I Forecast Medicine Gaps From Hormuz. The Gloves Got There First.

"In some cases 30 percent, 40 percent, 50 percent," Xavier Barreto told the news agency Lusa on 22 April. He was speaking for Portugal's association of hospital administrators about hospital consumables, and gloves were among the items whose price had risen most. Suppliers who had committed to a price could no longer hold it, and tenders already under way were struggling to find a supplier to award, because suppliers would not risk bidding a price they could not keep. The same day he told Rádio Renascença that some suppliers were no longer delivering under contracts in force. "For now they are not critical," he said of the difficulties, "but they may become so."

A month earlier a product manager at a German medical wholesaler had described the same thing from the other end. Almost three weeks into the war, he told Der Spiegel, six of ten Chinese manufacturers had stopped taking new orders, for disposable gloves above all, and his company was holding crisis meetings three times a week. A Chinese and South Korean chemical producer, LG Bohai, had told its customers that it no longer considered itself bound by its supply contracts, citing force majeure. The reason, as the trade paper that picked up the story on 20 March put it, was that the blockade of the strait had badly disrupted the supply of naphtha and ethylene.

None of that was in the forecast I published on 19 July. I wrote about medicines: active ingredients from India and China, and visible gaps in hospital formularies that could begin in September and October. The glove signal was already public, months old, and it pointed at the mechanism I would get wrong.

This is the first piece in a section of The Manifest Archive called The Track Record, where the archive scores its own work in public. It starts with the forecast that made the most promises with dates on them.

What I Wrote on 19 July

The piece was called "Hormuz Broke the Global Buffer. Nobody's Prepared." It argued that the disruption of the strait was moving toward European households through nine channels at once: container shipping, medicines, winter gas, banks and sovereign debt, trade finance, food imports, the dollar's role in oil, the precedent for other straits, and Iran's enrichment clock. It named a window, September through November 2026, in which buffers in several of those systems would give out together. It gave an order: prices first, inventories in September and October, a political energy emergency in November, sovereign and banking stress last.

At the bottom it listed seven signals that would confirm the forecast and seven that would disprove it. Those fourteen lines are what I score. The nine channels are context; the ones without a line get a verdict in prose and no score. I scored the list once already, on 6 October, in a shorter audit on Medium. This version updates the data to 11 October and asks the question that audit left open: why did gloves move and medicines not?

From Gulf Naphtha to a Nitrile Glove

The chain has to hold chemically before it can hold commercially. Naphtha is a light oil fraction. Steam crackers turn it into ethylene, propylene and butadiene. Nitrile gloves are made from nitrile butadiene rubber, a synthetic rubber built from butadiene and acrylonitrile, and acrylonitrile is made from propylene. A glove therefore depends on a cracker, and the cracker on naphtha.

Not every disposable glove is nitrile. Vinyl gloves, which China makes in bulk, are PVC, and PVC is made from vinyl chloride, which comes from ethylene and chlorine. LG Bohai, a joint venture of LG Chem and Tianjin Bohai Chemicals Group, makes vinyl chloride, the feedstock for PVC, so the warning Der Spiegel reported in March concerned that route. Both routes start at the same cracker.

Asia's crackers depended on the Gulf for that naphtha. According to Vortexa data reported in early March, the disruption put 57 percent of Asia's naphtha imports at risk, and close to 60 percent of Northeast Asia's. Crackers in South Korea and Indonesia declared force majeure. Within ten days, butadiene in China reached its highest price since 2017. From 19 March, according to the Malaysian broker HLIB, suppliers of nitrile rubber were demanding 20 to 30 percent payment in advance. Malaysian glove makers asked their government for relief, saying the blockade of the strait had constrained their supply of nitrile rubber. For acrylonitrile I found no separate price record; that link rests on the chemistry.

