The Cost of Indirect Costs
One can favor publicly funded science AND efficient use of public science funding.
Until indirect costs made the news, very few people outside the rarefied world of academic grants knew or cared about them. Much has now been written on the topic, so I might not have anything meaningful to add. Still, I’m in an interesting position because I wrote academic grants for many years, now write grants in the non-profit space, and, most importantly, have no particular stake in the outcome of the issue.
So I can tell you a little bit about how it all really works.
First, some basic economics.
You are the Dean of the College of Arts and Sciences at Poison Ivy University. Here is how you look at the grants that faculty in, say, the Psychology Department, might apply for. Let’s say Dr. Happy applies for $100,000 from the National Science Foundation to fund his “research” over the summer—and by this, Dr. Happy might very well mean that he will write a paper using data he already has and then say that the work was “supported” by the grant.
It’s possible to estimate the costs and benefits to you, the Dean, if the grant is approved. On the cost side is the extra money that the university will have to spend on overhead—utilities, administration, etc.—if (and only if) Dr. Happy gets the grant. After all, Dr. Happy will now need to be in his office during the summer months, which will add $48.73 to the University’s power bill. He will also use more pencils from the Department supply closet. Add another $3.27. You don’t have to pay the staff more; they are hired to clean up the place and so on, so there’s no additional cost in that sense if you get the grant. Now, maybe someone in the business office has to enter some numbers into some spreadsheets because there are some new assets and liabilities. We can assume some hours for that. Add it all up and let’s say, to be generous, it’s $1000.
What about the benefit side? This is where indirect costs come in. The overhead that institutions charge granting organizations is negotiated by the institution with the relevant granting entities. So, at Poison Ivy U, some committee, Dean, or Provost argues that for every dollar the National Science Foundation or one of the National Institutes of Health gives to Dr. Happy, the granting organization needs to give us $.65. That doesn’t mean that they get $.65 and Dr. Happy gets $.35. It means that if Dr. Happy gets $1.00, the National Science Foundation sends a check for $1.65 to the University, and then the University passes $1.00 on to the faculty member’s research fund for the project. In this case, if Dr. Happy gets the grant, the National Science Foundation will pay Poison Ivy University indirect costs to the tune of 65% of the amount of the grant, or $65,000.
Now, as a Dean, you are not very smart—you would not have gone into academic administration if you had the midi-chlorians to be a real scholar—so you go to the math department and ask whether $65,000 is bigger or smaller than $1,000. You are told that not only is it bigger, but it’s quite a lot bigger.
Therefore, you pay a visit to Dr. Happy and strongly encourage him to write the grant. In fact, you might subtly imply that if he doesn’t write grants, his future at Poison Ivy is not guaranteed, assuming that the good doctor is pre-tenure. Make no mistake, tenure committees know this. When I was on the clock, it was explained to me that of course we don’t really care if you get big grants, assuming you are publishing high quality work. What this meant, which everyone knew, is that they very much cared if I got big grants and might vote against me if I did not. In my case, I had some extra help, so the fact that I didn’t bring in the big bucks didn’t really matter.
Now let’s look at it from the perspective of the Program Officer at the National Science Foundation, the person ultimately responsible for approving or rejecting the grant. If the grant is approved, the federal government, through the National Science Foundation, spends $165,000 for $100,000 of “research,” and $65,000 to cover the $1,000 of actual overhead costs. The remainder goes to the Dean’s budget so that he can fly first class to his next conference where he can pretend to listen to talks about the actual, real mortal dangers of exposing students to any ideas to the right of Pol Pot.
Meanwhile, The Program Officer, knowing that the taxpayer is paying $165,000 for goods and services worth $101,000.00, could not possibly care less.1 Her job is to choose the grants that get funded and comply with a boatload of regulations. She herself is not out $64,000. You, dear taxpayer, are.
But now let’s consider a different world. Dr. Happy writes a grant for $100,000 and the extra cost associated with the grant is actually around $30,000. This is because Dr. Happy is running a medical trial which will require that another business unit at the university hire a part-time technician to support patient testing. (This is not necessarily realistic, but I suppose it could happen.) Now, in this hypothetical world, the indirect rate, let’s say, is 30%. If the grant is funded, the indirect costs will just cover the extra costs to the organization.
You are the Dean. How do you feel about this?
When you visit Dr. Happy and he asks if he should apply, you know that if he gets the grant, the extra costs associated with it will just be covered by the indirects. Your budget is not affected one way or the other. You tell the good doctor that you are indifferent to whether he applies or not.
Finally, suppose the government stops paying indirect costs. Run the conversation again, and now the Dean says, no, please don’t apply. If you do, you will be better off—you get the grant—but I will be worse off because I have to pay for the extra pencils out of my budget. In this world, it might be noted, scientific inquiry is genuinely reduced. Deans are telling faculty not to try to win support for their research.
