Clem Brooks and Jeff Manza's book essentially answers its title question by arguing that welfare states persist because the public wants them to persist. The book also argues that cross-national differences in welfare state efforts persist because differences in mass opinions persist. At first glance these do not seem to be controversial claims, but within the field of comparative welfare state research, the authors argue convincingly that this basic point has been given too little credence. The dominant theories have instead explained welfare state development by 1) functional needs either from “modernization” or “capitalism,” or 2) power resources and class coalitions, while welfare state persistence has been explained by 3) path dependency and other types of institutional “lock in.” Why Welfare States Persist offers a strong critique of such explanations.

Brooks and Manza's basic point is thus that there is a causal relationship between public attitudes and welfare state effort. The theoretical argument behind the causal relationship is fairly straightforward: Politicians will be responsive to mass attitudes in order to win elections. The main endeavor of the book is to provide empirical evidence for this relationship. The analyses are based on aggregated country-level data on 14 Western countries and Japan. The main independent variable centers on answers to two questions in the International Social Survey Program surveys: “On the whole, do you think it should be or should not be the government's responsibility 1) to provide a job for everyone who wants one? and 2) to reduce income differences between the rich and the poor?” The strength of this measure is that the questions have been asked at many points in time, which increases the number of cases in the analyses. The weakness, to which I shall return, is that the two items only give a very rough account of mass attitudes, which is particularly insufficient for the European countries. The main dependent variable is the total social expenditures divided by the GDP. These expenditure data also have problems. If the GDP rapidly increases, for example, as in the case of Ireland and the Netherlands, and social expenditures remain constant, then our indicator shows retrenchment. Or if unemployment suddenly increases, as it did in the Nordic countries in the mid-1990s, for example, then we have expansion. Nevertheless, in terms of the dependent variable, the authors do a good job in supplementing the main analyses with data on compensation rates and so forth. What follows is a sophisticated statistical model of key causal relationships.

In the first empirical chapter it is shown that the measure of mass attitudes influences social expenditures. The relationship is significant even when controlled for female participation rates, left-wing strength, and number of veto points. Thus, the argument is that public opinion is not only an intervening variable between the political strength of the left-wing parties and social expenditures, and that mass attitudes have an independent impact. On the basis of their statistical analysis the authors convincingly account for differences in social expenditures among countries and among welfare regimes. They are less successful in accounting for welfare persistence over time. By means of expenditure data the authors provide good empirical evidence for the fact that differences among welfare regimes persist over time. It is shown, for example, that differences in social expenditures become more similar within regime types but not across regime types. This is an interesting finding that has a value in its own right. However, in terms of linking this persistence to attitudes, the book only provides estimates of what would have happened if the attitudes of the Americans, for example, were like those of the Norwegians. Thus, it all comes down to the trust one has in the initial statistical model, which basically explains cross-national variation in expenditures in terms of mass attitudes.

This brings me to the main cause of my skepticism about the book: its lack of effort to employ a range of measures for mass attitudes, which reflects an overly simplified notion of “support” for public expenditures. The book simply assumes that the two selected items represent one underlying dimension. However, had the authors instead used attitudes toward pensions and health care—which actually make up the two largest entries in the public budgets—then the results would have been different. Measured in this way Americans—of particular interest for the book—would have been shown to be much more supportive of social expenditures. The authors might be excused by the fact that they wanted to compile as many cases as possible, but their result is nonetheless misleading. The lack of interest in the measurement of the main independent variable also generates somewhat strange discussions in Chapter 6, where Brooks and Manza argue that politicians have a “global” responsiveness (more/less state) instead of a policy-specific responsiveness. The empirical evidence provided is that attitudes toward state responsibility for redistribution from rich to poor and for providing jobs is more strongly connected to overall expenditures than to pension expenditures, health-care expenditures, and public-sector employment. However, in my opinion, a more proper test of domain-specific responsiveness would be to connect the domain-specific attitudes to domain-specific expenditures. After all, not many politicians argue for increasing total public expenditure per se; they normally justify expenditures for particular policies that purport to offer particular public goods.

Despite these points of discussion there is no doubt that Why Welfare States Persist will become a point of reference for future scholarly analysis of welfare state persistence and variation. Its statistical modeling is exemplary—even if more rigorous conceptualization and operationalization of the independent variable is in order. And its broad argument about the importance of public opinion in the analysis of public policy represents an important theoretical contribution. Naturally, one of the big questions likely to drive continued scholarly inquiry and debate regards the sources of public attitudes (and their variation) toward welfare policies. Brooks and Manza suggest an “embedded preferences” theory, which claims that attitudes derive from social-structural location (e.g., class), discursive communities present in major social institutions (e.g., family, schools, and churches), and collective memories of welfare state development. This argument is convincing but also rather broad. If the theory of political responsiveness is to be linked to the other theoretical explanations in the field, then we need more detailed studies of the origin of mass attitudes. By advancing a strong opinion-based account, and by highlighting the need for such further study, this book makes an important contribution to the social-scientific understanding of the politics of the welfare state in liberal democracies.