Does cointegration of prices of North American softwood lumber species imply nearly perfectly substitutable products?

Abstract

This study uses the Engle–Granger cointegration method to examine the correlation relationship among prices of North American softwood lumber species, wood-based panel prices, and US single-family housing starts. The methodology employed is identical to that used by US petitioners in the latest round of the Canada – United States softwood lumber trade dispute, whereby petitioners argued that all North American species of softwood lumber were considered to be nearly perfect substitutes for one another. Tests for nonstationarity using the augmented Dickey–Fuller unit root test, as well as our cointegration results, confirm petitioners’ results. We find evidence of a long-term cointegrating relationship among the prices of North American softwood lumber species. However, we found no clear evidence of nearly perfect substitutability among different North American softwood lumber species. Additional analysis reveals that a long-run equilibrium relationship also exists between North American softwood lumber species and four types of wood-based panel products, as well as US single-family housing starts. Given that the price movements of different wood-based panel products and single-family housing starts are highly correlated with the price movement of North American softwood lumber species, we fail to conclude that North American softwood lumber species can be claimed as nearly perfect substitutes. Furthermore, we find that the petitioners’ analysis is deficient in that it does not account for the fact that cointegration in prices among North American species of softwood lumber can be caused by common demand-side factors, such as residential construction activity.