Heteroskedastic Factor Vector Autoregressive Estimation of Persistent and Non Persistent Processes Subject to Structural Breaks

In the paper the fractionally integrated heteroskedastic factor vec- tor autoregressive (FI-HF-VAR) model is introduced. The proposed approach is characterized by minimal pretesting requirements and sim- plicity of implementation also in very large systems, performing well independently of integration properties and sources of persistence, i.e. deterministic or stochastic, accounting for common features of di¤erent kinds, i.e. common integrated (of the fractional or inte- ger type) or non integrated stochastic factors, also featuring condi- tional heteroskedasticity, and common deterministic break processes. The proposed approach allows for accurate investigation of economic time series, from persistence and copersistence analysis to impulse responses and forecast error variance decomposition. Monte Carlo results strongly support the proposed methodology. Key words: long and short memory, structural breaks, fractionally integrated heteroskedastic factor vector autoregressive model.