How to Shape a Market

How to Shape a Market

"It's a delicate exercise when you're dealing with a monopolist"

Today's interviewee, Professor Chris Snyder, is a Non-Resident Senior Fellow at the Institute for Progress (IFP). He teaches economics at Dartmouth College, where he specializes in industrial organization and microeconomic theory. He is also a research associate at the NBER, treasurer of the Industrial Organization Society, and a faculty director for the University of Chicago's Market Shaping Accelerator.

Chris played a pivotal role in the advance market commitment, or “AMC,” for the pneumococcal vaccine, which saved close to a million lives.

What you’ll learn:

  • How did the U.S. and Russia end up in the same funding coalition?

  • Why didn’t we design an AMC for malaria?

  • How do you place a market value on future innovations?

  • Why would cancer and Alzheimer’s be poor candidates for an AMC?

Chris, in a February 2020 NBER paper that you co-authored, you wrote that “‘pull’ funding is meant to supplement, not replace, direct R&D support, or ‘push’ funding.” What's “pull” funding? What's “push” funding? And why should we care? 

Let’s say funders want to encourage more innovation, and they’re worried that the private market might not provide the right incentives. Some sweetening of the pot is needed, and push and pull are different ways to do that. 

Innovation in pharmaceuticals, from idea to clinical trials to eventual product, could be a 5, 10, 20-year process. Push funding comes at the beginning of that pipeline: the funder picks promising innovators and provides grant funding for them. It's basically payment for attempts to get these innovations. 

Pull funding dangles a reward at the other end of the pipeline. It provides incentives, potentially to parties unknown to the funders, to move along the pipeline in order to receive that reward. So pull funding is a promise of a payment for success. It could be the reward for an idea, like the famous longitude competition, or it could be a reward for the development of a practical product.

The commercial market is the most familiar form of pull funding. If you develop an electric vehicle battery that has a much higher range and doesn't use a lot of rare earths, there'll be a huge market for that. But in some cases, commercial incentives may not actually match enormous social needs. There's a role for governments, organizations, and philanthropists to come in and try to shape those markets. 

Push funding is the default way to fund innovation — the NSF, NIH, and other agencies have used it for decades. Pull funding is less common and maybe needs a little bit of boosterism. At the Market Shaping Accelerator, we encourage people to consider pull funding because it has stronger incentives. For ideas that might not have a serious market, it can get contenders working hard to push the product to market as quickly as possible. 

Why is push funding the default? You mentioned the famous longitude prize, but that’s a counterexample. 

I think part of the reason is that it's hard to imagine paying for something that doesn't exist. Governments are used to signing a purchase contract for an existing product. How in the world do you even design this pie-in-the-sky product that might not exist for 10 years, and somehow get it to market? 

Sure. You were intimately involved in designing a particular kind of pull mechanism, the advance market commitment, for the pneumococcal vaccine. What is an advance market commitment? 

As you say, it's one of a variety of possible pull funding mechanisms. It's basically a bundle of features, some defining and some ancillary. The defining features are implied in the name. A funder promises to either purchase a product that is not yet developed, or to add a subsidy on top of other consumers’ purchases. The novelty is that it's the promise of payment for an innovation that doesn't yet exist. It might be a vaccine for a neglected tropical disease, or it might be a method for permanent atmospheric carbon removal at a moderate price. So the promise might be 5, 10 years in advance of when the product will arrive. The implied reward is not a lump sum, but is tied to the amount of the good that's distributed. And it's for a practical product that will be rolled out widely. 

It's not as simple as an advance purchase contract — we haven’t identified all of the producers to sign a purchase contract with. It’s firm-agnostic, in that you say, “We’ll purchase from any firm that comes along and produces a good that meets the technical product profile.” It could even be technology-agnostic: we specify that it has to solve these problems, and we’ll fund any technology that does the trick. The “commitment” piece is that the products may have limited commercial potential. It could be that consumers are in poor countries, and purchases are made on their behalf by donors. There might be climate externalities or epidemiological externalities, so the consumers aren't internalizing the full social benefit of their consumption. When the market eventually materializes, it might not be that lucrative. That's the whole reason for the market shaping mechanism: we specify in advance that there will be a subsidy or a more lucrative price than you might expect. That gives firms some confidence that investments in R&D and capacity will get compensated.

