, had favored such subsidies. A sociologist, Scott Lash, has just published an article entitled The End of Neo-corporatism?: The Breakdown Centralized Bargaining in Sweden' (1985). This article argues, from a somewhat Marxian perspective, 'that there is a long-term trend toward decentralization Swedish industrial relations' (p. 217). The economist Erik Lundberg, looking at a broader range of issues than are under discussion here, has depicted, in 'The Rise and Fall of the Swedish Model' (1985), a set of discouraging developments that are in a broader way generally consistent with the foregoing argument.

Most pertinently, the argument that has just been made is consistent with all of the foregoing review articles. If my argument here and in RADON is correct, we should not be surprised that Gustafsson found that more encompassing behaviour emerged from a large political party than from organized interest groups; in some sense, all branches of a political party will lose a lot if that party fails to win a national majority, but the constituent units of the interest groups may gain from cartelization and special-interest lobbying even though they undermine an encompassing policy. There has been a tendency in Sweden for encompassing bargaining to break down, but it is not difficult to imagine rules of the game that would make encompassing political parties quite stable in that relatively homogeneous country. Willis's emphasis on the way in which broad and ideologically based political parties can, to some degree, countervail narrow distributional coalitions also fits right in with this argument. The Rasch- Sørensen observation of ambivalence in encompassing interest groups in Norway also broadly consistent with what has been said here. So, of course, is the Lane-Ersson finding that 'adding other institutional factors like corporatism ... results in very minor changes^ whereas 'the length of time of institutionalization matters very much! The findings in the foregoing review articles and in this Appreciation' suggest that some of the future research of neo-corporatists ought to focus much more on encompassing political parties, on rules and arrangements limit monopoly price and wage setting, and on how to develop political resistance to special-interest lobbying.

It should now be clear why I have placed so much emphasis on the cumulative nature of science and the need for simplification. It should be obvious, too, why I am deeply grateful to the authors of these four most constructive review articles, to Jan-Erik Lane for eliciting them, and to the Editor of this Journal for welcoming this exchange.

REFERENCES

Lash, S. 1985. 'The End of Neo-corporatism?: The Breakdown of Centralized Bargaining in Sweden',
British Journal of Industrial Relations 23, 215-39.

Lundberg, E. 1985. 'The Rise and Fall of the Swedish Model', Journal of Economic Literature 23,
1036.

Olson, M. 1965. The Logic of Collective Action. Cambridge, Mass.: Harvard University Press.

Olson, M. 1984. 'Microeconomic Incentives and Macroeconomic Decline*, Weltwirtschaftliches Archiv
631-645.

Schmitter, P.C. 1974. 'Still the Century of Corporatism', Review of Politics 36, 85-129.