The glove price followed within weeks. Malaysian brokers put the average selling price of a nitrile glove at about $27 to $28 per thousand in May, against roughly $20 before the war, which is what Reuters' April report of a rise of about 40 percent to as much as $29 implies; Top Glove's input cost for nitrile rose 36 percent from the previous quarter in the March to May quarter, according to the broker Kenanga. NHS Supply Chain warned British hospital trusts of an expected 75 percent increase, according to the Health Service Journal. In Portugal the hospital administrators reported rises of 30 to 50 percent and prices that suppliers could not hold.

The sources support three different things, and they carry different weight. Chemically, a nitrile glove can be traced back to naphtha. As market disruption, each step from the feedstock to the European buyer has its own record, acrylonitrile excepted: the naphtha shortfall, the crackers' force majeure, the butadiene price, the rubber suppliers' demands for prepayment, the glove price and the hospital buyers. As causal transmission, the evidence is thinner. I found no figure for how much of the nitrile rubber in Malaysian gloves depended on Gulf naphtha, no estimate of how much of the European price rise came from the strait and how much from the restocking that followed, and no shipment followed from a Gulf cracker to one European ward. The closest thing to a statement of cause is the glove makers' own, made to their government while asking for relief.

Nitrile Up 36 Percent, Medicine Costs Absorbed in India

Medicines start partly from Gulf molecules too. Indian manufacturers reported raw material costs up 200 to 300 percent within weeks of the closure, and the main raw material for paracetamol rose from 250 to 450 rupees a kilogram. Yet no European medicines authority has attributed a shortage to the strait. The German government told the Bundestag in June that there was no fundamental impairment of health care and no reliable indication of direct effects on medicine supply; the European Medicines Agency's shortages group found "no critical shortages linked to the conflict."

The company reports for April to June show part of where the cost went. Dr Reddy's reported a gross margin more than ten points lower than a year earlier, citing among other things "elevated solvent costs arising on account of the Middle East crisis" and, in the same sentence, "price erosion in North America and Europe Generics." Cipla's finance chief told analysts the company was absorbing war-related costs of "maybe a 1% to 2%" of revenue. Lupin said the first quarter was cushioned because "we had inventories," and became "a little more cautious" on its margin outlook for the year; its chief executive described rising rebates in Germany. None of the Indian suppliers I checked withdrew from Europe. Several grew there.

The reason the cost stopped in India is written into European rules. Generic medicines in Germany are reimbursed under reference prices and rebate contracts, and a law passed in July extends the freeze on medicine prices to 2030. Medicines for Europe, the generics industry's association, said in September that price-only tendering applies in up to 84 percent of medicine procurement contracts. A supplier facing a higher input cost under those rules can absorb it, or stop supplying. The producers I examined show absorption as one of the mechanisms that kept the shock from the European buyer. Four company reports do not describe the whole European generics market, and Dr Reddy's margin fall has more than one cause.

Set side by side, the two cases show several buffers doing different work.

Each cell comes from a source cited in this section or listed below; the two marked "inferred" are my reading.

The glove row is the case that separates the kinds of buffer, because one failed and another held. Gloves had a thin price buffer and a thick capacity buffer: according to RHB, global glove supply is about 451 billion pieces this year. So the shock arrived mainly as a price; the scarcity Portuguese radio reported in April did not become a recorded stockout. Buyers slowed purchases from March to early May and restocked in May and June, the Thai maker Sri Trang said. By August the price had fallen back to about $20 per thousand. A thin buffer of one kind let the cost through; a thick buffer of another kind stopped the shortage.

A true counterexample would be a product with thin buffers of every kind that did not move. I looked for one among the products in this audit and did not find it. Since I chose the products, that absence counts for less than it seems.

Medicines show the reverse of gloves. German reimbursement rules fixed the buyer's price, so the cost stayed with the producer. Fertilizer, priced on the spot market, passed its cost straight to farmers: EU nitrogen fertilizer prices in April were 71 percent above their 2024 average, according to Commission data, and the Commission proposed a 540 million euro relief package. Helium was exposed, since Qatar produced about a third of the world's supply in 2025, according to the US Geological Survey. It had stock, managed supply and a substitute source, and the German government reported an average import price so far in 2026 below that of 2025.

