A crucial feature of the method used is that it allows jobs to change classification over time. The declining share of university graduates in non-university jobs could thus have been driven by two forces: university graduates could be flowing into university jobs; they could hold the same type of jobs, but jobs could be changing classification (from non-university to university). This section directly checks how much the results are driven by the changing classification of jobs. Consider
(GH):
• The actual share of university graduates in non-university jobs in period t: sht,baset;
• The hypothetical share of university graduates in non-university jobs in 1999 if jobs had kept their
classification: sh99,base86.