Dan Gross (2:36)

Research generally incurs two categories of costs, much as business operations do. You have fixed costs and variable costs. And so those variable costs are usually referred to as the direct cost of research. They're typically project specific, include salaries, consumable supplies, whereas indirect costs are not as easily assigned to any particular project. Things like lab space, data and computing resources, biosecurity, keeping the lights on, keeping the buildings cooled and heated, and even complying with regulatory requirements that the federal government imposes on researchers. It's the overhead cost doing research as a researcher.

Pierre Azou (3:12)

It's true, you will use those grad students and those mice and those test tubes. Those are your consumables, but you're also kind of implicitly using the lab space. You may be using a shared facility where the mice are actually kept and fed pieces of large kind of equipment that are shared by many other people to actually conduct experiments. So those are kind of fixed costs from the standpoint of your research project and the $1 million you were referring to, that's only the direct cost. They are directly attributable to what you promised the government you were going to do for the next four years, say. Right. And so indirect costs is a way for funding agencies to acknowledge that, well, actually, your research also has fixed costs, and we need to kind of make the institution that hosts you kind of whole from that standpoint. And the whole kerfuffle of the past few months has been all about how we should calculate that percentage and whether the currently used percentage is in some sense, correct.

Dan Gross (4:32)

And so in that context, indirect cost recovery, what we call icr, that is how the federal government has been paying for the fixed cost of research for really the past 60, even 80 years. And for those past 60 years, the way this has been done is by paying universities and other institutions institution specific fixed percentages over the direct cost of the research that the government fund funds. And that's the indirect cost rate. That rate is negotiated by institutions typically every two to four years, supported by several hundred pages of documentation around its actual incurred costs over the recent funding cycle. And the whole idea behind this is to be able to compensate federally funded researchers for the investments in infrastructure and other overhead expenses that are related to the research they perform for the government. Without that funding, universities would have to pay those costs out of pocket. And frankly, many would not be interested or able to do the kind of science the government is funding them to do without that additional support.

Host (Santi) (5:40)

So let's say I'm doing my mouse cancer science at MIT at peers parent institution. Sometime in the last two to four years, MIT had this negotiation with the National Institutes of Health to figure out what the MIT reimbursable rate is like. I don't have to worry about it as a researcher. I'm just like, I'm doing my mouse research, applying for a grant. I don't see or touch the broader debate over what our indirect costs that are reimbursable are. I'm just mouse research all day.

Dan Gross (6:11)

No, this is as much of a mystery to the researchers as it is to the public. I think to most of us, we have no idea where indirect cost rates come from. All we know is that, for example, when I was a junior faculty and I applied for an external grant from the National Science foundation and you're used to seeing, you can look up in the award search portal at NSF different awards folks have won and it doesn't break down in NSF what indirect and direct cost shares are at each grant. You just see the bottom line total and you say, wow, this person got $300,000, that person got $500,000. And then you go to write your own grant and you realize you can actually only budget in about 60% of what you thought you were going to budget for because the rest goes to overhead. And it comes as a bit of a shock the first time you apply for grant funding. And then even then, what actually goes into the overhead rates? Most researchers and institutions don't have clear visibility into that. The process itself is so complicated that it's hard even for those who are experts to keep track of all the pieces. We'll get back to what actual rates look like, I think in a little bit. The they don't always match the headline rates that appear in negotiated rate agreements or they get a lot of public attention. But suffice it to say that the system is complicated and hard to penetrate. And maybe peer has a different perspective and finds the stuff. Textbook.

Dan Gross (7:31)

Classic 101 research funding policy.

Pierre Azou (7:33)

I guess I do. I have. I don't know if it's a nuance or a disagreement, but I never remember applying for grants and I haven't applied for grants in a while. Kind of adjusting my budget based on the overhead rate of my institution. So there's a sense in which, you know, there's a budget constraint, which is the overall budget of NSF and nih. And so other things being equal, if there's more allocated to indirect costs, there is less allocated to direct costs. But that's not necessarily the attention you feel as an individual researcher applying for a project. You think about, you know, what the direct costs of your research projects are going to be. Maybe you kind of like forecast that they're going to adjust your budget down anyway, so you kind of like, you know, inflate a little bit in anticipation of that bargaining. But you're not really thinking about the indirect rate and the research administration of your institution when it actually kind of sends the application in, is going to apply the right rates to the right amounts. And, and yes, what will come as a surprise is what the total amount is. But I'm not sure if you know that in advance. Dan, you disagree?

Dan Gross (8:43)

It's very interesting. I think then every institution must have a different process for writing and preparing submissions to grant calls, because the budgets that I'm typically working with, I mean, they have that line item in there and they have the bottom line total. And this is one of the ironies. The POS Albert seems sensitive to what the grant total is on the proposed project is including the indirect cost recovery. On the other hand, when I talk to folks who are in the life sciences and who get NIH grants, they tell me specifically that indirect costs don't even enter scientific review. It's just the budget just isn't something that the peer reviewers pay attention to. They focus on the value of the science. And of course, it's anecdotal accounts from folks that are in my orbit and that I know and that even the scientific officer won't adjudicate or assess grants based on their budget and their overhead cost rate. It's really about the quality of the science and the peer review scores. And so if anything, I might be the outlier here.

