Economics as a Moral Science

As we have explained, welfarism is attractive to many economists. It clearly respects individualism and avoids problems of paternalism. It also fits well with the aim, clearly articulated by Lionel Robbins (1932), to make eco­nomics into a science, with positive economics clearly demarcated from ethics.

One result of Robbins’s injunction was that, from around the 1960s, welfare economics effectively disappeared from much of economics (Atkinson 2001). The background to that was that there was, for many years, great doubt about what welfare economics could achieve. Hicks (1939, p. 697) wrote of “the euthanasia of our science.” He believed that compensation tests provided grounds for reversing this, but this optimism did not last long. The work of Tibor Scitovsky, Kenneth Arrow, Ian Little, Paul Samuelson and Jan de V. Graaff seemed to imply that welfare eco­nomics was impossible (Mishan 1960, p. 218; see also Chipman and Moore 1978; Mongin 2006; Igersheim 2019). However, what Atkinson meant by “the strange disappearance of welfare economics” was not that economists stopped making normative statements, but that, from around the 1960s, they stopped analyzing the principles underlying normative statements, something that it had been common for earlier generations of economists to do.^[172] Where economists did discuss welfare issues from a theoretical point of view, the dominant approach centered on Pareto-efficiency and what came to be called the two fundamental theorems of welfare econom­ics, rigorously proved by Kenneth Arrow (1951) and Gerard Debreu (1951).

However, what happened was not that consideration of the principles underlying welfare economics disappeared, but that this became a specialized activity. Taking its cue from the work of Arrow, John Harsanyi, John Rawls and others, a new field of welfare economics emerged, placing the subject on an axiomatic foundation. This field, symbolized by the journal Social Choice and Welfare, developed as an interdisciplinary field straddling the boundaries between economics, phil­osophy and political science (see in this regard Fleurbaey 2000, 2007; Fleurbaey and Mongin 2005). This, of course, is the background to the work discussed in Chapters 7 and 11. In addition, in some other fields, explicit discussion of ethical issues was taken seriously. A major example is the problem of economic development, out of which capability theories emerged (see the discussions of capability theory in Chapter 12 and of Sen’s work in Chapter 13). For example, capability theorists and Sen have tackled explicitly methods of dealing with the issue of paternalism, focusing on the potential role of discussion and deliberation. Much work has also been done on how to conceptualize justice and how to balance the issue of justice against dimensions of welfare with which economists are more comfortable.

This is the context in which we need to think about the claim made in this book that, despite the preference of theorists for welfarist theories, when they have turned to practice, economists have frequently made non- welfarist arguments. This use of non-welfarist arguments strengthens the claim made by economists including Kenneth Boulding (1969), Amartya Sen (2004) and Anthony Atkinson (2009) that economics should become a “moral science.”^[173] What we mean by this is that economics should embrace discussions of ethics as one of the pillars on which normative economics should rest, abandoning the strict separation of economic science from ethics advocated by Robbins.

An important implication of such a change is that it may require a change in the way economists conceive their subject. At least until recently, eco­nomics was dominated by theory, conceived as a process involving the specification of assumptions in such a way as to make it possible to use mathematics to derive precise conclusions of general applicability. Welfarist criteria such as Pareto-efficiency, or other welfarist criteria such as utilitar­ianism or a modified Rawlsian criterion (maximizing the minimum utility achieved by any member of society) fitted conveniently into this approach, for they provided precise welfare criteria that can produce very general welfare conclusions. Such criteria can be used widely, even in macroeco­nomic models based on representative (identical) agents, a context in which ethical dilemmas relating to income distribution are absent. Welfare impli­cations can be derived using essentially the same methods as those used to derive comparative statics results from the models. In contrast, if a broader range of ethical criteria were to be used (for example, applying certain notions of justice or fairness), it would often be necessary to think about problems in different ways, for they do not always provide a criterion that can be applied mechanically to a model. As an example, take Rawls’s maximin criterion. Rawls applied this to the provision of primary goods, not to overall consumption, which implies that it becomes necessary to distinguish between primary and non-primary goods.

As a consequence, if economists are to take note of the conclusions reached in modern welfare economics, they may have to be open to arguments that cannot be fully expressed formally, or which rest on ethical judgments (e.g. ethical judgments are involved in deciding which goods count as primary goods). Atkinson (2009) provides several examples. For example, reducing social exclusion, a concept that, however meaningful, is hard to incorporate in an economic model, may be a powerful reason for targeting employment as a policy objective.

If economists are to be more open to conclusions reached by specialists in welfare economics and social choice theory, it is likely that they will need to be prepared to approach the subject in a different way. Welfarism, with its reliance on utilities, holds out the prospect of a completely general welfare that can be analyzed using the tools traditionally used in economic theory. Broadening the set of welfare criteria to include factors such as justice, which cannot be quantified in such a way that they can be measured against utility, may require going beyond economists’ traditional methods. It may be necessary to engage in deliberations over factors that are not quantifiable and have to be articulated verbally. In other words, being open to a broader range of non-welfarist arguments may involve a move away from economic theory as that is commonly understood.

This is potentially a radical change, but it would reflect a change that has already taken place in relation to empirical methods (see Backhouse and Cherrier 2017). When the 2019 Nobel Prize in Economic Sciences was given to Abhijit Banerjee, Esther Duflo and Michael Kremer for their use of experimental methods to alleviate poverty, they explained that their results were different from those traditionally sought by economists.^[174] Their work did not produce completely general results but results tailored to specific situations. This was important because the effectiveness of policies typical­ly depended on details of the situation that would not be apparent to a traditional economic theorist. Similarly, as large data sets have become available, economists have developed new econometric techniques in which a major concern, if not the predominant one, is identification. As a result, formal economic theory, of the type that fitted so well with the Pareto criterion and other welfarist criteria, has become less central to the field. Considering the numerous social and political practical consequences of any theoretical assertions, perhaps a similar change is required in relation to welfare economics. In the same way that some economists have become more open to psychology, and those parts of economics bordering on psychology, perhaps they should be more open to ethical philosophy and those parts of economics that engage with it, even if this means moving beyond traditional welfare economics. Only if economists move beyond welfarism and consider all relevant ethical values, will it be possible to deal with the full range of practical issues with which normative economics ought to be concerned.