Looking forward, central bankers

Jeffrey C. Fuhrer

Assistant Vice President and Econo-mist, Federal Reserve Bank of Boston.This article summarizes the Bank’seconomic conference held June 19-21,1994.

T he rate of inflation in the consumer price index over the pastthree years has been low and stable, averaging 2.8 percent andnever exceeding that average by more than one-half percentage

point in any quarter. Attending this success on the inflation front hasbeen a gradual decline in the unemployment rate to a level that mosteconomists agree is consistent with full employment. In broad terms,the Fed appears to have achieved the low-inflation "soft landing" that itsought."

Attaining this desirable economic state was not an easy task. Alongthe way, the Federal Reserve had to balance the often competing goals

10f’ price stability and full employment, relying on a broad set of.indicators to guide monetary policy in a changing financial environ-ment. Maintaining this desirable state presents significant challenges aswell. As Donald Kohn suggests in his comments on a paper by WilliamPoole, "’a central bank believing that it had learned sufficiently from itshistory to guarantee that it woi~ld not repeat its mistakes would besuffering a serious attack of hubris."

Looking forward, central bankers in the United States and abroadmust grapple with a broad array of questions about how best to conductmonetary policy. How much should the goal of price stability beemphasized relative to the goal of employment stability? Does centralbank independence aid in achieving either or both of these goals? Doesa stable, short-run trade-off between inflation and unemployment exist,and can it be exploited by the central bank? And whether such ashort-run trade-off exists or not, is there a long-run trade-off in thevariability of employment and inflation? What instrument should thecentral ba’nk manipulate in order to achieve its short-run and long-rungoals? What: indicators will prove most reliable in signalling the leveland direcfi6n of change of the central bank’s ultimate goals?

In June of 1994, the Federal Reserve Bank of Boston sponsored aconference to address these questions. The five papers presented at the

conference fall into three broad areas. First, JohnTaylor and Jeffrey Fuhrer each discuss the efficiencyof U.S. monetary policy, taking as given that policyhas both inflation and (in the short run) outputtargets, and that monetary policy adjusts an interestrate instrument in response to deviations of inflationand output from their target values. William Poole’spaper (which by itself constitutes the second group)suggests ways in which the monetary aggregates maystill be useful for the conduct of monetary policy. Thethird group, which comprises papers by CharlesGoodhart and Jos~ Vifials and by Guy Debelle andStanley Fischer, examines international evidence inorder to shed light on the questions of central bankindependence and accountability. A concluding panelconsidered ways in which monetary policy could be

Looking forward, central bankersin the United States and abroadmust grapple with a broad arrayof questions about how best to

conduct monetary policy.

improved, in light of the discussion in the precedingsessions.

As one might expect, it was impossible to reach aconsensus on many of the issues. Opinion rangedwidely about how much emphasis should be placedon stabilizing employment relative to prices. Oneview suggested that the Fed cannot reliably affect anyreal variables and thus should not try to control them;the other worried about the seemingly exclusivefocus on price stability and suggested that monetarypolicy must be responsible for prompt and appropri-ate management of real variables. Laurence Ball andJeffrey Fuhrer reached exactly opposite conclusionsabout whether gradual or "cold turkey" disinflationswere less disruptive. Finally, the assembled groupdisagreed about the nature of the "monetary trans-mission mechanism"--how changes in monetary pol-icy instruments, such as the federal funds rate, affectthe ultimate goals of policy.

Still, several broad conclusions emerged from theproceedings. First, many conference participantsagreed that U.S. monetary policy had been quite

successful over the past 15 years. The use of aninterest rate instrument to bring inflation under con-trol while minimizing disruption to output and em-ployment has been a winning strategy. Second, mostagreed that the role of the monetary aggregates in theconduct of monetary policy has been and shouldremain downgraded. Finally, most conference partic-ipants agreed with the broad conclusions of theDebelle and Fischer paper, namely that clear articu-lation of the central bank’s goals is desirable, whileconstraints that dictate how the goals should beachieved are not desirable.

How Efficient Has Monetary Policy Been?John Taylor’s paper, "The Inflation!Output Vari-

ability Trade-off Revisited," considers the trade-offsbetween inflation and output that monetary policyfaces in pursuing its ultimate goals. If no long-runtrade-off exists between inflation and real output, asMilton Friedman and Edmund Phelps first suggestedand most economists today accept, and if we ac-knowledge considerable uncertainty about the natureof the short-run inflation/output trade-off, then isthere any such trade-off that may be reliably exploitedby monetary policy? If not, then monetary policyshould focus exclusively on inflation (or the pricelevel) and ignore the consequences, if any, for thereal economy.

The Inflation/Output Variabilit~d Trade-off

Taylor suggests that we consider the inflation/output variability trade-off. Its essence is straightfor-ward: Keeping the inflation rate extremely stableabout a target may entail accepting much greaterfluctuations of GDP about potential (or unemploy-ment about the natural rate), even in the long run. Ifso, monetary policy may wish to balance its effects oninflation and output variability.

The Taylor paper provides a simple motivationfor the long-run trade-off. The motivation is based ona textbook macroeconomic model in which outputdepends on real interest rates, inflation responds todeviations of GDP from potential, and monetarypolicy sets the short-term nominal interest rate inresponse to deviations of inflation from target anddeviations of output from potential. The combinationof the aggregate demand equation and the policyresponse implies that the output gap is negativelyrelated to deviations of inflation from its target: If

4 SeptemberlOctober 1994 New England Economic Review

inflation exceeds its target, monetary policy will raiseinterest rates and depress output.