Host (Santi) (9:44)

Just to stick with this toy model, though, just so folks who have never applied for a federal research grant have a sense. My basic understanding is, okay, so I've got this $1 million experiment I want to run on mouse cancer. And what actually comes back if I get the grant is like a $1.5 million. And the university, before I see any of that, takes that half a million. That was for the indirect costs for the building I'm in and for the massive microscope we bought last year, and a tiny little bit for the groundskeeper, the janitor, just the maintenance of the building. And then I get my $1 million. Is that a rough as a toy model? Is that right?

Dan Gross (10:26)

It works like this. So Duke University has a 61% indirect cost rate. If I propose a Grant to the NSF for $100,000 of direct costs, it might be for data, could be for OpenAI, API credits, could be for some research staff salaries, research assistants that I hire, and things like this, I would need to budget in an extra $61,000 on top of the $100,000 of direct costs for ICR, bringing the total grant to $161,000. So it actually turns out, however, that in practice, Duke doesn't actually collect 61 cents over every dollar of direct costs on every grant. That's part of why ICR is pretty misunderstood. Super.

Host (Santi) (11:07)

And before we move on, just to set the stage here, my impression is most of the federal support for research that gets done in the US happens through project specific grants, Right. It's not these massive institutional block grants. Most federal funding for research that you or I do as researchers happens in this system, is that right?

Pierre Azou (11:30)

I think that's right. By and large, there aren't a lot of infrastructure grants in the science funding system in the U.S. there are other things than project grants. There are such thing as, you know, center grants that fund kind of groups of investigators, and sometimes those can get pretty large. Right. So like NIH grant for a major cancer center, like Dana Farber, you know, could be the tens of millions of dollars per year. Right. But even those will collect overhead. Right. So whether it's a project grant or not a project grant, the mechanics are kind of similar.

Dan Gross (12:06)

In the past, the US did invest more heavily. US science funding agencies did provide more funding for infrastructure and the kinds of instrumentation that you need to perform research through block grants. In the 1960s, really, the NSF and the DOD were kicking up major programs to establish new data collection efforts. Actually, I would say it's especially around data collection, whether it's the deep sea drilling project that the NSF ran, collecting core samples from the ocean floor around the world. Pretty cool. Whether it's radio astronomy, whether it's observatories, whether it's. For example, here's one that most folks may not know. It turns out that in the late 50s and into the 1960s, DARPA. Back then, ARPA was investing in nuclear test detection in order to be able to monitor adherence to nuclear test ban treaties. And some of these were satellite observation methods for atmospheric testing. Some of them were seismic measurement methods for underground testing. It turns out that ARPA supported the installation of a network of seismic monitors around the world. Those seismic monitors are responsible for validating tectonic plate theory. The readings that they recorded over the next decade essentially map the tectonic plates of the earth. And that's the kind of, let's say, large scale investment in research infrastructure that is not as common in the US research policy enterprise today. But we did see a lot of 60 or so years ago.

Host (Santi) (13:36)

That's fascinating. I think I learned last year how modern that validation of tectonic plate theory was that until, you know. Right. Until my well into my grandparents lifetime, we didn't really know if tectonic plates existed.

Dan Gross (13:48)

I know, right? So Santi, when I don't know. I'll ask you a question. When were you born?

Host (Santi) (13:51)

1997.

Dan Gross (13:52)

97. Okay. So I'm a good decade older than you. I was born in 1985, you know, when we were learning tectonic plate theory in the 1990s. Like it's something everybody had always known. Right. It turns out that it only been known for maybe 25 years.

Host (Santi) (14:03)

Right. It's so cool.

Dan Gross (14:04)

It's wild.

Host (Santi) (14:05)

There's so much, there's so much more to learn.

Dan Gross (14:06)

Yeah.

Host (Santi) (14:07)

But I think the kind of fact here that I'm taking away is like, I think there's an image of federal funding for science as like the Hubble telescope. Right. These massive big pieces of infrastructure. And although that does happen, and it's happened historically, the median dollar the feds spend on science today is you apply for a grant, we give you a grant. It's not we're building the Hubble telescope, we're installing seismic monitors all over the globe.

Dan Gross (14:32)

You apply for a grant to fund a specific project whose contours you've outlined in advance and we provide you the funding to execute that project, including an astronomy.

Host (Santi) (14:41)

Really?

Pierre Azou (14:41)

Yeah. You want to do some observations at an observatory in Chile and you're going to need to buy a plane ticket. Not first class economy, very much economy. You're going to have to get yourself there and you are going to probably also budget use of the instruments there as part of your grant budget. So this is kind of an interesting example because here in some sense, access to the infrastructure is typically built in the direct cost of the grant. But that's kind of unusual.

Host (Santi) (15:14)

Got it. Let's move to current events. In February of this year, the new administration comes in and the NIH after a couple of weeks announces it's capping indirect cost reimbursement. It's saying 15% reimbursement for indirect costs on all grants. Why? Like, explain to me, before we get into your view, what's the administration's argument here?

Pierre Azou (15:37)

I think, okay, so the argument is pretty simple. The argument is there are cases where foundations only charge 15% overhead rate on grants and universities acquiesce to such kind of low rates, and the federal government is kind of entitled to some sort of most favored nation clause where no one pays less in overhead than they actually pay. That's the argument in this kind of half a page kind of notice. I mean, it's not much more elaborate than that. So that's the stated rationale.

Dan Gross (16:18)

The stated rationale. I think that's a crucial qualifier.

Pierre Azou (16:21)

Yeah.

Host (Santi) (16:21)

Well, and just before we get into, you know, what you read as the subtext, the idea is like the Gates foundation says, hey, yeah, we'll give you a grant to do some sort of health research, and we're only going to pay 15% indirect costs. And some universities say, yeah, sure, thank you, we'll do that. So they don't need the extra indirect cost reimbursement, is the idea, right?

