A Skeptical View of the NSF

Tyler and I have a piece in the latest Journal of Economic Perspectives, A Skeptical View of the NSF’s Role in Economic Research. Here is one bit:

In considering the case for grant-based funding of economics research by the National Science Foundation, we find that a number of pertinent questions are rarely asked, let alone clearly answered. Instead, economists often put forward relatively weak arguments that they would likely dismiss if applied to government subsidies not reserved for economists.

For example, one common approach to defending NSF grants for economists is to list the prestigious individuals with whom the program has been associated. In his paper in this journal, Moffitt notes: “The program has supported every Nobel Prize winner in Economics since 1998 and almost every John Bates Clark medal winner since 1961.” (NSF economics funding started in 1960). Indeed, the list of grant recipients from NSF Economics is a literal “Who’s Who” of the top economists over the last half-century. But we don’t find the prestige of NSF recipients to be a good substitute for an estimate of the public benefits of research. Imagine a group of chefs who defended a hypothetical “National Food Foundation” on the grounds that it had provided grants to Alice Waters, Thomas Keller, Grant Achatz, and every winner of a James Beard Award since 1990. If these names are not familiar, rest assured that their published research output and training of students is very impressive. While we would not consider this information irrelevant (better to fund good chefs than bad ones), as economists we would be unimpressed by this case for government funding of chefs. Talk of how these grants brought about innovations in the culinary arts—such as sous vide, molecular gastronomy, and the introduction of quinoa to the American diet—would also not swing the argument. Instead, as economists, we would focus on how food markets would have operated without such grants and what else might have been done with the money.

We are not against any role for the NSF, however, but instead suggest that funds would be better spent on replication, public-use datasets and “far-out”, high-risk research. We also suggest other funding mechanisms such as prizes.

Here is one bit on replication:

Instead of pointing to the prestigious economists whose research they have funded, perhaps the NSF might point to the prestigious research that has been convincingly replicated—or not replicated.

Read the whole thing.

By the way, this entire issue of the JEP is excellent with important reviews of the research on charter schools and four papers on motivated reasoning.


In other words, what America needs is a Republic of Science.

More use of lotteries in the disbursement of research funds.

No, there should be a free market. Every econ research department should sell shares, which would be publicly traded. If they do good work, their share price would increase and they'd be able to hire more people. Otherwise, they'd diminish. NSF would buy a certain number of shares every year, but so would the Koch brothers and anyone else who wished to. If speculators think GMU looks promising, they might bid the price up in anticipation of good work.

Do you think the current system of awarding funding is worse than random? I agree that it is not optimal, but this seems extreme.

Worse than Random would be a great band name.

NSF grants seem to me to be a form of ex-post reward for scientists who have done good work in the past. This is not without value, but some ex-ante rewards are needed to.

Yes. My advisor had won NSF funding for previous work and a fraction of my salary during masters came from that founding. So, the "old" won the funding but the "young" had the benefits. Is it too bad that professors choose which students get the funding? It was a good experience, I could focus on science while the boss was very busy with all the paperwork.

Which describes the entire charter of NSF grants: fund the education of future scientists through apprenticeships with leading economists.

I too was paid in part by an NSF grant in 1968-1970 given to colleges to introduce computers into education in the most disciplines. I was a college drop out hired as an "operator". I knew a bunch of faculty and lots of students, and my advisor wrote half the grant. I worked with a lot of different kinds of students in the first year and some unusual for the 60s faculty disciplines. It turns out I took on a "side" project tied to fund raising that "paid" for the computer system for a couple of years - data mining it's called today.

Hundreds of colleges got similar grants in the late 60s and by the 70s, student access to computers was expected even at medium sized liberal arts colleges.

Oddly, with computers so cheap, use of computers is less exploratory than in the 60s and 70s.

I had students work on writing poetry on an IBM 1130 computer using 80 column cards in 1969. So, hearing a story on computer generated poetry this week just reminds me of how ambitious we were in the kinds of results we hoped for from that NSF grant.

"I had students work on writing poetry on an IBM 1130 computer using 80 column cards in 1969."

This doesn't surprise me at all.

The NSF hires a lot of survey statisticians. I wonder what their role is and whether they have any internal research showing the effectiveness of their programs.

Ooh! What a great idea! NSF should pit grantees against each other! Is this result statistically vaild? Are the p-values correctly calculated? Was the experimental design flawed? Maybe NSF needs to increase its investment in statistics as a share of the pie.

Sour grapes.

Koch brothers should fund economic research.

