|
|
|
Email this citation
A Strategy for Launching the Euro
Center for German and European Studies, University of California at Berkeley
February 1997
Abstract
This paper studies the constraints placed by the Maastricht Treaty on the rates at which member currencies will exchange against the Euro at the start of stage 3 of economic and monetary union (EMU). The paper shows that the stage 3 bilateral conversion factors for EMU member currencies must correspond to closing market exchange rates as of December 31, 1998; furthermore, currency conversion rates into the Euro cannot be determined until that date. Moreover, official announcements about intended conversion factors will carry no credibility with markets, as market rates must be chosen over any prennounced rates according to the Treaty. Unless there is heavy official intervention in the runup to stage 3, EMU members' bilateral market rates will exhibit excessive volatility and may induce beggar-thy-neighbor policy behavior. On the other, hand, exchange-rate targeting may open the door to speculative currency crises. The only feasible solution appears a widely-publicized institutional reform to subjugate national central banks' policies entirely to the goal of intra-EMU exchange stability in the final months of stage 2.
Maurice Obstfeld, Dept. of Economics, University of California, Berkeley
First draft: February 10, 1997
Revised: February 27,1997
On 1 January 1999, a core of European Union members will make a new European currency the basis for its unified monetary policy and payments system. Three years later, national currencies will be replaced entirely by the Euro. This paper discusses the available strategies for defining exchange ratios between the currencies of the first wave of EMU members and the Euro.
The first section below examines the Maastricht treaty's prohibition of a jump in the ECU's external value when the Euro is introduced. That rule alone, I shall argue, implies that the bilateral conversion factors derived from Euro conversion rates must equal the bilateral market exchange rates prevailing at the end of trading on 31 December 1998. The rule also implies, perhaps surprisingly, that earlier announcements about the rates to be set at the start of the third stage will carry no credibility. Unless amended, the Maastricht treaty thus may well force the authorities to choose conversion rates on 1 January 1999 quite different from those they might otherwise desire.
Section 2 expands on other drawbacks of this scheme and discusses the dangers of one way to overcome them, an intervention policy that rigidly targets bilateral exchange rates through the final months of the second stage.
Section 3 takes up the decision of the Madrid European Council that 1 Euro should equal 1 ECU at the beginning of the third stage. It is shown that this rule, too, forces the 1 January 1999 bilateral conversion rates volatilityto equal the closing bilateral market rates of the day before. Thus, this rule has all the drawbacks of the Maastricht treaty's rule on the ECU's external value. Because the 1:1 rule necessarily leaves the Euro's value uncertain until 31 December 1998, however, it leads to additional problems. Section 4 discusses a preferred strategy that would become available if the relevant prescriptions of the Maastricht treaty and Madrid Council could be revoked.
1. The Maastricht treaty and bilateral conversion rates at the start of Stage 3
The Treaty on European Union mandates that the 1 January 1999 adoption of conversion rates between the Euro and the currencies of the EMU members "shall by itself not modify the external value of the ECU" (Article 1091).
In this section I show that on a face-value interpretation of the Treaty, this requirement alone leaves the European Council no choice but to adopt on E-Day (Euro Day, 1 January 1999) Euro conversion factors such that the implied bilateral exchange ratios equal the market exchange rates prevailing at the foreign exchange market's close on Thursday, 31 December 1998. 2 I argue further that unless authorities are prepared to intervene so heavily as to peg bilateral "in" rates during the run-up to E-Day, even the most solemn promise to implement a desired set of bilateral conversion factors in the third stage need not drive actual 31 December 1998 rates anywhere close to the desired levels. Full discussion of intervention options is postponed until section 2.
Implications of the Treaty's mandate
The most natural interpretation of the Treaty is that the introduction of the conversion rates to be enforced by the ECB should not ceteris paribus imply a change in the exchange value of the ECU from its closing level of 31 December 1998. 3
Why does this stricture force the European Council to choose Euro conversion rates consistent with closing bilateral market rates for "in" currencies? Let ECU $ be the ECU's market value against an arbitrary currency (the dollar, say) at the last moment of Stage 2 when markets are open. The basket definition of the ECU implies
were a i is the number of i currency units in the basket and S $/i is the price of EU currency i in terms of dollars at the close of markets in 1998. Let there be I < 12 members of EMU, let the chosen conversion rates be C i/E units of "in" currency i per Euro, and let us initially focus on the DM, whose conversion rate against "in" currency i will therefore be C DM/i = C DM/E/C i/E. On E-Day the external value of the ECU implied by the new conversion rates, ceteris paribus, can be expressed as
Of course, the Maastricht treaty requires that the bilateral conversion rates for the ins satisfy
or, equivalently,
Now this equation alone doesn't obviously constrain the bilateral conversion factors very tightly. However, the preceding argument can be repeated for each of the I in currencies. For the French franc, for example, the conclusion is that
must hold, where the identity C FF/i = C FF/DM · C DM/i has been applied. But obviously the last two displayed equations cannot both hold true unless S $/DM = S $/FF · C FF/DM. By triangular arbitrage in the foreign exchange market, however, S $/DM/S $/FF = S FF/DM. Thus, invariance of the ECU's value with respect to external currencies requires that
that is, that the Euro conversion rates must reproduce the bilateral market rates prevailing at the last moment of trading (however defined) in 1998. It is clear that the same relation must hold for any pair of"in" currencies.
