Correcting the Inflation Bias

An inflation bias in monetary policy may reflect the need for seigniorage or the desire to create surprise inflation to alter the real value of nominal contracts.² The former may be important in less sophisticated economies in which tax collection is still difficult and the prevalent use of cash provides a significant tax base for the inflation tax as an alternative to even more distortionary taxes. This is not a good description of western Europe. Technical progress and increasing competition in financial services are steadily reducing the use of cash and hence the role of expected inflation in financing fiscal expenditure. It is unexpected inflation that increasingly holds the key to understanding any inflation bias in monetary policy.

Since Kydland and Prescott (1977), economists have understood that promises cannot be taken for granted. Decision makers face temptations that they may be unable to resist. Surprise inflation alters the real value of nominal contracts, especially in the labor market, where market power may imply undesirably high real wages, and in the bond market, where unexpected inflation appears to act as a lump-sum tax on holders of nominal government debt. However, since the private sector can foresee the temptation faced by government, the consequence is to raise inflation expectations by the appropriate amount.³ This is inefficient: the economy bears the costs of foreseen high inflation without the benefits of surprise inflation.

Subsequent research considered whether reputation might reduce this inefficiency when the game between government and private sector is repeated indefinitely.⁴ A reputational equilibrium can be sustained by the threat of punishment: to some extent, the private sector will trust a government promise to pursue low inflation because everyone knows that government cheating will trigger a period of high inflation expectations (loss of reputation), which is costly to the government.

Reputation cannot completely solve the problem of inflation bias. First, the more the future is discounted the less the deterrence today from the prospect of future punishment for succumbing to present temptation. Second, the period of punishment is ambiguous and the difficulties of coordinating private punishers substantial—since punishing is typically costly to the punishers, the incentive to free ride is powerful. But then equilibrium deterrence is small. Thus, although reputation helps, reliable attainment of low inflation may lie in taking steps that irreversibly precommit responsible monetary policy by the appropriate design of institutional infrastructure. Making the ECB independent of government is an example. E M U governments will delegate operational independence to the ECB as their agent. The ECB will be held responsible for achieving price stability.

A decision to assign the instrument of monetary policy to the agency of the ECB in pursuit of the objective of price stability does not imply either that price stability has become the sole object of macroeconomic policy or that monetary policy has become the sole instrument of macroeconomic policy. The monetary-fiscal mix continues to matter, and governments continue to care about deviations of actual output from potential output. Rather the case for delegation of monetary policy rests on the argument that the joint outcome for output and inflation will be better when the choice of the policy mix is not dominated by the fiscal authorities.⁵

Delegation is an attempt to mitigate the consequences of the fundamental distortions that give rise to the inflation bias, to find a way of living with them rather than eradicating them. In particular, the potential for inflation bias is larger the greater the output persistence exhibited by the economy:⁶ when sluggish wage and price adjustments lead to slow reversion of output toward its natural rate, surprise inflation has a larger benefit because it generates an output boom that is sustained for longer. Since the public foresee this, the expected (and hence the actual) inflation rate is higher.

An EMU characterized by sluggish adjustment of output and employment makes the ECB’s task of achieving price stability more difficult than otherwise it need have been. Although the literature on optimal currency areas has always recognized that greater wage-price flexibility can reduce the policy conflicts between member states, the systematic relationship between the temptation to inflate and the degree of inflexibility has attracted much less attention. It is striking that there has been such an asymmetry between the attempt to improve the fiscal fundamentals (e.g., the Maastricht criteria and the Stability and Growth Pact) while labor market fundamentals have been largely neglected.

A Trade-Off Between Price Stability and Output Stabilization?

Must the credible pursuit of price stability imply benign neglect of the business cycle? Initially, economists concluded that it must. Rogoff (1985b) showed that the easiest way to counteract the inflation bias inefficiency was to delegate monetary policy to a central bank that valued price stability more strongly, and therefore output stability less strongly, than did its citizens. Trading one inefficiency off against the other, appointment of a somewhat more conservative (inflation-averse) central bank would yield a social profit, despite the lower weight on output stabilization, and increased output volatility thereby entailed.

Yet this analysis does not fit the facts. For a large group of countries over many years, Alesina and Summers (1993) showed that countries with a more independent central bank not only enjoy lower inflation without lower real economic growth (the vertical long-run Phillips curve) but also that, despite lower inflation bias, they do not experience greater output volatility.

Subsequently, this puzzle has been resolved. Choosing central bankers with preferences biased toward price stability and against output stabilization is not the only way to ensure the central bank will counteract the inflation bias. Persson and Tabellini (1993), Walsh (1995), and Svensson (1997) discuss the use of inflation contracts that are most easily interpreted as financial transfers to the central bank governor, a performance- related bonus or penalty depending on inflation outcomes. In an ideal world, the optimal contract turns out to be linear in the inflation rate: since the key “externality” arising from surprise inflation is the attempt by the monetary authority to divorce actual and expected inflation—and in equilibrium, being equal, one is a linear function of the other—a penalty that rises in direct proportion to realized inflation is appropriate. However, the penalty “tax rate” on inflation will generally be period specific, since it should depend ideally on the initial conditions from which the economy begins and whatever shocks are known to have occurred.