The strait carries molecules. On this evidence, which buffer gives way, and to whom, appears to decide how far they travel and in what form they arrive.

Ten Ships a Day and Two Other Exits

Before the buffers can count for anything, the July piece's three exits have to be closed, because it was built so that it could not lose.

The first exit was a condition. The forecast held "under conditions unchanged from July 18," when I put traffic through the strait at about ten ships a day. If traffic had recovered, I could say the conditions changed. My first disproof line set the bar at "Hormuz returning to 60+ ships/day before September." No count came near it. Lloyd's List Intelligence found at least 346 crossings by large cargo ships not linked to Iran in August and at least 477 in September, about eleven a day and then sixteen. The exit stays shut.

The second was a disclaimer. In the middle of the July piece I wrote that the September to November convergence "is not a forecast. It is a structural interval." Near the end I wrote that the claim was "documentation of structural conditions, not prediction." The Evidence Map at the bottom of the same piece listed "Conditional forecasts," with dates. A text that says it is not predicting while listing predictions can claim whatever comes true. I score it as a forecast, because that is how it would have been read if it had come true.

The third was an answer to the strongest objection, that high prices would destroy enough demand to prevent shortages. I wrote that "demand destruction admits the thesis; it only argues about the label." That sentence made every outcome mine. The International Energy Agency now expects world oil demand to fall by about 2.5 million barrels a day in 2026, with next year's rebound only narrowly offsetting the loss.

Five Fields for Every Line

Every audit in this section follows the same standard, so that an outside reader can reach the same score from the same record.

Each line is recorded in its original words, with its publication date, the indicator it names, its deadline and its result; the register at the end of this piece quotes all fourteen. A line is scored on its own words; the author's intent does not count, and no line may be reworded afterwards to fit what happened. Where the original wording names no indicator, or names one that no public series measures, the line is marked "not independently scoreable" and counts as a drafting failure; no substitute measure is found for it after the fact. A confirming signal that occurred through a different mechanism, or in only some of the named cases, counts as half. A confirming signal already observable on the day of publication scores as a hit and carries no weight, so the verdict reports formal hits and informative hits separately. Unscoreable lines are listed apart, as failures of drafting. Disclaimers in the original are ignored. Results are hit, half, miss or not independently scoreable, and open before the deadline; disproof lines are met or not met. Every audit names the easiest line it hit. And confirming and disproof lines are often mirror images, so fourteen lines are not fourteen independent tests; the verdict weighs channels.

Fourteen Lines, Four Nobody Can Score

Counted separately. Confirming lines: two formal hits, of which one carries information (subsidies) and one carries none (LNG); one half; one miss; one open; two not independently scoreable. Disproof lines: three not met, two open, two not independently scoreable. The four unscoreable lines are drafting failures of the July piece and count against it as a forecast document. The verdict is reached by channel; the fourteen lines are not added up.

By channel: containers miss; medicines not demonstrated; winter gas open, with its scoreable half easy; production open; trade finance unscoreable; food not disproved; subsidies a hit on a line that named no country. Predicted order held: no. Verdict so far: the goods channels failed; the medicine shock has not been demonstrated; on winter energy the easy half is on course and the hard half, a political emergency, has not appeared. Easiest line: LNG diversion, which was already happening when I wrote it; it scores as a hit and tells you nothing. The subsidy line was nearly as easy, because it named no country. Easiest line still open: storage below 85 percent. Final audit: 1 December; the production line in January.

Gas: 72.97 Percent on 6 October

Qatar shipped 18 cargoes of liquefied gas in the first six months of the war against 509 in the same months a year earlier, according to ICIS data reported in late August. QatarEnergy had declared force majeure in March, which is why the diversion line carries no weight. On 6 October European storage was 72.97 percent full, according to the storage operators' AGSI+ platform as reported by TASS, against 82.9 percent a year earlier, the lowest level for the date since records began in 2011. If injections continue at the pace of September or early October, storage reaches between about 78.4 and 79.3 percent on 1 November. That range is a scenario at recent injection rates, my own arithmetic, and no independent forecast; I found no published projection. Reaching 92 percent by 31 October would take about 0.76 points a day, more than three times the September pace, which is why the disproof line is out of reach without being formally settled.

