Country profile methodology

Each /country/[iso3] page is a composition, not a new dataset. It joins layers FinWeave already builds and documents elsewhere, one canonical URL per country, so a reader or a link can reach a country's full financial profile in one place. No page here introduces a number that does not exist in a source parquet; every figure is reproducible from the section's cited source.

What is composed

The page draws on seven existing layers, each with its own module and methodology:

Which countries get a page

generateStaticParams returns the union of every ISO3 that appears in any layer: the crisis atlas, sovereign-debt stats, GFDD, and the four BIS files. That union is 242 economies as of the 2026-07-10 build. dynamicParams = false, so a code outside the union returns 404 rather than rendering an empty page. Country names resolve, in priority order, from sovereign_stats, then the crisis chronologies, then the BIS area_name; the single GFDD-only code with no country-level name (CHI, the Channel Islands) is completed from the World Bank economy list.

The layers added on 2026-07-10 (Findex financial inclusion, BIS central-bank policy rate, BIS real broad effective exchange rate, and the Penn World Table economic-scale section) introduce no new country pages: every ISO3 in findex (162), bis_cbpol (48), bis_eer real-broad (63), and pwt (185) already appears in the pre-existing union (verified 2026-07-10, zero codes outside it in any of the four), so the union query and generateStaticParams are unchanged.

The external-accounts layers added on 2026-07-11 (IMF balance-of-payments and IIP, World Bank reserves, the Chinn-Ito openness index, and remittance inflows) likewise add no country pages: the /country union is deliberately fixed by the risk layers (crises, sovereign debt, GFDD, BIS), and context or external series never expand it. All but two of their codes are already in the union; the two that are not (CWX, the IMF Curacao-Sint Maarten monetary-union aggregate in imf_bop_iip, and ZAR, the legacy Zaire code Chinn-Ito uses for the Democratic Republic of the Congo in kaopen) fall outside every risk layer and therefore have no page, so their external data is simply not surfaced. The generation query (generateStaticParams) is unchanged.

A section renders only where that country has data. An absent section is stated in words ("not in the BIS reporting set", "no GFDD series"); the page never draws an empty chart or implies a zero where the source is simply silent.

Per-source coverage and staleness (read this before comparing across sections)

The layers stop at different dates. A flat or missing recent tail in one section is almost always the source ending, not the underlying variable standing still.

