"Some Personal Reflections on Economic Well-Being and Income Distribution," in The American Economy in Transition, ed. Martin Feldstein (Chicago: University of Chicago Press, 1980).

Excerpt:

It is my understanding, from surveying various studies of trends in income distribution in the United States over the past three decades, that economists have found very little significant change to have taken place. There does seem to have been a slight increase in the proportion of national income received by the very poor, a slight decrease in the proportion received by the very rich. What goes on in between is such a complex muddle that economic analysis can tease few unquestionable inferences from the data. Moreover, the very methodology of studying income distribution has, over these decades, become ever more controversial. Just what is to be included in the concept of “income” becomes less clear every time a new governmental “entitlement” program is launched (whether it involves food, housing, medicine, or whatever). And it has become ever more apparent that in order to take account of normal age differentials in earnings, of changing demographies, and of economic mobility (both up and down), the distribution of “lifetime earnings” would give us a far more valid report than any cross-sectional survey at a moment in time. The trouble is that economists have not come up with any accepted procedure for measuring any such distribution of lifetime earnings, and there are even some grounds for thinking they never will.
Does it matter? What, precisely, is the point of all of these studies and of the interminable controversies they generate?

Online:
National Bureau of Economic Research [pdf]
Foreign Affairs