Measuring the Black Holes
Uncounted externalities of industry sum to about 30% of the entire global economy - an estimate it seems nobody has provided before.
There's a literal black hole at the center of our galaxy. It, Sagittarius A*, has a mass of about 4.3 million times that of our sun. It's big: some 60 million kilometers across: about 4,700 Earth diameters, or 43 solar diameters. The event horizon itself is smaller, maybe 25 million kilometers across..
Despite its vast size and mass, we seem safe. It's far away, and not drawing in much mass (a few percent of earth's modest mass per year, which seems remarkably modest). We don't have to do anything about it and we can't see inside it..
There's also a black hole in our economy. It lacks mass and physical size, but we are not safe. It's destroying value to the tune of tens of trillions of dollars per year: perhaps 30% of the global economy. What is this thing? It's the uncounted total of business externalities and government-caused harms, and consists of things like destroying the natural world, and causing sickness.
While real black holes are quite difficult to study and quantify, we know Sagittarius A*'s mass to a few percent. We do not know the net of the harms we inflict on ourselves even to within a factor of three. That is an odd state of affairs, because this one we live inside, can see inside, in detail, and we can study it. And we can fix it — we must fix it, if we're to have enduring prosperity.
One specific thing seems peculiar. There are a nice range of estimates of the size of Sagittarius A* from groups of astrophysicists around the world. There are, as far as we can see, no estimates of the size of this vast hole in the economy. Are we the first to describe this? It would seem so.
When industries behave like black holes
An accounting boundary acts like an event horizon. It's the edge beyond which things cease being reported: not destroyed, but no longer on anybody's books. A firm's balance sheet has that shape. Inside it, each cost has a line and an owner. Outside it, the cost still exists, it just stops being the firm's problem, and so stops being counted.
Harms can go from unremarkable to catastrophe, sometimes in very short periods. Sagittarius A* is quiet now: a few percent of an Earth a year is tiny in a galaxy of some 200 billion stars, many with planets of their own. It wasn't always so. The Fermi bubbles, two lobes of gamma ray-emitting gas 25,000 light years above and below the galactic plane, may be the exhaust of a Sagittarius A* outburst a few million years ago. Long and tolerable accumulation, then a fast catastrophe. This is also a fair description of an aquifer, a fishery, the permafrost, or the stability of a minority culture. They’re stable and then there’s a catastrophic failure.
Tidal forces are uneven, falling hardest on whatever is closest and weakest. Particulate matter, like soot, or poisoned watersheds, or factories built where land is cheap lands on those closest, who have the least power to object. Mercury from coal-burning power plants. Gradient is a central feature of gravity as it is for pollution, and it is always the small nearby thing that gets damaged.
Both things grow by eating. A black hole's cross-section rises with its mass, so the bigger it gets the more readily it gets bigger. Tolerating and normalizing externalities does the same. Once it is normal and cheap to put a cost outside the boundary, more of the economy reorganizes itself to do that, and the firms that try to do good are outcompeted by the ones that don't. The incentive structures push more and more into the black hole.
Externalities are not forces of nature
Let’s squeeze the last drops out of the black hole analogy to show its deficiencies and some challenges.
Sagittarius A* is nobody's fault. There was no meeting at which it was approved (but that might make for a cool scene in a sci-fi movie). In contrast, each item and action that contributes to the externality crisis has human causes — an old industry whose practices were grandfathered in, a siting decision, a regulation written to protect some sector, a banker who chose to look the other way. But there is good news in this: What humans cause, humans can comprehend and resolve.
Nobody profits from hypermassive black holes. Our economic black hole persists because it pays, and the institutions best to benefit work the hardest to sustain it. It isn't going to be dissolved by explanations or a never-ending series of sad stories. It isn’t a business model on its own, but is a large fraction of all business models, and has the defenders you'd expect a thirty-trillion-dollar position would have.
A real black hole conserves. Mass and information cross the horizon but are not destroyed. The mass is inaccessible, but the information may leak back out over time. Ours destroys. A fishery collapsed, a species extinct, a watershed exhausted, a life lost or one spent with the brain damage that lead imposed. there is no path back from those. In this way, our economic black hole is worse than our astronomical one.
