1 INTRODUCTION In Chapter 3 we explore the impact of policy research on policymaking. Obviously much of what we review is basic research. It rests on concepts so abstract and explains the causes of policy in trims so remote from immediate concerns of policymakers —for instance, economic level, demographic pressures, the structure of party systems and interest groups —that it can be of only indirect value. To know that the aging of a population will increase pension costs within a specified range is not to know other important sources of financial trouble —the rate of growth of taxable wages, productivity, prices, and unemployment, all of which are more difficult to forecast. Nor will an understanding of demographic pressures tell us what mix of policies will adequately resolve conflicts between the choice of work vs. nonwork for the aged, between employment opportunities for the young vs. opportunities for the old, or between disposable income for workers and benefits for the aged. Nevertheless, basic comparative research can make three contributions to top policymakers. First, policy deliberations can be improved by i better grasp of the degree to which social spending and program development are constrained by distant social, economic, and historical causes and the degree to which social policy is a matter of political choice. Second, by specifying broad policy options and program emphases chosen by diverse countries confronting similar problems, this research brings a wider range of policy options to view. Finally, insofar as this research uncovers the social, political, and economic consequences of different types of social policy and levels of social spending, it can improve the policymaker's understanding of leal opportunities and constraints. Thus the research covered in Chapter 1 provides background knowledge for policymakers but says little about actual program design. In principle, the research in Chapter 2, which focuses on issues of program design, implementation, and cost control, should be of much more immediate value to policymakers. In practice, the uneven quality of this research and the difficulties of drawing generalizations from it have probably limited its uses. Chapter 4 is an interpretive summary. [_ i l/«—/ (_/ ' , i " «—j v^ (\ (<—\J / * '*"- ' ' ' / ' ' '-*' ~ Chapter 1 SOCIAL POLICY DEVELOPMENT SOCIOECONOMIC THEORIES ECONOMIC LEVEL Socioeconomic interpretations of welfare-state effort emphasize the impact of economic growth and its demographic and organizational/bureaucratic correlates. Most studies rely on cross-sectional analysis of aggregate data, but a few provide time-series analysis. Although researchers have examined different samples of countries in different time periods and have used different concepts and measures, a basic Finding has emerged: economic level and its demographicttnd bureaucratic outcomes are major causes of welfare-state development. Using multiple regression techniques and including measures of economic level, age of a system or program duration, age of a population, or other correlates of affluence, several scholars have assessed the relative importance of structural forces in welfare effort or program development. As objects of explanation, most have emphasized social-security spending as a fraction of the gross national product (GNP) (or gross domestic product —GDP) —a measure of social welfare effort (elite decisions regarding the allocation of scarce resources). Some have used per capita social-security spending as their dependent variable (a measure of cash and services delivered to people by the government), which might better be seen as welfare output. Whether it explains effort or output, economic level has generally had a positive effect—if not directly, then through its effect on program duration or the aging of a population. SOCIAL POLICY DEVELOPMENT Variations in the findings result from differences in the range of affluence of the sample countries, in the measures, and in longitudinal vs. cross-sectional analyses. Moreover, with the exception of VV'ilensky (1975) and Hage and Hanneman (1977), almost all of the studies fail to straighten out the time order of the variables in causal models. Nevertheless, there is impressive convergence in the finding that economic level is a basic determinant of welfare-state development. As indicators of "level of social and economic development," Cutrigit (1965) relies on energy consumption, urbanization, and literacy; as a clue to welfare-state development, he relies on an index of "social insurance program experience" (the number of years a nation has had any of five Guttman-scaled programs in operation — work injury; sickness and/or maternity; old age, invalidism, and death; family allowance; and unemployment insurance). He shows that among seventy-six nations outside Africa social-security program experience most powerfully correlates with level of economic development. Cutright gives no attention to changes in levels of expenditure or to measures of ability to pay other than energy consumption (e.g., GNP). In a subsequent analysis, however, Cutright (1967b], using a sample of forty nations, finds a simple correlation coefficient of .61 between GNP per capita and social-security expenditures. In multiple regression analyses using different measures and controlling for age of a system, others also find that affluence is associated with social-security development, but their results are more ambiguous. Aaron (1967) uses per capita national income, per capita social-security expenditures, and social-security expenditures as a percentage of national income in twenty-two countries in 1957. He finds that per capita income is the most important determinant of per capita social-security outlays, even superior to age of system, but is insignificant as a determinant of the percentage of national income spent on social security. That finding held for a replication using 1960 data (Pechman, Aaron, and Taussig 1968). Sounding a discordant note, one study reports a negative correlation between economic level and social-security expenditure (as a percentage of SOCIOECONOMIC THEORIES GNP) for nineteen rich countries (1953 [r = -.51]; 1958 [r = -.44]; and 1963 [r=-.46] ; Taira and Kilby 1969). Comparing seven communist and seven democratic societies, Pryor (1968) finds that system age was the most important predictor of health, welfare, and education spending. In his cross-sectional analysis, economic level (GNP per capita) is unrelated to the share of national income devoted to social-security, health, and education expenditures, but in his time-series analysis (1950-62), it is important for expenditures in all three areas (pp. 179-80, 205, 219). Pryor concludes that economic level is an underlying but not an immediate cause of the level of social-security effort. Two studies based on longer time series and fewer countries come to similar conclusions. Peters (1972) analyzes expenditures in France, the United Kingdom, and Sweden from 1850 