The key implication of an optimally designed incentive contract for the central bank is that it deals with the inflation bias without inducing any distortion to output stabilization. It solves the Rogoff problem and would be compatible with the Alesina-Summers evidence cited above. However, performance-related contracts of this exact type do not abound, even in countries that have independent central banks. In any case, practicing central bankers give short shrift to the idea that performance-related pay could have much impact on the behavior of central bankers who have already volunteered to pass up much more lucrative job opportunities in the private sector.⁷

A more practical alternative to inflation contracts is the use of inflation targets, in which delegation prescribes a target rate of inflation in such a way as to counteract the inflation bias. In a natural rate model, in which output deviations last only a single period, Svensson (1997) confirms that a delegated inflation target, which may vary from period to period to reflect knowledge about recent shocks, can match the optimality properties of the inflation contract.

Once output persistence and sluggish adjustment are allowed for, policy needs two dimensions—one concerned with the optimal degree of accommodation of current shocks and one concerned with the optimal speed of unwinding of previous shocks,⁸ for example those demand and supply shocks outside the central bank’s immediate control. Nevertheless, a combination of targets and the appointment of central bankers with suitable “distorted” preferences about price and output stability can still achieve the optimal result: eliminating the inflation bias without distorting the ability to stabilize output. Since a high weight on inflation relative to output is one component of this twopronged approach, Svensson argues that the latitude for missing policy targets should therefore be relatively small. Any bands for intermediate targets should be relatively narrow.

Optimality of course is only for textbooks. For example, some of the shocks to which the central bank would ideally respond are almost certainly impossible for the public to verify even ex post. Monitoring difficulties preclude complex behavior contingent on shocks that are difficult for all to observe. Practical policy design will then favor robustness and simplicity. Simpler design of delegation rules, being second best, will reintroduce the need to trade off losses from the inflation bias against losses from an inappropriate degree of output stabilization.

The fear that the ECB will be insufficiently cautious should be balanced by a concern that it could be designed in a manner that forces it to be too cautious. EMU needs to get this balance right. Virtue is not exclusively on the side of caution. This section has argued that the design of EMU to safeguard price stability may, but need not, imply a requirement that monetary policy neglects any role for the stabilization of EMU output. An efficient design of the rules under which the ECB will operate can allow it to achieve price stability without exacerbating output fluctuations; an inefficient design may achieve the same degree of price stability but at the cost of greater output fluctuations. The evidence that in practice central bank independence has not implied additional output volatility may indicate either that national design of monetary policy has been efficient in such instances or that national fiscal policy has been able to take the strain. Whether fiscal policy can play an equivalent role within E M U is discussed in the section “Asymmetries and Adjustment Mechanisms in EMU.”

So far, the argument assumes that, apart from contemporaneous shocks, output is capable of central bank control. More realistic assumptions on the information structure—the long and variable lags in the effect of monetary decisions—raises other difficulties that will be discussed in relation to the choice of intermediate target for the ECB. First, however, a related issue must be addressed: the common perception that the ECB will initially have to engage in a period of “reputation building.”

Reputation Building

Even careful designs are more reassuring when they have been tried out in practice and withstood the stresses for which the architect had planned. There are two facets of the textbook account described above that require more reality. First, it is assumed that the principal can credibly delegate to the agent.⁹ Could there be circumstances in which EMU governments will so regret their precommitment of monetary policy that they may be tempted to try to reverse it? Second, the discretionary action of the ECB is sensitive both to the actual, but hard to verify, preferences of its decision makers and to its model of how the EMU economy works, on which there is substantial scope for disagreement. Simply asserting how the ECB is likely to behave is different from confirming that these predictions are reliably borne out. There is no substitute for a good track record.

Is economic theory helpful in contemplating how this is likely to be established? What is involved is a signaling game, originally introduced into macroeconomics, albeit in stylized form, by Backus and Driffill (1985) and Barro (1986). The main insight is that “soft” policymakers have an incentive to mimic “tough” policymakers for a while, thereby lulling the public into a false sense of security before springing their eventual inflation surprise.¹⁰ The cost to soft policymakers is initially having to pursue tougher policies than they would really like; the benefit is that they achieve a larger surprise when they do finally renege. Tough policymakers, by assumption, face no such temptation. Heterogeneity of preferences, as well as the public’s difficulty in assessing these directly, is what leads initially to a pooling equilibrium in which the type of policymaker cannot be distinguished. In a game with a finite horizon, the eventual temptation for the soft policymaker to cash in its reputation becomes irresistible.

In a game without a finish, the dynamics of when revelation takes place is less clear. It may be, for example, that when some shocks take on extreme values the cost of mimicking becomes unacceptably high and policymakers are forced to reveal their true colors. Tough policymakers also have an incentive to choose actions that soft policymakers would find it prohibitively costly to mimic, thereby bringing forward the date at which the benefits of a good reputation can be enjoyed. During the period of pooling, before revelation takes place, imperfect credibility can be very costly to a true “toughie,” since the public must hedge against the possibility that they are in fact dealing with a “softie”: inflation expectations, and nominal variables in which they are embedded, will therefore be higher than after the toughie is revealed.