What explains this result? On E-Day the dollar value of the ECU's "in" component can be computed equivalently as the dollar/DM rate times the DM value of that component, or as the dollar/franc rate times the franc value, and these must be identical at the new conversion rates if both are to match the prior dollar ECU value set in the market. If the new franc/DM rate departs from its prior market value, however, the first identity will not follow, as it normally would, from triangular arbitrage.
For example, suppose EMU depreciates the French franc against the DM as compared with the end-1998 market outcome. Then at 1 January 1999 conversion rates, the dollar value of the francs contained in the ECU's "in" component (at end-1998 market dollar exchange rates) will exceed the dollar value of the Deutsche marks contained in the same ECU sub-basket (and similarly for every other external currency). In short, there will be no uniquely-defined measure of the ECU's "external value" at the new conversion rates unless these are consistent with the prior market bilateral rates.
One could avoid facing the problem by insisting that the Maastricht treaty had in mind only one of the (at least) I possible ways of evaluating the ECU's external market value on E-Day. But it is not at all obvious which one to choose, nor would agreement in the European Council be straightforward from a political point of view. In any case, this apparent latitude is removed by the Madrid European Council's edict of December 1995 that 1 Euro shall equal 1 ECU as of the start of Stage 3. I defer discussion of this additional constraint until section 3, as a number of drawbacks are already implied simply by the Maastricht treaty's Article 1091.
The preceding analysis implies that a plan to base Stage 3 conversion factors on long averages of past market exchange rates--what De Grauwe (1996) and De Grauwe and Spaventa (1997) call the "Lamfalussy rule"--cannot satisfy Article 1091, apart from other disadvantages that those authors identify.
The impossibility of credible announcements
Keen-eyed observers have perceived the necessity of adopting prior market bilateral rates at the start of Stage 3 (for example, Kenen 1996a). Sometimes they console themselves by arguing that if the bilateral exchange ratios authorities desire can be signaled credibly, stabilizing speculation will automatically drive the end-1998 market rates to the desired levels. Thus, the market outcome simply will ratify official intentions; the bilateral exchange rates at the start of EMU will not be left to chance. Unfortunately, the stipulations of the Maastricht treaty itself undermine this hope. 4
Let us tentatively accept the assumptions of credibility and market rationality. Even so, it is still not true, strictly speaking, that the end-1998 rates would necessarily be close to solemnly preannounced Stage 3 conversion factors. Assertions to the contrary are implicitly based on the notion that trading always occurs in continuous time, so that even minor anticipated discontinuous jumps in asset prices--which imply instantaneously infinite arbitrage profits--are precluded. However, trading will cease on Thursday 31 December 1998 and not reopen until Monday 4 January 1999 (3 January in Bahrain). An expectation of some exchange rate change over that period is not ruled out by arbitrage considerations, and the expected change could be economically significant if for some reason fundamentals are expected to change markedly once the ECB opens its doors.
A likely minor discrepancy between preannounced bilateral conversion factors and market exchange rates on 31 December 1998 may seem like a small matter, but it is of critical importance because in case of any discrepancy the Maastricht treaty's Article 1091 compels the European Council to choose the market rates. That circumstance will lead economic actors to base their actions entirely on what they believe market exchange rates will be at the close of 1998 trading, not on what the authorities say they will be at the start of 1999 trading. Thus, a credible commitment to announced bilateral conversion factors is logically impossible unless the participating authorities actually lock their bilateral rates at the end of 1998, before the third stage begins (a strategy whose perils are discussed in the next section). Indeed, it can be shown that absent the most aggressive intervention, the announcement of intended bilateral conversion factors might well have no affect whatsoever on the end-1998 bilateral rates. Appendix A develops the analytics of this point in detail.
Of course, irrational foreign exchange trading also would leave the European Council with no choice under the Maastricht treaty but to accommodate the market outcome--a circumstance not in itself likely to enhance credibility either.
2. Disadvantages of basing Stage 3 on end-1998 bilateral exchange rates, and the role of exchange-rate management
To build Stage 3 on the foundation of end-1998 bilateral exchange rates has several grave disadvantages, some of which have been raised already by other authors (e.g., Begg et al. 1991, several of the essays in Kenen 1996b, or De Grauwe and Spaventa 1997). In this section I describe these, and then ask whether close targeting of exchange rates through intramarginal intervention is an adequate answer to the problems.
