Get an economic statisticGet one of twenty-six headline IMF statistics for a country and follow it through time - GDP and its growth, inflation, unemployment, government debt and deficit, the current account, population, the exchange rate. Pick the statistic and the country by name; the IMF dataset and its codes are already filled in. The World Economic Outlook forecasts five years ahead in the same column as its measurements, so every figure says which it is and the result carries the latest actual separately. No account or API key required.
Get the latest figureGet the most recent value of one IMF statistic for one country, returned flat so a later step can read the number, the period and the change on the period before without indexing into a list. Answers 'what is inflation in Brazil' in one step. Because the IMF publishes forecasts beside measurements, this returns the latest measurement by default and will only hand back a forecast if you ask it to.
Compare countriesPut several economies side by side on one statistic and rank them, in a single step rather than one per country. Pick the countries by name from the list, up to forty at a time. Leaving the year empty gives each economy its own most recent figure, which can be a different year for each - the result says when that happens and every row carries its own period.
Get a country's economic profileGet several headline statistics for one country at once - growth, inflation, unemployment, income per person, government debt and the current account unless you choose otherwise. This is the briefing-note step: one call instead of six, with each figure in the unit it belongs in and each marked as measured or forecast. A statistic the IMF does not report for that country is listed as not reported rather than left out.
Get a commodity priceGet the world price of a commodity from the IMF's Primary Commodity Price System - crude oil, gold, copper, wheat, coffee, natural gas and a hundred more, monthly since 1980. Ask for the price itself, the index, or the change on the month or the year. The result carries the high, the low and the average across the period as well as the latest, so a price alert needs no arithmetic of its own.
Find a datasetSearch the IMF catalogue by words in a dataset's name or its summary, and get back the code every other action takes. This is the way in to everything the headline statistics do not cover: trade by partner country, financial soundness indicators, reserves, government finance, the Financial Access Survey. Half the catalogue is dated archive copies frozen as a published report left them; those are left out unless you ask for them, and the result says how many.
Get a dataset's detailsEverything you need to query one dataset: what it holds, the filters it is keyed by in the order the key wants them, how many values each of those accepts and a few examples of each, how often it reports, how many economies it covers, when it was last updated and how the IMF asks to be cited. Run this before 'Get data from a dataset' and the filter boxes stop being guesswork.
List a filter's valuesList the codes one filter of one dataset accepts, each with the words behind it, so a filter box can be filled in rather than guessed at. The IMF shares its code lists across datasets - the country list has 343 entries and no single dataset covers them all - so the values this dataset actually holds figures for come first and every row says which it is. Search by words to find one in a list of hundreds.
Get data from a datasetGet figures from any IMF dataset. Name the dataset and its filters appear as boxes, labelled the way the IMF labels them. Comes back as figures to loop over - each carrying every filter in both code and words - and as a table ready to append to a spreadsheet, and it says which boxes you left open. A request that would select thousands of series is refused before it is sent, naming the box to fill in.
Get a time seriesFollow one series from any IMF dataset through time. Every point carries the change and the percentage change on the one before, and the result carries the highest, the lowest, the average and the change across the whole period - so a chart or a trend alert needs no arithmetic of its own. Where the figures are already percentages the change is given in points instead, because the percentage change between two percentages is not a number anybody wants. The filters have to pin the dataset down to a single series; if they do not, the step says which box is still open. Covers the last 24 periods unless you set dates.
