Comments on Bénassy-Quéré, Mojon, and Pisani-Ferry and on Masson and Turtelboom
- Thomas Krueger, Paul Masson, and Bart Turtelboom
- Published Date:
- September 1997
- Ralph C. Bryant
The title of this session as printed on the program is “The Euro as a Reserve Currency and Exchange Rate Volatility.” Yet the two papers begin to ask analytical questions about whether the prospective move to EMU will facilitate macroeconomic stability in a broader sense. This broader perspective is, in my view, more appropriate than a narrow focus on exchange rate volatility.
My main assignment is to discuss the paper by the three authors associated with the Centre d’études Prospective et dȁInfotmations Internationals (which for shorthand I will refer to as the CEPII paper). But I would also like to make some comments that apply to the Masson-Turtelboom paper.
The questions asked in the two papers are important and complex. There is no chance of addressing them satisfactorily without using an analytical framework—a model of some sort. The papers are therefore necessarily technical. And a full evaluation of them would have to dwell on their technical features. As in all analysis of this sort, there is a risk that the conclusions reached will be model-contingent—that is, that the results may depend sensitively on the particular models being used.
This conference is not the most appropriate place for an in-depth technical discussion of the strengths and weaknesses of the models used in the papers, important though such a discussion is to a full evaluation of the papers. For the most part, I am thus not going to discuss the technical features of the models used in the papers. Rather, I shall focus on the suitability of the analyses to the questions that are asked, and on the conclusions themselves.
As discussants inevitably do when fulfilling their task, I will concentrate on aspects of the papers where I have questions or doubts. But I emphasize at the outset that the two papers are thoughtful, stimulating, and helpful contributions to the growing literature analyzing the likely consequences of EMU.
How suitable are the models in the two papers for answering the analytical questions we want to ask about the likely effects of EMU on macroeconomic stability? The models are substantially different. It is an advantage of the CEPII model that it is fairly small, and can be solved analytically. But in several respects, I feel somewhat more comfortable with the MULTIMOD model of the staff of the IMF’s Research Department. MULTIMOD treats the individual European countries separately. And MULTIMOD explicitly incorporates key intertemporal stock-flow interactions that we believe to be important on both theoretical and empirical grounds. In the MULTIMOD model, the flows of investment add to capital stocks, which feed back into production functions; imbalances in government budgets alter the stocks of government debt, which have to be willingly held in asset holders’ portfolios and hence influence interest rates; and imbalances in current accounts lead to increases or decreases in countries’ net foreign asset positions, which in turn feed back into wealth and consumption, and indirectly into interest rates and exchange rates. More generally, there is an explicit specification of dynamic adjustments through time in MULTIMOD, which is missing in the simpler versions of the CEPII paper’s model.
Here is another pertinent point of comparison: the CEPII paper relies on comparative-statics analysis of its model, or at the end of the paper on an illustrative deterministic simulation. Such results can be interesting and useful. And they are perhaps a necessary prelude for understanding results from stochastic simulation analysis. For the types of questions being asked by both papers, however, I believe that stochastic simulations are a more appealing approach. For my taste, therefore, the methodology used in the Masson-Turtelboom paper has the greater potential for generating robust conclusions about the variability or stability of macroeconomic variables.
The CEPII paper examines EMU only after all the transitional problems are assumed to have receded in the past. To be sure, that focus is interesting and valid. In contrast, however, the Masson-Turtelboom paper acknowledges and raises some of the problems of transition from today’s arrangements to the brave new world. My instincts are that the hardest issues are transitional, and that it is those issues that merit our greatest attention in the next several years.
I now turn to some comments on the CEPII paper alone. Its analysis focuses exclusively on the consequences of EMU for the exchange rate of Europe vis-à-vis the rest of the world (in the model described as the United States, since there are only three countries in the analytical model). In a minute, I want to question that emphasis, asking whether that perspective is appropriate. But for a moment, let us accept that emphasis and consider several points about the paper’s approach and conclusions.