to 1965, using per capita spending as his dependent variable. Controlling for system age, Peters finds that the correlation between per capita GDP and per capita spending is significant for all three countries. In their longitudinal study of the United Kingdom, France, and Italy between 1870 and 1965, Hage and Hanneman (1977) identify two effects of socioeconomic development. They propose "that economic growth increases the relative as well as the absolute amount of resources available to the state and leads to increased demand for social-seairity provision as the number of unemployed and unemployable individuals rises. Based on an index of available resources (GDP per capita, total state expenditures as a percentage of GDP, and the size of the labor force employed in industry) and an index of demand (based on the size of the unemployed and aged populations and the median wage as a percentage of GDP per capita), they find that social-security expenditures both as a percentage of GNP and per capita are positively associated with the index of available resources in all four countries. (This is congruent with cross-sectional findings for thirty-three countries by Zöllner 1963 and with the findings for seventy-six countries by Cutright 1965.) The three studies that offer time-series analyses-Pryor (1968), Peters (1972), and Hage and Hanneman (1977) —consistently show that economic growth (as a clue to SOCIAL POLICY DEVELOPMENT available resources) is a powerful determinant of social spending, either as a fraction of the GNP (GDP) or per capita. Attempting to clarify the somewhat ambiguous findings regarding the relationships among economic level, system age, and spending, Wilcnsky (1975 and 1976) applies causal models to large samples. He first shows that for sixty-four nations in 1966 welfare effort (social security/GNP) varies by economic level: average spending goes up from 2.5 percent for the poorest quartile, to 4.0 percent for the next quartile, to 10.1 percent for the second quartile, to 13.8 percent for the richest quartile (1975:19). Then constructing causal models tested by path diagrams, Wilensky reports a quite strong correlation (.67) between economic level and social-security effort, but demonstrates that for the sixty countries with data on all variables, this relationship is largely mediated through two demographic and bureaucratic outcomes of affluence: the proportion of aged in the population and the age of the system. These findings stand up with or without the inclusion of measures of political system (totalitarianism vs. authoritarianism vs. liberal democracy). Including military spending and one measure of ideology (planning for equality) for a subsample of twenty-two countries and two measures of ideology (equality of opportunity and planning for equality) for fourteen rich countries added nothing to the variance explained. Wilensky interprets these results as follows: Over the long pull, economic level is the root cause of welfare-state development, but its effects are felt chiefly through demographic changes of the past century and the momentum of the programs themselves, once established. With modernization, birth rates declined, and the proportion of aged was thereby increased. This increased importance of the aged, coupled with the declining economic value of children, in turn exerted pressure for welfare spending. Once the programs were established they matured, everywhere moving toward wider coverage and higher benefits. Social security growth begins as a natural accompaniment of economic growth and its demographic outcomes; it is hastened SOCIOECONOMIC THEORIES by the interplay of political elite perceptions, mass pressures, and welfare bureaucracies (1975: 47). The inconsistencies that remain in the findings about the importance of economic level in welfare-state development can be explained by differences in samples and in cross-sectional and time-scries data. One source of these differences is the range of incomes captured by a sample. The greater the range of riches brought to view for any one year, the more important economic level will be. Compare Cutright's seventy-six country sample and Wilensky's sixty-four country sample with Aaron's twenty-two, Pryor's fourteen, and Wilensky's twenty-two richest countries. A second source of discrepant findings is that time-series data such as those used by Pryor and Hage and Hanneman capture shifts in demographic structure, ideology, technology, and social organization that can only be inferred from cross-sectional data. That a country's affluence typically has a stronger and more direct influence on per capita social spending than on welfare effort is readily explained: a small effort by a very rich country such as the.United States generally means more cash for each person than a large effort by a "rich" country farther down the economic scale such as Austria (Wilensky 1975: 17). In sum, even when these studies use different samples and measures, they confirm the importance of economic level and its demographic and organizational correlates as the root cause of welfare-state effort. Most important, this remains the case regardless of the type of political system. Despite their differences in methods and definitions, Outright, Pryor, and Wilensky have each found that economic level and its related processes overwhelm regime type as predictors of welfare effort. CONVERGENCE THEORY Closely associated with the discovery that economic level overwhelms regime type as a predictor of social-securitv effort over the 164 Age, class, politics, and the welfare stale Appendix 6.A. Varia/íle data sources and adjustments Variables .Sources GNP/pop. World Bank. 198.1. World Tallies. Baltimore: Johns Hopkins University Press. Uncnip. Inlcrnalional Labour Ol'liec. 19S0 and various years. Handbook of Labour Statistics. Geneva: International Labour Office. fem. cd. UNĽSCO. 1979 and various years. Statisticcil Yearbook. New York: United Nations. TFR United Nations. 1978 and various years. Demographic Year- book. New York: United Nations. fJf teen births United Nations. 1978. Demographic Yearbook. New York: United Nations. .ua lor 1955 and 1965 are estimated by midpoint interpolation. Data lor 1975 and 1980 arc Itoin the World Dank, except where clearly inconsistent with Davis, in which case dala arc projected. 'T'w Mime nations, it was necessary to subtract out denial surgeons from the figures in order to make them comparable to those of other nations. Data lor Belgium and Ireland obtained from regression-based estimates ol'Wcalhcrby ct al. 11983). 