The Meaning of Welfarism and Non-welfarism
Sen introduced the term welfarism as follows: ‘welfarism is essentially an informational constraint for moral judgments about states of affairs’ (Sen 1979b: 472). He did this in order to reorientate debates over welfare economics.
In his paper, ‘On weights and measures: Information constraints in social welfare analysis’ (Sen 1977a), Sen studied the informational properties of individual utility functions and used this way of thinking about utility functions to clarify the problems of intrapersonal and interpersonal comparisons involved in aggregation. The outcome was a major advance in the ability to analyse and criticise the framework of welfare economics, going beyond the former exclusive focus on aggregative issues. Questions of welfare economics were reframed in terms of information: what information is relevant for measuring social welfare and deriving policy recommendations? Is there a problem when we deliberately exclude certain information from our analysis?This approach is the basis for Sen’s key point:
The general approach of making no use of any information about the social states other than that of personal welfares generated in them may be called ‘welfarism’. I would like to argue that (i) welfarism as an approach to social decisions is very restrictive, and (ii) when the information on personal welfare is itself limited, it can be positively obnoxious. (Sen 1977a: 1559)
This leads to two distinct definitions of welfarism. The first, possibly less well known, defines welfarism as the claim that social welfare depends only on the welfare of the individuals in the society being considered, but does not need to consider how individual welfare is defined. This implies a bias towards individualism and consequentialism. Other sources of information, for example, relating to other moral values or deontological approaches, are not allowable.
Sen's second definition of welfarism involves restricting information about what ‘individual welfare' could mean and how it could be captured. Welfarism is often associated with a focus on subjective, ordinal, individual utilities, such as have become standard in microeconomic theory.This focus on ordinal utilities and the consequent refusal to make interpersonal utility comparisons very significantly limits the capacity of welfare economics to provide policy recommendations (see Coyle 2019). Welfarism, in this sense, is problematic for many reasons, including that people may adapt to unfair conditions, and because there are operational difficulties in gathering reliable information on utility. Coyle (2019: 8), citing Tony Atkinson, argues that, in order to respond to contemporary economic challenges, it is vital to ‘consider explicitly the moral consequences of particular models'. There already exists a substantial literature on axiomatic approaches to problems of social choice and welfare, in which ethical judgements are made explicit, much of which contributes explicitly to non-welfarist approaches. For example, there has been extensive debate over what is commonly known as the informational basis for justice. However, this literature, much of which is technical, involving philosophy as much as economics, has remained substantially separate from much of economic writing, which remains dominated by welfarist thinking.
If we take the first definition of welfarism, the restriction to utility information can be challenged from a normative perspective, turning to values other than utility. Non-welfarism can therefore involve acknowledging issues such as fairness, equity, solidarity and distributional concerns (Kolm 1971; Varian 1974; Fleurbaey and Maniquet 2011). Equality, for instance, may be the main value at stake (Atkinson 1970), or priority to the worst-off can be expressed by attaching a higher weight to worse-off individuals (Rawls 1971; Parfit 1995).
Non-welfarism might also involve taking into account the fact that skills are unequally distributed, that handicaps should be compensated or that responsibility needs to be rewarded (Pazner and Schmeidler 1974, 1978; Bossert and Fleurbaey 1996; Roemer 1998; Fleurbaey 2008). Goingbeyondwelfarism might also involve seeing both utility and freedom as intrinsically important. Such a move might mean defending pluralism in general (Sen 1985b, 2009), or consideration of equity as well as utility (Dworkin 1981a, 1981b; Fleurbaey and Maniquet 2011). Non-welfarism might also involve taking account of information about communities that would be neglected in a welfarist approach. For instance, Taylor's communitarism is clearly non-welfarist (see also Walzer 1983; McIntyre 1988). Some have also claimed that history matters (Nozick 1974), or raised other non-consequentialist concerns (Suzumura and Xu 2001; Suzumura 2016).Considering Sen's second definition of welfarism relative to the informational basis for justice, some welfare economists began to elaborate broader, non-welfarist approaches.[2] For example, considerations of justice have been used to focus on the provision of primary goods (Rawls 1971), resources (Dworkin 1981a, 1981b), capabilities (Sen 1985a; Nussbaum and Sen 1993), opportunities (Arneson 1990), advantages (Cohen 1989), rights (Gaertner, Pattanaik and Suzumura 1992), freedom of choice (Pattanaik and Xu 1990).
Restricting attention to utility rather than accepting a wider notion of welfare also implies that utility is unidimensional. Non-welfarism allows treating individual welfare as multidimensional (d'Aspremont and Gevers 2002). In this sense, synthetic indexes, such as the Human Development Index, belong to this family of non-welfarist evaluations.
The allowable information set is closely linked to the possibility of interpersonal comparisons of utility. For instance, if utilities are rationalised using revealed preference theory then it does not make sense to make interpersonal, or even intrapersonal, utility comparisons.
More generally, the ability to make comparisons, and to derive fairness recommendations, depends on one's chosen interpretation of utility. However, in order to overcome Arrow's impossibility theorem, it was often thought necessary to allow for a certain element of interpersonal comparability of utility. That would mean that the restrictions on information associated with welfarism create a risk of making social choice impossible, suggesting that the need to make policy decisions might make it essential to move towards nonwelfarism (see Sen 1977b, 1999).Rehabilitating interpersonal comparisons of utility for such reasons is a major development, for it rests on normative grounds, going against the idea that economics should be an ethically neutral, positive science. On the other hand, the acceptance of interpersonal comparisons of utility does not imply a commitment to non-welfarism. For example, some utilitarians (e.g. Harsanyi 1953, 1955; Ng 1999) accept interpersonal comparisons of utility. Conversely, non-welfarism does not imply acceptance for interpersonal comparisons, for some non-welfarist economics reject them (e.g. Fleurbaey, Tungodden and Chang 2003). It is even possible to justify a non-welfarist framework on the basis of ordinal utility only and no interpersonal comparisons, and still go beyond the Arrovian impossibility by introducing strong equity criteria (Mongin and Fleurbaey 2005).[3]
Almost all of these non-welfarist approaches take for granted that value judgements should be taken seriously and transparently explicated. This is usually done by capturing an axiomatic approach (see Thomson 2001; Mongin 2003). Normative principles, based on precisely specified value judgements, are expressed in a formal language, before studying the compatibility and their implications. This approach is clearly attractive but, at least so far, it has not led systematically to usable policy recommendations (Fleurbaey 2007).
3