When Economies Fall Together

What was measured. For each year, how many of twenty OECD economies had negative real GDP growth, compared with the number you would expect if each contracted on its own, at its own observed frequency. The data are World Bank growth rates; the comparison is exact.

What the record shows. Contractions are far more clustered than independence allows. In 2020 nineteen of twenty contracted. The aggregate economy is nine times more volatile than independence predicts. And when contractions are clustered, each country’s downturn is 2.3 times deeper than when it contracts alone (p < 0.0001, within-country in 17 of 19).

Where the cost sits. Average growth is identical in both worlds (2.69%), so clustering does not lower the average. It concentrates the losses: the worst year fell 4.3%, where independence puts the worst year near +1.2%. No country loses any growth—each still has all its boom years, just not at the same time as every other country. The twenty-country average in the best year drops from +6.2% to about +4.0%, but that is a change in timing across countries, not a loss of output. The median year barely moves. The gain is concentrated in the tail: the worst years improve by five and a half points while the best-year average falls by about two. On the combined output of these twenty countries, each of the two catastrophic episodes (2009, 2020) represents roughly two to three trillion dollars of excess tail loss.

What the brief recommends. Five policy actions, from measuring the clustering in any risk model (a single script on existing data) to sizing fiscal buffers to the clustered tail, stress-testing for the synchronized scenario, and making coordination mechanisms standing instead of improvised.