7. Conclusions: The causes and consequences of the welfare state As is line ol'any detailed, thorough empirical study, our results ;irc complex ami general conclusions must be qualified. Yet. the diverse empirical results in our study lend considerable support to our claim that age structure, politics, and development should be taken more seriously in the study of welfare- spending and social equality. Willi the perspective of the empirical analyses behind us. and a roughly defined interest group theory to make sense of the results, we can review these arguments and the evidence in favor of them. The aged and (he welfare slale •The. major influence on the rise of social welfare speiidinii from 1950 to 19X0. al least in political democracies, is n^c stmelím? - primarily Ihe rise of the aged population but secondarily the decline in the population ol young children. This inllueiice stems in pail from automatic entitlements for the increasingly larger number of eligible aged persons. As (he major program designed specifically for Hie aged? pensions respond in part to (he sheer growth in (he number of persons .entitled to benelils. Direct demographic forces likewise prove important lor medical care spending. The high rales of illness and medical care usage among older persons, and (he expensive and diflicull care needed for the very old. mean lhal Ihe aged receive medical care benelils in excess of their leprescnlation in Ihe population. Nol surprisingly, then, (hose societies wilh Ihe inosi developed welfare slates ha\e the world's oklest populations as well. At the same lime, something more than demographic accounting - perhaps ihe political inllucnce of a large, lilerale. and high-voliiig-age population - is involved in welfare spending. A number ol lindings emerge thai cannot be explained by demographic accounting alone. We lind lhal pension spending per aged person grows wilh percent aged in advanced industrial democracies, even though Ihe spending measure controls for the number of potential aged recipients. We lind lhal political participation and parly compelilion inleracl wilh the size of the aged populalion lo jointly raise welfare spending. We lind thai the cllcct of percent aged on welfare spending increases over lime, thus showing 165 32 53 166 Age, class, politics, unci the welfare state that the aged more effectively translate their numbers into greater spending in more recent time periods, ľinally, we lind that among lower- and middle-income nations, where the degree of democracy varies greatly, democratic political procedures amplify the inlluencc of age structure and other development variables - on welfare spending. All this suggests that democratic procedures furnish the opportunity for socioeconomic and demographic groups to influence public polky. Since the structure of such groups is determined by changes in the economy and in demographic composition, developmental change combines with democratic politics to raise welfare spending. Tlie aged, as both a demographic and a political force, are not the only inlluencc on welfare spending. Fconomic growth, unemployment, and inllalion'con-Irihtilc to the upward trend in spending in advanced industrial democracies, as do the aforementioned nonpartisan political participation and competition variables. Variables measuring the strength of organized labor and leftist parlies emerge important in explaining public assistance spending directed to the poor. Whereas social insurance programs for pensions and medical care favor the aged and middle-income groups, means-tested programs target the poor. Aj^iiľiely_oľ groups and forces thus inlluencc welfare spending It) one degree or another or forjiiuMype 0/progmnu^rjwojher. What appears special about the aged, however, is that the .social insurance programs they most inlluencc comprise such a large part of social welfare speiul-iiigTTcnsions and medical care spending, including that for government employees, make up 76 percent of the total. With other programs related to an old asc structure such as disability included, the proportion rises even higher. Programs with little connection to the aged, such as public assistance and unemployment, make up less than 15 percent of all welfare spending. Furthermore. whenjialiuns_slow the growjjj_or_c_ul the Ic^ofjweLfare benefits, programs for the aged fare best. Taken"together, these facts and findings add considerable credibility to the thesis that changes in age structure have contributed enormously to the growth of the welfare state over the past several decades. Among sociologists, this insight has sometimes been ignored but more often denied. The orthodoxy is that labor, capital, and class-based political parties drive welfare spending, regardless of the size or political inlluencc of the aged population. In this study, the orthodoxy fares poorly. We reject theories that dismiss the role of the aged and population, and recommend that researchers and theorists make room for age structure in (heir conceptualizations and empirical studies. .Several nvr" unirnil jippi"-:iiifin«: follow from the specific lindings.(\:\pi, conceptions of the stratification system must address more carefully the role of middle-class and ascriplive groups whose interests do not unambiguously co'm- Causes and consequences of the welfare slate 167 cidc with those of labor or capital. Dominant conceptions in the held emphasize / class relations to the means of production, occupational status, or work characteristics such as authority. When considering political components of slratilica-/ lion, however, nonclass ascriptive elements may also prove important. Since I age. sex, racial, or ethnic boundaries increasingly come to intersect class boundaries (Nielsen. 19X5). they contribute independently to a complex and diffuse pattern of social segmentation (Janowitz. 1985). In terms of the debate over | Marxian and Weberian conceptions of the stratification system, our study favors I ! a status-based theory that recognizes the importance of ascriptive and status groups ; (Parkin, 1979). Although some authors suggest the reemergence of class slrati-;' Mention in the 1980s (I'iven and Cloward. 1982: Walton. 1980). we lind the ''■ continuing importance of status and ascriplive groups in relation to the welfare state. i | ( Second..'although we lack direct evidence, our results support a view of the •'• ! , aged as an active political fbivejnjidvan<-i-d democracies. THěirpoliíícaí power j í j may come normative!}' and structurally. The aged have normative legitimacy to \ ! their claims on the welfare stale that is denied to most other groups. As a result, ! \ they have become symbolic representatives of the welfare state with full cili/.cn : t rights to public income maintenance. Beyond public support, the aged gain po-j litical power structurally through large numbers, effective organization, and common interests in higher benelils. The aged do not need candidates who spe-:] \ eialize in aged issues alone, but may use dieir symbolic and organizational power ], j to pressure candidates from all political parlies for support of their demands. t ' ( I'hinr. the size and political power of the aged suggest increased spending in j yearsjojairaje (or at least growth in the programs designated for the aged). The | simple projection of spending from numbers of potential recipients alone no doubt fails to capture all the contingencies involved (Furniss. 