Exchange Rate Management
Realigning targets

The European Monetary System (EMS) resembles most current and past systems of `fixed' exchange rates in allowing bilateral exchange rates to fluctuate by a few percentage points in each direction, and it has been argued that the global monetary system should be reorganized along similar lines. Formal target zone models of exchange rate behaviour explicitly allow both for exogenous uncertainty and for a pre-specified intervention policy rule. The very existence of sharp limits on fluctuations should reduce volatility within bands, even when there is no official intervention. Market participants form expectations about future interventions, and they engage in stabilizing speculation within the target zone.
In Discussion Paper No. 398, Research Fellow Giuseppe Bertola and Ricardo Caballero review and extend these models and apply them to the EMS. Most formal models suggest that exchange rates should be more variable when far from the band's boundaries, that an interest rate differential should decrease as an exchange rate approaches its upper boundary, and that over the long run exchange rates should more frequently be near the limits of the band than in the middle.
Bertola and Caballero show that the patterns of weekly closing spot exchange rates for the French Franc, Deutschmark and Lira and one-month Euromarket interest rate differentials are inconsistent with the predictions of such simple models, and they seek to develop a model more consistent with this evidence. They first seek to determine whether the specification of alternative intervention strategies may reconcile the theory with at least some of the facts by considering inframarginal, discrete intervention policies. Such intervention rules cannot, however, explain the behaviour of the EMS rates within target zones.
Inspection of the EMS data suggests that this inconsistency may stem from the fact that within-band exchange rate movements are dwarfed by relatively frequent and relatively large realignments. Bertola and Caballero propose a simple model of discrete exchange rate intervention that allows for recurring stochastic realignments, and find that these extensions actually reverse the model's implications for observables, pointing to a possible reconciliation of theory and reality.
In their model, the authorities intervene only at pre-specified, common-knowledge points on either side of the central parity. They may then either `defend' the parity by bringing it back to the centre of the current band or `realign' it by declaring a new band adjoining the current one. If the probability of realignment is assumed to be high, exchange rates should be highly volatile in the neighbourhood of the band's margins, depreciation should diverge rather than reverting to central parities, and the long- run distribution of the exchange rate should be symmetric, with more weight in the centre of the band. This simple model may be extended to allow for asymmetries in the probabilities of realignment and for a long-run tendency of one currency to depreciate relative to the other. These have characterized the early EMS period, and this extension allows the target zone theory to be reconciled with the evidence of the EMS, by allowing for the possibility of destabilizing expectations in the interior of a target zone.

Target Zones and Realignments
Giuseppe Bertola and Ricardo J Caballero

Discussion Paper No. 398, March 1990 (IM)