Another Look at U.S. Passenger Vehicle Use and the 'Rebound' Effect from Improved Fuel Efficiency

Recently, Greene (1992) analyzed vehicle miles travelled for U.S. passenger vehicles over 1966-89 to econometrically estimate the "rebound" effect in fuel consumption resulting from improved fuel efficiency. He found that a static AR(1) model could not be rejected, implying that the rebound effect is small (13%) with no significant long-run adjustments, regardless of the assumed functional form (linear or loglinear). Another look at the data from a different model selection approach shows that while a loglinear AR(1) model is acceptable, the linear version is not. Using either form, a lagged dependent variable model cannot be rejected on statistical grounds yet has insignificant GNP effects, yielding similarly small short-run rebound effects but significant long-run rebound effects of about 30%. Thus, the evidence from these competing models for a significant long-run adjustment process is mixed, so that its presence cannot be completely ruled out.