Pierre Azou (16:44)

I think so, yeah.

Dan Gross (16:45)

Well, it implies they're willing to accept a 15% rate and take the grant funding on that condition. That's true. Now, whether you can extrapolate from that to federal research funding is, first of all, a different question. Let alone if federal research policy supported less research, was funding less research including even maybe less overhead, would foundations that also fund research, first of all make up some of the difference, or second of all, even continue funding as much research when the resources provided by the federal government were lower? Those are also open questions. Right. I think it's reasonable to interpret that foundations complement federal funding as opposed to substitute for it, and maybe less interested in funding research if it's less productive.

Host (Santi) (17:28)

What are some reasons that argument might be misguided?

Pierre Azou (17:31)

Okay, so first, universities don't always say yes. Sometimes they say no.

Pierre Azou (17:39)

You're going to need to find some other way to take this money. But we don't want to be involved because we anticipate that if we say yes, the federal government is going to have a problem with that. This is not something that I know firsthand. But at mit, to get a grant like this means getting special authorization of the provost, and that special authorization is not always forthcoming. That's what I know. And I think that there's basically the provost as kind of a special fund, presumably funded out of the endowment that under certain condition will kind of dip into that fund to make up for the kind of missing overhead.

Host (Santi) (18:21)

So just to put color on it, you've got some research that for whatever reason the federal government won't fund or won't get funded by the public, and the Gates foundation is only willing to fund it at this low rate. And the university has kind of budgeted a little bit extra for those kinds of grants that it still wants to get funded.

Pierre Azou (18:39)

That's basically my understanding. This is kind of secondhand understanding. But I know that if you're going to get a grant like this, you're going to have to sit in many meetings and cajole any number of administrators to try to get your way. And you don't always get your way. Second, it's not an apples to apples comparison because many times there are ways to budget as a direct cost in a foundation grant budget what the government would actually consider indirect costs. So you might budget some fractional access to a facility so you end up getting more indirect costs, more overhead through the back door, because they kind of know, I mean, the foundations actually know that it's not really viable in that form.

Host (Santi) (19:25)

So when you say fractional access to a facility, you mean in a federal grant I might not list in my direct costs like I have an office in the, you know, psychology building and I need access to that office to do my research. And I might include that as a direct cost in my Gates foundation grant proposal.

Pierre Azou (19:45)

I think it's unlikely to work for the kind of the implicit rent or your office space, but it's going to work for other types of fixed costs that are more obviously research relevant.

Host (Santi) (19:56)

The mouse microscope that I have to use maybe.

Pierre Azou (19:59)

Yeah, some sort of the cryo em machine, whatever it is. But I think to me the more fundamental way in which that kind of approach is misguided is in some sense the government want its infrastructure that it has contributed to by having this structure of direct indirect costs. They would want that infrastructure leverage by other funders. It's already there, it's already been paid for. Right. In some sense it's sitting idle. And we can kind of get more bang for our bucks if we get those additional funders to kind of piggyback on that already paid for infrastructure investment.

Dan Gross (20:40)

That they might not be interested in funding themselves otherwise.

Host (Santi) (20:44)

Why wouldn't they be interested in funding it otherwise? I mean, what would be the problem with the federal government saying, hey, we're just going to pay less. We want somebody else to pick up the slack. If it's important research, somebody else will pay for it. What's the hole in that logic?

Dan Gross (20:58)

Well, I Think we're really talking about scale economies problem. Again, these are fixed costs. That means that the more they are utilized, the more they get spread over individual research projects and units that apply them. And there's no getting around the fact that at least for the past several decades, the federal government has funded an order of magnitude more university research than private firms or foundations have. If you go look at NSF collected survey data that actually collects information of how university R and D is funded. About 60 to 65% of university R&D in the US is funded by federal research funding, if I recall the numbers correctly. And more like 6% is funded by foundations. That really is an order of magnitude difference. And it just means that the federal government that really has the scale to support and extract value for whatever its goals are for American science investments in overhead costs that come from equipment, that come from facilities that come from the things you really need to perform research today.

Host (Santi) (22:03)

I should lay my cards on the table here for listeners who may not be familiar with the kind of pre existing beliefs I bring to the table here. I mean one is I've got a ton of problems with the American scientific enterprise as it currently exists. But I think that it's really obvious when you look at studies from a wide range of folks that like R and D research and development in American universities is hugely positive and valuable. Like whatever other critique you could have of the scientific ecosystem, it's a pretty clear econ finding that federal R and D dollars more than pay for themselves. Like we should probably spend a lot more on R and D. We get more back. So I just want to flag that for listeners as we get going. I want to leave room for all kinds of critiques of scientific ecosystem, of the universities, of individual research ideas. But at this 30,000 foot level, federal R and D dollars I think are really well spent. It's a good place to spend money.

Dan Gross (22:58)

That may be true, the evidence may suggest that, but that's not really where the political and public dialogue around science policy is. I think that's again, I'm going to bring in a long arc here in the 1950s and 1960s that is more or less where it was. We got to take chances, win a race with the Soviet Union. If we want to win this race, we're going to have to take some risky bets. And the US did. And it was in many ways more flexible with its resources and investments in science, university and industrial science, especially related to defense aims at that time. But over time, partly with the waning of these geopolitical pressures partly with new budgetary pressures. I mean, this is my read of the history. The tenor shifted from let's take chances to let's make science and also other parts of government more accountable. And I think the undercurrent of indirect cost recovery policy debates has much more of this accountability framing to it, which comes up in this comparison to foundation rates of, okay, is the government overpaying? Because clearly universities are willing to accept less from foundations. It comes up in this theme or this perception that indirect cost recovery is funding administrative growth that may not be productive or at least socially efficient. And, you know, reconciling what you just pointed out, Santi, with, with some of this context, I think that's the way I make sense of it. Accountability really does seem to be a priority in the current day.

