Erik Moberg ©


1 The problems
2 The key concepts
3 The empirical data
4 Rothstein on the second problem
5 Rothstein on the first problem
    5.1 Introduction
    5.2 Social dilemmas
    5.3 Are the welfare goods private or public?
    5.4 The logical structure of Rothstein's explanation
    5.5 Some further comments on Rothsteins explanation
6 A sketch of an alternative explanation
7 Summary



Bo Rothstein starts his paper "The Universal Welfare State as a Social Dilemma" with two alleged empirical observations. One is that the welfare state programs of the industrialized Western nations are widely different in "quality and scope". The other is that these differences have developed in the last few decades and thus are a quite recent phenomenon. "If we go back to the early 1960s", writes Rothstein (p 213f), "we can see that these countries spent almost the same as a percentage of GDP on welfare policies".

Rothstein finds both matters surprising. The great differences as such, although well-known, are surprising since the countries involved have "basically the same type of social, economic, and political structures; that is, they are all Western democratic capitalist market economies" (p 214). The recent emergence of these differences is also surprising, since it has occurred during a period when the international forces favoring convergence grew stronger. "Given the internationalization of values, increase in trade, globalization of capital, etc., this is", says Rothstein (p 214), "an unexpected development". Underlining the point he adds (p 214) that "[m]ost social scientists working in the early 1960s would probably, without the benefit we have of hindsight, have predicted convergence in social policy between these countries, rather than a dramatic divergence".

Rothstein thus formulates two problem:

  1. How come that the differences in social welfare spending are so great when the countries themselves are so similar in their social, economic and political structures?
  2. How come that the differences emerged during a period when the international forces favoring convergence grew stronger?
Having formulated these two problems it is furthermore quite obvious that Rothstein has an ambition to deal with at least the first of them – he really claims to have an explanation for the great differences. The development of this explanation is, in fact, his main undertaking in the paper. Whether he also purports to deal with the second problem is however less clear. In spite of his underlining of the importance of this problem in the beginning of the paper, he seems to forget it almost immediately. There is no further talk about it. Even so, I will however deal with both problems below, since they are related.


Since Rothstein’s discussion is about the "quality and scope of welfare state programs" (p 213) it is important to find out what he means with these concepts. Thus, and first, what are the programs which he calls "welfare state programs"? He is, in fact, not very clear on this point. One clue may be the other words which he sometimes uses for denoting his objects of study. He talks, for instance, about "social insurance and social assistance" (p 213), about "welfare policies" (p 214), and about "social policy" (p 214). Sometimes he is also more detailed. He says, for instance (p 218), that "social programs" include "old-age pensions, health care, child-care, education, child allowances, and health insurance". In another place (p 228), in a table showing attitudes towards "public expenditures", he uses the following eight categories: (1) Health care, (2) Support for the elderly, (3) Support to families with children, (4) Housing allowances, (5) Social assistance, (6) Primary and secondary education, (7) Employment policy, and (8) State and municipal administration. In that context he also writes as if most of the categories mentioned are included in what he calls "welfare state programs". The only exceptions, I think, are "state and municipal administration" and perhaps also "secondary education". Rothstein’s concept of "welfare state programs" is thus clearly, and independently of its exact interpretation, a rather wide concept.

The concept of "quality" of welfare state programs is also of great importance in Rothstein’s reasoning. His main distinction here is between universal welfare programs, which do not discriminate among citizens, and selective programs, which rely on examining the needs of those applying for support. Another way to put it is to say that universal programs include all citizens, whereas selective programs are organized just for those less well-off.

The distinction between universal and selective programs is, I think, quite common, but it may be elaborated in various ways. There is, for instance, a paper by Korpi and Palme (1998) in which an interesting and clarifying, and also somewhat more elaborated, typology for social welfare institutions is presented. Since this paper is of a general interest for the discussion here, and also used as a reference by Rothstein, I will describe parts of it in some detail here. Korpi and Palme distinguish between four kinds of "welfare state institutions" or "social insurance institutions", namely "encompassing", "corporatist, "basic security" and "targeted" institutions. A first comment on this typology is that the difference between "programs", which is Rothstein’s word, and "institutions", which Korpi and Palme talk about, is just incidental and, as I understand it, of no importance. Even Rothstein (p 215) writes about "the institutions of the welfare state programs" and, indeed, claims (p 230) to have solved his main problem by means of "institutional theory".