What are the problems?
One drawback of simply going with end-1998 bilateral rates has not been much emphasized. That procedure may magnify exchange-rate volatility in the transitional period between the selection of EMU members and the beginning of Stage 3.
Shocks to the fundamental determinants of an exchange rate may have various degrees of persistence, ranging from zero (a purely temporary shock) to infinity (a permanent shock). 5 Suppose traders expect that an end-1998 bilateral exchange rate will be frozen by Stage 3. A zero-persistence shock to fundamentals will induce no additional volatility in the exchange rate, even in that case, because it has no impact on expectations of the end-1998 rate. (The exception is a zero-persistence shock occurring on the last day before Stage 3; see below.) At the other extreme, a permanent shock also induces no additional volatility: its effects are already expected to persist into the indefinite future whether or not the exchange rate is frozen at the end of 1998.
Extra volatility comes, however, from shocks of intermediate persistence, because some fraction of their effects, which normally would die away completely over time, instead becomes embodied forever in the 31 December 1998 exchange rate. In effect, their persistence is enhanced, and because the resulting increment to persistence is greater the closer is the end of Stage 2, a given shock will have a progressively more pronounced exchange rate impact as 1999 approaches. The extreme case is that of a purely transitory shock occurring on 31 December 1998 itself. Stabilizing speculation will not mute the shock's impact, which will be as large as if it were a permanent change. Appendix B contains the full technical analysis, as well as a numerical example showing the possibility of substantially increased volatility in the weeks leading up to the third stage. 6
There are other reasons, moreover, to worry about increased volatility just prior to Stage 3. The prospect of EMU is likely to make the participating currencies more readily substitutable as transactions media, a development which could sharply heighted the variability of exchange rates (Kareken and Wallace 1981; Weil 1991). Whatever its causes, increasing exchange-rate volatility between prospective EMU members can only undermine credibility and heighten any political tensions generated by the final preparations for consummating EMU.
Market exchange rates are subject to exogenous forces, but they also respond to official policies. A second drawback of setting Stage 3 bilateral conversion factors equal to recent bilateral market rates, therefore, is an associated moral hazard problem. Once the core EMU members have been chosen, the exchange-rate stability convergence criterion becomes inoperative. While unilateral realignments can be ruled out, some countries might be tempted to depreciate their currencies within the existing ERM band. This beggar-thy-neighbor strategy would allow them to begin Stage 3 with more competitive real exchange rates. It would also render their public debts smaller when redenominated in terms of Euros. As Froot and Rogoff (1991) have stressed, national central banks might feel less need to safeguard low-inflation reputations at this stage, as they are about to be placed under new management.
A last disadvantage of delaying the specification of the bilateral conversion factors governing Stage 3 is the creation of needless uncertainty, which in turn could delay the achievement of precisely the kinds of trade gains that a single currency is supposed to yield. It is true that risks due to an uncertain future DM/FF rate, say, could be eliminated in the forward exchange market. But a portion of the gains from using a common currency (indeed, the most tangible and readily measurable ones) come from channeling into more productive uses some of the resources currently employed in forward exchange trading. It seems odd to use a strategy for launching the Euro that delays the exit of excessive labor and capital from the financial sector.
Intervention to peg exchange rates
In view of these problems, it may appear attractive for the participating central banks to adopt narrow bilateral exchange targets, which they defend through intervention from the naming of the initial EMU entrants through the start of Stage 3. Indeed, I argued at the end of the last section that this may be the only way to ensure the credibility of an announced set of bilateral parities, given Article 1091. This strategy is workable, at least in principle, if both countries concerned with a particular bilateral rate intervene symmetrically by extending intervention credits in unlimited amounts and forbearing from sterilization. In practice, however, it is rash to assume that credible monetary coordination can commence before the institutional framework ensuring credibility--the European Central Bank--has been put into place.
The major feature underpinning the credibility of the ECB's commitment to convert currencies at par during the 1999-2002 transition period is the consolidation of monetary decision-making power and national central bank assets in the hands of a supranational institution, combined with a seigniorage allocation formula that is removed from national balance of payments developments. 7
This feature will not obtain in the final months of Stage 2. According to Article 23.3 of the Protocol on the Statute of the European Monetary Institute appended to the Maastricht treaty, all intervention claims and liabilities arising from ERM support operations "shall be settled by the first day of the third stage." Thus, such claims will not simply be netted out on the books of the ECB on January 1, 1999. Intervention operations prior to that date will have natinal fiscal implications to which governments will not be indifferent. A particular perverse fiscal incentive would exist for countries that intervene in support of weaker currencies, refraining from sterilization and allowing their national money supplies to rise. By sterilizing instead, selling securities for domestic money, they can reduce the stock of national assets that must be effectively delivered into the ECB's hands at the start of Stage 3. This avoidance does not directly affect the national government's financial position--it parts with the securities in either case--but it is not a matter of indifference whether the income from those assets accrues to the ECB or to its own citizens.
