When a Strait Closes, Barrels Reroute: The April Jump in U.S. Crude Exports

A price round trip, a volume staircase

Two series that describe the same disruption are telling different stories, and the gap between them is the most interesting thing in the second-quarter energy data.

The price series has already normalized. EIA monthly average WTI spot ran 64.51 in February 2026, jumped to 91.38 in March, peaked at 102.13 in May, and fell back to 80.46 in July. That is a spike and most of a round trip inside six months — the classic shape of a supply shock that markets decided was temporary.

The volume series did not round-trip. U.S. crude oil exports ran 4,043 thousand barrels per day in March 2026, then 5,593 in April and 5,728 in May. The March-to-April move is +38% in a single month. Against the January 2024 – March 2026 average of roughly 4,049 thousand b/d, May sits +41% higher. And this is not a series that normally moves that way: across those twenty-seven months, no single month printed above 4,600. The closest was 4,593 in February 2024. Two consecutive months roughly a thousand barrels per day above that prior ceiling is not noise in a series with this history.

The disruption behind both series is documented. Hormuz transit volumes in the second quarter of 2026 fell to roughly 4.9 million barrels per day, down from about 21.6 million b/d in the fourth quarter of 2025 — a decline EIA has described as effective closure of the strait.

The framing that follows from this is simple to state and hard to resolve. Closing a chokepoint does not destroy barrels. It changes who loads them, from where, and over what distance. Price measures the panic. Volume measures the rerouting. Panic reverses in weeks; rerouting sometimes does not reverse at all. The question this piece is built around is whether the April–May export step is the visible edge of a durable rerouting, or the temporary bulge of a market clearing an emergency — and, more usefully, what would have to be observed in the next few monthly prints to tell those apart.

How a closed strait becomes an export statistic

The transmission chain from a blocked waterway to a monthly trade figure has more steps than the headline version suggests, and each step carries its own lag. That layered lag is the reason price and volume can point in opposite directions for a quarter without either being wrong.

Step one: physical shortfall at destination, not at source. A chokepoint closure does not reduce global production on day one. Producers upstream of the blockage keep pumping into constrained storage; consumers downstream of it face a delivery gap. The shortage is geographic before it is material.

Step two: price does the allocation. Buyers who cannot lift a Gulf cargo bid for whatever else clears customs, refinery configuration, and freight math. The March-through-May price path is that bidding, and it is fast because bidding is fast.

Step three: cargo re-nomination. Turning a bid into a barrel means finding an alternative grade a refinery can actually run, securing a vessel, and slotting a loading window. Spot substitution is quicker but competes for the same terminal capacity.

Step four: the voyage itself. Replacement barrels from the Atlantic Basin travel farther to reach the same refineries. Longer voyages consume more tanker capacity per barrel delivered, which tightens freight and stretches the interval between a decision and a delivery.

Step five: the statistical print. Only after loading does the barrel appear in monthly export data — and monthly petroleum data arrives with roughly a two-month reporting lag and is subject to revision.

Stack those steps and a rough expectation follows: a closure that bites in the second quarter should show up in price almost immediately, in volume with a lag of one to two months, and in the settled, revised statistics later still. That is precisely the pattern in the data. The price move led in March. The volume move landed in April.

The chart below plots the monthly export series through May 2026 against the two-year run that preceded it, which is the clearest way to see that the April step is a change in level rather than a change in slope.

Two features of that chart matter more than the peak itself. First, the pre-2026 band is narrow: months cluster between roughly 3,750 and 4,600, and the dispersion inside that band looks like seasonality and maintenance scheduling, not regime change. Second, the April–May pair does not sit at the top of the band — it sits above it, with daylight in between. A single month outside a two-year range can be a scheduling accident. Two consecutive months, the second higher than the first, is a harder thing to dismiss.

What comparable episodes did to volumes

Maritime chokepoint disruptions are rare enough to feel unprecedented each time and frequent enough to have a shared shape. Three broad precedents are worth holding in mind, stated at the level of mechanism rather than of specific figures.

Suez, closed in 1967 and reopened in 1975. The immediate effect was a freight shock. The lasting effect was structural: with the canal unavailable, the economics of routing crude around the Cape of Good Hope favored much larger vessels, and the tanker fleet was rebuilt around that geometry. When the canal reopened, the fleet did not un-build itself. The lesson is that a chokepoint closure long enough to justify capital commitments leaves behind assets and contracts that outlive the closure.

The 2022 reorganization of European crude sourcing. Sanctions and self-sanctioning did not reduce global supply so much as rearrange it — the same barrels reaching different buyers over longer distances. Prices spiked, then substantially retraced. Trade routes did not retrace. Voyage distances stayed longer, and the relationships built during the disruption largely persisted afterward.

The Red Sea diversions beginning in late 2023. Vessels rerouting around southern Africa added weeks to voyages. Freight repriced quickly; the routing choices proved stickier than the headlines that caused them, because re-planning a schedule twice carries its own costs.

The common pattern across all three is the same asymmetry now visible in the 2026 data: price is a fast, mean-reverting variable; routing is a slow, hysteretic one. Once a refinery has qualified an alternative grade, once a term contract has been signed, once a vessel has been chartered onto a longer run, the cost of switching back is real even when the original supply returns. That asymmetry is the reason a volume staircase is worth more analytical attention than a price spike — not because volume predicts price, but because volume records commitments while price records opinions.

The precedents also carry a warning. In each case, part of the initial volume move was inventory behavior rather than structural substitution, and that part did unwind. Distinguishing the durable portion from the pulse required several months of data, not one or two.

Three readings of the same two months

There are at least three internally consistent explanations for a 38% one-month jump followed by a second high month, and the honest position is that two data points cannot separate them.

