The challenges of classifying welfare state types : capturing the protective and productive dimensions of social policy

A growing number of comparative studies in social policy have addressed questions surrounding the reliability and validity of macro-quantitative approaches, indicating an increased awareness of what has been called the ‘Dependent Variable Problem’ of comparative welfare research. As Kühner (2007) has shown, this is more than a minor technical or methodological problem for the most commonly used social expenditure-based measures of ‘welfare state change’ often produce conflicting answers to key research questions about the nature of recent processes of welfare state change. Although there has been much progress in attempts to resolve weaknesses in key macro-level measures of welfare state effort, the ‘Dependent Variable Problem is still at the heart of important methodological debates in comparative welfare state research (see e.g. Castles 2004; Clasen and Siegel, 2007). One important outcome of these intellectual discussions in recent years is the understanding that single-epistemology approaches are less preferable than combinations of large-n quantitative and qualitative case study designs. Even within the quantitatively-informed literature, scholars have increasingly stressed the importance to account for the multidimensionality of the concept of ‘welfare state change’ – in other words: single-indicator research (and we should add: single-domain research) is prone to providing simplistic and – in the worst case – misleading assessments of welfare reform trajectories across mature welfare states.