First, a clarification about the “volatility” of exchange rates. It may be helpful to remind ourselves that the CEPII paper gets at only part of what most observers think of when the words “volatility” or “variability” are used. The CEPII paper is not directly addressing the issue of higher frequency variations in exchange rates, such as daily or weekly variability. Instead, it is concerned with the changes in the “levels” of exchange rates that would be produced by various types of shocks, persisting or transitory. The paper also does not deal directly with the asset-substitution effects of the sort emphasized by (for example) the Bergsten and the Alogoskoufis-Portes papers for this conference.
My greatest concern about the analytics of the CEPII paper is that it tends to downplay the important differences among alternative monetary policy regimes that the ECB might employ. (And there is no discussion at all of alternative fiscal regimes in European countries.) Even in the text of the paper there are some important clues about how important this issue is. Contrast, for example, the results with the two alternative loss functions they analyze in the comparative-statics model. They report the result that, under the L loss function (which has the monetary authority targeting inflation in producer prices and the real exchange rate), the dollar-euro exchange rate will move more under EMU than under a pre-EMU float for all the stylized shocks they consider. But under the alternative Λ loss function (which assumes the monetary authority is targeting inflation in consumer prices and an output gap), the results are revealingly different for supply shocks originating in Europe. European supply shocks result in less, not more, movement in the dollar-euro exchange rate.
More generally, I would place even greater emphasis than the CEPII paper does on differences among alternative monetary policy regimes that might be adopted by the ECB. (The Masson-Turtelboom paper goes a bit further than the CEPII paper in this regard, recognizing more explicitly that the choice of monetary regime may have very important implications for macroeconomic outcomes.)
A further clue about the importance of this issue: when the CEPII authors present simulations from a dynamic version of the model—see their Figure 4—there are quite significant differences across the ECB monetary policy regimes that they use for their single illustrative simulation. In that figure, the differences across the two alternative monetary policy regimes seem larger and more noteworthy than the differences for any given regime between the EMU and pre-EMU float cases!
I feel fairly sure that the CEPII paper properly solves its model to generate the conclusions it reaches about the movements of the dollar-euro exchange rate. But one can wonder whether other considerations, not treated in the model, might lead to amendments in the conclusions.
For example, suppose one focused more attention on the size and degree of openness of the interacting economies. In raising this question, I am influenced by having seen a recent paper by Philippe Martin, a CEPII colleague of the authors. Martin comes to a conclusion that, after the establishment of EMU, the variability of the EU’s exchange rate against the rest of the world might be lower, not higher, than the variability of the dollar-deutsche mark exchange rate in today’s international monetary system. The line of thought suggesting this outcome emphasizes the relative sizes of economies as a determinant of exchange rate variability, as the literature on optimum currency areas does. A small, open economy tends to be buffeted proportionately more than a large, relatively closed economy by shocks originating outside the economy (“foreign” shocks are more salient than “domestic” shocks). Thus a large, relatively closed economy’s output variability and price variability tend to be influenced by exchange rate movements less than is the case for a small, open economy. Hence, a large, relatively closed economy’s macroeconomic policies can be less concerned with exchange rate volatility. In a model such as Martin’s, such considerations appear able to overturn the CEPII paper’s main conclusion about the post-EMU variability of Europe’s exchange rates.
Martin’s analytical model is different from the CEPII model in several important ways. And he focuses on nominal rather than real exchange rates, I have not examined Martin’s model carefully, and do not mention it because I regard his model as preferable to the CEPII paper’s model. I bring up the two papers’ contradictory conclusions only as a reminder that an emphasis on different factors could modify the CEPII paper’s main conclusion.
When considering the CEPII paper’s results about exchange rate movements, we should also remember that what is called in the CEPII model the Ȝdollar-euro rate” is in concept the exchange rate of the EU against the rest of the world (ROW) as a whole. In actual practice, we have to deal with a ROW that is many dimensional. The dollar-euro exchange rate will not behave the same as the yen-euro exchange rate, for example, and a variety of other countries will be independently floating as well.
