Temkin on Inequality

The subject of this post–the first in a series on philosophical discussions of equality and inequality–is a fascinating essay (“Inequality”) by the philosopher Larry Temkin. Here’s Temkin in all his brainy glory:

temkin 002

I had heard of Temkin before–I knew him by reputation as a very good ethical and political philosopher–but I had never read any of his stuff.  In this essay, published nearly thirty years ago but obviously still timely, Temkin claims that

the notion of inequality turns out to involve a hodge-podge of different and often conflicting positions.  Moreover, and more importantly, many of these positions are fundamentally incompatible, resting as they do on contrary views….The notion of inequality may thus be largely inconsistent and severely limited.

In this post, I’ll try to convey Temkin’s argument for this claim.

Egalitarians believe that “it is a bad thing–unjust and unfair–for some to be worse off than others through no fault of their own.”  Temkin doesn’t want to raise doubts about this belief per se–in fact, my sense is that Temkin is himself an Egalitarian under some description–but rather show that it is very difficult to “unpack what [the belief] involves.”  Temkin’s argument may be summarized as follows (this is my presentation of his argument, not his):

1.  If we know what we mean when we say that it is a bad thing for some to be worse off than others through no fault of their own, then we should be able to determine, given any two situations wherein some are worse off than others through no fault of their own, which of these situations is worse than the other with respect to inequality.

2.  But we can’t do that.

3.  Thus, we don’t really know what we mean when we say that it is a bad thing for some to be worse off than others through no fault of their own.

Naturally, the bulk of Temkin’s essay is devoted to defending premise (2).  The problem, Temkin suggests, is not that we can’t find a principle on which to base judgments of relative inequality (as I shall call them), but rather that there’s more than one principle on which we might base such judgments, that these principles are incompatible, and that they seem equally intuitively compelling.  Temkin’s argument proceeds in two stages.  The first stage is concerned with finding a principle on which we could base our judgments about which individuals have a complaint concerning inequality–a principle, for example, that would make explicit the conditions under which I could plausibly claim that inequality is bad for me.  The second stage is concerned with finding a more general principle on which we could base our judgments about whether and, if so, why, an unequal distribution in one ‘world’ is worse than another ‘world.’

Stage One: Temkin’s Two Principles 

Temkin claims that there are two plausible but competing principles concerning the conditions under which I have something to complain about with respect to inequality (we’re assuming throughout that everyone is equally deserving):

Average Principle: I’m worse off than the average person.

Best-Off Principle: I’m anyone other than the best-off person.

In motivating the Average Principle, Temkin says that if I am “already…better off than the average member of my world, most people would tend to think that I have nothing to complain about with respect to my level of welfare.”

In motivating the Best-Off Principle, Temkin says that even if I am doing very well relative to the average, if there is even one person who is undeservingly better off than I, that seems like grounds for complaint.  That complaint may seem relatively trivial when compared with the complaints of the least well-off, but still, it seems I have a legitimate complaint.

Now Temkin shows that these intuitions about the conditions under which I have a justifiable complaint with respect to inequality are inconsistent, insofar as we can imagine circumstances in which they yield different verdicts concerning which of two unequal distributions are worse for me.  That is, we can imagine distributions which seem worse for me (I have a more serious complaint) according to the Average Principle, but not according to the Best-Off Principle, and vice-versa (Temkin includes a helpful diagram that illustrates this possibility, but I’m not tech-savvy enough to reproduce it here).  It seems, then, that we have to side with A1 or A2 over the other, but which should we choose?  They both seem compelling.

Stage Two: General Principles for Judging Relative Inequality 

Temkin asks us to consider a “group of artificially simple worlds….each of which consists of two groups of people, the better-off and the worse-off.”  We’re asked to imagine the worlds in a sequence such that “in the first world there are 999 people better off and 1 person worse off, in the second 998 better off and two worse off, and so on….[leading to the last world in the sequence where] one person is better off and 999 are worse off.”  Temkin asks: Do these worlds get better or worse with respect to inequality?