1986) or all the differences across nations (Myles. 1984). Still, the aged exert pressure to which governments must respond in some form or another. Since benelils are difficult to eliminate in (he face of the political power of the aged, governments must raise taxes or rely on deficit spending.-1 liven if the rale of growth cannot continue as it did during the 1960s and early 1970s, the welfare slate will remain strong by virtue of the political demands of the aged. Given these speculations, much remains to be done to investigate the inlluencc ol the aged in the welfare stale. Further research should examine how groups combine, compete, and intersect in the political process of implementing welfare spending. The aged must be given their due in such efforts, as should oilier groups. At the macrolevel. however, existing data make it dil'licull to provide more Ihan indirect evidence on participation in the political process. Studies of single nations, the organizations for the aged that exist in them, and their rela- 4�54 80 92 5224 0903 91 7 68 IQD rifit:, Lltf.1.1, ^/l/fllft.l, [frm ku t» if/lif L .min lionships with unions and political parties would supply valuable detail to go aforig with mullination. quantitative studies. Studies of low-income democracies m^'ht also provide an invaluable opportunity to study the treatment of the aged wiere their numbers are small and (he welfare state is just emerging. Another direction for research is to consider other types oľ expenditures and government activity. We limit ourselves to a set of programs based on social protection that make up a substantial part of nonmilitary government spending. Sncc mixing widely different programs in a single measure of spending hides the different causal processes that may be operating, delimiting the kind of spending lobe explained has important advantages. However, others might study different timponenls of government activity - regulation, educational spending, research nikl development, or a variety of oilier programs - and reach conclusions differed from ours. Ľľľorls to specify the domain of the various theories must continue. Rftiislribulive efficacy of welfare spending Ľo>nomicjujdj2roduclive structures have strong effects, whereas social welfare srcruling and hroadcrgovcTiímcTTnipcTKľmg variables have weaker effects, on two components of social equality - income inequality and infant mortality. High national product and an older age structure reduce levels of income inequality, high national product, female education, urbanization, and medical care resources reduce infant mortality. In contrast, political democracy, economic dependency, social welfare and other government spending, and leftist government o.ntrol have little effect on cither dependent variable. We do not show that developmental variables completely explain the cross-national variation in equal-it}, or that other variables are never important, but that the developmental vari-aHcs have generally beneficial effects, whereas the other variables do not. As a means to equality, political intervention appears to offer limited benefits. | Ifwelfaxejipenduig respondsJojJc.njocralic political processes, it may in large I p;rl rellect the demands of the politically powerful middle-income groups more l ilmi those of the poor. Having little inlluence in the political arena, the poor may mi be as cflcclive as higher income groups in pressuring for programs that most benefit them. If. as Olson (l')H2) argues, greater inequality exists in the opportunity to create distributional coalitions to protect interests than in the 'productive abilities of people, the ability of political programs to redistribute income is inherently limited. Iifforts to understand how differentials in wages and hours wwked may be more important to explaining national differences in income inequality than taxes ami transfers (Rainwater et al.. I98d). The limitations of public spending as a means to redistribution are further Louses ana consequences oj me weijare suite n.-? illustrated by considering the implications of our findings on the political importance of the aged as a determinant of welfare spending. The major part of pension spending is based on age rather than on need. Despite wide diversity in resources among the aged, benefits under dominant social insurance programs go to aged persons based on previous contributions rather than current need, liven if lower-income pensioners may receive higher returns on the contributions, higher-income workers receive higher benefits by virtue of their higher contributions. Rüther, cost-of-living increases, applied to all recipients, favor those with the highest benefits. Those nations that opt for some form of flat-rale benefits seldom offer them at high enough levels to redistribute income. Social insurance programs have grown precisely because they can obtain the support of persons at all income levels; more redislributive programs gain weaker political support. As a result, there may be a contradiction between high benelits and progressive distribution. The end result appears to be thai those vvhojArcjadj^ajMagcd duringjmiddlc1-age remain most advantaged during old age, cvcn_when their dependeneyon govenmient...liiiiisl'ers increases. Perhaps political pressure of the middle class \ leads to benefits tjiajjurjdjarcd by (lie aged poor, and without the force oflhe more aflluenl aged, benefits would be meager for those most in need. Yet, this precludes redistribution, as wealth transfers occur across generations rather than across classes (Page, ll)K.V. Ilcdstrom and Ringen. 1987). Despite its failure to reshape permanently the distribution of income, welfare spending may have other benelits. We in fact show that it can contribute lo lowering infant mortality, il may also contribute in important ways to I he well-being of the poor in many ways lhal we have not studied. There are many justifications for welfare spending, and we do not desire lo enter the political debate over the-, validity of these claims (see Ľisenstadt and Ahimcir, l°85, for a presentation of alternative sides of the debate). We focus instead on the more exagger-aled conceptions of the welfare stale as a means to achieve substantial reductions in inequality or as the road to socialism. Whatever the benefits of welfare spending, it does not replace economic and 1 productive_struclurcs as the iň"a~jóTlióujrč^^ Our results confirm the > existence of a relationship between economic development and equality, and show that il stands up lo commonly asserted criticisms. Some claim that with appropriate controls! the effect of development on equality disappears. The antecedent cause of economic dependency, or the roles of government intervention and union power, eliminate any association that development may have with equality, and need only be controlled in statistical analyses lo show the weak effects of development. Vet. our analyses show lhat these claims have lilllc validity. In nearly ail instances, economic dependency, social welfare spending, or leftist power have little inlluence on either income inequality or infant mortality. 