Host (Santi) (24:24)

So we're recording this on August 25, 2025. And I don't want to spend too much time on the specific happenings right now, but it's probably just worth flagging them for listeners, even though it might be outdated by the time we air this. But where are we right now on that 15% cap on indirect costs?

Dan Gross (24:44)

I mean, let's go back to when recent changes first kicked off. On February 7, NIH posted its supplemental guidance that introduced a policy that going forward, indirect cost rates that it paid on its grants would be 15%, particularly specifically to institutions of higher education. Ihes that policy was then adopted by NSF, by DOE, by DOD. Even all of these have gotten held up in court, been litigated by universities, university associations, and so on. And so, yes, right now things are stuck in some legal limbo. And Congress has presented its point of view that, at least for now, I'd like to keep things as they are. But this has been an object of controversy for long before the current administration even took office in January. And so, like, I don't think this is going away.

Pierre Azou (25:32)

You know, if I had to guess, the proposal as it first took shape is not what is going to end up being adopted, but I think what is going to stay relevant probably is the idea that overhead rate as kind of an object of controversy, it needs to be looked at, it needs to be reformed. The view that seems to have the upper hand is that overhead rates are too high, like it's too big of a share of the overall pie. And so by hook or by crook, we should find a way of reducing them.

Dan Gross (26:09)

And partly.

Dan Gross (26:11)

That'S because it's a complicated issue. Partly there's not a real benchmark of what indirect cost Recovery should be what's an appropriate policy is any way you approach this and any way you try to fund the cost of research, you're going to run into some trade offs. And those are complicated. I don't know that we're ready to get into those just yet. But, but any, any approach to funding overhead is going to have some pros and cons to it, as we've discovered in our work. As it is though, as it is, I think one thing that I wouldn't want to leave this thread without pointing out is that yes, indirect cost recovery does draw criticism. People think it's bloated or lacks transparency. And I think we would agree that some of these critiques are well founded. And yet I think it's also important to remember that F and A stands for facilities and administration. Those are the overhead costs that indirect cost recovery pays for. So it doesn't just fund administrative costs, which is what people usually at least mentally associate it with, even though we're very familiar here in this conversation with the research process and understand that facilities and infrastructure are part of how research gets done. Administration is also part of what ICR supports. Now it's not. At most universities, the majority of what ICR supports, the share of ICR that goes to administrative costs is actually capped. It's legally capped at 26% of direct costs. That cap has been in place since 1991 and many universities have been at that cap. You can see this in public records for many years. And so actually the idea that indirect cost spending and indirect cost rates are going up over time, and that's because of bloat at US Universities is it has to be factually incorrect because again, the administrative rate has been capped for three decades. And in any case, even then, many of those costs are incurred in service of complying with regulations that govern research. Right.

Host (Santi) (28:05)

What kinds of regulations?

Dan Gross (28:07)

Yeah, federal regulations governing federally sponsored research. And even just administering ICR is expensive. Compiling great proposals every two to four years and a new round of negotiations, all of that takes resources. And those are the kinds of things, among others, that indirect cost funding reimburses. And then, even then, universities appear to under recover their true indirect costs of federally sponsored research. We have examples from specific universities at different points in time which have reported more detailed numbers. And that kind of under recovery of the actual overhead costs associated with federal research means less incentive to invest infrastructure, less capacity for innovation, fewer clinical trials. So you know, there are, even in the current system, there's a case to be made that indirect cost funding is too low. Now we're not making that case. I think it's ambiguous what a best policy would look like. But this is all to say that first, public understanding of this complex issue is sometimes a bit murky. And second, a path forward has to embrace the trade offs that any particular approach to ICR presents.

Host (Santi) (29:25)

Yeah, you're flagging two things that I learned from reading your paper. One being, I don't know if I learned this exactly. I just got a much better sense of the fact that a ton of the administrative costs, the administrative bloat of the modern university is responding to federal regulations on research that a lot of the things that you have to do the paperwork class, this is famous, or at least famous in my professional circles fact that the average researcher spends reports spending like almost half of their time on paperwork. Some of that is a consequence of the research itself or the grant process and some of it is the regulatory compliance. And the other thing you're flagging, which I want to hear more from you guys on, is just the point about like the, the actual tools that we pay for in research seem to be becoming more expensive, more complex over time. So the, the facility maintenance and the microscope that I'm using today are, you know, an order of magnitude more expensive than the microscope I was using in 1950, say. And that has to, you know, you gotta recoup those costs somehow.