Reading one — structural substitution. Buyers who lost Gulf access signed replacement supply, some of it on term. If that is what happened, the new level reflects contracts rather than opportunism, and it should persist while those contracts run, largely independent of where the flat price goes.

Reading two — a restocking pulse. A closure of a major artery is exactly the event that makes importers rebuild safety stock. Restocking is a one-time flow: it raises volumes sharply for a few months and then stops, because a filled tank does not need filling twice. Under this reading, the April–May level is real but temporary, and the decay should already be underway.

Reading three — timing and measurement. Monthly export data reflects loading dates, and loading programs slip. Cargoes scheduled for late March that sailed in early April would inflate April at March's expense. The two-month March–April average of about 4,818 thousand b/d is elevated but far less dramatic than 5,593 alone. This reading is the weakest of the three, because it explains April by borrowing from March and then has nothing left with which to explain May — unless April cargoes also slipped forward, which begins to require a chain of coincidences.

Each reading implies a different future, and those futures are distinguishable in data that will arrive within a few months.

Note what the table does not contain: a probability weight on each column. Assigning one now would be false precision. The timing explanation is already strained by May; the other two remain live until at least three post-May prints are in hand.

Conditions under which the export step says nothing about Hormuz

The frame in this piece — volume as the durable record of rerouting — fails in several identifiable situations, and each of them is worth checking before leaning on the interpretation.

If domestic supply-side factors moved independently. Export volumes are a residual: production minus domestic refinery runs minus stock change. A heavy refinery maintenance season, an unplanned outage at a large domestic refinery, or a deliberate inventory draw can push exports up with no reference to any strait. If the April–May step coincided with unusually low domestic crude runs, the chokepoint explanation loses most of its force.

If the arbitrage window did the work by itself. When the domestic benchmark cheapens relative to waterborne benchmarks, exports rise mechanically, whatever the cause of the spread. In that case exports are tracking a price relationship rather than a routing decision, and they will reverse when the spread does — which makes them a coincident indicator, not a structural one.

If the data revises materially. Monthly petroleum statistics are estimates first and settled numbers later. A revision of a few hundred thousand barrels per day would not erase a 1,550 thousand b/d jump, but it would change how sharp the step looks and could shift the March–April boundary enough to matter for the timing reading.

If the strait reopens quickly and fully. Hysteresis in trade routing is a tendency, not a law. Short disruptions — measured in weeks rather than quarters — often do reverse cleanly, because no one has had time to sign anything durable. The structural reading depends on the closure having lasted long enough for commitments to form.

If exports rose while imports rose too. A simultaneous increase on both sides points toward grade-matching rather than net redirection of supply. Net trade, not gross exports, carries the rerouting signal.

What to Watch Next Week

• The weekly petroleum status report. EIA publishes it each Wednesday. The relevant line is the four-week average of crude exports, which is noisier than the monthly series but arrives with days of lag rather than months. A four-week average holding near or above 5,000 thousand b/d supports the structural reading; one drifting toward the low 4,000s supports the pulse.
• The next monthly petroleum supply release. This is where the June figure and any revision to April and May appear. Watch the revision direction as closely as the new print — an upward revision to March would be the first real evidence for the timing explanation.
• Domestic refinery utilization. Published weekly alongside exports. If runs were unusually low through April and May, part of the export step is a domestic story rather than an international one.
• Crude inventories at the Gulf Coast. A step in exports funded by drawing stocks behaves differently from one funded by production, and only the second is sustainable at the new level.
• Any update to Hormuz transit estimates. The Q2 figure of roughly 4.9 million b/d is the baseline. A move back toward double digits would begin the test of whether rerouting reverses.
• Tanker freight indicators for long-haul crude routes. Freight is the cleanest available proxy for voyage distance. Elevated long-haul rates persisting while flat price falls is the signature of routing change without scarcity.

Concrete Framework

A monitoring checklist that resolves within roughly one quarter, with thresholds set in advance so the conclusion is not written after the fact.

1. Set the reference line at 4,600 thousand b/d. That is the ceiling of the twenty-seven-month pre-shock range. Any monthly print above it is outside historical precedent; any print below it is inside normal variation.
2. Count consecutive months above 5,000. Two months is where the count stands now. Four or more consecutive months materially favors the substitution reading and effectively rules out the pulse.
3. Treat two consecutive prints below 4,600 as the end of the step. If June and July both land under that line, the episode was transitory regardless of how it looked in April.
4. Treat 4,600 to 5,000 as unresolved. Resist the urge to interpret this band. It is consistent with a decaying pulse and with a partial structural shift simultaneously.
5. Track the price-volume divergence explicitly. Record flat price and export volume in the same table each month. Volume elevated while price returns to pre-shock levels is the specific pattern this piece is about; volume and price falling together is an ordinary demand story instead.
6. Log every revision. Note the first-print value and each subsequent revision for March through June. If cumulative revisions exceed roughly 200 thousand b/d on either side of the March–April boundary, re-run the timing test before drawing conclusions.
7. Check the residual identity before concluding. Exports equal production minus refinery runs minus stock change, approximately. Confirm which of those three terms moved before attributing the step to a strait on the other side of the world.
8. Set a review date rather than a target. Three post-May monthly prints is the minimum sample. Before that, the honest answer to what the April step means is that it is not yet knowable — and saying so is more useful than picking a column of the table early.

What the data supports today is a narrow claim and a wide uncertainty. The narrow claim: U.S. crude exports in April and May 2026 stepped clearly outside a two-year range, at the same time a major maritime chokepoint was operating at roughly a quarter of its prior throughput, and that step did not reverse when the price spike did. The wide uncertainty: whether that step is a contract or a refill. Those two things look identical for about ninety days and then stop looking identical entirely.