A Record El Niño Is Coming: Why One Weather Pattern Can Eat Ten Trillion Dollars

Contents

1. What This Episode Is About
2. The Main Points
3. Going Further
4. 1. “A record El Niño, ten trillion in losses — so what do I buy?”
5. 2. “That number looks absurd. Can I believe it?”
6. 3. “My portfolio has nothing to do with the weather. Why should I care?”
7. Further Reading
8. One Thing To Take With You

In the third month no rain came, and a parching wind rose; the wheat shoots never headed, and most withered yellow and died. In the ninth month frost fell, autumn turned cold too early; the grain ears, still unripened, dried green on the stalk.

— Bai Juyi, “The Old Man of Duling,” Tang dynasty (translation mine)

What This Episode Is About

Odd Lots brought on Justin Mankin, professor of geography at Dartmouth and director of the Climate Modeling and Impacts Group. The hosts open by outing themselves: Tracy is a weather nerd who loves thunderstorms (except on airplanes), while Joe declares himself a “weather truther” who doesn’t believe wind chill can be measured at all — “it’s twenty-five degrees but feels like fifteen; how do you know what fifteen feels like?” And humidity: “there’s water in the air, give me a break.”

The joke turns out to be the best possible opening, because the guest answers it seriously. Wind wicks water off your skin; the phase change from liquid to vapor carries away latent heat, which is why you feel colder stepping out of a shower than getting in. Wind chill is measurable — and it belongs to the same family of problems as El Niño. Both are questions about where energy goes.

But the episode isn’t really about meteorology. It’s about a 2023 paper in Science. The old picture treated El Niño as a shock you take and then recover from. The evidence in that paper says otherwise: it pushes the whole growth trajectory down, and it never comes back up.

The Main Points

One: El Niño swaps a corner space heater for wall-to-wall radiators. Normally, easterly trade winds drag sun-warmed surface water westward along the equator and pile it into a warm pool parked next to Indonesia and Australia. Every few years the trades weaken and break down, and that pile — which runs deep, not just skin-thick — sloshes back east and spreads itself across the entire tropical Pacific. The guest’s analogy: you walk into a cold room with a small space heater in the corner. It’s generating heat, but only that corner is warm. During El Niño, the room has radiators along every wall. Since the tropics are the heat engine of global weather, this isn’t “somewhere got hotter.” It’s a wholesale redistribution of the planet’s energy.

Two: naming a phenomenon converts randomness into predictability. Climatologists call El Niño a mode of variability — it has characteristic length scales in time and space, and what they call teleconnections: the physical bridge between the equatorial Pacific and, say, the Indian monsoon or precipitation in the American Southwest. Princeton geophysicist George Philander called El Niño the trunk of the variability tree. Once it’s named and characterized, “what Peru should expect” becomes a sentence you can actually say. It stops being noise and becomes a source of forecasting skill. That move is worth stealing on its own: you can’t abolish uncertainty, but you can carve a structured piece out of it, and what’s left over is a smaller pile of noise.

Three: Peruvian potato farmers were forecasting El Niño five hundred years ago. A group of Columbia climate scientists working with an anthropologist published evidence that Andean farmers shifted their planting dates based on how clearly they could see the Pleiades. The mechanism is real: El Niño moves convection eastward from around Australasia, throwing high cirrus shields over the Andes and blurring the star cluster. The guest’s takeaway is that the heavily instrumented world may be leaving climate signals on the table that are still encoded in local cultural knowledge.

Four: what “record-breaking” measures matters more than the number. Every index ultimately benchmarks sea surface temperature against a normal — and the normal is drifting, because the tropical Pacific is warming on its own. That’s why Australia’s Bureau of Meteorology uses a relative index with a rolling baseline. El Niño also has flavors: the Niño 3, 3.4 and 4 regions sample different parts of the basin, and the spatial pattern of warming determines what a given event means for US hazards or Atlantic hurricane season. No two El Niños carry the same consequences. Right now, statistical and process-based models converge on the same answer: warmer than 2016.

