The Political and Moral Logic of the Universal Welfare State PDF

THE LOGIC OP THE UNIVERSAL WELFARE STATE 145 6 Variations in social policy between countries The political and moral logic of...

THE LOGIC OP THE UNIVERSAL WELFARE STATE 145 6 Variations in social policy between countries The political and moral logic of Seldom does the real world behave according to the predictions of social scientific models. If we go back in time, three decades say, we find that the universal welfare state the proportion of gross national product spent by OECD countries on social programs did not vary especially much. Up until the early 1960s, Sweden was one welfare state among others, and its social policy was neither more comprehensive nor more generous that that in other comparable countries. Three decades later, however, the variation is much greater. Comparative developments in social policy have, one might say, followed the shape of a fan. Instead of convergence between similar countries, we have seen divergence. At the start of the 1990s, Sweden was distinctive in several ways. One was in the unusual breadth and comprehensiveness of its social policy. 2 The remarkable thing about this widening social policy divergence between countries is that it has occurred at the same time as these countries have experienced an In the last chapter, I called attention to how human behavior is governed increasing internationalization of their economies. Indeed, the overall both by narrow self-interest and by the social norms emerging in an political and social interdependence of these countries has increased open, reasoning discourse. Put otherwise, human beings have dual utility during the last three decades. This ought to have led, one might reason- functions and their behavior is strategic. What is the result of this in ably conclude, to convergence in respect of social-policy ambitions, but evidently it has not. concrete politics? And what does it mean for the future of the universal welfare policy? The institutionalist approach I use here builds on the idea The quantitative measurement of welfare policy variation mentioned of a two-way relation between institutions and behavior.' This means we above is obviously a very rough way of describing what distinguishes behind the appear- must explain both what social forces and factors lie Western capitalist democracies from each other. Several attempts have ance and subsequent reproduction of a universal welfare policy, and what been made, with varying degrees of success, to capture the qualitative effects on these social forces the universal welfare policy as an institution differences between welfare models as wel1. 3 One usual distinction, gives rise to. If we recall here the idea of dual utility functions, moreover, mentioned earlier, is that between selective and universal welfare states. two questions arise. Firstly, what types of institutions and social norms Some of the ways in which these two types differ are summarized in sustain a macro-institution like a universal welfare policy? Secondly, table 6.1. which norms and interests are strengthened, and which are weakened, by As Diane Sainsbury has shown, the Swedish welfare state approaches the manner in which this institution operates? the universal model, even if considerable departures from the pure form It would perhaps be advantageous, pedagogically speaking, to examine may be observed. One unavoidable feature of a universal welfare state, it the political and normative logics each in themselves. As I stressed in the should be noted, is a heavy tax burden. If benefits and services are to be last chapter, however, the whole point of the idea of dual utility functions made available to the entire population, and if they are to be of such a is that the political and moral logics are intimately connected to each other, and so must be treated in a single context. The willingness of citizens 2 Esping-Andersen (1990); Stephens (1996). 3 to assist in realizing the objectives of a universal welfare policy depends on See, e.g., Esping-Andersen (1990), who identifies three different types: (a) the their regarding both the substance and the process of the policy as correct. liberal welfare model, which coincides broadly with the selective model; (b) the This depends in turn on how the political institutions charged with the social democratic model, which is largely the same as the general model; and (c) the Catholic-corporative model, which is a sort of middle category. Like the making and implementation of collective decisions are designed. general welfare state, the last-mentioned type covers the entire population, but the programs are often divided up, so that different social groups make use of Or to use a different terminology, how the dialectic functions in this context. different services and social insurance programs. 144  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 146 147 Table 6.2. The redistributive effect of the universal welfare state Table 6.1. Dimensions of variation between selective and universal welfare policy Group Average Tax Transfers Income after Selective Universal income (40%) taxes and Dimension transfers Proportion of GNP devoted to social A (20%) Small Large 1,000 400 240 840 purposes B (20%) Minimal Adequate 800 320 240 720 Benefit levels C (20%) 600 240 240 600 Range of statutory benefits and services Limited Extensive D (20%) 400 160 240 480 Proportion of population covered E (20%) 200 80 240 360 Minority Majority by benefits and services Ratio between Means-tested Universal groups A and E 5/1 Dominant type of program (= 1,200) (1,200/5=240) 2.33/1 Role of programs preventing need for Nonexistent Critical assistance in first place Fees Taxes groups takes place. 6 Some economists have therefore claimed that the Financing Great Small welfare system amounts largely to a costly bureaucratic roundabout.' Significance of private organizations Minimal Optimal Nothing could be further from the truth. This very common standpoint Favored role of state direction is, in fact, among the great follies of modem social science. Table 6.2 303. illustrates why. The redistributive logic of the model is as follows. In the Source: Prom Sainsbury (1991), p. 4; cf. Korpi (1960), p. first column, income earners are divided for the sake of simplicity into five groups of equal size, according to average income. We assume the quality as to satisfy also the better-situated, then high taxes are quite average income of the group earning most is five times that of the group simply indispensable. Naturally it is incomprehensible for a libertarian earning least. This difference, which we may call the inequality quotient, that a people could accept a tax burden of 57 percent of GNP. The is 5/1. We further assume, nota bene, not a progressive but rather a obvious answer to this puzzle is that public acceptance of this state of strictly proportional system of income taxes. We set the tax rate at 40%, affairs depends on the view being widely held that the results thereby 4 which corresponds roughly to that part of the public sector's presently obtained stand in reasonable proportion to the sacrifices incurred. 56.2% of GNP that is spent on social, educational, and other welfare This means that, if we are to understand the political logic behind the policy. Finally, we assume that all public benefits and services are universal welfare state, we must think in the terms suggested by the universal, which means that the individuals in each group receive on classic Italian system of double bookkeeping. This system is based on the average the same sum in the form of cash benefits and/or subsidized idea of a relation between costs and revenues, debts and assets. One public services. The result, as seen in the last column, is a dramatic should bear this approach in mind when considering Swedish social reduction in inequality between group A and group E, from 5/1 down to policy. The Swedish tax system is now, after the tax reform of 1990, only 2.33/1. The level of inequality has thus been reduced by more than half in 5 modestly progressive (and not just in fact but also in form). Many this model of how the universal welfare policy works. Note that this scholars have maintained that, since benefits and services are distributed redistributive logic works in the same way if you take the groups' (or in roughly equal shares to everyone, and since the tax system is persons') income over a life-time as well as if you compare them at one proportional on the whole, then no real redistribution between income 6 The following builds on many discussions I have had with Peter Mayers, whose 4 Cf. Agell (1992), pp. 248f. ideas and suggestions have been most valuable. Mayers was the one who first 5 Even before 1990, the tax system's real progressivity was not so great. See Steinmo pointed out to me the redistributive effects in the model below. 