One can envision a self-fulfilling crisis scenario in which a bilateral parity is attacked on the last day of 1998, but the "weak-currency" partner, perhaps hoping to secure the competitive or fiscal advantages of depreciation, reneges on its intervention commitment, leaving the "strong-currency" partner liable for the entire intervention burden. Since such intervention has a national welfare cost, it may not be carried out à outrance, and any resulting slippage in the market rate must be ratified after the fact if the letter of the Maastricht treaty is obeyed, to the enrichment of the speculators. There may be other reasons why intervention commitments could be viewed as noncredible prior to E-Day. It is conceivable that in some countries the central bank and the political leadership will take different views of the wisdom of proceeding rapidly to EMU, perhaps in light of the prospective list of first-round members. In that case, making pegged exchange rates the centerpiece of the transition process would give national central banks immense power to pursue independent agendas. 8
If the bilateral parities being targeted in the last half of 1998 are not entirely credible, however, they may be subject to speculative attack. The consequences would disrupt the economies involved, impair planning for Stage 3, and sap public confidence in and support for monetary unification. Under Article 1091 of the Maastricht treaty, an especially severe risk arises on the last day of Stage 2, 31 December 1998. Speculators will know that whatever bilateral exchange rates prevail at the end of the day must be maintained in Stage 3. Thus, central banks will not be willing to retreat easily, and can be expected to commit substantial resources to a battle with speculators. Since the bet can go only one way, such a commitment increases what speculators can expect to earn in a successful attack on the last day of Stage 2, while perhaps reducing the likelihood that they succeed. If the first factor dominates, however, then the bilateral rates of prospective EMU members will become particularly vulnerable to attack on the last day of Stage 2. By the usual backward induction argument, this fundamental vulnerability may emerge even earlier in the transition to EMU.
A less drastic intervention plan, advocated by De Grauwe and Spaventa (1997), is for the "in" countries to steer bilateral rates more gradually toward mutually agreed levels, announced or unannounced. Gros and Lannoo (1996) have sketched a similar scenario. The more flexible and discretionary is this steering process, the more it will be open to the problems of the last subsection. The more rigid it is, however, the greater the invitation to speculative attack. Presumably there is an optimal point on the tradeoff, but it is unclear for now how to find it in practice. An alternative to targeting specific preannounced rates would be to announce a range of possible conversion factors for the start of the third stage, but in that case the benefits of certainty would be compromised.
In any event, to hit a specific preannounced set of target conversion factors for the start of Stage 3, the authorities will have to peg the corresponding bilateral rates at the desired levels, with no margin of error whatsoever, on 31 December 1998. To discourage destabilizing speculation as the latter date approaches, they may find it necessary progressively to narrow the allowable fluctuation margins for bilateral rates. This necessity, in turn, would effectively imply the strategy of pegging within rather narrow bands, discussed above.
3. The Madrid Council's 1 Euro = 1 ECU rule
The Madrid Council decreed in December 1995 that upon the new currency's introduction the constraint 1 Euro = 1 ECU must hold. It is hard to interpret this constraint as applying to any ECU value other than the final market value of 31 December 1998.
But an analogous question to the one raised in section 1 above arises now: in what numeraire(s) must this constraint hold? Presumably the answer is: in any currency one chooses. If so, then the Madrid Council decree itself implies that bilateral conversion factors for the start of the third stage must equal the most recent bilateral market exchange rates. This, as we have seen, is necessary and sufficient for the start of the third stage to cause no jump in the external value of the ECU. In addition, however, the Madrid decision places a constraint on the value of the Euro itself, something the Maastricht treaty does not do.
Let ECUDM be the market value of the ECU in terms of DM at the close of trading for 1998, and let CDM/E again be the chosen conversion rate for DM into 1 Euro. In order for the constraint 1 Euro = 1 ECU to hold when both currencies are measured in DM, the following must hold (in section 1's notation):
where EuroDM is the value of 1 Euro in terms of DM. This constraint must hold as well, however, in terms of any other "in" currency, for example, the French franc. Thus
where triangular arbitrage among the end-1998 market rates has been assumed. The last two equations imply that
In words, the general constraint 1 Euro = 1 ECU again implies that bilateral conversion factors must equal end-1998 bilateral exchange rates, simply because it is the obvious intent of the Madrid European Council directive that the equivalence of the ECU and Euro should be independent of the yardstick used to measure their respective values.
