Canadian Forest Service Publications
Do forest products prices display long memory? 2012. Niquidet, K.; Sun, L. Canadian Journal of Agricultural Economics 60(2): 239-261.
Available from: Pacific Forestry Centre
Catalog ID: 33735
Available from the Journal's Web site. †
† This site may require a fee.
The behavior of forest products prices has large implications for forest management and policy. Decisions such as the timing of harvesting and investment in new plantations depend on future price expectations and policy makers often are concerned about the persistence of price shocks for stabilization purposes. Previous research on the topic has been mixed and focused on the question of whether prices are integrated of order one (i.e., have a unit root) and are nonstationary or if they are integrated of order zero and are stationary and mean-reverting. In an effort to understand the price process for two key forest products (lumber and pulp), time-series tests were applied to over 40 years of monthly pricing data. Using a modified Dickey–Fuller test, we rejected the null hypothesis of a unit root on most series. However, a separate test also soundly rejected a stationary null hypothesis. Such a finding led us to test whether forest products prices are fractionally integrated, exhibiting long memory. Estimates of the memory parameter (d) for various lumber price series ranged between 0.68 and 0.81 and for pulp it was 0.64. This implies that prices are both mean-reverting (in the sense that shocks die out) and nonstationary.