This simple characterization of the macroecon-omy makes it easy to see why a trade-off betweeninflation and output variability may exist. When theeconomy is subjected to a price shock that raises theinflation rate, for example, the amount of outputdisruption that will occur depends on the response ofmonetary policy to inflation and output deviations.The more vigorously the Fed moves the interest rate

The essence of Taylor’s inflation/output variability trade-off isstraightforward: Keeping theinflation rate extremely stable

about a target may entailaccepting much greater

fluctuations of GDP aboutpotential. If so, monetary policymay wish to balance its effects oninflation and output variability.

to offset deviations of output from potential, thesmaller will be the variability of output and the largerwill be the variability of inflation. The converse is alsotrue. Thus this simple model, coupled with alterna-tive monetary policy behaviors, suggests a trade-offbetween the variability of inflation and the variabilityof output that monetary policymakers may be able toexploit in the long run. Based on rough calibration ofthe model to recent history, Taylor suggests that anapproximately balanced response to inflation andoutput deviations would yield roughly equal varianceof inflation and output.

Taylor also discusses other potential long-runtrade-offs, especially the effect of inflation on poten-tial GDP, which is ruled out in the simple model thathe uses. Reviewing work by Fischer (1993) and Mot-ley (1994), he suggests that the link between inflationand productivity growth merits additional study...~

Finally, Taylor considers the possibility that out-put fluctuations affect long-run growth, an idea thatdates back to Schumpeter (1939). The notion thatrecessions might provide opportunities for firms to

make structural adjustments that enhance produc-tivity--a "cleansing effect"--has recently been ad-vanced by Davis and Haltiwanger (1990) and Cabal-lero and Hammour (1991). Taylor finds this link fromfluctuations to growth unpersuasive, since a gooddeal of restructuring (through "job destruction") oc-curs during years when output is at or above poten-tial. In addition, he suggests that greater outputvariability would have no net effect on the amount ofrestructuring, as larger positive fluctuations woulddecrease job destruction, while larger negative fluc-tuations would increase job destruction. The neteffect of increased variability on productivity-enhanc-ing restructuring would be zero.

Discussant Laurence Ball agrees with Taylor thatmonetary policymakers ought to focus more on me-dium- to long-term strategy than on the short-runtrade-offs involved in the Phillips curve. Thus, theattention to the variability of inflation and output isappropriate. He also applauds the simplicity of themodel used to motivate the variability trade-off butcautions that, while the model may be quite useful fornormative purposes, it may be less useful for positivepurposes. The reason is that the model assumes thatinflation always reverts to the monetary authority’sfixed inflation target whereas, over the past severaldecades, the inflation target appears to have movedaround with a good deal of persistence. Understand-ing monetary policy has largely been a matter ofunderstanding why the inflation target has changed,Ball suggests. Thus, while the model may fit thebehavior of the economy since the late 1980s quitewell, it is unlikely to fare well in explaining thebehavior of the economy from the 1970s, when theinflation target apparently drifted up, through the1980s, when the target declined precipitously underthe direction of Fed Chairman Paul Volcker.

Ball, however, expresses some doubt that poli-cymakers face a "painful trade-off between morevariable output and more variable inflation." Henotes that if demand shocks--shocks to Taylor’s I-Scurve and policy rule--are the only important sourcesof fluctuations, then it is, in principle, possible for theFed to eliminate all of the variability in both outputand inflation. In Taylor’s simple model, in the face ofa demand shock~an unexpected surge in defenseexpenditures, for example--the Fed can, by raisingthe interest rate tremendously, offset any effect of theshock on output and on inflation. Ball recognizes thatTaylor’s model abstracts from important features ofthe economy that make it very difficult in practice forpolicymakers to completely offset demand shocks.

September/October 1994 New England Economic Review 5

In the face of significant supply shocks--unex-pected increases in the inflation rate in Taylor’smodel--Ball professes agnosticism about the pres-ence of a trade-off between inflation and outputvariability. He notes, however, that in Taylor’s sim-ple model, the sum of the deviations of output frompotential after a supply shock is invariant to the

Ball agrees that monetarypolicymakers ought to focus moreon medium- to long-term strategythan on the short-run trade-offsinvolved in the Phillips curve.

particular policy response chosen. The timing of thedeviations can be affected: A policy that puts greaterweight on output will spread the output deviationsover a longer, smoother path. This reduces the vari-ance of output, but not the sum of the output losses.Simply put, Ball questions whether two years of 1percent lower output are preferable to one year of 2percent lower output. Measured by variability, thefirst outcome would be preferred.

Finally, Ball suggests that a policy that tried tominimize output variability might not actually pro-duce less output loss, although Taylor’s model im-plies that it would. The reason is related to Ball’sfinding (Ball 1994) that moving inflation back gradu-ally to its target is more costly than a rapid decrease ininflation. If so, then a policy that tried to minimizeoutput variability by gradually reducing inflation after asupply shock could actually, increase the output loss.

Optional Policy Responses to Inflationand Output Fluctuations

Jeffrey Fuhrer’s paper on "Optimal MonetaryPolicy and the Sacrifice Ratio" focuses on an age-oldquestion: Is it less costly to disinflate gradually, orrapidly? In the small macro model developed previ-ously in Fuhrer and Moore (1994), he finds thatgradual disinflation is less costly. The reason is that,in a world in which wages and prices are predeter-mined by contracts, previously negotiated contractwages and prices cannot adjust immediately to theannouncement of a disinflation. The more quickly

and vigorously the Fed disinflates, the more contractsit catches unexpired. When these contract wages andprices cannot adjust to a monetary contraction, quan-tifies of labor hired and goods produced must adjust,and thus the disinflation causes more disruption tothe real side of the economy.

