"Awesome liberal" Will Wilkinson was kind enough to respond to me and Matt Yglesias on the question of whether the diminishing marginal utility of consumption is a good argument in favor of redistribution, so I'd like to toss a couple of thoughts back in his direction.
Here's the background: one of the points in Will's original paper is that inequality in the United States doesn't look so bad when you stop talking about income inequality and start talking about consumption. The rich might have $11,000 fridges but, Will argued, they aren't much better off than the poor shlubs who spend $350 at Ikea. Matt and I pointed out that in making this point, Will seemed to be rediscovering a classic argument in favor of redistribution-- namely, that a rich person doesn't get as much utility from an additional dollar as a poor person. This is the basic concept of diminishing marginal utility: the next dollar of consumption won't get you as much satisfaction as the previous one did, and the one after that will get you even less, and so on.
Will made a couple of points in response to this, which can be summarized as (1) if you want to maximize utility, it's not necessarily you'd want to redistribute money from the rich (you might, for example, want them to save more); and (2) utilitarianism is "false," so we don't want to maximize utility, anyway. A couple of thoughts on these points:
1. I am totally open to the suggestion that redistribution might not be the thing to do when confronted with diminishing marginal utility. I confess haven't read the paper Will suggests on this point. (I'm in India with spotty Internet access, writing this post on a sticky note to publish when the wireless comes alive again.) But even if I accept Will's argument sight-unseen, it does not change the most important fact about the debate, which Will doesn't seem to dispute: it is very unlikely that the current consumption distribution is utility-maximizing.
So, sure, the utility-maximizing thing to do might be to have the rich save more. (I tend to think we should save more and can redistribute more, and don't think those things are necessarily mutually exclusive.) Or, the utility-maximizing thing to do might be to abandon this whole business of free exchange and consume government-provided pills that simulate the experience of hedonistic fulfillment. Open empirical question! But what I don't see Will disputing is that it would not be hard to improve on the present, even if there's some chance that tax-and-transfer redistribution isn't the best way to do it. Redistribution would still be an improvement.
2. I agree with Will that a strong utilitarian position is unacceptable (though I wouldn't call it "false") because it's not concerned with the procedures -- fair or unfair? moral or immoral? -- through which utility is obtained. And in the specific case of free markets and government redistribution, I would go one step further than most liberals: I believe there is always some inherent cost in limiting people from freely exchanging goods and services, just as I think there is always a cost in limiting people from freely exchanging words and ideas. It seems to me that there are very strong moral arguments in favor of a free market, in addition to the typical instrumental ones. So perhaps Will and I agree on that.
But let's not get carried away. There is no need to choose between Jeremy Bentham and Robert Nozick, and the fact that we might reject strong utilitarianism does not mean we stop caring about outcomes and start caring exclusively about rights. My sense is that most people think there are rights that hold irrespective of outcomes, and there are outcomes so dire that they always trump rights. Weighing any particular outcome ("lots of people don't have health-care right now") against any particular right ("my income is private property") doesn't have even the slightest veneer of an exact science.
(Jeremy Bentham, Photo credit: http://search.creativecommons.org/)So objecting to income redistribution because you put a priority on economic rights over economic outcomes (as Will does) strikes me as both perfectly fine and entirely optional. But I suspect Will and I agree on that too.