Early critics of the Exchange Rate Mechanism of the EMS believed that speculative capital flows would destabilize and possibly destroy it. In Discussion Paper No. 502, Research Fellows Andrew Hughes Hallett and Patrick Minford, with Anupam Rastogi, use stochastic simulations in the Liverpool multilateral world model to examine the ERM's stability by determining its welfare cost under a variety of operating rules. They assume rational expectations; high capital mobility; clearing labour markets that are significantly influenced by nominal trade union contracts; that money supply is invariant to shocks, monetary policy reacts strategically; and that fiscal policy cannot be used for macroeconomic stabilizations.

When the limits are set more tightly, there is effectively a clash between two monetary systems: member countries set monetary growth independently over the medium term but their exchange rates can respond only discretely, and there are no agreed rules governing the periods between realignments. With fully and permanently effective exchange controls, the central bank can resolve this clash by fixing the exchange rate through sterilized intervention while using monetary policy to set interest rates. With perfect capital mobility, however, the central bank cannot fix the money supply and the exchange rate simultaneously, so the money supply adjusts to whatever level the exchange rate peg dictates and uncovered interest parity is maintained.

The authors assume that the Bundesbank sets German money supply growth unconstrained by other member countries' monetary policies: the latter are are constrained, however, in so far as they maintain their parities against the Deutschmark. These assumptions do not rule out German short-run `strategic' reactions to the spillover effects of other members' policies. In the `default' model, the exchange rate must be held within a band of 3% around the parity for at least a year after a shock. A new parity can then be set to achieve expected exchange rate equilibrium, which can only be moved in multiples of 5%.

The authors argue that restrictions on the permitted frequency of realignments are potentially dangerous to stability: parity changes that are fully anticipated but stored up by a `two-year rule' will exaggerate the current disequilibrium and trigger a future backlash. The authors also examine the consequences of imposing restrictions on the size of realignments in a stochastic framework. In conclusion they propose a special adjustment of current wages in response to the prospect of future devaluation. For illustrative purposes they make an adjustment equal to approximately one-third of the parity change expected in the following year. They find some support for this scale of adjustment in the French and Italian experience of the EMS since 1979 and note further that these results are not sensitive to the parameter's size.