According to estimates presented in the paper,the U.S. central bank (the Federal Reserve) has re-cently chosen monetary policies that emphasize in-flation far more than they emphasize deviations ofoutput from potential. The consequence has beenthat the "sacrifice ratio"--the shortfall of outputbelow potential, per percentage point decrease in theinflation rate--has been quite high during the disin-flations of the past 12 years. The paper suggests thatthe sacrifice ratio could have been lowered substan-tially by increasing the emphasis on output fluctua-tions in the Fed’s reaction function.

If the Fed were already responding optimally toinflation and output fluctuations, increasing empha-sis on output fluctuations would of necessity yieldimprovements in the variability of output at theexpense of increased variability of inflation about itstarget. But could the responses required to reduce thesacrifice ratio also yield decreases in the variability ofboth output and inflation about their targets? Fuhrerargues that they could. Because vigorous inflationresponses of the Fed have been suboptimal--they did

Fuhrer finds that gradualdisinflation is less costly than

rapid, because previouslynegotiated contract wages and

prices cannot adjust immediately,and thus rapid disinflation

causes more disruption to thereal side of the economy.

not result in the smallest inflation and output vari-ability combination attainable--the Fed could alterits responses to inflation and output so as to lowerthe sacrifice ratio and decrease the variability of infla-tion about its target. Thus, the Fed could achieveimprovement on all fronts by suitable reaction to itsultimate goals.

6 September/October 1994 New England Economic Review

N. Gregory Mankiw finds three broad areas ofdisagreement with Fuhrer’s paper. The first is moti-vation: Why should we care about the sacrifice ratioin the way Fuhrer has defined it? In the typicaldiscussion of the sacrifice ratio, one wishes to mini-mize the output loss during a one-time reduction in

Mankiw doubts that gradualismis less costly than cold turkey,and he argues that credibility

effects may be extremelyimportant in determiningthe cost of disinflations.

the inflation rate. But this paper looks at the ongoingeffect of a particular monetary policy rule on thesacrifice ratio. In this context, a larger sacrifice ratiomeans a larger output loss when the inflation targetfalls, but it also implies a larger output gain when theinflation target rises. A better measure for this type ofongoing concern for output volatility is the varianceof inflation, also considered in the paper.

The second disagreement is with respect tomethodology. Mankiw suggests that, because expec-tations enter the model only through the wage con-tracting mechanism and through the effect of long-term interest rates in the I-S curve, the model maystill be subject to instability across policy regimes,that is, the Lucas critique. In addition, Mankiw findssome of the identifying restrictions imposed by therational expectations assumption in this model to beakin to Sims’s "incredible" identifying assumptions.Mankiw stresses that we do not know enough aboutthe price-adjustment process to trust the policy con-clusions that arise from a particular rendering of thesticky-price paradigm. He argues that we need to findrules that are robust across a wide variety of compet-ing models.

Finally, Mankiw doubts the paper’s main conclu-’sion that gradualism is less costly than cold turkey.Citing cross-country comparisons by Ball (1994, forth-coming) that impose little structure on the data,. ~efeels more comfortable with the empirical regularityfound there, which indicates that more rapid disin-flations are less costly. In addition, Mankiw arguesthat credibility effects, ignored in the Fuhrer paper,

may be extremely important in determining the costof disinflations. He cites the disparity between theCouncil of Economic Advisers’ forecasts of inflationfor the five years beginning in 1981 and the actualoutcomes for those years as evidence that the Volckerpolicy was "not credible even to the Administrationthat had appointed Volcker" and thus may haveplayed a role in the recession that accompanied thedisinflation.

Summary discussant Martin Eichenbaum pointsout the similarities between the frameworks used bythe Fuhrer and Taylor papers. Both assume thatmonetary policy uses the short-term nominal rate asits instrument, that the inflation rate responds slug-gishly to aggregate demand, that policy-induced risesin the short-term rates are mirrored in long-term realrates, that long-term real rates affect aggregate de-mand, and that monetary policy affects inflationthrough its effect on aggregate demand.

Eichenbaum points out that the common struc-ture employed by Fuhrer and Taylor ignores many ofthe financial market imperfections---credit crunches,liquidity constraints, and the like--that academicsand Fed Chairman Alan Greenspan have alluded toin recent policy discussions. He considers the lack ofdirect evidence in support of the assumed monetarytransmission a weakness of both papers.

Second, Eichenbaum suggests that while themodels used in both the Taylor and Fuhrer papersimply an inflation variability/output variability trade-off, both papers should have included some directevidence of the trade-off.

Eichenbaum concludes that anyempirical rendering of a Fed

reaction function should include areaction to the forward-looking

information in commodity prices.

Eichenbaum then explores a vector autoregres-sion (VAR) analysis of the three variables consideredin the Taylor and Fuhrer papers. He finds that, for aparticular ordering of the variables in the VAR, apositive shock to the funds rate causes a rise in theinflation rate. He suggests that this puzzling correla-tion arises because commodity prices are excluded


from the reaction function. The positive response ofinflation to an increase in the funds rate in thethree-variable model is really masking a positiveresponse of the funds rate to a rise in commodityprices--which preceded rises in inflation in the1970s--and a subsequent fall in inflation.1 Thus,Eichenbaum concludes that any empirical renderingof a Fed reaction function should include a reaction tothe forward-looking information in commodity prices.

