Résumé : The Granger Causality (GC) has been widely used in the field of both theoretical and applied econometrics. The stationary processes are the term of reference used in GC testing, which is generally studied by means of a standard F test. However, a considerable literature has proven an eminent presence of the unit root processes (UR) in macroeconomic series. Then, considering the important consecuences from spurious regression theory, it is important to understand the behavior of GC test if we have this type of nonstationary processes. In this work, we prove that, when the Data Generating Process (DGP) of the series is UR with or without drift, correct inference can not be drawn from an standard GC test and may identify inexistent causal relationship in the Granger sense. To show this we derive non-standard asymptotic distributions of F statistics and we study the properties in finite samples using several Monte Carlo experiments.

Cet exposé se tiendra en salle C20-13, 20ème étage, Université Paris 1, Centre Pierre Mendès-France, 90 rue de Tolbiac, 75013 Paris (métro : Olympiades).