Better Than the Euro? The European Monetary System (1979–1998)

ABSTRACT The euro crisis has provoked a debate on the pros and cons of adjustable exchange rate regimes that enable their participants to negotiate nominal de- and revaluations. To evaluate the functioning of such regimes, we revisit the EMU’s predecessor, the European Monetary System (EMS). We show that in the EMS, devaluations did indeed help more than revaluations did hurt. Assuming that the political-economic heterogeneity of the Eurozone will not vanish in the foreseeable future, the move to a more flexible exchange rate regime might therefore be economically advantageous. However, a purely economic view ignores the huge political ‘maintenance costs’ of negotiable realignments, costs that the EMS members aimed at overcoming when they opted for the euro. The re-politicization of nominal exchange rate policy in today’s Eurozone would therefore not end transnational political conflicts in the Eurozone but fuel new ones.


Introduction
The euro area has entered into difficult waters. Its core problem is finding solutions for the transnational macroeconomic imbalances that occurred since the introduction of the Euro (Scharpf 2013, Johnston and Regan 2016, Walter 2016. Today, the euro consists of a number of overvalued euros in the South and undervalued euros in the North. The usual answer to such constellations are nominal de-and revaluations. Given that such adjustments are impossible in a monetary union, critical economist such as Flassbeck and Lapavitsas (2015) have argued in favour of returning to an adjustable exchange rate regime such as the European Monetary System (EMS) that existed between 1979 and 1998.
The EMS aimed at sheltering its members from erratic financial markets, but also enabled its members to de-and revalue their currencies if necessary. The economic preferability of discretionary exchange rate systems under conditions of political-economic diversity, however, is far from undisputed. De-and revaluations, critics such as Mabbett and Schelkle (2015) argue, had only limited success in solving economic problems and rather contributed to new long-term distortions. We will revisit the EMS and show that the 62 de-and revaluations that were conducted among its members between 1979 and 1998 actually helped to minimise transnational macroeconomic imbalances. This insight, however, just opens up yet another puzzle: If the discretionary solution fits better into a heterogeneous Europe, why did the governments of the member states opt for the change from the EMS to EMU in the first place?
This article suggests that the answer lies in the trade-off between political and economic costs of exchange rate regimes. While we know today that the Euro's economic maintenance cost are much higher than its founders had expected, we tend to forget that the maintenance of the EMS was extremely costly, tooin political terms. There was, in other words, a discrepancy between the functioning of the exchange rate regime as a provider of macroeconomic stability, on the one hand, and the constant emergence of political conflicts while maintaining it, on the other hand. The EMS achieved economic results but failed on the politics. The monetary union thus appeared to be a relief on the political side of this trade-off by depoliticising currency politics through the removal of national currencies. Under conditions of political-economic heterogeneity, we conclude, exchange rate regimes fail to deliver political and economic expediency at the same time.
We will begin by summarising the features of the EMS (the second section). This gives us the basis for examining the political-economic processes in the EMS in its various phases (the third section). The next section (the fourth section) takes a closer look at the 62 exchange rate adjustments made in the EMS. To conclude our empirical observations, the fifth section looks at the political conflicts that accompanied the adjustment decisions. By way of illustration, this section selects a typical revaluing country (Germany) and a typical devaluing country (France) from the EMS countries. The concluding sixth section contains a summary of the findings of the empirical sections, which are then interpreted in light of the debate surrounding the future of the euro.

The structural elements of the European monetary system
The EMS was a discretionary exchange rate system which provided for quasi-automatic mandatory intervention to reduce exchange rate fluctuations, on the one hand, andthis is the discretionary elementpolitical negotiations on the redefinition of the parities, on the other hand (for more details, see Collignon 1994, Bernholz 1998). The EMS existed from 1 March 1979 up to the point where exchange rates for euro area countries were fixed at the turn of the year 1998/1999.
The crucial element of the EMS was the exchange rate mechanism (ERM), although the member states could opt out of the ERM if they had a valid reason. 1 The rationale behind the ERM was to peg national exchange rates to an internal accounting unit called ECU with a possible ±2.25 per cent deviation from the respective ECU parity. These permissible so-called 'fluctuation bandwidths' of ±2.25 per cent did not apply to all the participating currencies, however: the Italian lira (and later the Portuguese escudo, the Spanish peseta, and the British pound sterling) were granted a wider margin of ±6 per cent. In August 1993, the permissible bandwidths were widened to 15 per cent on each side for all the participating currencies.
