Methodology
How the figures on this dashboard are sourced, recorded and converted
Currency conversion
Figures are recorded in local currency as per official data. US dollar figures are strictly for comparison purposes only, and are converted from local-currency figures using annual period-average market rates from the IMF's International Financial Statistics, as published in the World Bank's official exchange rate series, matched to the year of the transaction data. Rates are fixed at the point of publication and are not revised. An alternative view converts every year at a single rate, which removes exchange-rate movement from multi-year comparisons. No inflation or purchasing-power adjustment is applied.
Please note that given the differing approaches to currency conversion — particularly for volatile currencies — non-local-currency data should be referenced with consideration to the above methodology, and referenced only as a guideline.
Local currency is the record
Every figure is stored in the currency its operator published it in. The dollar figure is always derived from that, never the other way round. Where the two disagree, the local-currency figure is the one to use.
Period average, not year-end
Transaction values are flows accumulated across a year rather than a balance struck on 31 December, so they are converted at the average rate over the period they accrued. Using a closing rate would materially understate a year of turnover in a currency that fell during it.
Where a period is not a calendar year, the rate applied is that of the calendar year containing the period's midpoint. Safaricom reports M-Pesa on an April-to-March financial year, so the year ending 31 March 2026 converts at the 2025 rate: nine of its twelve months fall in 2025.
Rates do not move once published
The rate for a completed year is written once and left alone. A dollar figure for 2021 shown on this site today will be the same figure next year. If a rate ever has to be restated it will appear in the changelog.
The two views
| Setting | What it does | What it answers |
|---|---|---|
| Each year's rate (default) |
Converts every period at the average rate for its own year. | What was this flow worth at the time? |
| Fixed rate | Converts every period at one year's average rate. | How much did the underlying payment activity change, setting the currency aside? |
Nigeria shows why both exist. NIP's naira value converted at each year's rate runs from roughly $678bn in 2021 to $723bn in 2024 — broadly flat in contemporaneous dollars. Converted at a single recent rate it runs from about $179bn to $705bn, a fourfold rise. Neither number is wrong. The first describes what the flows were worth when they happened; the second describes how much more money moved through the system in naira terms. The naira fell by roughly two thirds against the dollar over that period, and that difference is the whole gap between the two readings.
What is not adjusted
- No inflation adjustment. Converting at each year's own rate already absorbs much of the effect for high-inflation currencies, because those currencies tend to depreciate roughly in line with domestic prices.
- No purchasing-power parity. A payment moves financial value, and market rates are what value it. PPP is built for comparing output and living standards, not payment flows.
Limitations worth knowing
- Currency events are not payment events. Where a country floats or devalues, the dollar line moves for reasons that have nothing to do with payments. Nigeria unified its exchange rates in June 2023; Egypt devalued in March 2022, October 2022, January 2023 and March 2024. In each case the local-currency series is the one that describes the payment system.
- Official rates only. Where a country has run more than one rate, the official rate is used, because that is what the source publishes. Argentina's official rate has diverged sharply from its parallel market, so Argentine dollar figures should be read with that in mind.
- Pegged currencies. Several currencies here are pegged to the dollar, so their converted series carries no exchange-rate effect at all and the two views are identical.
Where the rate comes from
The primary source is World Bank indicator PA.NUS.FCRF, "Official exchange rate (LCU per US$, period average)", whose underlying source is the IMF's International Financial Statistics. Rates are held for 2009 onwards.
Four currency-years are not in it. Rather than borrow a rate from a neighbouring year or a commercial aggregator, each was taken from the country's own central bank, and each is listed below with what it actually is. Two are published annual averages. Two had to be built from published sub-periods because the central bank concerned does not publish an annual average at all.
| Currency | Years | Basis | Source and method |
|---|---|---|---|
| TWD — New Taiwan dollar | 2009–2025 | Published | Central Bank of the Republic of China (Taiwan), annual "Period Average" column of its N.T. dollar / U.S. dollar table. Taiwan appears in neither the World Bank nor the IMF dataset, so the whole series comes from its own central bank. |
| LKR — Sri Lankan rupee | 2024, 2025 | Published | Central Bank of Sri Lanka, Annual Economic Review 2025, average exchange rate table. CBSL and the World Bank differ by 0.008% in 2023, the last year both publish, so there is a hairline step at the join. |
| RUB — Russian rouble | 2025 | Derived | The Bank of Russia publishes a daily official rate but no annual average. Built as the mean of twelve monthly averages of that daily rate. The same method reproduces the World Bank's own figure to within 0.02% for 2023 and 2024, and the result agrees with the CBR's qualitative statement that the rouble averaged 6–11% stronger than in 2024. |
| VND — Vietnamese dong | 2025 | Derived | Eleven of the twelve monthly period averages are published by the IMF; August 2025 is missing, which is why no annual figure exists upstream. August was rebuilt from the State Bank of Vietnam's daily central rates. Vietnam runs a managed float, and this central rate — the one the primary series tracks — sits below the rates commercial banks actually quoted through 2025. |
Where a rate cannot be sourced at all, the dollar figure is left empty rather than filled with a rate from elsewhere or from another year. Local-currency figures are unaffected either way.
Downloads
Both conversions are included in every CSV download from this site, as separate columns, so a downloaded file is unambiguous about which basis produced which figure. The API returns both alongside the local-currency figure, the rate table itself, and the rate year applied to each data point.
What gets published
Confirmed figures only
A figure appears here only when the authoritative source has published it. Growth rates are not extrapolated, gaps are not interpolated, and no figure is estimated from an average transaction size or any other ratio. Where a source has not published something, the dashboard shows nothing for that period. A gap is an honest output.
Complete years only
A year is shown once the full year has been published. Year-to-date totals presented as annual figures are a common trap in press releases and are excluded — a December release describing the year "so far" is not a full-year figure.
Financial years are cut into calendar years
Where an operator reports on a financial year, the calendar year is reconstructed by summing the published months or quarters that fall inside it. A financial-year total is never presented as if it were a calendar year.
Volume and value are independent
Some operators publish transaction counts but have never disclosed value, as a matter of policy. Where that is the case the count is shown and the value left empty. Neither is derived from the other.
Sources are traced to the publisher
Every figure is traced to the operator, central bank or scheme that published it. Secondary aggregators are not used as sources, including where they cite this dashboard.