Korpi’s and Palme’s typology is however based on more dimensions than the simple distinction between universal and selective policies. They do, for instance, take the roles of employers and corporations explicitly into account. Still I think it is correct to say that Korpi’s and Palme’s categories, if taken in the order mentioned above, from the encompassing to the targeted ones, are successively, and roughly, less universal and more selective. A few quotations from Korpi and Palme may be used for substantiating this point and will also, in the process, shed some more light on the distinction between universal and selective institutions.

One aspect, in which Korpi and Palme are interested, is the role which the different institutions accord to markets and to politics in the redistributive processes. They argue (p 670 f) that the encompassing model interferes most with markets followed, in order, by the corporatist model, the basic security model, and finally the targeted model which "apparently involves the lowest degree of political interference with market distribution".

Another aspect which they discuss is the size of the public redistributive budget in the four cases. Here, again, the order is the same. Thus the encompassing model is expected to generate (p 672 f) "the largest budget size", followed by the corporatist model, the basic security model, and finally the target model which is expected to have the "smallest budget".


Concerning the "scope" of the welfare programs Rothstein is surprisingly scanty - the quantitative differences is after all a main object of his study. Concerning the situation in 1985, with the alleged great differences, he just writes (p 213) that "the Scandinavian countries spend about twice as much, as a percentage of gross domestic product (GDP), on social insurance and social assistance as the United States" and that "[m]ost other European countries’ spending falls somewhere inbetween". Obviously some more data is needed for a serious discussion. The easiest way to get it is to quote directly from the source on which Rothstein relies, which is the work by Korpi and Palme already mentioned. Their figures are those in the 1985-column in table 1. As we see the countries are also arranged in groups according to Korpi’s and Palme’s typology.

Nation and Type of
Welfare State Institutions
Total Welfare Policy Expenditures as Percentages of GDP
  1960 1985
Finland 8.8 21.3
Norway 7.9 28.0
Sweden 10.8 29.5
Austria 15.9 24.3
Belgium - 26.4
France 13.4 27.3
Germany 18.1 23.3
Italy 13.1 20.5
Japan 4.1 11.1
Basic Security    
Canada 9.1 15.6
Denmark - 27.5
Ireland 8.7 21.8
Netherlands 11.7 28.2
New Zealand 10.4 14.6
Switzerland 4.9 13.5
United Kingdom 10.2 19.4
United States 7.3 12.0
Australia 7.4 9.9
Table 1: Expenditures for social welfare systems in a number of nations 1960 and 1985.

Concerning the situation in the early 1960s Rothstein is even more abrupt. Without presenting any figures at all he just states (p 214) that the Western countries spent "almost the same as a percentage of GDP on welfare policies". Again he has a reference, namely an OECD report form 1994 (OECD, 1994). The figures given in that report are those presented in the 1960-column of table 1.


I will start my discussion with Rothstein’s second problem about the surprising appearance of widening differences during the period from the early 1960s to 1985. At first it is interesting to note that Rothstein, in spite of the fact that he formulates the problem, and stresses its importance, never, as I have already mentioned, really deals with it. The discussion in his paper, as I understand it, is rather exclusively devoted to the first problem. I will return to that problem later on, but I will start with the second problem. At first it may be worthwhile to see if it exists at all. As I have said, Rothstein himself does not present any figures but just states, that, in the early 1960s, the Western countries "spent almost the same as a percentage of GDP on welfare policies". Is this really true? Since we now have the relevant figures available in table 1, it is possible to settle the issue.

In order to make the two series of figures – the one for 1960 and that for 1985 – easily comparable they could be "normalized" around a common arithmetic mean, say 100. The mean of the figures for 1960 is 10.1, and in order to bring that to 100 it has to be multiplied by the factor 9.9 (100/10.1). Then, multiplying all the 16 figures in the 1960-column with that same factor, gives the result shown in the upper row of marks in diagram 1. The same kind of operations applied to the 18 figures for 1985 in table 1 give first an arithmetic mean of 20.8, then a factor for multiplying of 4.8, and finally the result in the lower row of marks in diagram 1.

Diagram 1: Differences in the studied nations' social welfare spending in 1960 and 1985.

