Go to Brad DeLong's Home Page
J. Bradford DeLong
University of California at Berkeley and NBER
Sebastian Edwards and Miguel Savastano have four major points:
- Throughout 1993-1994 most available information suggested that things in Mexico were getting badly out of hand.
- Most analysts, however, completely missed the magnitude of the disequilibrium--and thus the magnitude of the crisis.
- The nominal stability of the peso during 1996-1997 has been a source of alarm: will "it" happen again?
- Has the Bank of Mexico's monetary policy been geared to maintaining a degree of nominal exchange rate stability at odds with the requirements of the floating rate regime? No.
- The Bank of Mexico appears to have been using a feedback rule that takes account of the behavior of the nominal exchange rate.
- But Mexico's exchange rate behavior has been largely consistent with a (dirty, but not badly soiled) float.
As I read this paper, I became more and more impressed by the work that Sebastian Edwards and Miguel Savastano had put into it, and less and less impressed at our collective understanding of Mexico 1994-1995--why it happened, what it was, and what the chances are that it will happen again. They have done a very good job, yet I find myself feeling like--there is a Far Side cartoon, of dogs in lab coats, captioned, "dog scientists trying to discover the doorknob principle". An image that will resonate strongly with anyone who has had a dog and watched it try to open a door
They begin their paper by trying to debunk pieces--opposed pieces--of "conventional wisdom" that have become common rhetorical moves made either in looking back at Mexico 1994-1995 or looking across the Pacific at East Asia today. The first piece of conventional wisdom is that the peso crisis occurred because the Mexican government (assisted,of course, by their lackeys in the U.S. Treasury) lied about--concealed--hid--information about the state of its economy, and that when investors discovered that all was not pure, they recoiled in horror.
Edwards and Savastano have very little patience with this view, for which I thank them.
They also have little patience with the other--opposed--piece of conventional wisdom: that Mexican policy makers were simply doing their jobs, and doing reasonably well at their jobs, when due to bad luck the roof suddenly fell in. They write that: "...the Mexican authorities seriously underestimated the risks embedded in their chosen course of action. It is quite a stretch to claim... that this error in judgment... does not fall squarely in the category of policy mistakes." In other words, because Mexican economic policy led to catastrophe--did not keep the roof from falling in--Mexican economic policy was catastrophically awry.
From one perspective, this is clearly correct. Catastrophe happened. Policies to avoid it were not adopted. This was a policy catastrophe.
From another perspective this is unsatisfactory. The most interesting thing about the recent Mexican crisis was that the Mexican government's sins against the gods of monetarism seemed at the time--seem now--to have been relatively small. The magnitude of fundamental disequilibrium seemed at the time relatively small. Yet the punishment was swift, sure, deadly, and catastrophic: I do not know how deep Mexico's 1995-1996 depression would have been in the absence of IMF and (rather hastily paid back) U.S. Treasury support. I fear it would have been deep and long.
Now we do have models in which relatively small "fundamental disequilibria" can be followed by large currency collapses. But these are models of fiscal crisis (in which the government budget balance moves into massive deficit in such a way as to make everyone believe that large-scale monetization is likely) or models of policy fragility (in which a sudden fall in the currency induces a substantial shift in government policy toward accomodation and monetization). The Mexican case was neither of these: the government budget was not in substantial deficit; the fall of the peso did not lead to a substantial loosening but a tightening of policy--interest rates up to 80%.
Before the fact very, very few saw a crisis of the magnitude that actually occurred as even possible. As Edwards and Savastano write: "Very few, if any, observers would have predicted that merely a year after having abandoned the [currency] band, the peso would have lost almost one-half of its value. This inability to grasp the seriousness of the Mexican situation... clearly illustrates the shortcomings of the models commonly used by both private sector and academic analysts to assess the adequacy of real exchange rates."
