Global Macroeconomics
Policy Conflict and Cooperation

In recent years economists have devoted increasing attention to strategic aspects of interactions between countries, motivated partly by the growth of explicit intergovernmental negotiations over macroeconomic policies. This is an area of research which CEPR has emphasized since its foundation in 1983 and is currently pursuing through a major research programme with the Brookings Institution. In their introduction to the volume arising out of the 1984 CEPR/NBER conference on International Economic Policy Coordination (Cambridge University Press, 1985), Willem Buiter and Richard Marston remarked that policy coordination had not at that time generated as much attention among economists as it deserved, That conference proved to be a major step forward, however, and research into macroeconomic policy interactions has broadened and progressed significantly since 1984.

During the same period, the issues addressed by this research have also become more relevant to economic policy. The increasing tendency of the major countries to manage exchange rates has exposed the disagreement and conflict over appropriate relationships among exchange rates. There is also disagreement about the monetary and fiscal policies pursued by individual countries, The United States, for example, is alleged to pursue an excessively expansionary fiscal policy, while European and Japanese fiscal and monetary policies are alleged to be too contractionary .These macroeconomic policy conflicts also have implications for trade policy, as exchange rate fluctuations and persistent trade deficits create political pressures for tariffs and other forms of protection.

CEPR was therefore particularly pleased to organize a conference for the International Economic Association (IEA) on Global Macroeconomics: Policy Conflict and Cooperation', held in ,London on 12-13 February. The conference was divided into four sessions, each focusing on a major area of research and, policy debate. The first session examined applications of game theory, which forms the theoretic basis for many empirical approaches to policy coordination. The second session explored the potential for gains from macroeconomic policy coordination. The third session was devoted to the operation of the European Monetary System, and the fourth to macroeconomic interactions between the North and the South. The programme committee .acting for the IEA were Ralph Bryant (The Brookings Institution), Richard Portes (CEPR and Birkbeck College, London), Matthew Canzoneni (Georgetown University), David Currie (Queen Mary College, London, and CEPR), Dale Henderson (Georgetown University) and Marcus MiIler (Warwick University and CEPR). Financial support was provided by the Economic and Social Research Council, the Commission of the European Communities, the Bank of England, the Esmée Fairbairn Charitable Trust and Hambros Bank.

Game Theory and Policy Coordination
Game theory has provided insights which have been enormously valuable in the analysis of macroeconomic policy, but theoretical difficulties have limited its applicability to the actual design of policies. Ariel Rubin- stein (Hebrew University of Jerusalem and LSE)  opened the conference with his paper on 'The Complexity of Strategies and the Resolution of Conflict: An Introduction', Economists often argue that it is unrealistic to assume that economic agents choose arbitrarily complex rules of behaviour: rationality must be bound- ed in some way. It is only recently, however, that economic theory has begun to model the choice of complexity of rules of behaviour. Rubinstein reviewed recent work modelling 'games' in which the complexity of a strategy is assumed to impose a cost on the player who adopts it. The introduction of complexity had dramatic results on the outcome of a game: in such models there may be equilibria that exist even without highly complex strategies, but there may be fewer equilibria overall. Both results suggest that the game-theoretic approach may be more useful than it has so far appeared.

The need to enrich the structure of repeated games was widely acknowledged by participants at the conference, but John Roberts (Stanford University) noted that the paper was concerned only with the complexity involved in actually playing the game and executing the strategies, and not with the complexity of choosing the best strategy at the outset, Thus the model did not deal with the limited ability of players to evaluate alternative strategies. Rubinstein's assumptions about the defini- tion and costs attached to complexity were criticized by other participants as ad hoc. Rubinstein rejected this argument: he was merely providing, in one particular way, the additional information that the theory needed to produce a game with fewer equilibria.

