How the US can solve the global vaccine shortfall

We need to rapidly expand vaccine manufacturing around the world — are IP suspensions the best way to do it?

How the US can solve the global vaccine shortfall

The White House is currently considering whether to support a push to suspend drug companies’ patent rights to their Covid-19 vaccines. This is a delicate issue that requires policymakers to balance the importance of incentivizing medical R&D in the future against the need for a rapid vaccine rollout around the world — especially in developing nations. We need more manufacturing firepower, and the US can best unleash it by coordinating a global technology transfer through the purchase of the intellectual property and the creation of incentives for producers to share their know-how with the rest of the world.

The legal proceedings around this issue were triggered in October 2020, when India and South Africa circulated a World Trade Organization Trade-Related Aspects on IP Rights (TRIPS) petition calling for the intellectual property protections on Covid-19 vaccines to be suspended for the remainder of the pandemic. The movement has quickly gained momentum since.

More than 100 countries have signed on to the motion, some out of narrow concern with the pace of vaccine distribution in their country and others as a protest vote against a global IP regime they believe favors rich countries. But because WTO motions like this require unanimous approval from all members, the currently opposed group, which includes the US, EU, and the UK (not coincidentally the nations where the vaccine developers are located), is able to block the motion. In the last week or two, a group of House Democrats, with the support of Bernie Sanders and Speaker Nancy Pelosi, have started lobbying the Biden administration to switch course and support the motion.

The case for suspension:
The pro-suspension side is relatively straightforward. While wealthy countries have been able to receive quick access to the vaccine (indeed, near-miraculous progress compared to the normal vaccine creation timeline), deployment has been sluggish around the rest of the world and especially in developing nations. This is both a moral and a public health issue. It’s bad for people to die simply because they live in a country with fewer scientific resources. And the longer the virus hangs around, the higher the risk of a significant mutation that could restart the whole pandemic.

It’s difficult to project vaccine production timelines exactly, but a common estimate from experts seems to be that many low-income countries won’t be fully vaccinated before the end of 2022. Others are more pessimistic, predicting immunity as late as 2024. Either way, that’s still quite a while away, and plenty of time for new Covid variants to emerge. Already we’ve seen the B.1.351 variant from South Africa poke holes in our defenses as it appears the AstraZeneca vaccine is significantly less effective against this strain. The Pfizer and Moderna vaccines still appear to offer good protection, but even they may see a small drop in effectiveness.

We can always create new vaccines or new booster shots to help address new variants, but the distribution map is going to be similarly skewed towards rich countries. A cycle in which rich countries continue to get vaccines (and booster shots) first but neglect to accelerate production sufficiently to cover poorer nations in time to prevent mutations is both patently unfair and self-destructive. We are in a race against time, and the costs of accelerating global vaccine production are measured in the billions whereas the benefits of reaching global immunity a few months earlier are measured in the trillions.

The World Health Organization and pharmaceutical companies have responded to this situation through the creation of COVAX, a pool of pledged vaccines specifically to help vaccinate low-income nations. This is an important program and the recent Biden investment of 4 billion dollars to the fund should be applauded. But the scope is still lacking in the necessary ambition. The aim of the program is to supply vaccines to help participating counties vaccinate only 20% of their population.

Ultimately, we need to be producing more vaccines in less time. One way of achieving that is to make it easier for competing vaccine manufacturers to get in on the action. Specifically, if there is latent vaccine manufacturing capacity (more on this controversial point later), we could pull out all the stops by suspending the intellectual property protections. This should help speed the vaccination efforts around the world, and if the latent manufacturing capacity is located in the Global South, it could make the future waves of vaccine booster shots more equitably distributed given the apparent home-market bias of current producers.

