The currency crises of 1992 and 1993 that led to the effective collapse of the ERM pose a challenge to theories of exchange rate target zones. Neither the Krugman model of speculative attacks, which hinges on an unsustainable expansion of domestic credit, nor Obstfeld's model of self-fulfilling devaluation expectations, which Eichengreen and Wyplosz have related to the Maastricht conditions for EMU participation, captures the effects of German reunification credibility or some governments' leaving the ERM that could have remained inside, albeit at exorbitant interest rates. In Discussion Paper No. 879, Gulcin Ozkan and Research Fellow Alan Sutherland propose modelling the ERM as a fixed rate system in which a centre country sets its interest rate to achieve its own policy objectives; each non-centre country faces a foreign interest rate that is exogenous and subject to stochastic shocks, and its government maximizes a welfare function based on domestic output. Within the fixed rate system, it must set its interest rate to maintain the parity, so a rise in the foreign interest rate directly reduces output and welfare, but it can regain control of its monetary policy by exercising a once-for-all option to switch to a float.

Ozkan and Sutherland show that its optimal strategy is to maintain the parity so long as the centre country's interest rate remains below a certain `trigger' level but allow it to float once that level is attained. The private sector knows the government's preferences and policy options, so expectations of a regime switch as the foreign interest rate approaches this level will further raise the domestic interest rate and reduce output. This interaction between private expectations and the government's objective function creates a credibility problem. The government may wish to set a very high trigger level, for example by making a policy announcement that a regime switch is `very unlikely', which the private sector will not believe; the regime switch may therefore take place before it is ex ante optimal. Foreign exchange reserves therefore need not determine the timing of the regime switch, which instead reflects an optimizing decision by the government, albeit one that gives rise to a sub-optimal result.

A Model of the ERM Crisis
F Gulcin Ozkan and Alan Sutherland

Discussion Paper No. 879, January 1994 (IM)