My recollection is that the authors looked at the predictive power of a father's earnings on his son's earnings, but rather than measure changes in absolute (or inflation-adjusted) earnings, they measured earnings relative to the society (Sweden or the U.S.) as a whole. In other words, if my father was in the 50th percentile for earnings, I could use their "mobility" estimate to determine my odds of making it to the 75th percentile (or down to the 25th percentile).
That doesn't seem like a useful statistic on which to make qualitative comparisons when you're talking about societies with two very differently shaped income distributions (as the authors acknowledge is very much the case with Sweden and the U.S. on the freebie first page).
Suppose the middle 60% of Swedish males earn between $25,000 and $35,000 annually. A jump from the 20th to the 80th percentile (or vice versa) in absolute terms isn't that impressive. If the income curve gets narrow enough, any impact your parents have will get lost in the noise and you'll have a "perfectly mobile" society in which everyone makes almost exactly the same amount of money. But how is that mobility?
In contrast, if the earnings difference between the 20th percentile and the 80th percentile is, as in the U.S., a factor of four+, then moving from the 20th to the 80th percentile represents dramatically more "mobility" than in the Swedish case. AFAIR, the 1997 study would treat those movements equivalently.
http://www.jstor.org/pss/2951338
My recollection is that the authors looked at the predictive power of a father's earnings on his son's earnings, but rather than measure changes in absolute (or inflation-adjusted) earnings, they measured earnings relative to the society (Sweden or the U.S.) as a whole. In other words, if my father was in the 50th percentile for earnings, I could use their "mobility" estimate to determine my odds of making it to the 75th percentile (or down to the 25th percentile).
That doesn't seem like a useful statistic on which to make qualitative comparisons when you're talking about societies with two very differently shaped income distributions (as the authors acknowledge is very much the case with Sweden and the U.S. on the freebie first page).
Suppose the middle 60% of Swedish males earn between $25,000 and $35,000 annually. A jump from the 20th to the 80th percentile (or vice versa) in absolute terms isn't that impressive. If the income curve gets narrow enough, any impact your parents have will get lost in the noise and you'll have a "perfectly mobile" society in which everyone makes almost exactly the same amount of money. But how is that mobility?
In contrast, if the earnings difference between the 20th percentile and the 80th percentile is, as in the U.S., a factor of four+, then moving from the 20th to the 80th percentile represents dramatically more "mobility" than in the Swedish case. AFAIR, the 1997 study would treat those movements equivalently.