In the second volume of his Law, Legislation and Liberty, which is entitled The Mirage of Social Justice, Hayek makes the following claim:
Though it might seem reasonable so to frame laws that they will tend more strongly to improve the opportunities of those whose chances are relatively small, this can rarely be achieved by generic rules. There are, no doubt, instances where the past development of law has introduced a bias in favour or to the disadvantage of particular groups; and such provisions ought clearly to be corrected. But on the whole it would seem that the fact which, contrary to a widely held belief, has contributed most during the last two hundred years to increase not only the absolute but also the relative position of those in the lowest income groups has been the general growth of wealth which has tended to raise the income of the lowest groups more than the relatively higher ones. (p.131, emphasis added)
In the footnote to this paragraph, Hayek adds:
The chance of all will be increased most if we act on principles which will result in raising the general level of incomes without paying attention to the consequent shifts of particular individuals or groups from one position on the scale to another… It is not easy to illustrate this by the available statistics of the changes of income distribution during periods of rapid economic progress. But in the one country for which fairly adequate information of this kind is available, the USA, it would seem that a person who in 1940 belonged to the group whose individual incomes were greater than those of 50 per cent of the population but smaller than those of 40 per cent of the population, even if he had by 1960 descended to the 30-40 per cent group, would still have enjoyed a larger absolute income than he did in 1940. (p.188, emphasis added).
I think Hayek is comparing the mid-point (average income) of the 50-60 decile with the mid-point of the 30-40 decile. I couldn’t find data for 1940 on deciles, but the data on quintiles for the U.S. are readily available back to 1947. The picture they present is very interesting.
If you were at the top end of the 60th percentile in 1947 (the red line) and by 1967 you had dropped to the top end of the 40th percentile, you were still doing better. The same is true comparing 1977 with 1957, but not if we compare 1987 with 1967, 1997 with 1977, or 2007 with 1987. Starting roughly in 1970, what we see is that the upper limits of the 80th and 95th percentiles grow more quickly than do the upper limits of the 60th and 40th, while the 20th is essentially flat. What this means is that if you were dropping in the income distribution after 1970, from 60th to 40th, over a twenty year period, you were not doing better anyway. To put this the other way round, from the 40s to the 70s, people staying at the 40th percentile were doing better than were people at the 60th percentile twenty years before. After 1970, not.
This pattern in the U.S. income distribution doesn’t answer any fundamental normative questions. But it does help explain how someone like Hayek could have been a bit sloppy in his discussion of principles of social justice – how, for example, he could move very quickly from talking about inequalities benefitting all to maximizing the overall level of opportunity to maximizing overall (i.e. total) income. It doesn’t matter (as much) what exactly is the right principle of distributive justice if all reasonable principles suggest that the current system is better than any other feasible system. Before 1970, it was at least plausible that that was the case. After, not so much.
For more on this issue, see the lecture I recently gave on Hayek and Rawls.