Evaluating welfare state efforts and the model family approach: problems and promises Preliminary version – please do not quote Natascha Van Mechelen Gerlinde Verbist Karel Van den Bosch Centre for Social Policy Herman Deleeck University of Antwerp Prinsstraat 13 B-2000 Antwerpen Belgium fax +32.3.220.43.25 e-mail: natascha.vanmechelen@ua.ac.be gerlinde.verbist@ua.ac.be karel.vandenbosch@ua.ac.be Paper to be posted on the 2004 ESPAnet Conference website Abstract The aim of this paper is to contribute to the development of indicators for social protection that are helpful in the processes of 'benchmarking' and 'learning from best practices'. The importance of such indicators is underlined by the recent call by the Commission of the European Communities for more streamlining of policy co-ordination on social protection. Drawing on the principles underlying social indicators in general and input indicators (i.e. indicators about welfare state efforts instead of about outcomes) in particular, we suggest that the model family approach would be a most useful instrument to develop indicators of social protection arrangements. This approach involves calculating net disposable incomes resulting from packages of social protection measures, giving a clear indication of cross- national differences in the levels of minimum income protection and the financial stimulus to start working for various family types. Moreover, these models are very easy to develop and to maintain. Nevertheless the use of model family models is not unproblematic. Firstly, model family results are very sensitive to the definition of net disposable income. Secondly, model families are defined by numerous assumptions and because of cross-regional differences and a lack of standardised data it is often not possible to define a uniform method to deal with them. Thirdly, the model family approach only focuses on only one dimension of social protection, that is the level of fiscal and social costs and benefits. Therefore the model family approach produces only partial indicators of social policy efforts. Finally, it is not easy encapsulate the model family results for various family types – let alone for several social benefits – into one single indicator of minimum income protection. Except for the first problem, most disadvantages of the model family approach can be (partially) solved by the micro-simulation method.