John Roemer’s version of “equality of opportunity” seems like a reasonable one to me. I put it that way because there may be other reasonable versions. There is no reason why “equality of opportunity” should be uniquely defined. Roemer’s account has the virtue that it highlights the difference between “real” equality of opportunity and the mere absence of formal obstacles to individual achievement. Otherwise one is reminded of Anatole France’s line about freedom being the right of the rich as well as the poor to sleep under the bridges of Paris. Equal opportunity could be the right of the poor as well as the rich to sleep at the Ritz.

But I do wonder what the point of the exercise is supposed to be. It does not look like a practical recipe for income redistribution. (I will stick to income. Roemer’s choice of feelings of success as a criterion seems very odd. Suppose that I am predisposed, genetically or by influences I cannot control, to feel like a failure unless I become as rich as Ross Perot. Should people of more modest ambitions be taxed to help me?) The first source of impracticality is simply the enormous number of “types” across which equalization of opportunity is supposed to occur. Any of us could think of dozens of (more or less independently varying) factors that define the circumstances within which a person conducts her life. Suppose there are only six such factors: ethnicity, height, comeliness, gender, birth order, region. (I am being only slightly frivolous. There is a story that a major business school found height to be the personal characteristic most highly correlated with the postgraduate success of MBA students; and there is good evidence that earnings are correlated with good looks, for men as well as women.) Suppose these factors have, on average, four “levels” — only two for gender but more for some of the others. Then there are already 4096 types of people in principle. I am willing to make empirical compromises, but I think this difficulty will not go away.

A second practical difficulty arises because each person’s income follows a lifetime trajectory. At what stages is equalization supposed to occur? Surely not month by month or even year by year: the redistributional bureaucracy would be too large and too intrusive. But equalization at longish intervals introduces still other problems. It creates an incentive for each person to position himself for a big transfer at just the right time. Who is so handsome as to be unable to look ugly on any given day? What is more important, future earning power depends on past income, and not in the same way for everyone. So the timing of transfers is an important matter, and any simple rule will do an injustice to some.

A third set of complications arises because incomes, even earnings, are inevitably vulnerable to random fluctuations. Some people choose riskier careers than others, sometimes in ignorance but sometimes for good and sufficient reasons. The interaction of chance with the arbitrary timing of transfer payments can be a source of both trouble and injustice.

Some of these practical difficulties can be dealt with, but only by a complicated bureaucratic system. Roemer is far too good an economist not to have thought about them. Maybe the proper conclusion is that he does not mean to be proposing a practical redistributional scheme. Perhaps the purpose is only to call attention to the idea that the absence of formal barriers is at best a very partial step toward equal opportunity, and to show that there is no need for equal opportunity to submerge individual responsibility. If so, then the point has been made, though it could have been made less elaborately.