The participating central banks had two instruments with which to keep their own and other exchange rates stable within the limits of the set fluctuation margins. On the one hand, if a market exchange rate reached the borders of the fluctuation bands (the upper and lower intervention points), the central banks were compelled to intervene on currency markets to keep the exchange rates within the bandwidths ('obligatory interventions'). On the other hand, however, if 75 per cent of the possible fluctuation margin to the ECU parity was reached, the affected member states and central banks were expected to perform what are known as intra-marginal interventions. In theory, the capital required for obligatory interventions was unlimited and in principle had to be continued until such point as the exchange rate in question had stabilised or the parity adjustment had been renegotiated.
The interventions were carried out symmetrically by those central banks whose exchange rate parities were affected by the deviations. 2 The central banks of the countries whose currencies reached the lower intervention points had to buy their own currency by selling foreign currencies, while central banks of the countries whose currencies reached the upper intervention points had to buy foreign currencies. To make sure the national central banks had sufficient capacities for such market operations, the European credit facilities were extended with the help of the European Monetary Cooperation Fund (for details on this, see Bernholz 1998: 797-802).
The actual discretionary element of the EMS was its ability to adjust the parities between the individual currencies, as well as between the currencies and the ECU. While the aforementioned mandatory intervention on the part of the central banks was triggered by objective signals on the currency market, the adjustment of the currency parities was not subject to any comparable quasi-automatic processes. Permission to alter the paritiesto carry out revaluations and devaluations of the EMS currencies beyond the limits of the bandwidthshad to be applied for with the Economic and Financial Affairs Council (Ecofin Council) and had to be granted unanimously. In other words, the EMS united elements of both fixed and flexible exchange rate systems: short-term currency fluctuations were minimised with the help of interventions, but if the bandwidths turned out to be unfeasible due to short-term fluctuations, they could be adjusted by means of discretionary political interventions, a measure that had to be employed frequently, as will be shown below.

The EMS phases
As seen in the previous section, the architects of the EMS created a currency regime based on reciprocal intervention and assistance obligations for the participating countries. Moreover, the countries involved had the opportunity to tackle the root cause of the imbalance, for example, through measures such as internal deflation. If such steps failed to work, or if they were not carried out at all, there was also the option of reaching a mutual agreement on exchange rate adjustments. Table 1 shows that the discretionary function of the EMS had to be used frequently. A total of 62 de-and revaluations were negotiated in the EMS at 17 different points in time. 3 In addition, various countries joined and left the EMS temporarily (European Commission 2005).
At first glance, it can be seen that the adjustments are not randomly distributed across the years covered and the countries involved. It is possible to identify turbulent periods, (relatively) quiet phases, and, of course, typical revaluing and devaluing countries. With respect to the countries, Table 1 displays a pattern which is familiar from today's imbalances in the euro area. Germany and the Netherlands have exclusively revalued; other countries -Belgium, Denmark, and Luxemburg have revalued more than they have devalued. In the case of Italy, in contrast, the exchange rate adjustments were invariably devaluations. The same was true of Spain and Portugal, which joined the ERM in 1989 and in 1992 respectively. On the whole, France, too, was a devaluing country. But interestingly, all French devaluations took place in the first half of the EMS period.
The first four years of the EMS were marked by considerable instability and left doubt as to whether greater exchange rate stability between the participating countries was conceivable at all. The rates of inflation were so heterogeneous that maintaining the parities within the defined bandwidths seemed neither feasible nor desirable. 4 As many as 23 of the documented 62 exchange rate adjustments were made in this early phase of the EMS. The sheer scope of the negotiated exchange rate adjustments, which in many cases were above 5 per cent, is also remarkable. Three of the devaluations alone were carried out by France (see the fifth section).
