On 21 August the Gulf route earned 34% more per day than the long haul while charging less per tonne. The market is paying most for the voyage nobody will make — and that changes who keeps the arbitrage.
Shubham Vashist·September 2026·11 min read All figures publicly sourced
On 28 August the Baltic Exchange assessed the Houston–Chiba VLGC route at $263.33 a tonne, earning $154,458 a day. The same report put Ras Tanura–Chiba at $217.75 a tonne — and $210,291 a day.
The cheaper route pays thirty-six per cent more. And almost nobody will sail it.
The obvious reading is that money is being left on the table. It isn't, and understanding why explains where the 2026 propane arbitrage actually went.
The short version
A record US–Asia propane arbitrage opened in March at $307.58/t. Freight and terminal fees consumed nearly all of it before it reached anyone holding a molecule.
The Gulf benchmark reads high but is barely traded. At reported war risk of 7.5–10% of hull value, insurance alone is $188–250/t against a published rate of $217.75/t. The route isn't underpriced — it is uninsurable at a clearing price.
Asian PDH buyers paid the bill, cutting runs to a 47% trough, then restarting on a 3% feedstock move. Short-haul and tariff-free origins took share without adding a barrel.
Owners ordered 38 vessels in one quarter — but Dorian sold three ships into the spike while buying one for 2029. The specialists disagree with each other.
The fleet is not obviously long: ~936 kb/d of new long-haul US capacity lands in the same window as ~41% more tonnage. This turns on war risk repricing, not deliveries.
Freight, transits and capacity data current to 28–30 August 2026 · PDH run rates are the latest published, to June 2026 · For publication 10 September 2026
I · Where It Stands
Day 183, and No Resolution
Transit monitoring recorded three vessel movements through Hormuz on 23 August against a pre-crisis baseline near eighty-five a day, with 393 vessels holding position in late August. Lloyd's List Intelligence puts non-Iranian-linked transits at 68, 60 and 43 across successive weeks from 27 July. On 28 August QatarEnergy extended force majeure again, with Pakistani cancellations running into October.Verified
One signal points the other way: Bloomberg reported Ras Laffan loadings rising to a ten-day average near 80,000 tonnes in mid-August as Qatar prepared for a possible reopening. The exporter is readying; the market is not pricing it.
Fig 1The Chokepoint, Late August
Daily transits against pre-crisis baseline, and weekly non-Iranian-linked movements
Sources: PortWatch transit monitoring via public tracker, 30 August 2026; Lloyd's List Intelligence, Strait of Hormuz Brief, 19 August 2026; Bloomberg, 11 and 28 August 2026. Verified
II · The Barrier
Why the Gulf Route Does Not Clear
Almost nobody fixes BLPG1, so the assessment is not a profit pool. It is a reservation price — what the market would have to pay to get a mainstream owner into the Gulf. On a route that is barely running, the level measures the height of the barrier, not the size of the prize.Interpretation
And it is quoted against a cost that does not appear in it.
Fig 2The Gap Is Persistent — and Widening
Baltic VLGC time charter equivalent earnings, US$/day, by benchmark route
Source: Baltic Exchange weekly gas reports, weeks 31, 34 and 35 of 2026. Percentages calculated from published TCE assessments. Interpretation
War risk is the missing term. Gulf premiums ran 0.1 to 0.15 per cent of hull value before the conflict. S&P Global reported roughly 2.5 per cent in March, easing to about 1 per cent by month-end. By July, reporting put them at three to ten per cent, with S&P-sourced figures of seven and a half to ten. Howden Re calls the combined Red Sea and Hormuz experience a permanent structural repricing rather than a spike.Verified
Convert that into the unit freight trades in. A VLGC carries about 44,000 tonnes, and Dorian LPG contracted a newbuilding with HD Hyundai in June at approximately $115 million, so take $110 million as a working hull value. At the pre-war 0.15 per cent, war risk costs about $4 a tonne. At three per cent, $75. At seven and a half, $188. At ten, $250 — more than the entire published rate.Estimate
Fig 3Why the Gulf Route Does Not Clear
War risk premium per tonne of cargo against the published BLPG1 freight rate, US$/t
War risk levels: S&P Global (March and July 2026), Al Jazeera citing S&P Global (July 2026), The National (July 2026). Per-tonne conversion by Insight Stream assuming a 44,000-tonne cargo and a $110 million hull value, benchmarked against Dorian LPG's June 2026 newbuilding contract of approximately $115 million; AWRP is typically quoted per seven-day exposure period. Freight: Baltic Exchange, 28 August 2026. Estimate
So the picture resolves. A mainstream owner with conventional cover, a flag state to answer to, and crew who under July's IBF designations hold an explicit right to refuse to sail into the area cannot make that voyage work at $217.75 a tonne. The economics close only for operators outside that structure — which is precisely what Lloyd's List Intelligence describes: a bifurcated market west of Hormuz, with at least four VLGCs previously carrying Iranian LPG loading in the UAE and Qatar.
