《货币银行学》课程教学资源（文献资料）Inflation targeting - lessons from four countries

NBERWORKINGPAPERSERIESINFLATIONTARGETING:LESSONSFROMFOURCOUNTRIESFrederic S. MishkinAdam S.PosenWorkingPaper6126NATIONALBUREAUOFECONOMICRESEARCH1050 Massachusetts AvenueCambridge,MA02138August1997WethankBenBernanke,DonaldBrash,KevinClinton,JohnCrow,PeterFisher,CharlesFreedman,Andrew Haldane, Neal Hatch, Otmar Issing,Mervyn King, Thomas Laubach, William McDonough,Michel Peytrignet, Georg Rich, and Erich Spoerndli for their helpful comments. We are grateful toLaura Brookins for research assistance. The views expressed in this study are those of the authorsand not necessarily those of the Federal Reserve Bank of New York, the Federal Reserve System,Columbia University, the National Bureau of Economic Research,or the Institute for InternationalEconomics. This paper is part of NBER's research programs in Economic Fluctuations and Growthand Monetary Economics,Any opinions expressed are those of the authors and not those of theNationalBureauof EconomicResearch. 1997 by Frederic S. Mishkin and Adam S. Posen. All rights reserved. Short sections of text, notto exceed two paragraphs, may be quoted without explicit permission provided that full credit,includingnotice,is given to the source

InflationTargeting:LessonsfromFourCountriesFrederic S. Mishkin and Adam S. PosenNBER Working Paper No.6126August 1997JELNo.E5Economic Fluctuations and Growthand Monetary EconomicsABSTRACTIn recent years, a number of central banks have announced numerical inflation targets as thebasis for their monetary strategies.After outlining the reasons why such strategies might be adoptedin the pursuit of price stability,this study examines the adoption, operational design, and experienceof inflation targeting as a framework for monetary policy in the first three countries to undertakesuch strategies--New Zealand, Canada, and the United Kingdom. It also analyzes the operation ofthe long-standing German monetary targeting regime, which incorporated many of the same featuresas later inflation-targeting regimes. The key challenge for all these monetary frameworks has beenthe appropriate balancing of transparency and flexibility in policymaking. The study finds that allof the targeting countries examined have maintained low rates of inflation and increased thetransparency of monetary policymaking without harming the real economy through policy rigidityin the face of economic developments. A convergence of design choices on the part of targetingcountries with regard to operational questions emerges from this comparative study, suggesting somelinesofbestpracticeforinflation-targetingframeworksFrederic S. MishkinAdam S.PosenFederal Reserve Bank of New YorkFederal Reserve Bank of New York33LibertyStreet33 Liberty StreetNewYork,NY10045New York,NY 10045and NBERaposen@iie.comfrederic.mishkin@ny.frb.org

IntroductionThe key issue facing central banks as we approach the end of the twentieth century is whatstrategy to pursue in the conduct of monetary policy. One choice of monetary strategy thathas become increasingly popular in recent years is inflation targeting, which involvesthe public announcement of medium-term numerical targets for inflation with acommicment by the monetary authorities to achieve these targets. This study examinesthe experience in the first three countries that have adopted such an inflation-targetingschemeNew Zealand, Canada, and the United Kingdom-as well as in Germany,which adopted many elements of inflation targeting even earlier.Through closeexamination of the experience with inflation targeting, both how targeting operates andhow these economies have performed since its adoption, we seek ro obrain a perspective onwhat elements of inflation targeting work as a strategy for the conduct of monetary policy.Before looking in detail at the individual experiences of these countries, we firstdiscuss the racionale for inflation targeting and the design issues that arise in implementingan inflation-targeting strategy. Then, after the case studies of the individual countries, weprovide some preliminary evidence on the effectiveness of inflation targeting in thesecountries and conclude with an assessment of the inflation-targeting experience.1

monetary policy that is more expansionary than expected.As a result,policymakers whohave a stronger interest in output than in inflation performance will try to producemonetary policy that is more expansionary than expected. However, because workers andfirms make decisions about wages and prices on the basis of cheir expectations about policy,they will recognize the policymakers' incentive for expansionary monetary policy and sowill raise their expectations of inflation. As a result, wages and prices will rise.The outcome, in these time-inconsistency models, is that policymakers are actuallyunable to fool workers and firms, so that on average output will not be higher under sucha strategy; unfortunately, however, inflation will be. The time-inconsistency problemsuggests that a central bank actively pursuing output goals may end up with a biasto high inflation with no gains in output. Consequently, even though the centralbank believes itself to be operating in an optimal manner, it ends up with a suboptimaloutcome.McCallum (1995b) points out that the time- inconsistency problem by itself doesnot imply that a central bank will pursue expansionary monetary policy that leads toinflation. Simply by recognizing the problem that forward-looking expectations in thewage- and price-setting process create for a strategy of pursuing unexpectedly expansionarymonetary policy, central banks can decide not to play that game.Nonetheless, chetime-inconsistency literature points out both why there will be pressures on central banksto pursue overly expansionary monetary policy and why central banks whose commitmentto price stability is in doubt can experience higher inflation.A fourth intellectual development challenging the use of an activist monetarypolicy to stimulate output and reduce unemployment unduly is the recognition that pricestability promotes an economic system that functions more efficiently and so raises livingstandards.If price stability does not persist-that is, inflation occursthe society suffersseveral economic costs. While these costs tend to be much larger in economies with highrates of inflation (usually defined to be inflation in excess of 30 percent a year), recent workshows that substantial costs arise even at low rates of inflation.The cost that first received the attention of economists is the so-called shoe leathercost of inflation--the cost of economizing on the use of non-interest-bearing money (seeBailey [1956]).The history of prewar central Europe makes us all too familiar with thedifficulties of requiring vast and ever-rising quantities of cash to conduct daily transactions.Unfortunately, hyperinflations have occurred in emerging market countries within che lastdecade as well. Given conventional estimates of the interest elasticity of money and thereal interest rate when inflation is zero, this cost is quite low for inflation rates less than10 percent, remaining below 0.10 percent of GDP.Only when inflation rises to above100 percent do these costs become appreciable, climbing above 1 percent of GDP (Fischer1981).Another cost of inflation related to the additional need for transactions is theoverinvestment in the financial sector induced by inflation. At the margin, opportunities tomake profits by acting as a middleman on normal transactions, rather than investing in3

