The Biosecurity Reserve

The problem

The United States recorded roughly 1.2 million deaths from COVID-19. Work by Cutler and Summers put the total economic cost to the country at around $16 trillion, counting lost output, lost lives and lost health.

Against that, consider what was in place beforehand. The Strategic National Stockpile operated on a few hundred million dollars a year. Its N95 respirator inventory had been substantially drawn down during the 2009 H1N1 outbreak and was never fully replenished, because replenishment required an appropriation and no crisis was underway to justify one. The result in early 2020 is well documented and does not need retelling.

Then consider what worked. Operation Warp Speed spent roughly $18 billion and produced authorised vaccines in about 11 months against a historical baseline measured in years. Whatever one's view of anything else in that period, the return on that particular $18 billion was among the highest of any public expenditure in American history.

The lesson is not complicated: preparedness is cheap and response is ruinous. The problem is that preparedness has no constituency between crises. The pattern repeats with unusual reliability. A crisis arrives, emergency money flows, capability is built, the crisis recedes, the money is reprogrammed to something with a visible beneficiary, and the capability decays. It happened after 2001, after 2009, after 2014, and it is happening now.

There is a second and quieter failure running alongside it, and it is the clearest market failure in modern medicine.

Antimicrobial-resistant infections kill roughly 35,000 Americans a year and the number is rising. New antibiotics are desperately needed. But a genuinely novel antibiotic must be used as little as possible to preserve its effectiveness, which means its commercial value is close to zero. This is not a hypothetical. Achaogen brought a new antibiotic through FDA approval and filed for bankruptcy the following year. Several others have followed. The pipeline has largely emptied, and the reason is that we have asked a market to reward a product whose social value depends on nobody buying it.

Meanwhile H5N1 continues to circulate in dairy herds and poultry, which is exactly the kind of thing the watchtower is supposed to be watching.

The proposal

A permanently funded reserve, insulated from the appropriations cycle: standing surveillance, warm manufacturing capacity, a rotating stockpile, and subscription payments for antibiotics nobody can profitably sell.

How it would work

1. Surveillance that never turns off. Permanent national wastewater monitoring, which is astonishingly cheap and detects circulation days to weeks before clinical presentation, plus genomic sequencing of clinical samples at a defined sampling rate, plus systematic monitoring at the animal-human interface where nearly every novel pathogen originates. Data published openly and continuously.

1. Warm manufacturing capacity. Contracted, maintained, periodically exercised fill-finish and platform manufacturing capacity, paid for as standing capacity rather than per dose. This is the same instrument as № 047 applied to a different threat, and for the same reason: a line that has not run in five years is not a line.

1. A 100-day objective. An explicit national target of a candidate vaccine in clinical trials within 100 days of pathogen sequence, with prototype work maintained across viral families rather than waiting for the specific threat. The international 100 Days Mission has established this as a real engineering goal rather than an aspiration.

1. A stockpile that rotates. Inventory managed with vendor-managed rotation, so that stock is continuously used and replaced through normal commercial channels instead of sitting until it expires. The 2020 failure was as much an inventory management failure as a funding one.

1. Subscription payments for antibiotics. A delinked payment model: the government contracts to pay a fixed annual sum for access to a qualifying novel antibiotic, regardless of volume sold. The developer is paid for the drug existing and being available, and has no incentive to push volume, which is precisely the incentive we want removed. The UK has run this model since 2022. The PASTEUR Act has proposed it here repeatedly and has not passed.

1. Multi-year, ring-fenced funding. This is the load-bearing item. Authorised on a rolling 10-year basis, with a statutory floor, reported separately, and not available for reprogramming. Everything above has been proposed before in some form. What has never been done is protecting the money from the ordinary budget process, which is the actual mechanism of every past failure.

1. Oversight of dual-use research. A single, empowered, transparent review body for research that could plausibly enhance the transmissibility or lethality of a potential pandemic pathogen, with authority across federally funded and privately funded work alike. Reasonable people disagree sharply about where the line sits; almost nobody defends the status quo in which the line is drawn inconsistently and reviewed opaquely.