Rubinstein's focus had been on repeated games, in which the set of strategies and pay-offs which players can choose is assumed to be identical in each period. Chaim Fershtman (Hebrew University of Jerusalem) explored 'Alternative Approaches to Dynamic Games' in the conference's second paper. Fershtman set dynamic games and repeated games within a common framework, in which there were three possible sources of dynamics: structural, behavioural, and informational. Dynamic game theory has typically focused only on structural dynamics, in which actions chosen in one period affect the game's pay-offs and possible strategies in later periods. The analysis of repeated games has focused on behavioural dynamics, in which players condition their actions on the previous strategies of other players, but has ignored structural dynamics. Little attention had been given so far to informational dynamics, in which players use the history of the game to learn about the characteristics of the other players, such as their objective functions. In principle all three sources of dynamics may be present in .a game, and Fershtman gave an example in which both strategic and structural dynamics were important. Analyses of dynamic games have considered only 'simple' strategies, in which players were assumed to have no memory, This had the advantage of producing unique equilibria, but the set of possible solutions is clearly enlarged by allowing strategies that depend on the past history of the system.

Leonard Mirman (University of Illinois) focused on the issues faced by the users of game-theoretic results in deciding what kinds of models and games were appropriate. He suggested that the tractability of models was important to users: on this criterion, dynamic games scored heavily. 

Willem Buiter (Yale University and CEPR), in his overview of the first session, criticized the direction in which economic applications of game theory had been developing. He described it as being driven by the inner logic of game theory rather than by the needs of users. Both papers, he argued, contained ideas which could not be readily implemented. Rubinstein's notion of the complexity of a strategy was not a natural one, and his concept of the 'finite automaton' did not embody the idea of procedural rationality.

The Gains from Policy Coordination
The second session of the conference was devoted to measurement of the gains which could be realized through macroeconomic policy coordination. Empirical evidence can be brought to bear on the general question whether cooperation of any kind will increase the welfare of the countries concerned. In particular, if countries coordinate the full range of their macroeconomic policies, does this yield large gains relative to a complete absence of cooperation? David Currie (Queen Mary College, London, and CEPR), Paul Levine (London Business School and CEPR} and Nic Vidalis (Queen Mary College, London) found that a government's reputation for fulfilling its policy commitments was important in obtaining gains from cooperation; policy coordination without such a reputation was not always beneficial. In their paper, entitled 'Cooperative Rules for Monetary and Fiscal Policy in an, Empirical Two-Bloc Model', they used Minilink, a reduced version of the OECD's Interlink multi-country econometric model. The modified model grouped countries into two blocs the United States and the rest of the world. The authors also treated the private sector in the two blocs as a third independent 'player'.

This approach extended and applied their earlier theoretical analyses, reported in CEPR Discussion Paper Nos. 94 and 102, and allowed the consideration of four policy strategies. Each government can choose a policy which is 'fully optimal' over time: this is time-inconsistent, however, since the government will be tempted to renege on its announced policies at some future date. The private sector will realize this, and so the policy is not credible. Such a policy must also therefore be reputational: the government needs to establish a reputation to convince the private sector that it will not succumb to the temptation to renege. As an alternative to the best available reputational strategy, each government can pursue a 'non-reputational' policy: such policies are constructed at the outset so as not to present any incentive in the future for the government to renege on its policy announcements. This policy does not require the government to establish a reputation, but It generally results in smaller welfare gains. In addition, under each of these strategies (reputatlonal and non-reputatlonal), the two governments then choose whether or not to cooperate. 

The simulations demonstrated how these four strategies performed in response to a variety of shocks. Whether cooperation was beneficial (in the absence of reputation} depended on the duration of the shock: persistent shocks increased the desirability of coordination. The uncertain benefits of cooperation arose from the presence of the private sector in the game: the actions of a third player meant that cooperation between only two players (the governments) would not necessarily benefit them. The authors confirmed their theoretical predictions that of the four strategies the cooperative and reputational policy yielded by far the highest pay-off. The pay-offs from the other three policies were very similar but much smaller . The authors also showed that the cooperative reputational policy rule could, under certain circumstances, be sustained as an equilibrium of a non-cooperative game played by the two governments and the private sector.

Paul Masson (IMF) criticized some features of the empirical models used in the paper. Various wealth effects had been removed in order to accelerate the model's dynamic responses, and this had undesirable effects on the accuracy of the simulation results. Masson also noted that the price level in the policy simulations was fixed by assuming that a particular price level was an objective of policy. He recommended modelling monetary policy as operating through variations in the money stock rather than an interest rate instrument and by introducing more forward-Iooking expectations into the model, for exam- pie in the determination of inflation.