7 Tullock (1983). (1993), pp. if.  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 148 149 single point in time. It is only if you can argue that over time the persons of redistribution depends, in other words, not just "on accuracy of aim in groups A and B will switch with the persons in groups I) and E that but also on the sums transferred," as Walter Korpi and Joakim Palme the redistributive effect decreases. write. 12 Soderstrom's calculation of the redistributive effect of the Nevertheless, it should again be stressed that the redistribution in the Swedish social insurance system shows that, if one compares the 20 model is achieved with no progressivity in the tax system whatsoever, and percent of households earning most with the 20 percent earning least, the with no targeting of benefits and services on (as the usual term goes) the inequality quotient falls, after taxes and transfers, from 3.75/1 to 2.43/1. It truly needy. Such scholars as Amy Gutman, J. Donald Moon, Peter bears noting that SOderstriim's thesis — that the Swedish welfare state has Baldwin, Brian Barry, Albert Weale, and Gordon Tullock, are thus largely favored the middle class and has had no redistributive effect — is mistaken when they claim that a universal welfare policy does not fully contradicted by his elvn data, which show that the top (in the sense 8 contribute in any significant degree to achieving social equality. I believe of economically strongest) 60 pecent of households are those which pay their mistake results from their not recalling the simple fact that both in more than they get back from the social insurance system. 13 incomes and expenditures must be counted when judging the net impact When, therefore, a philosopher like Brian Barry claims that this type of of all economic transactions. They have instead simply pored over the welfare state "has no inherent tendency to bring about... net transfers" fact that "everyone receives the same," which obviously involves no between different social classes, he has no empirical basis whatsoever for redistribution. What they have missed is that different income groups his view. 14 Still stronger support for the thesis that the general welfare part with different absolute sums, even if their tax rate is the same in model is, as compared to its selective counterpart, a superior machine of 9 percentage tenns. redistribution may be found in empirical studies comparing income The central message of Italian double bookkeeping — that where there distribution in different countries.'' Thanks to the data bank built up in is a credit entry there must also be a debit entry — has passed these the so-called Luxemburg Income Data Study, we now know a great deal authors by. When, as in this case, expenditures (taxes) arc equal in more about these matters than earlier. As shown in table 6.3, the variation proportional terms, while incomes (benefits, services) are equal in between countries in the area of income redistribution is rather great. nominal terms, considerable redistribution takes place between those One interesting conclusion to be drawn from table 6.3 is that a strong with high and those with low incomes. When Amy Gutman, for example, relation of the following type apparently obtains: the more universal the writes that a system of generally available social services is "the least likely welfare system, the greater the redistributive effect. It is precisely the to meet the material needs of the poor," she lacks both theoretical and countries in the lower half of the last column which are usually classified empirical support for her conclusion.' as selective welfare states. This, it must be stressed, is a paradoxical and The stickler for detail can now point out that many benefits in the counter-intuitive result. One might reasonably expect that the countries universal welfare state are not equal for all citizens, but rather are income-related, and that this limits their redistributive effect. In some 12 1 Korpi & Palme (1993), p. 148; Weale (1990). cases, such as education, the middle class clearly recieves more.' This is 13 Calculated from table 3A in Lars SOderstram (1988), pp. 83 and 144. The data true in theory but not in concrete policy. Even if one includes benefits of are from 1982 and include households the adults of which are between twenty this type, the redistributive effect is considerable. The reason for this is and sixty-five years of age. If households whose members are over sixty-five are that, in nearly all of the income-replacement programs typical of the also included, the redistributive effect is strongly enhanced. According to universal welfare state, there is an income ceiling over which no further information communicated orally to the author, neither consumption taxes nor employers' fees are included in the calculation of what the different groups pay in compensation is paid, but over which high-income citizens must con- tax. If these are included as well, the redistributive effect in all likelihood becomes tinue paying charges nonetheless (in the form of wage taxes). The extent much greater, Lars SiiderstrOm's report has sometimes been taken as evidence that the Swedish type of welfare state largely redistributes between different Gutman (1990), B. Barry (1990), Goodin (1988), Moon (1990), Tullock (1983). periods in the individual's lifecycle, and not between income groups. 14 9 That some philosophers and historians make such a mistake is noteworthy, but B. Barry (1990), p. 505. 15 that economists of the trade do so as well is striking indeed. See Lars Soderstrbm See Korpi & Palme (1993). This is also stressed by Soderstrom (ed.) (1988), (1988), p. 83. pp. 84f., who with a number of reservations nonetheless writes that: "higher 1° Gutman (1990), p. 11. public transfers contribute to a more equal distribution, even when account is 11 Davidson (1989) and Goodin & LeGrand, and associates (1987). taken of various types of repercussions."  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 151 150 ° and after taxes and Table 6.3. Ranking based on Gini-coefficients before The political logic of the welfare state transfers, and resulting net redistribution How, then, can we explain this difference in scope and character between Gini-coefficients after Net redistribution the welfare states of the Western world? As indicated in chapter 1, the Gini-coefficients before taxes and transfers (percent) research in this area is very extensive, and it reveals (not unexpectedly) taxes and transfers 53 that a range of factors is at work. There does, however, seem to be 0.385 Sweden 0.197 Sweden Norway 39 considerable empirical support for the so-called power-resource model, 0387 Norway 0.234 Norway Canada 38 which stresses the importance of the mobilization of wage-earners in Germany 0.252 Germany United Kingdom 0.393 37 trade unions and social-democratic parties. The stronger trade unions 0.407 United Kingdom 0.264 Netherlands Germany 35 are, and the greater the I3olitical influence of social-democratic parties, 0.414 Australia 0.287 France Switzerland the more comprehensive and universal the welfare policy pursued. 18 The 0.414 Canada 0.293 United Kingdom 33 Australia 31 problem then remains of explaining why some capitalist democracies 0.417 Netherlands 0.293 Australia Sweden 25 have stronger social-democratic parties and higher rates of trade-union 0.425 France 0.307 United States United States 24 organization than others. Is there a connection, perhaps, between the 0.467 United States 0.317 Canada Netherlands 19 policies pursued and the strength of labor organizations? Have policies 0.471 Switzerland 0.336 Switzerland France been enacted, that is, that favor the appearance of a strong social democracy and strong trade unions? The answer to this question is a The Gini coefficient is a standard statistical measure of inequality. - affirmative.' 