But only when the faculty meets their approval - ' A conservative billionaire who opposes government meddling in business has bought a rare commodity: the right to interfere in faculty hiring at a publicly funded university.

A foundation bankrolled by Libertarian businessman Charles G. Koch has pledged $1.5 million for positions in Florida State University's economics department. In return, his representatives get to screen and sign off on any hires for a new program promoting "political economy and free enterprise."

Traditionally, university donors have little official input into choosing the person who fills a chair they've funded. The power of university faculty and officials to choose professors without outside interference is considered a hallmark of academic freedom.

Under the agreement with the Charles G. Koch Charitable Foundation, however, faculty only retain the illusion of control. The contract specifies that an advisory committee appointed by Koch decides which candidates should be considered. The foundation can also withdraw its funding if it's not happy with the faculty's choice or if the hires don't meet "objectives" set by Koch during annual evaluations.' http://www.tampabay.com/news/business/billionaires-role-in-hiring-decisions-at-florida-state-university-raises/1168680

What was more interesting was the disclosure statement in the paper and the lack of any influence by the Mercatus Center or its funders:

Tyler: "This research did not receive any financial support, and the author has not received support from the National Science Foundation, nor does he have a grant request pending. Over the last three years, I have received support from George Mason University, the Asian Development Bank, Leigh Speakers Bureau, Jane Street Capital, and St. Martin's Press. I am on the boards of the Mercatus Center and Institute for Humane Studies, both at George Mason University, unpaid positions."

Alex: "I have nothing to disclose."

'A Skeptical View of the NSF’s Role in Economic Research'

One hopes that page after page is spent concisely detailing why science and economics have nothing to do with each other, and mocking the National Science Foundation for ever thinking so.

The claimed NSF "skepticism" here is extremely weak.

The initial position of an objective review of the NSF should not be:
"We are not against any role for the NSF" -- but rather WHY should the NSF exist at all?

NSF is a classic government boondoggle of waste and rent-seeking. NSF was supposed to be the grand 'primary' government agency for funding and coordination of basic science, as envisioned by its naive supporters after WWII. But by 1950, major areas of research were already fragmented and dominated by specialized government agencies/fiefdoms. Despite that instant mission failure, the scope of the immortal NSF bureaucracy kept expanding (mission-creep) to include many areas not in its initial powers, including the social sciences, engineering, and mathematics education. The NSF is the only U.S. federal agency with a mandate to support all non-medical fields of research. What have they accomplished justifying their vast spending since 1950 ? Skeptical indeed.

Well, one "benefit" of sorts is that it's a secondhand subsidy for higher education. Universities are legally able to skim a cut off the top of the grant, and the grants often fund grad students and/or postdocs. Of course many fields suffer from an overproduction of graduate students, so there's that...

I gather you think students should pay to do grunt work for professors doing research they hope to profit from like businesses have tried to have interns pay to work?

In the 60s and 70s, TAs and RAs were slave labor working for professors who got to play with the toys to do their own projects, or at least become an experienced lab guy or writer or business startup. Or even go on to teach in the hopes of getting tenure and getting your own slaves.

The point of NSF grants then was not "results" but on getting research started in promising areas to the point it would get real funding.

I would describe NSF grants like angel investing. It gets promising people with promising ideas started so they can make a concrete research proposal to funders with real money.

The thing that gets me is conservatives say that free markets are so great because lots of risk taking and failure leads to great progress. So then they argue that 100% of NSF grants need to deliver ground breaking results and when the committees awarding grants pick the safe proposals they have failed because they should pick the highest risk totally unknown proposals and if any of them fail, then the whole system is a failure.

And we now see this in the private sector "free market" with record high profits but too much risk to make a single investment (uncertainty) except for insane people like Elon Musk taking massive risks with government money and the money of bamboozled rich people with too much profit.

The safe bet according to conservatives is ULA created out of the parts of failed aerospace government contractors who have suffered from cuts in government spending since the 80s tax cutting binge that has frozen or cut government spending that is really investment.

Conservatives want 100% safe investments that return high profits as if their investments have a 99% failure rate.

Try not to read too much into what I believe. I just got done being one of those people paid to do grunt work for four years not too long ago. I wish NSF had more money, frankly. That said, I also think that the overproduction of PhDs is a major problem, although it's not as prevalent in my field as some, and to the extent that NSF funding is subsidizing that outcome, I think there are improvements to be made.

Yeah, you're right. Not really many US contributions to non-medical basic science science 1950.

Evidently reading comprehension has retarded.

