A Post-Sample Diagnostic Test for a Time Services Model

Abstract
A post-sample diagnostic test for judging the temporal stability of the Box-Jenkins time series models has been developed. The proposed test is based on the stochastic properties of the errors of the forecasts, at different leads, made from the same origin. Its application has been demonstrated in the analysis of a time series consisting of the monthly demand for in-place telephone services in Australia. An important alternative use of the test has also been indicated.