An Examination of the Robustness of the Vector Autoregressive Granger-Causality Test in the Presence of GARCH and Variance Shifts

• The properties of the Granger-causality test in stationary and stable Vector Autoregressive (VAR) models are studied with different types of volatility processes imposed on the unconditional variance. For this test, it is examined how the size and power properties are affected by different magnitudes of GARCH processes and by structural shifts in the volatility. The study has been conducted by means of Monte Carlo simulations for different sample sizes. Our analysis reveals that substantial GARCH effects influence the size properties of the Granger-causality test, especially in small samples. The power functions of the test are usually slightly lower in the presence of GARCH disturbances compared to the case of white noise residuals. When a structural variance break is imposed, the size problem is rather severe, and the power functions are lower compared to the case with the pure GARCH processes.

Nyckelord

Nationalekonomi

Economics

SOCIAL SCIENCES Business and economics Economics

SAMHÄLLSVETENSKAP Ekonomi Nationalekonomi

Ekonomi

Business and economics

GARCH

Causality test

Size and Power

structural change

Publikations- och innehållstyp

art (ämneskategori)

ref (ämneskategori)