Again, how we answer this question depends on the principle underlying our judgments about relative inequality, and once again, Temkin says, there are several plausible but competing principles from which we might choose.  These are:

Maximin Principle: “How bad a world is with respect to inequality will depend upon how bad the worst off group in that world fares with respect to inequality…” (this principle implies that things get better as we follow the sequence).

Additive Principle: The more people who have complaints with respect to inequality, the worse the world is with respect to inequality (this principle implies that things get worse as we follow the sequences).

Weighted Additive Principle: Complaints are weighted, such that more serious complaints count for more when adding up all the complaints (Temkin says this principle implies that things get worse).

Deviation Principle: The greater the deviation from absolute equality, the worse the world is with respect to inequality (this principle implies that things first get worse, and then better).  

As with the principles concerning the determination of individual complaints, the Egalitarian cannot simply embrace each of these principles.  Since they yield different verdicts in different cases, they are inconsistent.  But which should the Egalitarian choose, and why?

This essay was very thought-provoking.  I’m looking forward to reading more Temkin in the future.







This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

6 Responses to Temkin on Inequality

  1. David #2 says:

    I think Temkin’s shirt, stating “a>b>c>a”, points to a serious flaw in his assertion number (1). Why should it be necessary that there be any ordering principle on the set of all possible worlds in which things are ‘bad’? Why can’t there be one or several tests of “unfairness” that all yield well defined results (we know what we mean) without any notion of the unfair worlds being more or less bad than each other?

    And why should I be concerned if one test does not detect “unfairness” in a scenario where another test might? Can’t I hold a perfectly sensible worldview in which there are many ways in which life can be “unfair”, all of which are bad, but without worrying about which ways of being bad are better or worse?

    Am I missing something, or doesn’t statement (1) amount to saying that a well-defined concept of “bad” requires an ordering principle based on “badness”? I don’t think I agree with that…

  2. David says:

    Hi David #2,

    Nice point! I’m not sure I know how Temkin would reply, but I’ll take a stab….

    You write: “Can’t I hold a perfectly sensible worldview in which there are many ways in which life can be “unfair”, all of which are bad, but without worrying about which ways of being bad are better or worse?”

    That sounds right, but Temkin’s claim is not that in order to claim to know what we’re talking about when we talk about ‘inequality,’ we must know how to ‘rank’ or ‘order’ the badness of inequality against the badness of other things, like non-inequality-related unfairness, or toothaches, or whatever. His point is that we should be able to rank or order distributions according to badness with respect to inequality.

    Schematically, I think his point could be put like this (and Temkin does seem to like schemas!): If I know what f is, then I should be able to make comparative judgments, faced with two or more ‘worlds,’ about the extent to which f obtains in those worlds. If I cannot, as Temkin thinks is the case when I evaluate and compare worlds according to the amount of ‘inequality’ they contain–Temkin thinks I’ll find myself wanting to say incompatible things about their relative levels of inequality–then I should consider the possibility that my concept of f is not as clear as I might have thought.

    Does that make sense?

  3. Josh says:

    Okay so the idea is we would need to have a theory of inequality that would tell us which of those worlds was more unequal than which, right? But can’t we just say they’re all unequal to the same degree, and then choose among them for other reasons? Like propose a theory that indicates what sorts of inequalities are better or worse than what other sorts of inequalities?

    I always thought Rawls’s derivation of the use of maximin reasoning in the original position, and also the notion of “chain connection” he uses were relatively arcane, but don’t they have something to do with this issue? Rawls argues we’d opt for a situation where the “worse off” group was as well-off as possible, given the options. We wouldn’t necessarily resolve where they stood in relation to the better off group though, or even how big each group was.

    • David says:

      Hi Josh,

      Yeah, I think I see your point, and I think it was probably what David #2 was driving at. Why not just say: ‘Look, there are many different ways of measuring ‘inequality,’ and depending on which measure we use we may arrive at different verdicts about cases. But we shouldn’t get caught up in arcane arguments about which of the various measures that recommend themselves to us is the One True Measure. Rather, we should focus on establishing which measurable form of inequality is the most urgently in need of rectification from the standpoint of justice. And here we may adopt Rawls’s maximin principle–inequality is objectionable from the standpoint of justice if, and insofar as, we could improve the lot of the least advantaged members of society by reducing the inequality.

