Two aims of forest economic modeling are to find optimal stand management and harvesting regimes and to set the optimal policy instruments. We studied the effect of specification of a size-class model on the solution for economically optimal forest management. We first focused on the
choice of the conversion method to convert the growth data of individual trees into size-class structure; i.e., we tested two alternative estimators (proportion and increment). Next we studied the effects of specification of size classes with different numbers of diameter classes and partitionings
within. Growth description was based on the MOTTI stand-level simulator built on data from extensive field measurements in managed forests in Finland. Optimal forest management with the size-classified matrix model included thinning and clear-cutting and was studied for even-aged forest stands.
We found that the proportion estimator better captured growth dynamics of small-diameter trees, whereas for larger-diameter trees the difference between estimators was less evident. The effect of the number and partitioning of size classes depended on the estimator method. With the proportion
estimator, the number of size classes did not have systematic effects on timber yield nor land expectation value. With the increment estimator, yield and land expectation value systematically decreased when the number of size classes was increased. The results show that specification of a
size-class model has clear effects on growth predictions as well as on optimized economic indicators. Therefore, it is important to check the robustness of results with different numbers of size classes, different partitionings, and alternative conversion methods and their combinations.