Monday 24 June 2013

More scientists?

I like science. But I'm not convinced that pouring money into producing more scientists is the most effective way of generating the great things that come from science.

Economists typically recommend "keyhole" interventions. If there's some market failure, intervene at the most direct level to fix it. It's plausible that basic scientific discoveries - the kind that aren't really patentable but that can yield all kinds of later commercial applications - are underprovided relative to some ideal. There are then a few policy options:
  • Award prizes for those making worthy discoveries.
    • But, if the basic research is expensive and if research teams have a hard time getting investor funding for prize-seeking, then this may still underprovide discoveries. Prizes are great when you know what achievement you'd like to fund but you don't know who's best-placed to provide it.
  • Award grants to research teams likely to provide discoveries.
    • This requires the granting agency to be able to pick winning teams and, in small countries, can yield nasty procedural tradeoffs between nepotism (the awards committee gives money just to their friends or to people doing politically favourable work) and administratively expensive application methods
  • Be really generous with baseline University funding, then revise slowly over time to focus funding towards institutions that produce a lot of discoveries so that Universities sharpen incentives.
    • We've moved toward this in New Zealand with PBRF, but the whole process is ridiculously administratively burdensome, the amount of money at stake is pretty small relative to overall University budgets, and because new and important discoveries are pretty rare, the whole thing seems to reward number of journal articles. If one University produced two field medalists, that would get it a couple of PBRF As, but plenty of other places will get plenty of As without field medals. There's basically a big upperbound truncation problem where truly stellar work - the kind that should wind up getting the really big prizes - is under-rewarded. 
  • Encourage lots of kids to take science degrees. Encourage the Universities to expand their science offerings by paying them extra for kids taking science courses and for degree completion in the sciences as compared to arts or commerce. 
    • Increasing the supply of scientists could reduce the cost of scientists and thereby increase the supply of discoveries from those places that hire scientists. In a small country, you cannot really have much of an effect on equilibrium wages in science as you'll just induce post-degree out-migration. And to the extent that science grads get hired in applied shops that, while great, largely internalise the benefits of what they're doing, you've not done as much to get the new discoveries.
I expect that optimal policy would involve some combination of the first three mechanisms. Prizes in combination with baseline university research funding would solve some of the problems we'd otherwise have in providing extra rewards for truly top-notch stuff. And grants can help if there are particular things you want done and you think you know who can do it. If those together work to build demand for scientists, then that can automatically start pulling more kids into science when they see rising salaries and better job prospects.

What do we get when we push a lot of kids through general science degrees instead? Andrew Norton at The Gratten Institute has a few numbers for Australia.
The annual Graduate Destination Survey (GDS) of people with recently completed bachelor degrees consistently finds that people with science qualifications have above-average difficulty finding work. The only exception is for people with degrees in the geological sciences.
Three years on from the GDS, the Beyond Graduation Survey shows that the job outlook of science graduates improves with time. After a slow start their employment rate is only slightly below average.
But this is not evidence of strong demand for specific scientific knowledge. Only 57% of science graduates say their qualification is important or a formal requirement for their job. It is the second lowest match between qualification and discipline.
Norton concludes:
Nobody doubts that science and maths skills are important to Australia’s future. In some specialised areas, employers might struggle to find suitable graduates. But no enrolment or employment data suggests that we are headed for general shortages of science and maths graduates.

The science applications boom may turn out to be a higher education bubble waiting to burst. If it does, thousands of intelligent and capable young people may be left with qualifications that hamper their ability to find meaningful and rewarding work.
Note here that the point isn't that "being employed in a directly relevant profession" is an important measure of a degree's fitness - in that case, dentistry would be the best degree ever. Rather, it speaks to whether there's any great shortage. Were there a generalised shortage of scientists, we would expect most graduating scientists to be snapped up by employers desperate for scientists. Instead, if we look at the underlying data, we see that only 63% of those graduating with life sciences degree and 65% of those graduating with a degree in the physical sciences say that their qualification is either a formal requirement or "important" for their job. 24% of those with a life sciences degree and 19% of those graduating in the physical sciences say their degree is not important for their job; the rest say "somewhat important."

I wonder what the NZ numbers would look like.

The running joke on Big Bang Theory is that scientists make little unless they manage to luck into a good job with Pharma. Bernadette waited tables while doing her PhD. Two young physics faculty members share an apartment; their neighbour, a waitress, hasn't got a roommate and consequently has more substantial money problems. Another physicist only has any money because his father sends him remittances from India. A university engineering technician doing NASA-level work lives with his Mom. Closer to real life, Noah Smith explained grad school in the bench sciences.

Want more of the good stuff that comes from basic science? Find better ways of funding basic science.


  1. Wouldn't the people most likely to make significant discoveries already be heading towards science?

    Instead of getting more people working as scientists, might there be solutions that aim to make science easier?

  2. Since you can never know which science is going to come with the next laser or transistor or DNa breakthrough why bother trying to double guess it by public funding science via committee. Why not just allow rich people to set up prizes to drive science forward in whatever way they like. So the X prize for space just like the original ones for flying in 20s drives research and industry forward. The financial gain isn't the goal, its getting there first that drives the research forward.