Can you tell me a little bit more about the history of advance market commitments? The Center for Global Development produced a working report in 2005 about how you might deploy an advance market commitment. Had the idea of an AMC been around in the economic community prior? What's the genealogy there? 

I think the idea was floating around in development economics circles back in the ‘80s and ‘90s. It really crystallized with the work of Michael Kremer, who eventually won the Nobel Prize in Economics for his work on using randomized control trials (RCTs). He later worked on another paper with Rachel Glennerster and Heidi Williams, the Director of Science Policy at IFP. Michael and Rachel also published a book called Strong Medicine, which talks about AMCs in general and why vaccines for neglected tropical diseases are a really good use case. Malaria was the main use case people had in mind at that time. So that was the intellectual genesis of the idea, which then caught fire and moved from theory, idea, and policy to implementation. 

When you say it caught fire, what happened? 

I think Michael Kremer, Rachel Glennerster, Ruth Levine, their co-author, and Nancy Birdsall at the Center for Global Development are persuasive people. Michael was the Gates Professor at Harvard University. That was a connection to the Gates Foundation, which was gaining a big head of steam. They have an innovative approach where they say, “Let's hire the best people and pay them a pretty good salary to pursue the best ideas that will do the most social good.” I wouldn't say CEO salaries or CFO salaries, but getting that quality of person. 

So I think they persuaded the Gates Foundation. There were also some finance ministers who Michael and Rachel had a connection to who wanted to get on board. These were farsighted people in the finance ministries of various countries — Owen Barder in the UK is a PhD economist — who said, “This is a winning idea that could save a lot of lives. Let's work on piloting it.”

You mentioned that malaria was one of the initial proposed targets for an AMC, but the pilot AMC that you were involved in was for a vaccine to prevent pneumonia, some strains of meningitis, and sepsis. That’s a very different set of diseases — what happened between the malaria idea and pneumococcal implementation?

The best candidate might not be an ideal pilot case. Malaria was very scientifically difficult: it has the insect factor, and it's not a virus, so it's gonna be really hard to deal with. Suppose that the advance market commitment was successful and generated 10 times more investment than has ever been seen in that space, increasing the chance of successful product development from 10% to 20%. Even if it's successful in spurring all this investment, 80% of the time, you're still not gonna end up with a product; it'll look like a loss and be hard to measure. So I think the idea was to have a near-term product that had a higher probability of success. And that's why they moved to pneumococcal disease. Malaria was responsible for far more death and illness, but pneumococcal disease was still a worthy target. It was responsible for the deaths of up to a million children under five each year. In high-income countries, it's less of a killer because we can get timely antibiotics. In poorer countries where timely antibiotics are not as widely available, it has a higher mortality rate. There were already first-generation vaccines against pneumococcal, so the AMC was for a second-generation vaccine. That’s not as hard of a scientific problem. It was more about, “Can we complete the R&D that we need for the late-stage clinical trials? Can we scale up capacity to support a fairly broad, almost universal campaign for low-income countries?”

What’s the difference between a first- and second-generation vaccine? 

There are scores of different strains of pneumococcal disease. The first-generation vaccine was marketed and distributed in high-income countries, and it targeted seven strains that were more prevalent there. That vaccine, Prevnar, was developed by Wyeth, which was later purchased by Pfizer. There was already work underway on a second-generation vaccine that integrated the remaining 10% of strains found in high-income countries. That marginal improvement was actually worth rationalizing the R&D. And so the thought was, “Could we piggyback off of this and get the capacity scaled up? Because this second-generation vaccine that will conjugate 13 different strains would be a lot more effective in low-income countries.” 