The observations made in the last section all apply to the 1 Euro = 1 ECU rule, even in the absence of the Maastricht treaty's prohibition on a jump in the ECU's external value. Thinking further about how the 1:1 rule would have to be operationalized reveals some additional consequences. The only way to implement this rule is to set CDM/E=ECUDM , CFF/E=ECUFF , etc., on E-Day, where the ECU values in terms of "in" currencies are market values at the close of 1998 trading. Clearly this procedure both guarantees that 1 Euro = 1 ECU, regardless of how the two currencies' values are expressed, and that bilateral Stage 3 conversion factors equal end-1998 bilateral exchange rates (ECUFF/ECUDM=SFF/DM=CFF/E/ CDM/E=CFF/DM , where triangular arbitrage in the foreign exchange market guarantees the first equality).
An immediate consequence of the 1:1 rule, however, is that the "in" currency rates against the Euro are quite unlikely to be known before 31 December 1998. De Grauwe and Spaventa (1997) have demonstrated that in all likelihood conversion rates of the "in" currencies into Euros cannot be chosen in advance of E-Day, and still be expected satisfy the constraint that 1 Euro = 1 ECU when the third stage begins. The reason is that the currencies of the "outs," which are included in the ECU basket, will be likely to fluctuate unpredictably in the interim. There are two distinct types of intervention scheme that might allow pre-fixing of the Euro conversion rates. First, all the "outs" could peg rigidly to the currencies of the prospective Euro area, whose members, in turn, peg to the ECU. But as there is no requirement (or incentive) for the "out" currencies in the ECU basket to peg, this outcome is extremely unlikely. A second strategy would be for Germany, say, to peg the DM to the ECU sub-basket consisting of "out" currencies, while the other "ins" peg their currencies to the DM. Needless to say, however, it is more than unlikely that the Bundesbank (or any of the other ins) will be willing to subordinate its monetary policy, even briefly, to those of the countries that did not qualify for EMU! And of course, any of these tightly defined intervention schemes opens the door to a currency crisis. For all practical purposes, therefore, the 1:1 rule guarantees that ultimate Euro conversion factors will be random.
Does this additional implication of the 1:1 rule really matter? Suppose it were somehow feasible to eliminate uncertainty about the bilateral conversion factors in the third stage, say through concerted intervention. The remaining uncertainty would concern only the scale or "size" of the Euro, i.e., whether it contains 1.9 DM, 2.1 DM, etc. But varying the scale chosen for the Euro amounts to a completely neutral currency reform, since it changes the internal price level measured in Euros and the exchange rate against external currencies in exact proportion.
The unpredictability of the Euro's scale matters in one important way, however. Uncertainty over the relationship between the Euro and national "in" currencies prior to the third stage will retard the process through which agents in the economy become accustomed to calculating in terms of Euro. The result will be to prolong the attachment to national currencies, and probably to make more costly or even delay the introduction of the Euro at the retail level.
The requirement that 1 Euro should equal 1 ECU thus has an unfortunate implication for the transition to the third stage of EMU. One must presume that a major intent of this requirement is to ease the translation of existing ECU-denominated contracts into Euro terms. If so, the measure is unlikely to be of much help, since the ECU and Euro will again be likely to diverge after the first moment of Stage 3, complicating this translation in any case (see Appendix C). One certainly would not want to assure all current holders of claims to future ECU that they will paid off in equal numbers of Euros. That would be too generous, as the presumption must be (and must have been when the ECU contracts were signed) that the ECU, which still will contain "out" currencies, will depreciate against the Euro as long as some of the 12 signatories to the Maastricht treaty remain outside EMU.
4. What is to be done?
Given the disadvantages summarized above, I strongly urge the European Commission to reconsider the wisdom of the Article 1091 stipulation on the ECU's external value and of the 1:1 conversion rule. By removing these constraints, which have no obvious economic rationale, the EU could instead adopt a strategy for launching the Euro that provides a window of certainty both with respect to bilateral conversion rates and Euro conversion rates, while not exposing the "in" currencies to disruptive speculative crises prior to the third stage. Given the period of learning likely to attend the inauguration of the ESCB and an integrated EMU payments system, it seems advantageous to eliminate as much uncertainty as possible, as early as possible. The current rather arbitrary constraints on the Euro's introduction prevent this.
After revoking the preceding constraints, it would be possible to proceed as follows.
- Upon the naming of the first wave of EMU members, the need for a general, and final, realignment of their ERM central rates could be decided. Such a move can actually raise credibility if it represents a significant move back toward equilibrium (see Drazen and Masson 1994); but on present data it is hard to make a case that such a realignment will be needed.
- Once central rates against the ECU are set, these should be announced as the conversion factors into the Euro, to be implemented on E-Day. This announcement should be viewed as a binding commitment. Policymakers can r complicating this taise the credibility of the promise by encouraging the private sector to reconfigure software on the basis of the announced rates, by changing their own internal accounting systems accordingly, etc. An important benefit of knowing the conversion factors in advance is that existing as well as newly issued government liabilities maturing in 1999 and after could immediately be redenominated in Euros, before the start of Stage 3, thus removing part of the major motivation for an ex post choice of conversion factors different from the one announced in mid-1998. Credibility would be further enhanced by such redenomination.