This last point connects to a larger question around self-sufficiency and national sovereignty in vaccine and pharmaceutical production. Much of this conversation is informed by the context of the HIV/AIDS epidemic of the 1990s to early 2000s which ravaged countries across Africa while the US and other wealthy nations were able to mitigate the damage using antiretroviral treatments. A combination of strict IP enforcement, a lack of foreign aid, and inadequate manufacturing capacity lead to millions of deaths that potentially could have been prevented. In 2003, the US created PEPFAR, an extremely successful public health campaign that was able to make massive dents in the crisis by sharing medicine and technical support with partner countries. To counter the concern that IP issues would hamstring future public health responses, the WTO signed the Doha Declaration which created rules around compulsory licensing of medical technology in the event of a public health emergency (but which advocates say are insufficient for this present situation). Given this history, developing nations are understandably skeptical of the idea that they should wait around for Western pharmaceutical companies to produce enough vaccines for them and would like to develop their own manufacturing capacity.

Finally, advocates for suspension typically point out that many of the pharmaceutical companies received R&D funding from the federal government and/or large market commitments through Operation Warp Speed, which guaranteed them some baseline level of confidence and profitability. Furthermore, the mRNA vaccines themselves are partially the result of decades of public investment in science, and indeed the National Institutes of Health may own some parts of the relevant IP in the case of Moderna. If you are inclined to believe that pharma R&D is mostly free-riding on the hard work of basic science, then this is a pretty clear-cut case for suspending the IP.

In short, we’re not safe from Covid until we are all safe. So let’s tear down any red tape or barriers that may be slowing the production and deployment of vaccines around the world.

The case against suspension:
On the anti-suspension side, there is the argument that these new vaccines (especially the new mRNA vaccines) required billions of dollars and decades of uncertain private investment to reach the technical breakthroughs that have given us a chance to end this pandemic. Yes, basic science work funded by the government helped a lot and established a strong foundation for much of this work. But enterprising individuals with a profit motive played just as large a role. Katalin Karikó, the Hungarian scientist who helped pioneer mRNA vaccines spent most of the 1990s receiving rejection letters for government grants and ultimately turned to the private sector where she co-founded her own company in 2006.

But even if all the basic research were federally funded, the argument for suspension would still create incentive issues. Operationalization and commercialization of scientific breakthroughs are essential and still cost a lot of money. There are a whole series of difficult engineering, logistics, and optimization problems that have to be solved when taking a complex biological product like a vaccine from research to reality. They do not pop out into the world fully formed from a peer-reviewed publication in Science. And it’s no accident that the countries with the most well-developed biotechnology and pharmaceutical clusters are the ones that produced these wonders.

Where most public health interventions failed miserably, the pharmaceutical companies worked around the clock to develop, test, and roll out a whole new genre of vaccine in record-breaking time and at high private cost. To expropriate the intellectual property that makes the whole investment pipeline worth it in the first place is to place future R&D investments under a great shadow. This is, unfortunately, unlikely to be the last global pandemic we face. And we got lucky that Covid-19 has a relatively low fatality rate and that we were able to so easily target its spike protein. To ensure that our biotechnology clusters are investing in R&D for new vaccine and therapeutic techniques for the future, we have to align incentives and make it profitable for them to throw billions of dollars at the problem years before it may ever appear.

Furthermore, all the focus on intellectual property conceals the point that the formalized information that can be written in a patent application isn’t that useful in isolation. Moderna actually pledged in October that they would not be enforcing the IP rights related to their Covid vaccine. And the underlying spike-encoding sequence was recently published online by Stanford researchers. However, no new Moderna knock-offs have sprung up since their October announcement. Why? Because the company has refrained from sharing any details about the manufacturing or design process, which indicates the underlying technical expertise and production process knowledge are just as important.

As Rachel Silverman from the Center for Global Development notes:

Observing their contents is insufficient to allow for imitation. Instead, to produce the vaccine, manufacturers need access to the developer’s “soft” IP — the proprietary recipe, cell lines, manufacturing processes and so forth. While some of this information is confidentially submitted to regulators and might theoretically be released in an extraordinary situation (though not without legal challenge), manufacturers are at an enormous disadvantage without the originator’s cooperation to help them set up their process and kick-start production. Even with the nonconsensual release of the soft IP held by the regulator, the process of trial and error would cause long delays in a best-case scenario. Most likely, the effort would end in expensive failure.

