Statistical forecasting of the balance of payments of Ukraine (Q2740059)

This paper deals with different methods of forecasting of the time series in the form \(y_{t}=tr_{t}+c_{t}+s_{t}+r_{t}\), \(t=1,\ldots,T\), where \(tr_{t}\) is a trend component, \(s_{t}\) is the seasonal component, \(c_{t}\) is the cyclical component and \(r_{t}\) is the random component. Using the root mean squared error for \(n\) periods as a criterion in exponential smoothing, the authors make a forecast for some main articles of Balance of Payments such as ``Export of Goods and Services'', ``Import of Goods and Services'' and other. Also the double and triple Brown exponential smoothing as well as Holt-Winter's smoothing are used. For the estimation of the trend and the seasonal components the authors apply the additive-seasonal model and Hodrick-Prescott filter. The modelling of time series subject to changes in regime is presented.