The European Monetary System is often cited as an example of effective international policy coordination. CEPR and the Department of Economics, Manchester University, jointly organised a recent Anglo-French conference on the EMS, policy coordination and exchange rate systems. The conference took place on 27-28 September at Wooton Hall, Manchester, and the 60 participants were drawn from France, Italy, and the United States as well as the United Kingdom. Financial support was provided by the Economic and Social Research Council, the French Cultural Delegation and the Maison des Sciences de l'Homme.

How does the EMS affect the impact of external disturbances on the member countries? In the first paper of the conference, 'The Dollar and the European Monetary System', Francesco Giavazzi (Venice) and Alberto Giovannini (Columbia) analysed the impact of a dollar appreciation on European domestic price levels. They found the effect to be substantial when monetary policy was accommodatory. They also constructed a theoretical model to analyse European policy responses to 'supply shocks' such as fluctuations in the value of the dollar. The model assumed two identical European economies, 'France' and 'Germany', linked by a fixed exchange rate which floated against the rest of the world. Both countries were assumed to be concerned with stabilizing employment and inflation. The essential feature of their model was the assumption that Germany set its money supply independently while France was then obliged either to adjust its money supply passively in response to interest rate differentials or to alter the exchange rate parity. The model suggested that cooperation in policy-making was preferable to non-cooperative behaviour. Parity realignments in response to exogenous shocks were still necessary, however, even with complete harmonization of monetary policies.

In his discussion of the paper, David Vines (Cambridge and CEPR) explained that the inferiority of non-cooperative policy was due to the over-reaction by each country to the dollar appreciation. This arose because each country failed to take into account in setting its own policy the policy adjustments of the other country. Vines also argued that parity realignments were necessary in the model because of the different inflation- unemployment trade-off in each country, arising from the asymmetrical effects of exchange rate changes. Vines questioned whether there was any guarantee that the non-cooperative solution was a stable one. Other participants suggested that it would be worth examining the implications of assigning particular policy instruments to different countries.

Inevitably there is a need for parity realignments in a system such as the EMS. In 'Exchange Rate Expectations and Interest Parity during Credibility Crises - The French Franc, March 1983' Sue Collins (Harvard) analysed market expectations concerning the timing of a devaluation by focusing on changes in the term structure of interest rates on assets denominated in different currencies. She argued that political and economic news conveying information about the expected timing of a devaluation would be reflected in shifts in the relative term structure of interest rates between countries. By assuming uncovered interest parity, she was able to construct from the relative interest rate structure a measure of the likelihood of devaluation in each period as perceived by the market. Evidence from the Franc- Dollar exchange rate suggested that large changes in the perceived likelihood of devaluation were induced by economic and political news. This news reflected the credibility of the exchange rate policy and, in particular, central bank activities and movements in the spot exchange rate.

The discussion pointed out that Collins's procedures depended on the assumption that all assets bore the same risk premium. Collins herself noted the measurement problems, especially the determination of the particular date at which news becomes available and is incorporated into expectations. It was generally felt, however, that the planned extensions of her analysis to cover longer time periods and to incorporate additional 'news' variables would be worthwhile.

It has often been argued that the EMS has brought about the convergence of its members' monetary and fiscal policies. Xavier Debonneuil (Banque de France), in 'The Effectiveness of Monetary and Fiscal Policies within the EMS: Reflections Based on the Case of France', examined the extent to which membership of the EMS was consistent with the pursuit of divergent policies by member countries. During the period 1981-83, expansionary budgetary policy in France was accompanied by three successive devaluations. With this experience in mind Debonneuil discussed a model in which exchange rates were governed by an (uncovered) interest parity relationship and inflation rates by an expectations-augmented Phillips Curve. Expansionary fiscal policy generated an increase in output and prices, which fed back into inflation through the expectations mechanism. The long-run effects of the policy depended crucially on the size of the initial fiscal stimulus and the presence of a threshold effect in the response of exchange rate expectations. At low rates of inflation, the expected exchange rate remained unchanged and the economy gradually returned to its initial output level. A larger fiscal stimulus pushed inflation past the point where expectations of a devaluation were triggered. This affected price expectations and set off a cumulative process which led inevitably to devaluation. If the expansionary policy continued, increasingly frequent realignments would be necessary until, eventually, the currency floated. Debonneuil argued that while the EMS may be sufficiently flexible to permit small and short- lived policy divergences, membership of the system was incompatible with broad or protracted divergence.

