Palm Reading

This is an oil palm, Elaeis guineensis, a perennial native to West and Central Africa now grown commercially across the equatorial tropics. Commercial plantations are concentrated in hot, humid lowlands, generally within about 10 degrees of the equator, where annual rainfall exceeds roughly 1.6 meters and is well distributed through the year. Palms dislike both sustained moisture deficits and prolonged waterlogging. Well-managed estates in good weather produce roughly 4–5 tonnes of crude palm oil (CPO) per hectare per year, and even at the 3–3.5 tonnes that Indonesia and Malaysia average nationally, the oil palm is the world’s most land-efficient oil crop. Palm oil is used in food, oleochemicals, and household products, while a growing share is also consumed as biodiesel feedstock.

The crop’s commercial center of gravity has moved decisively to Southeast Asia. Indonesia and Malaysia respectively account for 58.3% and 24.1% of total projected world production in 2026/27. As a percentage of gross international exports their share grows to 87.4%, though their export shares are not identical to production shares — Indonesia consumes a large and growing volume domestically, particularly in food and biodiesel, whereas Malaysia exports a larger proportion of its crop.

The oil palm has a long biological memory. The inflorescence that eventually becomes a harvested bunch develops over 2–3 years. Water or heat stress can reduce current bunch weight, disrupt fruit set, cause developing inflorescences to abort, or shift the palm toward producing more male flowers. Consequently, part of a weather shock may be visible immediately, but the largest production effect commonly appears about 6–12 months later and can leave a smaller tail into a second year.

The economic implication is that palm oil supply is geographically concentrated, susceptible to outlier weather events, and the impact of such events appears in the harvest after significant delay. A shock in the Southeast Asian palm belt cannot be replaced quickly by new planting elsewhere since replacement trees take roughly three years to enter production and close to a decade to reach peak yield. In the short run, the adjustment must therefore come from inventories, prices, substitution by other vegetable oils, domestic-use policy, or lower exports.

El Niño is the warm phase of the El Niño–Southern Oscillation (ENSO), a recurring interaction between the ocean and atmosphere across the tropical Pacific. It is not a single spell of hot weather but rather a basin-wide reorganization of sea-surface temperatures, winds, air pressure, and tropical rainfall that develops over months and influences weather patterns globally.

In a normal year the trade winds blow from east to west along the equator and pile the Pacific’s warmest surface water up against Indonesia, where it feeds rising air, deep convection, and heavy rain. During El Niño the trade winds weaken and that warm pool spreads eastward into the central and eastern Pacific. Rain follows the warm water and falls farther east, while the palm belt in the western Pacific sits under sinking air, fewer rain clouds, and a greater probability of dry spells. Events recur irregularly, on average every 2–7 years. A typical one develops in the spring or early summer, reaches its greatest strength around the end of the year or early in the next, and then weakens. Most last about 9–12 months, although some persist longer.

ENSO forecasts begin with the observed condition of the tropical Pacific’s surface and subsurface ocean temperatures, thermocline depth, trade winds, air pressure, and the location of tropical convection. The defining and most easily measured indicator is the sea-surface-temperature anomaly (SSTA), the difference between a location’s temperature and its long-term average. A positive SSTA means the water is warmer than normal, while a negative value means it is cooler. The benchmark is the Niño-3.4 region, the box in the east-central equatorial Pacific outlined on the El Niño panel of the schematic above.

NOAA’s historical classification uses a three-month running average of that index. A warm episode requires at least +0.5°C for a minimum of five consecutive overlapping three-month seasons, a test that by construction is only passed in hindsight. For real-time calls, NOAA issues an El Niño Advisory when the one-month anomaly reaches +0.5°C, the three-month index is expected to follow, and the atmosphere is responding the way a coupled event requires. The Advisory for this event came on 11 June 2026. Forecasts of where the index goes next come from dozens of dynamical and statistical models, which IRI and NOAA collate every month. NOAA’s August outlook, below, puts the odds that this event is very strong through the October-to-January seasons at roughly 90–95%.

