I show why, when the realized rates of depreciation within the exchange rate band are regressed on a given information set and conditioned on ("ex post") actual no realignment ("ý la" drift adjustment), a 'peso problem' is still encountered. The reason is that the frequency of realignments in...