1“Impressive” is the first word that comes to mind in describing this book. It weighs over a kilogram. At 635 pages (656 including the front matter) it is large. It is comprised of an introduction, eight self-contained chapters, a thirty-page bibliography, and a useful index. And “Impressive” is also how I would characterize the content of the book, as well as the project at which it is dedicated. “Well-Being and Fair Distribution provides a rigorous and comprehensive defense of the ‘social welfare function’ (SWF) as a tool for evaluating governmental policies. In particular, it argues for a ‘prioritarian’ social welfare function: one that gives greater weight to well-being changes affecting worse-off individuals. In doing so, the book draws on many literatures: in theoretical economics, applied economics, philosophy, and law.”
2Beyond being an impressive piece of work, this book is also an exemplar to be followed. It provides an excellent example of what needs to be accomplished by any defender of a particular type of normative analysis, in so far as it makes all its sources and its implications transparent. The justification of the use of prioritarian WF is like a colossal cathedral made of numerous stones. Most papers in the literature propose some marginal innovation, while taking for granted many of the stones which are already there. Here, however, Matthew Adler sets out to build the whole cathedral before our eyes. The interest of his book does not lie in the addition of a new stone; rather, each single stone is discussed, polished, and justified in every respect, i.e., in consideration of all the implications for philosophy, economics, law, and public affairs. Although I am not saying that the reader will be entirely convinced by every argument, I certainly appreciate the author's intelligence and courage in going through each step of the reasoning, while addressing each of the potential criticisms which might shake the construction in its totality.
3I shall here summarize the view defended by Adler. I shall review certain existing criticisms in the literature, and discuss certain further points regarding welfarism and legitimacy. I shall also explain why this book should be read, and pinpoint what side interests it may have.
4Adler's project is to justify the use of a prioritarian social welfare function in welfare economics and policy recommendations, as against the standard approach of Cost-Benefit Analysis (CBA). There are undoubtedly some points in common between CBA and Adler's proposal which could still be called into question. The interest of the book, however, lies in the distributive concern encompassed in the prioritarian SWFs, something which is deliberately missing in CBA: “One way to think about the book is that it's an argument for a weighted cost-benefit analysis, where dollar costs and benefits are first multiplied by weights before being added up, and weights are assigned in this `prioritarian' fashion,” says Adler (2013). In other word, “The prioritarian SWF gives us the intellectual foundation for distributionally weighted CBA” (Adler, 2012).
5The structure of the book is as follows.
6Chapter 1 discusses and defends consequentialist welfarism. On this view, public policies should be made in order to improve social welfare; and the only relevant information to compute social welfare is that which has led to individual well-being (34-35 and 38). This rules out any other sources of norms—such as liberty, rights or other values taken for their intrinsic importance—and the well-being of non-persons—such as non-currently-existing persons, or as animals. In welfarist approaches, it is not only social welfare that depends on individual utilities—a claim which would be hard to contest—but the resulting computed social welfare is also considered to capture judgments about fairness. Both on CBA and on Adler's proposal, fairness is hence presumed to be encapsulated exclusively in the consideration of individual welfares. This is a strong assumption—Dan Hausman (2012) has provided some illuminating criticism of this part of Adler's argument; and I must say I overwhelmingly share his doubts.
7Chapter 2 considers the competing views to his prioritarian SWF. After a short yet careful presentation, Adler studies and rejects each of the following positions: standard CBA, inequality, poverty, social-gradient metrics, tax incidence analysis, and QALYs.
8Chapters 3 and 4 defend the idea that “well-being should be understood in terms of extended preferences” (xix): Adler retains an objective account of the good, where well-being is based on an individual's convergent idealized ranking of their life-histories. Chapters 5, 6, and 7 comprise an argument for the claim that “the most attractive functional form for an SWF is the continuous prioritarian form” (xix). Notice that these two claims—welfarism on the one hand, and prioritarianism on the other hand—are separable: Adler defends both of them, but each can stand without the other.
9Three further problems with SWF are presented in chapter 8: the optimal design of legal institutions, future generations, and individual responsibility.
10Let us dwell on the welfarist prioritarian social welfare function that Adler defends. His fig 5.1 (312) is very helpful in this regard, providing a summary of its axiomatization as well as a number of critical arguments which I briefly describe here.