Of course, the world is not this clear cut. As Blinder (1995) observes, there is in practice a continuum of preferences, ranging from very tough to pretty wet; and revelation is never as decisive as the above account suggests. One act of weakness rarely dissipates the entire stock of reputational capital, just as one act of resolve rarely induces total trust. Rather reputational capital ebbs and flows in response to the repeated comparison of words and deeds.

Even within this more realistic view of how reputation can be developed, the ECB may initially devote considerable attention to investing in policies that are tougher than it would choose to pursue after its reputation has been established more reliably. Given the preceding discussion about the appropriate design to counteract the inflation bias, demonstrating resolve may imply both the pursuit of an average inflation target that is lower than it need be (perhaps even negative inflation), and insufficient accommodation of temporary disturbances, thereby exacerbating output variation. Given what is known about output persistence in Europe, there is every chance that the real effects will last a long time even if the original shocks wear off quickly.

Can anything be done to mitigate this difficulty? Although greater labor market flexibility would be beneficial, it is probably too late to secure significant changes by the time EMU begins; even so, there is no case for delay. More flexible labor markets remain an important requirement if EMU is to work well, and the sooner progress is made the better. One aspect of the ECB’s operational procedure that can make a difference and be achieved from the outset is the transparency of its decision making.¹¹ This operates through two distinct channels. First, the more evident is the basis of decisions and the manner of responding to shocks as they occur, the more difficult it is for an imposter to mimic the behavior of a truly tough central bank. Thus transparency facilitates revelation. Second, and related, it is important that the ECB not be sold to the public as a monetary straightjacket, for then it will have to behave in such a manner merely to avoid a dramatic loss of initial reputation. This argument views delegated monetary policy playing its appropriate role in dealing both with price stability and with output stability. Unless the role of the latter is explicit from the outset, its proper pursuit by the ECB will attract unwarranted criticism and hence will not be attempted.

The gains to attaining credibility are large. The Deutsche Bundesbank has often taken an active role in managing the German economy and has done so without inducing fears of higher inflation precisely because its assurances that its actions would be temporary were widely believed. For example, Begg and others (1991) document the behavior of central banks immediately following the stock market crash of October 1987. Fearing an adverse shock to demand and confidence, all central banks in the major countries cut interest rates, promising that they would be raised once the crisis was over. Which central bank cut interest rates most? The Bundesbank. Three months later, which was the only central bank to have raised interest rates to their former level? The Bundesbank.

Begg, Giavazzi, and Wyplosz (1997) argue that the Federal Reserve under Chairman Alan Greenspan has been similarly activist in the 1990s, again without adverse effects on inflation expectations. But this raises a second issue: how did the Federal Reserve acquire such a reputation, which allowed it to engage in monetary activism without precipitating fears of higher inflation? How much should be attributed to standing on the tall shoulders of the Federal Reserve under Chairman Paul Volcker, how much to the attributes of the current Board of Governors, especially of its Chairman? Answers to these questions help us think about how quickly reputation can be developed, and how much of it resides in individuals rather than institutions.¹²

ECB board members will not start with a blank history: as experienced central bankers in their own countries they already have an extensive track record. Indeed, given widespread agreement on the sluggishness of real adjustment in Europe and hence agreement about the cost of rapid disinflation, any reasonable assessment of post-Maastricht disinflation would surely conclude that what has been surprising is how much, not how little, disinflation has been achieved.

Of course, inflation control may prove difficult not by intent but because there will be uncertainty, especially at the beginning, about the demand for euro money and about the transmission mechanism of ECB monetary policy. Does it matter which procedure for intermediate targeting is adopted?

Monetary or Inflation Targets?

The instrument of monetary policy is the short-term interest rate. This takes up to two years to affect output and prices, a lag that raises the possibility of using an intermediate target for monetary policy. Brunner and Meltzer (1967) list the desirable properties of such a target: it should be reliably under the central bank’s control, it should have a predictable relationship with the final target, and it should be capable of affecting the public’s expectations.

For many years, targets for the growth of monetary aggregates have been used to fulfill such a role.¹³ Data on monetary aggregates is quickly available, allowing prompt evaluation of the actual monetary stance in relation to the stance previously promised. However, even for narrow money, fluctuations in demand have in practice meant that central bank control was usually only imperfect;¹⁴ for broad money, changes in the degree of banking competition and the nature of financial regulation have meant very substantial changes in money demand, and consequent uncertainty during this process, more so in some countries than in others. Hence, for a given level of the monetary instrument, the subsequent evolution of monetary aggregates has been only approximately under central bank control, and with a lag.

Lags between interest rate decisions and inflation and output consequences are longer still. Moreover, both money demand and the transmission mechanism to output and prices will be especially uncertain during the early period of EMU. There will be no direct history with which to compare, and construction of a “shadow history,” a composite of the behavior of member states, raises serious issues about the Lucas critique, whether the change of regime will itself induce changes in behavior.¹⁵

The EMI has recently recommended that the choice of intermediate target be narrowed down to a monetary target or an inflation target (European Monetary Institute, 1997). Monetary targets appear to accord more closely with the previous practice of the Bundesbank. An attempt to associate the behavior of the ECB with the track record of the Bundesbank might be easier if the ECB adopted the same monetary strategy as the Bundesbank has pursued. However, this argument can be overstated.

