• No results found

The re-discovery of institutions and outcomes outcomes

In document taming of inequality retirement (sider 38-43)

From institutions to outcomes

2.2 The re-discovery of institutions and outcomes outcomes

In the first generation(s) of macro-comparative research, most of the effort was concentrated on explaining welfare state variation - operation-alized as the relative size of public or sodal expenditures. Some studies supported a structuralist view of the world; others found that, indeed, politics did seem to matter. Among the potential political explanations, theories about the impact of working dass mobilization, operationalized as trade-union density and/or left-party strength (sometimes called the laborist approach) held a prominent position, but their effkacy vis-a-vis various structural variables appeared to be contingent on the general methodology employed (cross-sectional or time-series data), the choke of country cases Cinduding developing countries or not) and the set of control variables used in the regression equations. 23

It is fair to say that little attention was given to the second part of the

"politics matters" thesis, and the most important reason was probably that it was simply taken for granted that comprehensive (read: expensive) welfare states did in fact achieve the alleged goals of alleviating poverty and promoting economic equality.24 As pointed out by Esping-Andersen 0985; 1990), all sides to the debate entertained a linear conception of the welfare state itself, of the forces that were supposed to have shaped it and of the impact it was believed to have on sodal outcomes.

As a matter of fact, the exdusive reliance on a linear and one-dimen-sional conception of the welfare state was primarily a characteristk of large-scale statistical, or variable-oriented research efforts. In more theo-retical work and in the tradition of qualitative comparative research, there

23 The literature is massive. For an overview and critical examination of this first genera-tion of comparative welfare state research see Shalev (1983) and Uusitalo (1984).

24 One of the most influential contributions to this first generation of welfare state re-search is Wilensky's The Welfare State and Equality (1975), where, as the title suggests, the link between welfare state efforts and the promotion of equality is taken more or less for granted. See however Uusitalo (1985) for a review of "early" studies of the re-distributive impact of welfare state interventions.

was always astrong interest in institutional variation, inspired, for instance, by Richard Titmuss' (974) idea about the existence of qualita-tively different models (or ideal-types) of social policy interventions, or the (rather different) notion about the welfare state as an (expanding) set of social rights formulated by T.H. Marshall (950).25

It is only quite recently, however, that the re-discovery of institutional variation and the associated discomfort with expenditure data has led to serious attempts to operationalize the notion of institutional variation in ways that are applicable also for variable-oriented quantitative research.

Institutional features of general welfare states or of specific program areas have in many recent studies taken over as the preferred dependent varia-ble - especially among supporters of the "politics matters" thesis (see Myles, [1984] 1989; DeViney, 1984; Korpi, 1989; Esping-Andersen, 1990;

Palme, 1990; Castles and Mitchell, 1991; Abbot and DeViney, 1992; Huber and Stephens, 1993; Huber, Ragin and Stephens, 1993; Kangas, 1991;

1994; Korpi and Palme, 1994; Ragin, 1994a; Usui, 1994).26 We might also note that, as the focus has turned to institutional features of welfare state variation, structural theories have be en pushed to the background and the theoretical battleground is increasingly being occupied by contending

"political" theories. The working dass mobilization thesis is currently be ing challenged most seriously by state centered theories (Amenta and Skocpol 1986, Immergut, 1992), interest group theory (Pampel and Wil-liamson, 1985; 1989), theories claiming the importance of middle-class interests (Goodin and Le Grand (eds.), 1987; Baldwin, 1990), and theoret-ical contributions that stress the role of world system factors, diffusion and influences from the international environment. 27 Finally one could

25 In some qualitative work, however, the conceptualization and dassification of welfare state variation suffer from a tendency to be too abstract and holistic, at least for the purpose of systernatic causal analysis.

26 It is a further characteristic of the younger generation of macro-comparative research that a much more flexible and varied arsenal of analytical methods has come into play, some of which break with the linear and additive bias of the traditional regression analysis: pooled time-series analysis, duster analysis, event history analysis, probit and logit models for limited dependent variables, boolean algebra etc. See ]anoski and Hicks (1994) for a presentation of state-of-the-art analytical procedures in macro-com-parative research.