But the NSF has funded a lot of basic science, and the prior poster asked, "What have they accomplished justifying their vast spending since 1950?" so... pretty on-topic, I think.

Ethan responded to a post that said, "But by 1950, major areas of research were already fragmented and dominated by specialized government agencies/fiefdoms", not NSF. So Ethan, and your, responses are pretty nonresponsive to the claim that others generate most of the gains.

I have no idea, just being pedantic....

a) the NSF funds medical research too. b) let's play name dropping: Argonne, Fermi, Berkeley, Los Alamos, & Sandia National Labs. The least you can say about these places is weapons development during Cold War. If you don't like weapons, I remember being developed in National Labs is synthetic aperture radar (SAR), it's mounted on satellites and helps to make topography on Earth. The elevations you see in Google Earth comes from SAR measurements. Like this there thousands of things we use around the world developed on National Labs. Guys, appreciate the good things of the USofA.

Of the SBE disciplines, economics, by far, places greater value on pedigree over production. This begins quite noticeably in hiring, and it sets the course for most follow-on outcomes including funding and prizes.

Most simply, the big names argument is talking about average product when marginal product is what matters.

Not to be forgotten is that writing grant applications takes time. In the case of the big name receivers of grants, I bet it's actually the case that the NSF grants reduced their productivity because they each spent two weeks or a month of time and energy writing the application--- not to mention the time of the reviewers.

I had an NSF fellowship for graduate school at MIT. I would have gone to grad school anyway, but maybe not at MIT. Other students maybe would have chosen law school instead. How many, I wonder? That's the question.

If writing grants costs more than their ground breaking research, why don't they simply do and deliver their ground breaking results?

Are you arguing economists don't understand basic economics of the firm?

Seems like putting the old guys in charge of the biggest pot of money around would get a lot of reasearch to support old ideas. That's how orthodoxy perpetuates.

If the NSF grants were measured on failures instead of successes, would you praise the grants going unknown people with bizarre economic hypothesis to prove?

Say people arguing that the economy is best when economic profit is zero?

Or gdp can not exceed total aggregate labor costs?

Or gdp grows fastest when labor costs are zero and all people are eliminated if they work for income, and only people who have infinite cash to spend are retained in society?

Oh, wait, lots of famous economists argue the latter all the time.... my bad.

The "economic profit" is zero at "efficiency". Anything below "efficient" and you're talking an economic loss. Anyways, I wouldn't get distracted too much by the "zero economic profits" stuff - clearly in lots of industries the assumptions relating to perfect competition do not even remotely apply.

Spell out would have helped. I initially read NSF as "not safe for work".

The NSF is academic pork barrel. Send the Office of Polar Programs to the Interior department and eliminate the rest. It'll be very entertaining to read professor rent-seekers huff and puff in the Chronicle of Higher Education about it. ("blah blah blah in the opinion of I and my colleagues in the profession blah blah blah...").

So your view is that higher education as we know it should just collapse, then? The thing you are proposing is at least as big a change to a system as PPACA was to healthcare, and to a system of a comparable size. It's funny how people are suspicious of rapid change to institutions only when they like those institutions...

"Higher education as you know it" has a gross output of about $500 bn. The NSF last I checked disbursed just north of $7 bn in grants. You'll get a few billion from the national laboratories; a sum of a few billion more from the military, NASA, and the arts and humanities endowments; a billion or so from the National Institute on Food and Agriculture, and about $20 bn from the National Institutes of Health. So, you lose 6-7% of your budget (assuming the state legislatures and the foundations tell you all to bugger off), and you 'collapse'. Well, soooo looong....

Higher education as we know it is not an undifferentiated good. Distribution of resources also matters...

Take your case to the provost at your institution, or to the college president, or to the trustees.

Higher education sucks. No man of sense likes it.

As we are in the beginning stages of understanding how the public goods market system works, at a stage we will inevitable look back upon as crude and unsophisticated, the best we can do is clean up systems that exhibit traits of well known inefficiencies. Certainly if only 11 universities are receiving funding, the NSF has misplaced the 'national' in its name. Maintaining such a monopoly also discourages other researchers from using their voice to promote their cause and worthiness as they undoubtedly feel their efforts will go unheeded. And if the funding is only supplementing research that would have gotten private funding anyway, insult is simply added to injury, or rather inefficiency.

The law professor Phillip Johnson, who was for a time Peter Duesberg's lawyer, had an interesting article a while back on the social dynamics induced by federal research funding. I'm not sure where to find it now. His point was that it generates path dependency and bullies dissenting viewpoints out of existence.

Comments for this post are closed