Comparing Direct andIntermediate Targeting

William Poole provides a monetarist perspectiveon the question of where monetary aggregatesshould fit into the current policy process. Focusing onthe past dozen years, Poole acknowledges both theproblems with the behavior of monetary aggregatesand the success in using an interest rate instrument toconduct monetary policy. However, he counsels thatrecent experience does not preclude effective use of amonetary aggregate in the conduct of monetary pol-icy. He suggests that "there is a strong case forpaying much more attention to M1 than has been truein recent years."

Poole proposes a modification ofcurrent monetary policy that

builds on the successful use ofthe interest rate instrument

but allows a role formoney growth targets.

Poole suggests two explanations for the break-down between money growth and inflation in recentyears. The first is that, in an environment of lowinflation and low nominal interest rates, the penaltyfor holding non-interest-bearing money is small. As aresult, fluctuations in the stock of money created bythe central bank are largely absorbed by the public;they do not translate into higher inflation,a Thesecond is that a consequence of a well-executedmonetary policy is that the observed correlation be-tween monetary policy instruments and policy goalswill be zero. If the Fed has moved its policy instru-

ments (monetary aggregates) so as to pin its ultimategoals at their targets, then one will not be able toobserve any correlation between the instrument set-tings and the ultimate goal, since the goal has notmoved from its target. A corollary to this propositionis that a search for the best monetary aggregate bycomparing correlations of aggregates to policy goalswill be unsuccessful if the Fed is doing a good job.

Poole points out that monetary policy whenusing an interest rate instrument is less predictableand more difficult to communicate to the public thanmonetary policy when using a monetary instrument.Generally, a 1 percentage point decrease in’moneygrowth yields a 1 percentage point decrease in infla-tion and nominal interest rates in the long run. Thesimplicity of the monetary prescription for loweringinflation is lost when using an interest rate instru-ment, however. In order to lower inflation, the Fedmust first raise nominal interest rates, then lowerthem. And Poole argues that we cannot say with anyconfidence how much of an increase in rates is requiredto lower the inflation rate 1 percentage point.

Poole suggests that the difficulty of the Fed’s jobunder an interest rate regime is compounded by theinteraction of the Fed’s expectation of how its actionswill affect the credit markets with the credit markets’expectations of how the Fed will act. He asserts thatit may be impossible to build a model that incorpo-rates this simultaneity of expectations and implies areliable rule of thumb such as the 1 to 1 rule impliedby a monetary aggregates approach.

A Proposed New Role for Money Growth Targets

In light of the preceding observations, Pooleproposes a modification of current monetary policythat builds on the successful use of the interest rateinstrument but allows a role for money growth tar-gets. He suggests that the Fed should allow thefederal funds rate to "vary within a considerablywider band, perhaps 100 basis points, betweenFOMC meetings," as the demand for bank reservesfluctuates, keeping the supply of bank reserves on a

I In contrast, the Lrnpulse responses for Fuhrer’s model re-ported in Fuhrer and Moore (1994) show that inflation falls follow-ing a positive shock to the funds rate.

2 One standard description of the link from money creation toincreased inflation is as follows. If the Fed wishes to increase thestock of money, it must induce the public to hold the money byreducing the cost of holding money--the interest rate on alterna-tive means of storing value. A fall in the interest rate raises demandfor interest-sensitive spending, which may increase aggregatedemand sufficiently to put upward pressure on prices.


steady path. The advantages of this policy, accordingto Poole, would be twofold. First, the transition tohigher or lower interest rates would be smootherthan the discontinuous path followed by rates underthe current regime. Second, movements of creditmarket rates could once again provide importantinformation to the Fed, as rates would reflect themarkets’ assessments of the significance of incomingdata, not only "market speculation on how the Fedwill respond to the data."

Benjamin Friedman reads the history of usingthe monetary aggregates to guide monetary policysomewhat differently. In response to Poole’s two-pronged defense of monetary aggregate targeting,Friedman voices several objections. First, he arguesthat the objection that "no baseline prediction exists... as to how much ... inflation will rise if thecentral bank, say, lowers interest rates by 1 percent-age point" is invalid; the two papers in the first

Friedman suggests that theempirical support for the interestrate approach is arguably stronger

than that for the monetaryaggregates approach.

session of this conference provide examples of mod-els that do exactly that. Conversely, a stable moneydemand function, the cornerstone of the baselinemoney model, is nearly impossible to find in the U.S.data. Thus, the empirical support for the interest rateapproach is arguably stronger than that for the mon-etary aggregates approach. In addition, Poole’s ob-jection to a policy that permanently fixes the nominalinterest rate carries little force, because no one hasever suggested that the central bank pursue such apolicy.

Second, Friedman dismisses Poole’s explanationof the vanishing money-income correlation. Fried-man points out that, even if the Fed had pursued anoptimal monetary policy, the partial correlation~be~tween money and income--the correlation holdingthe effects of other variables on income constant--would not be driven to zero; in fact, it would in-crease. Thus, the estimates of the partial correlation

between money and income are not consistent withPoole’s optimal monetary policy story.