11First of all, Adler focuses on Paretian anonymous SWFs. According to the weak Pareto principle couched in terms of preferences, if everyone prefers x to y, then social welfare will be higher in x than in y. This condition captures some features of welfarism (52). According to anonymity, given an outcome x, if the arrangement of well-being levels in y is a permutation of the well-being in x, then x and y are morally equally good. For a given pattern of well-being, it should not matter which particular person is at which level (52). This condition captures the notion of impartiality, which is also necessary to define fairness. According to the Pigou-Dalton principle, the distribution of incomes in country A is better than that in country B if poorer individuals are better off in A than in B, and well-off individuals are not as rich in A than in B (339). It is fundamentally a distribution-sensitive condition; conversely, rejecting this axiom amounts to excluding any concern for distributional issues. Among the universe of Paretian anonymous SWFs, Adler favors the SWFs that respect the Pigou-Dalton principle. Otherwise—i.e., where the Paretian anonymous SWFs do not satisfy the Pigou-Dalton axiom—they correspond to a utilitarian or sufficientist SWF.
12When the axiom of separability holds, the well-being of unaffected individuals does not affect the ranking of outcomes (310). If the Paretian anonymous SWF respects the Pigou-Dalton principle, but does not satisfy the separability axiom, it is the rank-weighted SWF. In the case that it does, we are able to characterize the set of Prioritarian SWFs.
13Among the Paretian anonymous SWF respecting Pigou-Dalton, some may not satisfy the continuity axiom: this is the case for the leximin SWF and the Prioritarian SWF with lexical threshold. Continuity requires that small changes in individual well-being should not produce abrupt changes in the moral ranking of outcomes (326). For instance, continuity does not hold when there is a threshold such as a poverty line, for being very slightly under or above the threshold can totally change the ranking. As Adler considers that this is difficult to justify on the basis of fairness, he focuses on the set of all Paretian, anonymous SWFs respecting the Pigou-Dalton principle, separability, and continuity. These are the continuous Prioritarian SWFs, among which number Atkinson's SWFs, which are the ones defended by Adler.
14Several reviews of the book have already appeared. Hausman (2012), for example, praises the rigor of the book, which notably enables the reader to identify the fundamental and ineludable problem faced by the welfarist approach favored by Adler. He develops an external line of criticism based on the distinction between welfare issues and fairness, and specifically on the discussion of the separateness of persons. Voorhoeve (2013) also discusses the issue of the separateness of the persons, as well as providing comparisons with Harsanyi's account of the veil of ignorance and extended preferences.
15Sagoff (2012) does not insist on any line of criticism based on internal considerations of consistency and exposition, but rather recalls what the possible criticisms of Adler's welfarist commitments could be. Welfarism, he says, reveals itself to be incompatible with capitalism: rather, it is close to socialism in the sense that welfarist implementations may be captured by particular individuals or state councils, rather than serve the actual collective interest. Adler considers certain information restrictions that would guarantee that interpersonal comparisons are made possible and meaningful, on the premise that no collective evaluations are possible without them. Athanasiou (2012) contests the necessity of interpersonal comparisons of utility. He specifically notes that the banishment of interpersonal comparisons of utility remains topical today within the context of the new development of social choice theory. Since these appraisals, among others, are already well developed, I shall not repeat them here.
16This is obviously not the place to recount and discuss all the (other) arguments developed in the book. Rather, let us return to the underlying question which is both asked and answered by Adler: “Why should CBA as we know it be rejected?” Let us consider the three reasons evoked by Adler. Firstly, CBA does not provide a good account of well-being: rather than considering mere dollars—conceived of as willingness to pay or as willingness to accept—as a good proxy for well-being, Adler suggests that we should account for extended ideal rational fully informed preferences over life-histories. Secondly, as CBA offers no consideration of fairness, Adler suggests we move from mere efficiency to prioritanism. Thirdly, we cannot be sure that CBA is well applied, even though it is set on a legal basis in the US: here, Adler notably provides a description of both a discretion issue and an operational problem. My claim here is that taking the last argument seriously undermines Adler's strong arguments for his answers to the two first problems.
