A comparison of prediction efficiency for timber prices in Poland in times of economic crisis with the application of the linear approximation method and brown’s exponential smoothing model

An analysis was made of two prediction methods: the Linear Approximation Method (LAM) and Brown’s Exponential Smoothing Model (BESM). These two methods were investigated and compared in terms of their efficiency in timber price prediction. Models and price predictions were prepared based on three time series (5-, 7- and 9-year) for three years: 2015, 2016 and 2017. The analyses were conducted using data on mean annual timber prices from the period 2006-2017. This meant that the time series included the years of the 2007-2008 economic crisis. Prediction efficiency was evaluated by comparing the results obtained with actual timber prices in the years 2015-2017. It was found that the predictions generated by LAM were better than those produced by BESM. The smallest relative and absolute errors of prediction were obtained applying the linear function: Υt^ = 5.277t + 161.70. This function was constructed based on a 5-year time series. Absolute error amounted to 1.59 PLN (€0.35). Relative error was below 1%. The results of this work suggest that further studies are desirable to investigate the applicability of trend analysis to the prediction of timber prices with the inclusion of analyses of nonlinear trends. The present results of timber price modelling may provide a basis to search for a homogeneous model of timber price prediction adapted to specific conditions of timber sales.