For the EMS, the end of France's expansive policy and the subsequent policy of désinflation competitive (competitive disinflation, also known as the French tournant de la rigueur) implied greater inflation convergence. Anti-inflation programmes were also implemented in Italy, the participating country with the highest level of price instability. 5 In the years 1985 and 1986, all the EMS countries were recording single-digit inflation rates, although this relative convergence still masked clear differences that called for exchange rate adjustments. 6 Since our main interest lies in the implications of the EMS experience for the euro, the phase that followed, a quiet phase in the EMS that lasted from mid-January 1987 to early September 1992, is especially remarkable. In these five and a half years, the EMS countries dispensed with exchange rate adjustments almost entirely, with the sole exception of a moderate devaluation of the lira in January 1990. In a way, we can therefore interpret this phase as a pre-emption of the Economic and Monetary Union (EMU). 7 The purported end to tensions in the EMS benefitted the short-term enlargement of the EMS (Spain in 1989, the UK in 1990, and Portugal in 1992) as well as the 60 0 * * Note: All values represent changes of the member state's ECU-parity in per cent. *means that the respective country was not an ERM-member at the time. Source: European Commission (2005). medium-term plan to complete the process of introducing the euro. Overall, the stable phase seemed to indicate that the discretionary element of the EMS was no longer needed and that the EMS countries were in a position to peg their nominal exchange rates irrevocably.
But what was happening below the surface of the nominal exchange rate stability? The trend toward increasingly similar inflation rates among the EMS countries continued. The effects of this relative convergence, however, were cancelled out by the fact that no realignments occurred and remaining inflation differences, therefore, cumulated over a longer time period. In 1992, the last year of the quiet phase in nominal exchange rates, the price competitiveness distortions and real economic imbalances in the EMS area were basically no different than in any of the years leading up to the EMS where exchange rate adjustments were made. 8 Two further incidents caused tension beneath the, in terms of nominal exchange rates, quiet surface of the EMS. Contrary to international trends, in 1991, the reunification boom year, Germany saw growth of 5.2 per cent and a rate of inflation of 4.0 per cent; in 1992, this growth had slumped to 1.5 per cent while inflation had reached a rate of 5.1 per cent. By German standards, these were exceptional price increase rates although they were still less dramatic than those of Spain, Portugal, and Italy. At first glance, the reunification boom relieved the tension in the system, because it reduced the inflation range of EMS countries.
The EMS-wide consequences of the Bundesbank's monetary policy, however, turned out to be disastrous. In July 1992, in response to the spike in inflation, the German central bank raised the interest rate to what (in today's terms) is an inconceivable level of 8.75%. To contain capital flight to Germany, the central banks of the other EMS countries had no choice but to copy the rigid monetary policy of the de facto EMS-anchor, Germany. This meant raising interest rates despite their faltering economic conditions and, thus, doing the opposite of what was deemed macroeconomically necessary at the time. Consequently, the Bundesbank led Germany's neighbouring countries into a recession that culminated in 1993.
From mid-1992 on, these tensions reached a climax due to the unavoidable currency market interventions by the central banks and the return to the exchange rate adjustment mechanism that had been unused for several years. Between July and September 1992, the German Bundesbank interventions in support of other EMS currencies reached an all-time high of 87.2 billion deutschmarks. 9 This led the Bundesbank to urge the German government to push the finance ministers of the other EMS countries toward a comprehensive exchange rate adjustment.
The data in Table 1 give an impression of the events of September 1992. To begin with, all the EMS countries except Italy revalued their currencies by 3.5 per cent -Italy devalued by a further 3.5 per cent. 16 September 1992 went down in European economic history as Black Wednesday. Following considerable foreign exchange losses on the part of their central banks, the UK and Italy announced their decision to leave the EMS. The next day, Spain devalued the Peseta by 5 per cent. Not two months later, further 6 per cent devaluations occurred in Spain and Portugal. By mid-May 1993, substantial devaluations were additionally conducted by Ireland, Spain, and Portugal. Considerable instability ensued, as did further extensive interventions by the central banks, leading the governments and central bank presidents to extend the EMS bandwidths from ±2.25 per cent to ±15 per cent in late July 1993. According to several observers, this was tantamount to the de facto suspension of the EMS (see Busch 1993: 532).