9 As is usual in the social sciences, the relation between these Source:Castles & Mitchell (1992), p. 22. See also Fritzell (1991), p. 148. elements is not one-way, but two-way and dialectical. Karl Marx's theory has played many tricks on the labor movement, one of them being the mistaken prediction that the development of the concentrating their assistance on "the truly needy" — where the state literally takes from the rich and gives to the poor — would be those productive forces would lead, on the one hand, to the homogenization of achieving the greatest redistribution. Yet this is evidently not the case. It the working class, and on the other, to this class becoming a numerical is rather the countries that take from "everyone" and give to "everyone" majority. It did not turn out that way, of course, and in the time since the that exhibit the most marked redistribution. This is partly for the reason Second World War, social-democratic parties in the West have had to illustrated in the model above — that is, taxation is largely proportional, wrestle with the fact that their traditional voting group has stagnated and while most benefits are either flat or have a ceiling (i.e., they match thereafter diminished in size. It is instead the white-collar groups which incomes only up to a certain limit). The p9litical factor is also have rapidly grown. In the research on why different countries have different welfare important, however. Walter Korpi and Joakim Palrne put the point this way: "If we take from the rich to give to the poor, the rich simply will arrangements, the relation between the type of programs social-democratic 16 not part with especially large sums. " Or to paraphrase what Lyndon B. parties have established, and the electoral support they have enjoyed, has Johnson said, soon after becoming President, in explanation of why he come into focus. 2 ° According to this theory, social-democratic parties did not dismiss FBI chief J. Edgar Hoover: it is better to have the sharp must — if they are to attract a sufficient number of voters to achieve office elbows of the middle class on the inside of the system pressing it — design welfare programs so as to favor not just the (diminishing) 17 working class, but also that (increasing) part of the population working outwards, than the other way round. The critical thing for whether or not the welfare state redistributes, contrary to what one might think, is in various white-collar and service occupations (those termed in class that the system embraces not just the poor but also (and above all) the analysis the middle strata or the middle class). well-to-do. 18 Esping-Andersen (1990). 19 This theory was first launched by Przeworski (1985). For an empirical illustration, see Rothstein (1992c) and Svensson (1994). 16 Korpi & Pal= (1993), p. 148. 2 ° For a more extensive discussion and analysis on this point, see Rothstein (1990). 17 From O'Higgins (1987), p. 14.  JUST INSTITUTIONS MATTER THE LOGIC OP THE UNIVERSAL WELFARE STATE 153 152 This, however, has confronted social democratic electoral strategists of the Social Democrats to accept or even embrace the ideology of the with a difficult dilemma. They can, on the one hand, limit themselves to universal welfare state. 2 The first thesis, then, is that a universal welfare traditional working-class questions; this makes it difficult or impossible state can only exist if it enjoys support far up the social ladder. The to win office, but it also reduces tensions within the labor movement and "poor," the "underprivileged," the "working class," or any other such enables the party to retain a traditional working-class profile. Or they can social group is simply too small to constitute a sufficient electoral base for try to broaden their electoral base by launching a welfare policy designed a comprehensive universal welfare policy. And conversely, one can only in such a manner as to attract not just workers but white-collar employees reckon with support for this policy from white-collar groups and the as well. This latter strategy has the advantage of making it possible to middle class if it is so formulated as to serve their interests as well attract a sufficient number of voters to take power. However, it also (however these are defined). entails the risk of creating tensions inside the party, and of being rejected The critical political plisition occupied by the middle groups is also as non-solidaristic by the party's traditional working-class voters, who 2 apparent if we take another look at table 6.2. Group C, the middle group, might then abstain from voting or in other ways supporting the party. ' is central here for two reasons. The model shows, to begin with, that for However, the severity of this dilemma varies between countries, this group, the universal welfare policy is largely an economic roundabout. according to the composition of the class structure and the organization That is, they pay exactly as much in taxes as they get back in the form of of the labor movement. This, according to the theory, explains why social services and benefits. They neither gain nor lose economically.' This democratic parties have enjoyed varying degrees of success in different group could therefore be expected, on the assumption of rational utility countries. This theory is interesting in that it does something relatively maximization, to exhibit more uncertainty in their view of the universal rare in social science, namely combine structural factors (class structure), welfare policy than any other group. Yet this group may — even institutional factors (working-class organization), and actor-level factors considering the matter from a rationalistic standpoint — be won for a (choice of strategy in the form of social programs) in a dynamic universal welfare policy, under two conditions. process. 22 The first condition is that the members of this group consider the I shall not attempt to test the tenability of this theory here. I take it, value they get back to correspond tolerably well to the value they pay in. rather, to be a largely correct explanation of the two-way connection Be it granted, such a calculation may be exceedingly difficult for the between the structure of the social democratic welfare state and its individual to perform; still, let us assume that these persons perceive this electoral support. For further evidence, I refer the reader to the literature connection, and possess a certain ability to calculate rationally. If they cited. I would like rather to emphasize the point found in the research. consider the public services to be of low quality, or inaccessible, or of The Swedish Social Democrats seem hold the world championship in otherwise unacceptable character, one may expect a diminished will- handling this dilemma. They have succeeded in striking the best possible ingness on their part to support (i.e., pay for) the universal welfare policy. balance between blue-collar and white-collar interests; they have struc- This means, in political terms, that they will withdraw their electoral tured their various social programs in accordance with these post facto support from parties pursuing such a policy. According to the theory of handbooks in reformist political strategy. They have, in general, suc- contingent consent, moreover, the same thing can happen if the universal ceeded in framing social policy in such a way as to gain the support not welfare policy does not deliver what it promises, procedurally speaking, just of the party's traditional working-class supporters, but of white- or if there is a widespread belief that the various systems are abused. collar employees and the middle class as well. This two-way process has The other reason of a rationalist nature for this group to support a been thought to function like a perpetual motion machine, in which the universal system, despite gaining nothing from it economically, has to do social-democratic-governed state has built up a universal welfare policy, with its insurance character. For one not wholly sure of what the future which in turn has generated a continued increase in support for the party 23 Cf. Uddhammar (1993). (despite the fact that its traditional electorate has diminished). In an 24 Soderstrom (ed.) (1988) shows that the middle groups lose out as a result of the alternative interpretation, this process has forced the parties to the right social insurance system. However, in view of the fact that they make more intensive use of the educational system, publicly financed culture, and municipal a Przeworski (1985), Przeworski & Sprague (1986). day-care than the low-income groups do, they probably end up roughly at a 22 Svensson (1994) shows how this has functioned in the Swedish case. balance of plus minus nought.  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 154 155 has in store, there can be cause to support such a system, even if one will strategists have regarded the support of middle groups as altogether 26 not gain by it in a statistical material sense. In the event of unforeseen critical, and have consciously designed their policies accordingly. economic difficulties occasioned by unemployment or illness, it is However, this theory of why different countries pursue different probably much more agreeable to be taken care of by a universal than by welfare polices, and exhibit varying degrees of social-democratic strength, a selective welfare system. For the latter requires recipients to sell off all presents us with some problems. For one thing, if the Swedish Social their assets before receiving any help, and it only offers such assistance as Democrats have been so infernally skillful in building this cross-class suffices to maintain a minimum level. To be forced, on account of support base for their welfare policy, then why, in the election of 1991, temporary economic difficulties, to sell off large portions of one's assets did they turn in their worst electoral performance since 1928? In both may in fact be something which the middle class in particular finds 1985 and 1988, the party had presented itself as the main guarantee for extremely disagreeable. It can also make it harder to solve the situation or the preservation of the welfare state, and proposals from the bourgeois to return to working life. parties for (very modest) cutbacks gave the Social Democrats a great The other reason for the centrality of this group is that, if rationalist advantage in the election campaign. 