In his discussion, Salvatore Rebecchini (Banca d'Italia) argued that the EMS had provided a strong incentive for policy convergence. In practice, parity changes had been the result of collective decisions and had been accompanied by a package of supportive policies. Was it possible to estimate how long divergent policies could be pursued? Rebecchini argued that this was worth further study, as were the possible asymmetries between decisions to devalue and revalue.

The effectiveness of the EMS in stabilizing exchange rates and promoting convergence of economic policies was also examined in the paper by Michel Galy (Banque de France), 'An Empirical Evaluation of the Monetary Integration Process among EMS Members'. Galy first investigated how much the EMS had reduced exchange rate volatility. The evidence indicated that while variability in nominal exchange rates had been reduced, the results were less clear for real exchange rates. He argued that this implied some convergence in interest rates and in the monetary policies of member countries. This could be attributed to the spread of restrictive monetary policies and the increased integration of world capital markets, however, developments which were not directly related to the EMS. Galy further examined the convergence of monetary policies by estimating central bank 'reaction functions' for four member countries. These results indicated a significant increase in active intervention. This, he argued, was consistent with his earlier findings of reduced volatility in nominal exchange rates.

The rationale underlying these estimated reaction functions was questioned in the subsequent discussion. In particular, the linkages between the individual reaction functions were unclear and needed to take into account the behaviour of the entire system.

Jacques Le Cacheux (OFCE) writing on 'Asset Substitutability, Relative Size and Interventions in an Exchange Union', explored the effect of asset substitutability and the size of asset markets on an exchange rate union between two countries. His analysis concentrated on financial behaviour, using a portfolio balance model, in which asset markets cleared instantaneously. The model was used to examine the effect of various asset market disturbances on one country's interest rate and on the union's exchange rate with the rest of the world under differing assumptions about the type of intervention used to maintain a fixed exchange rate between member countries. Le Cacheux's results demonstrated the importance of both the relative size of the two asset markets and substitution between assets in determining the behaviour of the union. The author suggested that it was easier for a small country to respond to asset market disturbances because it enjoyed more flexibility in its choice of intervention policies.

The discussion indicated a clear desire, shared by the author, to see the analysis extended by combining the portfolio balance framework with a model describing the real economy.

The abolition of exchange controls on UK residents in 1979 has been a source of political controversy. Alec Chrystal (Sheffield) examined its effect on interest rates in his paper, 'An Investigation of Some Aspects of the Impact of the Abolition of Foreign Exchange Controls in the UK Financial System'. He discussed a range of empirical evidence, including a comparison of inter-bank rates in the UK with those elsewhere, an examination of portfolio movements since 1979, and tests of causality between domestic and off-shore interest rates. There was no conclusive evidence of the direction of change in UK interest rates resulting from the abolition of exchange controls. Causality tests indicated an interaction between domestic and off-shore interest rates which appeared to have increased in both strength and speed since the abolition of controls.

Lionel Price (Bank of England) pointed out several difficulties in interpreting these results. Chrystal's data suggested that any convergence between UK interest rates and those of other countries reflected a more general trend and not merely the abolition of exchange controls. In addition, the rather mixed results presented on real interest rate movements appeared highly sensitive to the data used. Using a different interest rate series and a different proxy for inflation expectations, Price produced equally inconclusive, but very different results. Finally, the evidence of a one-day lag between the Euro-sterling and the London inter-bank rate raised questions concerning lags in the data itself. One participant expressed surprise at the ambiguity of the results. His own work had clearly indicated that interest rates converged after capital controls were removed and diverged after they were imposed.

Several papers explored the consequences of cooperative and non- cooperative behaviour among countries. These grew out of new research into dynamic game-theoretic approaches to policy evaluation, which emphasize policy interdependence among nations.
Analysis of non-cooperative behaviour typically focuses on two types of solutions - the so-called 'Nash' (or Cournot-Nash) solution, in which individual players take the policies of other players as given, and the 'Stackelberg' solution in which one dominant player takes into account the effect which his policies might have on all the other players, each of whom continues to take his policy as given. These solution concepts can be further refined by considering both an 'open loop' version where policies are represented by instrument settings and the 'closed loop' or feed back version where policies are represented by rules.