Give me the timeline for your involvement in the pneumococcal vaccine development. How did you end up in this space? 

That was through my history of working with Michael Kremer. We noticed that pharmaceutical markets, and in particular vaccine markets, are just really unusual. It’s funny, but you can think of economics as the science of telling people that what they think is unusual about their markets is actually not. Vaccines do have some special features, though, so we spent some time just thinking about the microeconomic theory. That resulted in a theory paper about vaccines. 

When they decided to pilot this advance market commitment idea, which, in a sense, was Michael’s brainchild, he was part of the economic expert group to validate the program design. He brought me along because I was his co-conspirator in this theory and I had worked on vaccines; maybe I was there to do some proofs and think deeply about microeconomic theory. There was also Jonathan Levin, who was just named the president of Stanford. We were the theorists. We also had some economists who specialized in pharmaceutical markets, Ernie Berndt and Patricia Danzon. And then there were doctors from developing countries, industry officials, and scientists who knew about pneumococcal disease and vaccines.

Who was coordinating that expert group?

That was coordinated by Gavi. Andrew Jones was the head of the pneumococcal AMC part of their work. 

So you had interested funders in the Gates Foundation and others, and you had a coordination vehicle to talk across disciplines. How did you shop this idea around? 

That happened before I was brought on to the economic expert group, but I do think it was partly a matter of well-placed individuals in these organizations seeing the merits and getting the Gates Foundation and then Gavi interested. Together with the finance ministries of the countries — Italy, the UK, Canada, Russia, and Norway — they built a fund of $1.5 billion. That's orders of magnitude larger than the development programs we tend to see.

Did AMCs rub up against economists’ view of how markets are formed? Were you pushing for something that might not have been immediately intuitive to other area experts?

This funding was for future innovation and not subsistence consumption today, so it requires extra resources. $1.5 billion is a big fund. In a way, the timing was right. Countries were actually doing pretty well — it was a boom time and development aid was probably getting enriched because countries had extra money to throw in. This was before the global financial crisis actually hit in 2009, and some of these funds dried up, which is maybe part of the story. But as long as you can get countries when they're feeling a bit flush, the question is, “We want to expand development aid; what's the next idea?” 

Without making you name names, were there other development ideas that were rivalrous for the same funding? Were you arguing that this is a better deployment of resources than other interesting but maybe wrong-headed ideas in the space? 

There was some question about the sizing and pricing of the AMC, and one argument against it was that we were going to subsidize these vaccines that might have a pretty high price relative to, say, the polio vaccine. The pneumococcal vaccine was $7 per dose. This is multiples more expensive than a $0.50 per dose polio vaccine, which has been around for 50 years and is easy to produce. We know that polio vaccination is very cost-effective, and we're vaccinating every single person in the world that we can get a hold of. We try to feed people who are starving. What can we move to next that’s also cost-effective but might be more expensive? 

There was this new thought, “Why not just give money to poor people? Why give them vaccines?” I think that's an interesting consideration that should always be in the back of one's mind if you're spending $1.5 billion. Should we just do an income transfer and let consumer sovereignty reign? We could spend an hour making the arguments for why some of these income transfer programs are probably a good idea, but it's still worthwhile to have these vaccine programs. They have special benefits — if a consumer in a low-income country says, “I really want to vaccinate my kid,” they could scream as loud as they want, but that won’t create a market. These things require institutional moves to create markets and vaccines are typically purchased by governments and NGO programs. 

Was that debate between direct cash transfers and AMCs for vaccines a live issue in development at the time? 

Direct cash aid was growing in popularity at the time, so I think it was at the back of every developmental economist’s mind. It's a really good benchmark to compare other programs against, but the case for vaccines was pretty strong and there was funding for it. I don't think there was an either/or debate. 

Sure. You mentioned deciding how to size and price an AMC. How do you explain the economics to a layman? How do you make sure you’re not radically mispricing a subsidy for something that hasn’t been invented yet? 