It is worth asking: if the whole Pacific is warmer, does that not mean more rain, including over Indonesia? Mostly no. Tropical deep convection fires only where the sea surface is warmer than a threshold that itself rises with the tropical mean — that is, rainfall depends on SST gradients rather than absolute temperature. Absolute warmth still bites through two channels the gradient argument doesn’t cover: a warmer atmosphere holds roughly 7% more moisture per °C (thermodynamic amplification), which is offset by higher vapor-pressure deficit driving faster evapotranspiration and soil-moisture drawdown (evaporative demand).

Note the label in the subheading of the chart above: “index/RONI”. The distinction between the Oceanic Niño Index (ONI) and the newer Relative Oceanic Niño Index (RONI) matters when comparing to historical events. ONI measures the Niño-3.4 anomaly against its own climatological baseline. RONI first subtracts the average anomaly of the whole tropical ocean, then rescales the result to preserve the historical variance — that is, it tells us how warm Niño-3.4 is relative to the rest of the tropics, stripping out the background warming that lifts every raw ocean index. NOAA shifted its official ENSO monitoring and forecasting to the relative index on 1 February 2026. The choice changes the reading of this upcoming event. On the old ONI, 2026 sits well ahead of both prior analogs, while with RONI the lead disappears, though the trendline is steeper. Both indices are three-month means, and a three-month mean is slow. The raw weekly reading, the fastest gauge there is, shows August 2026 running unprecedentedly hot for the month. Subtracting the half degree of background ocean warming, it appears the trendline is continuing.

Everything above is a forecast of one variable: rain over the palm belt, which is what the tree actually feels — and this El Niño is delivering it differently from prior events. Rainfall from April–July is higher across every producing region, consistent with an event that started from a colder base on the warming-adjusted index and coupled to the atmosphere later. Most of Malaysia ran an outright surplus over those four months.

August rainfall flipped to the expected deficit. Riau — a fifth of Indonesian output by itself — collapsed to 31% of its normal rainfall pace. Even so, the south has received more rainfall than it had at this point in prior events and has been accumulating stress only since the second quarter, while Malaysia arrives with four months of banked surplus and full soil moisture. Yield damage follows rainfall failure with the 6–12-month lag described earlier, so southern Indonesia’s harvest weakens first, from late 2026, and Malaysia’s last, from mid-2027. The damage should therefore arrive later than it did in past El Niño events, and a market that prices this one off the analog calendar will be looking for the shortage months too early.

Everything to this point has come from international weather datasets. Turning to production data from the producing countries themselves, a disparity emerges between Malaysia, which has centralized and consistent monthly reporting, and Indonesia, which, despite being the larger producer and exporter, publishes far less and has no public monthly supply-and-demand balance. The Indonesian Palm Oil Association (GAPKI) estimates arrive on a 6–8-week lag, Statistics Indonesia (BPS) publishes customs exports monthly but production only annually, the agriculture ministry’s estate-crop statistics are annual and heavily revised, and USDA’s balances are analyst estimates rather than Indonesian administrative data.

The Malaysian Palm Oil Board (MPOB), on the other hand, has much more granular data because every mill, refinery, and exporter in Malaysia operates under an MPOB license, so the board can compel complete reporting from the entire supply chain. It publishes results on the 10th of every month, a release closely followed in commodity markets. The problem this year is that the country it measures is the one that dried last, and — for the first time in a strong El Niño — it did so from a rainfall surplus rather than a deficit. Through late 2026 and early 2027 the Malaysian government numbers could keep printing “fine” while a deeper Indonesian shortfall accumulates.

Keep this in mind when reading the Malaysian production and stock figures below, all from the MPOB open dataset linked in the sources. Both are drawn on an event clock, with April of the El Niño year as month zero. Malaysian production normally peaks between August and October and then drops steeply into the new year, so a seasonal rise is not by itself a sign of health: in 2015 the seasonal pattern looked intact a full year after the anchor even as output ran 13–25% below the prior year, and stocks kept building for seven months before halving.

Rainfall sets the damage to each producing hectare, but the national print is yield times hectares in production. Both prior El Niños arrived on the heels of a planting boom, so large cohorts of young palms crossed into maturity while drought was cutting yields on the estates around them. Those new hectares did not prevent the agronomic damage but they did pad the total. In Malaysia in 1998, oil yield per producing hectare fell 16.8%, but crude palm oil production fell only 8.3%, in large part because newly mature fields became fruit-bearing.