27 Did the Swedish voters suddenly assumptions about voting behavior as founded in self-interest are correct, tire in 1991 of being cared for by the state "from the cradle to the grave?" it is the electorally decisive group. Groups D and E will obviously support Or did the Social Democratic strategists miscalculate in some way? Either the universal welfare policy, according to the rationalist viewpoint, since something is basically wrong with our theory of the relation between the they gain from it materially. Similarly, groups A and B will oppose such a welfare state and its electoral support, or the party strategists did not system, as they pay more in taxes than they get back. (They could support study the handbooks closely enough this time.' I shall examine this the policy anyway, of course, but for other reasons.) It is group C that question more closely in chapter 7, where I shall attempt to give a determines the direction of policy, for its votes determine which side tentative explanation for this paradox. holds the majority. The middle group is therefore of absolutely critical Secondly, the theory under consideration proceeds on the premise — political importance for the maintenance of a universal welfare policy. In which earlier we examined and found wanting — that political behavior is part, this is because it is economically the most sensitive group for how founded in self-interest. It paints a harsh picture: policy is formulated so the system works; in part, it is because it is electorally the decisive group. as to serve the self-interest of certain groups, and therefore attracts their One could say, to simplify a little, that in those countries with a universal electoral support. Voters smile, in other words, on policies that grant welfare policy a political alliance between groups C, D, and E has been them economic advantage, and they do not allow matters of morality or successfully forged. In the countries pursuing a selective policy, by the common interest to disturb their calculations. contrast, the majoritarian political constellation consists of groups A, B, One reason the universal welfare state cannot be explained solely on and C. The simple mathematics of democracy tells us that three beats the basis of narrow rationalist assumptions is the following. In all of the two. Scandinavian countries, bourgeois parties have held the reins of power Our model would lead us, then, to expect political alignments of this for relatively long periods during the past two decades. Notwithstanding kind. And, in fact, this expectation is confirmed in the empirical research this, one cannot in any of these countries during the periods of bourgeois done on the Swedish welfare state's support base and design. Stefan governments discern a reduction in welfare expenditures or a change to a Svallfors' study (which is based on interview data) of the support enjoyed more selective social policy. If the universal welfare state has been the by welfare policy in different social groups largely confirms the assump- most significant factor behind the build-up of the parliamentary strength tions of our model: the higher up the social-class ladder, the thinner the of the social democratic parties, why then have the bourgeois parties in support for the welfare state. It is, moreover, precisely the gaup in the Scandinavia not been more successful in breaking up this system and middle that vacillates the most, ideologically speaking, and whose support replacing it with a more selective one? 29 As Paul Pierson has shown, even 25 for social policy is most conditional. In respect of the welfare state's 26 Svensson (1994). 27 design, moreover, Torsten Svensson has shown that Social Democratic 28 Bergstrom (1991). I am aware, of course, that there were many reasons — over and above the design of welfare policy — for the Social Democratic defeat in 1991. Cf. Rothstein (1993a). 29 25 Svallfors (1989 and 1991). Cf. Rothstein (1993a).  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 156 157 the Thatcher and Reagan governments got into problems with the "fair- reasoning can be assumed to be of especial importance for individual ness issue" when trying to cut back welfare programs.' behavior when the rational calculation of utility yields an unclear result. One way to keep the analysis within the "rational self-interest" If we are to understand this matter, then, our analysis must take account approach would be to argue that the middle and (to some extent) the of both rational and moral behavior. Before undertaking this analysis, upper classes support a universal system because they overestimate the however, I would like to stress one point. The discussion below attempts risks of becoming unemployed, etc." That would mean that even if it to ascertain - at the level of principle — the moral logic underlying support does not in economic terms pay back for them the self-interest model for (and opposition to) various types of welfare policy. As mentioned would still suffice to explain their behavior. My argument against this is above, the actually existing welfare system is founded, in all countries, on that they would, being rational and risk adverse, then try to organize their a mixture of principles. For instance, even if Swedish social policy may be social-insurance system and their need of social services separately on a characterized as largely universal in character, it exhibits many significant class or professional-group base instead of taking on the costs for departures from this principle as well. The discussion below therefore supporting the lower social classes that, in a capitalist society, pay less treats not the moral logic underlying Swedish social policy as such, but 32 taxes and are likely to need more support. There is thus strong reason rather some prominent features of this policy. to question whether self-interest is a sufficient explanation for political behavior (whether of politicians or of voters). The universal welfare state Substantive justice embodies a moral as well a political logic. What does the former look like? As mentioned above, the first condition of contingent consent has to do with the question of general fairness, or otherwise expressed, the problem of substantive justice. That is, can one argue that the goals of a particular The moral logic of the welfare state social-policy measure are just? We can find the answer to this question in I tried in the last chapter to show that social norms can be a function of the discussion of what the state should do in chapter 2: the state should institutions, as well as of the other way round. For instance, a norm that treat all citizens with "equal concern and respect," and it should furnish says state power is basically illegitimate weakens the ability of the state to them with "basic capabilities" so as to enable them to make autonomous function (to safeguard the common interest, for example), which in turn choices. The first principle lies, we might say, at the heart of a universal strengthens the norm of state illegitimacy, and so the cycle continues. welfare policy." Indeed, the whole point of a universal welfare policy is Now then, if social norms have the importance we have ascribed to them, not to discriminate between citizens, not to separate "the needy" and we must try to grasp the logic linking these norms with an institution like "the poor" from other citizens and to treat them differently. Social policy the universal welfare state. should seek instead to furnish citizens with basic capabilities according to I shall build here on the theory about consent, which I presented in the principle of equal concern and respect, thereby placing them on a more more detail in chapter 5. This theory seeks to explain how the greater or less equal footing in respect to their ability to act as autonomous number of citizens can choose to support a universal over a selective citizens. social policy, despite it not lying in their direct material self-interest. I In contrast to the situation under a selective system, the public wish to emphasize again that arguments of a moral nature may be discussion of welfare policy in a universal system cannot be conducted in assumed to carry the greatest weight with those in the middle group the terms indicated by the question: "what shall we do about these above, because their calculation of utility in the face of the welfare system deviant groups/individuals?" (Or as former US Vice-President Dan yields a balance of plus minus nought, at the same time as their electoral Quayle put it in a debate: "those people." 