That's a hard problem, and it was certainly an issue for the economic expert group for the pneumococcal AMC. With these programs, we're focusing on problems of great social need and with limited commercial incentives. That builds in a little bit of safety where you can say, “Okay, I should have paid $4 a dose, but I paid $5. I paid a dollar too much. But still, the social need is there.” 

We did some work on COVID vaccines, and when we wrote the paper, we estimated that a course of COVID vaccine had the social value of roughly $6,000 in terms of reduced mortality, morbidity, and getting people able to participate in the economy. The prices were ultimately more like 60 bucks a course for the most expensive ones. If you're getting $6,000 of social value, would you worry about paying $31 instead of $30? With problems like that, there's this asymmetric loss where if you go without the product, you lose so much social value relative to the small price you pay to the producers. But it's still an issue in these development areas where the funding is not so rich — dollars have quite a bit of shadow value to them. There's a political risk to being seen as giving away money to industry. 

Let's take the pneumococcal vaccine AMC. You can try to make some scientific estimates of how many people are dying and ill and how much of that would get mitigated by a vaccine campaign that's rolled out at a certain rate. Getting a monetary estimate there requires valuing a healthy year of life, but there are techniques for doing that. Even if we are conservative, you get some really high estimates of value. 

Then you try to think about the costs of R&D and manufacturing. A funder might have vague notions about that, but the firm producers have better private information. So there's this asymmetry of information that puts the funder at a disadvantage. Of course, the firms would like to say, “This is gonna be really expensive. We're gonna need a really high payment in order to justify the investment.” It is a hard problem, but there's a balance. 

For a program that has a really high social value, you're gonna err more on the side of having a high price, because that higher price per dose or top-up subsidy increases the chance that it's lucrative enough to induce the participation and investment of the firms.

You can actually work out some of the calculations if you have the time, but if you're rushing, you won’t have the two years that it would take to do this full-blown study. There was a huge range of uncertainty in the pneumococcal AMC pilot about the costs of capacity and production for the second-generation vaccine.

If you're a funder, how do you avoid getting snowed by pharmaceutical companies? If they tell you, “It's far more expensive than you think to produce this thing, and we'd like to triple that number,” what resources do funders, economists, and policymakers have to find a happy medium? 

You can try to look at analogous cases. If it's a vaccine that shares properties with other vaccines, we can compare the production costs. You can talk to former industry officials, which is somewhat fraught because you're not sure where their allegiances lie. Are they with their old industry? If they're talking to you pro bono, presumably it's because they're socially motivated, but it's hard to know. 

Another approach is to take a step back and, instead of identifying a specific disease as the target, run a competition for AMC candidacy among different possible diseases. If the disease is already chosen, the firms that are vaccine suppliers and doing the research have an incentive to say, “Oh, this is really going to be expensive.” But if you're competing to be the disease that the AMC is offered for, you don’t want to say it’s really expensive because that might actually disqualify you. 

If it's a really novel product, there might not be anything to compare it to, so there might be a huge range of uncertainty. If it's not a very high social benefit, then you'll really want to squeeze things tightly and economize on expenditures. You might not even run the advance market in that case because it's just too close of a margin. 

The pneumococcal AMC had an interesting mix of funders, including Italy, the UK, Russia, the U.S., and the Gates Foundation. What determined the coalition's makeup?

Yeah, that may never happen again. I think because this was a pilot program, there was some force of personality there. Some farsighted people working in different administrations saw the merits of the idea. I can't really say what was there for Russia. Norway was part of it, and they're actually the home of the Coalition for Epidemic Preparedness Innovations, which is a Gates-funded vaccine shop. So maybe there's some history there in Norway having a deep interest in vaccines. In the cases of Italy and the UK, they had forces of nature in the finance ministries who really liked the idea and pushed their governments to get involved. 

What was the effect of the pneumococcal vaccine? What do we know about the counterfactual world in which it does not exist?

