Math stimulus

Update: For a different take see David’s comment.

Peter Woit made an excellent point in the comments that I want to bring up to the main page and expand on:

In math, a short-term increase in NSF funding of conferences and summer salary is hard to justify, but more money for postdoc positions and graduate student fellowships to keep people employed as university budgets get cut is something that makes a lot of sense.

I think this is exactly right. The main purpose as I see it of a stimulus package (and I admit I don’t really understand stimulus stuff well, it’s not like universal health care where it’s pretty obvious what we should be doing cause you can just travel and see it) is to keep people employed and spending the same way they would be if the economy weren’t tanked. The single best way to do this is to keep people employed in the jobs they would have in a normal economy. This is why state aid is such an obvious component of a stimulus, it keeps school teachers and other state employees at the same jobs they had before the recession and the same jobs they’ll have after the recession.

In math what are the jobs people are losing because of the recession? Yes graduate students and postdocs as Peter points out. But even more than either of those it’s starting tenure track jobs that are getting cut. A quick perusal of the math jobs wiki shows that many more of those searches have been cancelled.

So if I were running the NSF and the math portion of the budget was expanded I would try to increase the number of graduate student and postdoc fellowships (but not their pays, no matter how much I personally would enjoy a pay raise), but my first priority would be to start a program whereby schools can get several years of bridge funds for making new tenure-track hires.


26 thoughts on “Math stimulus

  1. Noah,

    I agree with your point, though I’m not sure that “bridge funds for making new tenure-track hires” is the right way of going about it. That sounds rather hard to administer. Wouldn’t it be easier to just put more money into grants for new Ph.D.’s (maybe with large quantities earmarked for buying out teaching)?

    Alternatively, the NSF could initiate a post-post-doctoral fellowship. This might have worrying implications for the future “postdocification” of mathematics, but on the other hand, as long as the funds were flexible, I don’t think it would be a problem.

  2. I’m not sure why it’d be any harder to administer than the instructorship option for NSF postdocs. You’d have to build in a little bit more time for negotiations with schools, and perhaps make it a little bit easier to switch schools if the deans end up vetoing an agreement made in the fall. But basically I’m not sure what the problem would be with the NSF paying for the first couple years of someone’s tenure-track job.

    Post-post-docs are very scary for exactly the reason you suggest. The postdoc system is already bad enough for 2-body problems, life-work balance, etc.

  3. It’s not clear to me that anything like “bridge funds” would even be necessary. Roughly 35% of every NSF grant goes to the university as overhead, and an increase in overhead can help cover a temporary budget shortfall that might lead a university to try to cut faculty salaries through attrition.

  4. For state universities, I suppose this could just be part of federal aid to the states, giving them money to avoid the budget cuts causing hiring freezes. Having the federal government directly fund teaching positions at private universities would be tricky.

    Part of the problem with tenure-track hiring in this environment is the immediate university budget problems, and that can be improved by anything that gets them more money (more financial aid, higher overhead rates, etc.). But the other part of the problem is that a tenure-track position represents a long-term financial commitment for a university. In the present environment, like other businesses, university administrations have no idea how long the bad economy will last or how bad it will get, making new permanent hiring (of staff or tenure-track faculty) one of the first things they freeze.

  5. I’m not sure why it’d be any harder to administer than the instructorship option for NSF postdocs.

    A more apt comparison would be VIGRE postdocs. Had you said that, you would have had a point (the instructorship option works like this: you say “NSF, I would like only half as much money this year” and the NSF says “fine by us.”) Perhaps I should have stuck with pointing out (as David Savitt also said) that a new program is not necessary, just more grants to young people through existing channels.

    Post-post-docs are very scary for exactly the reason you suggest. The postdoc system is already bad enough for 2-body problems, life-work balance, etc.