34 ) support is decisive for who gains a majority. I assume, then, that moral In chapter 4, I called attention to the limits of public policy, and pointed out that the state cannot do everything. I stressed, furthermore, 3°Pierson (1996), p. 102. that such limits are particularly evident when it comes to implementing 31About the theory of the risk society, see Beck (1992). 2This is an argument against the otherwise convincing attempt by Moene & programs of a "dynamically interventionist" type. The idea of basic ' Wallerstein (1996) to explain the difference between selective and universal n Titmuss (1967/1987). 34 welfare systems based on a model of self-interested voters. Quoted in Katz (1989), p. 236. 158 JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 159 capabilities is in line with this. On the one hand, the state should supply are not to blame for their predicament (and so cannot legitimately claim all citizens with these capabilities; on the other, the state cannot have an assistance). We may refer to the first as the general and the second as the all-embracing responsibility for citizens' well-being. Those capabilities 35 not individual boundary-drawing problem. Contrary to the claims of such defined as basic are, in principle, the citizen's own headache. The philosophers as Gerald Cohen and John Roemer," there is often (as I precise content of the notion of basic capabilities obviously varies, both showed in chapter 4) no practical possibility of solving these problems at over time and according to political ideology, and it cannot be definitively the level of implementation. In the selective model, the discussion often established on the basis of political philosophy. As a general matter, comes to focus on "the undeserving poor." 4° In other words, an however, one can say that such capabilities are needed as enable citizens 36 unending debate is set in motion about how and where to draw these two to take part as free and equal partners in the political discourse. The boundary lines. It is therefore hard to argue that selective programs are critical thing is that we bear in mind what the debate over state fair, for the simple reason that they are not universal. Public consent to responsibility should treat with, and that we formulate a principle for the system is undermined, rather, because the social policy debate comes how these capabilities ought to be distributed. to turn not on what is generally fair, but on what is specifically necessary My point is that, depending on the institutions we select for furnishing (for the deviating individuals needing the help of the state). 41 Moreover, citizens with these basic capabilities, we create different types of moral since the majority can all the while question the way in which the two logic in the social-policy discourse. In the case of a selective policy, the boundary lines are drawn (since this can almost never be done with state separates out those citizens unable to provide such basic capabilities sufficient precision), the moral logic of the discourse tends to undermine for themselves, and furnishes them with said capabilities. To do this, the legitimacy of the system. Most selective types of policies that are however, it must first determine whether or not they belong to the needy structured to integrate a specific group with the rest of society, seem to group, and if so, how much they need. The problem, as I shall show entail a paradox of the following kind: to motivate selective measures of below, is that it is very hard to do these things without violating the whatever kind for d certain group, this group must first be singled out as principle that the state should treat all citizens with "equal concern and different from ordinary citizens. But if the group is that different, how respect." The very act of separating out the needy almost always stamps can they ever by any social policy initiative become like "ordinary them as socially inferior, as "others" with other types of social character- citizens." If the selective policy has only marginal effects, the usual 37 istics and needs, and results most often in stigmatization. Michael strategy for those advocating it is to argue that the group is even more Walzer argues that social policy of this sort is incompatible with the different (and thus has even more special needs) than what had initially maintenance of recipients' self-respect. And persons lacking self-respect been presumed. are scarcely able to act as autonomous citizens in the sense described in Under a universal system — in which the state furnishes all citizens with chapter 2. 38 these basic capabilities — the moral logic is altogether different. For the The public discussion of social policy in a selective system often distribution of basic capabilities is now designed in such a way that it becomes a question of what the well-adjusted majority should do about cannot violate the principle of "equal concern and respect." Since the the less well-adjusted or, in varying degrees, socially marginalized minor- universal welfare policy embraces all citizens, the debate assumes quite ity. The system's general fairness can thereby come under question by the another character: social policy is now thought to concern the entire majority, who might start asking (a) where the line between the needy community, and the question becomes what, from a general standpoint, is and non-needy should be drawn, and (b) whether the needy themselves a fair manner in which to organize social measures. No discussion of the type above — concerning how and where to draw the two boundary lines " This does not exclude the possibility that a universal welfare state can take — need ever take place, for the simple reason that no such lines need be responsibility also for matters above and beyond the resources defined here as 39 basic, but it need not then necessarily follow the principle of "equal concern and See ch. 2. " Katz (1989). respect." 41 StiderstrOm (1994) proposes the introduction of a selective system. They write 37 that: "The public commitment could be limited to some form of high-cost 36 Sen (1982), pp. 364ff. Salonen (1993), pp. 176-180. Walzer (1983), pp. 277f.; d Moon (1990), p. 35 and D. Miller (1978), p. 19. One protection for those groups unable to obtain satisfactory solutions within the can always, of course, consider this a reasonable price to pay for a lower overall framework of the insurance and credit markets." Unfortunately, however, they do tax level, as does Borg (1992). not discuss the problem of stigmatization.  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 161 160 drawn. Welfare policy does not, therefore, turn into a question of what (within the limits of such programs). The difficulty of finding usable should be done about "the poor" and "the maladjusted," but rather a criteria for selecting recipients can often become unmanageable. This question of what constitutes general fairness in respect to the relation creates a "black hole of democracy," in which citizens find themselves between citizens and the state. The question becomes not "how shall we faced with an administration or system of rules which no one really solve their problem?" but rather "how shall we solve our common understands, and in which no one can be held responsible. We may problem (health care, education, pensions, etc.)