Nonetheless, the governments of the EMS countries adhered to their plans to see the EMU through to completion, and indeed the EMS did stabilise within the new, wider bands. Between 1996 and 1998, Austria, Finland, and Greece joined the EMS, and Italy rejoined at the end of 1996. By late 1998, only two negotiated devaluations had been recorded: these occurred in March 1995 in Spain and Portugal. One of the convergence criteria set down in the Maastricht Treaty required countries wishing to be part of the euro to remain within the then-existing fluctuation margins for at least two years. For financial market players, this desire to stay within the given margins was credible, or defensible at the very least, putting an end to the speculative attacks seen in the early 1990s.
The observations made in this Section show that there was frequent occasion to make use of the discretionary scope of the EMS. Given the at times very high inflation differentials, the premature transition to a fixed exchange rate regime would have surely been a mistake. And similarly, clinging to the original time frame for the introduction of the euro seemed rather risky in light of the turbulent phases that followed in 1992/1993, only months after the Maastricht treaty had been signed in February 1992. A macroeconomic view on the exchange rate adjustments In this section, we examine the negotiated exchange rate adjustments in greater detail, removing them from their specific historical contexts and comparing them using selected key indicators. Our focus is on macroeconomic variables. Table 2 shows average key indicators of the 'magic square'price stability (inflation), the current account balance, economic growth, and the employment rate in each case, at the time of the exchange rate adjustments as well as five years before and afterwards. The interpretation of these values is, however, complicated by the fact that the selected key indicators during the period of the EMS were affected by time trends, as outlined in the third section above: the years 1979-1998 were a period of general disinflation. In order to be able to interpret the averages against the overall trends, we also show how the relevant key indicators developed, at the very same points in time, in all ERM countries (see the numbers in brackets in Table 2).
As in the third section, we will focus on inflation in particular because if nominal exchange rates remain constant, insufficient inflation convergence invariably leads to adjustment pressure. Developments in real interest rates are not considered here. It is evident that the devaluing countries in fact show a distinctly above-average inflation rate not only in the year of the actual exchange rate adjustment, but also in the years prior to this. Moreover, in contrast to the revaluing countries, no relative disinflation trend can be observed in the devaluing countries in the years preceding the adjustment. What happened in the following years? Nominal devaluations are a consequence of above-average price increases, but at the same timeif increasing import prices trigger wage-price-spiralsthey are a potential source of new spikes in inflation. In reality, however, devaluations negotiated in the EMS were generally followed not by reflation but by disinflation, which was even more pronounced than the disinflation of revaluing countries and also more pronounced than the overall time trend. The devaluing countries have been striving, at least relatively successfully, to minimise future pressure to devalue (see also the sixth section on this). However, the figures in Table 2 also show that despite successful relative disinflation, the devaluing countries always had a positive inflation differential against both the revaluing countries and the EMS average. In other words, the devaluing countries in the EMS managed to reduce future pressure to devalue, but not to eliminate it completely.
The current account balance data presented in Table 2 can be interpreted parallel to the inflation data. As is to be expected, the devaluing countries generally show a current account deficit during the exchange rate adjustment years, while the revaluing countries generated current account surpluses. While the current account of the devaluing countries was negative on average in the years preceding the exchange rate adjustment, no clear trend can be observed for the revaluing countries. In the devaluing countries, the exchange rate adjustments were mostly followed by a reduction in the current account deficit, in other words: by relative success. Nevertheless, the current account of devaluing countries remained consistently in the red. Pressure to either adjust economic policies or to devalue even further, therefore, was never completely eliminated. The corresponding development in the revaluing countries is worth noting, too. These were apparently able to cope well with the relative increase in prices of their tradable goods compared to other EMS participants. They were even able to further increase their current account surpluses over the course of time, which again suggests that revaluations in the EMS tended to be followed by new revaluation pressure. 10 We can conclude that while the exchange rate adjustments tended to help the devaluing countries to come closer to a balanced current account, the realignments have at least not generally proven detrimental to the revaluing countries as far as their current account balances are concerned.  (2015).
The data on real economic growth in Table 2 show that the devaluing countries tended to record declines in their original above-average (catch-up) growth before the exchange rate adjustments. These adjustments were then generally followed by increases in economic growth. While this is in line with our expectations, it should be noted that the revaluations were not generally followed by mirror-image slumps. It should of course be borne in mind here that in the revaluing countries (as is evident from the inflation data), revaluations cumulated in new upward pressure, in other words, a trend toward new real underpricing shortly afterwards.
