About 6.2 million Americans, 45.1 percent of all unemployed workers in this country, have been jobless for more than six months - a higher percentage than during the Great Depression.
I read this as saying that, relative to the Great Depression, a higher percentage of the unemployed have been out of work more than six months, as opposed to a higher percentage of total workers have been out of work for more than six months. Still, it is a telling statistic.
The argument most often used against the structural unemployment hypothesis is that unemployment is too widespread. How can unemployment be structural if jobs are being lost in so many sectors?
The argument against the aggregate demand hypothesis is that the jobs lost are not coming back. How can it be aggregate demand if workers have to find new jobs in new industries?
I can imagine that it was easier during the Great Depression for unemployment spells to be shorter than they are today. Back then, whatever jobs there were often required manual labor, and work could be given on a temporary basis. So, you might be unemployed for a few months, have a job for a few months, then be unemployed again, etc. Today, jobs require more skills and more on-the-job training. The fixed costs of hiring a worker are more significant. Therefore, fewer short-term jobs are being created. Unemployment is much more likely to be persistent than episodic.
On the long-term job structure, Mark Thoma found this chart. I am not sure that I find it so readable. But you can see that the big occupations used to be "farmer" and "farm laborer." The advent of the tractor lowered the price of agricultural produce to the point where tenant farming became uneconomical and the marginal farm laborer could not necessarily add enough value to feed himself. In my view, that is what made the Great Depression so severe. Up until that point, farm labor had always been an occupation of last resort. But in the 1930s, farm labor began to fit Tyler Cowen's Zero Marginal Product category. The Dust Bowl exacerbated that problem.
Median female income tracks real GDP per capita much more closely than does median male income. It's unclear which, if any, of the above explanations are consistent with this finding. Increasing inequality, for example, predicts an increasing divergence in real GDP per capita and female median income but we don't see this in the graph
To me, the explanation for the data in Alex's post is that the job structure has changed to diminish the relative importance of physical strength. This brought more women into the labor force, and it reduced wages and employment prospects for a significant proportion of the male population. This changed marriage, as Wolfers and Stevenson have pointed out, from something based on production complementarity (women do housework, men do factory or farm work) to consumption complementarity. This results in more assortive mating by income, creating lots of household inequality.