Counting, in public
Thirty trillion or so, or perhaps 30% of the global economy is a vast, vast sum, particularly when it’s not uniformly counted. Here are some examples of the pile it comes from. Each line is some institution’s serious attempt to price one category of harm.
Some recent estimates of industrial externalities
| Category | Estimated annual total of harms | Who estimated | Links |
|---|---|---|---|
| Hidden costs of agrifood systems | $12.7 trillion (2020, PPP); ~10% of world GDP in PPP terms | FAO, 2023: The state of food and agriculture | FAO report |
| Fossil fuels, explicit + implicit subsidies | ~$7.4 trillion (2024): $0.73T explicit + $6.7T implicit (~6.4% of GDP) |
IMF: Black et al. (2025), Underpriced and Overused: Fossil Fuel Subsidies Data 2025 Update, WP/25/270 The earlier 2023 Update gave $7.0T for 2022. About 39% of the total is local air pollution, which overlaps the next row. Calx estimates $5 Tpa from coal only, so this category will likely exceed $10 Tpa when completed |
IMF paper |
| Air pollution, health and lost output | $8.1 trillion (2019, PPP); 6.1% of global GDP | World Bank (2022), The Global Health Cost of PM2.5 Air Pollution: A Case for Action Beyond 2021 | World Bank report |
| Apparel and related | $2 trillion | Calx estimates, 2025 and 2026 | |
| Lead exposure | $6.0 trillion (2019; range $2.6–9.0T); 6.9% of global GDP mostly cardiovascular deaths (77%) and loss of neurological function (23%) |
Global health burden and cost of lead exposure in children and adults: a health impact and economic modelling analysis Larsen & Sánchez-Triana (2023), Lancet Planetary Health 7(10): e831–e840 |
Lancet paper |
| Land degradation and soil loss | $6.3–10.6 trillion (ecosystem-service value lost) | ELD Initiative (2015), The Value of Land | ELD report (PDF) |
| Biodiversity and ecosystem service loss | From $4.3 trillion / year to $20 trillion / year loss from land-use change, |
Costanza et al. (2014), Global Environmental Change 26: 152–158. Overlaps the land row above. |
Costanza et al. |
| Tobacco use | $1.4 trillion ($1.8T PPP); 1.8% of global GDP | Goodchild, Nargis & Tursan d'Espaignet (WHO) (2018). Tobacco Control 27(1): 58–64 | WHO authors’ paper |
| Plastics, health costs (including microplastics and plastic chemicals) | > $1.5 trillion |
Landrigan et al. (2025), The Lancet Countdown on health and plastics. Lancet 406: 1044–1062. Covers plastics across the life cycle, not microplastics alone. Calx 2026 projects >$3 Tpa by 2030 (internal) |
Lancet paper |
Some notes on the table
Global GDP is somewhere around $110 trillion, which is the denominator for the percentages. Some data are PPP denominated , but not all.
You cannot just add the estimates in each row, because — as is shown in some rows — some domains overlap, and there are other inconsistencies. Other overlaps come from some studies looking only at end use consequences, while others focus on elements of a supply chain. Even so, from just what is known about industry externalities, in informal counts, we know that it is hard to land anywhere under $20 trillion and fairly easy to land above $35 trillion, after overlaps are squeezed out. Thirty trillion and thirty percent are solid and conservative estimates.
The scale and scope of the external harms illustrate a key thesis of our work: the data on the harms do exist, and they exist at vast scale, in writings everywhere, because the harms to the environment, to human health, to the very survival and habitability of the planet, are real. People research them, they write about them, and their writings are scrutinized in public. The data do exist, but reading every paper about every harm in every sector of human activity, in every geography, all the time was a fanciful project. Until very recently. And now, it’s feasible. The scale that means the data are available also means that finding the reliable data is valuable. It’s feasible and it’s valuable. Therefore it becomes inevitable.
Dealing with the errors
The table double-counts some harms. Most of the fossil fuel figure is climate change and air pollution, which Calx separates into separate taxonomic classes; coal’s harms include vast human health costs (lead poisoning of infants!) and damage to the natural environment (coal spoil tips, bird deaths). The food figure contains some land degradation and obesity. Lead and air pollution both claim large slices of the same lost cognition and the same early deaths.