The likely hit is worth less than it looks. The European target is still 90 percent, but Regulation (EU) 2025/1733 lets a member state deviate by up to 10 percentage points in difficult conditions and lets it meet the target at any point between 1 October and 1 December. The law already allowed national targets of 80 percent or lower when I wrote that storage would end below 85, and my own text called 80 the practical threshold. I set a line the law had made easy to hit.

The harder half of the gas forecast is still open. The July piece said that if storage entered winter below 80 percent, a political emergency would follow: rationing, price controls, forced industrial cuts. My range sits just below that level. Nothing points to an emergency yet. The Commission has asked for voluntary demand reduction and sees "no immediate risks to security of supply," according to Euronews, and I found no member state that raised its crisis level this autumn. That claim is scored on 1 December.

Containers: Back Through Suez on 19 September

In mid-September Maersk and Hapag-Lloyd announced that four services would move off the Cape of Good Hope and back through Suez, with the first sailing on 19 September. Drewry's World Container Index for Shanghai to Rotterdam fell from $4,873 per forty-foot box on 16 July to $3,485 on 24 September, and Drewry linked the softening to the Suez return. Read literally, Maersk did withdraw services from the Cape route. I still score the line as a miss, because the July text defined a withdrawal as a route that "will not be served," and these routes were served faster.

Factories: A PMI of 52.9 in September

The euro area manufacturing PMI read 52.9 in September, the highest since May 2022. Stellantis is stopping four French plants for stretches of the second half of October. At Sochaux it cites a shortage of batteries for long-range electric models; at Rennes, Mulhouse and Poissy it cites industrial reasons and demand. None of the stated causes is the Gulf supply chain. Just Auto, citing Bloomberg, linked the battery shortage to a jump in demand for electric cars after the war pushed fuel prices to records, which earns the line half. The disproof line names capacity against July; the European Commission's quarterly business survey measures capacity utilisation for cars and for pharmaceuticals, and its January round settles the line.

Medicines: 546 Notices in January, 557 in October

Germany's drug regulator, BfArM, listed 546 supply shortage notices at the start of January and 557 on 5 October. The available German notifications did not show the surge the July forecast implied. That is narrower than proof that European hospitals saw no disruption: a national count of notices does not measure a hospital formulary, and new shortages can replace old ones without moving the total. It is enough to say that the predicted medicine shock has not been demonstrated.

Bonds: 150 Basis Points in Paris

The forecast put sovereign stress last and pointed at Italy. It came earlier and elsewhere. France's ten-year spread over Germany reached about 150 basis points on 2 October, its widest since the debt crisis of 2011 and 2012, after the government presented its 2027 budget. There is a path from the strait to French borrowing costs. The sources I read attribute the French move mainly to the deficit, the debt and next April's election. I do not claim this line.

Food, Subsidies, Payments and the Enrichment Clock

The FAO food price index rose through July, August and September to 136.0; FAO cites weather and transport disruption, naming the Strait of Hormuz alongside the Black Sea. Pakistan's cabinet economic committee approved 75 billion rupees of fuel relief on 14 September; in October the IMF told Pakistan to phase out costly fuel support schemes. I found no documented Gulf oil sale moved into yuan. The International Atomic Energy Agency said in September that it cannot verify Iran's stockpile, so the enrichment claim cannot be tested. The payment and enrichment claims were July channels without signal lines, and are not scored.

507 Million Barrels I Did Not List

The forecast said inventories would give way in September and October. One did, on a channel my list did not name, so it supports the July timing without scoring a line.

The IEA reported in September that observed oil stocks had fallen by 507 million barrels since February. Oil volumes came back in a form my disproof line could not see. By the last week of September, Kpler data reported by Reuters showed crude exports from the region above their pre-war level on four days, with a seven-day average of 18.5 million barrels a day on 1 October against about 18 million in the year before the war. Kpler's count includes exports from Red Sea terminals, which Saudi oil reaches by pipeline around the strait, and leaves out ships that crossed with their transponders off. A line written in ships at the strait could not register that the barrels had found their way out.