The numbers

Cost. Standing surveillance is on the order of $1 to $2 billion a year. Warm manufacturing capacity and the 100-day capability, $3 to $4 billion. Stockpile at genuine scale with rotation, $2 to $3 billion. Antibiotic subscriptions, at roughly $750 million to $1 billion per qualifying drug for a small number of drugs, $2 to $3 billion.

Total: roughly $10 billion a year.

Benefit. Set $10 billion a year against $16 trillion.

If this programme reduced the expected cost of the next pandemic by even 1 percent, it would return $160 billion against a decade of spending costing $100 billion. If it reduced it by 10 percent, which the Warp Speed experience suggests is not fanciful for the vaccine component alone, the return is an order of magnitude larger than that.

The right frame is not cost-benefit on a point estimate but insurance against a fat tail. Epidemiological estimates of the annual probability of a pandemic of COVID scale or worse cluster somewhere between 2 and 4 percent a year, which over a lifetime is not a remote risk, it is a likely one.

$10 billion a year is about 0.15 percent of federal spending. It is a third of what the country spends annually on pet food.

The honest objections

"The country just went through this and the political appetite is not merely absent, it is negative." The strongest objection and it is about trust rather than economics. Public confidence in health institutions fell substantially through the pandemic and has not recovered, and a proposal to hand those institutions a permanent, ring-fenced $10 billion will be received by many people as a reward for failure. We think the response has to be structural rather than rhetorical: this proposal deliberately funds capability rather than authority. It buys sequencers, factories, stockpiles and drugs. It confers no new power to close anything, mandate anything, or compel anyone, and it should be drafted so that this is unmistakable on the face of the statute. Anyone selling pandemic preparedness as a bundle with emergency powers is making both harder to get.

"Ring-fenced funding is how you get an unaccountable programme that nobody can cut." A real trade-off and we are choosing one side of it openly. The reason for item 6 is that the ordinary process has failed at this specific task four times in twenty-five years, with a consistent and well-documented mechanism. The mitigation is a hard sunset, a decadal reauthorisation requiring an affirmative vote, and published performance metrics, days-to-detection, days-to-candidate-vaccine, stockpile fill rate, that make decay visible while it is happening rather than after.

"Biosecurity research is itself a source of risk." A serious argument, held sincerely across the spectrum, and item 7 is a partial answer rather than a complete one. This proposal is deliberately weighted toward detection, manufacturing and stockpiling, which carry no such risk, rather than toward expanded work with dangerous pathogens. On the harder question of where to draw the line on enhancement research, we do not think this catalog can settle a scientific dispute by assertion, and we would rather say that than pretend to.

"$10 billion a year for a threat that may not materialise for decades." That is what insurance is, and the same objection applies verbatim to № 047, where we made the same argument about a different tail. The difference is that here the historical base rate is measurable, the last event is recent and documented, and the return on the one component that was actually funded properly, vaccine development, is among the best-evidenced in the whole catalog.

"Antibiotic subscriptions pay companies for doing nothing." They pay for a drug existing, tested, manufactured and available, which is not nothing, and they explicitly decline to pay for volume, which is the entire point. The alternative on current evidence is a world in which the drug is not developed, several firms that did develop one went bankrupt, and resistant infections continue to kill tens of thousands of Americans a year. The UK has been running this model since 2022 and it is the only approach that has produced any movement at all.

Sources

• US COVID-19 mortality; CDC (cdc.gov)
• Cutler and Summers, "The COVID-19 Pandemic and the $16 Trillion Virus," JAMA (2020)
• Operation Warp Speed funding and development timeline; GAO reviews (gao.gov)
• Strategic National Stockpile funding history and N95 inventory depletion following 2009 H1N1; ASPR and GAO reporting
• Antimicrobial resistance mortality in the United States, approximately 35,000 deaths annually; CDC Antibiotic Resistance Threats report
• Achaogen bankruptcy (2019) following FDA approval of plazomicin, as evidence of antibiotic market failure
• UK NHS/NICE antimicrobial subscription model, operating since 2022; PASTEUR Act as the proposed US analogue
• CEPI 100 Days Mission (cepi.net)
• CDC National Wastewater Surveillance System, established 2020
• Estimated annual probability of pandemics of COVID-19 scale; Marani et al., PNAS (2021)
• Proposals № 004 (CitizenCare); № 043 (The Formulary); № 047 (The Arsenal Account)