To begin to come to grips with these hard analytical issues, we will have to work with an analytical framework that has four or five regions, rather than merely three. For example, for some purposes a model will have to have three European regions—“Germany,” the other “ins” that participate in EMU core from the outset, and the European “outs” (“pre-ins”) that initially are not in the core. And outside Europe, the model might distinguish between at least two regions, say, the United States (or North America) and a Japan/Asia/developing country region. A model with four or five regions will not be nearly as analytically tractable. It is much easier to define the need for such a model than it is to build it and use it. But such a model may be a minimum configuration for getting at some of the issues.
To illustrate, consider the question of incentives that non-Europeans—either private investors or even official governments—will have to substitute out of dollar-denominated assets into euro-denominated assets, and hence what sort of pressure, if any, these substitutions would bring to bear on exchange rates. Many conference participants appear to believe that the euro would appreciate sharply against the dollar, perhaps even overshooting an equilibrium path, in the years after EMU’s inception. My own conjecture is like that in the Masson-Turtelboom paper. Such substitution pressures on dollar and euro exchange rates are not likely to be the dominant forces on exchange rates (as suggested in the Bergsten paper). Other factors, such as the relative cyclical positions of the European, U.S., and Asian economies and the relative monetary policy stances of major central banks (including the ECB), are likely to be more important.
Even so, if we want to get an analytical handle on the substitution pressures, we will have to consider the “relative variability” of the euro and the dollar vis-à-vis third currencies. For example, if the euro were much more variable against the U.S. dollar than against Asian currencies, the incentives of asset holders within Asia and the developing countries to substitute euro assets for dollar assets would be greater than if the dollar and third currencies fluctuated against the euro by relatively similar amounts.
The general thrust of my comments about the CEPII paper’s analysis of the relative size of exchange rate movements of the euro vis-à-vis the ROW is cautionary. Conclusions about exchange rate movements could be significantly modified by factors that are not considered, or not considered fully, in the CEPII paper. (Again I cite the example of the initial asset-substitution effects that figure prominently in several of the other papers for this conference.)
I have so far accepted the CEPII paper’s emphasis on the consequences of EMU for the stability of exchange rates. But we also should ask whether that emphasis is really appropriate—whether it addresses the questions of greatest interest to us.
My inclination is to believe that the stability of other macroeconomic variables is usually much more noteworthy and important. For Europe itself, should we not be much more interested in the variability of inflation, output, and consumption? Analogously, if we want to study the implications of EMU for non-European countries, should we not be focusing on those countries’ inflation rates, outputs, and consumption paths?
When one looks at the CEPII paper’s results on exchange rate movements, the differential effects on exchange rates of the pre-EMU float and the EMU environments tend to be rather small in magnitude. I believe Jean Pisani-Ferry used the adjective “moderate.” At first glance I thought of the word “small.”
In a limited way, it is possible to make a direct comparison between the CEPII paper’s conclusions about exchange rate variability and the Masson-Turtelboom stochastic simulations. The revised Masson-Turtelboom paper in its Table 1 reports standard deviations for the deutsche mark–dollar exchange rate under the ERM and the euro-dollar exchange rate under EMU. The CEPII model and MULTIMOD stochastic simulations do appear to agree on the point that moving to EMU could augment the variability of the exchange rates of European currencies vis-à-vis the outside world. Interestingly, the MULTIMOD results also show the variability of the weighted average exchange rate of the dollar increasing if Europe moves to implement EMU, whereas variability for the yen decreases slightly. In the MULTIMOD stochastic simulations, however, note that the differences between the ERM and EMU cases also do not appear to be large.
Nevertheless, suppose the differential exchange rate effects reported in the two papers are deemed “large” in some sense. Even then, we should probably not be looking primarily at the exchange rate effects when trying to make welfare judgments. We would be looking, again, at consumption, or at least outputs and inflation rates.