• GFDD (financial depth and soundness). World Bank Global Financial Development Database, September 2022 release. 215 economies, 1960 through 2021. GFDD values end in 2021; this is a documented, structural feature of the dataset, not a FinWeave truncation. Indicators shown: private credit by deposit money banks to GDP (GFDD.DI.01), bank deposits to GDP (GFDD.OI.02), bank Z-score (GFDD.SI.01), bank nonperforming loans to gross loans (GFDD.SI.02), and bank concentration, top-3-bank asset share (GFDD.OI.01). Codes and names are verbatim from GFDDSeries.csv.
• IMF FSI (current soundness). IMF Financial Soundness Indicators, core set, deposit takers, dataflow IMF.STA:FSIC(13.0.1), retrieved 2026-07-10 from api.imf.org (the SDMX backend of data.imf.org). This section exists specifically because GFDD's soundness series stop in 2021: the FSI panel is the current reading, annual through 2024 or 2025 depending on the country. 153 reporting countries, a subset of the 242-country union (it adds no new country page). Six best-covered core indicators shown: regulatory capital and Tier 1 capital to risk-weighted assets (FSI688, FSI626), nonperforming loans to gross loans (AQ12), return on assets and equity (ROA, ROE), and liquid assets to total assets (FSI283). Coverage is genuinely uneven across indicators and years; where an indicator has no observation for a country the panel says so rather than drawing a zero. Some countries report no FSI at all (for example Japan reports zero rows in this dataflow, a real IMF-side gap, not a FinWeave omission); those countries simply do not show this section. Codes and names are verbatim from the IMF.STA:FSIC codelist. Values are percent. The header's "Nonperforming loans (current)" fact is this dataset's NPL ratio (AQ12), shown only where reported.
• Findex (financial inclusion). World Bank Global Findex Database 2025 (5th edition), CC BY 4.0, 162 economies. Financial inclusion is measured in survey waves (2011, 2014, 2017, 2021, 2024; the 2024 wave is new to the 2025 edition), not annually, so each of the four headline series is drawn as a wave sparkline with the latest value and the change (in percentage points) since the earliest wave available for that country. The four indicators, codes verbatim from FINDEXSeries.csv: account ownership (account.t.d), made or received a digital payment (g20.any), saved at a bank or similar financial institution (fin17a), and borrowed from a formal bank or similar financial institution (fin22a); all are percent of adults age 15+. Not every wave reports every indicator for every economy (e.g. Bangladesh's formal-borrowing series begins in the 2014 wave), so a card's baseline year is the first wave present, not necessarily 2011. Countries not surveyed by Findex simply do not show this section.
• BIS (credit cycle). BIS credit and property statistics cover a fixed reporting set of roughly 44 advanced and major emerging economies, not the whole world. Countries outside it (for example Bangladesh) have no BIS credit-gap, debt-service, or property series and the section says so explicitly. The four series and their exact BIS selections:
  • Credit-to-GDP gap: bis_credit_gap, gap_dtype = 'Credit-to-GDP gaps (actual-trend)', private non-financial sector, all lenders. Actual credit-to-GDP minus its one-sided Hodrick-Prescott trend, the Basel III early-warning measure.
  • Private non-financial credit: bis_total_credit, borrowers = private non-financial sector, lenders = all sectors, unit = percentage of GDP, market valuation, adjusted for breaks.
  • Debt-service ratio: bis_dsr, borrowers = private non-financial sector, in per cent of income.
  • Real residential property: bis_property_prices, real index, 2010 = 100.
  • Each quarterly series is annualized to one reading per year by taking the latest available quarter in the year (year-end where a fourth-quarter value exists). BIS is the most current layer on the page: credit series run through 2025, and residential property prices reach 2026Q1 for economies with the freshest reporting (e.g. USA, GRC).
• BIS policy rate and effective exchange rate. Two more BIS series, wider than the credit set but still not the whole world. Central-bank policy rate (bis_cbpol, WS_CBPOL, "per cent per year"): the harmonized policy rate for 48 economies plus the euro area, more than the ~44-economy credit set. The headline figure is the latest value from the daily series (the most current reading; e.g. USA 3.625% on 2026-07-07, the FOMC target-range midpoint), and the sparkline is the last 10 years of the monthly series. Countries outside the 48 (for example Bangladesh, which has no BIS policy-rate series despite Bangladesh Bank setting a repo rate) show an explicit absence. Real broad effective exchange rate (bis_eer, WS_EER, real, broad 64-economy basket, index 2020 = 100): 63 economies; latest monthly value and a 10-year monthly sparkline, where a rise is real appreciation. Only the real broad series is shown; a country with nominal or narrow-basket EER but no real-broad series shows the EER absence. Both series are current into 2026 for the freshest reporters. These add no country pages (all 48 / 63 codes are already in the union).
• External accounts. Four series, none of them universal. Current-account balance (imf_bop_iip, indicator CAB_PCT_GDP): IMF Balance of Payments, percent of GDP, annual through 2025. Net international investment position (imf_bop_iip, indicator NIIP_USD): IMF IIP, shown as a US-dollar level (converted to billions), through 2025. The IMF IIP vintage carries NIIP in US dollars only, not as a share of GDP, so the level is shown rather than a ratio; no GDP denominator is introduced (a positive value is a net creditor position, a negative one a net debtor position). Total reserves (reserves, World Bank series FI.RES.TOTL.CD, "Total reserves includes gold, current US$"): a US-dollar level in billions, annual through 2025. Two early-year negative values (Chad, 1968-1969) are a WDI net-reserve-position artifact of the raw source, kept as-is. The current-account and net-IIP panels carry the same crisis-window shading as the credit-cycle section. Each panel renders only where the IMF or World Bank series has an observation for the country.
• Capital-account openness. kaopen, the Chinn-Ito index (Chinn and Ito 2006), normalized 0 (fully closed) to 1 (fully open), annual 1970-2023. Drawn as a sparkline where the country is in the Chinn-Ito sample. License: research use with citation only (not CC or commercial); this is the one external-accounts series that is not open-licensed, and it is surfaced as an aggregate display series consistent with the page's existing non-commercial crisis-atlas encumbrance. The USA reads 1.0 (fully open) and Bangladesh about 0.16 (relatively closed) in the latest (2023) year.
• Remittance inflows (context). remittances_flows, flow_type='inflow', World Bank / KNOMAD, US-dollar level in billions, annual through 2024. Shown as a sparkline and framed explicitly as external-flow context (household income received from abroad), not a FinWeave risk score, per the acquisition scope rule that admits remittances only as a context series. Bangladesh's 2024 inflow is about $27.5 billion; India's, the world's largest, about $137.7 billion.
• Crisis history. The five-chronology crisis atlas (Global Macro Database 2026_06, Reinhart-Rogoff via HBS BFFS, Laeven-Valencia 2020 and 2026, JST Macrohistory, ECB/ESRB). Distinct crisis years are merged into episodes bridging gaps of up to two years, the same rule the /crises pages use, so an episode link here targets a real /crises/[iso3]/[year] page. Crisis flags end around 2016 to 2021 depending on the source.
• Sovereign debt. sovereign_stats, the headline debt-to-GDP with its year and perimeter (IMF General Debt Database or, in fallback, WEO gross debt), the historical peak, the sovereign-crisis count, and the country's position relative to the 2000+ crisis-start median. IMF debt runs to 2024. The elevation comparison is descriptive, not a prediction; see the sovereign methodology for the debt-intolerance evidence.
• Economic scale (Penn World Table 11.0). pwt, CC BY 4.0, 185 economies, 1950-2023, a fixed balanced country-year grid. Population (pop, millions); real GDP per capita at chained PPPs (rgdpo / pop, 2021 US$, both source columns already in millions so the ratio needs no further unit conversion); and total-factor productivity (rtfpna, an index normalized to each country's own 2021 = 1, a within-country growth-accounting series, not a cross-country level comparison; ctfp, the USA=1 cross-country level series, is not shown). This section is a denominator reference for the rest of the page, not a financial-risk series. 185 of the 242 union countries are covered; the 57 outside PWT are mostly small territories and micro-states never admitted to its country set.

