In Measuring Social Welfare, Matthew Adler introduces and defends a general framework for policy-making, which he calls the Social Welfare Function (SWF) framework. The basic idea behind this framework is to define social welfare, or what is good for society overall, as a function of personal welfare, or what is good for individual members of society. Adler both defends this framework against its main rival, Cost Benefit Analysis (CBA), and defends a particular approach within this framework, a certain family of ‘continuous prioritarian’ SWFs. The book combines abstract mathematical analyses of these frameworks with more practical discussions of their implementation in policy-making scenarios.
Readers from a broad variety of backgrounds – including economics, philosophy, politics, public policy, and legal theory – will find much of interest in this book. Although it contains a good deal of technical material, this is presented in such a way as to be accessible to the average reader, yet without sacrificing rigour. Various views, conditions, and theorems are stated informally (or semi-formally) in the main text and illustrated by examples set out in tables, while precise formulations and proofs are provided in an extensive appendix. Moreover, despite this technical focus, Adler shows an impressive sensitivity to philosophical issues, including even a discussion of metaethics.
The book is structured as follows. Chapter 1 provides a synopsis of the SWF framework, lays out its ethical presuppositions (it is both consequentialist and welfarist), and contrasts it with CBA. Chapter 2 discusses well-being and its measurement, including the crucial issue of interpersonal comparisons. Chapter 3 gives an overview of the landscape of SWFs, presenting a variety of kinds of SWF – including utilitarian, prioritarian, rank-weighted (or egalitarian), and sufficientarian – and explains their different properties. Chapter 4 discusses ethical arguments for choosing between these types of SWF. Chapter 5 applies the SWF framework to a case study, evaluating a policy of reducing mortality risks. Chapter 6 considers the legal and political implications of adopting the SWF framework to replace CBA. Finally, Chapter 7 previews some areas of future research required to expand the SWF framework, removing some limitations assumed in the book.
Although Adler's definition of the SWF framework is generally very clear, there is one feature of this framework that I found somewhat odd. As Adler explains, the framework has three components: (1) a well-being measure, (2) a rule for ranking well-being vectors, and (3) an uncertainty module. A particular theory within this framework is formed by selecting a particular version of each component. For example, one could choose a well-being measure based on preferences or one based on hedonic experiences; a utilitarian rule for ranking well-being vectors or a prioritarian one; and so on. What I found odd, however, is that the second component is in a sense redundant. One cannot freely mix-and-match the second and third components. If two theories have the same uncertainty module, they must also have the same vector-ranking rule. (The reason for this is that the uncertainty module includes a ranking of lotteries or prospects, and a well-being vector can itself be considered as a degenerate case of such a lottery or prospect; so the uncertainty module also ranks well-being vectors.) It would seem more accurate, therefore, to say that the framework has only two components, the first and the third, with the second really being contained within the third.
The above observation is not really a criticism, but there are two further features of Adler's SWF framework that seem to me potentially more contentious. First, the ranking of well-being vectors is required to be transitive. Adler justifies this by saying that the ranking is intended to represent the ‘at least as good’ relation, which must be transitive by definition (p. 260). This seems fine. But then I wonder why the rule must determine a ranking as opposed to a choice function that specifies, for each possible set of options, which of these may permissibly be chosen. It is unclear why a choice function would be any less useful for policy-makers. Second, and related, Adler considers only SWFs that satisfy the well-known ‘Anonymity’ axiom. He justifies this by saying that it follows from a commitment to ethical impartiality (p. 97). However, if the requirement of transitive ranking were dropped, then I do not think this would follow. Within the broader class of theories that would then be admitted, there are some which violate Anonymity but which are nonetheless committed to ethical impartiality. I have in mind, in particular, theories that allow only restricted aggregation. To be fair, Adler does discuss the issue of aggregation (he considers adopting a ‘lexical threshold’ as a way to restrict aggregation). And, especially given that the book is intended as an introduction, he certainly can be forgiven for not covering everything.
One final, albeit still very mild, criticism: I feel that an important disadvantage of Adler's favoured family of SWFs, the continuous prioritarian SWFs, could have been discussed more prominently. Members of this family rank well-being vectors according to their sum of ‘priority-weighted’ well-being, where the weighting is determined by a strictly increasing and strictly concave function. As Adler states, in order for such an SWF to satisfy a condition of invariance with respect to ratio-rescaling (basically, the ranking should not be sensitive to the choice of unit in the well-being measure), this priority-weighting function must take a specific form (the subfamily of ‘Atkinson SWFs’). However, the cost of this is that negative well-being must be excluded. Adler does acknowledge this in a footnote (p. 155, fn. 33), but this limitation seems to me significant enough to have warranted some discussion in the main text.
There is much to commend this excellent book. It is extremely clear, informative, and relevant. As discussed above, it manages to be both accessible and technically rigorous. And it achieves a pleasing blend of both abstract analysis and practical application to real-world examples. Adler is also admirably careful in stating the assumptions behind the SWF framework as he develops it, and in indicating when and why these might be controversial. He openly acknowledges the limitations of the framework, and moreover discusses possible avenues for expanding it beyond these limitations. I would happily recommend this book, especially to those who are new to the SWF literature, but also to those already familiar with these topics.