Simulate Policy rules In ten empirical models

The case for coordinating policies is usually stated. In terms of efficiency gains. .Calculations of such gains usually assume that coordinated policies are designed using correctly specified models of the economies concerned, and simulations which demonstrate the benefits of policy coordination are often criticized for their sensitivity to errors in these econometric models. In 'International Policy Cooperation and Model Uncertainty', Gerry Holtham (The Brookings Institution) and Andrew Hughes Hallett (University of Newcastle and CEPR) explored the implications for macroeconomic policy coordination of uncertainty about the structure of the world economy. They simulated the performance of non-cooperative and cooperative policy rules in ten empirical models in current use, each characterized by a somewhat different economic structure. Having computed the optimal policy rules for a particular model, Holtham and Hughes Hallett assessed the consequences of using this set of policies in all the other models as well. This enabled them to look for policy rules that performed well on a variety of models and to study how the predicted gains from cooperation and the policies necessary to achieve them varied across the ten models.

The authors found that cooperation yielded consider- able gains but that the policies necessary to achieve these gains varied greatly across models. Cooperative policy rules derived from one model frequently performed worse when applied to other models than the corresponding non-cooperative policies! Non- cooperative policies appeared to be more robust with respect to variations in model specification. This result led the authors to conclude that risk-averse policy-makers may have little incentive to cooperate.

Gilles Oudiz (Compagnie Bancaire and CEPR) welcomed the attempt to compare optimal policies across a variety of empirical models. He suggested that the large gains from coordination found by Holtham and Hughes Hallett, relative to those found in previous studies, could have resulted from the large weight the authors had placed on output relative to inflation in the assumed objectives of policy-makers. Oudlz did not share the authors pessimistic conclusion that model uncertainty may discourage cooperative policy: he argued that model uncertainty rendered tentative any policy advice, not just that concerning cooperative policies.

Ralph Bryant (The Brookings Institution) observed that model uncertainty was of great concern to policy-makers and that this analysis of it was a step in the right direction. Participants disagreed, however, about the importance of the differences between the ten models used in the paper, Were they too similar, sharing common databases and in some cases common theoretical structures, as Ralph Tryon (Federal Reserve Board) argued? Or were they too dissimilar, as William Branson (Princeton University and CEPR) argued: the need to attract funding might encourage model builders to differentiate their products excessively.

Matthew Canzoneri (Georgetown University), in his remarks concluding the second session, drew attention to the large gains from cooperation reported at the conference relative to those found in earlier studies. The size of the measured gains depended on the extent of spillover effects between countries in the world model and on the source and combinations of the shocks assumed, He welcomed the effort made by Currie and Levine to incorporate history-dependent strategies into a dynamic game framework, but noted that if cooperative policy rules could indeed arise out of a non-cooperative game, international institutions like the IMF and the EMS would have little role to play.

The European Monetary System
The third session of the conference was devoted to analysis of the European Monetary System (EMS). Earlier research on the EMS had tended to focus on the operation of the exchange rate mechanism. In 'Exchange Rates, Capital Controls and the EMS', David Begg (Birkbeck College, London, and CEPR) and Charles Wyplosz (Institut Europeen d'Administration des Affaires and CEPR) sought to focus instead on the question of whether two countries ('France' and 'Germany') would have incentives to depart from free floating and to fix exchange rates through a mechanism such as the EMS. Begg and Wyplosz modelled the joint choice of an exchange rate regime and macroeconomic policies: the EMS could be viewed as one possible outcome of a game played between the two countries. Each country's policy-makers are concerned about inflation and output, and each country is assumed to inherit inflation, 'France' more so than 'Germany'.

The channels of interdependence between the two economies operate through trade, which affects aggregate demand for goods and aggregate price levels, and financial markets in which, in the absence of capital controls, each country's assets are perfect substitutes. Each country has only one policy instrument, monetary policy. The countries not only choose their monetary policies, but also decide whether to use capital controls and whether to establish a fixed exchange rate. The two countries' incentive to adopt an EMS mechanism is based on the mutual gains which it would produce for the two countries in the execution of an anti-inflationary policy.

Begg and Wyplosz noted that the incentives to form an EMS were sensitive to the initial conditions inherited by potential member countries. An EMS formed to fight inflation might therefore disintegrate once unemployment not inflation was the main concern of policy-makers. Begg and Wyplosz also explored how the

imposition of capital controls might affect the choice between the EMS and a floating exchange rate regime. They found that, by modifying the interdependence operating through financial markets, capital controls could make the EMS more attractive and thus more feasible. 

