Friday, September 18, 2009

Predicting oil prices from interest in electric cars

You have heard or read the public opinion theories that the oil companies are acting like monopolies (i.e., they conspire) to manipulate gas prices. While I have yet to see hard evidence that they collude, I find it troubling that little production capacity has been added despite higher prices. But that could be due to environment regulation, as is credibly claimed.

Jose Azar adds a troubling observation to the debate. He finds that whenever interest in electric cars increases, oil prices happen to decline. And not just a little, half of oil price changes can be explained by the frequency of Google searches about electric cars. But that can also show that markets really work: when there is interest in substitute goods, prices decline.


William said...

My quick impression (not having read the paper) is that this could just be a result of regression to the mean.

Easterly wrote that you will almost always be right if you say "I predict that the growth rate of country X will increase," where X is the country that currently has the lowest growth rate in the world.

People search for cars more when oil prices are especially high, i.e. the price is above trend value. Nothing nefarious going on here.

José said...


As you say, people get interested in alternatives when oil prices are high. I find evidence of this. The vector autoregression methodology is flexible enough to take that into account, as well as regression to the mean in oil prices. You can see in the impulse responses of Figure 5 that:

1- Interest in electric cars responds postively to oil price shocks (after a week).
2- There isn't much mean reversion in oil prices.

What I found is that, when search volume for electric cars is higher than would be predicted by a (linear) vector autoregression --which includes the price of oil as a predictor-- there is a negative effect on the oil price (beyond what would be expected by mean reversion).

A second point is that, even if the results were due to mean reversion, interest in electric cars could be acting as a transmission mechanism: perhaps oil prices would revert to the mean because there is high interest in alternatives.

Some caveats:
1- I'm using five years and a half of data, so this behavior could be specific to the period 2004-2009.
2- I provide some discussion of possible mechanisms that could be generating what I found, but I'm not attempting to be conclusive on this.

I hope this helps. Please, let me know if you have more comments or questions.

PS: The latest version of the paper can be found here:

Vilfredo said...

Now that this relationship has been discovered, will it disappear due to the efficient market hypothesis and arbitrage? Quite obviously, if it does not, there is money to be made.