First, continuity of reputation is more likely if the initial membership of EMU is small, centered on a German core, and seen to have fulfilled the Maastricht convergence criteria on which the Bundesbank itself has laid emphasis. If core countries such as Germany and France miss the Maastricht criteria by sizable amounts, or are thought to have achieved them only by creative accounting or one-off adjustments that give a poor indication of sustainability, and especially if one consequence of this is to make membership of EMU larger because other countries are then difficult to exclude, it will be difficult to believe that the ECB will derive much of a honeymoon effect merely by following the same intermediate target that was used by the Bundesbank.

Second, several empirical studies have confirmed that, whatever its rhetoric, the Bundesbank’s behavior has not been the inflexible pursuit of rigid monetary targets.¹⁶ Its revealed behavior has been consistent with the simultaneous pursuit of price stability and output stabilization, not unlike the reaction function implied by the Taylor rule,¹⁷ and similar in spirit to the vision of inflation targets elaborated above. As emphasized in von Hagen (1995), where previously announced monetary targets have proved inappropriate for the subsequent evolution of prices and output, it is the Bundesbank’s monetary targets that have quickly been adjusted. The transparency of the adjustment, and the reasons made public for the adjustment, have confirmed trust in the reliability of the Bundesbank’s pursuit of its underlying objectives. In some circumstances, they also blur the usefulness of distinctions between inflation targets and monetary targets. But it is important to understand when to worry about the distinction and when not to worry.

Here, some formal modeling helps. Persson and Tabellini (1993) show that it is possible to delegate monetary policy to a central bank and obtain the same outcome whether one uses a monetary target or an inflation target, but that it is easier for outsiders to monitor central bank behavior when assessed against an inflation target than a monetary target; if the central bank has inside information about monetary conditions, it is more difficult to assess the aptness of its choice of monetary instrument. One tends to judge the performance of a golfer by how close to the hole she hits the ball, not by direct assessment of club speed or hip rotation. Informational requirements are important. Given the priority that the ECB will inevitably place on early reputation building, procedures that complicate monitoring undermine the ease with which that reputation can be developed.

Moreover, the implied rule for monetary aggregates is a complicated rule conditional on the state of the economy and all available information: this is quite different from any simple rule for monetary growth, which would lead to quite different (and suboptimal) outcomes. At a time when money demand is likely to be especially uncertain, both because of changes in the medium of exchange and because the asset demand may be unpredictable until the ECB’s track record is better established, simple rules for monetary growth are particularly to be avoided: this is just an application of the analysis of Poole (1970).

Thus, if credibility were exogenous, there might be little to choose between an inflation target, implying a (complex) reaction function relating interest rates to the state of the economy, and a sophisticated monetary target, relating both interest rate and monetary announcements to the state of the economy in an equally complex way. Either would be preferable to simplistic rules for the evolution of monetary aggregates: the early years of EMU, when money demand and monetary transmission are bound to be uncertain, is not the time to put one’s eggs in the basket of simple monetary targeting.

Next, it must be recognized that credibility is not exogenous. This means that monitoring and accountability will matter.

Monitoring and Accountability

Advocates of monetary targets are well aware of the difficulties and drawbacks described above. The chief drawback of inflation targets is the length of time until actual inflation is affected by changes in monetary policy (see, for example, European Monetary Institute, 1997). This has two implications. First, it takes a long time to detect poor performance by the ECB. Second, the principals, EMU governments and their citizens, may find it difficult credibly to commit to punishing poor performance detected today but applying to decisions taken by the ECB two years in the past. Thus, the chief argument in favor of monetary targets is that decisions on interest rates affect this intermediate target more quickly and hence facilitate better monitoring. In turn this helps the central bank be more credible.

King (1994) and Svensson (1996) note that inflation targets do have an associated intermediate target that is immediately available, namely the central bank’s inflation forecast over the horizon until interest changes take effect on output and prices. But for model uncertainty and any possible inside information available only to the central bank, one could assess central bank performance by the effect of its interest rate stance today on today’s forecast of inflation say two years in the future. This of course is the procedure that all central banks adopt in private. The issue is whether it could be made public, or at least announced after private deliberations had been concluded.

Whether or not the ECB will have much private information of any value, model uncertainty (disputes in positive economics about how EMU transmission mechanisms will actually work) is likely to be significant during the early years, creating a mist within which public doubts may arise that announced ECB forecasts are not the central estimate of the ECB. Yet biases are unlikely to be substantial, and the interests of transparency are best served precisely by making such announcements and subsequently conducting discussions in public about why forecasts turned out to be inappropriate when this was the case. The Bank of England’s Inflation Report has been influential in stimulating such a discussion in the United Kingdom and surely helps explain Sherlock Holmes’s dog that didn't bark in the night—the fact that, despite leaving the exchange rate mechanism in order to engage in a substantial devaluation, the United Kingdom did not face significant inflationary pressures in the aftermath of 1992.

