How the country outlooks are set

Each country page in Strature GPS shows an outlook for seven categories: geopolitics and conflict; trade, tariffs and sanctions; supply chains and chokepoints; energy and resources; technology and cyber; regulation and compliance; and finance and markets.

Finance and markets: the chance, in the next 12 months, of a central-bank, bond, currency, budget, debt or banking event that raises the cost of money for the state, banks or firms in this country, or cuts their access to it.

Each outlook looks 12 months ahead. It has one of four levels: Low, Elevated, High or Severe. Two lines say why, and each line links to the evidence behind it.

Below each outlook sits a 0–10 index built only from open public data. Higher always means more risk. The outlook is our judgement; the index is the data. They are set separately and can disagree. When they disagree, we say so.

What each level means

Each level says how likely a material business impact is in the next 12 months, and how hard it would be to absorb. “Material” means the impact crosses one of our published materiality lines. Examples: 2% of GDP, US$1bn a year at stake, or 10% of a key import need.

We publish these chance bands so that every outlook can be scored. When a registered Strature forecast covers the same event, the outlook cites its number.

A level can sit at Elevated or High on a lower chance if the impact would be very large and hard to reverse. When that happens, the reason says so.

How a level is set

Each outlook goes through the same six frameworks as every Strature brief, in the same order.

1. Systems theory. We map the channels through which the country's business exposure flows: exports, imports, energy, capital, chips, straits. We note the buffers, such as stockpiles, spare suppliers and other markets, and the delays before a shock arrives.
2. Game theory. We name the players who can change the picture within 12 months, such as governments, regulators, armed groups and large buyers. For each, we ask which moves are credible and which outcomes are stable.
3. Political psychology. We ask where leaders may act outside a rational-choice reading: domestic pressure, public commitments that are hard to walk back, and past patterns under stress.
4. Futures and scenarios. We write two to four scenarios for the next 12 months, each with signposts. The level reflects the weight of the plausible scenarios, not only the central one.
5. Statistics and probability. We set a number on each scenario from base rates, data and our registered forecasts. The 0–10 index is part of the data reviewed here.
6. Optimisation. We ask what a business should do across the scenarios: monitor, prepare, act or decide. The level is the one whose action best fits the evidence. If two levels fit, we choose the one that holds up best in the worse scenarios, and the reason says so.

An outlook draws on our country readings, our briefs, our registered forecasts and other cited sources. Every level is drafted from cited sources and checked line by line against those sources. The Principal approves each level before it is published. No level is published without at least one cited line behind it. Until a level is approved, the outlook reads “Not yet set”.

When a level is reviewed

Every outlook is reviewed each quarter. A review can confirm the level; the page then shows the new review date. We also review sooner when a trigger fires. Triggers include:

• a registered forecast that touches the country moves by 10 points or more, or resolves;
• a country reading in the same category changes band, or a new reading is published;
• a new tariff, sanction, export control or regulation that crosses a materiality line;
• an armed attack, coup, blockade or closure of a strait or major port that affects the country's trade;
• a major cyber attack on critical infrastructure, payments or a national supplier;
• a change in the country's status on an international list, such as the FATF lists;
• the index moving outside its published range;
• a material correction to a source we relied on.

An off-cycle review is finished within 10 working days of the trigger. While it is open, the outlook shows “Under review since” and the date.

How a change is logged and checked

Every level change is logged, and the log is public. Each entry records the date and the category; the old level and the new level; the two-line reason, with links to the evidence; the trigger, or “quarterly review”; and the date of approval. Each entry is sealed with a SHA-256 fingerprint on the day it is published, as our forecasts are. A later edit would show.

Twelve months after each level is set, we check it against what happened: did a material impact, as our materiality lines define it, occur in that category? We score each outlook with a Brier score, counting Low as 5%, Elevated as 20%, High as 45% and Severe as 75%: the middle of each chance band. The results go on the public track record, with the misses first. Once each level has enough outlooks behind it, we also show how often a material impact followed each level.

How the index is built

Each indicator is turned so that higher means more risk, then placed on 0–10 between fixed, published goalposts: values below the floor score 0 and values above the ceiling score 10, so one extreme country cannot squeeze the rest. Where a materiality line exists, the goalposts are tied to it. Goalposts and weights change only in an annual method review; when they change, the past five years are recomputed.

The category score is the weighted mean of its indicators, with the fixed weights below, rounded to one decimal place. A score shows when indicators carrying at least 60% of the weight have data; the weights are then rescaled over those present, and the score is marked “partial”. Below 60%, the page reads “Not enough open data”. An indicator counts as missing when it has no value or its latest value is more than three years old (five for resource rents). Nothing is filled in from regional averages or similar countries.

The range shows how sure the number is. It combines the source's own uncertainty, a weighting check that recomputes the score leaving out each indicator in turn, and a term for the share of weight missing.

An arrow shows only when the new score lies outside the range published 12 months earlier. Otherwise the page reads “no clear change”. On the first publication there is no earlier range, so there is no arrow.