We can't really say that this program generated the second-generation vaccine. The R&D likely would have been carried out to create this better vaccine that you could sell for more money to rich countries. Prevnar 7, the first-generation vaccine, was sold at $180 for a three-dose course. It turned out to be a pretty lucrative market. 

The question is, what good did this program do? Did it speed up the rollout of the campaign to low-income countries? Michael Kremer got the Nobel Prize in economics for running randomized controlled trials (RCTs) in developing countries. You might want to run an RCT to learn the causal effect of the AMC, but it's hard to do when the pilot program is $1.5 billion. You can't necessarily run 10 of those, which wouldn’t be enough data anyway. 

But there are some measures you can take to try to evaluate it. You can look at the historical rollout of vaccines and the lag before wide-scale availability in poor countries. Dissatisfaction with that lag motivated the whole advance market commitment to begin with; we see historically that it could be a 20-year delay. So you could say that one of the benefits was that it took something like five years to scale up to the same coverage in rich and poor countries. 

By the way, that $1.5 billion was just the funder subsidy on top of the copayments Gavi made for the vaccines. Over the course of the program, Gavi spent $3.3 billion, so the AMC fund was actually a fraction of the total spend. If you throw $5 billion at a problem, of course you expect to get something for your money. The question is whether that validates the AMC design.

In the paper I co-authored with Michael Kremer and John Levin in the American Economic Review Papers and Proceedings, we compared the rollout of the pneumococcal and rotavirus vaccines. Gavi invested a similar amount in the rotavirus vaccine, but not with an AMC design. Those are just two data points, but they suggest that it was about five years faster.

How was the money for the rotavirus vaccine disbursed? Was it a push commitment directly into R&D? 

That wasn't any kind of committed program or subsidy. It was just an advance purchase commitment.

For vaccines, a unitary purchaser like Gavi negotiates prices with the manufacturers. The polio vaccine is now basically a generic product with many competing manufacturers, but for these new vaccines, you're dealing with maybe three suppliers. So it's this bilateral monopoly kind of bargaining. Firms are worried that bargaining too hard might reduce their investment incentives. That's one of the possible benefits of this advance market movement: the funder commits to a price or subsidy that sweetens the pot, and they can’t later renege on it.

Knowing what we know now, would you change anything about the design or execution of the pneumococcal AMC?

We actually did a little bit of work on it then. The economic expert group was brought in, as these groups often are, to validate the framework design. We did start to see some subtle issues: there were really only two suppliers who might have had a viable product. GSK and Wyeth, which was bought by Pfizer, were the two firms that had the second generation of vaccines in late-stage clinical trials. One of them was not optimizing production for a low-cost environment because they were supplying to rich countries and charging what the market would bear there.

So that really left one viable supplier, and we had to be really careful with mechanism design. If you expect 20 or 30 suppliers to come in, you're going to get competition, volume, and low prices. When you're dealing with a monopolist, it's a delicate exercise. We were a little bit worried about the monopoly firm not supplying the full amount that the program requested. Capacity is expensive, so the firm could economize and give you what you want in 20 years instead of in 10. So you get half of the target supply. We tried to tweak the design to take care of that problem.

The rollout could have been more ambitious. They did get initial responses from the two firms, and the Serum Institute of India later submitted a third. Eventually, it got to pretty high quantities, but it took five years to really scale up. You could have imagined hitting the target faster. Gavi started to get a little bit nervous that this thing was going to take all of their resources. And we were hitting the global financial crisis, so the largesse of development aid was drying up a bit. 

Where are AMCs in the development ecosystem now? There's the pneumococcal success, Operation Warp Speed, which is probably the most well-known to lay audiences, and you've got some private examples like Frontier, Stripe’s carbon removal AMC. Are AMCs just part of the firmament now? Are they fully validated? 

Some purists might ask whether Operation Warp Speed was really an advance market commitment or just procurement contracts with individual suppliers. It wasn't this firm agnostic mechanism that looked further into the future. There is some concern that we might have lost the momentum from the pneumococcal AMC. 

