Data as of November 2026. Register release 1.0.
Every claim below was computed from public official data by one script, and each one carries the number of observations behind it. We publish the claims that failed as prominently as the ones that held, because a research record that only reports successes is not a record.
Where a claim is about a relationship, three numbers are reported. The plain correlation mixes differences between countries with changes inside them. The second removes fixed differences between countries, so each country is compared only to itself. The third also removes each year's common shock. The third is the strictest and the one we quote.
Inflation is sticky. This year's rate is the best single guide to next year's.
Countries with more people online are richer, and the link still holds when a single country is followed over time.
The interest rate banks charge moves with inflation, closely enough that either is a fair proxy for the other.
When inflation runs hot the after-inflation interest rate falls, which is arithmetic before it is economics.
Five institutions hold 43.0 percent of insured bank assets and fifty hold 77.5 percent.
In monthly euro-area data higher unemployment does go with lower inflation, so the trade-off exists at higher frequency even though annual data hides it.
The share of output going into investment is only weakly related to growth the following year.
A high after-inflation rate is followed by slightly lower inflation a year later, but the size of the effect is small.
Countries with more women in parliament tend to have less income inequality, and that holds when the same country is followed over years.
Bigger banks run slightly leaner, but the difference is small.
Central government debt shows no relationship with growth the next year, which cuts against the simple story that debt slows an economy.
Foreign investment inflows do not predict growth the following year.
A larger government share does not predict slower growth the next year.
Countries with longer life expectancy grow slightly slower the following year. The effect is small.
A larger manufacturing share is not associated with more volatile growth here.
Military spending as a share of output shows no relationship with growth the next year.
Higher research spending is not followed by faster growth the next year.
Higher after-inflation interest rates are not followed by measurably slower growth in this data.
The tax take and the income level are barely related once a country is compared to itself.
More university enrolment is not followed by faster growth the next year in this data.
More open economies do not grow faster the next year in this data.
The trade-off between lower unemployment and higher inflation is not visible in annual data, though it is in monthly data.
The rule that unemployment rises when growth falls is not visible in annual country data.
Bigger banks are not more profitable. The gap is negligible.
Bigger banks do not earn wider lending margins.
Banks holding more capital earn the same return on assets in this snapshot.
The idea that growth widens inequality and then narrows it does not appear in this data once countries are compared to themselves.
Countries spending more on health live longer, but most of that gap is between countries, not within a country over time.
Richer countries are more urban, but a country's fastest-urbanising years are not its richest years. The headline comparison overstates the link.
Method version 1.0. Annual country panel from World Bank World Development Indicators, 1995 to 2025, 262 territories, 186,242 observations. Bank cross-section from FDIC quarterly financials for 2026 Q2, 4,313 institutions. Monthly series from Eurostat. Minimum thirty usable observations per test. Lags are chosen by hand, not estimated. All figures are association on official data; none of it identifies cause, and no instrument was attempted.
Corrections are logged publicly with dates. A claim that stops surviving new data is withdrawn here rather than quietly dropped.
29 claims ยท register release 1.0 ยท data as of November 2026