That cushion is gone in Malaysia. Immature palms were 13.1% of planted area at the end of 1997 and 13.9% at the end of 2015, against 11.5% at the end of 2025 — and the flow matters more than the stock. MPOB’s mature area grew 3.3% in the 1998 damage year and 2.9% in 2016; in 2025 it shrank 0.5%, as replanting rose to 3.4% of planted area. Immature area is rising again for that reason, which takes producing hectares out now and returns them in three years.

Indonesia’s cushion is thinner rather than gone. It holds nearly as many immature hectares as at the 2015 onset (2.48 against 2.57 million), but on a base 42% larger, so the share has fallen from 22.7% to 15.5%, and the official mature area, which grew 3.2% in the 2016 damage year, has been flat since 2023 as aging and replanting remove about as much as matures. If drought lowers yield in 2027, there is little acreage growth left to hide it.

Palm oil can be refined into fatty-acid methyl ester (FAME), a biodiesel whose production Indonesia subsidizes in pursuit of energy independence. For every barrel of diesel distributed, whether imported or refined locally, the government mandates that it gets mixed with FAME in a ratio that has been increasing since it was introduced in 2008 at 1%. The mandate got a funded subsidy pool in 2015, and since then the requirement has ratcheted up to 20% in 2016, 30% in 2020, 35% in 2023, and 40% in 2025. In July of 2026 Indonesia increased the ratio to 50% (hence the product’s name, B50) — a considerable, policy-made source of demand. It’s worth reviewing how the government organizes and finances this subsidy to determine how it may change policy in the event of a palm oil shortfall.

Biodiesel producers buy palm oil and process it into FAME; distributors such as Pertamina buy that FAME, blend it with petroleum diesel and sell the finished B50. For eligible deliveries, the palm-oil fund BPDP reimburses the FAME producer for the verified difference between the government’s monthly biodiesel reference price, HIP Biodiesel, and its petroleum-diesel reference price, HIP Solar. Eligible means the subsidized public-service-obligation (PSO) diesel sold at the pump to trucks, buses, fishermen, farmers, and small businesses; industrial and commercial users lost the incentive in 2025 when covering them became too expensive, and now bear the market cost of the mandate themselves. Realized blending has never quite matched the mandate — in 2025 only 90.9% of the allocated FAME was actually used — so the chart below assumes 91% realization. On that basis a first full B50 year draws about 16.0–16.5 million tonnes, roughly a third of all Indonesian output. Even without meeting the full biodiesel quota, the domestic claim on palm oil is about five times the one in place during the 2015/16 El Niño. Food demand and oleochemical use are also comparatively sticky, so exports have become the residual.

That leaves the question of how the subsidy is financed. The palm-oil fund, BPDP, pays part of the gap between HIP Biodiesel and HIP Solar using an export levy. Revenue therefore rises and falls with export tonnage, since fewer exported tonnes mean a smaller levy base, while the subsidy bill rises whenever palm gets expensive relative to diesel — the palm oil–gasoil spread (POGO) — because that gap sets the amount owed on every subsidized liter. Historically, export levies have generated more than enough revenue cumulatively to cover biodiesel incentives, but not reliably from year to year. When the spread moved against BPDP, Jakarta defended the mandate by consuming reserves, raising or redesigning the levy, or trimming producer support (e.g., only subsidizing PSO).

As the required ratio of FAME grows, the subsidy becomes less sustainable and more dependent on the price of diesel relative to palm. BPDP itself estimated that B50 would require Rp41.3 trillion of incentives even with crude at $85 a barrel, and that financing the program under normal conditions would require a levy of roughly 23.8%, versus 12.5% today. B50 therefore becomes hardest to finance at peacetime crude precisely when scarce palm is making it most powerful as an export amplifier. What breaks first is unlikely to be the headline mandate. In past periods of fiscal strain, Indonesia has quietly tolerated under-blending. In 2016 the realized national blend was only 10.5% against a B20 mandate. More recently, allocations have still missed full realization by about 9%. A B50 label can therefore conceal something closer to B40 in the physical balance, returning several million tonnes to food or export markets without the political cost of an announced reversal. If crude falls, the POGO spread widens, realized blending drops materially, and the export-amplification thesis weakens.

