SciRisk Calculate your ESCRC

Insights · Scenario analysis

The chokepoint illusion: what a Strait of Hormuz stress test actually tells you

Run against a hundred companies, the scenario ranked semiconductors and East Asian manufacturing far above the two oil majors whose real business is the commodity in question. Both results are correct, for reasons worth understanding before you present either one.

Engine ESCRC v5.0Paths 100,000Confidence 95%Horizon 365dScope All 100 SciRisk 100 companies

Run a geopolitical stress scenario against a hundred companies and you get a ranking that looks like insight: semiconductors and East Asian manufacturing hit hardest, energy and food barely touched. Read past the ranking and two of the companies at the very bottom of it are the two whose actual business is the commodity the scenario is about.

That contradiction isn't a data error. It's the most useful thing the exercise produced.

What the scenario actually does

The ESCRC engine has no concept of straits, oil, or Iran. It's a general-purpose supplier-disruption simulator: give it a network of suppliers, each with a disruption rate, a recovery-time distribution and a regional correlation parameter, and it runs 100,000 simulated years to produce a Value-at-Risk figure — the loss capital a company should hold against supply-chain disruption.

Geopolitical scenarios sit in a thin layer on top: a template that finds every supplier headquartered in a named list of countries and multiplies its disruption rate and recovery time by fixed factors. The shipped “Strait of Hormuz closure” template flags nine countries: the Gulf exporters whose oil and LNG transit the strait, plus Taiwan, South Korea and Japan, the energy-import-dependent manufacturing hubs downstream of it. Matched suppliers get their disruption rate raised 60% and recovery time extended 30%; the whole network's correlation assumption rises alongside them.

What came out, across all 100 companies

We ran the template against every SciRisk 100 network, unmodified, exactly as clicking it in the calculator's Network panel would.

Mean change in Value-at-Risk under the Hormuz scenario By sector, across all matching SciRisk 100 companies. Semiconductors +56.1% Technology Hardware +49.7% Automotive +33.0% Industrial Machinery +17.1% Chemicals +11.9% Consumer Goods +7.9% Retail & Distribution +7.6% Aerospace & Defence +6.1% Pharma & Biotech +4.4% Healthcare Equipment +1.4% Energy & Utilities +1.0% Food & Beverage +0.1% 0%+20% +40%+60%
Mean %(Delta)VaR by sector across all matching companies. Semiconductors and technology hardware lead because their networks are structurally concentrated in Taiwan, Korea and Japan.

Fifty-three of the hundred companies had at least one matched supplier. At the top: Honda Motor (+105%), Komatsu (+105%), Panasonic (+102%), Toyota (+84%) — all Japanese manufacturers whose parts suppliers cluster in Japan. At the bottom of the whole list: Chevron (+3.0%) and ExxonMobil (+2.6%), the two companies whose real-world business is Gulf oil.

The scenario's own ranking, inverted from reality Change in ESCRC risk capital under the Hormuz closure scenario. Honda Motor parts supply is Japan-based; sells no oil +105.2% ExxonMobil produces and sells the oil that transits Hormuz +2.6% 0%+40%+80%+120% The model prices supplier disruption, not the output-price windfall an oil major would actually see. It cannot see the channel that matters most for one of these two.
The scenario's magnitude ranking and the real economic-exposure ranking point in opposite directions for these two names.

Why the ranking is right and the size of it isn't

Three mechanisms are doing the work, and they don't point the same way.

Real concentration, correctly surfaced. Semiconductors and electronics top the list because those supply chains genuinely cluster in Taiwan, Korea and Japan — TSMC, Samsung, SK Hynix, Foxconn sit inside dozens of these networks. That's a structural fact the template exposes; nothing wrong with it.

Proximity mistaken for exposure. The same mechanism can't tell a components vendor with no energy dependency from a supplier whose costs actually move with Gulf gas prices — both get the identical multiplier for sharing a country. Honda's suppliers (Denso, Aisin, Nippon Steel, Bridgestone) are precision manufacturing and steel, not energy infrastructure; a closure would raise their costs at the margin, not double their odds of a major disruption. That collapse of “located near” into “exposed to” is what inflates Honda and Komatsu into triple digits.

A portfolio-wide dial, not a local one. The scenario's correlation increase applies to the whole network, not just the matched suppliers. Companies with zero Gulf, Taiwanese, Korean or Japanese suppliers — Shell, BP, TotalEnergies among them — still saw their loss estimates move, purely from this side effect.

Model assumption

ESCRC prices input disruption: the risk that a company's own suppliers fail to deliver. It has no representation of output price. A Hormuz closure would, in reality, likely be a revenue tailwind for companies that sell the commodity whose supply just shrank — removing roughly a fifth of global seaborne oil and LNG from the market moves benchmark prices up, and Chevron and ExxonMobil capture that almost directly.

Their showing near the bottom of this ranking isn't the model missing their exposure. It's the model correctly reporting that their component suppliers (oilfield-services firms, steel mills) are barely Gulf-sourced — while being structurally blind to the one channel, their own selling price, through which the scenario would move their numbers the most, and in the opposite direction from everyone else on the list.

What this means for how you read a scenario run

None of this is an argument against running scenarios. It's an argument for reading one the way an underwriter reads an actuarial table: a structured way to direct attention, not a number to repeat without translation.

  • Use the ranking, discount the size. The relative ordering — electronics and East Asian manufacturing more exposed than energy or food — reflects a real, auditable fact about supplier geography. The specific percentages don't survive contact with the template's own documented assumptions.
  • When a result contradicts domain intuition, ask what the model can't see. The single most useful output here wasn't the ranking, it was noticing that the two companies most exposed to Hormuz in reality came out least affected in the model, for a legible reason: the model only instruments one side of the ledger.
  • Treat a country match as a hypothesis, not a finding. “Headquartered in Japan” stands in for data that's expensive to collect — energy intensity, contract indexation, on-site generation. Fine for a first screen; not fine to present as a conclusion for the handful of names that actually matter to your portfolio.
  • Ask which part of the number is exposure and which part is a portfolio-wide assumption. A correlation bump that moves every company's figure, matched or not, will eventually teach users to distrust the whole output — including the parts that are well founded.

The takeaway

A chokepoint scenario like this is a lens on supplier-side disruption risk, not a forecast of financial impact. The companies it flags as most affected and the companies that would actually move the most money in a real closure can be, and here are, two different lists. Reading only the first list is the actual risk.

Run a chokepoint scenario on your own network

The Strait of Hormuz template, and three others, are one click away in the calculator's Network panel. Build the network you actually have and see which of your own suppliers the model matches — and which of your real exposures it can't.

Published 2026-09-19. Figures come from the production ESCRC engine on the worked example stated above, not from the SciRisk 100 dataset; the day-by-day comparison was built independently of the engine. For how the index is produced see the methodology, and for what it does and doesn't capture see research & disclosures.

SciRisk ESCRC is an analytical estimate. It is not a credit rating, investment recommendation or statement of financial condition.