      I have to admit that sounds awfully sensible to me, so I’m wondering now why I wasted a couple hours outlining Temkin’s argument (!). But maybe he would respond to this as follows–I can’t say for sure. Rawl’s maximin principle appeals to the idea of ‘reducing inequality,’ but how we do that is going to depend on our concept of inequality. For example, should we work to reduce the number of people whose welfare is below average for the society in which they live? Or, rather, should we work to reduce the ‘gap’ between the best-off and the worst-off? Or, rather, should we work to reduce the number of people who belong to the worst-off group? Temkin claims that depending on which of these conceptualizations of ‘inequality’ we adopt, plans and policies for ‘reducing inequality’ will look different. So we have to make a choice, and I’m not sure it will do to just appeal to Rawls’s principle here as the basis for the choice. After all, dyed-in-the-wool egalitarians might not like Rawls’s principle, insofar as it would seem to permit, and perhaps even mandate, increases in inequality (by some measure) under conditions where this led to increases in the welfare of the least advantages members of society.

      Anyway, I’m beginning to question my grasp of Temkin’s big picture, so I’ll stop.

      • David #2 says:

        What you said above does make sense, and I think I do grasp the argument. It’s worth considering, so I don’t think you’ve wasted your time.

        I don’t think restricting the ordering principle to ‘ordering of unfairness with respect to unfairness’ resolves my dislike of making ordering necessary for a usable concept of fairness. What if I’m talking about something closer to binary – the unfairness of a situation either attains or doesn’t. I don’t think there needs to be clear and ordered variation by degree in between ‘fair’ and ‘unfair’.

        If I remember reading Mill’s Utilitarianism hrmphrumph years ago, I think trying to quantify things by the number of people affected leads to all kinds of very undesirable logical consequences, like wanting to murder people and harvest their organs or some other such nonsense. So how about we eliminate injustice by *eliminating* it, as in, creating states of the world where our various ideas about unfairness simply don’t attain at all (as opposed to worrying about whether or not they attain more or less)?

  4. Josh says:

    The relationship between Rawls’s argument and the rest of the debate about “egalitarianism” (and what people seem to assume about it) has always perplexed me a bit. Rawls is lot less concerned with the conceptual analysis of “equality” and “inequality.” I think it might even be right to say that the Rawlsian account doesn’t want to “reduce inequality.”

    Instead of “reducing inequality” the Rawlsian account acknowledges that inequality is inevitable, and draws a distinction between good and bad inequality. A surface-level reading might lead people to think what he’s saying is “we need to reduce inequality to help the least well off.” Instead his argument is more like “social cooperation and the distribution of good is ALWAYS unequal. Or, maybe now and then, it produces equality too, but that’s not better or worse than systems that produce inequality. The issue is not that inequality is a problem per se. The issue is instead that we need a criterion to discern among the different kinds of inequality. Inequalities based on justice are better than inequalities based on other things. Therefore justice is the best way of confronting inequality.”

    In short – projects focusing on “reducing inequality” succumb to the problem Temkin points out. Projects which instead develop tools for understanding which kinds of inequality are better than which others do not.

    You are right that dyed-in-the-wool egalitarians might not like Rawls’s picture. It’s always seemed to me more centrist than most seem to realize: it’s about which inequalities help the least-well off, not reducing the inequality that made the “least-well off” in the first place. Hence G.A. Cohen’s disagreement: something like, SO WHAT if inequalities create incentives for things like doctors, and that makes our society better than it would have been without those incentives; why not just level the playing field and figure out another way to get doctors? Why take the selfishness that the incentives argument presupposes for granted?

    Or another paraphrase – Rawls is not a dyed-in-the-wool egalitarian. One of the reasons he’s not may be to sidestep the more conceptual-analysis-type debate being one seems to entail.

Leave a Reply

Your email address will not be published. Required fields are marked *