Even if blending slips, the tonnes it frees will not necessarily reach the world market. Earlier this year the trade ministry issued Permendag 16/2026, which centralizes palm oil derivative trading and gives Jakarta more control over it. Historically, Jakarta taxed the export trade without conducting it. A private shipper needed only an Export Approval, earned the right to it by meeting a domestic-market obligation that maintained price stability for domestic consumers, and paid a 12.5% levy to BPDP. After meeting these obligations it was free to negotiate its own cargo with its own buyer at the world price. Now, starting January of 2027, palm oil may only be exported by designated state enterprises through a new government desk, PT Danantara Sumberdaya Indonesia (DSI). Under the companion Government Regulation PP 24/2026, the foreign sale price is simply “determined by” the state exporter, which may also take a margin “within the bounds of reasonableness” — a bound the regulation never defines, which is to say no bound at all. Once that machinery is operating, Jakarta can ration exports by slowing approvals, tightening domestic-market obligations, or disputing reference prices, none of which requires walking back B50. It also inserts a new layer between Indonesian supply and world price discovery — one whose declared purpose, stopping “under-invoicing,” is to push realized export prices up, run by a desk that is permitted to take a margin on them. The agriculture ministry insists DSI will not profit from the arrangement; the regulation, and Danantara’s own investment chief, say otherwise.

Jakarta also owns a sharper instrument, and it has used it twice in modern history, including during the strong El Niño of 1997/98. In times of extreme price instability, Indonesian domestic politics has required Jakarta to remove palm oil entirely from the world’s balance sheet. The reflex exists because Indonesia is not only the world’s largest exporter of palm oil but also one of the largest consumers of it as food, so much so that minyak goreng (cooking oil) carries the political weight that gasoline prices carry in the United States. When world prices rally, producers prefer export parity, the staple gets scarce at the government’s ceiling price, and the government reaches for the export valve. The revealed hierarchy has been identical for thirty years: domestic food first, domestic energy second, exports last.

The first use came in 1997/98, in the middle of the Asian financial crisis and with the first well-documented El Niño yield shock unfolding. The government appealed to exporters to limit shipments (24 Nov 1997) → imposed export surcharges of up to 30% (17 Dec) → announced a full export ban (24 Dec), effective 1 January → lifted the ban in late April into a 40% export tax, raised to 60% by July. That year Indonesian exports fell 32% on a 7% production decline.

The second use came in 2022, when Russia invaded Ukraine, the world’s largest producer of sunflower oil, another cooking oil that substitutes for palm in the global market. Through late 2021 the global CPO rally had pulled cooking oil off Indonesian shelves to such an extent that 84% of Indonesians reported difficulty buying it, and President Jokowi’s approval fell from 73.9% in January to 59.9% in April. Mitigation came in five steps over four months: retail price caps and a 20% domestic-market obligation (DMO) in January → DMO raised to 30% in March → the whole apparatus scrapped mid-March for a levy-funded subsidy → Attorney General criminal charges against trade-ministry officials in April → and on 22 April, announced personally by the president, a total export ban on CPO and derivatives. The world’s largest exporter simply left the market, with world vegetable-oil prices already at all-time highs. CPO prices tripled trough to peak.

The ban was an easy political win but an economic blunder. Within weeks domestic supply exceeded domestic need to the point that storage tanks were at capacity, mills stopped buying fruit, smallholder prices collapsed, and farmers protested. Jokowi lifted the ban on 23 May, 25 days in, citing the industry’s 17 million workers, but the machinery was never fully rolled back. The DMO became a standing export-permit ratio that persists today. The levy became a standing dial, raised to 12.5% in March 2026 to fund B50, and Permendag 16/2026 lets Jakarta ration exports without the noise of ban announcements.