The valuation methods also do not fully align. Cost-of-illness / burden of disease accounting gives you one number; willingness-to-pay gives you a higher one; a statistical life valued at a country's own income level gives you a figure a hundred times smaller in Bangladesh than in Norway, which is a moral position hidden inside a methodological choice. And valuation defies reduction to simple rules for other reasons, some noted briefly below.
Discount rates do something similar across time. Choose a low rate (Frank Ramsey a century ago argued that future social benefits and harms should have a zero discount rate) and future harm dominates the ledger. Alternatively, pretend it’s a financial project only, and pick a bank-style rate, perhaps 4.5%, like WACC (weighted average cost of capital) and most of its impact vanishes. There is neither an empirical procedure, nor an international finance standard, that settles which to pick.
So the claim is not a specific number like "30.0%". Hyper-precision is unobtainable and cannot be our goal. The total today is between roughly 20% and 50% of global economic product; no serious estimate puts it near zero. The scope of the uncertainty itself is a scandal. We’ll narrow it down in subsequent work.
To deal with these, and other issues, Calx’s internal work relies on a multi-part effort at making the data accurate, and systematic, and open:
Creating an internal taxonomy that stabilizes harm data to a non-financial measure: megatonnes of CO2 or equivalent; DALYs and QALYs (disability-adjusted or quality-of-life-adjusted life years saved or lost).
Specifically, the taxonomy must support the key principle of being Mutually Exclusive and Collectively Exhaustive: MECE, pronounced mee-see. This term, developed by Barbara Minto in the 1960s and now used by business analysts everywhere, means that each item that needs to be counted shows up once (mutually exclusive meaning it’s not in two places) and that the count covers everything (at least everything that can or that needs to be counted). A trivial aside: this note’s author was taught the Minto methodology by Barbara herself, in the 1990s.
The internal taxonomy can be used to create a use-one tag system. While a harm as experienced shows up in multiple places, to make the data usable each harm has to be tagged by both the vector (what caused the harm), and the industry, and its role in a supply chain, and the locality, and more. Each analysis should draw only once on each data point. For instance: the apparel industry’s external harms include climate costs, post-production waste (including harm to the ocean from unsold clothes that appear dumped there) and social harms to poorly-treated workers. To count the industry’s harms, you include all of these. But for analysis that might be concerned with the harms to poorly treated workers in Southeast Asia, you need data on those for multiple countries in that region, and not the climate costs. Conversely, an ocean-only analysis does not ingest social harm data.
Use of reliable data, and expert human curation. We can build a search-and-review pipeline that uses good metrics to sift through all reports on a specific subject and still come up short. Any report has to be reviewed by independent experts tasked with challenging our data, until it reaches the standard needed to support decisions by policymakers, or bankers, by insurance executives and local governments.
Multiple valuation methodologies. A subject that requires far more than a sub-paragraph. Costs compound, fall unevenly, and have a non-linear trajectory. Small harms to any system can be net beneficial, since they provoke an immune response; large harms can be catastrophic, creating a literal version of the metaphor’s black hole outburst. And different groups or experts, even assuming all are acting with good faith and intent, can see things differently. A harm that is inconsequential to one community or species can be disastrous to another.
Openness and humility. We commit to showing our work, where the data came from, how we did the valuation, which experts reviewed it (and what they said). If someone says, in good faith, that our valuation procedure for something is wrong and that theirs is superior, that’s great — now we’ve got another valuation and perhaps a new partner.
Collectively we can solve an extraordinary information gap. We know Sagittarius A*'s mass to within a few percent, from watching stars orbit something we cannot see, 26,000 light years away. We do not know the size of the harm we do to ourselves even to within a factor of three.
The central thesis of Calx’s work builds on the recognition that the externalities of human activity are insanely vast. There’s nothing new in that. Beyond that, we hold that counting them, with decision-grade accuracy, and computing also the best paths to mitigation, can yield data and practical insights on pathways to enduring wealth, health. A real prosperity, measured in many trillions of dollars, secured.
The black hole, shrinking to insignificance.