Capital controls could make the EMS more attractive to members

The role played by capital controls in the EMS was widely discussed, William Branson believed that the viability of the EMS depended on minimizing the number of realignments. Capital controls could thus be seen as a way of allowing countries in difficulties to maintain exchange rate parities without an immediate realignment, while demonstrating a genuine case for an eventual realignment. Francesco Giavazzi (University of Venice and CEPR) argued that. In practice capital controls allowed the .EMS to survive realignments: they did not necessarily enhance the gains for countries from, EMS membership. Patrick Minford argued that capital controls entailed costly misallocations of resources, which the paper had failed to model. this contention was hotly contested. Rudiger Dornbusch (MIT) entered a plea that capital controls be considered as an ordinary policy instrument of public finance and pointed to the successful growth record of various countries which had used them wisely. Others argued that temporary controls aimed at curbing short-term movements of financial capital entailed no significant long-term costs to the efficiency of resource allocation.

Francesco Giavazzi (University of Venice and CEPR) and Alberto Giovannini (Columbia University Business School) adopted a more empirical approach in their paper, 'Models of the EMS: Is Europe a Greater Deutschmark Area?'. Economic theory suggested that asymmetry was an important feature of exchange rate systems. Asymmetry in the EMS, for example, theoretically allowed smaller European countries the benefits of pegging their exchange rates to a larger country, West Germany, which enjoyed relative price stability, Without this commitment to a fixed exchange rate, the smaller EMS countries would suffer a higher inflation rate due to the inflationary bias of their domestic monetary policies.

Giavazzi and Giovannini then explored whether in practice the EMS displayed such an asymmetry by examining data on movements in interest rates and on foreign exchange market interventions by the central banks of EMS countries. They concluded that although the EMS had not been designed to work asymmetrically, this was the key to its operation in practice. It had become essentially a 'Deutschmark zone', in which the monetary policy of the Bundesbank exercised a dominant influence. West German macroeconomic policy focused on its money supply, and virtually all the burden of intervention to defend exchange rate parities fell on the other countries. The behaviour of European interest rates appeared to support this interpretation. In April 1986 the market expected an appreciation of the Deutschmark relative to the French franc. As theory would suggest, interest rates on deposits denominated in francs were observed to rise in proportion to the expected appreciation (to a larger extent in the Euromarket, and to a smaller extent at home, because of exchange controls). In a symmetric system German interest rates should fall at the same time, but in practice they hardly moved, either at home or in the Euromarket. The authors' regression analysis also indicated that the advent of the EMS was associated with an increase in the volatility of interest rate 'innovations' in France, and a decrease in West Germanyand Italy.

Participants disagreed over the role played by the EMS and how it should be modelled. Richard Marston (University of Pennsylvania) was sceptical about the extent to which the EMS could be viewed as a means for the other European countries to hire a 'conservative central banker' in the shape of the Bundesbank, since it was always possible to realign the member currencies, David Vines (University of Glasgow and CEPR) interpreted the EMS as a mechanism for stabilizing exchange rates and preventing the appearance of speculative bubbles in exchange rate movements, though such a function might be better fulfilled by a 'target zone' system.

Some participants felt that Giavazzi and Giovannini's data on interventions were less useful because of their high degree of aggregation, It was also argued that the authors' use of the residuals from equations forecasting interest rates was not a reliable means of identifying the interest rate 'innovations' arising from EMS membership, since a variety of factors could have been responsible for the change in interest rate volatility.

Only 15 % debt service/export ratio required for solvency? 

North-South Macroeconomic Interdependence
The final session of the conference was devoted to interdependence between industrialized and less developed countries, Daniel Cohen (Centre d'Etudes Prospectives d'Economie Mathématique Appliquées à la Planification and CE.PR) discussed 'External and Domestic Debt Constraints for LDCs: A Theory with Numerical Application to Brazil and Mexico'. He examined the conditions necessary for LDC debtor countries to remain solvent, with regard to not only their external but also their internal government debt. Cohen showed that, in order to achieve long-term stability in debtor countries debt/export ratios, the fraction of export revenue, devoted to debt service should depend upon the initial debt/export ratio and the excess of the real interest, rate over the growth rate of exports. Cohen s calculations suggested, that the required ratio of debt service to exports was In fact rather low 15% or less, lower than the ratios which agencies such as the IMF have typically required of debtor nations. Cohen also noted that the evolution of total government debt is related to the size of the fiscal deficit. Increasing the trade surplus merely refinanced the national debt internally, unless the fiscal deficit was reduced at the same time.

