Via Greg Mankiw comes this suggestion of Yale economist Robert Shiller reported in Tax Notes Today:

The IRS should be instructed to automatically adjust tax rates to keep economic inequality from getting worse, according to a new proposal outlined by Robert Shiller, a Yale University economics professor.

“We have a serious problem, and it’s a problem of growing inequality,” Shiller said on December 6 at a Library of Congress discussion in Washington. Shiller developed the proposal with Len Burman, director of the Tax Policy Center, and the two are planning to write a book on the idea.

“We need a standard or principle of income inequality. We don’t have one now,” he said. Inequality provides motivation to work harder and benefits hard work, he said, so “we do want some inequality, but we don’t have any clear idea about where we’re going and what is appropriate.”

The standard, which Shiller calls “inequality indexation” of the tax system, would instruct the IRS to adjust brackets and rates whenever inequality worsened beyond an agreed-on level.

The question that leaps to mind is: why?

Shiller’s proposal illustrates the extent to which policy is a normative enterprise. In order to defend Shiller’s “inequality indexation,” you need some principled basis for believing that growing income inequality is a “serious problem,” and an explanation of why it is a problem. How greater inequality per se is a problem strikes me as utterly mysterious. There are many possible causes of income inequality. Some of them reflect injustices in the system, or barriers to the development of adequately fulfilling human lives. But in this case, inequality is a side-effect of some other injustice, and isn’t really the problem. And some causes of inequality don’t reflect injustice of any kind. How a change in the pattern of the income distribution over time can be a serious problem by its very nature is baffling.

Take an highly idealized example. Suppose for a moment that all of a society’s basic laws, institutions, and rules governing market exchange are fair. Everyone starts with a perfectly equal endowment of capital of various kinds, but with different preferences and goals. There will be, say, 1,000 rounds of exchange. Now start the clock. Some people will sit out some rounds. Some people will participate every round. Given different preferences, people will be motivated to invest in different forms of knowledge and skill, pursue different kinds of work, and purchase different kinds of goods and pick different kinds of trading partners. After round 1,000, stop the clock. Now count everybody’s money income over the period. There will be a certain amount of inequality in income. Start the clock again, and stop it after another 1,000 rounds. Suppose income inequality grew. What does that tell us?

If your answer is “It tells us that the laws, institutions, and rules” were not fair after all, then the question is “How does it tells us that?” And then, if the response is, “Because inequality is unfair,” that’s just begging the question. Justice and fairness, if those ideas mean anything, have something to do with giving people what they are due. If the basic rules aren’t keeping people from getting what they are due in each voluntary exchange (which would not have occurred unless the terms were satisfactory to each party), and each got what they were due every round in which they participated, then, when we stop the clock, what each ends up with over the period–their income–is just the sum of what they were due. But no one was due any particular sum. A fortiori no one was due a sum that is a particular ratio of someone else’s sum. So a change in the ratios between incomes across periods is irrelevant.

OK. Now suppose there are two social systems, A and B, with different basic rules running side by side, but in isolation from each other. A has lower inequality in each period, and lower inequality growth between periods that does B. Is A in any sense better? Maybe, maybe not. Suppose that the average income in each decile in B is higher than in A. Wouldn’t the denizens of A rather live in B, where there is greater inequality? I would. Indeed, this is the sense in which I think things like aggregate income is a matter of justice. People are due a system under which they can do as well as possible. If there is some alternative set of rules under which everyone could expect to be better off than in the status quo, failing to transition to that alternative system of rules would be a mark of injustice. Justice may require us to shoot for a system with greater inequality.

Shiller says “we do want some inequality,” since leveling would kill effort, leaving everyone worse off. But I don’t think we really want “some inequality.” We want a system in which everyone is doing as well as possible, and inequality is going to be a side-effect of that. Shiller needs to say why we should want less inequality. I am willing to believe that rising income inequality does reflect some injustices in the system. Perhaps some of the rules that regulate the governance of corporations. Or elements of the electoral and regulatory system that enable predatory rent-seeking. Or the system of monopoly public provision of education that systematically disadvantages certain classes of citizens on the basis of morally arbitrary characteristics, like the property tax rate in their neighborhoods. Or price floors and labor regulations that exclude low-skilled workers from the labor market. But in each case, inequality is the symptom, not the disease. The attempt to “correct” increases in inequality through the tax system is completely arbitrary, beside the point, and almost certainly itself unjust, if rising inequality is a side-effect of deeper injustice in the structure of our institutions. And if it is not a side-effect of injustice, but just a side-effect of exchange according to just rules, then it is a non-issue.