It is now easy to compare the differences in social welfare spending in 1960 and in 1985. We see, contrary to Rothstein’s claim, that there are very considerable differences between the nations’ spending in 1960. We also see, again contradicting Rothstein, that the differences have not increased. Rather, the differences have decreased - there has, in fact, been a convergence.

This result may perhaps be disputed in various ways. It may, for instance, be argued that the differences could be measured in other ways. Rothstein himself does however use exactly the measure used here. Thus, when describing the differences in 1985, he says (p 213, my italics) that "the Scandinavian countries spend about twice as much, as a percentage of gross domestic product (GDP), on social insurance and social assistance as the United States". Rothstein thus measures differences exactly as I have done in the comparison above, by means of factors. Using this method we thus see, for instance, that the highest figure in 1985 (the one for Sweden) is 3.0 times higher than the lowest one (the one for Australia), whereas the highest figure in 1960 (the one for Germany) is 4.4 times higher than the lowest one (the one for Japan). Thus, and again, we get the result that the differences, contrary to Rothstein’s claim, have decreased.

Perhaps it may also be argued that extreme values, for one reason or another, may be erratic and thus should be disregarded. We may, for instance, omit the two most extreme values at each end of each row. If so, we get a different result – the conclusion will really be that we have divergence rather than convergence. This procedure does however have its own problems. Why should the extreme values in the first place be considered as erratic? And even if that question could be given a good answer, why should we disregard just two values at the ends of the rows, and not for instance one or three?

Taking everything into account, Rothstein’s blunt assertions that there were no differences in the beginning of the 1960s, and that differences thereafter not only appeared but also, successively, widened dramatically, are therefore unsupportable. Rather it seems quite clear that there has been differences all the time, and convergence rather than divergence. Thus, as for Rothstein’s second problem, there is perhaps not much to be surprised about, taking "the internationalization of values, increase in trade, globalization of capital, etc." into account.

Furthermore, since there were differences between the countries not only in 1985 but also in 1960 it may be interesting to see how these differences are related to each other. Diagram 2 illustrates these relationships for the countries for which data are available. It shows the total welfare policy expenditures for the different countries in 1960 and 1985 as shares of GDP. The two lines in the diagram indicate the medians. Thus there are eight countries below the horizontal line and eight above, and similarly there are eight countries to the left of the perpendicular line and eight to the right of it.

Diagram 2: Welfare policy expenditures as percentage of GDP in 1960 and 1985.

We see that most countries fall either in the lower left part of the diagram or in the upper right part, and that most of the remaining countries lie fairly close to the median lines. The expenditures in 1985 are thus fairly well correlated with those in 1960. The countries which had relatively low (or high) expenditures in 1960 also had, on the whole, relatively low (or high) expenditures in 1985. The correlation coefficient r = 0.6. There are, though, some exceptions, among which Norway, which went from quite low expenditures to remarkably high ones, is the most striking.

This correlation indicates that the properties of the nations, which lie behind the differences, have some stability over time.


5.1 Introduction

Since Rothstein’s second problem is based on a false empirical observation, and in that sense is a pseudo-problem, or a non-existent problem, it is not much of a loss that Rothstein does not deal with it. There are differences between the Western nations’ welfare policies now (or at least in 1985), and there were such differences in the 1960s. We can therefore turn to Rothstein’s first problem about the reasons for these differences. That problem is certainly not a pseudo-problem. So, why do the social welfare programs of the Western democracies differ so much in scope in spite of some general similarities between these nations?

Rothstein starts his discussion of this problem by refuting two possible rather straightforward or conventional political explanations. He says (p 214) that the differences between the countries cannot be accounted for in terms of "the ideological orientation of dominant political parties", nor in terms of the "general norms and values" prevailing in the societies. Finishing his rather short discussion about these matters he claims (p 214) that "standard theories about economic development, political power, or social norms cannot explain the differences in welfare state programs." I will come back to these matters in part 6 below.

The explanation which Rothstein offers, and develops in the main part of his paper, is of a quite different nature. His main idea is that the different welfare programs in the Western nations can be considered as different outcomes in a particular kind of games called social dilemmas. Rothstein’s explanation is thus formulated within a game-theoretical framework.