62 99 82 06 27 29 23 170 Age, class, politics, and the welfare slate Economic development, when properly specified as having a curvilinear relation-' ship, has consistently strong effects regardless of the controls, as do some related developmental variables, such as age structure, fertility, or medical care. Others claim that low-income nations can never repeal the historical economic experience of Ihc Western nations or obtain the same benclits for equality from economic growth. The two groups of nations are fundamentally different, having begun industrial development during different historical periods and having occupied different positions in the world system. These differences make comparisons across the groups inappropriate and spuriously attribute to developmental differences what is really due lo fundamentally diverse historical experiences, liven if high-income nations have greater equality, it is not possible lo attribute it to economic development. In response to this criticism, we show that development has effects within these two groups of nations as well as across them. When analyzed separately, development strongly inlluences income inequality and infant mortality in developing nations. Similarly, development has strong effects on infant mortality among high-income nations. Only for income inequality among high-income nations, where variation is truncated due to lack of time-series data, are developmental effects weak. Otherwise, our results show clear effects of development within the historical classification of nations: the effects of development do not depend on artificial comparisons across strata in the world system. Although our results indicate that the effects of economic development are not spurious. Ihcy show thai the changes development brings about in equality conic slowly and with difficulty. Among developing nations, there is little evidence of Ihc ability of governments to unleash and maintain economic growtli and then enjoy immediate and substantial reductions in inequality; aiming high-income nations, the responsiveness-uf~incqu.al.itv lo economic growth decreases. Returns to economic growth in infant mortality are greater than in income inequality, but for neither has Ihc gap between low- and high-income nations been eliminated, expectations that equality should come quickly or easily make the slow pace of change seem unsatisfactory. Yet, we are unable to document empirically oilier means to permanent, sustained improvements in equality. The weakest link in these arguments, however, stems from problems of data availability and measurement. We, like all previous researchers, lack time-scries data on income inequality and are only partly able to overcome this gap with additional data on infant mortality, liven if multiple data points were available, problems in Ihc measurement of household income used to compute income inequality are serious. Alternative explanations based on measurement error might explain the empirical relationships we find (Gagliani. 1987). Since existing data on income inequality have already been fully analy/.ed. new data points and more Causes and consequences of t fie welfare stifle 171 accurate measures arc needed to move the debate forward. New and belter survey data would be especially helpful in this regard (e.g., Hcdstrom and Ringen. 1987; Rainwater el al.. 1986). Besides income inequality, scholars might fruitfully focus on the more valid and easily available measures of the consequences oľ income distribution such as infant mortality. Interest groups and political parties Control of governments by particular political parlies, of cither the right or the jeft. most inlluences the level of spending for relatively small public assistance' programs. For social insurance spending, the ideological perspective or unionI sujTjTojj_ofjhe viclori(ius_parly is less central Ijianjionpnrhsjm components* oľ| dcmocratie.pojiiics such as the participation of the population in elections and! the number of parties.compeling in the denuicrajic process. The class character-, is'tTcs traditionally considered responsible for the powcnTf partisan parties, such as union strength, monopolization, and strike activity, likewise have limited influence on spending. Our results show some cross-sectional relationships of class and parly variables with spending, bul lillle over-time relationship. The effects of the variables decline with controls for variables that track both over-time changes and cross-sectional differences in welfare spending. In developing nations, class structure operates through the forces of trade dependency and world system position. The effects of dependency we find in the analysis are not as strong or robust as those for percent aged combined with political democracy, and we do not waul to overemphasize (heir importance. Ľven so.r the. study of peripheral and semipcriphcral nations provides some interesting insights into the causes of welfare spending. We lind thai ties lo_corc naljons increase rather .than decrease social welfare spending^, thereby contradicting assumptions that welfare spending protects labor from the power of capital. Class-based parlies cannot be measured for most oflhese nations and cannot be expected to explain spending among them. But Ihc idea lhal welfare spending reflects lbe interests of labor and of labor parties is contradicted by the way dependency raises spending. Al a minimum, worker influence on public policy should be weaker and welfare spending lower in dependent nations, where the power of foreign capital is strong. We also were generally unable lo substantiate the causal impact ol political parties on social equality. The ^distributive impact of political parties should come through welfare spending advocated by leftist parties. Fven assuming that parties actually do increase welfare spending (at best, only partly true), the level of social welfare spending has no effect on income inequality and only limited effects on infant mortality. Moreover, the only direct association we lind be- L 5954 8513 45 4212 53 '72 Age, c7íi.v.y. politics, and the welfare state tween leftist governments andjneomc inc^ua]ilv_is shown among high-ineume democracies. This eľľect, however, is highly sensitive to the nations included in inis group and disappears with the addition ol'ncw cases. Since welfare spending itself has no direct inllucncc, it is more likely that .the effect of leftist governments on income inequality is spuriously due to the power of labor and interindustry wage differentials. Structural forces rather than political intervention emerge as most important for equality. All of this does not deny the importance of politics. Nonpartisan democratic politics remain central to welfare spending. Groups such as (he aged, which transcend the distinction between laborers anil capitalists, may pressure representatives of all