Definition and Measurement of Welfare Effort and Its Correlates in Cross-National Analysis: A Reply to Pampel and Stryker

Abstract

Pampel and Stryker's criticism of the O'Connor and Brym article (1988) is misdirected since the objective of that article was the identification of reasons for inconsistent findings on the determinants of welfare effort. Consequently, we focused on identifying the differences between previously established models, which had reached contrary conclusions on the determinants of welfare effort. We outlined the key arguments of two diametrically different models the logic of industrialism explanation exemplified in the work of Harold Wilensky (1976) and the working-class mobilization explanation proposed by John Stephens (1979) and identified major differences. We demonstrated that inconsistent findings relating to key variables derive from differences in conceptualization and operationalization of both key independent and dependent variables. We concluded that both measures of welfare effortthe ILO measure of social security expenditure used by Wilensky (1976) and civil public expenditure used by Stephens (1979) are problematic, the former one because it takes into account only a portion of welfare effort and the latter because its compass is too broad. An alternative approach is identified, which recognizes that the welfare state's redistributive effort comprises both a social transfer and a direct government civil consumption dimension, the relative importance of which varies cross-nationally. ' Consequently, an exclusive focus on one dimension is likely to misrepresent influences on welfare effort cross-nationally. Pampel and Stryker dispute none of our arguments relating to the differences between the Wilensky and Stephens models. However, they limit their analysis to the Wilensky model, thus making it impossible for them to speak to the point of the paper.' Despite the fact that no fully specified model is proposed they attempt to read the paper for its substantive conclusions. In particular, we made no attempt to model temporal changes in welfare effort in any theoretically thorough way. Our purpose was clearly stated to be methodological and conceptual. The key argument of Pampel and Stryker is that old-age dependency is underestimated in the covariance analysis with dummy