Macroeconomic Policy
Coordinated losses?

The well-known result due to Rogoff, that international monetary cooperation does not necessarily provide a welfare improvement over the outcome under non-cooperative policy, was derived in the framework of a static repeated game. In a previous study of international cooperation, Programme Director Marcus Miller and Mark Salmon found that this possibility of `counter-productive coordination' also arises in a dynamic model of open economies with floating exchange rates. The appealing intuition that countries must necessarily be able to secure a welfare improvement by cooperating to internalize the externalities generated by monetary policy actions with floating rates is thus subject to a form of the `Lucas critique': it ignores the impact of the act of coordination itself on the expectations of market participants, and hence of the constraints this imposes on policies that are determined jointly as a result.

In Discussion Paper No. 425, Miller and Salmon use a continuous time model of two symmetric open economies with floating exchange rates to demonstrate the sensitivity of the welfare effects of policy coordination to the initial inflationary positions of the potential partners. They restrict their attention to the examination of time-consistent policies, since it is well known that these may be welfare inefficient, and it appears that coordination increases the potential for such inefficiency, at least when the inflationary conditions differ between the two countries.

The authors also find that a stochastic interpretation of their results shows that coordination is effective when inflationary supply-side shocks are highly correlated but may not be effective otherwise. Hence uncorrelated or negatively correlated inflation shocks may pose a time-consistency problem that is exacerbated by coordination.

By showing how the pay-off to coordination depends on the initial conditions (or distribution of repeated shocks), Miller and Salmon resolve the seeming contradiction between the `positive' pay-offs reported by Oudiz and Sachs and the `counter-productive' results they obtained for essentially the same dynamic model.

Since the welfare losses of counter-productive coordination are essentially those induced by the `time inconsistency' of optimal policy, many of the suggestions that exist already in the literature to ameliorate this problem may be applied to this case. The authors dismiss the argument that `if coordination doesn't pay then it won't happen' and proceed instead to consider two of the proposed remedies in the context of their formal model: the `delegation' of national authority to an anti- inflationary coordinating authority; and the operation of `simple rules' that distinguish between symmetric and asymmetric inflation shocks.

When Does Coordination Pay?
Marcus H Miller and Mark H Salmon

Discussion Paper No. 425, July 1990 (IM)