Much of the paper is devoted toexpounding the standard model of the exchange rate acceptedby most economists today. This regards the exchange rate asa forward-looking asset price. Its steady-state level isdetermined by the need to have a current account balancethat will keep the debt/gross domestic product (GDP) ratioconstant, while the path of adjustment toward thissteady-state level is determined by the representativeagent's rational expectation of what will happenbetween now and the long run. The paper then examines anumber of criticisms of this model: that exchange ratechanges are driven by 'news' and will benonexistent in the absence of news; that it implies thatchartist rules will systematically lose money; and that itleaves no room for 'bubble-and-crash' dynamics,which appear to have occurred. An alternative'behavioral' model that gives room for suchbehavior is presented. The paper then argues thatovervaluation can thwart development through an attack of'Dutch disease,' and discusses the role thatexchange rate policy may play in avoiding this outcome. Thisdemands primarily the use of nonmonetary instruments likefiscal policy or capital controls, but the behavioral modelof the exchange rate implies that intervention can also playa role. The paper also includes a discussion of thealternative exchange-rate regimes available.