Asymptotic behaviour of temporal aggregates of time series

摘要: SUMMARY The problem of modelling and forecasting temporal aggregates time series is discussed. Let zt be the basic XT m-component aggregates. Forecasts future may constructed from data on (i) or (ii) XT. It shown that, for large m, there no gain in using if ztis stationary, but considerable can obtained when Zt nonstationary.