Pierre Azou (30:34)

Everything costs more than it used to. Right? I mean research is in a lot of respect, not all respects, but subject to the bommel cause disease. So I think there are some areas where there's been productivity gains. I mean software has had a lot of impact on research in multiple fields. So I think it's just complicated. I think the bottom line is like I don't feel we know if there is under recovery or over recovery of kind of indirect costs. We should recognize that it's of course the incentive of the university administrator to claim that there's under recovery. So I take that with a huge grain of salt. But I think the most important thing from a policy perspective is that we used to think that this is the least sexy meta science topic possible to study. So for a long time we were sitting on this data and we were not doing much with it because we thought there were five people in the world interested in this topic. But the stakes are high because if we get this wrong, we're sending participants in the research ecosystem at large that they should bias the type of research that they're going to invest in the type of training that they're going to undergo with an eye to what is cheaper. Right. So if we end up reducing overhead rate by a lot, then over time we should expect research that in some sense has less fixed costs and more variable costs to gain in favor and research that is more kind of scale intensive to lose favor. And there's no reason that we should imagine a benevolent social planner to find out a good development. Maybe it's the scale intensive research that's going to be more fruitful in the future, or maybe it's not. The point is that the government should kind of really be neutral with respect to kind of the cost structure of research activities.

Dan Gross (32:39)

Is it possible to be neutral, Pierre? I mean, a choice has to be made.

Pierre Azou (32:43)

Well, I'm not sure we, you know, I'm not sure we can be fully neutral, but we could aspire to not be biased by design. No.

Host (Santi) (32:52)

Well, let me ask you a question I think I know the answer to, but I think it's still helpful here just to get you to say it explicitly. Well, why wouldn't Kritik respond? Okay, we're just going to fund, you know, a little bit on top of indirect costs, but we're not going to subsidize stuff that takes huge amounts of overhead. And if universities want to build that fancy new telescope because it's valuable, they'll do it. Like, why is that wrong when it comes to science funding?

Pierre Azou (33:22)

It's not completely wrong. I mean, there's a grain of truth to it.

Dan Gross (33:26)

With what resources though? That's the problem. It's who's incentivized to invest in this infrastructure. There's not a market for, not a paid market for science. Now universities can generate some licensing fees from any patents that result from science, that it's research is produced and so on, but those are meager revenue streams, realistically. And we don't need to go on a whole tangent about why commercial firms are under incentivized to invest in basic scientific research. But there are lots of reasons to believe that. And this was the state of the world before 1940 to 1945, when really the exigencies of a war.

Dan Gross (34:08)

Drew the federal government into funding research in order to win a war. And then it was productive enough that folks decided we should keep doing it. And yes, you know, prior to 1940, the scientific enterprise was just dramatically smaller because there just wasn't funding the way that there is today, let alone even in the 1950s, shortly after. And so I think history and economic logic tells us that you're just not going to see as much science, including in these, and especially in these fixed cost heavy endeavors when those resources aren't provided by the public.

Pierre Azou (34:44)

I fully agree, but I think the answer of the. You were kind of cosplaying the administration. Right? You know, I guess one possible direct answer to the question is, oh, the endowment is going to pay for it.

Host (Santi) (34:56)

Sure. Okay.

Pierre Azou (34:57)

To be clear, yes, MIT has an endowment, but many other universities do not. So what does that mean for them? Right. And second, the administration also wants to tax the heck out of the endowment. So, yes, with what resources is this research going to be done?

Host (Santi) (35:14)

Sure. This is maybe a good opportunity to actually look at the empirical work you guys did. In this great paper. You pulled together data that I think represents 85% of all NIH funding over the past 20 years. As far as I can tell, this was one of the first real looks at what do indirect cost rates actually look like in real life? Not what's the sticker number? What did you guys find?

Dan Gross (35:40)

So let me kick off by saying one of our main findings, probably our most important finding, is that there's a large gap between the sticker rates, the negotiated ICR rates that are visible to the public that get floated around on Twitter as examples of university exorbitance. The big difference between that and the rates that universities are actually paid in practice, at least on NIH grants that we think likely the case for NSF and other agency grants too. And so let me explain how we get there. So, Santi, as you were beginning to introduce, in order to better understand and direct cost funding, at least for NIH, we pulled together data for around 350 universities and other research institutions. Together they account for around 85% of all NIH research funding over the last 20 years. We looked at both their negotiated indirect cost rates and what we call their effective rates, which is, as we see in NIH funding data, the ratio of the indirect cost funding an institution receives for every dollar it spends on the direct costs of research. We get the negotiated rates from institutional indirect cost agreements that each university enters into with the government. And look, Pierre and our other co author, Bobin Sampat, I mean, Pierre, it must have been a decade ago, you and Bobin, maybe more than a decade ago.

Pierre Azou (37:03)

Two decades ago.

Dan Gross (37:04)

Two decades ago.

Pierre Azou (37:06)

I foia, I see for the first time in direct castrates. Yeah, more than 20 years ago.

Dan Gross (37:12)

