The Mysteries of Labor Composition

Print
Monday, 14 March 2011 11:10

David Leonhardt had a blogpost last week that left some of us here at CEPR stumped. It had two graphs, one on top of the other, showing patterns in wages since the start of the recession. The top graph showed wage gains by educational attainment. This showed that college grads had an increase of about 1.5 percent in their real weekly earnings, while everyone else saw modest declines. The second graph showed real wage growth by income cutoffs. Those at the 90th percentile saw real wage gains of 8.0 percent, but everyone else also saw modest wage gains as well.

At first glance, these seemed inconsistent and we thought that Leonhardt had made a mistake. After checking his data, we saw that he was exactly right. The explanation was a change in the composition of the employed workforce. There was a sharp drop in employment among workers without high school degrees and those with just a high school degree between 2007 and 2010. On the other hand, the number of people employed who had advanced degrees actually increased slightly.

 

educational_mix2_31489_image001

Source: Bureau of Labor Statistics.

 

What happened here is that the change in composition means that much of the bottom portion of the workforce is no longer employed. Therefore the 90th percentile worker in 2010, might have been the 92nd percentile worker in 2007. And, in 2007 the 92nd percentile worker earned 6.3 percent more than the 90th percentile worker.

So, what looks like a big rise in wages for higher-end workers is in fact the result of comparing different workers. This is worth keeping in mind. The wage growth at the middle and lower-end of the income distribution in the late 90s looks even better when we consider that many less educated workers found jobs in this period.