From a speech by Electronic Arts CEO John Riccitiello, reported by Andy Katkin.
I see an unfortunate gap between the economic performance of key parts of the tech sector, and the rest of our economy. I see two economies on such radically different trajectories that it's hard to believe they co-exist in the same country.
In tech -- Silicon Valley, North Carolina, Austin and Boston - we are witnessing the most dynamic period of creative destruction and economic construction in history. And that construction is creating new and very high paying jobs.
Economic disruption is a constant everywhere....but today in tech, economic cycles turn, not over multiple years, but in months....large companies and fortunes are built, erased, and built again with frightening speed.
In tech, the most impressive and profitable operations are constantly being destroyed...and replaced with even better businesses. And these businesses are hiring a rapid pace. Job openings exceed the number of qualified workers and wages are skyrocketing. And companies with only a few years in operation are assigned multi-billion dollar valuations.
Also, Tyler Cowen links to a paper by David Autor and others that looks into the way trade with China affects specific labor markets. (This is actually a revised version of an earlier paper.)
If "labor" were a homogeneous, fungible aggregate, we would see unemployed manufacturing workers taking the high-paying jobs that tech companies are having difficulty filling. Thus, the assumption of a single labor market is clearly false.
Even in the DMP type of search models, this sort of structural mismatch is not found. It is as if there is a "matching technology" that can fit blue-collar pegs into high-tech holes, but the process takes time and effort.
A macroeconomist would argue that structural unemployment always exists. It represents a sort of background noise. Models that assume a single aggregate labor market are only approximations, and if the economy acts as if there were a homogeneous labor market, then this approximation causes no more harm than any other simplifying assumption in economics.
It's an empirical question. My view is that the variation across sectors and occupations is more dramatic and important than the variation within sectors, even for the period 2007 to the present. That is, for some fairly large categories of workers you will see a higher unemployment rate in 2007 than for other categories in 2011, because the structural factor swamps the supposed aggregate-demand factor.