I’ve written about the severe burden that many researchers experience from having to comply with endless administrative requirements, as well as all of the attempts to address this problem.
But let’s work through a concrete example of the dollars-and-cents impact of administrative compliance, with NIH as the setting. By a back-of-the-envelope calculation, we could free up nearly $2 billion in researcher time every year at NIH, if we could cut the administrative burden by half.
That would be a steep challenge, to be sure.
First, the problem is pervasive across federal science funding agencies, and goes way beyond anything that the agencies themselves can fix by themselves (at a minimum, others at the table would need to include universities, OMB, and Congress).
Second, much of what we’re talking about isn’t mindless red tape, but instead consists of oversight requirements (e.g., budgetary, ethics, and the like) that are actually a good idea. In other words, reducing the burden of bureaucracy here would require making some difficult tradeoffs.
So, here’s how I got to that estimate.
By the 2018 survey, federally-funded researchers across many agencies say they spend 44% of their research time dealing with administrative requirements.
The NIH has 92,957 active grants right now. Some of these are large centers or other large projects with many PIs; some are single-PI grants. Let’s assume, by way of being conservative, that each grant has 2 people (a PI and a co-PI or post-doc) who end up spending administrative time on proposals, reports, budgets, etc., etc., etc. Helpfully, when one PI has multiple NIH grants, they aren’t allowed to double-charge, and there are no economies of scale — if they spend 20% of their time on each of 5 grants, then there would be an administrative burden for each grant (reports, budgets, etc.).
What are these folks paid? The American Association of University Professors puts out a yearly salary survey. The survey for 2021-22 is here.
The average full-time salary for Associate Professors at doctoral-granting universities was $106,564. This seems like a reasonably salary to attribute to all personnel — on one hand, some personnel might be post-docs with a lower salary, but many are full professors with an average salary of $163,485. Indeed, to the extent professors with long-established careers are more likely to get funding in the first place, federal grants might be overly weighted towards the higher-paid professors!
According to a contact who heads up research compliance at a large university, the typical benefit rate is around 25%. That seems low to me (having reviewed many research budgets/grants while at Arnold Ventures where I recall seeing fringe rates up to 47%), so it may be a conservative assumption as well.
That brings total compensation to $133,205 (or $106,564 * 1.25).
Then, let’s assume that federally-funded researchers spend an average of 50% of their time on federal research grants. This may be a conservative assumption as well (some researchers spend 100% of their time on research grants). 50% of $133,205 is $66,603.
So, we have 185,914 researchers (92,957 * 2) spending an average of $66,603 of their time on NIH grants. Total amount spent for their time: $12.4 billion in the past year.
That compares to an NIH budget in 2021 of $42.93 billion. So the amount spent on researcher time is probably around 29% of the total budget.
That seems a little low, but remember that the total NIH budget includes a large intramural program, and that its extramural grants support lots of expenses that aren’t researcher time (e.g., indirect cost rates that can be 69%, in the case of Harvard).
In any event, as with any Fermi calculation, it’s probably in the ballpark (e.g., I might have overestimated the 50% researcher time but underestimated salaries or benefits). It would help if the NIH officially reported the actual amount of money spent on researcher time, but when I asked, they just told me to file a FOIA request (now pending for three months with no reply). If I get an answer, I’ll update my estimate accordingly.
Just to be even more conservative, let’s assume that whoever answered the 2018 survey and claimed to spend 44% of time on bureaucracy was an outlier, and the true figure is about a third lower: 30%.
Put that all together, and we have a tentative estimate (0.3 times $12.4 billion):
In other words, NIH spends some $3.7 billion each year just paying for researchers to comply with administrative requirements.
If we could cut that burden in half, that would be like getting an extra $1.85 billion in biomedical research every year.
NIH budgets are going up, so let’s round this up to a nice, even $2 billion. But let’s be clear: that is still a highly conservative estimate in several ways. Some you can see above, where I repeatedly chose lower numbers just to be conservative.
But it’s conservative in another way I haven’t even mentioned yet: The 2018 survey on researcher time does not account for the administrative costs/burdens incurred by anyone other than the researchers themselves.
To quote a correspondent of mine who has studied these issues in some detail:
There are, in fact, many additional costs that are borne by the institution. To ensure that they and their faculty are in compliance with federal requirements, universities have had to add significant numbers of additional university employees. They now have expanded research officers to pay countless staff to ensure compliance with animal regulations, human subjects protections, export controls, academic integrity, CIC/COC requirements, environmental health and safety, accounting, etc. . . . So, there needs to be some recognition of these costs as well, many of which can’t be and aren’t costs that the federal government will cover due to the existing F&A cap on administration.
Think about all of those costs. Then extrapolate to the rest of federal science funding.
It’s staggering to imagine how much we could improve science — just in dollars and cents, never mind the effect on scientists’ imagination and focus — if Congress, NIH, NSF, universities, OMB etc., were able to significantly streamline bureaucratic requirements.
To repeat, that would require us all to deliberate as to difficult tradeoffs and priorities. But nothing in life is for free. We can’t have endless oversight of federally-funded researchers without making those researchers spend tons of time on our oversight requirements.
We have to decide where to let go. A future installment will explore ways to do that.