    There’s some truth to this, though of course, it’s a much less severe problem if the money is tied to the person rather than the school. After all, having a fellowship which is flexible about location is vastly preferable to a tenure-track job if you have a 2-body problem. At some point, AJ and I were discussing how one could fix the two-body problem for postdocs. The best answer we came up with was “a lot more NSF postdocs.”

  6. Good point, VIGRE might be a better model.

    In terms of helping with 2-body problems, aside from more NSFs, increased flexibility for where to take NSFs would also be a good idea.

    A weirder idea I had once was to have some sort of fellowship that was competitive but lower paying (thereby allowing more of them to be offered). One postdoc salary is typically a pay raise from two graduate students, but since you can’t professionally afford to have a gap like that in your CV people end up having to make worse arrangements.

  7. A weirder idea I had once was to have some sort of fellowship that was competitive but lower paying (thereby allowing more of them to be offered).

    Of course, one could try to accomplish this by giving differing quantities of NSF postdoc fellowship to different people, and then giving them more flexibility about how to dole it out. If you only wanted $10,000 a year, you could pay it out over 10 years, or could take it in 2 years straight.

  8. About NSF budgeting:

    1. The NSF should not be involved in the tenure-track business,
    or in granting loans to universities private or even public (as they are
    state and not federal institutions).
    2. It costs less to create a postdoc than a tenure-line position,
    and job creation statistics are (rightfully) cited in numbers of
    jobs, not salary dollars.
    3. You are right to discuss “overhead.” Many institutions charge
    ridiculous overhead costs to the NSF, i.e. the taxpayer. Can your
    institution rightfully say that it costs 40% of your grant amount
    just to furnish you with a desk and computer/admin support
    over the summer? The NSF doesn’t seem to scrutinize these
    amounts. They should. The numbers are strictly analogous
    to the “standard” (even legislated) $400-ish “consultation”
    fees that doctors charge the insurance companies.

    My recommendation to making government leaner and
    meaner (assuming that a budget decrease is mandated):
    crack down on overhead fees; continue funding
    postdocs; lower dollar amount but not number of grants.

  9. “it’s not like universal health care where it’s pretty obvious what we should be doing cause you can just travel and see it”

    Perhaps mathematicians should stick to their own profession. As an economist interested in health care policy, I can tell you that “what we should be doing” is anything but “obvious”.

    Comparing different countries’ institutions is extremely difficult, if not impossible, as seems “obvious” to me. (Presumably this kind of comparison is what you mean by “you can just travel and see it”.) As academics, I thought you understand the nature of scientific research and how difficult it is to advance the frontier of knowledge. Optimal government policy is no different, and if one thing is certain, it’s that we don’t know what the optimal policy is (if it even exists, given the difficulties that arise when trying to aggregate preferences).

    As someone who studied mathematics as an undergraduate, and in a field that is essentially applied mathematics, I enjoy your posts about math. But in non-math posts, please exercise some caution when making such statements.

    P.S. These posts do give me great examples of rent seeking behavior, however!

  10. My understanding of economics is very poor. But I think that Noah’s statement that the stimulus should keep “people employed and spending the same way they would be if the economy weren’t tanked” is wrong. I think we want to keep people employed and spending at the same rate that they would be (so that there will be enough money for investment, and so that people don’t suffer from unemployment). But we want to keep them employed in short term jobs, from which they will easily be able to move into whatever sector of the economy starts growing again first. Moreover, we want to make sure that, while in these jobs, they will be able to rapidly produce goods that are of clear benefit to society; this is why there is so much interest in infrastructure spending. A tenure track position strikes me fulfilling neither of these goals.

    The idea of extra postdocs is nice. But let me throw out some crazier ideas: How about “expository postdocs”? Hire recent grads on 2 year fellowships with the primary duty of recording and simplifying the research of the last decade. Like infrastructure spending, this is something that the academic community has a huge need for, and which no institution has an incentive to create. Or “retraining postdocs”? We are going to have a ton of auto workers and mortgage dealers looking for new work, and mathematical skills would help them. Why not take the bottom portion of the graduate pool and put them to work doing retraining?