?" As Hugh Heap has put accordingly conclude that, from the perspective of legitimacy, universal it, the best way to help the poor is not to talk about them, that is, not to (i.e., broadly targeted) programs are to be preferred to selective (narrowly 42 set up programs targeted specially on them. The delineation of certain targeted) programs. In other words, it is better to have a state that does a citizens as "poor" entails portraying them as different in nature from little well than a lot badly. other citizens, and therefore violates the principle of equal treatment. The difficulty of handling the discretionary power of administrators in And in deciding how much such persons need, the state assumes a selective programs has two important consequences. These consequences paternalistic role, which cannot be done while respecting the integrity of are often thought to be opposites, but in fact they are two sides of the the individual at the same time. same coin. They are the bureaucratic abuse of power, and fraud on the part of clients. Applicants in a selective system have a natural tendency to claim their situation is worse than it actually is, and to describe their Procedural justice prospects for solving their problems on their own as small to nonexistent. Condition number two concerns, as mentioned earlier, the implemen- The administrators in such a system, for their part, tend to be suspicious tation of policy. Can welfare policy be carried out in a fair manner? This of clients' claims. (They are often pushed in this direction by their is the question of procedural justice. In addition, the successful operation superiors, moreover.) Empirical studies show, interestingly enough, that of a welfare system presupposes that citizens believe its proudly pro- in counties with selective systems, the authorities devote great resources claimed aims are actually implemented in the intended manner. This is to checking up on clients and front-line officials, while this is a non- the question of what the state can do. question in countries with universal systems, even in their means-tested How does the choice of a universal or selective welfare policy affect the programs.43 public's view of state capacity? To begin with the former: one should bear The question of procedural justice looms large in selective systems, in mind that a typical universal welfare program — like flat-rate pensions which tends to undermine legitimacy. Even if cases of cheating, fraud, or child allowances — is a great deal simpler, cheaper, and easier to and the abuse of power are in fact relatively rare, we now must live with implement than its selective counterpart. This is largely because there is the logic of the mass media, which means that it is just such cases that no need, in a program of a universal type, for an administrative apparatus will receive attention in the public discussion of social policy. This charged with carrying out the two types of 'eligibility tests which are a undermines public support for the welfare state. Moreover, the selective necessary concomitant of a selective program: for ascertaining (1) model leads, as even one of its champions — Robert Goodin — has been whether a given applicant is entitled to support, and (2) if so, to how forced to admit, to "unavoidable," "insurmountable," and "insoluble" much. In general programs, eligibility criteria can often be framed so problems in respect of the arbitrary treatment of citizens seeking simply (e.g., age) that the process can be automated. Social policy can assistance." The research on policy implementation has highlighted the thus be given the form of specified citizen rights, and the social duties of difficulties of exercising democratic control over the fate of clients in the the state can be rigorously defined. Such an institutional arrangement implementation stage. Suspicion of clients and the abuse of power are respects the integrity of citizens more than any beautiful (but vague and 43 non-binding) official declaration of social policy objectives could ever do. Weatherly (1991). In a conversation about an upcoming research report, As shown in chapters 3 and 4, selective programs present serious Weatherly said that in countries with a selective social policy, control and supervision of clients is a significant question, and such tasks take up a large part problems of implementation, for they allow administrators a wide field of social workers' time and attention. By contrast, in countries with a largely for discretionary action. There is often no solution to this problem general welfare policy, like Denmark and Sweden, this is to a great extent a non- question for social workers. 44 42 Quoted in Skocpol (1991), p. 430. Goodin (1988) pp. 219f. 162 JUST INSTITUTIONS MATTER THE LOGIC OP THE UNIVERSAL WELFARE STATE 163 built into the selective welfare model. The possibility of applying such a On the basis of figures supplied by the Association of Local Authorities model in accordance with the principle of "equal concern and respect" on the number of employees in various occupational categories, I have does not exist in practice. It is impossible to combine means-testing with calculated that, between 1975 and 1984, the number of persons opera- this principle, for means-testing itself entails a violation of citizens' tively taking care of children (so to speak) in the municipal day-care integrity — either in the means-test itself, or in the verification checks system increased by about 100 percent. Those involved (in various ways) which often follow. in administering the system's queue increased, during the same period, by In order to solve these "unavoidable," "insurmountable," and "in- 270 percent. In view of the low esteem in which "administration" is held, soluble" problems, then, Goodin offers an interesting solution. The according to the studies which have been done (see table 6.4), this would bureaucratic arbitrariness and uncontrolled interference into the lives of also seem to undermine the legitimacy of needs-tested public programs. clients, to which a selective social policy gives rise, can be moderated by instructing front-line officials to apply a certain method. When it is The just distribution of burdens unclear whether or not a client's claim satisfies the requirements of the rules, administrators are to be generous, and are to give the applicant the Condition number three in the theory of contingent consent has to do 45 benefit of the doubt. I am convinced, however, that such a policy would with whether or not all citizens bear their share of the costs of a given severely undermine the legitimacy of the system. The mass media and the policy, that is, it concerns the just distribution of burdens. Citizens are popular mythology would be full of stories about how clients with highly portrayed here as players in a so-called insurance game: they are prepared doubtful claims succeeded in reeling in benefits anyway, and the majority to support the program in question — even if they cannot be sure they of citizens would conclude that the requirement of procedural justice is will themselves directly gain by it — as long as they can be convinced that not fulfilled. The selective policy leads, then, to an "administrative all (or almost all) other citizens will also contribute to carrying it out. nightmare," 46 which in turn undermines the legitimacy even of this The willingness to contribute depends, that is, not just on the fulfillment li mited welfare model. of the requirements of procedural and substantive justice; it also assumes This can be explained, in game-analytical terms, as an effect of the so- a credible organization of the collective efforts (so that such efforts are, in called control game. It would be best both for the official and for the truth, collective). The other side of the question, of course, is how to client if they could trust each other: the official would avoid having to discourage the unsolidaristic use of the benefits the welfare policy brings. check up on the client, and the client would be left in peace. But since the One can also express this in the following way: citizens who wish to client (in Goodin's proposed system) can assume that the official will contribute to the common good are only willing to do so if they do not grant him the benefit of the doubt, he has a strong incentive to try to believe others will take undue advantage of their solidarity. A minority obtain benefits to which he is not entitled. And since the official can will not behave solidaristically if the majority is unsolidaristic. It is reasonably assume, after having experienced a number of such "games," heroic, but meaningless, to be the only one who defends the country. that the chances are great that the client intends to cheat, he will consider There is a certain threshold that must be crossed: citizens must be it necessary to carry out control checks. The result is an eternal round- persuaded that others will also contribute before they are willing to pitch about; the actors agree that it would be best to avoid having to administer in themselves. control checks or to submit to them, but neither one can avoid this The universal model differs from the selective on this point as well. outcome. Expressed in game-analytical terms, the actors lack dominating Typical for the latter is that assistance is granted only to those citizens strategies, and since there is therefore no point of equilibrium, the game who cannot in some other way provide for themselves or meet their lacks a solution. 