BIS locational banking statistics: the download that answers who owes whom across borders

One fully-keyed series; the quarterly values run out to its right. Two real rows, exactly as they arrive:

freq          : Q
measure       : S  (amounts outstanding)
position      : C  (total claims)
instrument    : G  (loans and deposits)
currency      : JPY (yen)
reporting cty : 5A (all reporting countries)
cp sector     : N  (non-banks, total)
cp country    : YE (Yemen)
quarters      : 1977-Q4 ... 2026-Q1   # one value column per quarter

freq          : Q
measure       : F  (FX and break adjusted change)
position      : C  (total claims)
reporting cty : BR (Brazil)
currency type : F  (foreign currency)
position type : R  (local lending by foreign-owned offices)
1977-Q4 value : 1643.983

Codes arrive paired with labels - FREQ carries both Q and Quarterly, measure both S and 'amounts outstanding' - so joins happen on codes while humans read labels. Swap the counterparty code and the same key returns China, Switzerland or any offshore centre; swap the measure and stocks become flows with the valuation noise removed.

The deliberate absence informs as much as the presence: no institution identifier appears anywhere. Rows describe a reporting country's banking sector in aggregate, which is precisely what lets the collection publish near-census coverage without waiting on entity-level disclosure cycles. The leverage lives in the counterparty detail instead - more than 200 countries by sector and instrument.

Fifteen documented fields form the verified spine, captured row-by-row during the August 2026 research pass - nothing inferred from column names alone. Eight address a banking stock before any value appears:

• L_MEASURE separates outstanding stocks (S) from FX- and break-adjusted changes (F). Pick the measure before the analysis, not after.
• L_POSITION chooses the side of the balance sheet: total claims against total liabilities.
• L_INSTR splits loans and deposits from debt securities and the all-instruments roll-up.
• L_DENOM and L_CURR_TYPE say which currency the position is booked in and whether it is foreign-currency business relative to the reporting country - the two cuts behind most global-dollar-credit work.
• L_PARENT_CTY and L_REP_BANK_TYPE carry bank nationality and institution class, the funding-chain axes headline aggregates hide.
• L_POS_TYPE distinguishes cross-border positions from local lending by foreign-owned offices - the book most risk registers miss because they count only cross-border claims.

Three more place and value the observation: L_REP_CTY names whose banks hold the claim, while L_CP_SECTOR and L_CP_COUNTRY say against whom. Aggregates sit beside individual codes - 5A all reporting countries, 5C/XM euro area, 5J all counterparty countries - so global totals are a filter rather than a merge.

The indicator groups that share the wider BIS catalog fold under additional fields on request: consolidated-basis netting for ultimate-risk views, issuer residency and maturity buckets for international debt securities, derivative asset-class breakdowns, credit-to-GDP gap components, property-price index bases. Name the ones your workflow touches and they arrive as populated columns.

Temporal: quarterly observations from 1977-Q4 through 2026-Q1, 195 quarters per series. Long enough to hold Latin America's lost decade, the Asian crisis, the euro area's sovereign stress and the post-2022 retrenchment of offshore lending inside one continuous panel - which is why backtests of funding-stress and carry strategies keep returning to it rather than to shorter commercial panels.

Geography: about 50 reporting countries across Europe, the Americas, Asia-Pacific and the offshore centres, against counterparties in more than 200 countries, with aggregates beside the individual codes.

Granularity: reporting country x counterparty country x counterparty sector x instrument x currency x position type x measure, per quarter, in millions of US dollars or growth ratios where the measure calls for them.

Two seams deserve planning rather than discovery. First, Russian public-authority data stopped flowing after February 28, 2022 - those series end there rather than trailing off gradually, so recent prints carry a hole that is a collection fact, not a revision. Second, officially compiled figures get revised between releases: provenance per pull belongs beside the values themselves, not in a folder name, and point-in-time research wants each vintage archived as it lands. Both are handled upstream of your warehouse when delivery runs through Datadory; the breaks are flagged before they corrupt a backtest.

Global-liquidity modeling. Dollar- and yen-denominated cross-border credit to non-bank borrowers is the canonical gauge of the global financial cycle. Measure F strips the valuation swings that make raw stocks lie about momentum, so credit booms read as lending rather than as exchange-rate artifacts.

Country-risk screening. Counterparty-level claims by sector let credit committees size what a given sovereign's banking system actually owes abroad before the headline number gets politicized - China as a counterparty code is one filter, not a modeling exercise.