- The ± 15 percent fluctuation bands of the current ERM should be retained until the third stage. The participating authorities may wish to prevent drastic departures of exchange rates from central rates, as De Grauwe and Spaventa (1997) suggest, but they should not offer a target to speculators or to political detractors of the EMU project by staking their prestige on narrow exchange-rate targets. In any case, given the high degree of inflation convergence among the prospective first cohort of EMU entrants and the sluggishness of inflation, it is unlikely that any significant misalignment of central rates will emerge in the six to eight months between the selection of initial members and the beginning of the third stage.
Under the preceding approach, the conversion rates introduced on 1 January 1999 will probably differ somewhat from the market rates the day before, but not by much. Gros and Lannoo (1996) argue that private sector actors will not regard the conversion rates as fair unless they correspond precisely to market rates, and that such a perception could lead to legal challenges down the road. Apart from the extreme implausibility of the hypothesis that private agents ever have or will in the future accept market outcomes as equitable ipso facto, exchange rates are not determined independently of government policies, and any realized rates at the end of the second stage are likely to be perceived as resulting from sins of commission by governments, or from sins of omission--actions that could have been taken to affect the market result but were not. Feigning passive acceptance of the market outcome does not let governments off of the hook of public opinion.
The legal issue is perhaps more serious, which is why it is all the more pressing to amend current decisions formally so as to allow an orderly advance agreement on the conversion factors ultimately to be adopted. If the conversion rates have been announced well in advance, and are implemented as promised, there will really be little ground for private-sector compliants. Only those who speculate against the authorities will take substantial losses.
Appendix A The impossibility of credible announcements about Stage 3 conversion rates
Section 1 claimed that the Maastricht treaty's provisions preclude credible announcement of the bilateral conversion rates to be set at the start of Stage 3. Only by pegging the 31 December 1998 exchange rates at precisely the desired Stage 3 levels--a task which may prove difficult, as I argued in Section 2--can authorities ensure a specific constellation of conversion factors for EMU.
A convenient (if somewhat oversimplified) vehicle for illustrating these points is the monetary model of exchange rates (see Mussa 1976). This model states that the equilibrium date-t exchange rate S t obeys the difference equation
where E t(·) is a conditional expectation and the random variable M t represents the exchange rate's "fundamentals" on date t. Above, λ is defined as η/(1 + η), where η is the nominal interest rate semielasticity of real money demand.
Let us suppose that the authorities announce in advance that the bilateral conversion rate S 1 T is to be implemented on date T = 4 January 1999 (the first trading day of EMU). Then if time T - 1 corresponds to the time of the market's close on 31 December 1998, and on the tentative assumption that market actors take the announced conversion factors as 100 percent credible, the market exchange rate S T-1 will be
There are two immediate consequences of this result:
- There is no presumption that S T-1 = S 1 T. Making this substitution above, one can see that the two rates will be the same only if, through pure chance or through official intervention, the fundamentals are so aligned that M T-1 = S 1 T.
- The Maastricht treaty requires the rate S T to be set equal to the market exchange rate (1 - λ)M T-1 + λS 1 T rather than to S 1 T. This fact means that the tentative assumption that the announced rate was credible is untenable.
We can find the "rational expectations" exchange rate in the following way. Any market actor who thinks all others believe in S 1 T will expect the conversion factor
to materialize instead. This is another way of seeing that the announcement S 1 T generally will not be credible absent official adjustment of the fundamentals. If one agent finds it rational to believe S 2 T, however, so will they all. But if all agents believe S 2 T, each individual will figure out that the Maastricht treaty really implies the conversion factor will be
Continuing to argue in this manner, one sees that the only credible exchange rate announcement is that the Stage 3 conversion factor will be the fixed expectations point
or
This is a crucial result. Given the provisions of the Maastricht treaty, market participants will expect the conversion factor to equal the 31 December 1998 fundamental, regardless of any supplementary pronouncements the participating authorities might make. One implication of this finding, or course, is that the ending level of the fundamental will have an inordinately large influence on the Stage 3 conversion factors specified on 1 January 1999. Indeed, market expectations about possible future levels of fundamentals will play no role whatsoever in "smoothing" the exchange rate. Instead, it will be up to the authorities to smooth fundamentals themselves through intervention.
When the trading interval is of arbitrary length h (it was assumed above that h = 1), then the equation of monetary equilibrium becomes a generalization of the one used so far,
where λ in equation (1) above is interpreted now as η/(h + η). It is clear that as h → ∞, so that the limit of continuous trading is approached, S T-h → E t-h(S T). This result implies that in continuous time, closing exchange rates for 31 December 1998 will indeed approach the rates promised for the start of Stage 3, if those rates are believed. This is the result asserted in some of the literature, but I have shown that it does not generally hold without continuous trading.
