Loss function analysis of private forest owners’ timber price change expectations

ABSTRACTLoss function analysis was introduced to timber market price analysis. Methods found in Bayesian statistical decision analysis were applied to expectations of market price changes. These were obtained from a forest owner survey conducted in Finland in 2009. Forecast errors of heterogeneous price expectations of the individual private forest owners were derived with observed price changes in six different timber price regions. The forecast errors were complemented with random number simulations to control for unobserved response heterogeneity. The properties and estimation of generalized loss functions – LIN-EX and flexible loss functions – allowing for asymmetry in individual forecast errors were introduced into the analysis. Estimation results for different price regions implied that forest owners’ perceived costs related to positive forecast errors of negative price changes that took place in 2009 were larger than those related to negative errors. This under-prediction vs. over-prediction asymme...