    I should say that my inspiration here is the historians and linguists who were hired by the Works Progress Administration to record the oral history of the US. That’s work no university would fund, but it is a gold mine of data for people working on US history today.

    The worry here is that these jobs would not be good preparation for be hired at a top level research university. My response is two-fold: (1) Part of that is stigma; universities are nervous about hiring from nontraditional tracks. If this were a large government program, universities would find it less intimidating. (2) It may very well be that we have created too many math Ph. D.’s, and the economy needs to shift them into other fields. I have seen a lot of economists speculate that this is the case with autoworkers and investment bankers. I don’t know if this is the case in math, but I would be opposed to programs that cut off that possibility.

  11. The Mellon Foundation gives grants to some liberal arts colleges, and one of the purposes grants can propose spending money on is to “support curricular continuity by ensuring some overlap of new and retiring faculty members”(*).

    The Mellon Foundation is not by itself a good model for the government to follow, because it is quite non-transparent and dedicates most of its funding to very-good-but-not-quite-elite institutions, but…

    (*) the Mellon Foundation website

  12. Let’s define “good stimulus” as “something that brings benefits that people who vote for it describe”. I’d like to agree with both points that were pointed out by various, usually different, people:

    (1) I don’t see the good argument that any short-term increase in NSF funding is a good stimulus;

    (2) I think that $100b short-term increase in NSF funding is as good stimulus as $100b short-term increase in infrastructure and other spending.

    You can probably connect the dots.

  13. Matt: sure this is rent-seeking. But take a look at how much math teachers and professors — at all levels — make compared to mathematicians in industry, and then at the claims for how workers in the coming economy will need better math, science, and technical training. Then tell me, with a straight face, that shoring up their salaries in whatever way isn’t a good thing for society as a whole.

  14. @David: Thanks for the clarification on how the stimulus is supposed to work, as I said I’m still quite confused by the whole thing. Questions about infrastructure spending and mathematics is pretty fascinating. Expository postdocs could be combined with the creation of a large-scale math research wiki (like the nLab but bigger), or with Gowers’s tricki idea. Open it up for proposals and see what people come up with.

    @Matt: I was being partly tongue-in-cheek there, my point was mostly that I don’t have well-formed opinions on the stimulus. Certainly you’re right that there are lots of difficult details in healthcare. I do think having a strong opinion on healthcare is easier than the stimulus because I can see what people who live in other countries think of their healthcare while I can’t visit several other global economies and see the results of no stimulus vs. stimulus. But I don’t want to get bogged down in a discussion of health care, there are better places on the internet to do that.

  15. Matt-

    I think “rent-seeking” is a bit harsh. Mathematicians aren’t hoping to get grants from the goverment to go sit on the beach and drink piña coladas. They’re hoping to get grants in order to give more people more opportunities to do math. You can dispute whether the value of mathematical research is worth what the government pays for it, but the vast majority of mathematicians seek research funding in good faith.

  16. Ben: the NSF system works for US citizens only. At my department, between half and two thirds of the students in a given year are not Americans, most of whom will look for jobs at American universities after graduating. Many of the top PIs are foreigners as well. The more universities rely on NSF grants, the more they’ll have to hire researchers based on citizenship rather than merit. Ironically, the one thing Americans are systematically better at than foreigners, teaching, is the one thing NSF grants don’t require…

    This effect spills over, and hurts some Americans as well if they have foreign-sounding names. I know a Chinese-American first-year grad who the Northwestern admissions committee admitted around April because they thought she was on a student visa; they explained to her that if they’d known she had a green card, they’d have admitted her in January. But her name is Chinese, so no matter what she wrote in her application, they assumed she couldn’t be permanent US residents.