Applying this analysis to the situations of Brazil and of Mexico, Cohen rejected the notion that a rational debtor country should ever wish to reduce its debt so as to make voluntary lending possible in the future. Several participants disputed this view, arguing that after an adverse real interest rate shock during which outstanding debt had risen, the level of debt relative to GDP should be reduced to allow for subsequent borrowing in the event of a similar adverse shock in the future. Jonathan Eaton (University of Virginia) pointed out that Cohen's analysis had omitted capital productivity and international mobility of private capital, so that the effects of investment and of capital flight on the behaviour of debt were excluded.

Ravi Kanbur (Princeton University and CEPR) and Sweder van Wijnbergen (World Bank and CEPR) analysed two aspects of North-South interactions. In their first paper, 'On the Fix-Price Foundations of Structuralist North-South Models', they used the methods of fixed-price disequilibrium models to explore the effects of supply shocks in the 'South' on the industrial economy of the 'North'. They argued that, by causing the economies of the North to switch between regimes of Classical and Keynesian unemployment, such supply shocks may have larger effects on the North than is commonly supposed. In a second paper, entitled 'Protectionism and the Debt Problem', van Wijnbergen used a general equilibrium model to examine the way in which Northern trade policy, aimed against exportable goods from the South, exacerbated the transfer problems faced by the Southern debtor countries in their attempts to repay their external debt. Van Wljnbergen explored the further costs induced by non-clearing markets. and by intertemporal aspects of the debt problem. His analysis shared common elements with Cohen's, in that costs arose from excesslvely rapid debt repayment and, effectively, from capital-market rationing.

The discussion following the papers by Kanbur and van Wijnbergen centred on the empirical significance of their theoretical models. Willem Buiter thought more information was needed on the orders of magnitude involved in North-South interactions, and Dale Henderson (Georgetown University) asked for examples of regime switches in the North that could have been induced by Southern harvest fluctuations. John Williamson (Institute for International Economics) re- marked that the interactions were sufficiently large to convince many economists that the reduction in OECD inflation since 1979 had been achieved largely by squeezing real incomes in the South.

Cohen's careful modelling of the domestic as well as the external aspects of debt problems was welcomed by Rudiger Dornbusch, in his overview of the final session. Dornbusch noted that German hyperinflation in the 1920s had been partly associated with a failure to control the government budget deficit. He was not convinced, however, that North-South issues were related to regime switching in the North. In his view, the North had for some time found itself in a regime characterized by Keynesian unemployment.

Dornbusch argued for the potential benefits of debt repudiation. In his view the costs to debtor nations been exaggerated; in the 1930s, many countries had repudiated debts and had, he claimed, subsequently enjoyed much faster growth and development than they would otherwise have done. John Williamson disagreed: he insisted that repudiation by big debtors, would be considered solvent on any sensible criteria, would make it harder for the genuinely hard-pressed debtor countries, which tended to be both much smaller and much poorer, to obtain relief.

Michael Kaser (St Antony's College, Oxford, and IEA) concluded the conference by placing it in its historical context, in regard to both its topic and the long series of important IEA conferences. He noted that this was the first of the 80 IEA conferences held so far to take place in London. This was surprising, if only because of the prominent role played, by the Royal Economic Society and by Sir Austin Robinson in the IEA. The conference was also noteworthy as the first to be explicitly devoted to international policy coordination and indeed the first since 1974 to be devoted to international interactions. Kaser thought that the conference' s examination of the gains from policy coordination would prove valuable and represented an important harnessing of theory to the fundamentals of economic policy, one of the goals of the IEA.

The conference proceedings will be published in late 1987 for CEPR and the IEA by the Macmillan Press, under the title Global Macroeconomics: Policy Conflict and Cooperation, edited by Ralph Bryant and Richard Portes.