The piece is by Christopher L. Smith, at the Board of Governors, and from this work we learn a few things:
1. Once we adjust for demographics, labor market polarization is a stronger effect than the raw data themselves suggest.
2. Polarization is driven both by more people leaving the middle income category and fewer people entering it.
3. The actual changes in polarization often arrive during cyclical downturns but then stick.
4. “…the share in middle-type jobs has fallen for all types of workers.” (wrt education) The declines have been the worst for younger workers and for non-college workers. Even for college-educated workers, the share employed in low-end jobs has risen since the early 2000s.
5. The inflow into middle-type jobs, from the previously unemployed, is declining steadily with time.
6. Unemployed, previously middle-type workers are tending to remain non-employed rather than to trend into lower-type jobs.
7. The rate at which persons in low-type or (especially) middle-type jobs transition into high-type jobs is rising over time.
8. Yet the rate at which persons in low-type jobs transition into middle-type jobs is falling over time.
9. The core trends on labor market polarization date from the mid-1980s.
For the pointer I thank Claudia Sahm.