Typically this kind of soft IP is transmitted to the new company in a technology transfer process that happens during a licensing deal. Process experts from the licensing firm will sometimes travel to the facilities of the licensee and oversee operations for the first several batches to help convey the sort of tacit knowledge that is difficult to transmit unless you have physically done the operation yourself. In exchange, the licensing company will get a cut of the revenue from each dose, typically 5-10%. However, even when these licensing deals are in place, it’s easy for the new firm to make critical mistakes because the margins for error are razor-thin.

All this means that if the TRIPS proposal were passed in isolation, it would likely lead to little appreciable change in vaccine production unless it was paired with significant incentives for the manufacturers to play ball on the more formalized technology transfer process.

The open question: is there latent capacity?
Does this mean we are already at the frontier of vaccine production and there’s nothing more to be done? I doubt it.

In some sense, the easiest thing to do would be to plow many billions more into programs like COVAX, which reserves vaccine doses for developing nations. That would be good, but it may just extend the length of vaccine production rather than increase the rate of production. Maybe we could specifically help pay for existing vaccine manufacturers to scale up their capacity even more? This too, would be worthwhile. After all, it’s difficult to imagine having too much vaccine capacity (from a social welfare perspective). But even here, building new factories and machines can take months or years, and it would be nice to have something that works more quickly. Secondarily, this solution will only continue to concentrate vaccine production in existing firms and in rich nations, which undermines the self-sufficiency concern that helped fuel this debate.

The main advantage that some version of IP suspension offers is activating manufacturing capacity that already exists but isn’t currently being deployed. But how much latent capacity is there actually?

Especially for the mRNA vaccines of Pfizer and Moderna, it appears that the production bottlenecks are quite severe. A great piece by Derek Lowe outlines how specialized the supply chains and manufacturing processes for mRNA vaccines are, especially in their use of lipid nanoparticles. Lowe argues that only a handful of firms globally could have jumped in and started making mRNA vaccines immediately, and it seems likely that all of them already have.

However, this may be too fatalistic about the supply elasticity over a slightly longer time horizon. Production bottlenecks can ease over time with enough investment and learning by doing. Due to the massive social and economic costs we face during each additional day of this pandemic, even a slight increase in production can justify a significant price tag.

It seems more plausible, however, [I have no special expertise in vaccine manufacturing, so take this for what you will] that there is latent capacity using the more traditional adenoviral vector vaccines like Johnson & Johnson and AstraZeneca. Because adenoviral vector vaccines have the advantage of having simply existed for longer, more vaccine manufacturers around the world have the capacity to produce them than can produce mRNA vaccines right now. The related supply chains are more mature as well, which means a sudden surge in demand can be better accommodated. We saw this play out with the Serum Institute of India, the world’s largest vaccine manufacturer, which opted to take a voluntary licensing deal to manufacture the AstraZeneca vaccine at scale.

There may also exist capacity that is not exactly latent, but is not being used very efficiently. If an opportunity presented itself, we could see some vaccine manufacturers who are currently pursuing vaccine trials that will otherwise be late to the game drop their research trials and focus on producing known (and already approved) vaccines. Sanofi and GlaxoSmithKline, for example, began a phase 2 clinical trial in February that (if everything goes right) will finally be ready in Q4 of this year. There’s no particular reason to think this new vaccine will be better than the existing offerings. It’s more likely that they are pursuing this line because they know the vaccine rollout around the world will be slow and they want to have proprietary IP to offer. While both Sanofi and GSK have signed deals to help manufacture other vaccines in the meantime, these are quite limited and the companies are likely saving capacity so they can rapidly retool if their own trials are promising. But it would certainly be more socially valuable if we could encourage them to drop these trials and fully ramp up the production of proven vaccines right now. There are also companies like Merck that appear to be mostly sitting out the Covid vaccine rush altogether after their own efforts failed.

The case for buyouts:
The economist Michael Kremer wrote a paper in 1997 formalizing the idea of using patent buyouts as a way of maintaining strong incentives for innovation while still getting crucial information into the public domain ASAP. Essentially, a government could offer to pay the present value of the expected future revenue stream that would result from the temporary monopoly that a patent grants. While the patent-owning company or individual should be indifferent to the outcome, the general public could receive more value from having unabridged access to the information and the ability to modify it without permission before the patent expired. In these cases, a patent buyout can clearly improve outcomes for everyone, and Kremer notes that pharmaceuticals may be a particularly appropriate case.