In the first of these papers, 'Policy Design in Interdependent Economies: The Case for Coordinating US and EEC Policies', Andrew Hughes-Hallett (Rotterdam) examined empirical evidence of the possible gains and losses from cooperation. He used a modified version of the COMET model, which has three economic blocs: the US, the EEC and the rest of the world. The US and the EEC blocs both include Keynesian-type real sectors, with inflation generated by 'cost-push' mechanisms, and a detailed monetary sector involving financial relationships between the central banks, commercial banks, the government, the private sector and the overseas sector. Capital is internationally mobile and exchange rates float. The model also allows asymmetries between the US and the EEC, the most important of which is the methods of financing government expenditures. In the EEC this involved bond sales to banks, while in the US tax measures were used. In addition, markets responded at different speeds in the two blocs, with policies acting more quickly in the US than in the EEC. Hughes-Hallett assumed that the US and the EEC had the same policy targets, but attached different priorities to particular objectives.

Three policy regimes were considered for the period 1974-78. The first involved 'isolation' policies, in which both the US and the EEC set policies ignoring others' reactions - the open loop Nash assumption. The second involved the closed loop Nash game. The third regime assumed that policies were set cooperatively.

The results indicated that policies in the US and the EEC tended to converge as one moved from the first to the second and then to the third regime. On moving from the first to the second policy regime, both the US and the EEC benefited, although the gains were greater for the US. The further move to the cooperative regime was relatively more beneficial to the EEC, though these benefits were small compared with the previous move. Hughes- Hallett also found that the EEC would still be better off under cooperation than under non-cooperation, even if the US cheated subsequently by reneging on its agreement to cooperate! Overall, Hughes-Hallett's conclusion was that the gains from explicit cooperation were small. The largest benefits seemed to come from policies which were non-cooperative, but which took account of the interdependence of the two blocs.

In his discussion, Elias Karakitsos (Imperial College) questioned the absence of 'time-inconsistency' problems, which seemed to have been avoided by the informational assumptions made by Hughes-Hallett. It was also unclear how sensitive the results were to the particular objective functions assumed for the US and the EEC. Nevertheless, the paper constituted an important first step towards measuring the gains from cooperation. It would be worthwhile to undertake further work which used other informational assumptions as well as tests of robustness of the results with respect to different model specifications.

Daniel Laskar (CEPREMAP) discussed a similar issue - the effect of cooperative behaviour on exchange rate volatility - in 'Excessive Fluctuations of the Exchange Rate and Lack of International Cooperation in Foreign Exchange Interventions'. He outlined a simple, though fairly general, two-country model. Both countries were assumed to be identical, and policy-makers in each country minimized fluctuations in the terms of trade as well as a domestic variable such as output. To achieve these objectives, a single policy instrument was assigned to stabilizing the exchange rate.

Laskar compared two forms of policy: a cooperative solution and a non-cooperative 'Nash' equilibrium. He demonstrated that non- cooperative policies meant less intervention as a whole than did cooperative policies. This implied that there was greater exchange rate volatility under the former.

David Currie (Queen Mary College and CEPR) argued that this result was not surprising given the 'public good' features of exchange rate intervention. Could Laskar's results be generalized? For example, with more than one policy instrument, or with many countries, the problems he considered might simply disappear. In the former case, different instruments can be assigned to the internal and external goals. With many countries, a change in one country's exchange rate may have very little effect on other countries. The question then turns on whether the EMS is sufficiently large for this to be the case. It was argued that Laskar's analysis may be more suited to examining the behaviour between the EEC and the US, as in Hughes- Hallett's paper. Nevertheless, it was felt that Laskar's appraisal could be extended in other ways by the introduction of dynamics or the assignment of costs to variations in policy.

Daniel Cohen (CEPREMAP) and Philippe Michel (INSEE) continued the discussion of policy evaluation in a stimulating paper entitled 'Towards a Theory of Optimal Pre-Commitment I: An Analysis of Time-Consistent Equilibria'. They examined a dynamic policy 'game' between the government and the private sector and devoted special attention to the problem of 'time-inconsistency'. Optimal policies in dynamic rational expectations models are often 'time-inconsistent' - there is no incentive to carry out previously announced policies as time moves on from the date at which the plan was formulated. The problem with such policies is that no one else will believe them when they are announced! A time-inconsistent strategy therefore requires that the government must somehow bind its future actions in order to sustain its credibility. But, given the inability to pre-commit the actions of future governments, this requirement is unrealistic. Cohen and Michel investigated time-consistent policies which would avoid this credibility problem.

