This book addresses a timely question: why, when confronted with external pressures, are some governments more able than others to make unpopular reforms? European welfare states have faced intransigent pressures from population ageing, global competition and post-industrial shifts in recent years. Things can only get worse after the financial crisis. Analysis of reforms in the book shows substantial change in a number of directions in different countries, but not sufficient to shift individual nations out of the regime theory categories to which they are most closely related. General theories such as Jessop's Schumpeterian model, which assume a universal tendency in a particular direction leading to convergence, are not supported. Neither are more mainstream theories of path dependency which imply the reproduction of existing structures, but regime theory cannot be the last word, because nations move in different directions within regimes.

A moment's reflection can make most puzzles more confusing and this is the case here. From a naive perspective, one might expect democratic governments to do things that were popular with the electorate. However, there are many examples where governments gratuitously pursue unpopular reforms. The desire to avoid even more unpopular outcomes might be one explanation. Vis applies a variant of Tversky and Kahnemann's prospect theory: governments, like people, are loss averse. When things go downhill either economically or politically, they experience a strong desire to make up the lost ground, and may pursue policies that are not in their electoral best interests.

The book also uses an ingenious and helpful methodology. Causal patterns in welfare state development are complex. Much research shows that factors like left government, level of industrialisation, size of welfare state, constitutional and institutional arrangements, religion, social class structure, the politicisation of gender divisions, cultural or imperialist traditions and political history all make a difference to outcomes in this area. The problem is that these factors combine in complex ways, with the result that various combinations of different factors may lead to similar outcomes.

The widely used regression-based statistical methods find such complexity hard to manage, since they operate by pursuing the best-fit statistical model to a data-set. If there are multiple routes to welfare state reform, it is not clear what this will be. In practice, use of this methodology simply discards most routes, treating them as noise in the data. Regression methods can deal with complex causation, principally through the use of interaction terms and through causal path analysis, a possibility Vis does not consider. However, these approaches are clumsy, statistics are hard to interpret and they are not routinely used. The alternative of case studies does justice to individual circumstances, but requires endless labour to execute and to absorb.

Vis's study uses fuzzy-set analysis to test hypotheses about the necessary and sufficient conditions that will lead governments to pursue popular or unpopular reform programmes. This procedure analyses the role of particular factors in combination as necessary and/or sufficient conditions for a particular outcome. In this case, the hypotheses considered posit a deteriorating socio-economic or political position for the cabinet as necessary conditions for the pursuit of unpopular reform, with the reverse circumstances necessary for the converse direction in reform. The political complexion of a cabinet is relevant but not decisive. Right partisanship is itself an insufficient condition for unpopular reform, but it is part of an unnecessary and sufficient combination of conditions. Left partisanship relates in a corresponding way to more popular directions of reform.

The author uses existing political-economic data sets to code the various factors as variables. The method then uses algebraic models that take into account the extent to which relationships between the various factors are necessary and sufficient to model outcomes. The main point to come out of the analysis is a confirmation of the argument: governments are more likely to do unpopular things when faced with political or economic losses.

The book makes valuable contributions to method and to theory. Fuzzy-set analysis has attracted nothing like the attention that other approaches to cross-national comparative modelling enjoy. It does offer a way of taking seriously the potential complexity of the relationships between the various explanatory factors in welfare state political economy that the most popular alternative finds difficulty in handling. The practical outcome in comparative work is that these complexities receive less attention than they merit.

However, fuzzy-set approaches have their own problems. There are practical difficulties in constructing the algebraic models that express relationships between large numbers of interacting variables. The result is that analysts typically compress variables heroically, in this case reducing such complex issues as ‘a deteriorating political-economic situation from the perspective of government’ to one number. The analyst must also calibrate and establish cut-off points for the influence of a factor, and this can be no more than an exercise of (informed) judgement. Some critics would say that the approach is an ingenious way of dressing up intuition in academic clothing, but this is too harsh. More realistically, considerable developmental work needs to be done to address these issues, an excellent reason for expanding the use of fuzzy-set approaches.

Prospect theory provides economical explanations of individual behaviour under some circumstances. There are issues in applying it to collective decision-making, as an extensive decision theory of group choice makes clear. In this context, it is possible to think of alternative explanations of the evidence. Governments might do unpopular things to advantage particular groups of supporters against the mass (and here constitutional differences that determine how many voters must shift to shift a government are clearly relevant). They might hope to produce a positive outcome come election time which will enable them to reward the electorate. They might draw on ideological resources (which appear helpful in getting citizens to go to foreign countries and be killed so that others may get access to oil, for example) which are not considered in this account. They may wish to reward influential groups, attract funds to fight elections or win the approval of media barons.

Despite these criticisms, this is a scholarly and worthwhile book and one that all serious comparative social policy academics and their students should read with care. It will help to establish fuzzy-set analysis as an important method in comparative work, also applicable at the national level. It will broaden the use of theoretical frameworks that draw on areas outside political economy. Given the equivocal success of current analyses in providing good explanations of where welfare states in Europe are heading, this is all to the good.