Patent buyouts present at least a theoretical solution to the problem of maximizing the number of players that can legally produce vaccines while maintaining strong incentives for innovation in the future. Of course, in this situation it’s only partly about the intellectual property and partly about the manufacturing know-how that has to be transferred, which means we need a broader conception here — a full stack “technology buyout” which includes both the IP and the promise to transfer process knowledge.

Essentially, the US government (or conceivably a set of governments, but that would take longer to negotiate) could offer a lump sum payment to the accepting firm or firms to write down as much about the scientific and production process as can be made explicit and then make it publicly available. There could then be additional payments made for the sharing of tacit knowledge and aid in setting up the manufacturing operations either on an individual factory level, or on a per vaccine dose administered basis. The advantage of additional payments of a per vaccine dose administered basis is that it properly aligns incentives for the firm(s) sharing technology to maximize their impact by transferring to partners that can actually get shots into arms as quickly as possible and to make sure they do a good job.

To make this concrete and put some back-of-the-envelope numbers on this, I would suggest the initial lump sum payment to make the IP public would be in the range of $10-20 billion, and the additional per dose administered prize would be in the realm of $0.50 - $2. Assuming this program was able to administer vaccines for an additional 4 billion people (8 billion doses) across the developing world, we are talking in the range of $36 billion dollars.

And we should in some sense actively try to overpay. In unique situations like this, we should err on the side of overcompensating and risking some economic rents rather than inadvertently under compensating and hurting the long-term incentives for innovation. The most important thing here is that we do not kill the goose that lays the golden egg. In any event, we should be willing to pay an order of magnitude more than $36 billion to definitively end Covid, so this program should be a bargain under a wide range of potential cost assumptions. One estimate from a group of economists and public health officials ballparks the global cost of the pandemic at around $1 trillion per month.

Some readers may note that “payments for each vaccine dose administered” sounds very similar to the voluntary licensing agreements that manufacturers around the world have already signed. And they are correct, it’s a closely related mechanism. Indeed, commentators like Rachel Silverman (quoted above) have actually suggested that the best way forward may be having the federal government use its political leverage to lean on US pharmaceutical companies to accept more licensing deals with manufacturers in developing countries. Which raises the question, why bother doing the additional lump sum buyout to make the information public? Shouldn’t we concentrate all our efforts on encouraging licensing?

I would differentiate the technology buyout I’m suggesting from voluntary licensing on a few dimensions:

  • First, buying out the IP may be a determining factor for encouraging large, mature manufacturers like GSK and Sanofi to abandon their own duplicative vaccine trials and go full steam towards producing existing vaccines as it removes any competitive disadvantage in paying the licensing fee.
  • Second, because these voluntary licensing deals are typically assigned as a percentage of the dose cost, it creates an incentive for licensing firms to prioritize deals that will charge a higher price to the buying local governments where this may not be optimal. If the US government is instead paying for each dose administered in these developing nations, then the transferring firm(s) should be neutral with regard to the sale price.
  • Third, this structure allows the federal government to selectively overpay on a per dose basis if the US firm is genuinely helping a developing nation jumpstart new manufacturing capacity as opposed to helping an existing manufacturer retool.
  • Fourth, individual countries and foreign manufacturers know better their own capacity than US government officials or even US firms do, so opening up the IP could help identify latent manufacturing opportunities more effectively than a top down approach.
  • Fifth, opening up the IP at least gives every nation the ability to try and make their own vaccine if they so wish, which helps address the self-sufficiency concerns.

A buyout also presents several key advantages when compared to IP suspension as well, even putting the incentive issues aside.

  • First, speed. Even if the US were to reverse course and support the WTO proposal, it would take quite a bit of time to negotiate and wrangle all the other countries to the table. Remember, the proposal has to be supported unanimously, so there is no guarantee that the EU, UK, Australia, and other opposed nations will reverse course just because the US does. A buyout, in contrast, can be done unilaterally by the US (at least for the US based vaccine firms).
  • Second, in the unfortunate event that a new variant of Covid requires a booster shot, a buyout ensures the incentive to quickly create a solution (so that the new booster shot can also get bought out). Under the WTO petition, the IP suspension would remain in play for the duration of the crisis, which would reduce the urgency and resources that pharma companies are willing to throw at the problem, given fewer opportunities to recoup costs.

