Listen Get

Bombs vs. Biotechnology: The ROI

Author
Affiliation

Mike P. Sinn

International Campaign to End War and Disease

Keywords

ROI analysis, 1% treaty, GDP growth, disease queue, cost-benefit

Every number below is sourced from SIPRI, WHO, NIH, and the companies’ own SEC filings. The derivations are at GDP Trajectories182 and parameters-and-calculations. If a number is wrong, the linked source says so.

If a link below is broken, the counter is fiction; break the link and improve the book. If every link holds, then every day of delay gives disease about 139 thousand extra murder victims, and the counter is your to-do list.

Summary

Metric Status quo After 1% redirection
Diseases without effective treatment

6,650 diseases (95% CI: 5,700 diseases-8,232 diseases)

 Same queue, 12.3x (95% CI: 4.92x-50.8x) trial capacity
New first treatments per year

15 diseases/year (95% CI: 8 diseases/year-30 diseases/year)

185 diseases/year (95% CI: 63.8 diseases/year-816 diseases/year)
Years to clear the disease queue

443 years (95% CI: 255 years-841 years)

36 years (95% CI: 8.15 years-106 years)
Global GDP at year 15

$167 trillion

$238 trillion (95% CI: $204 trillion-$261 trillion)
Global GDP at year 20

$188 trillion

$322 trillion (95% CI: $245 trillion-$404 trillion)
Per-capita lifetime income gain Baseline +$518,879 (95% CI: $221,703-$860,930)
Healthy life gained per person Baseline +12 years (95% CI: 8 years-18 years)

The disease queue

6,650 diseases have no effective treatment. The world discovers about 15 first treatments per year. At that rate, clearing the queue takes 443 years. 99.7% (95% CI: 99%-100%) of the potential uses for the 9,500 known-safe compounds have never been tested.

Redirecting 1% of global military spending ($27.2 billion per year) to pragmatic clinical trials increases trial capacity by 12.3x (95% CI: 4.92x-50.8x), compressing the queue from 443 years to 36.

The cost: one of our 122 apocalypses worth of nuclear capacity. It takes about 100 warheads to cause nuclear winter. We have 12,241.

Why GDP grows 1.43x (95% CI: 1.22x-1.56x) at year 15

The treaty’s GDP effect comes through four compounding channels, each sourced independently:

1. Multiplier differential. Every dollar spent on the military returns approximately 0.6x (95% CI: 0.4x-0.9x) in GDP (CBO, Moody’s Analytics). Every dollar spent on healthcare R&D returns approximately 4.3x (95% CI: 3x-6x), plus a 2x (95% CI: 1.5x-2.5x) spillover into adjacent sectors (biotech, AI, manufacturing). Redirecting 1% moves spending from the low-return category to the high-return category.

2. Disease burden recovery (dominant channel). Disease drags GDP by 13% annually ($400 trillion (95% CI: $252 trillion-$544 trillion) per year, WHO). The redirected funding scales trial capacity 12.3x (95% CI: 4.92x-50.8x), progressively curing diseases and recovering that drag. This channel alone accounts for most of the GDP gap.

3. Peace dividend. War costs the global economy approximately $11.4 trillion per year in direct and indirect damages (IEP Global Peace Index). A proportional reduction from reduced military posture recovers part of that drag.

4. Cybercrime recovery. Cybercrime costs approximately $10.5 trillion per year (Cybersecurity Ventures). Reduced military tensions partially reduce state-sponsored cyber operations and their economic drag.

The treaty ratchets from 1% to 10% over 15-20 years. The compounded effect: GDP grows at an effective 5.28% (95% CI: 3.85%-6.48%) instead of the baseline 2.5%. At year 15, the treaty economy is 1.43x (95% CI: 1.22x-1.56x) the baseline ($238 trillion (95% CI: $204 trillion-$261 trillion) vs $167 trillion). At year 20, 1.71x (95% CI: 1.3x-2.14x) ($322 trillion (95% CI: $245 trillion-$404 trillion) vs $188 trillion). Full derivation: GDP Trajectories182.