    Matt: the way I understand it, rent seeking is a term invoked for natural resource states, or other environments in which the government has an external source of money to distribute among people. When the money has to come from taxation, or from future taxation in the form of issuing debt, it’s ordinary economic policy.

  17. We have a friend who had a similar issue with getting misfiled as foreign because he went to college in Canada. I’m not sure, but I suspect these sorts of mistakes happen because of overworked administrators use the wrong sticker and then no one double-checks it. It’s certainly a problem, but I think saying “But her name is Chinese, so no matter what she wrote in her application, they assumed she couldn’t be permanent US residents” is a bit too harsh. It’s not as if everyone on the committee looked at her name and thought “obviously foreign!” instead it’s probably that one person going through a stack of hundreds of forms made that mistake and no one else involved at a later stage checked it.

  18. Alon, you’re using too narrow a definition of rent-seeking. When any politically powerful constituency manipulates the legislative process for gain, that’s rent seeking. An actually good example would be the copyright regime in the US, where it is essentially impossible for works to enter the public domain. I don’t think that the NSF money in the stimulus qualified as such, but the important distinction is the lack of a powerful mathematician lobby.

    As for the NSF funds only being for US citizens, I don’t find this so surprising or upsetting. It’s annoying, but it’s also annoying that I can’t go to college free in Germany, or that I can’t get a job in Canada without a visa. Part of the point of NSF grants is to develop the skills of Americans so that we aren’t entirely dependent on foreigners (and that enough mathematicians can get security clearances). This is a bit unfair, but not nearly as offensive as ridiculous immigration restrictions.

    Ok, I’m not really sure where I’m going with this, since I can’t quite tell which point you’re refering to.

  19. Note: comments primarily about universal health care, on either side, are off-topic and will be deleted. I like scrapping about politics as much as the next guy (more, probably) but there are plenty of political blogs for that.

  20. In the hopes of tamping down people’s enthusiasm about discussing UHC, it’s worth pointing out that the narrow point I was trying to make (“understanding stimulus packages and what they should do is relatively hard because there are few easily understood comparisons and little empirical data”) is just as true if you come to the conclusion that UHC and stimulus spending are both wrong. I’d still expect it to give conservatives pause that the stimulus is such terra incognita, while with UHC you can just visit the nearest socialist hellhole and realize what a joy it is to live in the good old US-of-A.

    If you disagree with this point and think that what the stimulus should be and do is obvious, then you can make that point without referring to health care as I probably should have done originally.

  21. Ben: the upsetting thing about NSF grants is mainly that they’re almost the only source of external funding in the US. This results in ridiculous things, like Columbia’s having three number theory scholarships it can’t give to anyone because no American here is a number theorist. In principle it’s supposed to develop Americans’ skills, but in practice there just aren’t enough qualified citizens.

    Making it easier for marginally qualified Americans to have academic jobs actually hurts the US, in three ways. First, they replace foreigners, who, unable to find work in the US, return to their home countries and apply there the skills they learned in the US; this is essentially a transfer of about $250,000 in scholarship money out of the US. The only way around it, restricting stipend payments, would just ensure the most qualified non-Americans go elsewhere to study, such as the UK or Canada.

    Second, NSF scholars don’t teach, making the pool of undergrad-level math teachers more foreign. Americans tend to make better teachers because of their better English skills and understanding of US education mores. This makes the teacher pool worse, making it harder for college students, most of whom are Americans, to learn the necessary math skills.

    And third, apparently the people who go to the NSA are those who couldn’t find academic jobs. If the national security interest is that the NSA have the best people, then NSF scholarships ensure that more Americans will find non-security related jobs, making the pool of mathematicians in intelligence worse. A better way to address the issue is to increase NSA salaries, or allow people at the NSA to engage in some unclassified, publishable research.

