This paper describes the implementation of monetary policy by the Banque de France during the period beginning in January 1987. This period was characterised by a quasi-fixed exchange rate for France, as the franc was never realigned within the European Monetary System (EMS).

The behaviour of the Banque de France is modelled with VAR models, which are a means of breaking down the dynamics of interdependent time series of macroeconomic variables, into the various sources of exogenous fluctuations. In the case of an open economy the modelling of the monetary policy requires a careful account for the exchange rate impact on the short-term interest rate. This is even more tricky in the countries like France who target a fixed exchange rate to the DM. Even in the very short term, a variation of the interest rate can be interpreted either as a response to a demand shock for currencies, or as a genuine shock of monetary policy.

I propose a "procedural" identification of French monetary policy shocks in the spirit of Bernanke and Mihov (1995, 1996). This approach helps to solve the problem of the simultaneous determination of the exchange rate and interest rate instruments of the central bank. I use two interest rates of the French interbank market: the intervention rate, which I assume, represents the instrument of the monetary policy and the day-to-day market rate. In the model, the short run reaction function of the Banque de France was to raise its intervention rate in case of a German short-term rate innovation or when the franc depreciated and to let the day to day rate increase steadily in periods of stress on the Franc.

I build indicators of the stance of French monetary policy. These show the influence of the ERM constraint on French monetary policy in an original way. I also show that since 1987, the shocks of purely domestic French monetary policy (i.e. the variations of the interest rate around a rule of defence of the ERM peg of the franc), have only had very little impact on the French economy.