Five: the counterintuitive core — this is a growth effect, not a level effect. The 1997–98 El Niño was scored at the time at roughly thirty-six billion dollars, treated as a shock followed by recovery. The Science paper recomputes it with causal inference methods and finds growth was systematically depressed, so that by 2003 — five years out — cumulative global losses were about five point seven trillion. The distinction is everything. A level effect means you lose a chunk this year and make it back next year. A growth effect means you were on one trajectory, and now you’re on a lower one, and you stay there. Losses accrue indefinitely. Mankin’s estimate for the event now unfolding: roughly ten trillion dollars over five years, conservatively — up to fourteen if you take the tails.

Six: the mechanism is simultaneity. A single regional disaster gets diluted. A heat wave in the Upper Midwest tells you nothing about Russian grain or cotton in Uzbekistan’s Fergana Valley; those shocks are far enough apart in space that they don’t land together. What makes El Niño expensive is that it fires in many unrelated places at once, through many different hazard types — heat waves, drought, floods, landslides, wildfire. It isn’t one supply chain disrupted; it’s a dozen. Moving labor and capital toward their most productive uses gets hard when hazards are everywhere at the same time. That, the guest notes, is exactly why El Niño is such a good analog for broader global warming: highly correlated, simultaneous impacts.

Seven: we’ve been exposed to this for centuries and we have barely adapted. This is the heaviest line in the episode — there isn’t much evidence that we’re well adapted to the climate we already have. If adaptation were working, costs should fall over time; instead the measured costs are astounding. The guest splits the problem in two: making global markets resilient so weather shocks don’t drag prices and interest rates around, and protecting people from the hazards themselves through early warning, emergency response, and insurance markets that actually price region-specific geophysical risk. They need completely different tools, and blurring them means doing both badly. Incidentally, while discussing reservoirs he refers to Lake Powell as “the central bank of the American West water” — Tracy’s immediate verdict: he knows his audience.

Going Further

1. “A record El Niño, ten trillion in losses — so what do I buy?”

For most readers this is the first instinct: find the trade. Ags, fertilizer, natural gas — something.

Start with the first layer. Ten trillion is not a price forecast for any market. It’s the five-year accumulation of a downward shift in the distribution of global growth. It’s scattered across an uncountable number of small places: a road taken out by a landslide, a planting delayed two weeks, a mine-to-port link severed, a premium repriced. It sums to something enormous, but no clean financial instrument maps onto it one-for-one. Translating a macro narrative straight into a ticker is exactly where that gap gets lost.

The second layer matters more, and the episode answers it in the opening minutes. Tracy notes that in just the past few days she’d seen a Citi note on Thai rates, a Morgan Stanley Bank of England outlook, and a BofA emerging-market FX outlook — all invoking El Niño. When something shows up across three sell-side desks on three unrelated topics, it isn’t an information edge anymore. By the time you read it, some of it is in the price.

So where is the gap? I’d argue it’s in the time structure. Markets are good at pricing one-off shocks: hurricane lands, plant goes down, inventory drains, next quarter recovers. Those have a definite verification date, so pricing is efficient. Markets are bad at pricing a permanent shift in the growth rate, because verifying it takes five or ten years — and, as the next section covers, the researchers themselves say the statistics can’t establish permanence. Something everyone agrees matters but nobody can confirm or refute within a few quarters gets eaten by the discount rate. Not because it’s unimportant, but because nobody can settle a trade on it.

So next time you see “event X will cost N trillion,” ask two questions. Is N a level effect or a growth effect? Is it concentrated in a handful of companies, or dispersed across tens of thousands of invisible places? Together, the answers tell you whether you’re holding a trade or a worldview. Both have value — mistaking one for the other is expensive.

2. “That number looks absurd. Can I believe it?”

Flat disbelief is as lazy as flat belief; both skip the reading. This episode happens to contain a complete worked example of how to do the reading.

Joe raises a genuinely elegant objection. If El Niño has always been with us, then “a world without El Niño” isn’t a coherent counterfactual at all. It’s like saying cold winters depress growth, so if there were no Decembers we’d have grown faster. Logically true, and absurd, because the premise of a year without winter doesn’t exist.

Mankin’s answer is worth writing down. El Niño is cyclic but not perfectly so — which makes it as good as randomly assigned. That sentence is the hinge of the whole method. Because assignment is quasi-random, you can split the historical record into El Niño years and non-El Niño years and compare economic outcomes, the way a randomized controlled trial compares a treatment arm against a placebo arm. The counterfactual isn’t a planet without El Niño. It’s the same countries in the years the treatment didn’t fire. The first is science fiction; the second is data.

