The Case for Aggregative but Non-Anonymous Distributive Principles
A Comparative Version of the Priority View?
Disclaimer: This post is wonkish and dull! If you just want to have an idea of what it is about, read the first and last paragraphs.
As I have discussed in a previous post, many moral philosophers are wary of “aggregative” distributive principles of justice that, in general, contend that the benefits received by some members of the population can compensate for the burdens afflicted on others. In particular, one implication of aggregative principles is that “numbers count”: if enough individuals receive even tiny benefits, it can justify that one single individual suffers the greatest burden. John Rawls and Thomas Scanlon are among those who oppose those principles. In the post I have just linked to, I conclude however that opposing all forms of aggregation is not plausible and, actually, is not reflected in our actual moral and political practices. Now, I think that if you look at Rawls’s, Scanlon’s, and other contractualists philosophers’ positive accounts, you can see that their objection is not necessarily with aggregative principles per se, but rather with the fact that distributive judgments can be formed anonymously, i.e., without making any reference to the identity of the persons benefited and burdened. From a contractualist perspective, moral justification must be addressed to individuals. Moral justification entails more than pointing out to a person that her losses are compensated by the fact that others are being benefited. You need also to provide a reason that is related to this person’s situation and circumstances. So, can we imagine a distributive principle that, while aggregative, get rid of the kind of anonymity that is problematic from a contractualist perspective?
I will sketch such a principle here. Suppose that we are concerned with the distribution of individual wellbeing in the population and that individual wellbeing can be measured cardinally. I denote X the set of social states and S the subset of feasible social states. We have a population N of n individuals with a level wellbeing in any social state x given by
for any individual i. I will say that individual i has a complaint in state x if there is a feasible state y such that
The magnitude of the complaint is a function of the difference between the wellbeing of i and the wellbeing she would have in the best feasible state. For simplicity, we can just assume that the relationship between the magnitude of the complaint and this difference is linear.
Therefore, complaints have a numerical value given by their magnitude and we can use an aggregative principle to compare social states in terms, for instance, of the sum of complaints. Note however that we have already introduced a non-anonymity element by measuring the magnitude of each individual’s complaint taking into account the highest level of well-being this individual could achieve. But we can go further. In an interesting article that develops similar considerations, the philosopher Michael Otsuka introduces a principle that distinguishes between the magnitude of a complaint and its strength.[1] I call this the “Otsuka principle”:
Otsuka’s Principle. For any given magnitude, the strength of a complaint in a social state x is lessened by the fact that there is another feasible social state y where one could be or could have been living and that is worse for her. The worse one could have been in another social state y, the less one’s complaint in x has strength.
In other words, the fact that one could have been in a worse situation makes her complaint less strong, everything else equals. These definitions of the magnitude and strength of complaints imply that if one’s situation is the best that is possible, she has no complaint and that her complaint is stronger the worse her situation is compared to other feasible outcomes.
I think this conception captures the essence of at least one aspect of the contractualist objection to aggregative principles, namely the fact that in comparing the burdens and benefits distributed in the population, we abstract from how each person’s life could have been worse or better. To take this aspect into consideration, we have to give up anonymity. That Ann is doing good and Bob isn’t necessarily is not equivalent to the converse state of affairs. That depends on how much Ann and Bob each could be doing even better or even worse. Indeed, Otsuka’s principle echoes Rawls’s non-constructivist argument for the difference principle that implicitly suggests that the fact that the worst-off could have been even less well-off in the absence of inequality provides a justification for this inequality.[2]
Let’s formulate this idea more formally. Denote
the measure of the strength of the complaint of individual i in state x given the set S of feasible states. Following Otsuka’s principle, the strength of the complaint depends on whether there is another social state z in S where i would have been worse. The strength also depends on the complaint’s magnitude, so that we have
We suppose that c is a continuous and twice differentiable function that is linearly increasing in its first argument and decreasing in its second, with
Now, we obtain an aggregative but non-anonymous distributive principle that I call the Complaints Aggregation Principle:
Complaints Aggregation Principle. Given a set S of feasible social states, state x is strictly better than state y if and only if
\(∑_ic_i(x,S) <∑_ic_i(y,S) \)
Let’s discuss some interesting features of this principle. First, it is obviously an aggregative principle. It is obviously aggregative in complaints but also in wellbeing, at least in the sense that a small increase in the wellbeing (and so the diminution of their complaint) of a large enough number of persons must compensate for any decrease in wellbeing (and so the increase of her complaint) of another person. As I noted above, despite contractualists’ objection to aggregation, it is hard to oppose the idea that numbers should count at some point, as even someone skeptical toward aggregation like Thomas Nagel admits.
Second, the Complaints Aggregation Principle is comparative in the sense given by Larry Temkin.[3] That means that the value the principles give to social states is not intrinsic to those states. It depends on the other social states to which it is compared. This is, I argue, an essential feature both of the contractualist critique of aggregative principles but also of their constructive accounts of distributive justice. As I point out above, Otsuka’s principle is indeed reminiscent of Rawls’s defense of the difference principle based on Pareto-improvements.