The data on employment rates paint a similar picture to those on real growth. Not unlike the inflation data, these, too, can only be usefully interpreted by first examining the general time trend: the EMS existed during a period of increasing inclusion of women in the labour market and, consequently, of rising employment rates. Bucking this trend, average employment rates of the devaluing countries stagnated in the years preceding the exchange rate adjustment, before subsequently rising slightly. The revaluations also do not appear to have been detrimental to the revaluing countries with regard to growth in employment. Against the background of the exchange rate adjustments, their employment rates grew approximately in line with the trend at the time and, contrary to expectations, even slightly more favourably after the adjustments than prior to them.

Political conflicts in discretionary exchange rate regimes
In the previous sections, we focused on the economic rationale behind the EMS, ignoring the political tensions accompanying the decisions required to run this system. We will examine these below, first by turning our attention to the domestic dimension, looking at the conflict between the German government and the Bundesbank. Next, we will demonstrate the transnational dimension of the tensions and the lack of political expediency of the adjustment mechanisms using the example of the conflict between France and Germany regarding the realignment in March 1983.

The conflict between the German government and the Bundesbank
The EMS was established on the initiative of the heads of the German and French governments, Helmut Schmidt and Giscard D'Estaing (using elements of the 1974 Fourcade Plan, see Hoffmeyer 2000: 18f.). What is striking about the history of the EMS is that the central banks in general and the German Bundesbank in particular did not participate, for instance, in the preparations for the Bremen council meeting in July 1978 where the initial planning took place. Helmut Schmidt and Finance Minister Hans Matthöfer were afraid the German central bank would be obstructive to the plans if they were involved at an early stage. It was only after the key features had been politically negotiated that the German government informed the Bundesbank (Emminger 1986: 357;Marsh 1992: 194-5).
Why did the architects of the EMS anticipate resistance from the German central bank? The Bundesbank had been fixing money supply targets since 1974. However, it proved impossible to reach an agreement on how to achieve money supply targets and the planned mechanisms for stabilising the exchange rates without conflict. If central banks intervene because their own exchange rate is getting close to the lower intervention point, they run the risk of losing their currency reserves and having to make use of credit facilities. The central banks of countries with strong currencies do not have this problem because they intervene in their own currency. But new money injected into the market counteracts money supply targets if they exist, as in the case of the Bundesbank. In general, Helmut Schmidt was suspected by the German Bundesbank of disregarding stability objectives in favour of economic policy considerations, as well as general objectives of European integration (Emminger 1986: 364). Consequently, it could be safely assumed that the Bundesbank Council would interpret the plans for the EMS as attacks on the autonomy of the central banks and the primacy of the stability policy.
According to Bernholz (1998: 800), in the early stages of the EMS preparations, the Bundesbank attempted to find allies among German stakeholders including the labour unions. Their objective was to maximise the discretionary function of the EMS, i.e. the option of making exchange rate adjustments, and, conversely, minimise the obligation to intervene. The nearer they got to this objective, the closer the EMS was de facto to a flexible exchange rate regime. 11 In June 1978, the Bundesbank sent a letter to the German Chancellor pointing out the limits of the obligation to intervene: an obligation that was capable of undermining the Bundesbank's stability objective could not be reconciled with the Bundesbank's mandate. In fact, the German Bundesbank managed to obtain the Chancellor's agreement to release it from the obligation to intervene in favour of weak currencies in the face of imminent violation of the stability objective (Bernholz 1998: 801-2;Stoltenberg 1997: 236). It was only after this assurance that there was a majority in the Central Bank Council in support of the government's objectives. The following illustrates just how serious the tensions were during the preparatory phase of the EMS: as Emminger (1986: 363) reports, German Chancellor Schmidt had agreed to visit the Central Bank Council in fall 1978 to clarify the status and background to the EMS preparations. Otmar Emminger (Bundesbank president from 1977(Bundesbank president from to 1979 advised the Chancellor against visiting the Bundesbank: each attempt to sway opinions in the Central Bank Council could be counterproductive and result in even more vociferous opposition. 