There’s a portable test in here. For any claim that X causes Y, ask first: where does the variation in X come from? If part of it is driven by Y, or by something correlated with Y, what you’ve estimated isn’t causation. This paper holds up because global economic activity doesn’t reach back and set sea surface temperatures in the equatorial Pacific — the arrow only points one way. And because El Niño produces so many hazard types, while countries differ in how tightly their weather is coupled to it, the design gets two layers of variation: whether it happened, and how strongly each place is connected.

Then there’s a third thing, which I find more instructive than the headline. The team ran what’s called a perfect model framework: assume the effect really is permanent and exactly the size we estimated, then ask whether our actual data record could recover that known effect. It couldn’t. So the claim they make is “we cannot rule out permanence” — not “we have proven permanence.” A researcher volunteering that their data is insufficient to prove the thing they’d like to prove is a strong credibility signal. Conversely, a report that never says what it failed to establish usually isn’t silent because it established everything.

Three things to check in any study: where the variation comes from, how the counterfactual was built, and whether the authors say plainly what they couldn’t show. The third gets read the least and carries the most signal.

3. “My portfolio has nothing to do with the weather. Why should I care?”

The portable idea in this episode isn’t El Niño. It’s simultaneity.

Read the guest’s contrast again: a Midwest heat wave has nothing to do with Russian wheat or Uzbek cotton, so that shock gets diluted by the system. That dilution is what diversification means. El Niño is expensive precisely because it fires in many unrelated places at once — not one broken supply chain but a dozen. In other words, El Niño is a correlations-go-to-one event.

Which is also the definition of diversification failing. A portfolio holding semis, medical devices, software and packaging looks well spread across sector labels. But if those names share an underlying variable — the same customers’ capex cycle, the same shipping lane, the same currency, the same power source, even the same embedded assumption that rates will fall — then on the day that variable breaks, the diversification was never real. Diversification isn’t a static distribution of labels. It’s a claim about what would have to happen for things to break together.

So the way to audit it isn’t counting sectors. It’s asking the inverse: is there a single event that would make three or more of these positions worse at the same time? Joe reaches for this himself at the end of the episode — we saw a globally synchronized shock like that during the pandemic.

That’s the most practical homework this episode offers an ordinary investor. Not to forecast El Niño. Go back to your own holdings and try to write that sentence down. If you can write it, you know what your real risk looks like and can decide whether to keep it. If you can’t, there are two possibilities — you’re genuinely diversified, or you just haven’t thought of it yet. Keep those two apart, because the next step differs completely.

One closing detail from the episode. Tracy points out that all of this lands against a backdrop of elevated fertilizer prices and constrained supply, some of it stuck in the Strait of Hormuz. That’s the simultaneity point in miniature: each item alone is manageable; the stack is the problem.

Further Reading

• Bloomberg Odd Lots, this episode: “A Historic El Niño Is Coming That Could Cost the World Trillions” (2026-08-14), with guest Justin Mankin
• Callahan, C. W. & Mankin, J. S., “Persistent effect of El Niño on global economic growth,” Science, 2023 — the paper at the center of the conversation
• Mike Davis, Late Victorian Holocausts: El Niño Famines and the Making of the Third World (2001) — the 1877–78 event and the famines in India, northern China and northeastern Brazil
• Orlove, B., Chiang, J. & Cane, M., on Pleiades visibility and Andean crop forecasting, Nature, 2000
• NOAA Climate Prediction Center ENSO Diagnostic Discussion, updated monthly
• Australian Bureau of Meteorology ENSO updates and the relative index documentation

One Thing To Take With You

The line that stayed with me was the definition of diversification: not how many kinds of things you hold, but what would have to happen for them to break together. That sentence is how I realized the spread-out-looking row of things I owned all hung on one assumption — invisible to me because that assumption was the very reason I’d picked them.

Here’s a small thing I tried, if you want to try it: today, pick a friend and describe three things in your life you couldn’t easily replace — an income, the place you live, the one relationship your whole social world runs through — but don’t tell them what you’re worried about. Ask one question: what single event do you think would make all three worse at once? Save their answer, word for word, in your phone. The first time I did this, my friend named something within three seconds that I had never once considered, and that to me didn’t even count as a risk. You can’t get there on your own, because the person who chose and the person who’s looking are the same person.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.