parties for higher spending, particularly for programs such as pensions and medical care. Thus, the issue is not whether politics are important: clearly, they are (Burslein. j9H5j-_Xl1.c issuc concerns the organizational units of political action relevant to welfare policy. Our findings suggest that the political fartjaj}aj^oji_oJj^ and_dcmograpmTg7öTip~sTth'e number and_^ajjcty^»f3^Supeting parties, and live" structural changes in group resources <-"''!UrÍ'HLfLJiKire.l«-viiriati()^ií7HM^Iic \wfránľj^rL^Mľi;iňpiirty ideiilog'y. Neither structural forces alone nor class-based parties are sufficient lo account for growth of Ihc welfare state. Taken together. Ihe empirical results reveal an apparent contradiction in the functioning of democratic politjcs: It raises spending but not equality. As shown by the political determinants of welfare spending, democratic procedures have thcjTojauMajj^aĽdisjNhutc income. At the same timeríľdcmoo'aTícTnTýniiďuml public benefits go to thlTrnldOTFincome groups rather than lo the poor, they maintain levels of equality rather than increase them. Transfers across generations occur, but few involve shifts of income from the middle classes to the poor. From this viewpoint, democracy does not necessarily offer a means lo achieve socialism or transformation of the stratification system, although il certainly increases welfare spending. Seen as a competitive struggle for people's voles (Scliiunpeler. 1975). democracy may instead limit direct class competition. In order lo obtain sufficient votes for election, compromises in socialist ideals (or any ideological position) must be made (Downs, 1957; I'r/.eworski. 1980). Once in power, aller having compromised to win an election, parlies become further dcradicali/.ed. either unwilling or unable lo implement tangible policy outcomes (Jackman, 1980). Rather than being a group based on some unifying principle or common class position, parlies end up morejisjiractical coalitions that act in \ cí'nĽcrtjogajnjioJitia^ Rather than leading to truly \ rttorn/iujjv^pplicics. democratic noliticsand the "strugglMor~vôTeZIlĽÍur lo \srtmlhigJI^£a^onyhe^ y lies may have the potential lo reshape the stratification system, bul in practice. Causes and consequences of the welfare state 173 such redistribution has not occurred given the strong political base ol the middle; class in allvíínčed"industrial democracies. i Alternatively, parlies may prove unimportant only for the post-World War II period we study. Perhaps all parties can favor welfare spending during periods of prosperity, but during periods of crisis, rightist parties more actively oppose spending, since it- represents a drain on scarce prolits. Partisanship has been more clearly defined in battles over initial welfare legislation before World War II. Although Ihe major issues of contention between leftist and rightist parties, and the social composition of party support, have changed dramatically since then (Ingleharl. 1977. 1987). partisanship may have emerged with renewed strength dining Ihe 1980s anil may grow in the years to follow as il becomes increasingly difficult lo expand the welfare system continually and to maintain economic growth. Indeed, some indication of Ihe negative effects of rightist parties appeared in the. late 1970s in omjiiodeis'.' lislimation of models i'or later time periods would provideTinc means lo confirm this possibility. Additional investigation of Ihe political inllucncc of noncla.ss groups acioss a variety of nations could also add lo the credibility of these arguments. We have focused on ihe aged here, while ethnic and linguistic politics may be more important in oilier arenas. Research on ethnic mobilization offers a prolilable research strategy (e.g.. Nielsen. 1985). as would research on political participation of the aged. We have not been able lo measure (heir actual political participation here, or Ihe activity of their representative organizations (e.g.. Pratt. 19701. bul the potential for such study exists.' Issues of research design One source of Ihe differences between our findings and previous research is methodological in origin. We have explicitly considered longitudinal variation (whenever possible), along with cross-sectional vnrialion across nations. One form of variation is not superior to Ihc oilier, and il is not possible to use one type of model lo make inferences about ihe other (l-'irebaugh. 1980). Moth Is pes of variation require explanation, and models relying on both types of dala are necessary. We assume thai variables thai explain both changes over time within nations and differences across nalions are superior lo those dial explain only one form of variation, for example, our data on welfare spending in 18 advanced industrial democracies cover a .10-year lime span. The growth of percent aged wilhin nations corresponds closely lo Ihe growth of welfare spending: also, ihe percent aged is highest in high-spending nations such as Sweden or Austria. In contrast. (iNP explains the trend in spending wilhin nalions. lull not ihe dillcr-ences across nalions: rightist government control explains differences across na- 02 82 0226 85 445 11 37 i 2.90 Conclusion interests of the poor, but equally so the middle classes' desire to be favored by statutory generosity. On the Continent, in contrast, legislation did aim to benefit the unfortunate at the expense of the well-off, but precisely for this reason it failed as the bourgeoisie mobilized to protect itself against the reformers' redistributive predations. With no immediate interests in a reallocation of risk, the self-reliant here still sought to leave each social group to its own devices. Later, in the 1960s and 1970s, however, the bourgeoisie was prompted by demographic and economic evolution to develop solidaristic interests and accordingly changed its stance in favor of a redistributive approach. Class interests did determine the outcome of battles over welfare policy, but these varied. The laborist social interpretation suggests a particular and static connection between one class and redistributive reform that is not borne out with much consistency. Those with most to gain from solidaristic legislation have differed both between nations at any given moment and within each over time. Two further factors make the identification of any particular class with redistributive social policy even less convincing. First, the necessary developments following from initial choices that restricted later reformers' freedom of movement. Second, the battles between shifting constellations of interests that determined the fate of solidaristic reform - an element of social logic. (z) Once made, choices in social policy exerted a determining influence on the course of events later possible. Decisions taken easily in certain circumstances eliminated later disputes. After the implementation of universalist systems that granted the poor of all classes tax-financed benefits already in the nineteenth century, subsequent changes in