Gasoline Demand, Part 3
[See preceding posts for more information.]
After writing yesterday’s post, I decided to update my data to include the last quarter of 2009. This definitely changes things, as you can see from this graph:
What makes these record-low consumption rates interesting is that they are not completely explained by high fuel prices. The same observations are again highlighted in orange in the graph below:
What is different about these observations and 2009 in general? What could be lowering gasoline demand, even if prices are lower?
The answer, of course, is the recession. One good way proxy for the recession and its effect on people’s economic behavior is the unemployment rate. When we do this, the demand equation becomes:
where δ equals the monthly variation (remember that gas consumption fluctuates month by month) and U equals the unemployment rate (for example, 9.7% in March of 2010.)
Adding the unemployment rate greatly improves the accuracy of our model. That’s not to say that a simpler model, using only unemployment, would do a better job than a model only using price — the unemployment-only model explains just 25% of the variation in gas demand. When combining unemployment with gas price and time of year, however, we’re able to explain about 83% of the variation in gas consumption [click here for more details]. This is really good.
***
How does this change the elasticity discussion from before? It actually makes demand slightly more elastic than I had previously estimated. However, it is still within the range estimated in the paper I referenced yesterday. Here is a graph depicting ε from 1/2002 through 12/2009:
***
Why go through all this trouble? Why care about demand equations and elasticities? Tomorrow I will show a way to reduce carbon dioxide emissions from US vehicles by 45 megatons while raising over $210 billion in revenue. Stay tuned!
