Group autoregressive modeling using a smoothed matrix of indirect growth rates

Abstract

Purpose: the author describes a new approach to forecasting multidimensional economic processes on the basis of a combined matrix model, the construction of which uses the procedure of smoothing indirect growth rates. Discussion: group autoregressive modeling is carried out in two stages. At the first stage, an autoregressive model is constructed for each predicted indicator. And in the second stage, these autoregressive models are combined into a group using the matrix of indirect growth rates. It allows to carry out forecast calculations with orientation on group dynamics of predicted indicators. The group dynamics is formed on the basis of its own growth rates. They are adjusted with the help of indirect growth rates in the direction of balanced development for all indicators. The second stage of model construction provides balance of forecast dynamics, at the same time concedes on reliability to the first stage. The author forms a matrix of indirect growth rates using the exponential smoothing procedure to equalize the reliability. Results: the author offered and made changes to the methodology of group auto-regression modeling for multidimensional economic processes. The results of the computational experiment confirmed the feasibility of the exponential smoothing procedure, as well as improving the reliability of predictive calculations.