This is exhibit 1 in the report that Noah Smith linked to yesterday. Anyone trying to make a labor market policy claim on the basis of a single supply number and a single demand number should raise eyebrows. Supply and demand is a schedule. If actual demand exceeds actual supply it suggests we have a shortage. That requires an answer to why we have a shortage. What is artificially keeping STEM wages low? These numbers come from a report from the Georgetown Center on Education and the Workforce that came out with much fanfare (I believe) last summer. I was at a conference where Tony Carnevale presented it around the release. It (the Georgetown report) does a lot of the same thing. It projects "supply" (a single number), projects "demand" (a single number) and then declares a shortage. Policymakers love this report, but I know a lot of economists are skeptical.
According to the footnote the supply number comes from the IPEDS. I use IPEDS data a lot... nothing in the IPEDS will tell you how many STEM degrees we'll be producing in 2018. There's some projections going on here. Growth in this number is probably safer to project, though. I'm skeptical of the demand number and I really don't like treating these as numbers rather than schedules.
When you think about the STEM workforce from a market perspective and you see these claims about shortages you start asking questions about what would be preventing labor market adjustment. Research by economists back to Blank and Stigler's study in the fifties have pretty much had the same answer: nothing. STEM labor markets are responsive to wage signals just like other labor markets. There are adjustment lags, it's true, because of longer training periods. But we do not have any evidence of persistent shortfalls of STEM workers. My take is that this means that rather than dumping one specific group of workers on the labor market or engaging in manpower planning with temporary visa programs, we should have an immigration policy that welcomes and settles lots of new Americans regardless of how they fit into a policymaker's manpower plan.
Hal Salzman (Rutgers), Lindsay Lowell (Georgetown), and I are waiting to hear back on a grant (should hear back within a week now) to study "loose coupling" in the STEM workforce. A lot of people assume this is a straightforward pipeline, but it's not true. A lot of STEM occupations are filled by non-STEM degree holders, and a lot of STEM degree holders aren't working in STEM jobs (in many cases because they can't get a STEM job at a good enough wage). A lot of this "just give us more STEM graduates!" view is contigent (or at least has historically been contingent on) this linear pipeline view of this labor market.