Donald Kohn focuses on Poole’s proposal to fixthe supply of reserves and allow the federal fundsrate to fluctuate within a band in response to changesin the demand for reserves. Kohn argues that theunpredictability in the demand for reserves wouldyield a volatile funds rate that often hit the upper orlower end of its band, imposing significant uncer-tainty on financial markets. He also asserts that it

Kohn argues that no feasiblealternative is available to thepresent practice of using a

short-term interest rate as apolicy instrument and looking

at all kinds of informationto gauge progress.

would be neither more nor less difficult to obtaininformation from asset prices under the fluctuatingfunds rate regime; market prices would still be deter-mined in part by expectations of short-rate move-ments, now with the added burden of anticipatingreserves demand.

Policymakers have drawn two important lessonsfrom the experience of the past 25 years, Kohnargues. First, no feasible alternative is available to thepresent practice of using a short-term interest rate astheir policy instrument and looking at all kinds ofinformation to gauge their progress. Second, giventhe lags in the monetary transmission mechanism,policymakers must be ready to move their instrumentquickly in response to new information.

Lessons from International ExperienceCharles Go0dhart and Jos~ Vif~als’s paper "Strat-

egy and Tactics of Monetary Policy: Examples fromEurope and the Antipodes" provides a comprehen-sive taxonomy of the current and projected issuesfacing monetary policymakers in Europe, Canada,and Australia/New Zealand. They first document thatin virtually every country, price stability has becomethe primary objective for the central bank. Interest-


ingly, where legislation has accompanied the focuson price stability, it is rare to find a precise definitionof price stability. Most, although not all, arrange-ments allow the central bank to respond to othereconomic conditions, often with the stipulation thatthe prime directive be accomplished first.

Price Stabilit~y: The Central Bank’s Pri~namd Goal

As Goodhart and Vifials note, much of the sup-port for an independent central bank with a primaryobjective of price stability has come from the theoret-ical economics literature. The time inconsistency ar-gument, for example, asserts that central banks un-der pressure from the electorate will consistentlyaccept unexpected output gains at the cost of in-creased inflation, thus building in an inflationarybias. While this bias towards inflating is widely citedas an argument in favor of legislating price stability asthe only goal of the central bank, relatively littleempirical backing for the inflationary bias exists, andthus some have questioned the exclusive focus onprice stability. As an alternative, many economistshave suggested a nominal GDP target, which givesequal weight to prices and to deviations of output

Goodhart and Vi~als documentthat in virtually every country,price stability has become the

primary objective for the centralbank, and point out the possibility

of a deflationary bias.

from potential. Goodhart and Vifials point out thatcentral banks nonetheless have overwhelminglyopted for the price stability goal, perhaps becausepotential GDP is hard to estin~ate; data on GDP areavailable only with a lag and are subject to revision;and a focus on price stability underscores that centralbanks cannot be responsible for real variables in thelong run.

The paper goes on to review the more detailedissues pertaining to the achievement of price stability:Should the central bank target the price level or therate of change of prices? Should central banks adopttarget ranges for prices, rather than point targets? At

what horizon should the central bank announce thatit intends to attain its goal? Which index (producerprices versus consumer prices, for example) shouldbe used as the measure of price performance? Shouldexplicit contracts that reward central bankers for goodperformance be used to provide the incentive toachieve the goal?

Next, Goodhart and Vitals address the merits ofdirect versus intermediate targets in achieving pricestability. Citing Persson and Tabellini, they arguethat "An inflation contract . . . generally dominatescontracts based on intermediate monetary targets."Nonetheless, relatively few direct inflation targets areobserved among central banks clearly concerned withprice stability. Apart from historical accident, onereason may be that the effect of monetary policy onprices occurs with considerably more delay than theeffect on monetary aggregates or other financial vari-ables. Thus, use of a financial aggregate as an inter-mediate target could provide an earlier signal thatpolicy has deviated from the agreed-upon course.Most European countries have made the exchangerate their primary target, on the grounds that itresponds instantaneously to interest rates and iswidely understood by the public. The larger and lessopen countries, such as Germany, France, and theUnited Kingdom, have chosen monetary targets,primarily in their belief that monetary aggregates arereliably linked to nominal variables, can be controlledby the central bank, convey information to the publicabout the stance of monetary policy, and thus facili-tate monitoring by the public of monetary policy.

Goodhart and Vifials point out the possibility ofa deflationary bias among central banks committed toprice stability.3 Given uncertainty about both thestructure of the economy and the shocks that mightperturb the economy during the delay between policyaction and its effect on prices, central banks mayattempt to lower inflation to its target level quickly,so as not to suffer derailment at the hands of unpre-dictable events. In fact, the experience in both Can-ada and New Zealand is consistent with this hypoth-esis: Both central banks have reduced inflation to, orbelow, their target levels in advance of the agreedhorizon.

Finally, Goodhart and Vi~als discuss the impactof a monetary union on monetary strategy and tacticsin Europe. Countries currently differ significantly

3 Tl-ds hypothesis provides an interesting counterpoint to theinflationary bias of central banks suggested by Barro and Gordon(1983).


with regard to implementation of monetary policy:Reserve requirements, the discount window, andopen market operations are used to differing degreesacross Europe. Considering the diversity of currentpractice, the need to unify both policy formulationand policy execution remains a daunting task for theEuropean Monetary Institute.

Cooper stresses that a centralbank should remain accountable

to the political processeven though reasonably

independent of it.

Richard Cooper points out that the excellentprice stability performances by the central banks inthe United States and Japan--the first an indepen-dent bank with no explicit targets, the second acentral bank with little independence--run counter tothe generalizations drawn in the Goodhart and Vifialspaper. He also criticizes the easy acceptance thatGoodhart and Vifials grant to price stability as thecentral bank’s primary objective. Cooper stresses theimportance of the central bank’s role in maintainingthe smooth functioning of the financial system in theface of large real and financial shocks, and also the"lubrication" that inflation can provide in allowingreal wage adjustments when nominal wages aredifficult to reduce.

