Sixteen countries, and what it costs to compare them
This panel offers two measures of the same subject, and they come from different sources counting different things. Toggling between them changes the question, not the units.
IN DOLLARS, the source is the International Monetary Fund’s monetary survey — depository corporations’ claims on the private sector — for sixteen countries, monthly since December 2001. One indicator code, one definition, every country. It includes banks’ holdings of private securities alongside their loans.
AS A SHARE OF GDP, the source is the Bank for International Settlements and the Consejo Monetario Centroamericano, for twelve countries, quarterly. That measure is bank credit to the private NON-FINANCIAL sector, a narrower perimeter: at the first quarter of 2026 the IMF leg reads about 1.2 times the BIS-implied level for Brazil. Both are correct and neither supersedes the other.
Why not put both on one perimeter? Because this project holds no standalone GDP series, only ratios. A share of GDP for the IMF figures would need a denominator reverse-engineered out of somebody else’s published ratio, which is a guess dressed as arithmetic. The same refusal is why Nicaragua is truncated rather than rescaled — see below.
The BIS publishes its own all-lender measure too, and it is not used here. That series runs between 23 and 91 per cent above the bank-only one: at the first quarter of 2026 it puts Chile at 130.3 per cent of GDP where the bank-only figure is 69.5. Mixing the two would not simply raise five of the twelve bars, it would reorder them, which is the one thing a reader takes from a ranking.
THE DOLLAR CONVERSION uses a monthly average rate for every country, so the method is uniform even where the source is not. Nine currencies have a daily rate in this platform — the same one every dollar figure elsewhere on this site uses — averaged over the calendar month. Six have none, and take the Fund’s own monthly period average. Panama is dollarised and its source already reports dollars, so it is divided by one. A chart converted at fifteen different vintages would move for reasons that have nothing to do with credit.
NICARAGUA’S SHARE OF GDP STOPS at the fourth quarter of 2025. From January 2026 its published ratio jumps from 28.9 to 1,146.2 per cent, and the cause is visible in the source: the credit figure keeps growing by about one per cent a month while the GDP it is divided by falls by a factor of thirty-nine — from roughly 814 billion córdobas to roughly 20.7 billion, which is Nicaragua’s GDP counted in US dollars. The numerator stayed in córdobas. Its dollar series is unaffected and runs to its own end.
Ratios are never stacked. A share of GDP is measured against each country’s own denominator, so Brazil’s 74.9 per cent and Chile’s 69.5 do not add to 144.4 per cent of anything, and a hundred-per-cent stack of them would be a share of a total that was never a total. The control withholds the stacked options for that measure rather than trusting the reader not to reach for them.
An annual figure is a December, or a fourth quarter — selected, never averaged. These are stocks and ratios of stocks, levels at an instant, and the mean of twelve month-ends is not a level anything held. Countries end at different dates: Venezuela in August 2014, Peru in August 2025, Nicaragua in November 2025, Paraguay in January 2026. Those are the sources’ extents, and an absent period is a gap rather than a zero.