27 These kinds of factors are often found to be salient in qualitative case-studies, but they have not until recently been subjected to testing in a quantitative framework. See, however, Abbot and DeViney (1992) and Usui (1994) for very promising attempts to investigate the role of international trends on national social policy developments by the application of event history analysis to rnacro-comparative data.

mention that catholic institutions and Christian Democracy have received much attention in recent years as explanatory factors both in qualitative and quantitative studies (Wilensky, 1981; van Kersbergen, 1995; Misra and Hicks, 1994; Huber and Stephens, 1993; Huber, Ragin and Stephens, 1993).28

lf the first part of the "politics-matters" question has not yet been set-tled, the second part is even more open, but it has begun to appear more frequently on the agenda for comparative research. The optimism on behalf of the welfare state that characterized most of the first generation of macro-comparative research, has been seriously challenged by neo-lib-eral and radical critics. These critics maintain that welfare state interven-tions are either ineffective, futile, or even counterproductive with respect to the goal of a progressive redistribution of living standards and life chances. This has helped provoke a proliferation of comparative studies that attempt to treat welfare state variation as independent (or intermedi-ate) variables and focus on distributive outcomes as the main dependent variable.

Among the most ambitious attempts to evaluate the distributive impact of welfare states in a comparative perspective are the works by Hicks and Swank (1985), Ringen (1987), Pampel and Williamson (1989), Muller (1989), Mitchell (1991a) and FritzelI (1991), Huber et al. (1993), Korpi and Palme (1994). These are all highly aggregate studies, as they try to esti-mate the total impact of public welfare provision (and tax systems) on the general income distribution in different countries. 29 The highly inconsist-ent results that have emerged from these studies clearly demonstrate that als o this part of the grand debate is very far from settled. Ringen (1987), FritzelI (1991) and Korpi and Palme (1994) all conclude that the public transfer and tax systems redistribute from rich to poor and that the effect is stronger the bigger the welfare state. Mitchell's conclusions are also

28 The finding that countries with a strong catholic and/or Christian Democratie legacy are high soeial security spenders is a disturbing anomaly for the laborist or soeial dem-ocratic interpretation of the we1fare state. It is not clear, however, whether the fact of this anomaly could be accommodated within a more general (less Swedocentric) la-borist perspective, or whether it requires a radically different theoretical framework (see the discussion in Kersbergen (1995:6ff.).

29 The basic methodology of these studies varies considerably. The studies by Muller (1989) Pampel and Williamson (1989) and Huber, Stephens and Ragin (1993) are based on statistical analyses of macro variables, while the other three rest on analyses of the redistributive impact of taxes and transfers in a number of countries. For a cri-tique of the latter approach see Pedersen (1994).

highly optimistic about the welfare state's capacity to alleviate poverty and to reduce inequality, but she tends to emphasize efficiency aspects (the degree of progressivity of taxes and transfers) rather than the size of welfare spending (Mitchell, 1991a; 1991b). At the other extreme, Pampel and Williamson find no effect of the size of public spending on the level of aggregate income-inequality (1989).

These two more recent tendencies in comparative welfare state research - the focus on institutional variation and the interest in outcomes - have not yet be en fully integrated. Most of the existing efforts to explore the link between welfare state variation and distributive out-comes still use the size of sodal expenditures as their main independent (treatment) variable. Hence, they have not followed the lead to replace expenditure data with mea,sures of institutional variation as a theoretically more adequate operationalization of welfare state variation.30

On the other hand, much of the research effort that has be en devoted to measures of institutional variation stops short of a direct attempt to investigate the relationship between institutional characteristics and sodal outcomes (Castles and Mitchell, 1991). This is perfectly understandable in view of the serious methodological problems that face such attempts -especially where the impact of entire welfare states is at issue - but there is also a tendency in parts of the literature to rationalize this failure by suggesting that the measured institutional variation represents a relevant outcome in its own right. Hence, it is suggested that a search for traceable effects on sodal and economic structures is in some sense redundant.