The market did not split by geography. It split by insurability.
This leg reopens when war risk reprices, not when shipyards deliver. Watch the AWRP quote, not the order book.
III · Who Kept the Arbitrage
Four Claimants, and the Middle Took It
In late March, OPIS assessed the Mont Belvieu–Far East propane spread at $307.58 a tonne, up from $227.26 a month earlier. US LPG exports hit a record seven million tonnes in April. That should have been a windfall.Verified
OPIS stated where it went: shipping costs came to command the lion's share of the arbitrage value, and when the arb says go but the netback says wait, the bottleneck is in the water. RBN's assessment was blunter — terminal fees closed the spot Asia arb. The dock's own windfall proved fleeting, with USGC terminal fees falling roughly seventy per cent from their early-April peak within the month.
The buyer paid the balance
A PDH plant earns the gap between propylene and delivered propane. Delivered propane is the Far East Index plus freight — so inflated freight raises the input while the output stays priced in a domestic Chinese market that does not care what a VLGC costs. Below breakeven, the rational move is to shut.Interpretation
Chinese PDH units ran near sixty-four per cent in February and March, fell to fifty-five in April on feedstock shortage — Argus put the trough at forty-seven — and recovered to sixty-five to sixty-eight per cent in June as the prompt Far East swap eased from $762 to $737 a tonne. The IEA recorded Asian petrochemical run cuts of ten to thirty per cent of capacity, with regional naphtha demand down 450,000 b/d and LPG and ethane down another 320,000 b/d.
Read the recovery carefully. Rates came back on a roughly three per cent move in feedstock. That is an extremely price-elastic buyer: it will absorb very little further cost, and it will bid hard the moment delivered propane cheapens.Interpretation
Fig 4The Buyer Absorbed It by Switching Off
Chinese PDH operating rates through the disruption, 2026
Sources: OPIS, April and June 2026; Argus Media, June 2026; IEA via Oil & Gas Journal, April 2026. Verified
And proximity got paid
Vortexa data reported by OPIS shows Canadian cargoes into China rising fifty-seven per cent month on month in March, from 184,800 to 289,400 tonnes, with West African volumes up seventy-seven per cent. Neither origin added production. Two advantages stacked: shorter voyage, and no exposure to the ten per cent Chinese tariff on US cargoes.Verified
Fig 5Short-Haul Origins Took Share Without Adding Supply
LPG cargoes into China, February to March 2026, tonnes
Source: Vortexa data reported by OPIS, June 2026. Verified
IV · The Response
Steel Ordered Against a Strait
Veson Nautical reports 38 VLGCs and very large ammonia carriers ordered in the second quarter alone, against ten in all of 2025, taking orderbook-to-fleet near twenty-eight per cent and implying roughly forty-one per cent more tonnage by 2029.Verified
Fig 6A Year's Worth of Ordering in One Quarter
VLGC and VLAC newbuilding orders
Source: Veson Nautical, Shipping Market Outlook Q3 2026. Verified
The aggregate conceals a split worth seeing. On 23 June, Dorian LPG announced one 90,000 cbm newbuilding for 2029 delivery at about $115 million — and in the same release agreed to sell three older vessels for roughly $256 million. BW LPG went the other way, ordering eight Panamax newbuildings in May. One specialist is monetising near-term tonnage into the spike; another is building scale. Dorian's filing also repeats the realisation gap: 99 per cent of second-quarter days fixed above $68,000 against a spot average nearer $110,000.Interpretation
But the fleet is not obviously long
Every tonne returning to the Gulf removes about fifty-eight per cent of the tonne-miles it was generating — 6,700 miles against 15,900 via the Cape. Taken alone that reads as overhang. It counts one side only.