5.2 Social dilemmas

For the clarification of the concept "social dilemma" Rothstein refers to a paper by Elinor Ostrom (1998). In that paper Ostrom writes (p 1) that "[s]ocial dilemmas occur whenever individuals in interdependent situations face choices in which the maximization of short-term self-interest yields outcomes leaving all participants worse off than feasible alternatives. In a public-good dilemma, for example, all those who would benefit from the provision of a public good – such as pollution control, radio broadcasts, or weather forecasting – find it costly to contribute and would prefer others to pay for the good instead. If everyone follows the equilibrium strategy, then the good is not provided or is underprovided. Yet, everyone would be better off if everyone were to contribute."

Ostrom also writes (p 1 f) that "[s]ocial dilemmas are found in all aspects of life, leading to momentous decisions affecting war and peace as well as the mundane relationships of keeping promises in everyday life" and that "[t]he prisoners’ dilemma has become the best-known social dilemma in contemporary scholarship".

In a social dilemma there are thus two outcomes:

Now, the problem is to understand when the one or the other outcome will result. Ostrom starts her discussion by considering (p 2) "[o]ne of the most powerful theories used in contemporary social sciences – rational choice theory – [which] helps us understand humans as self-interested, short term maximizers." The prediction of this theory, she says (p 2), is "that no one will cooperate" in "one-shot or finitely repeated social dilemmas". But all real dilemmas are "one-shot or finitely repeated", and consequently all real dilemmas should end up in the first outcome above, the non-cooperative one.

But that is not the case; people do not always behave like that. According to Ostrom people sometimes, for one reason or another, really succeed in solving their dilemmas. Sometimes, and somehow, people succeed in being less self-interested and more far-sighted than assumed in the rational choice model. Indeed, if it were not like that, the present society would not exist. "In prehistoric times", writes Ostrom (p 2, my italics), "simple survival was dependent on the aggressive pursuit of self-interest and on collective action to achieve cooperation in defense, food acquisition, and child rearing". Indeed, in the very beginning of her article (p 1), she addresses her readers with the claim that "[y]ou would not be reading this article if it were not for some of our ancestors learning how to undertake collective action to solve social dilemmas".

The conclusion drawn by Ostrom, and by many other social scientists, is that the rational choice theory is somehow wrong or insufficient. In order to be able to account for the frequent happy outcomes of social dilemmas, the theory has to be complemented, or elaborated, in some way. How should this be done? There are many answers to this question, taking many different aspects of the dilemmas into account, but it is quite common that the presence, or non-presence, of "trust" is emphasized. Rothstein, for instance, writes (p 222) that "one of the things needed to solve social dilemmas is trust and other similar norms of reciprocity, which ensures that enough others will behave cooperatively."

5.3 Are the welfare goods private or public?

Rothstein writes (p 215) that "[f]rom a mainstream neo-classical economic perspective, most of the things provided by modern welfare states are essentially private goods. Such goods as health care, social insurance, and education, for example, can be privately consumed. This means that A, who owns the good, can exclude B from consuming the good in question." All of this is quite correct, but it is also a problem for Rothstein. The reason is that the social dilemma approach, according to Rothstein (p 215), "by definition, relates only to public goods". In order to be able to explain the different spending on welfare policies in different countries in terms of social dilemma outcomes, he thus feels compelled to show that neo-classical economics is wrong and that the welfare goods are in fact public goods.

In a rather lengthy discussion he therefore argues (p 215 ff) that "most parts of what is known as the modern welfare state are in fact a public good and thus a candidate, as good as any, for the problems we know as social dilemmas and collective action". The main argument is that the state provides the welfare goods more efficiently than markets, which are hampered by adverse selection, moral hazard and high transaction costs. I have two comments on this discussion about the nature of the welfare goods.

The first one is that, considering Rothstein’s undertaking, it hardly matters whether the goods are private or public, and therefore the discussion is superfluous. The social dilemma approach, as I understand it, is not limited to situations involving public goods. A prisoners’ dilemma, for instance, may very well be, and usually is, a game about private goods. The essence of a social dilemma, rather, is that short-term, self-interested behavior will lead to an outcome which is worse than other available outcomes. Such dilemmas may concern private as well as public goods.