17Putting aside the open issue of whether preferences or utilities—however “laundered” they may be—do or do not entirely correspond to fairness, who wouldn't agree that ideal rational fully informed preferences are important to satisfy? Yet this so far remains only an ideal. Not only may the ideal be hard to reach, it may even be hard to identify. Along with the father of welfarism, Bentham, and indeed with Adler himself (see, among other instances, p. 282), I agree that it is better to have a bad thermometer of ideal welfare than no thermometer at all (University College Bentham Manuscript, Box 27, p. 37, reproduced by Goldworth (1979, 13)). Even so, the proxy may be substantially different from what it stands for. And if it proved incompatible with individual welfare in some significant cases, this would raise an issue with far-reaching consequences. In other words, I fear that we claim to respect welfarist fairness, whereas this principle may in fact be violated whenever it is implemented. Therefore, for operational reasons, I question whether the kind of preferences retained by Adler do indeed respect welfarism.
18 Adler recognizes that actual preferences may be wrong, based on false beliefs, incomplete information, or irrational thoughts; and he concludes that these do not constitute relevant information for assessing social welfare as fairness. To tackle actual well-being, we need to make sure preferences are conceived as if individuals were fully informed, and fully rational, even over time. As well as this, to make sure we tackle justice issues, preferences should be impartial, in the sense that individuals' convergent idealized preferences should be favored. Thus Adler considers that the relevant preferences for justice issues are idealized extended preferences over life-histories rather than actual preferences. On Adler's account, as in contemporary utilitarian theory, preferences should be “laundered”. Let us take the scope of this “laundering” process seriously.
19To estimate extended preferences, Adler uses “a strategy that uses existing data about an individual's preferences over outcomes and choices, to estimate her own-history preferences, and thereby to estimate her extended preferences more generally” (205). Let us consider spectator k's preferences. This strategy says that: If spectator k shares with respect to outcome z the same essential attributes (such as similar chromosomes) and the same contingent attributes (such as income, leisure, happiness, health) as individual i does with respect to outcome x, then spectator k shares the same preferences in z as i does in x. While they appear as mere proxy tools, attributes cannot but play a role in identifying which actual preferences are relevant for our purpose (246). Concretely, if somebody with some given characteristics or in a given situation likes a better than b, then you may consider that anybody with similar characteristics or in a similar situation will like a better than b. With different characteristics, you may come to consider a different preference. Even if you expect preferences to eventually converge to define idealized preferences, the eventual identification of characteristics is of great importance to determine what preferences are retained for the analysis. I wonder, however, whether this strategy is faithful to genuine welfarist fairness, i.e. whether the resulting evaluation is always justifiable to each person.
20First, this strategy at least raises a vocabulary problem. This operationalization strategy is literally non-welfarist in a technical sense, since it is using information that is different from utilities to assess social welfare. It does not (yet) represent a conceptual inconsistency with welfarism since Adler does not conceptualize the preference in this way; he is content simply with suggesting it for the sake of operationalization (240). Notice that a similar issue has been raised with respect to Bentham's utility calculus: the implementation of the calculus is eventually based on descriptive circumstances, rather than only on utilities as such (Bentham, 1789, chap. 3-4 ; Baujard, 2009).
21Second, this strategy may raise an ethical problem in certain situations. Because preferences may be defined differently when the identification of attributes is changing, it may happen that real preferences (whether idealized or not) may be violated from one set of attributes to the other. If you use an assessment of utilities that is not strictly welfarist as a proxy for utilities, the actual context of data observation is likely to induce a situation where a majority of preferences is not satisfied (e.g. Baujard, 2006). In an ethical sense, welfarism may hence be violated in these specific (but not insignificant) instances.
