This is the third in a series related to Forisk’s Q3 2016 forest industry analysis and timber price forecasts for North America.

Prior to opening our spreadsheets and shutting out the real world, let’s remind ourselves of practical matters that influence hardwood forecasts. Operationally, hardwood supplies by specie and the hardwood mills by type dictate the local situation for hardwood log markets. In confirming the facts on the ground, we verify supportable relationships between total demand and supply at the specie level, if possible, for the given market. In cases, specs and data for specific products vary to such a degree over time that statistical relationships have little meaning. So we work to confirm that a reasonable supply base exists for the mills in the market.

Forisk has built hundreds of market-specific, forward-looking models for timber prices, forest industry production, wood bioenergy capacities, and the U.S. economy. Our team studied data sources, developed models, and back-tested results. From this experience, we apply key lessons as guidelines for developing forestry-related models associated with timber pricing and demand projections.

Lesson one is “understand the underlying data.” When analyzing timber markets or wood baskets, paranoia regarding price data serves us well. Why?  First, each stumpage price data set – no matter its source, perceived quality, consistency, depth and structure – reflects a (biased) sample that suffers from practical, unavoidable limitations when applied to statistical analysis.  Second, stumpage prices cannot be generalized across markets.  Like timber markets, stumpage prices reflect a snapshot of uniquely local circumstances.

Price data for hardwood logs and stumpage vary in quality, consistency, and reported scale (e.g. Scribner versus International), especially when compared to softwood prices. The confidence and robustness of hardwood log forecasts built for Maine differ from those for Michigan or Minnesota or Massachusetts. Sample size truly matters, and small samples – a common issue with specie-specific hardwood price data – can be unreliable. When working with hardwood prices, plot the data and calculate summary statistics to provide a sense of the data depth. If the price data is insufficient for a meaningful time series, spend less time on the model and more time confirming demand (healthy mills) in the marketplace and on the phone or in the archives confirming price levels today and historically.

 

This post includes excerpts from Q3 2016 FRQ featured research by Brooks Mendell and Jack Lutz on building forecasts for hardwood stumpage prices for local timber markets, which includes a case study and forecast of the Lake States region. To learn more about the Forisk Research Quarterly (FRQ), click here or call Forisk at 770.725.8447.