Scandinavia necessarily differed from developments on the Continent, where contributory social insurance limited to wage earners had come first. Entitlement already earned by French and German workers prevented these nations from imitating the flat-rate, tax-financed egalitarianism of Anglo-Scandinavian developments during the Beveridge era. Only later, in the 1960s and 1970s, when the self-employed recognized their interest here, was reform along these lines possible. Neo-universalist legislation on the Continent was eventually necessitated by the choice made after the war in favor of socially fragmented arrangements that tied together for the sake of risk redistribution both the rich and poor of a single group. Separatism had been the demand of affluent independents and white-collar wage earners who refused to share burdens with workers. Once the self-employed began to decline, however, and wage earners, as a group, to Ilourish, well-off The Social Bases of Solidaristic Reform 191 independents recognized the advantages of irfimersion in a broader risk pool and reached to grasp the helping hand of solidarity that they had earlier spurned. Conversely, the flat-rate approach to social insurance that had characterized early measures in Britain and Scandinavia later required a painful reorientation towards earnings-related arrangements of a Bis-marckian bent as the inadequacy of state intervention limited to a minimum became increasingly apparent. One consequence of this evolutionary logic was that the changes which determined interests in or against solidaristic social policy reform were as often developments in the welfare system as in society at large. A triangular relationship evolved between social insurance, society and class. Only after —^ social and demographic (or in Titmuss'scase, foreign policy) developments ^.--' revealed to the bourgeoisie its potential need for redistribution did it formulate an interest in solidarity. But by themselves, such changes were insufficient. Not until also the workings of the social insurance system reinforced the middle classes' potentially redistributive interests were solidaristic initiatives possible. Universalist reform in Britain and Scandinavia <—• after the war was prompted not just by widespread feelings of vulnerability, but more specifically and prosaically by the threat means tests posed to middle-class efforts at self-help. The desire among French and German ?/ able wave of elite beneficence, explain redistribution beyond a certain ^— minimum - what is needed tojjreserve the status quo or maximize total ..,(,, utility. Thesccon^l^xn^úvcjiquld^m theory, account for a real reapportionment of burdens,, but raises new questions. Given a complete revolution, the disfavored might restructure the hierarchy of allocation, both primary and secondary. Attention in this case would turn to the major change embodied in the new structure of reward, of which social policy reform would be but one element. Put even more consistently, given broad and radical change in the primary distribution, social policy would now need to serve only less pressing redistributive functions. However, precisely because solidaristic welfare, even where most successful, has not been part of such thoroughgoing change, this caliber of explanation faces /difficulties. Either redistributive social policy reform is evidence of a I significant shift of power from favored to disadvantaged (the laborist, j Social Democratic view), or it is part of a sophisticated Bonapartist ploy j (the orthodox Marxist approach), the concession that allows elites to .' maintain their position. In the first case, cause and effect have been f reversed. Radical change does not precede and thereby allow solidaristic sociajjpolicy; welfare reform is a constituent element of this significant change. The problem, then, in.the.absence of any larger revolution to which social policy reform belongs, is how to defend the laborist view l '■■••" • ■ • The Social Bases of Solidaristic Reform , ,.. Z95 I'WLfrV (,,■ •■:■;••/■ / A^':':LL ' " ' ' '- -"'■•;•'.' > against the Marxist version of Bonapartism,JWithout the revolution, how i' dowe know that even generous social pplic.yis.nDl just, the price of stability ; { and order? How plausible is the conclusion that, without upheaval, the , Jf favored classes only reacted to pressure frornbelow, without themselves | Ý leaving a mark on reform? These two explanations (the Marxist/Bonapartist and the Social Demo-cratic/laborist- the welfare state as a manipulation of, or as a real victory for, the oppressed) are, at their most extreme, mutually contradictory. They cannot be reconciled except by qualifications that weaken their force: reform may uphold the status quo, yet also help the disfavored; elites may sometimes object to reform because they are unaware that ultimately they benefit, not because real concessions are being offered. The middle ground between the two is not a stronger argument than either extreme, but a logically vitiated version of one outlier or the other. What is needed instead is an argument with the empirical verity of the middle-ground explanations and the logical momentum of the extremes. Both the Bonapartist and the Social Democratic approaches ultimately assume only two actors, fortunate and needy, pitted in combat with each other. Their particular identity has varied. The advantaged, once Junkers and industrialists in Wilhelmine Germany, are today Thatcher's constituency. The disfavored, earlier the industrial working class, now include also ethnic minorities, the handicapped, single mothers and other- from a traditional blue-collar point of view - marginal groups. Nevertheless, this basic dualistic mold persists in social explanations of the mystery of how the have-nots ever get more than a pittance from the haves. It is frustration with the inability of social interpretations to break this polar model that has prompted the search for other agencies, above all the state, with interests and aims transcending such trench warfare. Nonetheless, an appeal to this sort of deus ex machina, though tempting, is not yet necessary. Socially based interpretations have tended to elide between class and risk category. If these are freed of any necessary identity, however, a different caliber of social explanation is unsheathed, one that is able to account for how genuinely solidaristic reform was possible without a major victory of the disfavored, how redistribution may well have represented a real concession from some among the fortunate while still buttressing the power and privileges of others. Such an unravelling reveals the circumstances in which subgroups from among both the disfavored and those who in other senses were advantaged held solidaristic goals in common that they were able to realize against equally heterogeneous, but / 2.