    The point you make about free college in Germany (though tuition in Germany is still almost two orders of magnitude less than in the US) and working in Canada (though getting a visa if you have a degree and a job offer is trivial) doesn’t really improve research quality. Developed countries should be racing to encourage scholars from lower-paying areas to relocate. Britain already is attracting more Indians than the US; Europe is also gaining on the US in attracting Chinese grad students. At the very top the US has the advantage of paying stipends and having free tuition regardless of citizenship, but beyond that it’s falling behind.

  22. Isn’t the solution then not to allow foreigners to get NSF money, but rather to make immigration easier? I certainly think we should be giving out green cards to Ph.D. candidates at american universities like candy.

  23. Frankly, I would argue for money for Classics long before I would argue for more money for math. The discipline of Classics is in danger of going extinct in the US over the next couple of generations. Math is not in anywhere near so perilous a position.

  24. While I don’t know about the state of classics in the US, I agree with Alexander Woo to some extent — I don’t think math needs more money, and I think society would benefit more from money put elsewhere.

    To quote Paul Halmos:
    “I don’t think mathematics needs to be supported.”
    “If the NSF never existed, if the government never funded American mathematicians, we would have half as many mathematicians as we now have, and I don’t see anything wrong with that.”

  25. Alon,

    You make some good points, and fundamentally, I agree with you; the policy is silly jingoism. I guess I’ve just reached the point where I’m not only unsurprised by ridiculous jingoism coming out of Congress, but it’s hard to summon any outrage at all but the most extreme examples.

    Also, there are several points on which your arguments are less than convincing.

    In principle, it’s supposed to develop Americans’ skills, but in practice there just aren’t enough qualified citizens.

    For which purpose? I, of course, can’t speak to your Columbia-specific example (I assume you mean this is funding for graduate students through a number-theory specific RTG grant), but I haven’t seen much evidence that the large NSF programs (the graduate, postdoctoral, and general grants programs) are in any way lacking for qualified candidates. I know many quite promising American citizens who didn’t get the postdoctoral or graduate fellowships (I don’t think it would be a crazy state of affairs if the NSF was funding every American graduate student who was good enough to get into Berkeley. We are nowhere close to this point), and several NSF grant reviewers have told me that competition is intense enough that many good proposals get turned down, consistently. My understanding is that the NSF funds a much smaller percentage of proposals than, say, the Canadian analogue, though I can’t speak to other countries.

    First, they replace foreigners, who, unable to find work in the US, return to their home countries

    The primary problem here is silly immigration policy, not NSF funding, and I couldn’t agree more that our immigration is completely fucked and needs to be essentially scrapped and redesigned. Good luck getting that through Congress.

    Second, NSF scholars don’t teach, making the pool of undergrad-level math teachers more foreign.

    While this is true, broadening the applicability of NSF doesn’t sound like a good solution at all. Rather, making sure that people who teach classes in the US really speak comprehensible English, and then raising the attractiveness of teaching to the point where they can attract enough Americans is. Of course, the effect of this would be to make it much harder for foreign students to get funding from their universities, but it might be worth it. Broadening NSF applicability would then be a (perfectly sensible) response to the drop in university funding for foreigners.

    And third, apparently the people who go to the NSA are those who couldn’t find academic jobs.

    Yes, but they had to make it through graduate school and maybe a postdoc first, which NSF funding might make possible. I certainly know some people who would have been much more likely to drop out of grad school if they had not had external funding from the NSF or DOD.

  26. A minor comment about NSF funding, just in case there is some
    misconception: NSF grant support (including summer salary) is
    not limited to US citizens or permanent residents.

    The NSF distinguishes between two missions: research and training.
    (I may not have the correct names here, but I think I am capturing
    the correct spirit.) Individual grants are part of the research mission,
    and can in principal support anyone, US citizen/permanent resident
    or not. On the other hand, grad student support and post-doc
    support via RTGs, the NSF post-doc program, etc., counts as training,
    and only can be used to support US citizens/permanent residents.

Comments are closed.