This paper revisits the sticky-price pricing-to-market model of Devereux and Engel [Devereux, M.B., Engel, C., 2003. Monetary policy in the open economy revisited: price setting and exchange-rate flexibility. Review of Economic Studies 70(4), 765–783], in which fixed exchange rates are optimal even in the face of country-specific nonmonetary shocks. We show that this result hinges critically on the Devereux–Engel model's prediction that international consumption levels are perfectly synchronized under flexible prices. Realistic modifications of the model that produce nonsynchronous consumption movements – such as, the presence of nontraded goods – upset the fixed exchange rate prescription even in the absence of an expenditure-switching role of exchange rate changes.
This paper revisits the question of optimal exchange rate variability when prices cannot adjust immediately after country-specific real shocks. The traditional approach to this question dates back to Friedman (1953) and even earlier writers, who argued in favor of flexible exchange rates. In conventional open macromodels, dating back to the seminal works of Fleming and Mundell in the 1960s, imports are priced in the currency of the producer and the law of one price holds for tradable goods. The pricing assumption implies that the pass-through of an exchange rate change to import prices is complete and immediate: when a currency depreciates, for example, the prices of all imports rise immediately in proportion to the depreciation. This relative-price change generates an expenditure-switching effect between home and foreign goods and lends a stabilization role to exchange rates in the face of country-specific real shocks. Empirical evidence, however, suggests that the assumptions of costless international trade and rapid, unitary pass-through are in general oversimplifications, lending further interest to the study of macromodels featuring market segmentation and pricing-to-market in international trade.1
In a pioneering paper, Devereux and Engel (2003) – DE hereafter – extend sticky-price models in the “new open economy macroeconomics” vein to incorporate price-setting in buyers' currencies by price-discriminating exporters. A key feature of the DE model is that exchange rate changes are not associated in the short run with changes in the relative import prices that confront consumers and, thus, do not generate an expenditure-switching effect between local and imported goods. One might infer that in models such as DE's, featuring pricing-to-market and local-currency pricing, exchange rate variation cannot stabilize the economy as it does in the Mundell–Fleming model, by switching aggregate demand between home and foreign goods. And indeed, DE conclude, on the basis of a welfare analysis in their model, that fixed exchange rates are optimal even in the presence of idiosyncratic national productivity shocks. This inference would seem to overturn the conventional wisdom that country-specific real shocks make exchange rate flexibility desirable. See Engel (2002) for an elaboration of this theme.
In this paper, we demonstrate that flexible exchange rates are optimal in realistic variants of the DE model which still feature a complete absence of expenditure-switching effects of exchange rate changes. In our model, it is optimal for monetary authorities to affect domestic aggregate demand differently in response to country-specific real shocks, implying a flexible exchange rate under optimal policies. In DE's model, the optimality of fixed exchange rates is primarily due to a knife-edge and unrealistic symmetry restriction embedded in their model that eliminates the need for distinct effects of monetary policy on aggregate demand across countries. We emphasize that DE's result is not due to the absence of expenditure-switching effects of exchange rate changes. It is important to stress that we are not simply making the point that an absence of expenditure-switching exchange rate effects on consumer spending leaves room for exchange rate flexibility.2 Instead, we show that even a complete absence of expenditure-switching effects need not nullify the case for flexible exchange rates in more realistic variants of the DE model. The specific modification we make to the DE model is to add nontradable goods, although our analysis suggests that a number of alternative plausible modifications would have a similar effect on the model's predictions about optimal monetary policy.3 Our conclusion is that while more detailed theorizing about open economy price rigidities is extremely valuable, the channels of monetary policy transmission can be subtle and researchers should accordingly be cautious in leaping to radical policy conclusions
Taken together, DE and this paper suggest caution in analyzing the transmission mechanisms for monetary policy in open economies with complex price rigidities. The DE paper makes an important advance in demonstrating how alternative price-setting arrangements in open economies can alter the transmission mechanism. A distinct advantage of the new open economy macroeconomics approach is its accommodation of the detailed modelling of price-setting regimes, coupled with an exact analysis of the general-equilibrium welfare implications. In reality, national consumption movements are asymmetrical and international asset markets are incomplete. We therefore think it unlikely that optimal monetary responses to country-specific real shocks would ever imply rigid exchange rates in practice. Both DE and this paper, moreover, suggest that the sources of low exchange rate pass-through to consumer prices affect the optimal degree of exchange rate volatility. Like DE, we have limited our analysis to the qualitative dimensions of monetary policy, leaving for the future a close quantitative study of how alternative pricing arrangements and economic structures affect the optimal amount of exchange rate volatility.