<)6 Conclusion I politically less powerful, coalitions of redistributive losers. Actors from / each side have on occasion shared sufficiently coincidental interests to / constitute a solidaristic quorum. A social explanation of the center of what ! is usually seen as an irreconcilable polar conflict between fortunate and ill-starred is possible. In fact, it is not until one examines the middle between the have-alls and the have-nothings that solidaristic social policy and its vacillating fortunes can be explained. : ,- In highly stratified societies, what redistribution took place played !V' "between rich and poor, channeled through the institutions of charity and '| ' public assistance. Disputes here pitted disfavored against fortunate, low ; 1 against high. As extremes of poverty and affluence gave way to a ■ I flourishing of the middle strata, the bourgeoisie could not remain uncon-'; / cerned by redistributive measures. Because royal bureaucrats had on occasion assisted the poor courtesy of a taxable, but still disenfranchised bourgeoisie and because the middle classes, confident in their own self-reliance, were eager for others to follow their example and reluctant to help those who did not, reallocation was at first restricted to a minimum. This liberalist bourgeois solution to the need for a reapportionment of risk was, however, only transitory; nor, in any comparative account of the welfare state not locked in an Anglo-centric orbit, has it been particularly characteristic. Risks multiplied in extent and effect with the growth of modern economies, necessitating some form of redistribution. Self-reliance became an increasingly untenable ambition, even for the bourgeoisie. Workers, confronted daily with industrial technology's dangers and bereft of'personal resources beyond their labor, were in certain countries the first clients of a system that promised to spread the consequences of new risks. But they did not long remain the only.ones..Solidaristic measures_quickly began to appeal also to groups once proud of their ability to take on fate and circumstance singlehandedly. To the extent that redistributive measures began to concern a significant class at the fulcrum of society, and were no longer a matter exclusively for the extremes of high and low, they of necessity redirected the flow of resources within a group distinguished as much by risk as social position. Social insurance, especially of a solidaristic bent, was possible only given a certain degree of homogeneity. In highly stratified populations, class and risk ran too parallel to each other for there to be any common agreementto redistribute burdens without at the same time restructuring the status quo. The large, impoverished, needy mass would profit, the small, favored, prosperous minority only lose from a system of reallocation with The Social Bases of Solidaristic Reform 197 ambitions greater than charity. The possibility of reciprocity was a precondition for social insurance, especially as it aimed beyond a limited group to cover much of the population. Those not among the poorest but nevertheless prone to certain risks had to be convinced that, potentially vulnerable, they too stood to gain. This was not possible unless class and risk were conceptually distinguished. At the least, the redistributiye pool j had to be sufficiently homogeneous that risk, not class, was its primary / differentiating characteristic. Nor would it have been possible for the\ middle classes to recognize their interests in reform were social insurance not limited to reapportioning merely the effects of risk, and restrained from any further ambitions also to correct underlying social imbalances. Unpacking the solidaristic welfare state's partially middle-class baggage does not reveal why this route was chosen in the first place. Bourgeois interests can be and have been served in many other ways, some involving different forms of statutory intervention, others, conversely, an attempt to limit the state's role. Insome cases, an all-embracing risk pool addressed middle-class concerns for reasons that in the broader scope of things seem fortuitous or at least unpremeditated.jContinental independents sought universalist measures in the 1960s to resolve the dilemma inadvertently created_by the separatist approach taken in postwar reform. In others, the middle classes discovered immediate pecuniary motives for welcoming the embrace^ofstatutory concern, as when.p.ensions w.ereextended eyeaio.the welkejidoyred_inBntaiaand Scandinavia. Most overarchingly, however, the decision taken by the bourgeoisie in certain nations for statutory intervention in the guise of social policy, rather than other means of pursuing its concerns, was determined by much broader motives of an ideological cast. For a member of the middle class, average in both fortune and risk, social insurance of sufficient actuarial orthodoxy was not especially distinct from private efforts at risk redistribution. It offered no particular advantages beyond certain considerations of efficiency and administration, and threatened no fearsome disadvantages. For such a person, it mattered little whether public risk redistribution was limited to the poorest, leaving the self-sufficient to their own devices, or whether statutory intervention broadened in scope, with the bourgeoisie both the main source and primary recipient of reallocation. For the average middle classes the 1 distinction was largely a matter of indifference: whether tneyTnsured I themselves or paid taxes for statutory provision was materially inconsequential. The reasons behind a choice between these alternatives, should I 198 Conclusion . -,J therefore be sought in the ideological realm^Where reliance onjhgjtate was accepted and commonplace,.the second was the obvious joludon. In other nations, long and troubling battles over the virtues of statutory intervention were and continue to be fought. Where the state was regarded as the appropriate agentfor such matters, social policy had.to embjacealsp the middle classes as it evolved from a means of keeping.thej>oorest afloat to a wide-spanning system of risk redistribution.2 For the needy, on the other hand, the difference between these alternatives was much greater. Where the state was excluded from or restricted in the management of personal risk, they were left to the last resort jof public assistance. In nations where the state became.the main .insurance broker of the bourgeoisie, in contrast, the disadvantaged gained from clinging to the coattaiís of the favored. The middle classes arranged things! first arid foremost for themselves, the unfortunate were the beneficiaries of a comparatively successful.trickle-dowrju_Thanks to such self-interestec motives, social legislation aimed also at the middle classes has been mor< stable and firmly supported than measures reserved for the needy. Even -perhaps especially - the cutbacks of the 1970s confirm the point here, having hit most harshly those programs and policies addressed particularly or only to the poor, while largely sparing middle-class entitlements.3 In the ong run, the unfortunate have.gained most from those welfare states