More than 20 years ago, Santi, it's science that sat on the shelf. They collected these data. They never quite made their way to a specific paper to write with them. And so this is just a resource that was quietly sitting on the shelf waiting for its day. And that day came this past February. So Bob and npier collected information on negotiated ICR rates, really for the past 60 years for as many institutions as they could get their hands on. And that came from the agreements themselves during this project in this past year. We also collected the most recent versions of those agreements from university websites and other places in order to try to bring the numbers up to the current day. And then we use public NIH grant funding data to measure the effective rates. Right? So our key finding then was the following. We found that negotiated indirect cost rates have been rising over time. Those rates, as we've discussed, are determined through this complex audit process and a negotiation with the federal government. But concretely, what the rate comes down to is comparing the actual incurred indirect costs of research associated with federally funded activity in recent years and dividing that by the direct costs of that federally funded research in order to come up with the rate. So you have indirect costs on the top and direct costs on the bottom. You take that ratio and you say, okay, if that ratio goes up over time, as we've seen in the data, as indirect costs have gone up, that has led to concerns that the overhead cost research is going up, including these claims, right, that it's funding administrative bloat. So, you know, that's what it looks like on the negotiated official indirect cost rate side. If you look at the actual NIH funding data and you look at institutions effective ICR funding, that is their indirect cost funding over the direct cost funding, what's actually paid, it turns out their effective ICR rates are much, much lower than their negotiated rates and they haven't really changed much for 40 years. Likewise, if you look at NIH's annual budget, the share of total extramural grant funding that goes to indirect costs has been roughly constant at 27 to 28% of total funding for a long time. And that implies an effective rate of around 40% of over direct costs. So look, even though many institutions have negotiated rates today between 50 and 70%, what they actually receive is usually more like 30 to 50% on average, or at the median, around 40%. And here's where the story starts to get even more nuanced, but in a very important way. I'll try to explain this as crisply as I can, and hopefully you and your listeners can follow along. The difference between those negotiated rates and the effective rates seems to us, from what we can tell, to be due to limits and exceptions built into NIH grant rules. Those rules exclude some types of grants from full indirect cost funding, such as training grants. They also exclude some types of direct costs from the denominator of these rates, from the cost base used to calculate ICR rates and payments. So the implication of that is that institutions receive ICR payments based on a smaller portion of of their actual incurred direct costs than typically assumed. And then what happens? Well, mechanically, as the denominator of that negotiated rate falls, that is to say, the included direct cost base shrinks, the rate mechanically has to rise. Even with the same indirect costs, same overhead costs. If the denominator is going down, the ratio has to increase. You have to see a university being paid a higher indirect cost rate off a smaller modified direct cost base to recover the same amount of overhead costs.

Host (Santi) (41:03)

So let me back up here. What's the historical trend that's happening here that you guys are seeing in the numbers? Is it that over time, more and more things are being carved out and the federal government is saying for grants generally, hey, we're not going to, we're not going to reimburse that as an indirect cost?

Dan Gross (41:20)

So this is a great question, and it's one we've really wrestled with. The negotiated rates have gone up over time, and now look, at least to the layperson, large. But the effective rates haven't changed for 40 or 45 years and are more modest, at least around 2/3 of those negotiated rates. And you're asking the question of why? What would really drive this gap? And here it gets tricky because without actually seeing the specific direct costs that go into every grant that NIH funds, which we can't see, that's not visible publicly, it's hard to fully diagnose that. I mean, at the end of the day, we found this growing gap between effective and negotiated rates. We have to find some way to, to make sense of it. This is where we shift, have to shift a little bit from assessment to speculation, at least in my view. And one could say, okay, so let's think. What's excluded from modified total direct costs? Well, one is researcher salaries above a certain level. That level is actually relatively high. Not many researchers are at that level. Especially when you consider that a lot of research salaries are really being paid to postdocs and grad students.

Host (Santi) (42:23)

What is that level? Can you give me a dollar amount?

Dan Gross (42:27)

I don't recall exactly, but I think it's between 200 and 300,000 annual salary. So high. Yeah, yeah, no, that's high. And for what we understand about scientific salaries, and also who supported these grants. There aren't enough people being paid that kind of a salary on these grants for that to explain the difference. It just can't be there.

Host (Santi) (42:46)

You're looking around at the scientists in your academic institution and thinking like, yeah, that's not where the money.

Dan Gross (42:51)

No, it's not. I mean, it's like even if you consider PIs, but even beyond that, if you consider postdocs and grad students, it certainly isn't.

Dan Gross (43:00)

My best hunch is that research projects have become a bit more capital intensive and only a certain level of expenditure on equipment can be included in the modified total direct cost base. And so again, this is a hunch. I don't even have smoking gun evidence on this, but it's my intuition for what we understand about the problem. My hunch is that that's the cost category that's driving it. So you have some equipment that's being budgeted into projects where the indirect cost recovery that can be charged against that equipment is capped and that could drive the winch.

Host (Santi) (43:36)

Well, let me ask. In the paper you guys did, there's this really fascinating chart where you show basically the institutions that would get hit hardest by a 15% cap tend to be the institutions that do the most valuable medical research, most useful clinical R and D. What did you find?

Pierre Azou (43:58)

One thing that we can quantify is kind of the relevance of that publicly funded science for private sector R and D in the form of patents built on the original NIH research. And then we can very, very simply just correlate those two things. So the takeaway is that the more a university or research institution sends to lose under the administration policy, the more it contributed over the past 25 years in research that the private sector found relevant in terms of patents related to pharmaceuticals.

Host (Santi) (44:44)

So this would be, this is kind of counterintuitive if your whole kind of model of funding for science is, oh, let's just cut the subsidies for the stuff that the private sector doesn't care about. All this big stuff that's this big equipment. Turns out when you cut those subsidies, the stuff that suffers most is the stuff that the private sector likes. The patentable clinical, the drugs. I'm painting with a really broad brush here, but I guess there's like this funny relationship, right, where like the, the private sector likes the stuff that's getting subsidized. It pays for that stuff on the back end.

Dan Gross (45:15)

Right?

Pierre Azou (45:16)

So I actually, to me, it makes perfect sense. It's not counterintuitive at all. This is the stuff that the private sector would not be willing to invest on its own. But that research having come into being, it now is a very, very valuable input into activities that profit minded investors are going to find kind of interesting and relevant and worth taking a risk on.

Host (Santi) (45:42)

This is basically like just the argument for basic research, right. For the government to fund, support basic research.