Brent was still around $100 in the first week of October, after a renewed wave of attacks: Lloyd's List Intelligence counted at least thirteen attacks on tankers and gas carriers since 28 September, the most in seven months. The cost of the strait now sits in the freight, the insurance and the risk of sending oil through it.

Four Corrections Beneath the July Text

Four claims in the July piece do not hold as written. The Singapore figure of seven days and 450,000 TEU comes from a Linerlytica report of May 2024 and was not measured in June 2026. The 80 percent threshold derives from Regulation (EU) 2025/1733, which the July piece credited to ACER. The pre-war baseline of 88 ships a day and the Brent price of $88.10 for 18 July could not be matched to a source. A dated correction is published beneath the July text, which is otherwise unchanged.

From 2.6 to 3.8 Percent Without a Line-by-Line Review

The institutions missed too. In March the European Central Bank's baseline put euro area inflation at 2.6 percent for 2026, with severe paths up to 4.4 percent; the flash estimate for September alone was 3.8 percent, and the IMF's July update assumed the strait would reopen from mid-July. They review their errors in aggregate; for this crisis I found no line-by-line comparison, and a forecast that is never scored is a story with dates in it.

Three Objections and Six Registered Bets

The strongest objection is that this is still the forecaster grading himself. I chose to score before the window closed, which lets me set the story while the gas line still looks good. I chose the explanation after seeing the results, and a pattern found after the fact is easy to find.

There is a harder objection on the medicines line. The cost of Gulf inputs landed on Indian manufacturers in the spring, and Lupin said its first quarter was cushioned by inventory and that the higher cost would show in later quarters. If that squeeze ends with producers stopping lines, the shortage would arrive after my window, and the miss scored here would be a timing error.

The hardest objection is that the buffer reading can explain anything after the fact: wherever a shock did not arrive, a buffer can be found that stopped it. The table above answers that only in part. It names four kinds of buffer, assigns each from a cited source, and contains a case, gloves, where one kind failed and another held. That makes the reading falsifiable, and the bets below are the test. Each is registered separately, so that one result cannot be blurred by another, and each states what a failure would say about the reading.

Each of the following was registered on 11 October, with its indicator, its threshold and its deadline.

Gas. The EU aggregate on GIE's AGSI+ platform for gas day 1 November, as first published, will be below 80 percent. It is disproved at 80 percent or above. This bet tests the July line and says nothing about buffers; a failure would mean injections outran my extrapolation.

Medicine attribution. By 31 December none of the EMA, BfArM, ANSM, AIFA or Infarmed will attribute a medicine shortage to the strait. A failure would mean the absorption buffer gave way somewhere and was recognised as doing so, the direct refutation of the medicine half of the reading. Authorities rarely name causes, so a pass is weak evidence.

Medicine count. BfArM's count of supply shortage notices on 31 December will be no higher than 585, within 5 percent of the 557 recorded on 5 October. A count above 585 would show availability worsening while producers are still said to be absorbing costs, which is what the timing objection predicts. On its own it would say nothing about the strait, and it would send me to the stated reasons of the new notices.

Glove price. Nitrile rubber costs rose again in September, from about $950 to $1,100 a tonne according to the broker Kenanga, and Top Glove expects its prices to climb from about $20 to $22 or $23 per thousand. The average selling price for the quarter to November, in Top Glove's results due in December or in broker notes on them, will stay below $25 per thousand, at least $2 under the May peak. A price at or above $25 would mean the surplus no longer caps the price, which is the core of the glove case.

Glove supply. By 31 December no European health authority or hospital association will report a glove stockout. A report would mean the capacity buffer failed somewhere, whatever the price did.