12 It could be argued that conflicts concerning the structure of the EMS are not necessarily an indication that its modus operandi was also accompanied by internal political conflicts of the type described. In fact, however, these types of conflict were a leitmotif running through the history of the EMS. A discretionary exchange rate regime can be viewed as an institutional set-up whose position between the poles of 'fixed' and 'flexible' becomes the subject of negotiation in each individual case. For instance, the Bundesbank repeatedly urged the German government to enter into international negotiations to put an end to the obligation to intervene through exchange rate adjustments. The Bundesbank for its part was regularly subjected to criticism both at home and abroad when it carried out tough interest rate decisions with the aim of achieving internal price stability, thus leaving its neighbours with the choice of either following suit or having to devalue (see Baltensberger 1998). 13 The tensions between the government and the Bundesbank are just one example of the potential for domestic conflict surrounding discretionary exchange rate regimes. In general, the use of the discretionary scope has to be justified to stakeholders and voters (Schmidt 1990: 233). For politicians, who thus find themselves in the line of fire of clashing interests, this is a drawback of discretionary flexible regimes compared to both completely flexible exchange rate regimes (in which the nominal exchange rate is a market outcome) and currency unions (in which nominal exchange rates are fixed anyway). More so than other monetary regimes, discretionary exchange rate regimes in 'normal operation' (as opposed to in crisis situations) are consequently exposed to the tensions between opposing domestic interests as well as being forced to make unpopular decisions that need to be justified. 14

Negotiation of the realignment of March 1983
Let us now turn to the transnational dimension of the conflicts surrounding negotiated exchange rate adjustments. In order to illustrate the amount of political energy which the participating governments had to expend on conflict management, we will look at one example, namely the dispute over the exchange rate adjustments in March 1983, represented by the data on Round 7 in Table  1 and which primarilybut not onlymanifested itself as a conflict between Germany and France. 15 The tensions between them are of particular interest for the purposes of the present paper because these countries' commitment to integration since the Treaty of Rome has served as the engine of European integration. At the same time, the two countries represent two different views on the desirable interaction between the state and the economy which outlasted the EMS and still continue to exist to this day in the context of the euro (Kauffmann and Uterwedde 2010: 13-4): the prevailing French view insists on the primacy of politics over the economy, is interventionist, and generally oriented towards the domestic economy, and, conversely, the German view is geared towards stability, rule-based, relying on autonomous authorities, and export oriented. 16 In the early 1980s, against the backdrop of the second oil price shock and the attempt by the Mitterand/Mauroy government to counteract the problems affecting the French economy with a Keynesian expansive economic policy, France had to contend with two-figure inflation rates as well as an inflation differential against their German neighbour that was persistently over six percentage points (see Hall 1986: chap. 8). Consequently, the French franc had already come under repeated pressure to devaluate and the parity of the franc was adjusted downwardsonce in October 1981 and again in June 1982. France found itself in the same situation again in the first few weeks of 1983. In principle, the view had already become widespread that another exchange rate adjustment was inevitable. One unresolved issue, however, was the distribution of the burden of adjustment among the EMS participants (Hoffmeyer 2000: 58). At home, the French government was urged to work towards a revaluation of the currencies of the deutschmark bloc and consider leaving the EMS rather than see its own economic policy thwarted or have to devalue its own currency.
What happened during the negotiations on the distribution of the burden? Accounts of these events can be found both in the memoirs of the German Finance Minister at the time, Stoltenberg (term of office: 1982-1989, Stoltenberg 1997 and in the memoirs of the French Economics and Finance Minister at the time, Delors (term of office: 1981Delors (term of office: -1984Delors 2004: chap. 5). In the first half of March, Mitterand aspired to achieve a fast revaluation of the deutschmark, having 8-9 per cent in mind. Stoltenberg travelled to France on 17 March to present the negative German position on this to the French President. So as not to give any indication of an imminent realignment to the financial markets, this visit was kept secret. Stoltenberg landed at a military airfield and was taken to the Elysée Palace via a normally unused side entrance. Becausealso for reasons of confidentialityno interpreter was to be used, a trusted friend of Mitterand's translated what his guest said to him, with great difficulty as Stoltenberg recalls. While Germany for its part was willing to revalue its currency, it called for France to reciprocate with a move to devalue its own currency, something Mitterand wished to avoid.