My second comment is that I agree with neo-classical economics and thus think that Rothstein is wrong in his claim that the welfare goods are public. Even if the state provides welfare goods more efficiently than markets, which may be disputed, that does not prove that welfare goods are public goods. The definition of public goods, as Rothstein is well aware, is hardly in terms of efficiency but rather in terms of consumption possibilities. Thus a good is public if it may be consumed by several people, or, as Rothstein himself puts it (p 215), if "one individual cannot exclude others from using" it. This is obviously something else than being produced efficiently by the state. In fact, some public goods may be best produced privately (for instance club goods), and some private goods may perhaps be best produced publicly (possibly welfare goods).

My position that the welfare goods are, in fact, private, is important, since, contrary to Rothstein, I believe that the Western nations’ differences in welfare spending may be explained in terms of this very property. I will come back to this topic in part 6 below.

5.4 The logical structure of Rothstein’s explanation

Rothstein’s main undertaking, as I have already emphasized, is to explain the differences in the Western democracies’ spending on welfare policies. But, in spite of this, the exact details of Rothstein’s explanation are not clear to me. The explanation is hardly chiseled out with clarity and in detail. What follows is therefore partly an interpretation of Rothstein’s arguments, but I hope and believe that my interpretation is reasonably fair.

One important element in Rothstein’s explanation is his association of high spending with universal welfare policies, and low spending with selective policies. This seems quite reasonable. Taking the similarities between Korpi’s and Palme’s typology, and Rothstein’s distinction between universal and selective policies, into account, it is in accordance with the figures in table 1. It is also in agreement with Korpi’s and Palme’s explicit contention, quoted above, that the encompassing model generates the "largest budget size".

But the association of high and low spending with universal and selective policies respectivly is obviously not enough. After the establishment of that association it is necessary to explain why nations sometimes adopt universal policies, and sometimes selective ones.

It is in his efforts to deal with this crucial task that Rothstein evokes the discussion about social dilemmas. If I do not misunderstand him completely, he associates universal and selective policies, respectively, with the two types of outcomes of dilemmas which Ostrom, as quoted above, writes about. Universal policies are thus the happy outcomes which make everybody better off, and selective policies are the unhappy outcomes which leave all participants worse off. That this really is how Rothstein thinks is, I think, evident from his writing (p 229 f, my italics) that "if the welfare state is a public good", which he considers it to be, "then it is obvious that the Western countries have had different success in finding solutions to this social dilemma". Thus, there are good and bad outcomes, but all nations have not succeeded in finding the good ones. Some nations have failed to solve the dilemma.

There are thus two possible outcomes to this particular kind of social dilemma:

Before continuing I shall just say that I have had some difficulties with Rothstein’s terminology. Thus, he sometimes calls both types of outcomes equilibrium outcomes. He writes, for instance (p 230), that "both a selective and a universal welfare state are likely outcomes, which means that we face a situation with multiple equilibria". Sometimes he also seems to characterize both types of outcomes as solutions. He writes, for instance (p 230, my italics), with obvious reference to the two types of outcomes, that "structurally similar states manage to solve this social dilemma in very different ways". I do believe, however, that these variations are unintentional and without any real significance. I think it is possible to stick to the terminology explained above, and I will try to do so in the following. So, why do the nations sometimes end up in the equilibrium outcome, and why do they sometimes solve the dilemma?

In his discussion of this problem Rothstein uses (p 218) a schematic example of a universal welfare state as shown in table 2. There are five income groups, each constituting 20% of the population. The taxes – all of which are income taxes - are proportional and 40% for all citizens, and the welfare transfers are – in absolute terms – the same for all citizens, namely 240 units. With this system the two uppermost groups – A and B – contribute more to than they get, and they will thus, assuming self-interested behavior, vote against system, whereas the groups D and E, which gain more than they contribute, will support the system. Assuming the use of the majority rule, the members of group C, who gain as much as they contribute, will thus become "swing voters" – "they will decide what the majority will be" (p 220).

Group Average income Tax (40%) Transfers Income after taxes and transfers
A (20 %) 1000 400 240 840
B (20 %) 800 320 240 720
C (20 %) 600 240 240 600
D (20 %) 400 160 240 480
E (20 %) 200 80 240 360
Table 2: Taxes and transfers in a universal welfare state.