Other conclusions, however, are possible, and seasoned observers have reached differing views. Svensson (1996), having made the analytical case for using inflation forecasts as intermediate targets, then concludes that it may be “safer” simply to rely on ex post inflation outcomes, despite the delay involved. Von Hagen (1996) jumps the other way and concludes that (sophisticated) targets for monetary aggregates may be more reliable.

Whichever balance is reflected in this final judgment about the target that is primus inter pares, and for which the ECB should be held formally accountable, transparency of procedure is paramount and inflation forecasts should be announced.

Accountability will in fact operate through several channels. The ECB will formally and regularly report to the European Parliament. Its activities will also be subject to implicit review when the fixed-term contracts of ECB members come to an end and the Council of Ministers appoints successors. But, in addition to those formal channels, there may be more frequent accounting for decisions, both to financial markets and to electorates.¹⁸

This is not simply wishful thinking. There is a sound political reason to anticipate that the ECB will be driven into an early and frequent accounting for its actions. At least in the early years of EMU, and in particular until 2002 when banknotes are expected to be widely introduced, the threat of a deeply dissatisfied country leaving EMU will be more credible than, say, the threat of Alabama to secede from the Union because it dislikes the monetary policy of the Federal Reserve. In practice, therefore, the ECB will face pressure to inform and to explain, both in relation to changes in interest rates and in relation to whether targets have in fact been achieved. Member states anxious that the early years of EMU avoid a deflationary bias may wish the ECB to explain why it missed the bottom end of its target range as well as why it overshot the top end of the range.

Conclusion

This section has discussed the monetary strategy by which the ECB will pursue its primary and overriding goal of price stability and subsidiary objective of supporting the economy. To make this operational, the ECB needs to have an idea of what price stability means,¹⁹ over what period it is to be achieved, with what precision it should be achieved, and whether its performance will be evaluated by intermediate targets or final outcome. Each of these questions has yet to be resolved.

Delegated monetary independence, although the solution to the inflation bias of discretionary government control of monetary policy, does not imply zero weight on output stabilization. Coupled with a realistic model of output persistence, this implies that past deviations of inflation from target, arising because central bank control of the economy is only imperfect, should ideally be eliminated steadily rather than in a single step. If model uncertainty could be resolved sufficiently to make agreement on the degree of persistence practical, it would of course be desirable to formulate a normal (i.e., optimal) rate at which this gap should be closed. Given model disagreements and the need to find a procedure on which EMU participants can agree, an implicit band for inflation outcomes may offer some of the same advantages, albeit in a more ad hoc manner. If such a band is adopted, it should be symmetric around the central target, and the ECB should have to account explicitly for crossing either edge of the band; the ECB would also be wise to account for its policy actions (or inaction) within the band.

Provided knowable dangerous rules, such as simple targets for money growth, are avoided, the precise details of the monetary strategy may be less important than the perceived transparency with which operational decisions are reached, including the readiness of the ECB to explain why decisions were reached, to admit when previous decisions were erroneous, and to take prompt action to remedy them. Since the environment will be one of uncertainty, the reliability of response to new information will be more reassuring than adherence to central estimates that are unlikely to prove robust.

Are Shocks Asymmetric?

If structures of different economies are sufficiently similar, common shocks can in principle be dealt with by common policy, at least to the extent this is compatible with price stability as discussed in the section “EMU’s Monetary Strategy.” However, different shocks may call for different policy responses. A robust judgment about the likelihood of a shock to a particular country being resolved by pursuit of the appropriate policy at the EMU level thus requires analysts to be as specific as possible about the shocks and their correlation across countries.

One evident distinction is between demand and supply shocks. Bayoumi and Eichengreen (1993 and 1996) use the Blanchard-Quah methodology to identify demand and supply shocks affecting output in each European country.²⁵ Examining the correlations across countries for each shock, they conclude that there is a clear group of “core countries” in which both demand and supply shocks are highly correlated across countries (Germany, Austria, Switzerland, France, Denmark, Belgium, the Netherlands, and Luxembourg) and a clear periphery that is less integrated with the core (Italy, the United Kingdom, Spain, Portugal, Ireland, Finland, and Greece).²⁶ By implication, a common monetary policy is more likely to meet the needs of the core than the periphery.²⁷

Kenen (1969) noted that a likely source of shocks that are asymmetric across countries is regional specialization. If shocks are largely sector-specific, and countries have a diversified collection of sectors, asymmetries across countries may be small. Stockman (1988) examined the pattern of sectoral and national shocks in seven European countries and concluded that they were of roughly equal importance. Of course, one could try to examine whether the sectoral/national division bears any correspondence to the supply/demand division. Bayoumi and Eichengreen (1996) confirm that “sectoral” shocks are much more closely identified with supply shocks and “national” shocks with demand shocks.

Suppose this is true. Can one make an educated guess about how the Lucas critique is likely to operate here? How will the advent of the EMU regime change the likely pattern of shocks? The most obvious conjecture is that demand shocks will become more highly correlated because of the common monetary policy pursued, a conclusion likely to be correct despite the following qualifications.