Third, if we adapt the definition of the magnitude of complaints by now assuming that it depends on the difference between an individual’s actual level of wellbeing and an absolute best level that is common to everyone, then the Complaints Aggregation Principle takes the shape of a comparative version of prioritarianism. In standard prioritarianism, improvements of wellbeing targeting a person are given more or less priority depending on the absolute level of wellbeing of that person. By contrast, the Complaints Aggregation Principle weights improvements in wellbeing based on a comparison with how much worse-off a person could have been or could be. The idea of priority remains but it is not attached to absolute levels of wellbeing, but rather to how individuals fare off compared to what would be feasible. Interestingly, while in a society where the worst positions are opened to everyone, the Complaints Aggregation Principle approximates standard prioritarianism, in structurally inegalitarian societies where only some individuals can end up in the worst positions, the complaints of these individuals are automatically weakened. This can be seen as a serious objection to this principle.[4] In return, the comparative version of prioritarianism escapes some of the objections addressed to the standard version regarding its incapacity to account for the distinction between intra- and interpersonal tradeoffs.
Fourth, it is not difficult to see that the Complaints Aggregation Principle violates rationality requirements like acyclicity and the so-called basic contraction axiom. This is not surprising given the comparative nature of the principle. Its comparative assessment of any pair of social states depends on the set of feasible social states. Therefore, a contraction of the set S into a subset S’, because it may change the best and worst feasible states for each individual, can alter the complaints of individuals even though their levels of wellbeing remain by assumption constant. It follows that while x maybe judged better than y when z is part of the feasible set, the evaluation may be turned upside down once z is no longer feasible. An option is to follow Temkin and others who argue that this just signals that rationality requirements should not be considered absolute constraints on distributive principles. A more formal answer is just to point out that what the Complaints Aggregation Principle emphasize is that the feasibility or not of z affects the nature of the states we are comparing for relevant reasons. In other words, what we label states x and y are formally not the same alternatives whether or not z is feasible. Proper individuation of social states makes this kind of trouble disappear.
Finally, the Complaints Aggregation Principle can be extended to distributive issues in risk and uncertainty. In this case, we can distinguish an ex post from an ex ante version of the principle. The latter measures complaints by looking at the expected wellbeing of individuals for every feasible policy/action. The former evaluates policies/actions based on an expectational measure of complaints, measuring complaints in each state of nature and weighing the measures by the probability that the policies lead to each state. Similar issues as those faced by ex ante and ex post forms of prioritarianism then surface.[5]
***********************************
Summary for those of you who have stopped at the first paragraph
The contractualist’s objection to aggregative distributive principles is partly based on the fact that those principles are generally based on an anonymity condition. The identity of those who receive the benefits or suffer harm is not morally relevant. However, from a contractualist perspective, moral justification must be addressed to individuals. The distribution of benefits and burdens must be assessed by taking into account the actual and counterfactual situations of individuals. The Complaints Aggregative Principle illustrates how we can preserve aggregation while taking the contractualist concern with anonymity seriously. The measure of complaints depends on how individuals fare in terms of wellbeing compared to how they could have fared in other feasible states of affairs. Hence, the moral value of a benefit is not the same whether it is distributed to a person who is already doing almost as well as she could or to someone who has huge opportunities for improvement. Also, even if someone is not doing very well in relative terms, the fact that her situation would be even worse off contributes to justifying the current distribution. On the Complaints Aggregative Principle, we should minimize the level of complaints. This entails giving a priority to those whom a benefit will contribute the most to the diminution of their complaints. This can be interpreted as a comparative version of the priority view.
[1] Michael Otsuka, “How It Makes a Moral Difference That One Is Worse off than One Could Have Been,” Politics, Philosophy & Economics 17, no. 2 (May 1, 2018): 192–215. Otsuka is mostly concerned in this article with distributive issues with variable populations.
[2] Ibid., p. 202.
[3] Larry S. Temkin, Rethinking the Good: Moral Ideals and the Nature of Practical Reasoning (Oxford University Press, 2012).
[4] This can be partially corrected by assuming that the c function is convex and decreases steeply in its second argument for low absolute values of wellbeing. That necessitates however to modify the function to make complaints depend on absolute levels of wellbeing, which partially defeats the point.
[5] Essentially, in so far as a policy can be interpreted as presenting individuals with a “lottery of complaints”, the ex post version of the Principle is “lottery-neutral” and so is not sensitive to the distribution of chances. The ex ante version does not have this issue but on the other tends to downplay large complaints that may emerge in some state of nature.
“at least in the sense that a small increase in the wellbeing (and so the diminution of their complaint) of a large enough number of persons must compensate for any decrease in wellbeing (and so the increase of her complaint) of another person”
I wouldn’t say this principle would be generally accepted without looking to the underlying causes of welfare increases and decrease. As you said, contractualism argues that we would agree to impartial principles. And these principles would impose general rights and duties on free agents, rather than guarantee a relative amount of well-being (which I would argue, be limited to Sufficientarian concerns). So any benefits or burdens without a breach of these rights or duties wouldn’t be an ethical concern for contractualists.
Similarly, the magnitude of a complaint only matters when it is a consequence of a breach of a moral duty.