That same week, the participating finance ministers and central bank presidents met in Brussels to negotiate the modalities of the exchange rate adjustment. France was eventually planning to devalue its currency and subsequently pursue a strict stability policy. In the eyes of the French, the lion's share of the blame for the troubles in the EMS lay with the Germans. If the German side would not budge, France said they would leave the EMS (Delors 2004: 180-1). 17 According to Delors, France was supported by the representatives of Belgium and the UK. The 48-hour negotiations were repeatedly interrupted by bilateral meetings and consultations between the negotiating parties and their governments. These negotiations ended in the compromise shown in Table 1: a revaluation of the deutschmark by 5.5 per cent, slight revaluations of the Dutch, Danish, Luxembourg, and Belgian currencies, devaluations of the French franc and Italian lira by 2.5 per cent and of the Irish pound by 3.5 per cent.
These descriptions illustrate the political energy that has to be invested in maintaining a discretionary exchange rate regimeand explain why, despite dubious success at convergence (to say the least), replacing the EMS by a single currency with a multilateral central bank was nevertheless an attractive option. The EMS worked moderately well (in the words of Hall 2012: 356), but due to its domestic and international conflict intensity, the discretionary moment of the EMS was not only a welcomed economic buffer, but also a political burden. 18 The simultaneity and intertwined nature of such conflicts multiply to produce a constant complexityinevitable in a discretionary exchange rate regimethat can apparently only be controlled by means of arduous political efforts. Hence, in the trade-off between political and economic expediency, the EMS achieved economic results but failed on the politics.

Discussion
In the present article, we have taken the debate on exchange rate adjustments as a starting point for a review of the European Monetary System. Due to the difficulty of resolving the real exchange rate distortions occurring from 1999 onward solely by means of internal adjustment measures (disinflation and reflation), some of the participants in the debate proposed the transition to a new discretionary exchange rate regime. Our retrospective study and our insights on the trade-off between economic and political expediency allow us to cautiously hypothesise on how such a regime would work today.
Looking at the twenty EMS years, it is striking how frequently its member countries had to make use of exchange rate adjustments (see the third section). Since inflation convergence sufficient enough to justify fixed exchange rates was obviously absent not only during the EMS years but also since the introduction of the euro, it is fair to conclude that realignments would often occur in a new EMS, too. As in the old EMS, re-and devaluations would always only temporally curb imbalances, without being able to prevent them from recurring in the medium term. If, however, the objection to discretionary exchange rate regimes is raised that nominal devaluations are pointless, counterproductive in fact, because they fuel inflation (Schulmeister 2013: 108), we fail to find empirical support. The problem in the EMS was rather that the general disinflation trend did not produce inflation convergence sufficient enough to bring about more nominal exchange rate stability.
With respect to current account imbalances, the nominal exchange rate adjustments were not in vain as well. Hassel (2014: 9) assumes that the current account imbalances observed in the euro area would have also accumulated in the context of a new EMS. We only find modest support for this view. It is true that even in the EMS, countries remained in deficit over extended periods of time. But it was possible to keep these deficits und surpluses within relatively tight limits. Compared to the current account imbalances in the euro area, the conditions in the EMS were clearly advantageous. In particular, there was no scenario in the EMS where one country -Germanywould have been able to achieve a current account surplus of over 4 per cent 13 years in a row (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) and even over 6 per cent in 10 of these 13 years. Moreover, double-digit current account deficits (five euro years in Greece and seven in Portugal) never occurred in the EMS. In terms of ability to keep imbalances of this kind within limits, the available data clearly indicate that the discretionary regime of the EMS was economically preferable to the EMU.
With only these economic aspects in mind, it seems to be somewhat of a mystery in hindsight why the EMS participants decided in the 1990s to abolish the discretionary exchange rate regime in favour of a monetary union (a clear overview is provided by Sadeh andVerdun 2009, see also Frieden 2014). A look at the chronology of events in the EMS reveals that the signing of the Maastricht Treaty coincided with the end of the five-year quiet phase in the EMS during which it could be reasonably assumed that the need for nominal exchange rate adjustments was a thing of the past (the third section). But the question arises as to why the Maastricht resolutions were not revised after it had become obvious that the countries concerned had not achieved a level of inflation convergence that was able to justify the irrevocable fixing of the exchange rates.