47 "basic needs." This means as a rule that such citizens have no income, It also bears mentioning that needs-testing is administratively costly. and therefore pay no tax. They constitute a category, then, which does not contribute economically. In sum, to the extent that the welfare system is designed so that even 45 Goodin (1988), pp. 220f. 46 The expression is Hans L. Zetterberg's. See his article in Dagens Nyheter, net beneficiaries can play a role as partners who contribute, according to 09-20-90, p. 2. Cf. Pierson (1996), p. 105. their ability, to the defraying of costs, the legitimacy of welfare policy will See further Hermansson (1990), pp. 118f. increase. It becomes a question of how citizens shall undertake to solve  JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 164 165 their common problems, rather than a question of what "we" shall do When asked, "If you could be guaranteed that increased government 48 spending would be efficiently and effectively used to address society's about "them." problems, would you agree to an increase in your taxes?", the vast We observe here, then, two wholly distinct moral logics. The difficulty of implementing selective programs in such a way that their objectives are majority of people I interviewed, even the Americans, answered "Yes!" attained, and their processes considered fair, undermines public support The respondents, especially the Americans, quickly add, however, that for social policy in general. For example, the majority might be open to they do not believe that revenue from higher taxes would be used supporting social policy in principle, but constant reports of cheating, efficiently or effectively and therefore they would not approve tax increases. 52 fraud, bureaucratic abuse of power, waste, inefficiency, and other irregu- larities lead to their taking the view that the policy's implementation is so In interviews with trade-union leaders and Democratic Party representa- deficient as to make the whole affair a waste of time and money. It is very tives in the United States, moreover, Steinmo asked whether they would hard to imagine, moreover, that a population with such a negative view consider trying to introduce a value-added tax of a Swedish type, in order of welfare policy would be receptive to proposals to give it a more to raise revenue for the kind of social program which political groupings universal form (for this would involve expanding social policy). Instead, a of this type in the United States generally advocate. The answer, according suspicion of state measures — to which Nathan Glazer, among others, to Steinmo, was a unanimous "No." "In each case the respondent 49 gives clear expression — becomes the dominant attitude. A state that explained that given the way in which politics worked in the United fails to take care of "the poor" cannot of course be entrusted with the States, they would not trust the system to use the increased tax borne by larger task of attending to the welfare of the entire population. Citizens the poor to pay for programs that benefited the poor." Or as a high-level are more willing, on the other hand, to agree to collective undertakings if senatorial staff member answered: "This is not Sweden. How can we be 54 they have confidence in the state as an institution. sure that that extra money won't just be used to cut the taxes of the rich In a universal system, by contrast, this suit of discussion does not even more, or to buy more B-1 bombers?" 53 Steinmo explains the arise, because needs-testing is either unnecessary or can be done so difference as reflecting "the faith in the legitimacy of the state's role in 51 simply that it can be automated. The simplicity of universal programs, society" in the Swedish case. 54 Glazer's observations about the lack of the clarity they facilitate in the matter of who bears political responsi- faith in state legitimacy in the United States seem, then, to have some bility, and the possibility of organizing them on the basis of specified empirical basis. rights — these qualities enable such programs to fulfill (and to be seen to In the popular Swedish debate, many good examples have been given fulfill) the requirement of procedural justice. Sven Steinmo's comparative of how individuals have tried to take undue advantage of an excessively study of tax policy in the United States, Sweden, and the United generous system of income compensation." Healthy persons have Kingdom confirms this picture. Steinmo, who interviewed a large declared themselves ill, those able to work have pretended not to be, etc. number of centrally placed actors in the three' countries, maintains that The good intentions behind high compensation levels have sometimes these actors would make the same sort of choices were they faced with come to naught in the stage of implementation: generosity has invited similar institutional settings. But of course they are not: abuse (i.e., unsolidaristic behavior). Even if these examples are few (but they are telling), and even if the economic consequences of fraud and "it was foolish overuse maybe have not been so great in relation to total costs, such Even Glazer (1988) argues, on p. 34, for this type of universalism: behavior damages, in a fundamental way, the values of solidarity which... to have a system in which the welfare poor received free medical care, while those who worked at low paying jobs generally did not have medical insurance - originally provided the foundations of the welfare state's legitimacy. and lost their insurance when they lost their jobs. This alone made welfare more Moreover, we must now live with the logic of the mass media, which attractive than work." means that individual cases of cheating are brought into public view and 49 See ch. 4. 50 Levi (1997); cf. Levi (1993). are highlighted. This leads easily to the prevalence of the view that fraud 51 One must distinguish here, as I indicated earlier, between needs-testing is the rule, not the exception. Unfortunately, if this view — that "everyone undertaken in order to establish the client's ability to pay (which is in focus here), and needs-testing performed by professionals, as when a doctor in the 52 Steinmo (1993), p. 199. 55 public health care system seeks to establish what kind of medical attention is Ibid., pp. 199f. My italics. " Ibid., p. 207. 55 required. (The latter type of test is not at issue here.) Isaksson (1992). 166 JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 167 else" cheats — gets established by the mass media as a truth, this in itself Table 6.4. Attitudes towards public expenditure leads to increased cheating (or, as it is also called, "concealing"). As Laurin has shown in his investigation of the extent and causes of tax Year 1981 1986 1992 fraud in Sweden, most people say they are willing to act solidaristically Health care +42 +44 +48 (i.e., to pay their taxes) on the condition that others do so as well. Support for the elderly +29 +33 +58 Laurin's study also indicates that the beliefs citizens hold about the Support to families with children +19 +35 -1-17 morality of other citizens has a greater impact on their preparedness to Housing allowances —23 —23 —25 try to withhold taxes than does their estimation of their prospects for so Social assistance —5 —5 —13 doing without being caught. The conception of which social norms Primary and secondary education +20 +30 +49 prevail, in other words, seems to be more significant in this context than Employment policy +63 +46 +55 self-interest. 56 State and municipal administration —54 — 53 —68 One consequence that modern social science has brought is that citizens are asked now and again about their attitudes towards various Note: The material in this table comprises answers to following question: matters. So also with welfare policy. How does the empirical evidence "Taxes are used for various purposes. Do you think the revenues spent on look, then, in relation to the theory of contingent consent? Can empirical the purposes mentioned below should be increased, held the same, or support be found for the proposition that, if the institutions of social reduced?" The figures in the table represent the percentage of those wishing policy are structured according to the principles of this theory, they will to increase expenditure minus the percentage of those wishing to reduce it. create norms forming a basis for the reproduction of the policy? Some Source: From Hadenius (1986), p. 85; Svallfors (1991 and 1992a). opinion surveys include questions of a general nature about what the Swedish people think about the public sector as a whole. Without specifying whether fighter planes or day-care centers are intended, they basis of the Swedish welfare model in popular opinion lacks empirical ask whether the respondent wishes to expand or reduce the public support. The goals seem still to be held dear, at least by the population. sector. 