Second, Esping-Andersen makes claims for the predictive capacity of his model, arguing that ‘a theory that seeks to explain welfare-state growth should also be able to understand its retrenchment and decline’ (1990, p. 32). That is a challenge taken up in his later edited work (1996), emphasizing differences in the ways the different regimes have responded to global economic challenges, giving attention to issues about family arrangements (a point of criticism for its neglect in his 1990 work) to predict responses to demographic change and increasing female labour market participation (1999), and exploring future scenarios (2002). The claim that the model has a predictive value will be explored further in Chapter 4.
Esping-Andersen’s work has spawned an enormous literature (see Emmenegger et al., 2015) and poses the question as to why a study conducted in the 1980s using a limited sample of OECD member countries (when that organization itself was much smaller) should be so influential in shaping scholarship. Inevitably, some of the responses to his work criticize his methodology. Bambra (2005) and Scruggs and Allan (2006) have even found errors in his original work. Studies have been developed with alternative indices and other methodologies (see Arts and Gelissen, 2010, table 39.1, for a summary of these).
Much discussion has centred on the implications of using a typology. As indicated above, this approach has been presented as advancing a superior approach to the analysis of welfare state development to theories that postulate uniform or unidimensional processes. Esping-Andersen (1999, Chapter 5), in an answer to his critics, makes the case for parsimony and questions whether the presence of ambiguous cases matters for the strength of the overall approach. A less defensive response could be that while arguing against a multiplication of types, the recognition that ambiguities can throw up important questions about what is happening in specific countries is productive in theory development.
Clearly there is an alternative response to the inadequacies of global generalizations. This is to stress the unique characteristics of each individual system (see Schubert et al., 2009). But there is a midway point between these alternative perspectives. Comparisons may be made between detailed developments in individual countries, developing ‘middle range’ theories. There is a good case to be made for this approach, recognizing that social policy ‘systems’ are complex mixtures of policies than may bear relatively little relationship to each other (see Kasza, 2002).
Arts and Gelissen (2002, pp. 138–9) provide a valuable discussion of reasons to be cautious about the use of a typology for theory development. Their starting position is as follows:
Do typologies based on ideal types have theoretical and empirical value … ? The conclusion emerging from the philosophy of science literature is clear: not if ideal types are goals in themselves, but only if they are a means to a goal; namely, the representation of a reality, which cannot yet be described using laws (Klant, 1984). This means that typologies are only fruitful to an empirical science that is still in its infancy. In contrast, a mature empirical science emphasizes the construction of theories and not the formulation of typologies.
They, not surprisingly, see the ‘sociology of welfare states’ as in its infancy. In this sense, typologization is ‘a means to an end – explanation – and not an end in itself’ (Arts and Gelissen, 2002, p. 140). In a similar vein, van Kersbergen and Vis (2014) point out that Esping-Andersen confuses his typological method with ‘ideal type’ analysis, but they go on to argue for the use of typologies as way to ‘reduce complexity for analytical and comparative purposes’. Their argument is that reducing complexity through typologization is the purpose of the exercise rather than a flaw in the approach per se (van Kersbergen and Vis, 2014, pp. 67–74). Simplification, therefore, inasmuch as it is located in the ideas and evidence important for the explanation of differences, is significant in propelling further research. Here lies the answer to the question posed at the beginning of this section about why Esping-Andersen’s work is so important.
It is long time since Esping-Andersen’s original book was published, and the central issue is surely that regime theory aims to go beyond simple typification to make suggestions about how to explain differences. In that sense it may be used in ways that are not bound by Esping-Andersen’s theory. The development of regime theory can be separated into two kinds of work. One of these is work that – using essentially the countries featuring in Esping-Andersen’s original theory – suggests alternative ways of identifying regimes. The other is work that explores issues and problems about extending ‘worlds of welfare’ outside the original sample of nations (or those closely similar to them). While these tend to be mixed up together, particularly in the first of Arts and Gelissen’s incisive reviews (2002), some of the issues, notably those concerning gender divisions, arise under both headings, and are considered here in separate sections.
Reshaping the original regime types?
Esping-Andersen’s underlying explanatory approach involves the suggestion that decommodification has been a goal of social democratic parties. He thus correlates the political strength of social democracy in various countries with decommodification. This is a proposition that is challenged in Castles’ and Mitchell’s work (see Castles, 1985; Castles and Mitchell, 1992). In describing Australia and New Zealand as perhaps belonging to a ‘fourth world of welfare capitalism’, Castles and Mitchell (1992) draw attention to the fact that political activity from the Left may, in some places, have focused not so much on the pursuit of post-income equalization through social policy, as the achievement of equality in pre-tax, pre-transfer incomes. Hence the authors challenge what they term ‘the expenditure-based orthodoxy’, which assumes that only higher social spending will lead to more and better distribution of income. Castles’ (1985) earlier work also drew attention to the particular emphasis on protecting wage levels in Australian and New Zealand Labour politics. Castles and Mitchell also make a second point, again about Australia but also with relevance for the UK: that the Esping-Andersen approach disregards the potential for income-related benefits to make a very ‘effective’ contribution to redistribution. Australian income maintenance is almost entirely means tested, but potentially challenges the liberal principles of this mechanism because it is not simply concentrated on redistribution to those with the lowest incomes.
Mitchell (1991) brings to the argument an interest in exploring the relationships between income differences in societies before and after government interventions, suggesting that it is the size of the ‘gap’ between rich and poor, and the extent to which policies close that gap, that needs to be the object of attention, rather than simply aggregate expenditure. She examines income transfer policies in terms of their contribution to both the reduction of inequality and the eradication of poverty – alternative social policy goals which need to be interpreted in their wider political contexts. She also compares income transfer systems in terms of efficiency (the relationship between outputs and inputs) and effectiveness (the actual redistributive achievement of systems).
This work, and that of Mitchell in particular, while offering what is perhaps now a historical account of Antipodean public policy (recently economic liberal trends in these countries have been strong), is important for raising questions about the wide range of influences on incomes, and the variety of policy options available to political actors who want the state to try to manage income distribution. Arts and Gelissen (2002), in their survey of alternatives to the Esping-Andersen model, identify an additional category that some authors have distinguished within the ‘liberal’ regime group, which is called ‘radical’ (see Kangas, 1994) or ‘targeted’ (Korpi and Palme, 1998). Its characteristic is an absence of social insurance, but some evidence of the use of means tests to effect substantial redistribution. It bears some resemblance to the