Abstract
State-space models were created based on the prices of domestic or North American roundwoods or lumbers from January 1975 to December 2014. Trend, autoregressive, and seasonal components were identified. The resulting trends were approximately consistent with actual measured data, and the rates of autoregressive and seasonal components were comparatively small. Some autoregressive components produced large fluctuations around 1980, 1987, 1997, and 2013. Among seasonal components, Japanese cypress, Japanese cedar, and Japanese pine roundwoods produced large seasonal fluctuations. Furthermore, some autoregressive and seasonal components produced similar fluctuations. Hierarchical clustering analyses were performed on both autoregressive and seasonal components, and similarities were found between Japanese cedar and Japanese cypress roundwoods, or among Japanese larch roundwood, Douglas fir rectangular lumber, and Western hemlock roundwood. One explanation for the former was their similar circulation and production systems, or their strong reciprocal substitution potential. However, the latter was that the autoregressive and seasonal components had similar patterns though their relationship is still unclear.
