Tuesday, December 15, 2009

Employment and Oil Demand

I wanted to take a look at the relationship between employment and oil demand. I was rather hoping to use this as a basis for forecasting the recovery in oil demand, but actually the correlation is too poor for that.

Here's the raw data. The employment data is non-farm payroll employment from the BLS establishment survey. The oil demand data is the EIA weekly US product supplied series discussed in this post.

To make is easier to see, I re-expressed both series as a fraction of their peak value. (The oil demand is probably not going to go above 1 again, but hopefully the employment will!)

As you can see the oil demand peaked earlier (which makes sense given the price shock in 2005-2008), but also has fallen proportionately far further than the employment series.  This I need to understand better - I will take a look at the various subcomponents of oil demand shortly.


kjmclark said...

Turns out there are many economics papers on the effects of oil shocks, in part started by some guy named James Hamilton in a 1983(!) paper. This is a good summary.

I was looking for something that compared the effects of oil shocks on different industries, due to their different oil intensities. Turns out the most influential paper on this was done by Lee and Ni in 2002. However, they ended up concluding that Hamilton was basically right, "The findings of this study lend support to the theory that increased operating cost of durable goods and heightened uncertainty are major reasons for oil price shocks to induce recessions."

I also found this, on industry-level effects, and this, which argues that unemployment effects are more the result of the way industries pass along their costs or are affected by demand for their products due to the oil shock, and how linkages between industries are important.

Stuart Staniford said...

KJM - I have read at least some of Jim's papers on the subject, and I kind of agree with his perspective, but only up to a point. I think he's identifying the specific paths through the economy that oil shocks often take, but to me the more important point is that economic activity is physically constrained by the amount of energy available and the efficiency that it can use energy to create goods and services. So an oil shock mandates that less is produced in the way of goods and services, unless/until the economy can be made more efficient. That to me is the first order thing going on in the situation, and the fact that people get uncertain and don't buy cars is more of a second order thing. An oil shock induced recession must go on long enough and be deep enough to eliminate enough economic activity that society can now function within the reduced oil budget available. That could happen in a lot of different ways (eg in the most recent case, there happened to be a housing bubble at the same time, so it was possible for that to burst and the physical constraints the economy was running into be satisfied that way).

Nonetheless, the fact that auto demand always goes down in recessions *is* an important issue, because it places a constraint on how quickly society can improve vehicle fuel economy.

Stuart Staniford said...

I guess to say a little bit more, I've always found the way economists look at this subject to be quite strange because they think entirely in terms of prices and incentives, and not very much in terms of physical flows and physical constraints. Still, their perspective is valuable and important and it's a mistake to dismiss it out of hand without making an effort to understand why they think what they do (as I know many scientists tend to).

josswinn said...

You might want to look at employment data from ShadowStats (http://www.shadowstats.com/), which shows, for example, twice the level of unemployment than the official figures.

Half Empty said...

Economics seems always to emphasise price, while never considering the amount of energy needed within the system to give those prices strength or meaning.

This comment on a recent peak oil article in The Economist discusses the kind of possible feedback loops that economics often seems to overlook.

JackRussell said...

I would like to second the motion to look at employment data from ShadowStats. The government has learned to manipulate the unemployment figures (and they have been doing this for decades - nothing new here).