The authors first demonstrated the time-consistency of a Cournot- Nash solution in which all agents simultaneously announced their strategies. They then showed that in a 'Stackelberg' game, in which the government assumed the role of the leader and the private sector acted as the follower, the optimal policy was time-inconsistent. It also had the peculiar feature that the government would announce its pre-commitment to pursue policies which would run counter to its objectives! Both the government and the private sector were assumed to have the same objective functions, and so by announcing such apparently perverse policies, the government gained by forcing the private sector to bear the main burden of adjustment.

These solutions involved the unrealistic assumption of pre- commitment. Consequently, Cohen and Michel applied their analysis to a case where time-consistent policies can be derived. Here the policy involves a game among an infinite number of players or governments in which policies are chosen to maximize an objective function subject to the constraints of the economy and given the policies of other governments. Attention was focussed on the simplest case, where each government assumed the policies of its successors to be dependent only on the current state of the economy. Such policies are termed 'memoryless' and avoid the interesting problem of 'reputation' effects, an issue briefly taken up later.
Cohen and Michel found that the Stackelberg (leader-follower) solution converged more slowly towards a 'policy equilibrium' than did the Cournot-Nash solution. This was also a feature of the time-inconsistent solution, though in that case it was more pronounced. In both cases the presence of a leader forced the follower to increase its share of the burden of adjustment.

The paper prompted a lively debate. The discussant, Marcus Miller (Warwick and CEPR), noted that there were other time- consistent policies, such as that proposed by Buiter. Miller argued that the Cohen-Michel approach was preferable to Buiter's solution, because it did not alter the strategic structure of the game. Miller also pointed out that time-consistent solutions are often linear approximations to time-inconsistent ones, but that this was not true of the Cohen-Michel framework, in which the economy actively 'overshot' its equilibrium before returning there.

The final paper on this theme was presented by Gilles Oudiz (INSEE) and represented joint work done with Jeffrey Sachs (Harvard and CEPR) elaborating on the work of Cohen and Michel. Their paper was given previously at the recent CEPR/NBER conference on 'The International Coordination of Economic Policy', and a full discussion of it can be found in CEPR Bulletin No. Four. In his comments on the paper, Richard Portes (CEPR and Birkbeck) saw the need for more empirical studies of policy coordination to match the growing analytical literature. This was readily acknowledged by Oudiz, whose intended paper, 'European Coordination: Some Empirical Evidence on the Potential Gains' was incomplete at the time of the conference. Some participants were concerned about the common assumption that there exists very little conflict between the policy objectives of different nations. Others felt that the effects of coalitions would be worth further study, though progress on this would be slow.

Exchange rate volatility was the subject of the paper by Adrian Blundell-Wignall (OECD) and Paul Masson (IMF), 'Exchange Rate Dynamics and Intervention Rules under Regressive and Rational Expectations'. They examined whether official intervention in the foreign currency market dampened the exchange rate overshooting caused by exogenous shocks. They first investigated the issue analytically, in a Dornbusch-style model of overshooting in which assets bore a risk premium. This gave a model of portfolio balance. Intervention was conducted to stabilize real exchange rate movements, subject to the constraint that losses in reserves arising from intervention could not continue indefinitely. The analysis indicated that intervention would dampen exchange rate overshooting and was unlikely, by itself, to induce cyclical behaviour. Both of these results were offered as support for such intervention.

These preliminary findings were then investigated in an empirical model of portfolio balance for Germany, in which the current account was endogenous and expectations were assumed to be rational. Intervention would indeed dampen the initial exchange rate response to disturbances. Subsequently, however, intervention had the effect of slowing down the adjustment towards equilibrium. This result indicated that evaluation of foreign exchange intervention should take into account a trade- off between lower exchange rate volatility and more prolonged misalignment.

In his discussion, Jeff Frankel (Berkeley) argued that the presence of a significant, though small, risk premium contradicted most other studies. It appeared to be due to the particular definition of the supply of foreign assets adopted by the authors. In addition, if a high exchange rate were due to the fiscal-monetary policy mix then the relatively small effect of the risk premium might be insufficient for intervention to operate effectively. Alternatively, if other factors such as taxes were responsible for the high exchange rate, then these might be reflected in the equilibrium exchange rate, in which case the need for intervention would disappear. Some concern was also expressed at the absence of any wealth effects in the goods and asset markets and the absence of an explicit government budget constraint.

