Comments and Discussion Barry Eichengreen and Jeffrey A. Frankel Barry Eichengreen: The first rule of forecasting is, "Give them a forecast or give them a date, but never give them both."1 Michael Dooley and Peter Garber have given us a forecast, namely, that the dollar will fall and U.S. Treasury yields will rise. Bravely, they have also given us a date. Unfortunately for those of us interested in the future, that date is 1971. Like Dooley and Garber, I agree that what cannot go on forever generally will not. But unlike them I do not believe that recent events in financial markets can help us pin down the timing. The middle of March, just before the Brookings Panel meeting, saw an increase in noise about the possibility that foreign central banks might diversify out of dollars. The governor of the Bank of Korea made some widely reported comments about the need for more-active reserve management. Prime Minister Junichiro Koizumi of Japan told a parliamentary committee that reserve diversification was "necessary."2 Y. V. Reddy, governor of the Reserve Bank of India, said that the diversification of reserves was under active discussion.3 Ukrainian economy minister Sergiy Teriokhin argued publicly that the country should diversify its reserves out of dollars and into euros.4 This upsurge in noise was associated with an eight-month high in Treasury yields, reinforcing the belief that reserve diversification could eventually force the dollar down and Treasury yields up. At the same time, that eight-month high in Treasury yields was not all that high. I would acknowledge that this is a troubling point. I am not alone, of [End Page 188] course: Federal Reserve Chairman Alan Greenspan has commented on this issue extensively, to the point where it is now known as the Greenspan conundrum. Factors invoked to help explain it include the relatively short supply of new long-term Treasury debt coming onto the market as the debt managers at the U.S. Treasury shorten maturities, and the inelastic demands of various institutional investors for government securities. In Dooley and Garber's view, the proper interpretation is that financial market participants are attaching a positive probability to Asian central banks continuing to support the dollar by making massive purchases of Treasury bonds. This is the substance of the first of the authors' three notes. Who am I to second-guess the markets, much less to second-guess our esteemed authors? Well, I'm an economic historian who can recall a substantial number of previous episodes where major imbalances leading to sharp changes in exchange rates were not obviously factored into financial markets until immediately before the event. For example, the January March 1933 run on the dollar, a suggestive precedent, had virtually no discernible impact on interest rate differentials or forward exchange rates until almost immediately before it occurred, despite the fact that the possibility had been actively discussed for the better part of a year.5 The 1992 attacks on the pound sterling, a currency that commentators regularly cited as ready for a fall, were similarly not preceded by the emergence of noticeable interest rate differentials or a forward discount in the foreign exchange market until a couple of weeks before the denouement.6 Particularly interesting, given the context, is that in 1968-71, in the run-up to the collapse of the Bretton Woods system, the forward discount on the dollar was very modest, as was the interest rate differential between the United States and Germany.7 Then, in the summer of 1971, the forward discount jumped upward. Although one can always ascribe such behavior to the arrival of new information, it is not as if people failed to see the collapse of the Bretton Woods system and a substantial devaluation of the dollar coming. To the contrary, there was an immense contemporary literature warning [End Page 189] that the system would dissolve and that the dollar would have to fall substantially. Yet there seemed to be a striking reluctance to take a position on this basis until one minute before the clock struck midnight. This behavior may be hard to reconcile with perfect foresight, but, if it exists...