Welfare Economists Have Some Catching Up To Do

Let’s start with an assertion: I claim that moral philosophers are about 30 years ahead of economists when it comes to understanding social welfare. The next few paragraphs will attempt to qualify that assertion.[1]

When economists talk about social welfare, they usually rely on what’s known as the social welfare function (SWF). The SWF maps the set of individual utilities in society into an aggregate utility function. As such, maximizing the SWF means maximizing the aggregate amount of utility in society.[2]

To the economist, the SWF will usually take one of three forms: a utilitarian form, a prioritarian form, or a Rawlsian form. As I’ll argue later, each of these forms is problematic for the reason that none incorporates a historical account of welfare.

Let’s start with the utilitarian form. A utilitarian SWF assigns an equal weighting to each person’s utility in society. The appeal of a utilitarian SWF is that such a function treats each person equally: the utility of the poorest member will count just as much as the utility of the richest member when the function is maximized.

There are many obvious problems with a utilitarian SWF. The first is that, by counting each person’s utility equally, the function can lead to questionable conclusions. For example, say that there’s a society dominated by Romans and that within the society there is one Christian. Moreover, suppose that the Romans would derive a tremendous amount of utility by feeding the Christian to lions for all to see. The Christian would obviously derive a tremendous amount of disutility from such a proposition. Nevertheless, since each person’s utility is counted equally under a utilitarian SWF, the utility gained by the Romans from watching the Christian suffer would likely outweigh the Christian’s disutility from suffering. Therefore, in order to maximize the utilitarian SWF for this hypothetical society, one inescapable recommendation would be to feed the Christian to the lions.

There are many other problems with a utilitarian SWF – and with utilitarianism in general – and I won’t repeat them here. But to get around such problems, economists have typically employed either a prioritarian SWF or a Rawlsian SWF, both of which are similar. In short, a prioritarian SWF assigns higher weights to the utilities of the least advantaged members of society, whereas a Rawlsian SWF puts maximum weight on the utility of the least advantaged member of society.[3]

Both SWFs are a step in the right direction. To take the Roman-Christian example, under either a prioritarian SWF or a Rawlsian SWF, if the Christian is among the least advantaged members of society, then the maximization conclusion may not involve feeding the Christian to the lions; as the higher weighting on the Christian’s utility might make his or her disutility outweigh the combined utilities of the Romans.

This is more or less where welfare economics is today. When economists make policy recommendations consistent with, say, optimal tax theory, which seeks to provide a framework for redistributing the economic pie in such a way that maximizes social welfare, those recommendations usually start from assuming some sort of SWF. In Mankiw et al., for example, an influential paper about implementing the conclusions drawn from optimal tax theory into practice, a utilitarian SWF is assumed.[4]

But here’s the thing, none of the aforementioned SWFs – utilitarian, prioritarian, or Rawlsian – takes into account the reason for differences in welfare. They each simply take the distribution of welfare in society as given, after which aggregate utility is maximized.

In other words, while a Rawlsian SWF places more weight on the utility of the least advantaged member of society, it doesn’t say anything about why that person is the least advantaged member of society. If that person made bad choices, and if those bad choices are indeed the reason why that person is the least advantaged member of society, then why would we want to place a greater emphasis on maximizing his or her utility? If, for example, Donald Trump spent all of his money on cocaine and strippers, in turn making him among the least advantaged members of society, would we really want to place a greater emphasis on Trump’s utility in the SWF?

This is why I asserted that moral philosophers are about 30 years ahead of economists when it comes to understanding social welfare; for moral philosophers have been talking about issues of choice and personal responsibility in social welfare since at least the 1980s. For example, the luck-egalitarianism movement that spawned largely from Rawls’ theory is based on the idea of equality of opportunity for welfare, effectively attempting to tease out luck factors from effort when it comes to understanding why some members of society are worse off (or better off) than others. And so, a luck-egalitarian SWF, as it were, would seek to place less weight on the utilities of those whose welfare positions were a result of luck factors and more weight on those whose welfare positions were a result of controllable effort. Intuitively, this seems more morally attractive than a SWF that takes the distribution of welfare in society as given.

The point is, economists need to move beyond their narrowly defined view of social welfare. Quantitative tools such as SWFs are of course welcome; but they need to be constructed properly, and on sturdy moral grounds. In this case, economists need a historical account of welfare before they employ their maximization methods. Otherwise, the maximization conclusions, however fancy or sophisticated they may be, will be wrong.


[1] I first heard of the criticism proposed in this post from Joe Mazor at LSE.

[2] In this post, I’m purposefully not saying anything about what utility is. It might refer to preference satisfaction, “purified” preference satisfaction (à la Hausman), money accumulation, or some hedonic measure like happiness.

[3] The Rawlsian SWF obviously comes from Rawls’ Difference Principle, which is explained in his Theory of Justice. It’s important to note, however, that economists have actually distorted Rawls’ views somewhat by incorporating them into welfare economics. In short, the Rawlsian SWF, as economists know it, follows a maxi-min specification, whereas Rawls’ Difference Principle was actually based on a lexi-min specification.

The difference is that a lexi-min specification puts lexical priority, in strict increasing order, on the welfare of the least advantaged members of society. So, for example, after the welfare of the least advantaged member of society is maximized, the focus then shifts to the second-worst-off member of society. And after the welfare of the second-worst-off member of society is maximized, the focus then shifts to the third-worst-off member of society. And so forth until the focus lastly shifts to the best-off member of society. The implication is that the weighting for the best-off member of society will be the lowest in a Rawlsian SWF that follows a lexi-min specification. This is not true in a Rawlsian SWF that follows a maxi-min specification, for which the weightings for better-off members in society become indifferent after the utility of the least advantaged is maximized.

[4] Mankiw, N. Gregory, Matthew Charles Weinzierl, and Danny Ferris Yagan. (2009). “Optimal Taxation in Theory and Practice.” Journal of Economic Perspectives.

Leave a Reply