Long-term trends in the Real real prices of primary commodities: Inflation bias and the Prebisch–Singer hypothesis

Abstract

In his recent article on measuring the long-term trends in the real prices of primary commodities, Cuddington (2010) extends in several important respects our earlier efforts ( Svedberg and Tilton, 2006) to correct real commodity price trends for biases in the Consumer Price Index and other deflators. First, he argues for a log–linear relationship between prices and time. Second, he proposes a simple and quick method for obtaining corrected price trends from the published but uncorrected estimates. Finally, he illustrates, for the case of copper and presumably for many other commodities as well, the difficulties of obtaining real price trends significantly different from zero when the log values of the price data contain a unit root, requiring the use of difference stationary models. We welcome these insights, which should improve and make easier efforts to estimate correctly real commodity price trends over the long run. We would stress, however, that it is still important to correct for the biases in inflation indices, notwithstanding the failure of difference stationary models to obtain long-run real price trends (both corrected and uncorrected) significantly different from zero.