An import counts as material from US$1bn or 2% of the country's imports, whichever is lower. The external threat counts only conflicts that target the country or sit near it, within 1,000 km of a target; operations a country runs abroad do not raise its score. Taiwan, Hong Kong, Bermuda and the Cayman Islands are scored only from sources that cover them. The EU gets no index; its member states do.

In finance and markets, a currency pegged to another moves little and can hide the risk of a break; reserves partly correct for that. Inflation also moves bond yields, so the two can count one strain twice. The euro members share one currency. No open data exist for credit default swaps or bond index spreads, and credit ratings do not feed the index.

The index is compared with the outlook only to flag a gap: 0 to under 2.5 sits in the Low band, 2.5 to under 5 Elevated, 5 to under 7.5 High and 7.5 to 10 Severe. A note shows when the two are two or more steps apart, or one step apart with the whole index range outside the outlook's band. The gap is never closed by moving one to match the other.

Indicators and weights

Geopolitics and conflict

Trade, tariffs and sanctions

Supply chains and chokepoints

Energy and resources

Technology and cyber

Regulation and compliance

Finance and markets

What the outlook is not

• It is not advice. It does not tell any reader to invest, divest, enter or leave a market. Your own decision needs your own facts and advisers.
• It is not a credit rating. It does not measure the chance that a government or company defaults on its debt. It is not issued by a credit rating agency.
• It is not a single-event forecast. Our register scores those. The outlook sums up a whole category for a whole country.
• It is not a travel or security advisory. For travel, read your government's official advice.
• It is not legal advice on sanctions or compliance. Check the official lists and take legal advice.
• The index is not the outlook. The index measures what open data show. The outlook is our forward judgement.

Sources for the index

Every source in the index is listed below, with its publisher and licence. We use only sources whose licence allows commercial reuse with credit, read on the publisher's own page.

Waiting for data

• US duties paid by country of origin (US Census international trade API) (US Census Bureau): waits for access to the US Census Bureau's trade data interface; until then the input counts as missing and is never estimated.
• Long-term interest rates (OECD Main Economic Indicators) (OECD): its licence is to be read on the publisher's own page (CC BY 4.0 under the OECD terms since 2024, not yet confirmed); until then the input counts as missing.
• Daily Treasury par yield curve rates (US Treasury): its reuse terms are to be read on the publisher's own page (a US federal work); until then the input counts as missing.
• Euro area long-term interest rates for convergence purposes (ECB Data Portal) (European Central Bank): commercial reuse is to be confirmed on the publisher's own page; until then the input counts as missing.
• Foreign exchange rates (Federal Reserve H.10) and the broad dollar index (Board of Governors of the Federal Reserve System): its reuse terms are to be read on the publisher's own page (a US federal work); until then the input counts as missing.
• Bilateral exchange rates (BIS) (Bank for International Settlements): commercial reuse is to be confirmed on the publisher's own page; until then the input counts as missing.
• General government gross debt (IMF World Economic Outlook) (International Monetary Fund): commercial reuse needs the IMF's written permission; until it is given the input counts as missing.
• IMF lending arrangements and members' financial data (International Monetary Fund): commercial reuse needs the IMF's written permission; until it is given the input counts as missing.

Considered and not used

• V-Dem dataset (V-Dem Institute): V-Dem dataset (V-Dem Institute): not used. Its licence (CC BY-SA 4.0) would require the scores we build from it to be shared on the same terms; we have not yet decided to do so..
• IMF data (trade, PortWatch) (International Monetary Fund): commercial reuse needs the IMF's written permission.
• UN Comtrade (UN Statistics Division): redistribution and commercial use need written permission.
• BACI international trade database (CEPII) (CEPII): dropped until CEPII confirms in writing that commercial publication is allowed.
• OFAC Specially Designated Nationals list (US Treasury): no reuse statement found on the publisher's own pages; the EU list is used alone.
• UK Sanctions List (UK Foreign, Commonwealth and Development Office): licence not confirmed for the list file; not built.
• UN Security Council consolidated list (United Nations): reuse limited to non-commercial use.
• Liner Shipping Connectivity Index (UNCTAD): licence not confirmed on the publisher's own page.
• Ember electricity data (Ember): open licence, not yet added to the index.
• ND-GAIN country index (University of Notre Dame): no licence named by the publisher.
• FATF lists, OECD FDI Restrictiveness Index, World Bank B-READY (FATF; OECD; World Bank): licences not confirmed on the publishers' own pages.
• SIPRI military expenditure; IEA data; Global Peace Index; ACLED; ITU Global Cybersecurity Index (various): not open for commercial reuse.
• Five-year sovereign credit default swaps (S&P Global (IHS Markit); LSEG; Bloomberg): no open source exists.
• EMBI sovereign spreads (J.P. Morgan): no open source exists.
• Credit ratings (S&P, Moody's, Fitch) (the rating agencies): not used in the index; a rating action may still be cited as a dated fact in an outlook reason.