US export capacity is expanding on a comparable scale, and all of it long-haul: roughly 300,000 b/d at Enterprise's Houston Ship Channel terminal by end-2026, Targa's Galena Park rising from 444,000 to about 625,000 b/d by Q3 2027, a 400,000 b/d Oneok–MPLX terminal at Texas City for early 2028, and 55,000 b/d more at Nederland. That is about 936,000 b/d — on the order of forty per cent more US export volume, against forty-one per cent more ships. Argus puts total US capacity beyond three million b/d by 2028; the EIA expects the US propane surplus to keep growing through 2027.Estimate
Canada adds roughly 217,000 b/d through REEF, its optimisations and Trigon — but ten to eleven days from northeast Asia rather than twenty-five-plus. Canadian growth is excellent for the netback and tonne-mile dilutive for the shipowner. And a reopening does not automatically restore Gulf volume: Iran shipped 11.35 million tonnes in 2025, 84 per cent to China, and those barrels sit behind a naval blockade that is a separate question from the strait.Interpretation
Fig 7Whether the Fleet Is Long Depends on Two Political Variables
Tonne-mile demand by 2029 against roughly 41% fleet growth
Scenario framing by Insight Stream. Inputs: Veson Nautical (fleet growth, Middle East export recovery); Enterprise, Targa, Oneok/MPLX and Energy Transfer capacity announcements; Argus (US capacity beyond 3mn b/d by 2028); EIA (growing US propane surplus through 2027); Clarksons via Splash247 (route distances); Baltic Exchange, 28 August 2026 (Panama transit and draft restrictions). Quadrant outcomes are qualitative, not modelled volumes. Estimate
So the fleet is long in one corner of that matrix and broadly balanced in two others — and which corner arrives is decided by Hormuz and Panama, not by shipyards. Note that the Panama Canal authority advised in the same 28 August report that low water levels are again forcing reduced transits and draft.
The honest objection
Three arguments cut against this. The strait shows no sign of reopening, and a closure that persists keeps the long haul intact well into the delivery window. A meaningful share of recent ordering is for ammonia carriers serving a different trade, so reading the order book as pure LPG overhang overstates it. And BLPG1 is thinly traded, which makes any single assessment noisy. The reply to the last is that it held and widened across five consecutive weeks — a persistent reservation price on a route nobody sails is itself the information.
V · The Position
What Follows
1 · Underwrite the two premiums separately. Today's blended freight contains a distance component that more tonnage will erode, and an insurance component no yard addresses. A multi-year charter at today's level is long both.
2 · US Gulf sellers: do not assume the freight saving arrives. Roughly 936,000 b/d of new long-haul capacity lands alongside the new ships. Relief requires Hormuz or Panama to normalise, and neither is in your control — so price term business freight-indexed rather than freight-inclusive, and expect a price-elastic PDH buyer to capture part of any saving.
3 · Short-haul and tariff-free origins: convert the advantage now. Canadian volumes into China rose fifty-seven per cent on distance and tariff position alone. Both edges narrow when the Gulf returns.
4 · Watch in this order: the Gulf AWRP quote, then transit counts, then the share of US–Asia cargoes still routing via the Cape. Freight assessments move last, not first.
Our call, stated so it can be checked
We expect BLPG1 to stay elevated and thinly traded while Gulf war risk sits materially above one per cent of hull value, and to fall sharply within weeks of AWRP normalising — ahead of and independent of vessel supply. We expect the freight share of US–Asia delivered cost to peak in the first half of 2027, but not to compress straightforwardly thereafter, because the new long-haul capacity lands in the same window. And we expect Chinese PDH rates to recover above pre-war levels before US producer netbacks do. Revisited in twelve months against the same public sources. Estimate
What would change our mind
A sustained reopening. If transits recover toward the ~85/day baseline for four consecutive weeks and the BLPG1 premium does not collapse first, the premium is capacity-driven and our reading is wrong.
The gap closing while the strait stays shut. BLPG1's premium narrowing below roughly 15% with no change in transit counts would mean something other than insurability is setting it.
US capacity slipping. If Texas City or Galena Park slips materially, long-haul tonne-mile growth undershoots the fleet and the overhang case strengthens well before 2029.
PDH stalling below 60% on cheaper feedstock. That would make the demand loss structural rather than price-driven, and the value would not return to the buyer as we expect.
VI · Closing
Freight Ate the Arb
2026 produced one of the widest US–Asia propane spreads on record and one of the most profitable freight markets in the history of the trade. Those are usually two stories. They are one.
The spread was a gross number four parties negotiated over, and the scarce link took it. Producers exported records into an arbitrage they largely did not keep. Petrochemical buyers switched off rather than pay. The dock captured a windfall and surrendered most of it in a month. The ships kept the rest.
It is tempting to conclude that forty-one per cent more tonnage settles what happens next. It does not — nearly a million barrels a day of new long-haul US capacity arrives in the same window. The fleet is clearly long only if Hormuz reopens and Panama normalises, two political outcomes on nobody's construction schedule.
Which is why the number worth watching is not a freight assessment or an order book. It is the war risk premium on a Gulf transit. Steel is being ordered against a strait — but the strait is not priced by shipyards. It is priced by underwriters, and they can change their minds in a week.
LNG contract economics, LPG benchmarks and Pacific trade flows, VLGC freight, China PDH and NGL export infrastructure. Independent market intelligence, built only on sources you can check.