That’s why Michael Kremer, Rachel Glennerster, and I formed the Market Shaping Accelerator at the University of Chicago. We provide policy advice to governments, NGOs, and philanthropists who might be interested in running some kind of market shaping mechanism, whether it's an advance market commitment or others.

I've got a question from the audience that gets at an interesting tension: you want to pick targets for AMCs where you're fairly close to the technological breakthrough rather than things that are very far away. But presumably, in these capital-intensive domains, you are more likely to have big incumbents like Pfizer and GSK. 

How do you encourage free market competition and not just pick the big winners that are closest in the R&D space? 

When I tried to explain why pneumococcal disease was picked over malaria, I hope I didn't give the impression that that was the only thing that advance market commitments can be used for. They can be used for these fairly late-stage products, but also for pie-in-the-sky, far-flung innovations. It's a challenge to specify the technical product profile for something that doesn't exist, but that can be overcome and the social value may be enormous. Think about things like nuclear fusion where the benefits could be incalculable. It might be 50 years away, but it’s still worth considering how to structure those. 

We're thinking about this for carbon removal and climate. The Frontier AMC is buying offtakes from current producers year after year and meeting them at their price point. So one producer might remove carbon durably for $1000 per ton, another for $500, and another for maybe $350. Why not pitch an AMC that might be 10 years into the future at a $50 price point, but a much larger quantity? 

One of the things we’re thinking about is how much money to dedicate to near-term technologies to help them march down their cost curve versus setting up a much bigger prize further in the future to encourage the big players who can produce at scale. The AMC doesn't necessarily select for incumbents — it can cut both ways. But I think the question is exactly right to point out that the big benefits might be for long-term projects that need a lot of R&D. 

From a political perspective, those long-term projects seem trickier to justify to a democratic public today than during COVID. Whether you call that an AMC or a procurement of some other kind, the benefit was widely understood. Is it harder to defend billions of dollars for nuclear fusion that will arrive when your kid grows up?

Right now, you’re actually using push funding for these things, and who knows if they're going to pass? In a sense, this problem is almost mitigated by the AMC, because the payments might not come until the product is actually developed. There's a question about how to bureaucratically commit yourself to making payments in the far future, which is a greater challenge for certain governments. There are some facilities that can be designed to get around those problems. 

Is there a risk that firms delay investment in technology, let's say a carbon removal project, because they expect the government may establish an AMC for it? And in a sense, publicizing that you're interested in AMCs may delay R&D commitments from the private sector in the short term? 

I'm not so sure how much I'm worried about that. In a sense, it's just adding resources to the market, and that spurs investment and innovation. If the AMC is designed as we envision, it's firm-agnostic. It's not like they're waiting to sign up for something. There's not really any kind of waiting game that might be spurred, as far as I can see. 

Cancer and Alzheimer's have been floated as ideas for AMCs. Are they good, reasonable, or poor targets? 

We're typically targeting areas where the social need is great, but the commercial incentives are less so. Those are two cases where the social needs are enormous. Let's just take Alzheimer's — think about the loss of quality of life and reduced life expectancy. How much would we pay to avoid the scourge of that disease? Trillions of dollars per year? The net present value is almost incalculable, but the commercial incentives are pretty great too. If you come up with a blockbuster drug for Alzheimer's, it will pay for any investment costs. 

Plus, there's a ton of push funding already happening. I heard that because it’s one of the big priorities for the NIH, you should sprinkle the word “dementia” into your otherwise unrelated grant proposal. So it doesn't necessarily need any extra tweaks or market shaping. I would say the same is true for cancer. 

There's an interesting parallel here. Your colleague, Jonathan Levin, just replaced the last president of Stanford, Marc Tessier-Lavigne, who was heavily implicated in the amyloid plaque hypothesis for Alzheimer's that received a ton of push funding. My understanding is that it's now largely seen as a failure and a misallocation of resources.