Cooper points out the importance of the distinc-tion between the independence and the accountabil-ity of a central bank. The central banks of the UnitedStates and Germany, he claims, are reasonably inde-pendent of the political process, but they are stillaccountable to it. The design of the European Systemof Central Banks essentially makes the central bankcompletely independent of the political process. Coo7.per finds this institutional arrangement "highly un-desirable" because it removes a degree of longer-termaccountability to the political process from the centralbank’s actions. Finally, he dismisses other rational-izations of the focus on price stability--money oialyaffects prices in the long run, inflation decreases realgrowth and productivity--as lacking in empiricalsupport.

The Costs and Benefits ofCentral Bank h~dependence

Guy Debelle and Stanley Fischer’s paper "HowIndependent Should a Central Bank Be?" answers thequestion with a blend of sensible interpretation ofempirical regularities and compact theoretical analy-sis. The authors stress the multidimensional nature ofcentral bank independence. In particular, they distin-guish between goal independence and instrument in-dependence. They argue that the optimal outcomemay be one in which a legislative body sets thecentral bank’s goals, but the central bank sets itsinstruments however it believes it can best attain theprescribed goals.

Debelle and Fischer begin by reviewing the re-sults that relate measures of central bank indepen-dence and macroeconomic outcomes for variouscountries. They find that independence is negativelycorrelated with the rate of inflation: Countries withmore independent central banks generally experiencelower inflation rates. In addition, countries withgreater central bank independence appear to attainbetter economic performance, perhaps because theyare generally better disciplined and thus suffer fewerand smaller self-inflicted shocks. Thus, indepen-dence appears to be a "free lunch": Increased central

Debelle and Fischer distinguishbetween goal independence andinstrument independence for acentral bank, arguing that theoptimal outcome may be one in

which a legislative body setsthe goals, but the central bank

sets its own instruments.

bank independence yields better inflation outcomeswith no loss to output.

Having said this, Debelle and Fischer turn to acomparison of German and U.S. performance duringrecent disinflations. Many believe that when a morecredible central bank announces a disinflation, ex-pected inflation will fall, prices will adjust in line withthe newly expected inflation rate, and output will not


suffer. Thus disinflations should be noticeably lesscostly in countries with credible central banks. TheBundesbank widely viewed as the most crediblecentral bank in the world--should have earned a"credibility bonus" that would allow it to disinflatewith less cost than a central bank without suchcredibility. Debelle and Fischer, drawing on work byBall (1994), find that German disinflation has beenpurchased at a higher cost than U.S. disinflation,particularly in the case of the 1981-86 episode. Inaddition, they find that this relationship extendsbeyond the U.S.-German comparison. For the coun-tries in their sample, the output loss associated witha disinflation is higher for countries with greatercentral bank independence. This finding suggests acost to greater independence, and is consistent withtheir conclusion that independent central banks mustbe held accountable for their actions, so that they donot pursue price stability to the exclusion of aggre-gate demand management.

In discussing Debelle and Fischer’s paper, RobertHall points out an intriguing irony in the evolution ofmacroeconomic theory and monetary policy imple-mentation. Soon after the academic communitywarned of the inherent inflationary bias of central

Hall suggests that the bestsolution is to appoint central

bankers with our preferences andto build in incentives thatpenalize chronic inflation.

banks~which arises "for the same reason that ajudge will impose too lenient a sentence on a miscre-ant-the crime has already been committed and thesentence can’t deter it"-~central banks proceeded torelentlessly wring inflation from most of the devel-oped countries in the world. Thus, the predictionmade by believers in the inflationary bias not onlywas not borne out, it was sharply contradicted bycentral banks around the world.

Hall regards the conclusions drawn by the De-belle and Fischer paper as "schizophrenic" withregard to the relationship between central bank inde-pendence and output volatility. Early in the paper,they suggest that the pursuit of hawkish policies hasno cost in terms of real performance. On the other

hand, their final figure shows that hawkish countriesappear to have more severe recessions. Germany andthe United States have low output variances but thelargest output sacrifice ratios during disinflations.Thus, any conclusion about the costs of maintainingcentral bank independence depends critically on themeasure of output loss used.

With regard to the theoretical section of thepaper, Hall points out that the Debelle-Fischer modelviolates Friedman’s natural rate law. Sustained andfully anticipated inflation stimulates output in theirmodel and creates a bias towards inflationary mone-tary policies.

Finally, Hall emphasizes that he agrees with thebasic conclusion of the paper. We should not appointcentral bankers who reflect our own preferences,since they will tend to produce too much inflation.One approach is to appoint inflation hawks, as inRogoff (1985); the problem with this approach is thathawks will consistently underrespond to recessions.The best solution is to appoint central bankers withour preferences and build in incentives that penalizechronic inflation.

Panel DiscussionThe conference closed with a panel discussion

among five eminent macroeconomists. The panelrevisited and expanded upon many of the themestaken up in the preceding sessions.

