Money Supply Data Source: How This 185-Country Dataset Is Built

Two sources feed this money supply data source: the IMF’s Monetary and Financial Statistics database and the World Bank’s World Development Indicators, with the euro area taken straight from the ECB. As of 2026-06 the combined series covers 185 economies, 161 of them converted to USD at period-average rates.

The series codes are MFS_MA/BM_MAI and MFS_DC/DCORP_L_BM at the IMF, FM.LBL.BMNY.CN at the World Bank, and BSI M3 at the ECB. This page states what is inside that 185 and what is not. Here’s why: the audience reading a methodology page is the audience most likely to catch us being wrong.

What “money supply data source” means on this site

Two primary series, one derived layer on top.

IMF Monetary and Financial Statistics (MFS) is the IMF SDMX-format broad-money database behind 157 of our 185 rows. Its standard series code, MFS_MA/BM_MAI, supplies 155 of them. Two rows — India and Somalia — use the alternate depository-corporations series, MFS_DC/DCORP_L_BM, because the standard series doesn’t cover them. MFS defines broad money as currency in circulation plus transferable and other deposits held by resident sectors. The aggregate goes by different names by region: M2 in the US, M3 in the euro area, M4 in the UK. The exact components differ slightly by country even when the underlying idea is the same. Monthly, updated on the IMF’s own release schedule, published at data.imf.org.

World Bank World Development Indicators (WDI) supplies the annual backfill, FM.LBL.BMNY.CN — broad money in current local-currency units. We use it only where the IMF has no monthly series at all: seven economies, listed below. WDI also publishes broad money as a share of GDP under a separate code, FM.LBL.BMNY.GD.ZS. That’s a ratio, not a level — not what we use here. Published at data.worldbank.org.

ECB Balance Sheet Items (BSI) M3 supplies the euro-area series directly from the European Central Bank — more granular and more current than either institution’s own euro-area row (more on that below).

None of this is raw redistribution — we compute YoY growth, 10-year CAGR, doubling time and the USD conversion ourselves, on top of the source series.

Coverage: 185 economies, 161 of them in USD

185 economies carry a broad-money observation in this dataset; 161 also carry a USD-converted value. The row count was 158 before a mid-2026 repair pass. That pass added seven economies through the World Bank backfill route below, and gave the 20 euro-area members an explicit redirect row each instead of leaving them silently absent. 158 plus 7 plus 20 is the 185.

All six arrive through the same backfill methodology. The IMF’s monthly series doesn’t run for any of them. So the price of coverage is annual frequency and, in four cases, a last observation more than 24 months old, which carries a visible staleness badge. Canada is the exception that proves the rule: it used to sit in this table on a 2008 World Bank figure, and now comes monthly from the Bank of Canada’s own Valet API (V41552801, M2++) — a live national source beats an annual mirror whenever one exists.

Sixty of the 185 rows carry a coverage start year of 1960, the earliest year any source reports. A 1960 start is not the same as a report every month since. Several of those series have gaps. So each country page shows its observation count beside its coverage start year, and that count is the honest measure of depth. The rest start later, at independence, a currency reform, or wherever the IMF’s own coverage begins.

The euro-area rule: why 20 countries don’t get their own row

Germany does not print euros separately from France. None of the 20 euro-area member states issues a national currency of its own any more: the eleven founding members stopped in 1999, Greece in 2001, and Croatia, the most recent joiner, in 2023. They are Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Croatia, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Portugal, Slovakia, Slovenia and Spain. None of them has its own money supply to report, so none of them gets its own row in this dataset. Each one carries redirect_to: EMU and points to the Euro Area series instead.

That’s not a gap in our coverage. It’s a fact about how the currency works — and it’s why the Euro Area is one page, not 20 pages fabricating numbers that don’t exist. The Euro Area series comes straight from the ECB’s BSI M3 database: monthly, 558 observations, 1980-01 through 2026-06, €17.61T as of 2026-06 — $19.90T converted. That is deeper and fresher than any free euro-area series we found while building this site. If you know a longer public one, tell us and we’ll switch to it.

No competitor states the redirect explicitly. TradingEconomics lists euro-area members with their own local-currency M2 rows (checked 2026-08), as though Germany and France still ran their own printing presses — a framing the euro retired in 1999.

Three more EUR-denominated economies in our data make the useful contrast. Kosovo, Montenegro and San Marino report their own broad-money figures in euros. Each uses the euro unilaterally or by agreement, without belonging to the Eurosystem or the ECB’s monetary-policy area. They keep an independent row, with their own observations and USD conversion. The distinction that matters is currency-union membership, not currency choice.

How the USD conversion works — and its one real limit

Every USD figure here is the local-currency observation multiplied by the World Bank’s PA.NUS.FCRF period-average annual exchange rate for that observation’s calendar year. Not a spot rate, not a monthly rate — an annual average, applied to a monthly (or annual) observation.

That is a real limitation. Say a currency moves 20% against the dollar between January and December of one year. Every month inside that year converts at the same blended rate. January and December both carry the full year’s average, even though the true dollar value of each moved throughout the year. The distortion is largest exactly where it matters most: fast-devaluing currencies. In the Argentine peso, the Venezuelan bolívar and the Turkish lira, a single year’s FX range can exceed the annual average itself.

Two consequences follow. First, our dimensionless metrics (YoY, 10-year CAGR, doubling time) are computed pre-conversion, so this limit never touches them. It is a level problem, not a growth problem.