Pierre Azou (45:49)

And the thing is like that argument has been made kind of at the macro level forever. But the bibliometric revolution of the past 15 years really allows you to kind of look at this at, I don't even say at a micro level, it's at the nano level.

Dan Gross (46:08)

So.

Pierre Azou (46:09)

So like recently I've, I've been able to look at, for a different paper, I've basically looked at the history of Ozempic and it's completely fascinating. So if you look at the, the patent for Ozempic, the main patent, it hardly cites any research, any publicly funded research at all. Actually it cites zero publicly funded research at all.

Host (Santi) (46:32)

Okay.

Pierre Azou (46:32)

But it cites a bunch of patents, including patents taken up by academics. And those patents in turn cite the foundational research performed by Joel Habener and his team at MGH in the early 1980s that basically elucidated the role of GLP1 kind of as a target, as a potential target. In the 80s I saw the research as well. Yeah, in the early 1980s. This is a grant, actually. I looked at this example in detail. I think this grant was first awarded to Habener in 1979 and then was. Every four or five years is being renewed and I think it died like in 2008. 2008, Joel Habanero kind of moved on to other things. Right. Wow. So those entire exchange are kind of like very kind of complex, but we can actually put them under the microscope. Like we can actually figure out what happened. And so like the macro picture that we always believed in, based on a general understanding of what basic research is good for, you know, we can now validate it at this more kind of granular level and we can now leverage this data to say things that are kind of more specific to try to kind of think about the likely impact.

Dan Gross (47:53)

I do want to add one qualification and I think it's a useful one because it also suggests some opportunity and directions for the future, which is there are things we still can't see despite peers zeal, which again I relate to quite a bit. It's for example, in our projections of what the consequence of 15% rate or rate cap might be. Our projections are still pretty coarse. We're just supposing what's going to happen if NIH funding to these institutions declines when their effective rates might drop from whatever they currently are to 15% at a cap like this. And when that's the case, we don't quite know what research might not take place. We don't know, for example, what are the indirect cost categories that are exposed, if any. We don't know how universities would reallocate. And all those things are first of all going to be difficult to project without a proper experiment or experience. But one thing that I think I would have loved to have more visibility into is what is the structure of indirect costs at universities across the country. Now, that might be proprietary, insensitive, but even at a slightly disaggregated level from the mainline numbers we see in NIH grant data, where you just see the indirect cost total and the direct cost total, we don't really know what lives inside the indirect costs in practice. That also makes it hard, for example, to assess what share of paid indirect costs actually are going to administrative expenses, let alone what are those being incurred for and by. Or what are the direct cost categories that are being excluded from modified total direct costs might be generating this wedge between negotiated and effective rates that we talked about. We would need a little bit more transparency into the system to be able to know the answers to these questions. And so, you know, that's all to follow up on Pierre's observation that there's a lot we can do today at a very disaggregated level by saying that there's even more.

Host (Santi) (49:53)

I would like to know, does that information have to be proprietary and behind closed doors? I mean, it's part of negotiations of the federal government about how much the taxpayer will pay, you know, Harvard or MIT or Duke or what have you for it's overhead on these grants, like which piece of this is so special and proprietary that it can't be shared? It's like a trade secret that Duke is buying a big new piece of lab equipment.

Pierre Azou (50:22)

You're talking to the wrong people here because, like, we're meta scientists. So our answer is none of it should be private. None of it should be private.

Host (Santi) (50:30)

I think we all agree, right? We'd love to see this stuff.

Dan Gross (50:33)

That's right. But now you have to ask the university lawyers and they'll tell you what they think, right?

Pierre Azou (50:37)

I don't have a good explanation.

Host (Santi) (50:39)

What would, what would the, the case from the university side be of? Oh, no, we can't tell you what we're spending. We can't tell the public what we spend.

Pierre Azou (50:45)

I, I think my sense is that there are lots of institutions of academia that strike most lay people as completely bizarre, hard to explain without context, hard to explain, illegitimate. You know, people who don't understand it and haven't thought about it will find it so bizarre that they will typically jump from the odd aspect of it to basically that must be corruption. And I think university administrators are hugely attuned to that aspect of things. And so the natural defensive approach is to shroud it in secrecy. This way we don't see how the sausage is made.

Dan Gross (51:32)

I think transparency can be both a blessing and a curse. I mean, the reality is, in my view, that more information supports more considered decision making and assessments. On the other hand, it also opens the door to misinterpretation, misrepresentation by critics who have their own agendas. And so how to weigh off those trade offs is that. That's a complicated problem that I don't have the answer to. And I think Pierre's right that there are some practices you see at universities that to the public might look unusual or even some that might be familiar. But one might say, how is that a useful expense? I mean, even a simple thing like having an administrator, right? Having someone who, let's say, might manage a faculty's calendar, that might seem excessive. Many people out in the world manage their own calendars. And at the same time, when you think about how someone's time is best used given limited time of the day, given their respective expertise and very heavy investment in specialized human capital and science, is emails, calendaring, note taking, like, are these the right things for scientists? Paperwork? Right. All the things you talked about, where scientists spend a large chunk of their time now on administrative tasks, especially administering grants. Is that the right use of time? Or does it actually make sense to outsource that to someone else and preserve the scientists time for doing more science? These are things we can debate in the abstract. When you actually put forward data that show that some share of federal research funding is going to fund administrative costs, at first glance it might look wasteful or unnecessary, and yet it might still be productive. These are the kinds of things that I think are at the same time challenges that transparency could open up. But you know, by and large, to me, at least, my view is that I would understand the system better and be able to make a more considered judgment over what a reasonable path forward might be if I had access to more facts, including what indirect costs look like under the hood.