They go on to criticize our collective model building, writing: "Indeed, most of these models are strictly based on flows considerations.... Models that, on the other hand, pay attention to stocks in general and to the foreign demand for securities issued by emerging markets in particular are, in principle, better equpped to gauge the magnitude of the disequilibrium in circumstances where credibility collapses." (p. 7)
But is this correct? What is the fundamental foreign demand for securities issued by emerging markets? It depends on risk, covariance with industrial-market returns, and expected return--and expected return is largely linked to the expected future growth rates of emerging-market economies.
I don't know anyone who has a clear visualization of the distribution of growth possibilities for emerging market economies. I don't know how news about today--about politics and economics today--should affect my estimate of the future growth of emerging market economies, and thus my fundamental demand for emerging-market securities. Nor do I know why a--relatively minor, 125 basis point--increase in medium-term real interest rates in the U.S. in 1994 should have had a large effect on industrial-economy demand for emerging-market securities.
I do know that the amount of capital crossing borders today is smaller, relative to the size of the world economy, than back before 1914. We have vastly more information now than they did then. And our markets and political systems are more open and more honest--back in the old days E.H. Harriman refused to let J.P. Morgan vote by proxy the shares of his British clients at the Illinois Central annual meeting, on the ground that even though the bylaws of the corporation admitted proxies the laws of the state of Illinois did not.
That makes me think that "fundamental" demand for emerging-market securities is considerably higher than the current value of such securities: that capital markets have been liberalized, and we are on a trajectory toward a steady-state in which cross-border investment holdings are much larger than they are today.
I also know that when you don't know much, and when news doesn't teach you much, your beliefs don't vary much--that conditional expected values are and remain close to the unconditional means of probability distributions. Under such circumstances large swings in the stock demand for emerging-market securities seem a deep puzzle: if we know next to nothing about long-run returns, how can our demand based on expectations of long-run returns suddenly and significantly change.
And this brings me to the second part of the paper. Edwards and Savastano ask: "Will 'it' happen again? Is the Mexican government informally pegging its nominal exchange rate in a way that is leading to sustained real appreciation and real overvaluation, thus setting Mexico up for a repeat crisis?" Some prominent and influential international macroeconomists who write columns for Business Week, have offices about half a mile from here, and have initials "R.D." have worried that it is.
Edwards and Savastano analyze this question by comparing the actual behavior of the Mexican peso over the past couple of years to what would have been expected had the currency been undergoing a relatively "clean" float.
Their first problem is that it is not at all clear how a currency undergoing a clean float behaves. We have good models of how it should behave. When we test these models, the data rejects them. We have very good normative models of exchange rates--but not good positive models of exchange rates.
So Edwards and Savastano begin by comparing the short-term variability of Mexico's exchange rate in 1996-97 to the variability of other exchange rates, and find that Mexico is not out of line. They go on to worry about the sudden shifts in volatility regime apparently experienced by Mexico--using the "virtual fundamentals" methodology of Flood and Rose (although it is not clear to me whether "virtual" is used in the sense of "virtual particle," "virtual reality," or "virtual presence"). They find that--in sharp constrast to Flood and Rose, who found that changes in exchange-rate volatility for OECD countries appeared unconnected with changes in macroeconomic volatility--that macroeconomics have changed: that Mexico's money demand has become more unstable in the period since the crisis.
This is very interesting: Flood and Rose, as I understand them, had set up their model--and the contrast between traditional and virtual fundamentals--to make it as hard as possible for changes in the exchange rate regime to be paralleled by changes in "traditional" fundamentals. The point was to demonstrate the inadequacy of monetary models of exchange rates. Yet Edwards and Savastano, using Flood and Rose's methodology, actually find standard monetary models... somewhat useful.
They also provide interesting evidence that the Bank of Mexico has not focused on domestic conditions entirely, but has kept one eye on the exchange rate in 1996-7--tightening monetary policy in response to large depreciations of the peso.