Offshore-centre exposure mapping. Position type R isolates local lending by foreign-owned offices, the exposure most risk registers omit because they count only cross-border claims.

Funding-stress and carry backtests. Five decades of quarterly claims and liabilities give strategies multiple regimes to prove themselves against - not just the last cycle.

Citable macro framing. The series names its compiler, method and residence basis, which is why central banks, academics and financial journalists cite it directly. Worked examples of the citation path appear in citation-grade research; the quant applications extend naturally into quant backtesting.

The BIS ships two views of the same international balance sheet, and picking the wrong one produces plausible-looking nonsense. The locational statistics are unconsolidated by design: positions include intragroup funding between offices of the same banking group, which makes them the right lens for funding-flow work - where money moved, in which currency, booked where. The consolidated statistics re-cut the picture on an ultimate-risk basis, netting intrabank flows out to show where risk actually comes to rest once guarantees and branches are accounted for. Nationality-based questions belong to the consolidated view; residence-based ones to the locational.

A second routing decision sits closer to home. For US bank health - charge-offs, deposit mixes, domestic balance-sheet detail - the FDIC's Quarterly Banking Profile owns the question and the locational panel cannot answer it, because it stops describing US-resident banks' foreign business only where the border crosses. The head-to-head works through that pairing cell by cell in BIS LBS vs FDIC Quarterly Banking Profile.

API, files, or your warehouse. Daily, weekly, or hourly.

Because the hard part was never finding the release - it is keeping six decades of revisions flowing into a model without the pipeline quietly breaking. The raw artifact is a half-gigabyte wide file built for archivists, one column per quarter, every code abbreviated, and most questions want one slice of it. Teams that ingest it directly re-derive the same three fixes on every project: decode the code-and-label pairs, normalize the wide grid into tidy columns, pin the vintage so restatements stop invalidating last quarter's extract.

Request a sample first: name the reporting countries, counterparties, currencies and measures, and the extract arrives shaped identically to the standing feed, so anything prototyped on it survives into production unchanged.

Ranked by how directly the panel answers their day job:

1. Investors & quant researchers track cross-border credit build-ups as a risk-appetite signal, watching foreign-currency claims and bank-nationality mixes shift quarters before spreads do. See investors & quants use cases.
2. Risk & treasury teams at internationally active banks benchmark their own book against the system view - the mirror image of their own regulatory returns, seen from the outside.
3. Economists & central-bank watchers treat the panel as the measurement layer under global-financial-cycle arguments, pairing it with credit-gap and debt-service indicators from the same compiler.
4. Data scientists & ML engineers engineer features from eight-way-keyed stocks and 195-quarter histories, with enough depth for regime models rather than last-quarter deltas. See data scientists use cases.
5. Policy researchers, academics & journalists cite a record every central bank already treats as canonical, without stitching together national publications.

Ranked by how often a cross-border banking workflow reaches for them:

1. Federal Reserve Z.1 Financial Accounts of the United States - the US flow-of-funds record with flows back to 1945:Q4, the domestic counterpart when global BIS totals need a national balance-sheet breakdown.
2. ECB Statistics - Consolidated Banking Data (CBD2) - the euro-area aggregated balance-sheet view overlapping the consolidated (not locational) side of the BIS catalog.
3. FDIC Quarterly Banking Profile (QBP) - the domestic health check: charge-offs, deposit mixes and income statements for US institutions the locational panel deliberately abstracts away.
4. World Bank Development Indicators (Financial Sector) - dozens of financial-sector indicators across more than 200 economies, supplying the macro ratios - domestic credit to GDP, bank capital - that contextualise claim growth.
5. Federal Reserve Yield Curve (FEDS) Daily CSV - the rate layer, thousands of daily rows reaching back to June 1961, for valuing or stress-testing the exposures the locational panel measures.

None of these describe the cross-border book for you; the locational statistics own that. They supply the rate, flow and structural context that turns 609,000 rows of stocks into an argument about global liquidity.

Start with the BIS LBS dataset page for sample rows, the full field dictionary and a sample request cut to your counterparties and measures. For the wider pool, the diversified financial services data hub holds all 25 pooled records and best diversified financial services datasets carries the ranked comparison; the diversified banks hub lines up the neighbouring banking shelves. When the choice between lenses gets contested, read BIS LBS vs FDIC Quarterly Banking Profile before committing either way.