British concessions. Where reimbursement follows market prices, the cost should show as price, and Britain is the clearest large European case. The count of drug price concessions granted by the Department of Health and Social Care, summed from Community Pharmacy England's updates, will exceed 200 in each of October, November and December, against about 107, 100 and 112 in the same months of 2025 and about 200 in March 2026. It resolves with December's final list in early January. A month at or below 200 would mean price pass-through had faded while the strait stays disrupted, which weakens the price half of the reading. Concessions were already rising before the war, passed 250 for the first time in June and fell to 226 in July, so this is probably the easiest of the six and the weakest test of the cause.

Back to $20 per Thousand

The July forecast measured the size of the shock. It did not measure the distance between the shock and the people expected to feel it. On the record so far, that distance was filled with stock, contracts, spare factories and balance sheets able to absorb a cost. I saw the disruption and assumed the consequence would arrive with it.

The evidence for how it did arrive had been sitting in a German trade paper since March: a wholesaler whose Chinese glove makers had stopped taking orders, and a chemical plant that had declared itself free of its contracts. By August the glove price was back near where it started.

No medicine shortage was traced to the strait. The gloves went up.

Evidence Map

Facts, interpretations, forecasts, and disconfirming signals.

Core claim. Verdict so far on the forecast of 19 July 2026: the goods channels failed; the medicine shock has not been demonstrated; on winter energy the easy half is on course and the hard half, a political emergency, has not appeared. The audit's working hypothesis is that the shock reached European buyers according to which buffer gave way: as a price where contracts and prices were thin (gloves, fertilizer), as producer losses where buyers' prices were fixed (generic medicines), and with no recorded effect where stock and substitutes held (helium). The July piece also contained three devices that protected it from being wrong, which this audit closes before scoring.

Source uncertainty. Low for container rates, Lloyd's List Intelligence counts, the EMA and German government statements, company results and the regulation. Medium for gas storage, taken from AGSI+ as reported by TASS because I could not query the platform directly, and for the glove chain, which rests on trade press, broker notes and associations.

Causal uncertainty. High for the buffer reading. It was formulated after the results; it assigns buffers from sources, but two cells in the table are inferred; and it cannot yet be fully separated from differences in how markets set prices.

Forecast uncertainty. The gas prediction rests on a short extrapolation close to the threshold. The glove price prediction depends on oil and nitrile costs over the next two months. The medicine attribution prediction has little power, because authorities rarely name causes. The British concessions prediction uses an indicator that was rising before the war.

What would confirm this. Storage below 80 percent on AGSI+ for 1 November; no attribution of a medicine shortage to the strait by the five named authorities by 31 December; BfArM shortage notices at or below 585 on 31 December; a nitrile glove price below $25 for the quarter to November; no glove stockout reported by 31 December; British price concessions above 200 in each of October, November and December.

What would disprove this. Storage at or above 80 percent on 1 November; any attribution to the strait by the five named authorities; more than 585 BfArM notices; a glove price at or above $25; any glove stockout report; any month of British concessions at or below 200. Each is scored on its own.

Watchlist. AGSI+ storage data weekly; EMA, BfArM, ANSM, AIFA and Infarmed shortage notices; Community Pharmacy England concession updates; Top Glove's quarterly results in December; Lloyd's List Intelligence Hormuz briefs (next on 22 October); the IMF October World Economic Outlook; the ECB decision of 29 October.

Audit Register

Each line of 19 July in its original words, the indicator used, the measurement date, the outcome, the main uncertainty and the numbered sources in the Source Register below.

Source Register

Every claim that carries an audit outcome or the buffer analysis, numbered in the order it appears. "Primary" is the issuing body's own document; "secondary" is a report of it by another outlet. The Audit Register above cites these numbers.

Related from The Manifest Archive

• Hormuz Broke the Global Buffer. Nobody's Prepared.
• Everyone Is Counting Ships in Hormuz. Nobody Agrees on What Counts.
• Six Industries Published Reports About Hormuz. None of Them Talked to Each Other
• Oil Falls While Hormuz Stays Closed: The Arithmetic

Jerry van der Laan writes The Manifest Archive, Forensic Narrative Intelligence Writing on the systems beneath power, money, and history. He traces the structures beneath them.