First, until one considers fiscal policy, and hence the monetary-fiscal mix, one cannot be sure that monetary policy will dominate demand at the national level. Whether there is much scope for national fiscal policies, and whether or not they will be correlated, will depend, inter alia, on the Stability and Growth Pact and how countries decide to operate within it (see below).

Second, asymmetries in structure may impart asymmetric effects to symmetric policies. In addition to the domestic aspects of monetary transmission cited earlier, there are external aspects: different economies face different exposure to the euro exchange rate against the rest of the world (a topic largely neglected to date). If the fact that EMU as a whole is relatively closed induces benign neglect of the exchange rate, the significance of differences in exposure to the external exchange rate may be magnified by the larger fluctuations in the exchange rate that are allowed to occur.

Against these concerns must be set the possibility that EMU increases the correlation of demand (and possibly supply) shocks. Frankel and Rose (1997) demonstrate the clear cross-country relationship between the degree of trade integration and the correlation of countries’ business cycles. Presumably this can take place through two quite distinct mechanisms. First, greater openness increases cross-country spillovers and induces incentives for greater policy coordination (the process of which EMU is the extreme case).

Second, greater openness may reflect lower trade barriers and greater specialization. Specialization can take place within countries or in regional clusters that span national frontiers of small, continuous economies. Why does all this matter? Because the eventual pattern of sectoral location may increase or reduce the sectoral similarity of different countries and hence the likely correlation of shocks faced by their national economies. Bayoumi and Eichengreen (1996) discuss this issue and conclude the evidence may be compatible with increasing sectoral specialization at the national level, a recipe for somewhat greater vulnerability to asymmetric or “country-specific” shocks. Had the Frankel and Rose evidence been drawn from the same sample of (potential EMU) countries, this would have provided some reassurance: whatever is happening to sectoral location, even if this operates in an “unhelpful” direction, the finding that overall output correlations rise with trade integration (presumably through the dominance of demand mechanisms in this instance) provides some comfort. However, the Frankel and Rose evidence draws on a much wider group of countries to reach this conclusion, so, without further research, one cannot be sure it applies to potential EMU members alone. In particular, it should not be assumed that EMU countries are representative of the wider sample: the decision by a particular group of countries to embark on a currency union reveals a belief about the nature of their integration that is quite unrepresentative of the world in general.

Labor Market Adjustment to Asymmetric Shocks

If the common monetary policy will not deal adequately with country-specific shocks, can labor market adjustment within countries play that role? Blanchard and Katz (1992) conclude that in the United States interstate migration plays a major role in adjustment to state-specific shocks; real wage flexibility is next most important; changes in unemployment and labor force participation are much less important. In contrast, Decressin and Fatas (1995) find that in Europe changes in labor force participation provide the principal adjustment mechanism.

Sometimes labor force participants will be close to indifferent about whether they work or not; the private costs of changes in participation will then be small. But this is not the norm. Even if it was, there may be beneficial externalities from being in the labor force. The social cost of having labor force participation take the strain is therefore likely to be large. If the United States finesses this problem by its greater real wage flexibility, could EMU do likewise? The section “EMU’s Monetary Strategy” noted the failure of EMU preparations to impose any conditions for greater labor market flexibility as part of the Maastricht convergence conditions. Any reliable increase in real wage flexibility will therefore have to rest on the Lucas critique. Such a case is not difficult to make, as in Begg (1990): in a model of staggered wage setting, as in Taylor (1980), less accommodative policy by government changes the degree of persistence in the real economy and thereby induces some compensation via the incentive to set forward-looking wage contracts. Nor is this a theoretical curiosum: Alogoskoufis and Smith (1991) show that a century of evidence, during which exchange rate regimes changed markedly, provides powerful empirical support. Whatever the modeling details adopted, Bayoumi and Eichengreen (1996) are surely correct to conclude that past evidence on real wage flexibility in Europe may be only an imperfect guide to future behavior. By the same token, induced labor market responses are likely generally only to offset part of the difficulty. Fiscal policy will therefore come under pressure to deal with remaining asymmetries.

Fiscal Accommodation of Shocks

Could monetary union quickly set in train forces for much closer fiscal integration? Some fiscal integration has of course already occurred through pressures of fiscal competition and mobility of tax bases across countries: to the largest extent in relation to capital taxation, to a lesser extent in relation to indirect taxation;²⁸ and, since labor mobility between European countries is not high, to a much lesser extent in relation to income taxation and welfare provision. However, the focus here is not on the wider impact of European integration on the structure of national fiscal systems but on the narrower impact of monetary union on the ability of national governments to mitigate the effects of country-specific shocks on output once they have dispensed with country-specific monetary and exchange rate policies.

During the last two decades, there has been downgrading of the role of discretionary fiscal policy in output stabilization (see, for example, Mishkin, 1995), reflecting both concerns about long-run fiscal solvency and about the timeliness with which discretionary fiscal policy can be deployed in relation to short-run shocks. Solvency will be discussed shortly in relation to the Stability Pact; the prior concern is timeliness. Two considerations suggest fiscal policy may have an important role to play in relation to accommodation of shocks in Europe. The first is that, in comparison with the United States, lower real wage flexibility and greater output persistence in Europe make timeliness less of an issue: unaccommodated, shocks have effects that last long enough to organize a policy response. Second, the fiscal response need not be discretionary. In principle, automatic stabilizers can do the same job. If one therefore proceeds on the assumption that a fiscal response to shocks is feasible, is it desirable?