How does all this bear on their major question: is 'it' likely to happen again? Unfortunately, the answer is not clear. The Bank of Mexico appears to be following a policy like that of many other countries--float the exchange rate, but pay some (but not exclusive or even primary) attention to the exchange rate in setting monetary policy. There is no reason why such a feedback rule should lead to a period of pronounced real overvaluation followed by a currency crash.
But then a year before the 1994-1995 collapse of the peso there seemed to be no fundamental reason for the peso to undergo a catastrophic collapse either.
Some quotes from Edwards and Savastano:
"it can hardly be argued that there is consensus on what caused the December 1994 devaluation of the peso.... [W]e... take issue briefly with two rather extreme views.... The first one is the view that ascribes a negligible role to policy mistakes in the unraveling of the Mexican crisis [but instead]... attributes the coccurrence (and size) of the collapse of the peso to the combination of a sequence of adverse and unexpected shocks... and various sources of multiple equilibria that gave rise to a self-fulfilling run on the currency.... The second... ascribes a central role to the... paucity and unreliability of offical data on key economic variables in triggering the speculative attack..." (pp. 8-9)
"[U]ntil the very end, foreign investors simply underestimated the probability of a crisis." (p. 11)
"The flaws of the ... 'bad luck' view... become apparent when one examines the policy response to the sequence of shocks... during 1994.... There is a fundamental difference between using debt swap operations to smooth transitory hikes in interest rates, and using those operations to keep interest rates below their equilibrium level for a prolonged period of time.... Moreover, the decision to maintain the course of policies unaletered after the August presidential election... leaves little doubt that the authorities deliberately opted for a policy stance which they hoped would enable them to continue lowering interest rates and preserve the exchange rate band." (p. 12)
"By the third quarter of 1994, Mexican authorities might well have been under the impression that they had weathered off the worst... and that it was a matter of time before capital inflows would resume." (p. 12)
"Whatever the reason, there is no question that the Mexican authorities seriously underestimated the risks embedded in their chosen course of action. It is quite a stretch to claim... that this error in judgment... does not fall squarely in the category of policy mistakes" (p. 13)
"On December 20... the authorities... lifted by 15 percent the ceiling of the exchange rate band. To everyone's surprise, the announcement of the new exchange rate ceiling was not accompanied by other supportive changes in macroeconomic policies.... In disbelief, investors... fled, rendering the change in policy ineffective... the mayhem had just started." (pp. 13-14).
Currency Crises Conference, Royal Sonesta Hotel, Cambridge Massachusetts, February 6-7
- 9:00 AM February 6
- Alan Drazen, "Political Contagion in Currency Crises" (Carmen Reinhardt)
- 10:15 AM
- Barry Eichengreen and Olivier Jeanne, "Currency Crises and Unemployment: Sterling in 1931" (Michael Bordo)
- 12:15 PM
- Robert Flood and Peter Garber, "Is Launching the Euro Unstable in the End Game?" (Peter Kenen)
- 2:15 PM
- Robert Gordon, "Macroeconomic Responses in the 1929 Breakdown of the ERM" (Paul Krugman)
- 3:30 PM
- Sebastian Edwards and Miguel Savastano, "The Morning After: The Mexican Peso in the Aftermath of the 1994 Currency Crisis" (Brad DeLong)
- 9:00 AM February 7
- Guillermo Calvo, "Balance of Payments Crises in Emerging Markets: Large Capital Inflows and Sovereign Governments" (Roberto Rigobon)
- 10:15 AM
- Gian Maria Milesi-Ferreti and Assaf Razin, "Reversals in Current Account Deficits and Currency Crises" (Jaume Ventura)
- 11:15 AM
- Jeffrey Sachs, "Currency Crises and the Lender of Last Resort" (Frederic Mishkin)
- 12:15 PM
- Panel: Richard Portes, Jeffrey Frankel, Stanley Fischer, Jeffrey Shafer
This document: http://www.j-bradford-delong.net/Comments/Comment_E&S.html