The behaviour of real exchange rates, as well as real interest rates, was taken up by Charles Wyplosz (INSEAD and CEPR), in his paper 'International Aspects of the Policy Mix in Six OECD Countries'. His main concern was to explain movements in these variables by measures of monetary and fiscal policy stance. Wyplosz conceded difficulties at the outset. The theoretical links between economic policies and the variables studied had not been well established, and there were severe problems in the measurement of the variables.

Wyplosz argued that by historical standards the recent levels of interest rates and exchange rates were high only for the US and to some extent the UK. Moreover, whereas these two variables displayed a positive relationship in the US and UK, a negative correlation was evident in other countries. One possible explanation of these results was that the tightening of monetary policy in the US and UK had provoked corrective action by other nations to reduce any exchange rate depreciation. If the ratio of the monetary base to GNP was used as a measure of monetary policy, there was evidence to support this interpretation for most countries though not to explain the US experience. Consequently, Wyplosz turned to the influence of fiscal policy. A variety of measures of fiscal stance yielded no firm conclusions. Wyplosz did find, however, that the level of public debt could explain some of the movement of interest rates and exchange rates.

Wyplosz then discussed the results of regressing real interest rates and real exchange rates on each of the policy variables. It was readily acknowledged that the exercise was fraught with difficulties; the policy variables were obviously endogenous, and were closely interrelated through the government budget constraint. The results indicated that the only significant variable had been the level of world debt. Wyplosz also reported unsuccessful attempts to estimate a dynamic IS-LM model with a government budget constraint.
The paper was criticized by the discussant, David Begg (Oxford and CEPR), who felt that the problems of measurement had not been adequately treated. Wyplosz's procedure could yield a variety of results depending on the definition of the economic aggregates used. In addition, the government budget constraint had not been given sufficient attention. One participant thought that if world deficits were the main cause of high interest and exchange rates, it might be possible for one country to 'free-load' by pursuing an expansionary fiscal policy with very little effect on its own interest rate. Despite these problems, the general feeling was that Wyplosz had addressed an extremely important issue which deserved further work.

One difficulty in the analysis of fixed exchange rate systems is that, faced with the prospect of exchange rate realignments, agents with foresight will attempt to avoid capital losses by switching their currency holdings, and this causes the system to collapse. Jacques Melitz and Philippe Michel (INSEE), analysed this question in their paper 'The Dynamic Stability of the European Monetary System'. The solution they explored involved the authorities introducing sufficient uncertainty about the timing of realignments. They employed a two-country ('Germany' and 'France') Keynesian model, in which capital holdings were assumed to depend entirely on transactions motives. The model determined the earliest and latest ('critical') dates at which realignment could occur but not the precise timing of the realignment. Melitz and Michel demonstrated that such uncertainty would make realignments possible without inducing a collapse of the system.

A major criticism of the model raised by the discussant, Jeffrey Sheen (Essex), was the omission of any speculative motive for determining capital movements. Wealth effects were also absent, which was undesirable given that inflation played an important role in the model. There was also some concern about the manner in which the two 'critical' dates were determined. These seemed to be points where policies were becoming perverse. For example, the earliest date at which realignment could take place was when the French authorities raised interest rates even though French output was falling. Similarly, the latest date at which realignment could take place was when the German authorities lowered interest rates though German output was rising. It was suggested that a more obvious determinant of the latest date would be the imminent collapse of the system.

The conference closed with a paper by George Zis (Manchester Polytechnic) 'The European Monetary System and the UK', which discussed the arguments advanced against UK membership of the EMS. He conceded that the role of sterling as a reserve currency could create strains within the EMS but felt that the benefits associated with the monetary autonomy afforded by the present flexible exchange rate were not substantial. Moreover, he argued that there had been marked convergence in economic policies and reduced exchange rate volatility since the establishment of the EMS. Zis partly attributed the failure of the UK to join the EMS to the general hostility of academic opinion towards British membership.

The discussant, Brian Tew (Nottingham) concentrated on the arguments against full UK membership presented in the 1984 Report of the House of Commons Treasury and Civil Service Committee. There had been a concern that the UK might be required to join the EMS at an unsuitably high parity. There was also evidence which suggested that the strains already imposed on the system by Germany's reserve currency role might become intolerable were the UK to join. One participant suggested that the preference for a monetary target, rather than an exchange rate target (which would arise in the case of EMS membership) was possibly due to the greater credibility of the former in reducing inflation; the alternative of using a persistently overvalued exchange rate would be accompanied by speculative pressure on sterling.

The conference brought together economists working on common themes on both sides of the Channel, underlining the scope for further joint research.