Published online by Cambridge University Press: 06 March 2015
Prioritarianism is the moral view that a fixed improvement in someone's well-being matters more the worse off they are. Its supporters argue that it best captures our intuitions about unequal distributions of well-being. I show that prioritarianism sometimes recommends acts that will make things more unequal while simultaneously lowering the total well-being and making things worse for everyone ex ante. Intuitively, there is little to recommend such acts and I take this to be a serious counterexample for prioritarianism.
1 Parfit, D., ‘Equality and Priority’, Ratio, ns 10 (1997), pp. 202–21CrossRefGoogle Scholar. This article arose from Parfit's Lindley Lecture of 1991 (published as Parfit, D., ‘Equality or Priority?’, The Ideal of Equality, ed. Clayton, M. and Williams, A. (Basingstoke, 2000), pp. 347–86Google Scholar. Versions of this idea have been examined much earlier within the study of social welfare functions in economics. See e.g. Atkinson, A. and Stiglitz, J. E., Lectures on Public Economics (London, 1980), p. 340Google Scholar.
2 For formal definitions of several different forms of separability, see Broome, J., Weighing Goods (Oxford, 1991), pp. 60–89Google Scholar. What I refer to here is also called strong separability and is provably equivalent to additive separability.
3 Broome, J., ‘Equality versus Priority: A Useful Distinction’, Fairness and Goodness in Health, ed. Wikler, D. and Murray, C. (Geneva, 2003)Google Scholar.
4 See McCarthy, D., ‘Utilitarianism and Prioritarianism II’, Economics and Philosophy 24 (2008), pp. 1–33CrossRefGoogle Scholar.
6 See Parfit, ‘Equality and Priority’, p. 213. For further discussion of such interpretations of prioritarianism, see Williams, A., ‘The Priority View Bites the Dust?’, Utilitas 24 (2012), pp. 315–31Google Scholar.
8 McCarthy, ‘Utilitarianism and Prioritarianism II’.
9 For example, if the function were log2 we could replace the numbers 4, 36, 49, 100 with 1, 4, 8, 16. In general we have four numbers which we shall call a, b, c, d. For the example to work, we require that a < b < c < d, b + c < a + d, and f(b) + f(c) > f(a) + f(d). This can be achieved for any f if we choose a to be less than d, then set b to be 2/3 a + 1/3 d – ɛ and set c to be 1/3 a + 2/3 d – ɛ, where ɛ is a small number that has to be closer to zero the closer f is to linear.
10 I wish to acknowledge the Oxford Martin School for funding this research and Marc Fleurbaey, Andrew Lister, Derek Parfit, Larry Temkin, Alex Voorhoeve and an anonymous referee for helpful discussion.