57 The problem with this approach is that it does not capture the Another surprising result is the marked and stable difference in dimension — which is of particular interest to us here — of how much support for different types of programs. Support for the universal support the various types of programs enjoy. Axel Hadenius and Stefan programs is unambiguouly strong and stable, while the opposite is true Svallfors have done investigations of the required sort, however. They for the three selective programs. Two programs in particular are weakly have asked identical questions of representative samples of the Swedish supported — social assistance and housing allowances. 58 And both are population in 1981, 1986, and 1991. The results may be seen in table 6.4. dearly selective. At least two results of these studies are worthy of note. The first is the One way of interpreting this is according to the pure model of self- marked stability of public support for the major types of welfare policy. It interest, that is, only those programs which benefit (at least potentially) a appears that the right-wing wave which swept over the Western world majority of the population are supported by a majority of the population. (including Sweden) during the 1980s, and which had so striking an The support for different social programs, in this view, is directly related impact on the character of public debate, had no effect at all on the views to whether or not one uses subsidized services or cash benefits. This of the population regarding social policy. Insofar as studies of this sort would seem to explain the public support for child allowances, pensions, are able to measure such things, it is clear that all talk about the weak education, and health care tolerably well. One point counting against such an interpretation, however, is the strong support for employment (i.e., active labor market) policy to be 56 Laurin (1986). Laurin's data can also be interpreted, as he himself agrees, as observed in these surveys. The risk of becoming unemployed, after all, is indicating that the opposite causal sequence obtains, i.e., that the persons in question want for reasons of self-interest to withhold taxes, and seek to justify 58 this by pleading that "everyone else cheats," even if they do not really know if Another study shows that the increase in negative attitudes towards social policy this is so. detected in opinion surveys during the 1970s did not continue in the 1980s. The 57 See, e.g., Nilsson (1991), pp. 38f. trend was rather in the opposite direction. See Nilsson (1991). 168 JUST INSTITUTIONS MATTER THE LOGIC OF THE UNIVERSAL WELFARE STATE 169 something one can on good grounds expect to be extremely unevenly legitimacy, and made it possible for the Reagan administration to distributed in the population. During this period, unemployment in mobilize sufficient political support to abolish it. 62 Sweden has been below 3 percent which means that a majority of citizens This also seems the most reasonable explanation for the low level of will probably not need to make use either of unemployment compensa- public support in Sweden for such programs as housing allowances and tion or of other, mare active labor market measures. Employment policy public assistance. Both are selective; but in addition, both are very thus clearly deviates from the expectations of the theory that self-interest difficult to implement in such a manner that the requirement of is a sufficient explanation for variations in support for different welfare procedural justice is fulfilled. Fraud, waste, and bureaucratic abuse of programs. It is evidently not always the case that the programs enjoying power are particularly marked in such programs as these. 63 In contrast to broad support are those benefiting a broad range of the population. the case with labor imarket policy, no great efforts have been made to 64 Swedish employment policy enjoys very strong support, despite the fact implement these programs in such a way as to reinforce legitimacy. that a majority of citizens do not benefit from it. In conclusion, it seems that the example of Swedish welfare policy One reason for the broad support enjoyed by this policy is, pre- provides empirical support for the theory of contingent consent. It is sumably, that its founders attached great importance — when they were precisely the universal programs — which fulfill the conditions specified in building up the administrative structures for its implementation — to the this theory — that command widespread support in the population. At the question of legitimacy. One critical method for achieving this was to give same time, it is the two programs (public assistance and housing representatives of the groups towards which a policy was directed (e.g., allowances) which appear most clearly to violate the principles of this trade unions and employers) a very strong influence over the implemen- theory that enjoy the least support. It is hard to argue on behalf of these tation process. These groups have had strong interests (which have programs by appeal to a conception of substantive justice. Moreover, balanced each other, moreover) in supervising the procedures of policy these programs are difficult (not to say impossible) to implement in a 59 procedurally fair manner. They make it easy, finally (at least in the case of implementation. There is reason to compare Sweden with the United States on this public assistance), to argue that those receiving benefits do not contribute point. As Margaret Weir has noted, it is striking that no form of active according to ability to defraying the costs of the program, that is, the labor-market policy has been successfully established in the United States, benefits generally go to people who do not work and therefore do not pay 6° despite the fact that a strong work ethic pervades American society. The income tax. In other words, citizens have reason to distrust both the attempt made beginning in the 1960s CETA (Comprehensive Employ- government institutions and their fellow citizens. The strong support for ment and Training Act) — was that social program which the Reagan needs-tested employment policy, on the other hand, can be interpreted as administration found easiest to dismantle upon assuming office in 1981. resulting from (a) the widespread view that justice demands that all be This was because CETA was equated, in public opinion, with waste, allowed to work; (b) the common belief that unemployment is not self- bureacracy, and a focus on helping just certain socially distinct minority inflicted; (c) the relatively high level of confidence, historically speaking, groups; it was, in short, a program exhibiting all the problematic features in the organs of implementation; 65 and finally (d) the fact that those in of selective policies. An American scholar puts it this way: relief jobs and labor-market training give something back, that is, contribute according to ability. The legitimacy of CETA was seriously eroded by the stream of "bad press" it was receiving — adverse publicity on waste, nepotism, patronage and This indicates that even social programs with the following character- corruption. Perhaps nothing contributed more to the loss of confidence istics can achieve a broad legitimacy: (1) they benefit only a minority of and legitimacy in CETA and, ultimately, to its demise.' the population; (2) they apply needs-testing; and (3) the relation between the resources they expend and the effects they achieve is often uncertain. Many other researchers have pointed to the deficiencies in CETA's This depends on how well the founders and administrators of the implementation as the factor which undermined the program's political 62 Weir (1992), pp. 125ff.; Janoski (1990), pp. 268ff.; and Donahue (1989), p. 181. 59 Cf. Rothstein (1996). For a comparison of Sweden and United Kingdom in 63 Cf Svallfors (1989), pp. 53f. regard to this policy's legitimacy, see King & Rothstein (1993). " For an examination of problems encountered in implementing the Swedish Law 60 Weir (1992), ch. 1. on Social Services, see Astram (1988), Sunesson (1985). ss Cf King & Rothstein (1993) and Rothstein (1996). 61 Mucciaroni (1990), p. 176. 170 JUST INSTITUTIONS MATTER program succeed in (1) arguing for the program's general goals in terms of justice; (2) finding (or founding) an organization capable of imple- menting it in accordance with procedures generally regarded as correct; and (3) designing an organization capable of handling the uncertainty and friction arising in the implementation process due to uncertainty in the policy theory (i.e., the causal link between inputs and outcomes in the program). This accords well with the conclusions drawn in chapter 4 about how uncertainty in policy theory can be compensated for in the stage of implementation, and that legitimacy can be achieved by other means than through fhb parliamentary process.