German dominance in the European Monetary System: a reprise using robust Wald tests

Abstract

This study re-examines the German dominance hypothesis in the EMS. A VAR in levels approach is adopted and modified Wald tests, which are robust to cointegration properties of the variables, are used. The findings indicate that, although Germany played a measurable role, the extent of monetary cooperation in the system was greater than previously detected.

Notes

The study contributes to the literature in two additional aspects. First, the GDH is examined in five subperiods, each is characterized by different developments in the EMS. Prior studies suggest that conclusions concerning the GDH are sensitive to the period examined. Second, daily interest rates are used, rather than the monthly or quarterly rates often used in the literature. The use of these higher frequency data helps to detect dynamics in the system when policy response is rapid. With the gradual dismantling of capital controls, lags in the response of interest rates to innovations are likely to have been reduced to a few days. Hence, statistical causality tests based on monthly data are likely to lose too much information to be reliable.

The power of the modified Wald test may be questioned because there is inefficiency when estimating a VAR with more lags than contained in the true structure. However, Zapata and Rambaldi (1997) compare the conventional F-test in the context of a VECM with the Dolado and Lutkepohl (1996) approach in a Monte Carlo study and find that the modified Wald test has favourable performance in size and power in samples of 50 or more observations.

The UK was a member of the EMS from October 1990 to September 1992.