From this we see that we can infer which world we are living in by the Dean’s behavior. We will know when indirect costs are about right—exactly covering the actual marginal expenses of a new grant but no more—when Deans don’t care if faculty apply for grants.2
I will be very pleased if anyone is aware of cases in which a university dean or provost indicated indifference to whether or not their faculty wrote grants. I certainly know of no such case, which tells me which world I, at least, lived in.
Now some caveats. Of course in some areas, getting a grant funded genuinely does add significant expenses beyond the direct costs covered by the grant. The medical example above illustrates how that might happen.
I never saw the Dean’s budget at Penn, so I can only relate what the experience looks like from the faculty side. I’ll provide one example. To be clear, this is something of a confession. I fed at the trough.
I was part of only one large grant back in my Penn days. My part of the sizable federal grant was probably a couple hundred thousand—I don’t exactly recall, which is embarrassing—meaning that Penn got somewhere in the neighborhood of $130,000. Other than the time for accounting, I can’t for the life of me figure out what that hundred grand or so might have been used for. I didn’t use more pencils or electricity. Yes, I worked on that project because of the funding, but Penn didn’t need to pay some technician to service a fancy machine that went bing.3 I ran all subjects in my laboratory, so it’s hard to see how this added costs to the institution. I do that all the time anyway. Most of the money I think went to paying me in the form of summer salary and supporting a graduate student so she wouldn’t have to spend time as a teaching assistant. That can’t have changed Penn’s overall overhead. I would think that Penn barely noticed the difference between my having the grant and not having the grant, and bore no additional real costs. Puzzlingly, this did not lead them to turn down the overhead. Some mysteries can never really be solved.
Keep in mind that the pile of money that went into the Dean’s pocket could have funded another research project had the overhead been lower than 65%. High overhead carries an important opportunity cost, one of my favorite ideas from economics. Imagine if that money went to pay a graduate student stipend to do science instead of the Dean’s travel budget. That is, if you like science, wouldn’t you like less money to go to Deans and more money to go to scientists’ budgets for their research? If you genuinely want to maximize science, you want to be in the world in which Deans want their faculty to get grants, but are very close to indifferent about it. At that point, scientists have incentives to get grants and the least possible amount of money is going to administrative overhead. It’s worth thinking carefully and critically about why various people and organizations loudly come out in favor of status quo, high overhead. Are they really in favor of maximizing the amount of science funded by the government? Or might they have ulterior motives?4 If you were a Dean and you saw your faculty indifferent to—or even supporting—the loss of your slush fund, how would you feel about those faculty? Just asking questions here.
I think a good comparison is the non-profit world, where I now work. (Technically, Penn is a non-profit, but everyone knows that’s just pretense.) At the Garces Foundation, we provide low-income newcomers to the United States living in Philadelphia education, dental, and health services. Last year, the William Penn Foundation generously funded the Garces Foundation and allowed 15% overhead. Now, we’re small and don’t have economies of scale, so in this case we did bear additional costs because of the grant. In fact, we actually have negative economies of scale: as our programs get larger, our per student cost rises because we’re a genuine charity and don’t charge students enough to cover our costs. Therefore, each additional student adds to our net loss.
I might also note that the William Penn Foundation grant was unusual for us. Most grants include 0% for overhead. You can only ask for direct costs.5 That requires planning very carefully in terms of what you’re asking for. If you add a program, you usually can only request money for that program. So as you grow, adding more and more programs, you might well have more overhead—administration, rent, etc.—but the granting organizations aren’t paying it. This is why we always have our hand out. In addition, this dynamic has caused me to avoid applying for some grants: without the ability to add some overhead to a grant proposal, getting a grant could actually put us in a worse financial position. It is hard to me to imagine that was the case at Penn in the last several decades.
Why does Penn get 65% overhead, when it doesn’t really, at least in the case of grants like the one I got, need it to cover additional costs when small non-profits get little or no indirect costs?
I will leave it to wiser minds than mine to ponder.
However, I will venture the perhaps crazy hypothesis that it might be due to incentives. I will gently remind the reader about the Program Officer. Consider how much she cares about the amount of other people’s money being poured into the Dean’s travel fund. (Hint: she literally could not care less.) Now consider the incentives of the Dean and Dr. Happy. The Dean wants more money for his slush fund and Dr. Happy just wants more money.
It’s important, I think, to note that in some sense, there’s no bad guy here. Everyone is following the rules of the game they are playing. Dr. Happy wants to do research. The more research he does, the more his star rises. I myself faced these incentives and behaved as predicted. The Program Officer wants to choose good research but doesn’t care about indirect costs. The Dean wants to fly first class, even if that means putting the bite on the faculty. (Ok, yeah, maybe the Dean is a bad guy here.)