The biotech margin argument

Biotechnology companies have net profit margins of 18.5% (95% CI: 15%-22%). Military contractors have net profit margins of 4.99% (95% CI: 4%-6%). That is 3.72x higher. A military contractor board that sells 1% of its bomb-making assets and buys biotechnology shares with the proceeds has given itself a raise on the reallocation alone, before the treaty’s GDP growth compounds the gain.

The economy the treaty produces is 1.43x (95% CI: 1.22x-1.56x) larger at year 15. Every diversified portfolio scales with it. A military contractor is worth far more operating in an economy that large than in the economy its current lobbying is working to keep small.

Historical precedent

After World War II, the United States cut military spending by 87.6% in two years. That reduction was followed by the greatest sustained rise in living standards in American history. Current US military spending is 30.6x the pre-war baseline. In 1939, before the United States won the war, military spending was 96.7% lower than it is now. A 1% reduction is manageable.

The spending ratio

You have about a 1 in 30 million chance of being killed by a terrorist, and approximately a 100% chance of being killed by a disease. We spend 604 (95% CI: 453-888) times more on the military than on clinical trials. That ratio is why the cures do not exist yet.

The ratio persists because military contractors spend approximately $198 million (95% CI: $190 million-$210 million) per year on lobbying. The lobbying is effective: global military spending is $2.72 trillion per year (SIPRI). Global government clinical trial spending is $4.5 billion (95% CI: $3 billion-$6 billion) per year (NIH). Nobody has ever ROI-analyzed the lobbying expenditure.

Why it has not happened

CEOs and board members in the military industrial complex are also humans. The average age is about 60. In about 21 years (95% CI: 19.6 years-22 years), on average, they will be dead, and their net worth will be zero, unless they are buried with their money, which their children are unlikely to do.

If they are trying to maximize their net worth, and their net worth is going to be zero when they are dead, they should be fighting for this 1% reallocation that could eradicate disease 12.3 times faster. Instead they block it, because this set of facts has never gotten into their brains. If it did, and they are rational agents interested in maximizing their net worth, they would reinvest their personal money into biotechnology and use their lobbying budget to redirect 1% of military spending to pragmatic clinical trials.

Model objections

Objection: More trials does not mean proportionally more cures (diminishing returns).

Doubling biomedical research dollars does not double cures. Diminishing returns on research are well documented (Bloom et al., “Are Ideas Getting Harder to Find?”). The 443 → 36 years compression assumes linear scaling, and the cure rate is sublinear in trial count.

Refutation:

You are basing this on a situation where trials are only done on profitable novel molecules. These trials are done on all potentially medicinal molecules, regardless of whether they are profitable, and the cost is about 44.1 times cheaper than traditional trials. Additionally, under this system, it is a decentralized FDA183,184 where all information is published, giving us additional public and shared information about the effectiveness of each intervention on each target. This allows us to better understand human biology and improves our ability to target and select the next intervention. You can aggregate all of this data, rank interventions, and always be trialing the most promising intervention available based on the entire universe of data. Therefore, the number of cures produced per trial should increase over time.

We are not running out of low-hanging fruit. 99.7% (95% CI: 99%-100%) of the potential uses of 9,500 known-safe but generally unpatentable interventions have never been tested. You can’t run out of low-hanging fruit if you haven’t even gone into the orchard.

Objection: GDP growth does not accrue evenly.

The gains may concentrate at the top. The reader you most need to persuade may hear “the average rises” as “I get screwed because I am not average.”

Refutation:

Fair, so run it on the median instead of the mean. The treaty’s targets are median disposable income and median healthy life expectancy, not GDP per capita. The disease-cost reduction (out-of-pocket care, lost workdays, caregiving) lands disproportionately on the median household, which is why the median rises even under historical patterns where the mean outruns it.

