Pemodelan Time Series Multivariat secara Automatis

Abstract

This research aims at establishing model of multivariate time series by means of econometric instruments. Four instruments in use are vector auto regressive (VAR), structural vector auto regressive (SVAR), vector error correction model (VECM), and structural vector error correction (SVEC). VAR and VECM are employed to estimate and construct models and, subsequently, predict the future values of an object. SVAR and SVEC serve to analyze innovative structures of a model. VAR and SVAR can be implemented only to stationary data whilst VECM and SVEC can be applied to non-stationary inputs. The identification and estimation of the model in this research are specifically designed by R software. Based on this software, all the aforestated models are conclusively able to identify dynamic relationship of endogenous variabel in a model well.