I think the work by Paul Romer is fundamentally flawed from the outset.

Should the Government Subsidize Supply or Demand in the Market for Scientists and Engineers?

Paul Romer http://www.nber.org/~confer/2000/ipes00/romer.pdf

Why do we need to train more scientists and engineers when we are not gainfully employing the ones we have now?

As you know, the NSF gathers statistics on the employment of S&Es [see References below]: The NSF statisticians put a positive spin on the fact that there are "Twelve million workers have an S&E degree as their highest degree and 15.7 million have at least one degree in an S&E field....However, only 4.9 million were in occupations formally defined as S&E." See:

http://www.nsf.gov/statistics/seind06/c3/c3h.htm

This would seem to indicate a vastly underutilized S&E workforce-by some 10 million people who already have S&E degrees but who are not actually working as S&Es. Even those 4.9 million S&Es employed as S&Es are not usually employed in an efficient manner. Most S&Es, even at the PhD level, are in such plentiful supply that their time is often spent doing work that could be better taken up by tech assistants and administrative assistants. Employers know this, but it is more cost effective to hire more advanced degreed S&Es and far fewer support staff than the other way around.

When was the last time you saw a doctor or a dentist answering the phone, filling out paperwork, returning phone calls, fixing the Xerox machine, etc.? Never, because he has a staff that preserves his time for seeing patients. This is an extreme example, but I feel S&Es are completely on the other side of the spectrum concerning the utilization of their time because there is such a glut of S&Es. The underutilization of S&Es, relative to other professions, is obvious to those on the inside, but not to some economist writing a "working paper".

I think some of the statements and statistics cited above can be used to make the case that it would be difficult for a "shortage" of S&Es to materialize anytime soon (within 5-7 years), even if hiring were to pick up. A more efficient utilization of those with existing S&Es, would take up increased demand for the foreseeable future.

Here is an interesting graph from the NSF:

http://www.nsf.gov/statistics/seind06/c3/fig03-13.htm

Note the strong uptick in unemployment for those with 30 years since their degrees (people in their fifties). It might be interesting to plot unemployment vs. salary demands for S&Es.

These readily available statistics from the NSF are also ignored by the NAS/NRC in their "Gathering Storm" report. This is part of the case that needs to be made to Congress, not the ones by Paul Romer.

Bob

References:

For link to NSF's Science and Engineering Indicators:

http://www.nsf.gov/statistics/seind06/front/about.htm

Here is a great list of links to very useful figures:

http://www.nsf.gov/statistics/seind06/figures.htm