It is certainly possible to argue that the very existence and quality of a set of social rights is significant in and by itself - as an extension of civil and political rights (Marshall, 1950), as a contribution to "the social wage"

(Myles, [1984]1989) or as providing for "decomodification" of labor (Esp-ing-Andersen, 1990). However, I would agree with the consequentialist position taken by Castles and Mitchell (1991), who argue that what ulti-mately matters is the actual use of these rights and their traceable impact on sodal outcomes. If variation in welfare state interventions cannot be shown to make a differenee to some aspects of sodal stratification (defined and measured independently of the social policy instruments

30 The studies by Fritzell (1991), Mitchell (1991a), Huber et al. (1993) and Korpi and Palme (1994) are exceptions to this generalization, but in particular the first two of these studies suffer from serious methodological problems in their own right - see Pedersen (1994).

themselves), then the "politics matters" thesis has failed and contending structuralist interpretations continue to command the high ground.31

It is true that the range of potentially relevant social outcomes extends far beyond measures of poverty and income inequality that are the focus of the outcome oriented literature referred to above. As suggested by Esping-Andersen in the quotation below, the range of relevant social out-comes should be wide enough to inc1ude all aspects of dass and status cleavages, i.e., phenomena like the occupational structure itself, condi-tions for occupational mobility, stratification by gender and ethnicity, res-idential segregation, etc.:

In any case, poverty and income distribution constitute only one (albeit impor-tant) aspect of welfare-state stratification. Even if inequalities in living standards dedine, it may still be the case that essential dass or status deavages persist.

What concerns us here is not so much incomes as how nations differ in the struc-turing of social citizenship. (Esping-Andersen, 1990:57)

In other words, it is perfectly possible that contemporary welfare states exhibit differences in their institutional charaeteristics that do have signif-icant repercussions for some aspects of social stratification, while they turn out not to bear on the eapacity for eeonomic redistribution. How-ever, the distribution of economic resources is sueh an important feature of the social structure and a salient criterion for social evaluation (Rain-water, 1974:4fO, that it would significantly restrict both the political and the sodal sdentific significanee of those institutional differences.

The more general point is that the ambition to link welfare state varia-tion with observable variavaria-tion in patterns of social stratificavaria-tion should be reinforced rather than abandoned. Without this ambition we lack clear criteria for distinguishing between trivial and non-trivial institutional dif-ferences and for sele eting between the myriad of possible classificatory sehemes. There is always a danger that attempts to conceptualize institu-tional variation in welfare state institutions will be culturally biased. There is an undeniable tendeney in the literature for disagree)llents about clas-sification to correlate with the national baekground of the contending authors (see for instance Castles and Mitchell (991), Mitchell 0991a) and van Kersbergen (995) for complaints about "Swedoeentrism" in the

31 "Making a difference" does not require that the outcome is dose to some abstract ideal. The question is whether the outcome is less harsh than the counterfactual out-comes without (or with an alternative type of) we1fare state interventions.

some of the very influential contributions by Scandinavian scholars).32 It is difficult to see how debates of this kind can be settled unless one agrees that the ultimate criterion for the relevance of a classificatory scheme is taken to be its efficacy in explaining certain outcomes.

To insist on the relevance of measurable outcomes helps to bring wel-fare state research in closer dialogue with policy oriented research efforts that are by their very nature concerned with the link between goals, insti-tutions and outcomes. One might as weU recognize that the ultimate pur-pose of any social scientific research effort is to gain instrumental knowl-edge about the social world (Popper, [1957] 1986: 58ff) , and therefore the distinetion between purely academic and policy oriented research cannot be a deep one.

However, this does not mean that the outcome variables under study should necessarily correspond to what appear to be the primary goals and interests of the policy-makers or the most powerful actors in the social policy field. For instance, one might very weU argue that vertical redistri-bution of economic resources is at best only a (contingent) side-effect of existing social policy institutions, while their main raison d'etre is to pro-vide the middle classes with insurance against various social risks and to smooth out income streams over the life-cycle (Baldwin, 1990; 0verbye, 1995), Even so, this does not make the actual degree of vertical redistribu-tion less relevant or legitimate as an object for research. The responsibility for choosing outcome variables rests with the researcher, and the researcher must take the specification of outcome variables very seriously and be able to justify their normative relevanee (Rein, 1974:62 ff).

2.3 The welfare state and

In document taming of inequality retirement (sider 38-43)