In his recent blog, Roberto Rigobon has directed attention to the role of fiscal policy in preventing real exchange rate appreciation in the inflation targeting (IT) regimes of Brazil, Colombia, and others. I would like to suggest a slightly broader focus, however, on the issue of what, if any, should be the role of exchange rate management in an inflation targeting regime. This is an aspect of inflation targeting regimes where the discrepancy between theory and practice has been most evident (at least to me), and where the extent of our ignorance may be quite damaging.
Inflation targeters in Latin America generally claim to have flexible exchange rates, as required by the conventional wisdom on IT. At the same time, however, IT central bankers openly admit that they reserve the right to depart from flexible rates under some circumstances. For example, Peru’s central bank “has the policy of moderating excessive exchange rate volatility in order to minimize the negative effects of large exchange rate movements in a financially dollarized economy” (Adrian Armas and Francisco Grippa, Targeting Inflation in a Dollarized Economy: The Peruvian Experience, April 2005, page 27). As stated in the quote, central bankers in IT regimes often feel the need to prevent exchange rate movements that may not be warranted by fundamentals, and to fight large depreciations that may jeopardize financial and payments systems. As these concerns emerge relatively often, active exchange rate management has not been an occasional departure from the IT rulebook. Instead, it is an integral part of the rules.
What would conventional IT gurus say? In the well framework of Svensson (1998, 2000), for example, a central bank chooses its instrument path to minimize an expected loss function of inflation and the output gap. Svensson shows that the instrument should then be set as to minimize deviations of the inflation forecast from target. A corollary is that the instrument should respond to any variable that provides news useful to forecast future inflation. The exchange rate could be one such forecasting variable, and therefore the conventional IT wisdom may rationalize the need for policy to respond to exchange rate fluctuations.
But of course this argument, while theoretically tight, is clearly not the reason why Latin inflation targeters have engaged in active exchange rate management. The Svensson framework does not allow for either of the two concerns most frequently by our central bankers: that exchange rates may be driven by nonfundamental forces, and that large depreciations may destroy our financial systems. Two questions then emerge: What do we know about how these two concerns affect our views on how the economy works? And, what are the implications for exchange rate policy in an IT regime?
While we are very far from having a satisfactory theory, recent research has made considerable progress on the second concern. There is now a very active literature on liability dollarization, currency mismatches, and financial frictions, which emphasizes that a currency depreciation may indeed exert a contractionary effect on aggregate demand via net worth effects. This research has been rich on policy lessons. Unfortunately, it has not settled the question of how exactly the central bank should adjust its policy instrument in response to exchange rate news. Luis Cespedes, Andres Velasco, and I have developed models in which the contractionary effect of depreciations can be very strong, yet it is optimal for the exchange rate to float; this finding has been echoed by others (e.g. Devereux , Lane, and Xu 2004). But David Cook has presented models with opposite results. Therefore, while it is now widely accepted that balance sheet effects and dollarization means that exchange rate depreciation may be strongly contractionary, the theory has not yet provided a rationale for active exchange rate management.
The situation is much less satisfactory on the question of how nonfundamental exchange rate movements may affect the role of exchange rate management under inflation targeting. This reflects, of course, the lack of a satisfactory theory of “nonfundamental” phenomena. In addition, the few attempts I am aware of suggest that allowing for such phenomena has ambiguous implications for exchange rate policy. In particular, a few years ago Ben Bernanke and Mark Gertler asked whether allowing for asset price bubbles in financial accelerator models is likely to affect standard prescriptions for monetary policy. The answer, they found, is negative. Much more research seems to be needed on this front, however, as it is extremely hard to dispute the contention that exchange rate movements are (perhaps most of the time!) not justified by fundamentals.
My preliminary conclusion (and please correct me if you disagree!) is that, while IT central bankers may be right to be concerned about excessive and large exchange rate movements, I am unclear about whether their common reaction (to fight such movements) makes sense. And, by extension, I am unclear about how inflation targeting works in the tropics, if it does.