Second, we made a choice. A rough USD figure with the method stated beats the alternatives. One alternative is to skip conversion, which is what theglobaleconomy does across all 174 of its rows (checked 2026-08). The other is to convert and say nothing. TradingView’s public money-supply page shows Malawi at 4.25T USD, larger than Germany, and Tunisia at 48.07M USD, smaller than Comoros (both checked 2026-08). No FX method is stated anywhere on that page.

Why 24 rows can’t enter a level ranking (but do enter a growth ranking)

24 of 185 rows carry no USD value. They split into two very different categories, and mixing them would misstate our own data.

Twenty are the euro-area members from the section above. They don’t get a level figure or a growth figure of their own. Every number attached to them, growth rates included, comes straight from the Euro Area series, because they carry zero independent observations. Showing Germany’s “own” 4.0% YoY growth would just be the Euro Area’s 4.0% wearing Germany’s flag — we exclude them from every ranking, not only the level ranking, for that reason.

The other four are a real, separate gap: Anguilla, Montserrat, the Eastern Caribbean Currency Union (ECCU) and the West African Economic and Monetary Union (WAEMU). Each has its own observations from the IMF — Anguilla’s series runs 291 months back to 2001, WAEMU’s runs 295 — and its own growth rate computed from them. What they lack is a USD conversion. The World Bank’s exchange-rate table carries no rate indexed to these four entities. That is a hole in our sourcing, not a design choice. It is also why they are safe in a growth ranking. We keep them out of every level ranking and out of the world total until we find an FX source that covers them.

Stated once, so it doesn’t need repeating on every ranking page: growth, CAGR and doubling time are comparable across every currency without conversion. 8% a year is 8% a year, pesos or pounds. That’s why the growth ranking here is more defensible than any level ranking, including our own.

How current is “current”

110 of 185 rows (59.5%) carry an observation from 2026; 36 more are dated 2025. The remaining 39 are older, and the two oldest are Panama (2008-05) and Syria (2011-12), both IMF MFS.

That spread is published, not smoothed over. A ranking table that mixes a 2026-06 observation with a 2008 observation without saying so would be misleading. Ours says so on every row, with a visible staleness badge on anything older than 24 months — 36 rows carry that badge today. Headline claims (front-page totals, “current” language) draw only from rows inside the 24-month window.

Switzerland states the trade-off plainly: the most recent free figure we can source is the World Bank’s 2016 annual reading. The honest options are showing nothing, or showing 2016 with a loud staleness badge. We choose the second — a labeled old number beats a blank cell, and a missing country reads to our readers as an error, not caution. Canada used to sit in this paragraph on a 2008 figure; it is now monthly from the Bank of Canada, which is what we do whenever a national source can be reached.

Why IMF and World Bank give you two different numbers for the same country

The euro area shows how the same currency can carry two different figures on the same day. The IMF’s own euro-area row (G163) reports €16.79T as of 2025-05 — $18.97T converted. Our primary Euro Area row, sourced from the ECB’s BSI M3 database directly, reports €17.61T as of 2026-06 — $19.90T converted. Same currency, same currency union, two different figures, because they’re two different collection pipelines with two different cutoff dates. We show the ECB figure and mark the IMF one exclude_from_rankings: true so it can’t be counted twice.

The same mismatch shows up, less sharply, wherever both institutions cover a country. Different methods, different revision schedules, different frequencies. We don’t blend the two inside one country’s history. The rule is a strict hierarchy: IMF’s monthly MFS is primary wherever it exists; World Bank’s annual WDI is the backfill only where IMF has none. Each country page states which source is live for it — a source switch would be logged as a correction, not silently absorbed.

What we compute ourselves vs. what we redistribute

World Bank WDI data (including all seven backfilled economies here) is published under CC BY 4.0: free to redistribute, with attribution. IMF’s Monetary and Financial Statistics terms are more restrictive and don’t clear bulk redistribution of the raw series.

The data licence splits by source, and that split decides what’s downloadable. The seven World Bank rows redistribute the raw local-currency series directly, attributed. The 178 IMF- and ECB-sourced rows never ship as a bulk file. Instead we publish what we compute on top of them: YoY, CAGR, doubling time, the USD conversion, each labeled as our own. We add a deep link to the primary series on data.imf.org or the ECB’s Statistical Data Warehouse.

Most competitors don’t draw this line visibly. CEIC and TradingEconomics gate the underlying series behind a sales form. Ours is free, no email required — it ships our derived metrics plus source links, not an IMF bulk export we don’t have permission to distribute.

Three steps to check a number yourself

In order of effort:

• Read the Specs block on that page. Every country page, ranking and comparison carries source, series code, observation date, frequency, unit and FX method inline. It is the same block shown at the top of this page, scoped to that one figure.
• Download the CSV. The data page ships the full dataset as one CSV, with the same source/date/unit columns as the site itself, free, no email gate. One file, no per-country split, no API key.
• Go to the primary source. Every derived figure links to data.imf.org, data.worldbank.org or the ECB’s Statistical Data Warehouse. Our revision policy: a source correction updates the row and is logged publicly, never silently edited. If our number disagrees with the source on the same date, that’s a bug — tell us, and it goes into the public log, not a quiet edit.

Cross-check this money supply data source against the primary series any time — we’d rather be told we’re wrong in public than right by luck.

FAQ