The Masson-Turtelboom paper, in its sections describing the simulations with MULTIMOD, has a broader focus of the sort I have in mind. It reports the consequences for exchange rate variability, but more of its focus is on other macroeconomic variables. Their focus seems to me more interesting and appropriate.
The revised Masson-Turtelboom paper understandably does not highlight how the move to EMU and alternative monetary policy regimes within Europe might affect the stability of macroeconomic variables outside Europe. But there are a few results in their Table 1 that bear on the stability of variables outside of Europe. The European move to EMU appears to raise slightly the variability (measured by the standard deviation) of short-term interest rates in the United States and Japan. The reported consequences for real GDP growth and the inflation rate in the United States and Japan, on the other hand, appear quite small. It will be interesting to see whether further research on the effects of EMU on the United States and Japan confirms this finding of minor effects on outputs and inflation rates.
I have another suggestion for the Masson-Turtelboom paper, or more properly for extensions of their research that might be attempted in the future. It could be valuable to do further work on stochastic simulations of the sort they have carried out, but with an “unbundling” of the types of shocks. As implemented so far, the authors formed a variance-covariance matrix of all the key residuals for the 15 EU member countries and other major countries. All the 350 trials in the exercise (35 simulations for each of 10 years) involved random draws from the entire matrix for all the residuals. Hence, no distinctions were made between shocks originating inside Europe and those originating outside. And no distinction was made between demand shocks versus supply shocks versus shocks to the risk premium in the key interest-parity equations determining exchange rates.
In the Brookings-sponsored stochastic-simulation analysis of monetary policy regimes carried out several years ago, resulting in the volume on Evaluating Policy Regimes referred to by Masson and Turtelboom, the participating model groups also did not do such an “unbundling” of the shocks. But with the wisdom of hindsight, we very much wished that such an unbundling analysis had been done. Focusing separately and sequentially on a few key types of shocks, one type at a time, but with stochastic simulation techniques, would be a helpful halfway house between the deterministic simulations used at the end of the CEPII paper and the full-blown stochastic simulation analysis in the Masson-Turtelboom paper for this conference.
I would like to make a final point applicable to the analysis in both papers. The specification of monetary policy regimes, which are used in the simulations in both papers, can be very sensitive to the values of the “feedback parameters” that are used. (The feedback parameters are the coefficients in the policy rules that govern how aggressively or how mildly the policymakers move their instrument, the short-term interest rate, in response to deviations of their target variables from desired values.)
The values of the feedback parameters used in both papers are probably still fairly arbitrary, and the authors have probably not been able to subject them to much if any sensitivity testing. Other research on policy rules tends to underscore the importance of such sensitivity testing. Conclusions of simulation analysis can in some instances be changed quite substantially by varying the sizes of feedback parameters.
As an example, consider the inflation-plus-real-output targeting rule (in the last several years referred to as a “Taylor-type rule”). Both the CEPII and Masson-Turtelboom paper report some results with this type of monetary regime. Implicit if not explicit in most macroeconomic models, there is probably a “trade-off frontier” between output variability and price variability. Along this frontier, policy can only reduce price variability further by accepting somewhat greater output variability, and vice versa. But if the analyst implementing an inflation-plus-real-output targeting rule assumes relatively weak values of the feedback parameters for the rule, model simulations will yield fairly high values of both price variability and output variability. Moving to larger values of the feedback parameters can reduce price variability and output variability simultaneously. Misinterpretation of such results could lead to the incorrect impression that a trade-off frontier between price variability and output variability does not exist in the model.
My conjecture is that the conclusions in the two papers about the consequences of EMU for the stability of macroeconomic variables could be significantly influenced by the particular values of the feedback parameters chosen for the simulations. (The authors of both papers are aware of this issue. I emphasize it not as a criticism of the papers, but as a reminder of further research to be done.)
To conclude, I note again that both of the papers in this session are thoughtful and stimulating. They provide much food for thought, and help to carry us forward in our exploration of the possible consequences of European monetary union for macroeconomic stability.