One conclusion could be that the rational strategic competence and forecasting abilities of the euro's architects should not be exaggerated (see Walter and Willett 2012). Interestingly, however, precisely this point is fiercely disputed by participants in the debate such as Moravcsik. According to this interpretation, the euro's founders in southern and northern Europe had both strong forecasting abilities and a rational capacity to actand 'both groups got what they wanted out of the euro': 19 The south not only a multilateral central bank instead of the Bundesbank but also the opportunity to accrue private and public debt under more favourable conditions, and the north free rein for durable real undervaluations and resultant competitive edges on the export markets. If, however, these effects were anticipated, why not also the upheavals that ultimately arose from this scenario?
Rather than by assuming strong forecasting abilities on the part of those involved, we argue that it makes more sense to focus on the level of dissatisfaction with the status quo ante. This dissatisfaction was driven by a sense of unease with the constant political conflicts which inevitably went hand in hand with keeping the discretionary regime runand which would presumably also accompany the establishment of an EMS II. Yes, we argue, the respective political actors had to decide under conditions of far-reaching uncertainty. But the point is not so much that politicians at the time wrongly expected overall economic gains without cost. The point is rather that they faced a sharp trade-off between the political and economic expediencies of their decisions.
We illustrated the political cost of maintaining the economically viable discretionary exchange rate regimes in the fifth section. The heads of government, finance ministers, and central bank presidents involved all had reason to view the prospect of a never-ending series of politically negotiated exchange rate adjustments as a nightmare scenario, and so were justified in their decision to save themselves this by irrevocably fixing the exchange rates and making the transition to a common currency. Although the EMS provided the means for macroeconomic adjustment, the high political 'maintenance costs' provided a strong incentive for the involved governments to opt for monetary union and hope for the economic side to work out eventually.
These insights make it clear that the ideal exchange rate regime for heterogeneous Europe does not to exist. As an irony of history, however, the lack of economic expediency of the monetary union (Höpner and Lutter 2017) led to such a severe crisis that the conflicts surrounding the politics of macroeconomic adjustment during the EMS re-emerged even more dramatic during the negotiations over European rescue packages and the overall crisis management. Eurozone governments now seem to be locked in a regime that neither provides economic nor political expediency.
Is the policy implication that euro members should dismantle the monetary union and move back or perhaps, forwardto an EMS II? Our reexamination of the EMS and our implicit comparison to the euro give no indication as to the likely transition cost from one regime to the other. Obviously, the amount of such cost would have to be a core factor in the decision. But at the very least, our analysis encourages us to advise the participants of the euro debate not to ignore that option. If transition costs turn out to be manageable, a planned, step-by-step transition of the Eurozone might be preferable to both the status quo and the danger of a politically uncontrolled break-up of the euro.
However, we also emphasise that one should not expect too much from discretionary exchange rate regimes. Such a regime may indeed fit better to a heterogeneous Europe. But it would still be a regime that would aim at stabilising nominal exchange ratesfor good reasonsand that would therefore still constrain domestic economic policy. Any regime solution will always require concessions with respect to either internal room for manoeuvre or external exchange rate stability, and any thinkable regime change will definitely disappoint those who think that the move to another regime may completely and permanently make this conflict of aims disappear. Also, as we showed in detail, keeping a discretionary exchange rate run would require permanent political investment. But since the euro clearly failed not only in providing economic stability, but from 2009 onwards, also with respect to political maintenance costs, the need for permanent political investment should, in our view, be among the smaller problems of a possible transition to an EMS II. Notes 1. For example, after joining the European Community, Spain and Greece decided not to participate in the ERM immediately. 2. The obligation of symmetric intervention, however, was only a result of many years of conflict-ridden negotiations between the EMS member states (see Collignon 1994 andHoffmeyer 2000). 3. The events of mid-September 1992 are regarded as one point in time here. 4. In 1980, the year after the second oil price shock, the EMS countries experienced rates of inflation between 5.4 per cent (Germany) and 21.1 per cent (Italy). Until 1982, France had recorded double-digit rates of inflation, too. 5. The Italian rate of inflation went from an exorbitant 18.0 per cent in 1981 to 10.8 per cent in 1984. For a critical account of the Italian (and Spanish) experience in the EMS, see Perez (2000).