One might argue that trading time is all that matters, and that trading time is continuous notwithstanding the closure of markets for several days after 31 December 1998. But the discrete institutional shift coming into effect on 1 January 1999, including the opening of the ECB, make that position untenable. Market actors will perceive a sharp demarcation line between the final trading seconds of 1998 and the first ones of 1999.
Appendix B Excessive volatility of bilateral exchange rates prior to the third stage
This appendix demonstrates the claim in Section 2 that bilateral exchange-rate volatility for the "ins" is likely to be accentuated by the knowledge that conversion rates in the third stage will equal 1998 closing market values.
Continue to assume the exchange rate equilibrium condition of equation (1) in Appendix A, let date T once again be the first trading day of the third stage, and let t be any prior date. Then as is well known (see, for example, Mussa 1976), the equilibrium exchange rate can be written as
where S T is the conversion rate to be set inn Stage 3. (Thus, the date t < T market exchange rate is a weighted average of fundamentals between dates t and T - 1 and the rate expected to prevail from the start of the third stage on date T).
Appendix A showed that the rational expectation of the conversion factor for the third stage is
under the Maastricht treaty. Suppose the fundamental M t follows an autoregressive stochastic process of the form
where 0 ≥ ρ ≥ 1 and εt is a random white-noise shock. Then for j + t ≥ T - 1,
while
From equation (2), it therefore follows that for j + t ≥ T - 1,
In contrast, under a hypothetical free float with no prospect of EMU (which a perpetual regime of ± 15 percent bands approximates), the exchange rate solution would be
It is evident from comparison of equations (3) and (4) that unless ρ = 0 or 1, equation (3) implies a more sensitive response of the exchange rate to the fundamentals than does (4), and hence higher volatility in the exchange rate. (Even if ρ = 0, however, exchange rate volatility eventually becomes sharply higher, but only on date t = T - 1.) It is also apparent that because the factor (λρ)T-t grows as t approaches T (recall that λρ < 1), the excessive volatility (compared to a free float) will grow as the third stage nears.
Figure 1 shows the ratio of actual to "normal" (i.e., free-float) volatility when the model parameters are λ = 0.95 and ρ = 0.7. Just before Stage 3, volatility is nearly seven times what it would be absent the prospect of a monetary regime shift. While the precise numerical settings used in the figure have been chosen in part for their dramatic effect, the results still are indicative of what could happen.
Appendix C Problems of redenominating ECU contracts in Euros at the start of Stage 3
In its December 1995 conclusions the Madrid European Council decreed that:
The substitution of the Euro for national currencies should not of itself alter the continuity of contracts, unless otherwise provided in the contract. In the case of contracts denominated by reference to the official ECU basket of the European Community, in accordance with the Treaty [on European Union], substitution by the Euro will be at the rate of one to one, unless otherwise provided in the contract.
This appendix illustrates difficulties of redominating ECU contracts in terms of the Euro, given that the ECU basket contains currencies that will not be pegged to the Euro on 1 January 1999.
Consider a previously issued bond promising the stream of ECU payments
payable (say) on January 1 of 1999, 2000, 2001, 2002, etc. In terms of the notation of Section 1 (and ignoring any distinction between the public and private ECU), the bond is a claim to a i P t units of currency i on date t, for each of the 12 currencies comprising the ECU basket.
If 1 Euro = 1 ECU on 1 January 1999, then bondholders will receive the 1999 payment promised to them if instead of getting P 99 ECUs they get P 99 Euros. But the only way to redenominate this bond in Euros while maintaining the "continuity" of the contract is for subsequent payments to recognize changes in the exchange rate of the ECU against the Euro that will in all likelihood occur after the start of 1999. Let there be I < 12 "in" currencies, let C E/i,t be the conversion rate of "in" currency i into Euros in Stage 3 (Euros per currency unit), and let S E/i,t be the market exchange rate of "out" currency i against the Euro on date t. Then on date t in the third stage, 1 ECU will be worth
Figure 1
Euros. The Madrid Council ruling requires that the sum above should correspond to 1 Euro at the very beginning of Stage 3, but since the "out" exchange rates against the Euro will probably fluctuate thereafter, the 1:1 equivalence will not hold for long. In that probable case, the date t Euro payment on the ECU bond should be
and not P t Euros to safeguard the continuity of the original agreement.
The preceding analysis implies that the Euro payments due on an ECU bond after the start of Stage 3 are actually random variables rather than known quantities. Thus, a simple value-preserving redenomination scheme for an ECU bond would convert it into a risky Euro security. A much more complicated alternative is to convert it into a (safe) Euro bond the present value of whose payments equals the market value in terms of Euros of the ECU bond.
References
Begg, David, et al. (1991). Monitoring European integration: The making of monetary union. London: Centre for Economic Policy Research.