22Third, this strategy raises a legitimacy problem. The way these attributes are determined, observed, and considered depends on the experts who implement the evaluation or the policy. Similarly, in Bentham's utility calculus, descriptive circumstances are not given by the individuals themselves, or by any impartial spectator (who does not, as such, exist) ; the selection, identification, and assessment of these circumstances are made by certain “experts” independently of actual individual welfares. This cannot but bring up an important issue of discretion, as Adler has rightly pointed out with respect to the implementation of CBA—e.g the actual possibility of properly enforcing a moral code (557-558). Should we leave this consideration entirely up to politicians, as is suggested by Adler (e.g. with regards the choice of the inequality-aversion parameter, 396)? The answer would be ‘yes’ only with the proviso that a normative demarcation between politicians' and experts' value judgments is truly possible. But, in spite of the exemplary exercise of transparency conducted by Adler in his book, I fear that this demarcation is not always possible (Baujard, 2013). We would need to carefully study the stakes of the choice of certain attributes rather than others in the process of implementation of preferences for evaluating policies; if indeed this is possible, it is certainly not something that has already been accomplished. The power of experts may be growing in the shadows, while the degree of empowerment of the people concerned is declining. Unless we assume that experts always know and want what is good for the people (and not for themselves), this strategy is unlikely to be faithful to welfarist fairness, and in particular to the ideal of the separateness of persons. The issue of democracy remains open.
23To be clear, I am not suggesting that this descriptive information is irrelevant to measuring individual welfare; it is. My claim is rather that the move from the theory—and its impressively minute, and possibly even convincing justification—to its operationalization, also entails the loss of certain essential properties: technical welfarism, ethical welfarism, and legitimacy. I think it may be worth taking these issues seriously for the sake of normative transparency.
24The interest of the book lies primarily in its thorough justification of prioritarian SWF. The rigor, completeness, and minuteness of the analysis must be underlined. Economists, philosophers, lawyers, public affairs scholars, and public decision makers should all be interested in the book. They should avail themselves of it both for their own practices, as well as a way of opening their minds to the need for a pluridisciplinary approach to social welfare issues and policies.
25The book is deeply and personally linked to the creators of alternatives to CBA welfare economics, as testified by the list of acknowledgements, encompassing notable authors in post-CBA welfare economics as well as in the field of “economics and philosophy”. As is established at the end of the introduction of the book, its scope corresponds closely to the editorial lines of the journals Economics and Philosophy and Social Choice and Welfare.
26It is worth emphasizing that anybody ought to be able to read this book without facing problems due to a lack of pre-requisites. The technicalities are there, but these do not obscure or replace any arguments—axiomatizations and formal arguments are mostly contained in (important but skippable) footnotes. The standard of philosophical discussion is high, but all theories are presented from the basics. This means that anybody who is interested in the justification of public policy making, and not only academics specializing in social choice theory or political philosophy, can benefit from reading it.
27Reading the whole book from first page to last is a rich experience. But it also seems possible and fruitful to read parts of the book independently of the others, without encountering problems.
28The interest of the book goes beyond its primary ambition. In order to be able to present his argument rigorously, Adler considers other competing theories. This thorough attention makes the book a good survey of the literature related to the subject. For instance, chapter 2 is an excellent (brief) introduction to many approaches in social welfare, including CBA, inequality metrics and other equity metrics, and even QALYs. Even if oriented towards making a certain point, this short yet systematic presentation of competing theories is a rare and valuable aid for students of this area. Furthermore, Adler provides a historical overview of the field, and especially of the evolution of the status of interpersonal comparisons of utility, and the origins of cost-benefit analysis. Of course, this part is brief and not representative of the book overall—we might even regret it is too short and insufficiently deep—but nevertheless the book is relatively rare in the current literature on welfare economics to make this effort. This feature, beyond helping the reader follow Adler's argument, is another of the book's qualities.
29Above all, this book may be the work that will truly raise the profile of post-CBA welfare economics within economics—including prioritarianism, of course, but perhaps beyond. It is worth recalling that until now, the area has existed more or less silently. It has been developed within a circle of a small number of scholars—namely social choice and welfare theorists, or equity theorists. The research has been scientifically rigorous and, as much as possible, normatively transparent—notably unlike CBA in both these respects. Yet it has never spread through the rest of economics, which contents itself with applying CBA in most welfare analyses: in economic theory, applied economics, and in actual policies. This situation is regrettable, not only for intellectual reasons, but also for the formation of real policies—and above all for the people who are actually concerned by the implementation of these policies.
30With Adler's book and his successful challenge to frame a truly pluridisciplinary, non-technical and complete analysis, we can hope that this considerable area of research may become better understood and more widely recognized. At the very least, we may say that this book is what was needed to mark the start of a new era for welfare economics.