[From The Stanford Daily: “Two of Tessier-Lavigne’s influential neurodevelopment papers published in Science and a third published in Cell were withdrawn after they were found to contain manipulated images.”

It seems both push and pull funding have their challenges for Alzheimer's. Is it just that hard problems are hard to find solutions for, or is there something else going on? 

Yeah, I've been talking a lot about pull funding, but that doesn’t mean that we advocate against push. There are good reasons for both. With Alzheimer's, it's incalculable how valuable it is. Were grants for these research angles misspent? Probably not. We can't just rely on commercial incentives. There are reasons for funding basic research. There’s a spillover of ideas: maybe an idea is very basic and doesn't directly lead to the product, but it provides an onramp to do the development. You're not going to necessarily get the returns, and that needs to be subsidized. 

For these huge problems, that’s worth doing. So we're not necessarily against push, but we're just saying it sometimes it can be complemented by pull. In the case of some of these endemic diseases that devastate high-income countries, there are pretty good commercial market incentives. Sure, you want some push to get the basic research done, but I think the pull is already there. You don't need a special program. 

So far you've talked about how pull mechanisms are more attractive to funders in a resource-rich environment. If you're a finance minister in a low- or middle-income country and you're interested in pull mechanisms, are there ways for you to apply these lessons, or are they reserved for major funders that can outlay on the order of billions of dollars for future commitments?

Some of the lowest-income countries need to focus on day-to-day needs, and it's hard for them to imagine funding the R&D for these far-away innovations. That said, there are things that they could do for pandemic preparedness. During the COVID pandemic, there was concern that rich countries were getting their vaccines first and that there was much delay for the poor countries because they didn't get in the procurement queue. 

One possibility for them to think about is reserving capacity to try to get their place in the queue when the next pandemic or regional epidemic hits. Cholera and dengue are circling the globe now, and there's actually limited capacity for the new vaccines for both of those. Vaccine capacity is insurance against future pandemics, or can be repurposed for current epidemics. 

Another question from the audience: wouldn't AMCs for future spare capacity only benefit incumbents who have existing spare capacity or the funds to build out that capacity without additional revenue? Would more market-oriented options like prizes for multiple winners or ARPA-E-style loan guarantees be more appropriate mechanisms for market innovation? 

Let's take durable carbon as an example. Suppose we have an auction in 10 years. Who would the players be? Are they the ones that Frontier is funding, Big Oil, or some other firm? It's hard to say who the incumbents are. If it takes a big, well-funded firm to finally push innovation ahead, sometimes we have to live with that. That's not necessarily a disaster. I’m not going to shed tears over the fact that Pfizer was a big company entering into COVID vaccines. If it happens to be some big oil firms that come up with the best low-cost technologies in the end, how many tears should we shed about that?

You've been very reasonable on the need for both push and pull funding, so let me push you a little bit. Broadly speaking, should we rebalance the portfolio more towards pull funding? And if so, could you give us a rough pie chart for how to think about that? 

Given that the balance is almost exclusively push, it seems like some boosterism is needed for pull. I am doing some work with Caleb Watney, who is co-CEO of IFP, Matt Clancy, who's involved in IFP and Open Philanthropy, and some people from my accelerator team on an innovation atlas. Funders and government agencies don't have to reinvent the wheel for every problem they face. This innovation atlas would be a guide to the types of problems that are best solved by pull or push. Among pull mechanisms, what are the right ones? Is it a prize, an advance market commitment? It's a complex problem, but we'll have a nice map to support innovation.

To get updated on the Innovation Atlas when it’s released, subscribe to IFP’s monthly newsletter.

Thank you to Rita Sokolova for her judicious edits to the written transcript.

In this podcast, we interview top appointees and civil servants about how they managed to achieve a particular policy goal.
Subscribe at to get interview transcripts in your inbox once a week.
Listen on
Substack App
Apple Podcasts
Pocket Casts
RSS Feed
Appears in episode
Santi Ruiz