Generalizing from this schematic example Rothstein then writes (p 220) that "[s]ome modern capitalist democracies have more universal welfare systems, and others have more selective arrangements, and this follows to a large extent from the electoral behavior of the middle class". Consequently he concentrates his following analysis on the behavior of the middle class.

Before turning to that analysis a comment on Rothstein’s use of table 2 is however necessary. It is true, as Rothstein says, that a universal system will be established if a majority votes for the system shown in the table. But he is wrong when he says that a selective system will appear if a majority votes against the system in the table. The immediate result will rather be no welfare system at all. On this point Rothstein thus mistakes a situation with no welfare policy at all for a situation with selective policies. This, of course, is a serious error, and I will come back to it in the next section. Here, however, in order to be able to continue the discussion, I will disregard this error, and assume, with Rothstein, that people either vote for universal policies, or for selective policies.

So, how come that the middle classes sometimes cast their votes in favor of universal systems, sometimes in favor of selective systems? In his efforts to answer this question Rothstein first argues that people are less self-interested than usually assumed. "The willingness", he writes (p 221), "to act out of norms of solidarity is simply much higher and more widespread than the standard economic theory about human behavior predicts". The condition, however, is that they are not cheated – they want to know that the government, and their fellow-citizens, play their parts of the game in an equally fair way. They have to feel trust in those respects. "What they do not want to be", writes Rothstein (p 222), "is ‘suckers’, contributing when others are not." Thus, and again, trust is the crucial factor.

So, when is trust present, and when is it not? This question, I think, is most important, since the answer should tell us when, according to Rothstein’s explanation, we get universal policies, and when we get selective ones. But Rothstein’s answer, which is quite strange, fails to do so. What Rothstein basically says (p 223 ff), as an answer, is rather that universal systems enhance trust, whereas selective systems enhance distrust and suspicion. The truth of this answer can certainly be disputed, but I will leave that point till the next section. Here, where I am basically interested in the logical structure of Rothstein’s explanation, I will rather concentrate on the relevance of the answer. From this point of view the answer is quite strange since it places, as it were, the cause within the effect to be explained.

To be more specific I think that the trust, or the lack of trust, which is used for explaining the emergence of universal or selective policies respectively, has to be present in the system before the appearance of the policy to be explained. The reason, of course, is that the cause has to appear before its effect, since otherwise it cannot be a cause. Rothstein does not, however, say anything about the systems before the appearance of the policies he is interested in. Therefore he does not demonstrate what he claims to demonstrate, but rather something like the following:

First he demonstrates, that if universal policies for some reason emerge, then they are likely to remain. The reason is that universal policies, according to Rothstein, not only require trust for their appearance, but also enhance trust. Hence they are inherently stable. Similarly, Rothstein’s argument also seems to imply that selective policies, once in place, will remain there. These policies appear when trust is not present, and once in place they do not produce trust but rather distrust. Therefore these policies also seem to be inherently stable.

The implication of all this, of course, is that Rothstein has not explained why universal policies sometimes emerge, and selective policies sometimes. He has, at most, shown that if one or the other of these policies emerge, for whatever reason, it is likely to remain in place.

This, of course, is a major deficiency in Rothstein’s intended explanation, but it is not the only one. Another one is that he does not, in any way, link or attach his argument to the different nations and their properties. Therefore, even if his intended explanation was correct, he would at most, as his text now stands, have explained why there are differences. He would not have shown why, for instance, Sweden, has one kind of welfare system, and Australia has another. This deficiency is, of course, closely related to the first one. Since Rothstein places the causes within the effects he cannot, by definition, find any properties among the different nations which are independent of, and prior to, their welfare policies, and which might therefore explain the nature of these policies.

It may perhaps be argued that criticism is unfair since Rothstein in a few places talks about national properties apart from, and prior to, their welfare policies. Thus he writes (p 215) that "[m]y argument is that the solution to how this dilemma has been solved in different countries is related to the way in which the institutions of the welfare state programs have been historically established." After that there is however, unfortunately, not much talk about history - just a few references to the Swedish political history (p 229). So I think that my criticism above is fair and valid.

