Say crash and people tend to think of The Great Depression and the Global Financial Crisis: severe economic contractions triggered by financial crises that caused enduring slumps. But a paper out this week from the Bank for International Settlements suggests, rather worryingly, that oil price shocks tend to have the greater scarring effects on long-term growth.
BIS lays out its methodology as follows, drawing on data from 24 countries from 1970 to 2019 from which they glean 4,908 “observations” (for a more detailed breakdown, read the paper):
We first define contractions as time periods where the (standardised) annual real GDP growth rate is below the median, and order such events in terms of their severity. We then calculate multiyear real GDP growth rates (up to 10 years) and compare those from the origin of contractions — the quarter immediately preceding the drop in GDP — with those calculated from all other points in the sample.
First, authors David Aikman, Mathias Drehmann, Mikael Juselius and Xiaochuan Xing find a “tipping point in recovery dynamics” at around the 20th percentile: ie, only the top-fifth most severe contractions have observable effects on GDP levels a decade later.
Economic contractions are grouped into four categories — those caused by banking crises, restrictive monetary policy designed to combat high inflation, oil shocks, and everything else. Of the “extreme contractions” contained in the dataset (at or below the fifth percentile), 100 are banking-crisis driven, 51 are associated with monetary policy, 19 are caused by oil shocks and 9 have “other” causes.
The researchers summarise these findings as follows:
The difference in 10-year growth rates is 0.9 standard deviations following the 5% largest annual falls in GDP, but a more modest 0.2 standard deviations for a contractions between the 15th and 20th percentiles. These reductions in the 10-year growth rates translate approximately into permanent losses in the level of real GDP of 4.75% and 1.05%, respectively, for a typical economy in our sample. This is below the average loss estimate of 8.4% reported by Ball (2014) for the Great Recession.
The team next sort crises by type, focusing on the the most severe contractions at or below the fifth percentile:
Perhaps counter-intuitively, BIS finds that . ..
. . . while the point estimates of the growth shortfall following monetary policy tightenings is somewhat smaller than that following financial crises, oil price shocks generate materially larger growth shortfalls with 10-year growth rates 1.5 standard deviations weaker following such shocks . . .
Overall, these findings challenge the notion that it is only financial crises that generate scarring effects; the perhaps surprising message from [the chart above] is that all severe contractions have this characteristic.
Imagine the hit to growth were an oil shock and a financial crisis to hit not too far apart . . .