Tyler Godspeed (former acting CEA Chair Trump 1.0; chief economist ExxonMobil) observes that shocks as determinants of recessions means they are essentially unpredictable.
If you look back over not just the past 80 years, but indeed over the last three and a half centuries, energy is one of those sectors that has generated, or has been subject to, a lot of shocks that then permeate the rest of the economy. It’s not hard to see why, because energy is an input into a lot of other sectors, and it is very difficult over a 12-month or even 24-month time horizon to find substitutes for fuels, for heating, for the materials that use petroleum products.
I don’t have at hand 350 years of energy prices, but here’s the one relevant for the US over the past century past.
Figure 1: CPI-deflated price of oil (WTI), in 2026$/bbl, annual (blue), monthly (red), both on log scale. NBER defined recession dates shaded gray. Oil prices 1913-1970 annual. March oil price data is through 3/30; March CPI data is Cleveland Fed nowcast. Source: EIA, CPI via FRED, EIA, NBER, and author’s calculations.
As Godspeed correctly points out, the highest price (real as well as nominal) is recorded in 2008; However, there’s a question — as always — whether level of change is the relevant variable. This is particularly true if one is thinking of “shocks”. If one thinks of oil prices as partly endogenous in the smooth run-up in prices, then one would more likely want to focus on differences. Below in Figure 2 is the year-on-year log difference in real oil prices.
Figure 2: Year-on-Year growth rate of CPI-deflated price of oil (WTI), in 2026$/bbl, annual (blue), monthly (red), both on log scale. NBER defined recession dates shaded gray. Oil prices 1913-1970 annual. March oil price data is through 3/30; March CPI data is Cleveland Fed nowcast. Source: EIA, CPI via FRED, EIA, NBER, and author’s calculations.
In this context, the 2008 year-on-year change of 640% is larger than the 1979-80 increase which presaged the 1980 recession, but on a month-on-month basis does not come close to the 1973 or 1990 changes. And (thinking of Sims’s work, I’d be more inclined to use this object in a VAR), maybe the month-on-month or quarter-on-quarter changes would be more appropriate.
Figure 3: Annualized month-on-month growth rate of CPI-deflated price of oil (WTI), in 2026$/bbl, annual (blue), monthly (red), both on log scale. NBER defined recession dates shaded gray. Oil prices 1913-1970 annual. March oil price data is through 3/30; March CPI data is Cleveland Fed nowcast. Source: EIA, CPI via FRED, EIA, NBER, and author’s calculations.
If the month-on-month changes more closely resemble a shock, then we’re in for some “propagation” in the next months (the graph reads “400%” shock).