United States

The US is the one country with FinWeave's bank-level, systemic-risk, and financial-conditions modules built out (FDIC call reports, market returns, FRED). The /country/USA page links to /banks, /systemic, and /conditions rather than duplicating them, keeping each number in one canonical place.

International debt securities (BIS IDS, added 2026-07-11)

Every /country/[iso3] page with BIS coverage now carries an "International debt securities" section from the BIS debt securities statistics (bis_debt_sec.parquet, view bis_debt_sec; countryDebtSecurities in src/lib/debtSecurities.ts, DebtSecuritiesChart component). It sits after the external accounts and before the sovereign-debt section.

• What it is: the amount outstanding of international debt securities (IDS) issued by a country's residents, by issuer sector, quarterly. IDS are bonds and notes placed outside the issuer's home market (a residence-basis, cross-border measure). The four issuer sectors, verbatim from the BIS sector codes: all issuers (code 1, the total), general government (2), financial corporations (B), and non-financial corporations (J). "All issuers" is the total; the other three are its components, each shown only where the BIS reports it.
• International-only caveat (read this first): this is not a country's total debt securities outstanding. Domestic-market issuance is excluded, so the figures are far smaller than the whole bond market and much smaller than the government debt-to-GDP ratios in the sovereign section. The clearest illustration is the United States: US general-government international debt securities are only about $2.5 billion (2026-Q1), because the US Treasury issues in its own deep domestic market, not offshore, while the same government's total debt runs into the tens of trillions of dollars. The two are different universes and the page never implies otherwise. The section copy and the source line both state "international issues only (residence basis); domestic issuance excluded".
• Units: source is US$ million (outstanding_usd_mn); the page divides by 1000 and displays US$ billions ($T above 1000 bn). Never a percent, never a share of GDP.
• Coverage: 158 economies report to bis_debt_sec (a subset of the 242-country union, so it adds no country page). NULL-iso3 rows are BIS regional and aggregate groupings (euro area, "all countries", and similar); they are excluded by construction (iso3 = ? plus length(iso3) = 3) and never leak onto a country page. Quarterly series run from as early as 1962-Q4 to 2026-Q1 for the freshest reporters. Each quarterly value is plotted directly (no annualization); the multi-line chart thins its x-axis to whole-year ticks so a 60-year quarterly axis does not collide. Countries outside the 158 (many small economies) simply do not show the section.
• License: BIS statistics, free to use with a "Source: BIS" attribution (same terms as the credit-cycle, policy-rate and EER sections).
• Citation: Bank for International Settlements, Debt securities statistics (international debt securities). https://data.bis.org/topics/DSS
• Anchors (checked against the parquet 2026-07-11): USA all-issuers 2026-Q1 = $2,986.8 bn, of which general government only ~$2.5 bn; United Kingdom all-issuers $3,989.9 bn; Brazil all-issuers $104.7 bn (general government $62.7 bn, the largest component, an emerging-market pattern opposite to the US).

Licenses

The composition inherits each source's license, and a combined page is bound by the most restrictive input it displays.

• GFDD: CC BY 4.0 (World Bank).
• Findex: CC BY 4.0 (World Bank Global Findex Database 2025).
• Penn World Table 11.0: CC BY 4.0 (Feenstra, Inklaar and Timmer, 2015, American Economic Review 105(10): 3150-3182).
• BIS: free to use with a "Source: BIS" attribution (covers the credit-cycle, policy-rate bis_cbpol, effective-exchange-rate bis_eer, and international-debt-securities bis_debt_sec sections).
• IMF balance of payments and IIP (imf_bop_iip, current account and net IIP): free with a "Source: IMF" attribution.
• World Bank total reserves (reserves, FI.RES.TOTL.CD): CC BY 4.0.
• Chinn-Ito capital-account openness (kaopen): research use with citation only; not CC or commercial. This is the most restrictive external-accounts input and, like the crisis atlas, binds the full page to non-commercial research use.
• World Bank / KNOMAD remittances (remittances_flows): CC BY 4.0.
• IMF sovereign debt: free with a "Source: IMF" attribution.
• IMF Financial Soundness Indicators: free with a "Source: IMF, Financial Soundness Indicators" attribution.
• Crisis atlas: non-commercial research only (inherits the Global Macro Database and JST licenses); cite all five chronologies when publishing.

Because the crisis section carries the non-commercial GMD/JST encumbrance, treat a full country page that includes crisis history as non-commercial research. The GFDD, BIS, and IMF sections on their own are attribution-only.