Movements in the parameters of the gentan [Production controls] probability distribution

This paper considers the possibility of movements in the underlying gentan probability distribution. It is first shown that the mean and variance of the felling age are unlikely to be equal to the mean and variance of the underlying gentan probability distribution. The mean and variance of the felling age will increase if the growing stock is aging even if the gentan probability distribution is stable. It is concluded that movements in the mean and variance of the felling age cannot be used to infer movements of the underlying gentan probability distribution. A likelihood ratio test is developed as a rigorous test of whether the gentan probability distribution is moving through time. In 88 data sets the distribution showed a movement that was significant at the 5% level. Thus, if the gentan probability method is to be used as a forecasting model such shifts must be predicted. However, the problem of forecasting shifts is confused by the fact that they show no pattern, and because the proportion of the total area of forest under active management is not known and might, itself, be changing over time. It is concluded that much more research is required before the residual error in the forecasts can be reduced