Paul Samuelson warns against lashing ourselvesto the mast of a fixed policy rule; having seen anynumber of proposed rules come and go, he is skepticalthat any rule is likely to perform well in practice. A littlegood sense goes much further. He sees no necessitythat the Fed pursue a single goal, argtting that "Godgave us two eyes and we ought to use them both."Rather, he suggests that to run the Fed, you need tofocus on both the price level and the real output profile.He argues against reading too much from movementsin the bond markets; they are, after all, only a reflectionof our own actions. To do so would be to behave like amonkey who discovers his reflection in the mirror and"thinks that by looking at the reactions of that mon-key-including its surprises~he is getting new infor-mation." Finally, he counsels against trying to isolatethe central bank too much from the democratic process.This strategy cannot work in the long run; if the peopleare sufficiently displeased with the actions of the centralbank, any legislation that shields the Fed will be over-turned.


James Tobin agrees with many other participantsthat monetary policy did "pretty well" in the Volckerera. However, he observes that the economy has spentconsiderably more years producing below its potentialthan above it. Part of the explanation for this phenom-enon, Tobin asserts, is that the public believes that arecovery is defined as a period of nonnegative growthin GDP, instead of growth at or above the rate ofpotential. Tobin suggests that the result of this miscon-ception is that "pressure for expansionary policy van-ishes once the quarterly real growth report is positive."Tobin advises further that, because the link betweenthe federal funds rate and the real economy is some-what tenuous, the Fed should consider conductingopen market operations in longer maturities that are"closer to the points of meaningful contact betweenthe financial and real economies."

Tobin expresses dismay at the widely supportedproposition that central banks ought to ignore realgrowth and employment and focus exclusively onprice stability. Monetary policy must worry about realoutcomes, Tobin argues, because it is unlLkely thatfiscal policy will be flexible enough to effectively man-age them. Finally, Tobin cautions against using zeroinflation as the default target, citing several argu-ments~the downward rigidity of nominal wages, thepolicy constraint of the zero floor of nominal interestrates, and upward biases in standard measures ofinflation in favor of a positive target rate of inflation.

Robert Barro urged the central bank to focusexclusively on control of nominal variables such asthe price level, monetary aggregates, and nominalGDP, rather than real variables such as employmentand real GDP. Nominal variables are the properdomain of monetary policy, he asserts, because mon-etary policy has "uncertain, and usually short-livedand minor, influences over . . . real variables." Butfor a price stabilization program to be successful, itmust be attended by a credible commitment to thegoal. Otherwise, the temptation will always be toaccept ex post the real-side advantages that attendunexpected and unfavorable price shocks, thus devi-ating from the path of price stability. A commitmentwill likely be viewed as more credible the morebinding are its legislative underpinnings; therefore,"Barro cites the growing support of legislated, inde-pendent central banks as a reasonable means ofcommitting to a rule.

Lyle Gramley also emphasized the successe$.0fmonetary policy in the 1980s, suggesting that theywere attributable to the sharper focus on price stabil-ity as the goal of monetary policy, and to more

Panel Com~nents

Samuelson counsels againsttnying to isolate the central

bank too much from thedemocratic process.

Tobin observes that the economyhas spent considerably moreyears producing below itspotential than above it.

For a price stabilization programto be successful, Barro noted, itmust be attended by a credible

commitment to the goal.

Gramley strongly advocates the useof an interest rate instrument to conduct

monetary policy, and favors legislateddefinition of the Fed’s goals.

McCallum suggests that the Feduse policy rules, not as externalconstraints, but as benchmarksin the decision-making process.

forward-looking monetary policy. In addition, Gram-ley strongly advocates the use of an interest rateinstrument to conduct monetary policy. This woulddecrease the cost to businesses of highly variableinterest rates and improve overall performance rela-tive to a monetary aggregates strategy. Finally, heargues for legislated definition of the Fed’s goals, assuggested by Debelle and Fischer.

Bennett McCallum suggests that the Fed usepolicy rules, not as external constraints imposed onpolicymakers’ behavior, but as benchmarks for use inthe decision-making process. McCallum favors a rulein which the monetary base is adjusted so as to attaina nominal GDP target. He suggests a GDP targetbecause keeping GDP growth close to target would


ensure a low average rate of inflation; the samecannot be said of achieving a target growth rate for amonetary aggregate. Using the base as the policyinstrument is desirable, McCallum argues, primarilybecause it requires a very simple policy rule: Increasebase growth when nominal GDP is below target, anddecrease it when nominal GDP is above target. Bycontrast, an interest rate instrument requires a morecomplex rule, in part because what constitutes a restric-tive interest rate depends on the rate of inflation andthe state of the rest of the economy. For example,McCallum cites the confusing rule he tells his students:"If the Fed wants interest rates to be lower [throughlower inflation], then it must raise the interest rate."McCallum has found that, in model simulations, hismonetary base rule performs quite well.

ConclusionAt the first Federal Reserve Bank of Boston Con-

ference in 1969, Paul Samuelson opened his commentswith the declaration: "The central issue that is debatedthese days in connection with macro-economics is thedoctrine of monetarism.., the belief that the primarydeterminant of the state of macro-economic aggregatedemand . . . is money." Twenty-five years later, thestatus of money in the thirty-eighth conference is farfrom central; indeed, William Poole’s paper striveshard to find any role for the monetary aggregates inthe conduct of monetary policy.

In his opening remarks for the 1978 FederalReserve Bank Conference, Federal Reserve Bank ofBoston President Frank Morris expressed dismay that"it will be a long time before we again have thecomplete confidence which we had in the early1960s--that we knew exactly what we were doing."

Judging by the comments of many of the 1994 con-ference’s participants, we should have regained inthe 1990s some of the confidence that we lost in the1970s: "the Fed has performed well indeed in recentyears" (William Poole); "the results of monetarypolicy in the 1980s were remarkably good" (LyleGramley). At the time of the conference, it appearedthat inflation was under control, real growth waspositive and sustainable, and the Fed had found apolicy strategy that could keep it that way.

