Have you ever wondered if vehicle kilometers traveled (VMT) does a good job of predicting recessions? You should have stopped after watching this Econbrowser publication of January 4, but I thought an update to the most recent data would be nice since we get the December data. First take a look at what VMT does during recessions, compared to heavy truck sales (suggested by calculated risk at times) and Sahm’s eponymous rule (real-time version).
Figure 1: 12-month growth rate of vehicle-kilometres traveled, nos (teal), truck sales, sa (tan) and Sahm rule indicator – real time (black). Sahm’s rule is the 3-month moving average unemployment rate relative to the lowest unemployment rate in the last 12 months. The red dotted line indicates the threshold for the Sahm rule indicator. The NBER has defined peak-to-trough recession dates as shaded. Hypothetical 2022H1 recession shaded lilac. Source: FHA via FRED, census via FRED, FRED and NBER.
It’s hard to see, but the 12-month change in MTV declined a few months ago before recovering in December (it’s this decline Mr. Steven Kopits pointed to), while heavy truck sales were up through January, year-over-year. The Sahm rule is exactly zero according to February data released yesterday (it takes 0.5 points to cross the threshold).
In any case, VMT growth is a poor predictor of recession (McFadden R2 of 0.07) compared to heavy truck sales (0.28) (see regression results in this job).
Looking ahead, I would be even more cautious about using the VMT as an indicator, given that the relationship between the VMT and GDP has apparently gone through a structural break. In Figure 2, I plot vehicle miles traveled (seasonally adjusted) at a quarterly rate by US GDP at a quarterly rate (so the units are vehicle miles traveled/GDP in real dollars). There is a clear trend of a decline of 1.15% per year over the period 2000-19. (I’m estimating a stochastic trend since I can’t come close to rejecting a unit root in the logarithmic ratio.) Using the estimated trend to project forward, I get:
Figure 2: Vehicle miles traveled per Ch.2012$ GDP (blue line) and estimated stochastic trend over the period 2000-19 (beige line), on a logarithmic scale. The NBER has defined peak-to-trough recession dates as shaded. Hypothetical 2022H1 recession shaded lilac. Source: FHA, BEA, NBER and author’s calculations.
In other words, Mr. Kopits viewed the downgrade to a seemingly new trendline as a cyclically induced reduction in VMT.