5.5 Some further comments on Rothstein’s explanation

Rothstein’s explanation of the differences in welfare spending between the Western democracies is, as I have just tried to show, a failure. The basic reason for this is that he fails to attach his main two types of outcomes, that is universal and selective policies, to independent, and prior, properties of the nations discussed. Therefore, he demonstrates at most, considering just his logic, that universal policies are inherently stable, and that the same is true for selective policies. This, however, requires that the material content of his explanation is correct - what he says about the properties of universal and selective policies has to be true.

Since the logic of his explanation is faulty anyway, I shall not spend many words on this truth-issue. It is however interesting to note that Rothstein seems to think that universal policies are better than selective policies in all respects. Thus, in the three sections (p 223 ff) "Substantial Justice", "Procedural Justice" and "The Just Distribution of Burdens" he has a lot to say in favor of universal policies, and a lot against selective policies, but nothing, as far as I have been able to find out, in the opposite direction. These sections, in fact, seem to me to be much more political agitation than intellectual analysis. The fact noted above, that Rothstein equates selective policies with no welfare policies at all, also testifies to his admiration for universal policies.

I wrote in part 5.2 that a main general problem with social dilemmas was to explain the occurrence of solutions rather than equilibrium outcomes. In his discussion about the welfare state as a social dilemma Rothstein turns this whole problem upside down. He stresses the good properties of universal policies, the bad properties of selective policies, the human tendency to be loyal and cooperative, and so forth, so intensively that we are, in fact, left wondering why any nations, ever, adopt selective policies. But we know that they do.

This, of course, is still an indication of the inadequacy of Rothstein’s explanation.


In the beginning of his paper Rothstein dismissed, we remember, explanations of the differences between the Westerns countries’ welfare policies in terms of standard political science variables, and he also claimed that the differences were surprising since all the countries are "Western democratic capitalist market economies" with "basically the same type of social, economic, and political structures". When reasoning so Rothstein did however not take constitutions into account – surprisingly, it may be added, since Rothstein is a political scientist. And certainly the constitutions of the Western countries, although they are all democratic, differ in other important respects. Perhaps some clue to the differences may be found here?

In diagram 3 the nations are classified according to their constitutions. What I have taken into account is the system used for electing the members of the parliament – whether it is proportional or majoritarian – and the method used for appointing the executive – whether it is parliamentary or presidential. As we see there are no countries with presidentialism and proportional elections. The reason is that this kind of constitution mainly is a Latin American phenomenon and therefore there are no such countries among the OECD countries studied here. It should also be noted that some countries, although they appear in the statistics studied here, nevertheless are absent from diagram 3. The reason is that their constitutions in some way or another are difficult to place in any of the four boxes in the diagram, for instance because they use some kind of mixture of parliamentarism and presidentialism. The countries omitted for such reasons are France, Germany, Ireland, Switzerland and Japan. The remaining countries thus are fairly clear-cut representatives for the constitutional types they are associated with in the diagram. In the diagram I have also indicated the types of social welfare system used by the different nations, as well as their expenditures for social welfare – this information is exactly the same as the one already given in table 1.

Diagram 3: Thirteen countries arranged with respect to their type of constitution, their type of social welfare system, and their social welfare spending as percentage of GDP. E stands for encompassing welfare systems, C for corporatist systems, B for basic security systems and T for targeted systems.

The diagram shows a considerable correlation between constitutional types and welfare expenses and institutions. All encompassing and corporate systems appear in countries with proportional elections and parliamentarism, and all basic security systems, except Denmark’s and the Netherland’s, and all targeted systems appear in countries with majoritarian elections. Furthermore all countries with proportional elections and parliamentarism have higher expenses than any of the countries with majoritarian elections. This pattern, I think, is far more distinct and clear than anything brought forward by Rothstein. Is the pattern, we should therefore ask, a mere coincidence or are there some plausible, systematic causes behind it?

I think that some interesting causes may be identified. Since I have written extensively about these matters in other places (Moberg 1998, 2000a, 2000b, 2001-) I will however just give a few hints about my thinking. First, I think, as I have said above, that the welfare goods are private, not public. Second, I think that people always, and in principle, can use the political system, or political decision making, for acquiring private goods which they otherwise would have to buy in the market. Politicians may also use private goods for "buying" votes. The practical possibilities in these respects do however depend on the nature of the constitution. The best possibilities for satisfying private interests, or special interests as they are sometimes called, are present in countries with proportional elections and parliamentarism. The basic reasons for this are twofold.

