Thiswork examines the conditions for non-violation of the basic assumptions on the error component of a multiplicative time series model when inverse square transformation is applied to the error term. To achieve this, the curve shapes of the probability density functions (pdfs) of were compared for some values of and it was found that the distribution of the transformed variable loses its symmetry when Rolle’s Theorem was also used to find the region where satisfies the bell-shaped condition; and this is met Use of the simulated error terms shows that the transformed variable is normal for Finally, from the functional expressions for it was observed that the mean of is one and the increased variance is approximately 4 times the variance of Therefore, the condition for successful inverse square transformation with respect to the error component of the multiplicative time series model is.
展开▼
