LINEAR DYNAMICAL MODELS WITH TIME LAGS

摘要

Dynamical processes in the economy are usually expressed either as a dependence between system variables with some cause-effect relations or in the form of a time series model based on statistical data. In the last case they can select autoregressive models and an optimal choice of the representative from a function class given. This report is devoted to linear autoregressive models for multivariable time series with one or more time lags. The optimal autoregressive coefficients are found as a point of a square functional minimum at fixed lags. The optimal lag corresponds to the best forecast. Both closed macroeconomic systems and control models are considered. The algorithms given are tried on macroeconomic parameters of one European country.