Building for the long term
Finally, there is the issue of whether a technology buyout would be better suited for the mRNA vaccines of Modena and Pfizer or for an adenoviral vector vaccine like Johnson & Johnson. To answer that we have to ask a more fundamental question: what is it we are trying to achieve here? There is likely more short term capacity to scale adenoviral vector vaccines, so if we are narrowly trying to get the world vaccinated against Covid-19 as quickly as humanly possible, then a technology buyout for the J&J vaccine probably makes the most sense.

But if we have a larger vision of using this crisis as an opportunity to bootstrap new, flexible vaccine manufacturing capacity around the world for the future, then mRNA vaccines have a host of advantages that will make them the better long-term pick. Of course, we can try to do both to take advantage of the distinct advantages of each vaccine type, but it helps to have a longer term anchor goal to build towards.

A few points in favor of the future oriented approach. First, hoping the adenoviral vector vaccines hold up long enough against possible variants is a significant risk. As alluded to earlier, the mRNA vaccines appear to be more resilient to the recent Covid variants and are significantly easier to modify in the event a new variant arrives that our current vaccines can’t handle.

Second, there have been a series of exciting news developments in the last few months indicating that mRNA vaccines could unlock a much broader wave of medical improvements, including possible vaccines for multiple sclerosis, some forms of cancer, malaria, and HIV. All of which means mRNA is likely to be a more general purpose vaccine technology and makes it almost impossible to imagine having overbuilt capacity at this point.

Third, separate from the supply chain issues, new mRNA facilities are actually much cheaper to build and operate than traditional vaccine factories. From an article in the Journal of Advanced Manufacturing and Processing:

Based on our techno‐economic assessment, the RNA vaccine production process can be two to three orders of magnitude smaller than conventional vaccine production processes in terms of facility scale, and can be constructed in less than half the time with 1/20 to 1/35 of the upfront capital investment… It therefore presents a strong advantage of requiring small‐scale, high‐capacity facilities, which can be constructed more rapidly and could make wide use of single‐use disposable equipment. Due to its small scale, the RNA vaccine drug substance production process could be placed in a small part of an existing conventional vaccine facility, for example in a room, and still produce more doses worth of drug substance than the entire original conventional vaccine production facility.

Fourth, we should view this as an opportunity to build good will around the developing world. The total US foreign aid budget was around $40 billion in 2019, right in the range of what we are proposing here. I’m inclined to believe sharing new, highly effective technology around the world during this unique crisis would generate a significantly higher return diplomatically than the projects we usually get with this scale of funds. Already we’ve seen China and Russia attempt to leverage their vaccine exports for diplomatic purposes. Indeed, another lens you could use for this would be as a kind of liberal counter-weight to the Belt and Road initiative that has helped China make inroads across Africa. Instead of physical infrastructure investment, the US would be helping them with technological infrastructure investment.

Finally, it is in the long-term interest of all of humanity to have a developed and coordinated ability to respond to new diseases around the world. Viruses don’t respect national borders and the risk of a global pandemic has only increased over time as our world has become more interconnected. This capacity inherently needs to be distributed around the world for it to be maximally effective, so we may as well start building it out now. On the high-end of effectiveness, this capacity to rapidly create and distribute new mRNA vaccines could help us eliminate much of the long term risk of bioweapons and natural pandemics. That is the true goal worth aspiring toward here.

The world needs more vaccines and we need them quickly. It seems unlikely that suspending the IP rights will do much to accelerate production in isolation and it could significantly diminish the incentives for investing in the future. On the other hand, a full technology buyout — which includes the IP rights and incentives to work with foreign manufacturers — could simultaneously activate any latent capacity that may exist while also giving the developing world an opportunity to bootstrap their own vaccine manufacturing capabilities with the help of American know-how.

The US should be aiming not only to vaccinate the entire world, but to teach the world how to make vaccines. As the famous saying goes: give the world vaccines and you stop one pandemic, teach the world how to manufacture mRNA vaccines and you stop pandemics forever.