And the people negotiating the indirect cost percentage—the people at the Universities—have an incentive to try to make that number as high as possible to grow their budgets, but below the level that would draw attention to the con.
As for the other side of that negotiation, the people at the federal agencies, they were very pleased with what university administrators did with the extra money, including indoctrinating educating their students into voting the correct way, at least from their perspective. The arrangement worked just fine.
Juuuust fine.
Right up until the moment that someone decided to make a federal case out of it.
You’re probably wondering at this point what my solution is for setting the rate of indirect costs. I confess I don’t know.
I would say that it might be difficult to use different indirects for different areas, but it is odd that the indirect rate is constant from the humanities to fields such as the hard sciences and medicine. My experience suggests that the overhead costs of a big grant vary by discipline. Perhaps by a lot. In my case, I think it was pretty close to zero, but I don’t know that for sure. Ask Gus in the business office how many hours he put into the administration of the grant. Maybe 20, total?
I would also say that it doesn’t seem all that crazy to me to run some experiments—when does the Dean’s office not care about the next grant?—or even run some analyses to determine the actual cost to the university when it gets an additional dollar in grants. I would wager there are already datasets that inform this. Perhaps an economist can look at administrative costs for universities that move from a teaching focus to a research focus, and see what budgets look like. Maybe there are analyses already out there.
I do think that it’s important to frame the debate properly. Reducing overhead is reducing the money that goes to the administration, for overhead, not to the scientist, for research.6 If one supposes that the budgets of NSF and NIH, for example, are fixed, and the amount paid for administrative overhead decreases, then the amount of science getting done will increase, not decrease, as money moves from indirect lines to direct lines. Once the overhead rate gets below the actual indirect costs of a grant then, yes, research at places where the indirect rate does not cover overhead costs will decline for the reasons indicated above.
My guess is that way more money is being paid to institutions for “overhead” than is actually needed to cover the cost of the overhead for grants. That money goes to administrators, not scientists. Now maybe some would argue that more research gets done if we spend a ton on administrators and less on scientist, but color me skeptical of that. I, at least, never experienced that.
One can be both in favor of more scientific research, even strongly in favor of publicly funded research, and at the same time be in favor of the efficient use of resources—public and private—to do that research.
This note was added after posting. Please see Rosemary’s correction on this topic in the comments. My remarks about this are hyperbolic. Some Program Officers do indeed care. My experience is that many people in the ranks of the federal agencies do not, but Rosemary is correct that the remark here as I have it is not accurate. I regret the error.
Obviously it’s more complicated than that. Grants add to the greater glory of the organization and thus the Dean. So they might encourage grant-writing even if there is a financial loss. Fine. But an economist would argue, ok, so the grant has benefits beyond the dollar amount which should be included in the calculation of what the granting entity is giving to the university.
You might be wondering about course buy-outs. If I teach less because I am doing more research, now the university has to pay to cover that cost. Except, no. If I put a course buy-out in the grant, the granting organization pays the institution what they would have paid me, say $50,000 for argument’s sake. So in effect Penn is paying me less and, at the same time, gets the indirects ($30,000, say) and pays an adjunct $8,000 to teach the course I am now not teaching.
The non-profit world distinguishes between “restricted” funds and “unrestricted” funds, which are also called “general operating” funds. Charities such as the Garces Foundation want general operating funds, which can be used for any expense, including overhead, but funders generally prefer to grant only restricted funds, to support a specific program.
During my career, I became aware that some organizations kick some of the overhead money back to researchers in various ways. I will say nothing on this topic.
So, I've never written a grant proposal. I have however, written or participated in writing quite a few commercial equivalents, ie offers/quotes/proposals prepared in response for RFPs from potential clients, mostly working for small businesses providing services. The clients varied from a local brewery to government ministries, charities and national bureaus of supranational organizations under the UN umbrella.
Quite a few of those explicitly didn't allow for any kind of proportional overhead but required everything to be itemized even if in gross terms like project management or supervision or training or purchase of pencils and paper and I might be naive but is there a reason why this couldn't be dealt in the same way with grant applications
I'm not sure this is entirely fair. You are right that overhead varies by discipline, but one thing I think you're overlooking are the *common* goods that indirects can contribute to. For example, if there are enough individual grants coming in, and thus enough indirect costs going to the university, that university can start offering services and resources to the researchers as a body that it previously could not afford, such as more research librarians, expensive data subscriptions, or grant support personnel (this doesn't beg the question: such personnel are justified by the *direct* costs they help researchers win that those researchers otherwise might not).
Your argument seems to have it that deans would be indifferent to *these* monies, since they don't do things like let them fly first class. But I think this also overlooks that many such administrators also like having larger teams and operating budgets because this is its own form of social cachet. Maybe that also makes them the bad guy, or it is just a case of aligned incentives.