Abstract:
The standard Vector Error Correction Model (VECM) approach to investigating the underlying dynamics of economic variables assumes a constant co-integration space. This paper relaxes this assumption by implementing a regime switching VECM that allows for shifts in both the drift and the long-run equilibrium. Applying this more flexible formulation to a study of UK meat consumption, we can clearly identify several shifts in meat consumption. These can be explained by significant shocks in consumer confidence in meat safety, such as BSE. Although it is possible to model these explicitly, since the approach adopted models the regime shift in terms of an unobserved state variable, it can be useful in identifying such shifts, thus allowing them to be modeled in subsequent steps.