Objection: Compounding model uncertainty makes the joint claim weaker than the chain reads.

Each link is probabilistic: 1% → 12.3x trials, more trials → more cures, more cures → 1.43x (95% CI: 1.22x-1.56x) GDP at year 15, bigger GDP → the reader’s personal net worth rises with it. A sophisticated reader can accept the direction of every link and still rationally disagree about the magnitude. “I agree disease eradication is undervalued; I disagree that the multiplier is 1.4x rather than 1.05x.” That is not irrational.

Refutation:

Run the conservative case. Assume 1% reallocation → 2x trials (not 12.3x). Assume 2x trials → 1.5x cures. Assume the GDP gain comes in at the bottom of the model’s confidence interval rather than the center. You still get an enormous reduction in disease burden, an enormous personal benefit to every human alive, and a treaty that pays for itself many times over its 1% cost. The math is overwhelming even at conservative parameters.

The person who says “I disagree about the multiplier” has to specify a multiplier so low that the conclusion flips. There is no honest multiplier that low. The cost of being wrong about the multiplier is “we got fewer cures than hoped.” The cost of being wrong about not doing it is “we continued blowing up cities and prolonging disease for another century.” Those two costs are not the same size.

Compounding uncertainty cuts both ways, by the way. It is just as likely that the multiplier is higher than estimated as lower, because we are at 99.7% (95% CI: 99%-100%) untested potential uses for known-safe interventions. The point estimate is the median of the distribution, not the worst case.

Objection: Motivated reasoning beats spreadsheets in the empirical record.

The cognitive science is bad for the strategy. Forty years of “smoking causes cancer” warnings produced slow gradual behavior change, not immediate compliance, even though the math is personal and overwhelming. Cardiovascular non-compliance, climate-risk complacency, financial discount-rate failures: humans systematically underweight low-probability catastrophic personal outcomes even when their own life is at stake. The “rational agent who reads the math and updates” is an economist’s model, not a psychologist’s.

Refutation:

Smoking is the wrong reference class because it requires the smoker to change their own behavior, which is hard. The treaty requires no behavior change from the public. It requires a budget line item to be adjusted in 190 spreadsheets. The friction is institutional, not personal.

Even taking smoking on its own terms: US adult smoking declined from 42% in 1965 to 12% in 2023. That is a 71% reduction over 58 years through exactly this kind of multi-channel, multi-decade campaign. If the disease-eradication campaign achieves the smoking trajectory, the treaty passes. Slow and gradual is fine. Disease eradication does not need to happen next year; it needs to happen eventually, and 36 years is an acceptable timeline even with substantial campaign drag.

Also: the people who matter (CEOs, judges, senators) are an unusually rational subset of the population, and they are exposed to spreadsheets professionally. They are more responsive to math than the median smoker is, because their job is to read math. Motivated reasoning is strongest when the cost of changing your mind is high. For a senator, the cost of changing their mind on the treaty (once their constituents are wearing the shirts) is low and the cost of not changing it is career death.

The Funniest Joke in the Universe, Mathematically

The total funniness of this campaign, measured in laughs prevented from being lost to disease and aging, is approximately 3.51 quadrillion laughs (95% CI: 971 trillion laughs-9.71 quadrillion laughs). For comparison, the average joke produces 1 laugh. This is the funniest joke in human history by a margin of approximately 3.51 quadrillion to 1.

The chain:

  • The average human laughs 17 times per day.
  • That is 6,205 laughs per healthy life-year.
  • The treaty recovers approximately 565 billion DALYs (disability-adjusted life-years) by compressing the disease eradication queue from 443 years to 36 years.
  • 565 billion recovered healthy life-years × 6,205 laughs per year = 3.51 quadrillion laughs.

Every objection above is asking you to weigh some friction (diminishing returns, distributional effects, model uncertainty, motivated reasoning) against the funniness of this joke. The friction is measured in academic papers and confidence intervals. The joke is measured in quadrillions of laughs. The math does not require a calculator.