Despite increasing capital mobility and the subsequent difficulty in controlling exchange rates, intermediate exchange-rate regimes have remained widespread, especially in emerging and developing economies. This piece of evidence hardly fits the "impossible Trinity" theory arguing that it becomes difficult to control the exchange rate without a "hard" device when capital flows are freed. Calvo and Reinhart (2000) have suggested several explanations for such "fear of floating": exchange rate pass-through, liability dollarization, dollar invoicing of domestic and external transactions, and an underdeveloped market for currency hedging make it more desirable to stabilize the nominal exchange rate. However, the New-Keynesian model, which has become the main workhorse for studying exchange-rate regime choice since the 1990s, typically opposes fixed nominal pegs to free-floating regime, without considering intermediate regimes. We intend to fill this gap here by comparing the performance of "extreme" regimes to that of an intermediate regime where monetary authorities care both about inflation and about nominal exchange-rate deviations from the steady state, when a small economy is hit by several types of shocks. Without nominal wage rigidities, our results are in line with the New-Keynesian literature arguing in favor of inflation-targeting regimes. However, when nominal wage rigidities are taken into account, we find the intermediate regime to be appropriate for an economy that is mainly hit by productivity and foreign-interest shocks, which is often the case in emerging and developing economies. The free-floating regime (with inflation targeting) seems more adequate if the economy experiences mostly demand shocks and foreign prices shock. Finally, the fixed peg regime is always dominated by either the free-floating or the intermediate regime. A fully-fledged analysis of intermediate regimes should of course account for the fear-of-floating-type advantages of such regimes, as well as for their shortcomings in terms of costly reserve-accumulation and/or recurrent crises. Our results however suggest that, by concentrating on two extreme regimes (fixed nominal pegs and free floats), by neglecting wage rigidities and/or by assuming that floating countries can engineer an "optimal" interest-rate feedback rule, the existing New-Keynesian literature may have exaggerated the merits of free-floating regimes to the detriment of "soft" pegs.