Now overlay the current political landscape. Prabowo runs a personalist, statist government that analysts file under resource nationalism. His coalition controls ~80% of the DPR, the national parliament, so no legislative obstacle exists, and there is no national vote until February 2029; the Constitutional Court’s June 2025 ruling pushed local elections out to 2031, so 2027 has no ballot at all. In August–September 2025 a protest wave hit 107 cities, left at least ten dead and over 3,300 detained, and within days Prabowo had sacked his finance minister, Sri Mulyani, the cabinet’s last fiscal hawk. Protests returned in June 2026 after Pertamina raised Pertamax prices by a third, with the rupiah and the free-meals program on the same placards. Prabowo’s approval has fallen from 81.2% last November to 51.1% in July. MinyaKita, the government-backed cooking-oil brand, carries a retail ceiling of Rp15,700 a liter; it sold at Rp19,000 during Ramadan and Kompas found it at Rp21–22,000 in stores. On 30 June 2026 the trade minister rejected raising the ceiling, conceding that CPO, logistics, and packaging costs had moved past it, while the palace said affordable cooking oil was the president’s priority. The mechanism that is supposed to fill shelves at that price has already failed once this year: the DMO earns cooking-oil supply as a ratio of export permits, so when exports fell about 30% in February–March the domestic allocation fell with them, from roughly 1.9 to 1.0 million tonnes on GAPKI’s count, and by July the monthly obligation was 53,059 tonnes against 215,359 a year earlier. A price ceiling the government admits no longer covers costs, a supply mechanism that shrinks as exports do, and a president down thirty points in nine months: that is the 2022 configuration, going into a strong El Niño.

The wars in Ukraine and Iran both play a part in palm oil prices. Iran because crude sets B50’s financing gap, the cost of running the equipment that harvests palms, and the price of the fertilizer used to grow them. Ukraine because the two countries at war export most of the world’s sunflower oil, a cooking oil that substitutes for palm. Hormuz has settled into a stalemate that keeps oil prices elevated, while the stalemate in Ukraine appears to be breaking down, which threatens Ukrainian exports.

The strait remains, functionally, closed. Gulf oil exports have plateaued at about 15 million barrels per day, roughly 40% below pre-war, sustained by pipeline bypasses and a transponders-off shuttle trade moving Iraqi crude at $25–30-a-barrel discounts. Global inventories are draining at 1.8 million barrels per day and Brent sits near $90. Iran has attached six conditions to reopening and Washington has hardened in response. Neither side’s existence is at stake, both can stop when the price is right, and prices have stopped moving much on each new headline.

The war in Ukraine, on the other hand, is existential — for Ukraine literally and for Putin personally. He cannot end the war without something he can call a victory, and is thus structurally incapable of accepting a stalemate the way Tehran and Washington have. Now the war’s own long stalemate appears to be breaking, and in Ukraine’s favor. Russia’s answer has been to change tactics and go after Ukrainian exports. Over three weeks from late December 2025, at least nine strikes hit Ukraine’s vegetable-oil sector specifically — Bunge’s Oleina plant in Dnipro, the Allseeds terminal at Pivdennyi, Kernel’s oil-and-fats plant in Chornomorsk — leaving about a quarter of Ukraine’s crushing capacity idle and pushing sunflower-oil export prices to their highest since 2022. In June the campaign moved to the Odesa ports, through which more than 90% of Ukraine’s agricultural exports flow. Since early July, weekly grain and oilseed exports have collapsed by more than 90% and farmers — whose 2022-era financial reserves are spent — are calling this worse than the 2022 blockade. Ukraine could exhaust its grain-storage capacity by early November. And the escalation runs both ways: Ukrainian strikes on Novorossiysk on 11–12 August suspended two major Russian grain terminals. The Black Sea produced about 60% of the world’s sunflower oil and three-quarters of its exports going into 2022, and Russia and Ukraine are still the two largest exporters. Both ends of that trade are once again under fire.

Today’s FCPO price curve — Bursa Malaysia’s crude palm oil futures — remains elevated above prior years, and indeed has already increased 21% year to date. This has mostly been driven by the war in Iran; however, as El Niño became more certain over the last few weeks, the trendline has steepened. The relative prices of the contracts over the coming year are more instructive, though, when determining a position. In prior El Niño years, the low was set during the El Niño — a two-year low in July 1997, 3–4 months after the event was public, and a multi-year low in August 2015, five months after it was declared. From those lows the rallies ran +42% and +38% in dollars. Then the peaks came late — May 1998, 6–8 months after the drought, and January 2017, eleven months after the production trough and eight after that El Niño had officially died. Today’s curve peaks early, in April 2027, and gives back 4.9% by January 2028, putting its entire premium in the headline months. Both the historical examples and the nature of the current El Niño suggest prices should be higher at the later end of the curve. Thus the position:

Long September 2027.

Disclosure: I am long this contract, talking my book, and may add or trim at any time.