Sala-i-Martin and Sachs (1992) first suggested that the existence of the federal tax system in the United States provided the automatic stabilizer par excellence, insuring individual states up to 40 cents on the dollar against state-specific shocks to income. Since Europe patently has nothing like this is place, the idea that monetary union in the United States is underpinned to such an extent by fiscal policy has provoked a profusion of further research.

Subsequent research has tended to qualify heavily the original conclusion. Von Hagen (1992) noted that the empirical methodology of Sala-i-Martin and Sachs failed to distinguish temporary and permanent shocks, or in his terminology failed to distinguish insurance and redistribution. Von Hagen’s estimates indicated that perhaps only 10 cents on the dollar reflected insurance. Bayoumi and Masson (1995) and Fatas (1997) confirm that redistribution and insurance are both reflected in flows of fiscal transfers. Von Hagen and Hammond (1995) argue that separating out the insurance dimension is complicated, difficult to verify, and hence improbable politics.

There is also a deeper concern, namely whether or not any fiscal federation can commit to make permanent transfers in response to permanent shocks. Ex ante, this is the risk that it is most valuable to insure, but ex post such insurance may or may not be available. In a sense, what defines an optimal political area is one in which such commitments are credible.²⁹ However, plans for European integration in the 1990s have taken place against the same backdrop as hopes by some for disintegration of some European countries: separatist movements have gained ground in such countries as Italy, Belgium, and the United Kingdom. What this paper calls systematic regional redistribution can also be called the ex post fulfillment of an ex ante insurance contract to a region whose shocks happened to be adverse and very permanent. The point is that there is a great deal of evidence that such contracts are difficult to honor even within countries whose idea of nationhood and mutual responsibility is at least a century older than EMU: one should not expect substantial amounts of insurance for future but permanent shocks, what were earlier called supply shocks, through any pan-EMU federal mechanism.

What about temporary shocks? Here too the previous benefits of mutual insurance through federation may have been exaggerated. The degree of integration that makes monetary union attractive simultaneously implies a positive correlation of output shocks that yields little gain through mutual insurance. Putting the same point in another way, one country may get federal help in a slump but partner countries are also likely to be in a slump, so the federal budget will go into deficit and the present value of the recipient country’s future tax liabilities will rise correspondingly. With Ricardian equivalence, net gains would be small, a point emphasized in Bayoumi and Masson (1995) and Fatás (1997). Of course, Ricardian equivalence does not hold exactly, so there are some liquidity benefits from such a policy. Moreover, the more one wishes to rely on the high correlation of output shocks to dismiss the efficacy of mutual insurance, the more one is diminishing not merely the solution but the problem: with high output correlation, the single monetary policy is increasingly well placed to respond to the concerns of all member states. Thus, part of the concern that lies behind this strand of research is the fear that this single monetary policy will pay insufficient attention to output stabilization, either because the monetary strategy of the ECB has been formulated more narrowly than the discussion in the section “EMU’s Monetary Strategy” suggests is optimal or because, despite appropriate design for the steady state, early emphasis on reputation building requires the ECB to signal early toughness.

It is therefore likely that countries will face asymmetric shocks and wish to use their own national fiscal policies. For temporary shocks, national fiscal policies provide a natural means of intertemporal smoothing; for permanent shocks, intertemporal smoothing is largely inappropriate. If supply has fallen permanently, demand needs to adjust one way or another.

This prescription glosses over the very real difficulty of diagnosing shocks soon after they occur. Nor does reliance on automatic stabilizers rather than fiscal discretion provide any solution to this dilemma: permanent shocks require (discretionary) changes in the setting of tax rates and spending levels. Automatic stabilizers are therefore no panacea. Their continuing attractiveness rests on an empirical judgment that demand shocks are sufficiently prevalent to make them part of normal operating procedure; governments (and economists) also have to be alert to the situations in which other action is called for.

The Stability and Growth Pact: How Much Will It Impede Output Stabilization?

The Stability and Growth Pact, agreed in Dublin in 1996 and confirmed in Amsterdam in 1997, provides a framework for maintaining and enforcing the Maastricht fiscal criteria after EMU has begun. The pact restates the commitment to a maximum budget deficit of 3 percent of GDP and, except in special circumstances, applies penalties to deficits that exceed this level. Exemption from penalty will normally apply to countries whose real GDP has fallen 2 percent within a period of four quarters; in cases of output falls of between 0.75 percent and 2 percent, penalties may or may not be applied at the discretion of the Council of Ministers; overshooting the deficit limit without the alibi of even an output fall of 0.75 percent will automatically trigger penalties. Eichengreen (1997) calculates that, among the 15 individual countries that make up the EU, GDP has fallen 2 percent in a four-quarter period only 13 times in the past 30 years, an average of less than once per country in three decades.

