On Thu, 14 Oct 2004, Doug Henwood wrote:
> Wojtek Sokolowski wrote:
>
> >Provided that linearity assumption is reasonable. As I understand, the
> >regression equation your cited was determined from historical data, but the
> >projections go beyond the range of the historical data i.e. the price did
> >not go beyond $2.30 at the pump or $50 crude.
>
> True, but real crude prices today are well below the $80 real (CPI)
> level of 1980.
>
> You could also argue that nonlinearity would work the other way -
> that as oil prices rise, the ratio of retail gas prices to crude
> could decline, since some of the markup is attributable to refining,
> transportation, and other overhead costs (and taxes) that bear little
> or no relation to oil prices.
Frankly, I'm amazed that a linear model fits up to this point. (I wonder about the homoscedasticity of the residuals, though. Any chance we could get the raw data?)
Miles