Host (Santi) (53:36)

Totally. Well, let me leave you Guys, with one last question about what other changes you guys would like to see to the indirect cost system. This is one of them. More transparency. Let's actually know what's being reimbursed under the hood. In a world where you guys have the ear of Senate appropriators and political leadership at the nih, maybe you guys have some sway in what the universities will lobby for or accept. You know, in a world where you guys occupy that position, what would you be pushing for on indirect costs?

Pierre Azou (54:09)

So you're giving us a magic wand.

Host (Santi) (54:11)

Not fully magic wand, let's say, you know, in the realm of the possible. But you guys get to, you get to talk to everybody involved and you get in the room with every major player here. You get to make your. Make your case. What are you making the case for?

Pierre Azou (54:23)

Okay, so let me start. This is not something that's in our paper, right, because we try to give kind of alternatives, but my favorite policy reform, you know, I've come to think that this indirect cost rate is very much a second best institution, you know, very much terrible, and may yet be superior to many of the alternatives that are being contemplated. My favorite alternative would be one where there would be kind of a flat rate that's applied not to modify total direct cost, but just to the direct costs. And that rate would be something that's more or less like the average effective rate that is currently observed in the data. So something on the order of 40%.

Host (Santi) (55:13)

And what would the point of that be? You're swapping out this complicated system to basically, in the end, reimburse universities 40% for just a formal. Yep, we're reimbursing you 40%.

Pierre Azou (55:24)

The idea is we know there are fixed costs. Those fixed costs need to be paid. We could have an elaborate bureaucratic apparatus to try to get it exactly right, but it's kind of mission impossible. So why don't we give up on that right from the start and we set a rate that's at a level such that it's unlikely to lead to kind of large errors in either under recovery or over recovery. And so I'm not particularly attached to this 40% rate, but certainly something that's kind of higher than the 15% that was contemplated. That to me seems absurdly low. Dan, come at me.

Dan Gross (56:06)

Okay, come at you. I don't know that I'll have to actually aggress that strongly, at least at up here, but I'll point out a few things. First, just to reiterate, in the work we've done, we do lay out a number of different approaches that we can imagine to this or that have been discussed in the past or present to this problem. And the reality is I see a lot of gray. It's hard to take a strong point of view given that every option has some pros and cons. But I'll give you something anyways, Santi, so let me talk through just a few things here. The 15% rate that NIH has attempted to implement in other agencies too. This is simplifying. It's flat, right? It generally cuts out the negotiation. Well, actually wouldn't fully cut out the negotiation process in the sense that to even receive that 15% you have to document your overhead costs and demonstrate that they reached that level or exceed it for that matter. In any case, a 50% rate is simplifying. It forces more cost sharing and maybe more judicious investments by universities. But it's also so low that in my view, it's likely to make a significant amount of high value and ultimately life extending and improving research economically unattractive for universities. The current system, on the other hand, is complicated and burdensome on the margin, might encourage investment in less productive things. And that's particularly because universities can get it paid back through future icr. At the same time though, you know, it provides pretty good incentives, pretty robust incentives to take on expensive, high cost, but also high value research on behalf of the public. In between the current system and the proposed 15% rate, there are other options. And yes, I think if you put my feet to the fire, I would land on one of maybe two alternatives. One of those is close to what Pierre said. This would be a case where we have fixed rates though allow them perhaps to vary by institution type. So maybe one for universities, one for medical schools, one for independent research institutions. Because we do see some variation in their cost structures across them. And we might set those rates around their historical average effective rates since those haven't actually changed for quite a long time. We would ditch this complicated modified direct cost accounting that mucks up our understanding of the whole system, let alone creates a lot of cost administration and just pay ICR over the full direct cost of research as we calculate the effective rates to be. So that's close to what Pierre said, but maybe with one dimension of variation across institution type. Now, of course, if you refine and cut the categories of institutions and set different rates for different categories, the more finely you slice the pie, the closer you end up to actually the current system. So that's why I said maybe just at a very high level, four categories. The other adequate imagine is to shift more of these costs above the line. So in other words, to adapt the system to enable more of these indirect costs to be budgeted at as direct costs in grants. This isn't always easy or feasible at all, but presumably some things we currently call indirect costs could be accounted for in a direct cost manner. Foundations do it a bit more than the federal government does and so that could be another path forward. And so there's no silver bullet here. I think at the very least, our goal with this work was to try to bring some understanding, both conceptual and empirical, to this thorny problem and to this long running policy dialogue or debate over how to fund the indirect cost of research and what appropriate rates should be. It's been a recurring question for several decades and of course now is in the hot seat again. And so hopefully through this work we've been able to help push that dialogue along. Sure.

Host (Santi) (60:03)

Pierre, any last thoughts before we close?

Pierre Azou (60:06)

What Dan said is the man excellent.

Host (Santi) (60:09)

Well, you guys are great. Thank you to the both of you and to your colleague Bhavan Sampat for doing this research. Thank you so much for joining.

Dan Gross (60:16)

Thanks.

Pierre Azou (60:17)

All right, I can see this thing is going to trend on Spotify.

Dan Gross (60:21)

That's right. Only if it cuts out my answers and keeps it to you. Then we've got a real, a real blockbuster.

Pierre Azou (60:28)

I would have said the opposite, but.

Host (Santi) (60:30)

Okay, you guys are a good pair.