The regular pattern shown in diagram 3 may thus be explained in constitutional terms. The information provided by diagram 4 lends, I think, some further support to this point. Along the y-axis the diagram displays the welfare policy expenditures for the countries concerned as percentages of GDP – these are thus the same figures as those in table 1 – and along the x-axis all other, remaining public expenditures in 1985, also as percentages of GDP (these data are calculated from figures in OECD, 1985). The expenses for some really public goods, such as defense, are thus present here.
Diagram 4: Welfare policy expenditures, and the other remaining public expenditures, both as percentages of GDP, in 1985, for the countries studied.

As we see, there is no obvious or striking correlation between the two categories of expenses. A possible interpretation of this is that the "other expenses", which include expenses for public goods, are unrelated to the constitution, whereas expenses for private goods, such as welfare policy goods, are closely related to the constitution. In countries with parliamentary, proportional constitutions, these latter expenses are not only very high, but they are also much higher than the other expenses. Parliamentary, proportional constitutions thus seem to have a particular propensity to make the provision of private goods a public concern.


Rothstein starts his paper by formulating two problems. Both are about the Western democracies’ spending on welfare policies. One of the problems comes from the alleged empirical observation that there were no spending differences at all in the early 1960s, and that, after that, differences have appeared and widened dramatically. Rothstein’s finds this surprising since the ongoing globalization should have caused convergence rather than divergence. I have however tried to show that Rothstein’s initial observation is wrong. There were considerable differences in 1960, and, since then, we have in fact witnessed convergence rather than divergence, so there is not much to be surprised about.

The other problem concerns the reasons for the differences. Rothstein dismisses some rather conventional explanations, for instance in terms of the political parties’ ideologies, and sports a quite different explanation of his own in terms of so called social dilemmas. The idea is to consider the welfare state as a social dilemma which sometime could result in a good and cooperative outcome, sometimes in a bad and non-cooperative one. The coopertive outcome is characterized by universal welfare policies, and is also comparatively expensive. The non-cooperative outcome is, on the contrary, cheap and based on selective policies.

I have argued that this explanation is untenable for two reasons. First, and primarily, the whole explanation, as put forward by Rothstein, fails logically. The reason is that he does not attach the two types of outcomes to any independent, or prior, properties of the nations concerned. He therefore fails to show why some nations have expensive universal policies, and other ones have cheap selective policies. At most he shows that universal policies, if for some reason they appear, tend to remain in place, and similarly for selective polices. This, however, requires that his claims about the properties of universal and selective policies, respectively, are empirically true. That, I think, can be disputed, and that is a second, but secondary, reason for refuting Rothstein’s explanation.

But even if I refute Rothstein’s explanation I think that the whole problem about the differences in spending is interesting, and I have therefore suggested an alternative explanation. In spite of Rothstein’s remarkable claim that the Western democracies have "basically the same type of social, economic, and political structures", I think that there are important differences, for instance constitutional ones. My alternative explanation is thus in constitutional terms. It may be right or wrong, but anyway, and so far, I think it results in a much more distinct and clear pattern than anything brought forward by Rothstein.


Korpi, W & Palme, J, "The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries", in American Sociological Review, 1998, vol. 63, pp 661-687.

Moberg, E, 1998, "The expanding public sector - a threat to democracy?", in G Eliasson & N Karlson, ed., The Limits of Government - On Policy Competence and Economic Growth. Stockholm: City University Press. Also available on this website.

Moberg, E, 2000a, "The Swedish Model: A Comment on Mancur Olson's Analysis", in M Olson & S Kähkönen, ed., A Not-so-Dismal Science - A Broader View of Economies and Societies. New York: Oxford University Press.

Moberg, E, 2000b, Ideological Politics and Interest Politics, available on this website in the section "Arguments.

Moberg, E, 2001- , A Theory of Democratic Politics, available on this website.

OECD, 1985, Economic Studies, No 4, Spring 1985.

OECD, 1994, New Orientations for Social Policy. Paris: Organization for Economic Development, 1994.

Ostrom, E, "A Behavioral Approach to the Rational Choice Theory of Collective Action, Presidental Address, American Political Science Association, 1997" in American Political Science Review, March 1998, volume 92, number 1, pp 1-22.