Let us agree that exemptions will be rare and that incentives to avoid penalties will be large: countries will therefore try to keep budget deficits below 3 percent of GDP.³⁰ One implication of the preceding analysis is that, since countries will want some room for fiscal maneuver in order that they can respond to adverse country-specific output shocks, they should aim for an average budget position that allows scope for substantial loosening in a crisis.³¹ Suppose for the sake of argument this turned out to be roughly budget balance on average. Thus in the long run, nominal debt of the government would be constant, while nominal GDP continued to grow. Despite occasional cycles, the trend in debt/GDP ratios would be inexorably down. For countries whose debt/GDP ratios are already well within the 60 percent level, the conclusion may be that the Stability Pact is unnecessarily tight.

Those uncomfortable with this conclusion might wish to base their case on the poorness of current budget balances as an indicator of longer-run sustainability, notably because the (large) extent of unfunded future pension obligations of the government can already be calculated. While this does indeed provide a motive for further fiscal adjustment, it also undermines the case for a uniform ceiling on current fiscal deficits; countries such as Germany and France have much larger unfunded pension liabilities than say the United Kingdom.

Tight fiscal policy on average need not pose any problems for output in the long run, especially if accompanied by looser monetary policy (which in such circumstances would not jeopardize price stability) But fiscal tightening imposes a substantial deflationary bias in the short run if monetary policy is not correspondingly relaxed. Indeed, given the difficulty that many (most?) countries have had squeezing under the Maastricht wire, there is a presumption many EMU members will initially be very close to the 3 percent limit: it may be hard for them quickly to engage in another bout of fiscal tightening.

The objective of the Stability Pact and Growth is not to promote short-run output volatility but to safeguard fiscal solvency, intrinsically a long-run consideration. This invites the response that the Stability Pact should apply to structural (i.e., cyclically adjusted) deficits not to raw unadjusted deficits.³² Any redesign in this way would of course have to recalibrate the whole exercise: an average or structural deficit of 3 percent yields quite different long-run properties from a maximum deficit of 3 percent. It might, for example, be recast as a requirement to pursue budget balance over the cycle.

The long-run characteristics of such a policy might not differ dramatically from the existing Stability Pact, provided, under the latter, countries aim for a “precautionary fiscal safety margin” in order then to have scope for accommodation of adverse countryspecific shocks. However, the former variant is preferable: by putting too little pressure on countries’ fiscal position during an upswing, the Stability Pact may leave countries with too little scope to accommodate downswings.³³

A more limited innovation, suggested in Eichengreen (1997), would be to utilize the evidence from cyclically adjusted budgets to adjudicate, within the existing framework of the Stability Pact, when a penalty should or should not be applied within the range of moderate recession, that is, when there is a four-quarter output fall in the range of 0.75–2 percent.

Conclusion

Integration of Europe’s labor and product markets, achieved more through trade and capital mobility than through movement of workers, has proceeded substantially. There is a clear group of “core countries” and to date they behave differently from those in the periphery. Participation in EMU will change historical patterns of correlation and is likely, though not guaranteed, to increase these correlations. It is unlikely quickly to change the geographical and cultural realities that make the periphery distinct from the center, and, even within the core, country-specific shocks will remain.

Although, as a consequence, real wage flexibility and factor mobility are likely to increase somewhat, it would be optimistic to expect rapid changes in behavior. A role will remain for fiscal policy, more through automatic stabilizers than through discretionary changes, except when obvious supply shocks have been identified, in which case a discretionary response is needed. Federal fiscal policy will play little role in the foreseeable future, and countries close to the 3 percent limit for budget deficits will be constrained by the Stability Pact. In practice, therefore, early EMU participants face the unpalatable choice of engaging in yet another round of fiscal tightening in order to build up precautionary fiscal surpluses, as a baseline from which countercyclical fiscal policy is then possible, or else risk entering EMU with budget deficits sufficiently close to the permitted maximum that any setback to recovery in Europe would quickly force members to take discretionary action to switch off the automatic stabilizers that would normally kick in during a recession.

The latter is a distinct possibility. It would render immediately irrelevant empirical studies of historical correlations of output: countries with spare fiscal capacity would allow fiscal policy to loosen as in the past, countries hard up against their constraint would not. Knowledge of the fiscal position would be a key piece of information in predicting which countries were correlated with which; it would imply a fiscal core and a fiscal periphery, not well correlated with the grouping with which one has become familiar.

In contrast, redefining the Stability Pact with explicit reference to structural budgets would have the short-term conjunctural advantage of distinguishing the continuing need to improve underlying budget positions over some medium run from the necessity of doing so precisely at a time when Europe is in recession.

Fiscal tightening need not imply recession. In principle, what is taken away with fiscal policy can be given back with monetary policy. Indeed, one of the key arguments for a regime in which operational control of monetary policy is delegated to an independent central bank is that it should remove the presumption that fiscal tightening invariably leads to output contraction: in pursuit of price stability, the central bank will undo the effects on EMU aggregate demand that fiscal authorities create. However, the “conjunctural” concern is that, until the ECB has established its reputation, it may be unwilling to do this to the extent required.