Nevertheless, participants expressed concernabout whether the current success could be main-tained in a dynamic, changing economy.~ As thisconference pointed out, we are still quite ignorantabout much of the way the economy works. Econo-mists do not agree on the degree of emphasis mone-tary policy should place on prices versus output; theydo not agree on the size of the output loss associatedwith further decreases in the inflation rate, or how tominimize that loss; and they do not agree on themechanism by which monetary policy affects outputand inflation. If monetary policy had to respond to asizable supply shock, for example, these areas ofignorance would become more obvious weaknesses.

As with the 1978 conference, we did not expectthis conference to produce the new synthesis thatwould dispel our ignorance. But we hoped that itwould, as Frank Morris hoped, "generate a buildingblock or two upon which a new synthesis will bebased." The building blocks that emerged from thisconference include a beginning understanding of theinflation/output variability trade-off that monetarypolicymakers face, a better understanding of theconsequences of using a short-term interest rate asthe instrument of monetary policy, and preliminaryinternational evidence on the costs and benefits ofcentral bank independence.

ReferencesBall, Laurence. 1994. "What Determines the Sacrifice Ratio?" In

N. Gregory Mankiw, ed., Monetary Policy. Chicago: University ofChicago Press, forthcoming.

Barro, Robert, and David Gordon. 1983. "A Positive Theory ofMonetary Policy in a Natural Rate Model." Journal of PoliticalEconomy, vol. 91, pp. 589~10.

Caballero, Ricardo, and Mohamad L. Hammour. 1991. "TheCleansing Effect of Recessions." National Bureau of EconomicResearch Working Paper No. 3922.

Davis, Steven, and John Haltiwanger. 1990. "Gross Job Creationand Destruction: Microeconomic Evidence and MacroeconomicImplications." In Blanchard, Olivier, and Stanley Fischer, ed.,NBER Macroeconomics Annual, vol. 5, pp. 12~68.

Fischer, Stanley. 1993. "The Role of Macroeconornic Factors in

Growth." Journal of Monetary Economics, vol. 32, pp. 485-512.Fuhrer, Jeffrey C., and George R. Moore. 1994. "Monetary Policy

Tradeoffs and the Nominal Interest Rate-Real Out-put Correla-tion." Forthcoming, The American Economic Review.

Motley, Brian. 1994. "Growth and Inflation: A Cross-CountryStudy." Prepared for the Center for Economic Policy Research-Federal Reserve Bank of San Francisco conference, March 1994.

Rogoff, Kenneth. 1985. "The Optimal Degree of Commitment to aMonetary Target." Quarterly Journal of Economics, vol. 100, pp.1169-90.

Schumpeter, Joseph A. 1939. Business Cycles: A Theoretical, Histori-cal, and Statistical Analysis of the Capitalist Process. New York:McGraw-Hill.


Goals, Guidelines, and Constraints Facing Monetamj Policymakers

At the Federal Reserve Bank of Boston’s most recent economic conference on June 19, 20, and 21, 1994,bankers, economists, and other financial specialists met to consider three broad questions about theconduct of monetary policy. First, how efficiently has U.S. policy balanced the goals of price stability andfull employment? Second, have rapidly changing financial markets made the use of intermediate targets,such as monetary aggregates, obsolete? Third, what can domestic policymakers learn from the tactics andstrategies employed by foreign central banks? The conference agenda is outlined below.

How Efficient Has Monetary Policy Been?John B. Taylor, Stanford UniversityDiscussant: Laurence M. Ball, The Johns Hopkins UniversityJeffrey C. Fuhrer, Federal Reserve Bank of BostonDiscussant: N. Gregory Mankiw, Harvard UniversitySummary Discussant: Martin S. Eichenbaum, Northwestern University

Comparing Direct and Intermediate TargetingWilliam Poole, Brown UniversityDiscussants: Benjamin M. Friedman, Harvard University

Donald L. Kohn, Board of Governors of the Federal Reserve System

Lessons from International ExperienceCharles A.E. Goodhart, London School of EconomicsJos~ Vifials, Bank of SpainDiscussant: Richard N. Cooper, Harvard UniversityStanley Fischer, Massachusetts Institute of TechnologyGuy Debelle, Massachusetts Institute of TechnologyDiscussant: Robert E. Hall, Stanford University

How Can Monetary Policy Be Improved? A Panel DiscussionRobert J. Barro, Harvard UniversityLyle E. Gramley, Mortgage Bankers Association of AmericaBennett T. McCallum, Carnegie Mellon UniversityPaul A. Samuelson, Massachusetts Institute of TechnologyJames Tobin, Yale University

The proceedings, Conference Series No. 38, will be published late this year. Information about orderingwill be included in a later issue of this Review~


New EnglandFiscal Facts

This newsletter is designed to brief readers on fiscal developmentsin the New England states. Published three times a year, Fiscal Factspresents short analyses of fiscal issues especially relevant to NewEngland. Tables and text also provide the most recent information aboutstate budgets and spending. The Fall 1994 issue features an articleon the rising cost of operating state prisons. There is no charge forthis publication. For a copy of Fiscal Facts and for subscriptions, phone(617) 973-4252 or write to the Research Department, Att: Fiscal Facts,Federal Reserve Bank of Boston, P.O. Box 2076, Boston, MA 02106-2076.

Looking forward, central bankers

Documents

Transcript of Looking forward, central bankers