De Grauwe, Paul (1996). "How to fix conversion rates at the start of EMU." Discussion Paper Series, no. 1530, Centre for Economic Policy Research (November).
De Grauwe, Paul and Luigi Spaventa (1997). "Setting conversion rates for the third stage of EMU." Photocopy (February).
Drazen, Allan and Paul R. Masson (1994). "Credibility of policies versus credibility of policymakers." Quarterly Journal of Economics 109 (August):735-754.
Froot, Kenneth A. and Kenneth Rogoff (1991). "The EMS, the EMU, and the transition to a common currency." NBER Macroeconomics Annual 6: 269-317.
Gros, Daniel and Karel Lannoo (1996). The Passage to the Euro. Centre for European Policy Studies Working Party Report No. 16, Brussels (December).
Kareken, John and Neil Wallace (1981). "On the indeterminacy of equilibrium exchange rates." Quarterly Journal of Economics 96 (May): 207-222.
Kenen, Peter B. (1996a). "Sorting out some EMU issues." Jean Monnet Chair Paper No. 38, Robert Schuman Centre, European University Institute (December).
Kenen, Peter B., editor, with John Arrowsmith et al. (1996b). Making EMU happen, problems and proposals: A symposium. Princeton Essays in International Finance, no. 199 (August).
Mussa, Michael (1976). "The exchange rate, the balance of payments and monetary and fiscal policy under a regime of controlled floating." Scandinavian Journal of Economics 78 (May): 229-248.
Rolnick, Arthur J. and Warren E. Weber (1989). "A case for fixing exchange rates." Federal Reserve Bank of Minneapolis 1989 Annual Report: 3-14.
Weil, Philippe (1991). "Currency competition and the evolution of multicurrency regions." In Alberto Giovannini and Colin Mayer, eds., European financial integration. Cambridge, UK: Cambridge University Press.
Notes
Note 1: Professor of Economics, University of California, Berkeley, U.S.A. I thank Luigi Spaventa for several e-mail exchanges and for access to his unpublished notes on the Euro. Back.
Note 2: The Treaty would seem to imply that the relevant closing exchange rates would be those in the New York market. In practice the requirement may be finessed by defining the relevant external value of the ECU as its average for December 31, 1998, or over that day and the preceding three, etc. As the argument of this section will make clear, such averaging would not satisfy the letter of the Maastricht treaty, since triangular arbitrage of currency cross rates does not hold with respect to arithmetic averages. Back.
Note 3: This claim is controversial. But it is hard for me to see any other meaningful interpretation. One might hope that an E-Day realignment could be designed to leave the market value of the ECU on the first trading day of the third stage at its 31 December 1998 value, but because the former value depends on the rates of external currencies against EMU member currencies, the desired result could be guaranteed only if the European Central Bank intervened initially to peg the ECU's market value. It is difficult to believe that this is what the drafters of the Maastricht treaty had in mind. In any case, if the Maastricht treaty itself does not constrain Stage 3 conversion factors to equal final market bilateral rates, the Madrid Council's rule, discussed in section 3, certainly does. Thus, the basic thrust of my analysis still applies. Gros and Lannoo (1996) seem to interpret the Maastricht treaty in the same sense as I do. Back.
Note 4: De Grauwe and Spaventa (1997) point out that a decision taken at the Dublin European Council (December 1996) requires the E-Day conversion rates to be expressed as rates against the Euro. Bilateral conversion rates between any two "in" currencies are to be derived by converting the first currency into Euros, then converting the resulting sum of Euros into the second currency. The reality of the situation, however, is that if the constraint 1 Euro = 1 ECU is to be respected, conversion rates into the Euro will have to be chosen so that they yield end-1998 market exchange rates when used to derive bilateral conversion factors. Nor would the Dublin constraint in itself preclude authorities from informally announcing the bilateral conversion factors that their ultimate choice of euro conversion factors will imply. Back.
Note 5: Technical note: I rule out persistent shocks to the growth rate (rather than level) of fundamentals. I also ignore shocks with negative serial correlation. Back.
Note 6: The mathematical reasoning yielding these results is related to that underlying De Grauwe's (1996) discussion of the Lamfalussy averaging rule. However, that rule causes exchange-rate volatility to spike upward (upon adoption) and then decline over time. Back.
Note 7: This same structure links the United States regional Federal Reserve banks, see Rolnick and Weber (1989). Back.
Note 8: Gros and Lannoo (1996) rightly observe that the national central banks of the "ins" can carry out nonsterilized interventions without changing the EMU-wide monetary aggregates. They conclude that the EMU-wide aggregate will become more important as 1999 nears, and that governmental commitments to large-scale nonsterilized interventions in support of exchange-rate targets therefore will become more credible. However, the declining importance of nation-specific money targets pertains only to certain aspects of exchange-rate credibility. Back.