productive activities, increase with instability in prices. A number of estimates put therise in the financial sector share of GDP on the order of 1 percentage point for every10 percentagepoints of inflation up to an inflation rate of 100 percent (English1996).The transfer of resources out of productive uses elsewhere in the economy can be as largeas a few percentage points of GDP and can even be seen at relatively low or moderate ratesofinflation.The difficulties caused by inflation can also extend to decisions about fucureexpenditures. Higher inflation increases uncertainty about both relative prices and thefuture price level, which makes it harder to arrive ar the appropriate production decisions.For example, in labor markets, Groshen and Schweitzer (1996) calculate that the loss ofoutput due to inflation of 10 percent (compared with a level of 2 percent) is 2 percent ofGDP.More broadly,the uncertainty about relative prices induced by inflation can distortthe entire pricing mechanism. Under inflationary conditions, the risk premia demanded onsavings and the frequency with which prices are changed increase. Inflation also alters therelative attractiveness of real versus nominal assets for investment and short-term versuslong-termcontracting.5Themost obvious costs of inflation at low to moderate levels seemto come from theinteraction of che tax system with inflation. Because tax systems are rarely indexed forinflation,an increase in inflation substantiallyraises the cost ofcapital, causing investmentto drop below ics optimal level.In addition,higher taxation, which results from inflation,causes a misallocation ofcapital to different sectors,which in turn distorts the labor supplyand leads to inappropriate corporate financing decisions.Fischer (1994) calculates that thesocial costs from the tax-related distortions of inflation amount to 2 to 3 percent of GDP atan inflation rate of 10 percent. In a recent paper, Feldstein (1997) estimates this cost to beeven higher: he calculates the cost of an inflation rate of 2 percent rather than zero to be1 percent ofGDP.The costs of inflation outlined here decrease the level of resources productivelyemployed in an economy, and thereby the base from which the economy can grow.Mounting evidence from econometric studies shows that, at high levels, inflation alsodecreases the rate of growth of economies. While time series studies of individual countriesover long periods and cross-national comparisons of growth rates are not in total agreement,the consensus is that, on average, a l percent rise in inflation can cost an economy o.1 to0.5 percentage points in its rate of growth (Fischer 1993). This result varies greatly withthe level of inflation-the effects are usually thought to be much greater at higher levels.6However, a recent study has presented evidence that the inflation variability usuallyassociated with higher inflation has a significant negative effect on growth even at lowlevels of inflation, in addition to and distinct from the direct effect of inflation icself.?The four lines of argument ouclined here lead the vast majority of central bankersand academic monetary economists to the view that price stability should be the primarylong-term goal formonetarypolicy.&Furthermore,to avoid the tendency to an inflationarybias produced by the time-inconsistency problem (or uncertainty about monetary policy4

goals more generally),monetary policy strategy often relies upon a nominal anchor to serveas a target that ties the central bank's hands so it cannot pursue (or be pressured intopursuing) a strategy of raising output with unexpectedly expansionary monetary policy. Aswe will see, this anchor need not preclude clearly delineated short-term reactions tofinancial or significant output shocks in order to function as a constraint on inflationarypolicy over the long term. A number of potential nominal anchors for monetary strategy canserve as targets.CHOICEOFTARGETSOne nominal anchor used by almost all central banks at one time or another is a targetgrowth path for a monetary aggregate such as the monetary base or M1, M2, or M3. Ifvelocity is either relatively constant or predictable,a growth target of a monetaryaggregate can keep nominal income on a steady growth path that leads to long-term pricestability.In such an environment, choosing a monetary aggregate as a nominal anchor hasseveral advantages. First, some monetary aggregates, the narrower the better, can becontrolled both quickly and easily by the central bank. Second, monetary aggregates can bemeasured quite accurately with short lags (in the case of the United States,for example,measures of the monetary aggregates appear within two weeks), Third, as pointed out inBernanke and Mishkin (1992), because an aggregate is known so quickly, using it as anominal anchor greatly increases the transparency of monetary policy, which can haveimportant benefits. A monetary aggregate sends almost immediate signals to both thepublic and the markets about the stance of monetary policy and the intentions ofpolicymakers, thereby helping to fix inflation expectations. In addition, the transparency ofa monetary aggregate target makes the central bank more accountable to the public to keepinflation low, which can help reduce pressures on the central bank to pursue expansionarymonetary policyAlthough the targeting of monetary aggregates has many important advantages inprinciple, in practice these advantages come about only if the monetary aggregates have ahighly predictable relationship with nominal income. Unfortunately, in many countries,velocity fluctuations have been so large and frequent in the last fifteen years that therelationships between monetary aggregates and goal variables have broken down. Someobservers have gone so far as to argue that attempts to exploit these relationships have beena cause of their breakdown. As a result, the use of monetary aggregate targets as a nominalanchor has become highly problematic, and many countries that adopted monetary targetsin the 1970s abandoned them in the 1980s.Not surprisingly,many policymakers have beenlookingforalternativenominal anchors.Anotherfrequently used nominal anchor entails fixing the value of the domesticcurrency relative to that of a low-inflation country, say Germany or the United States, or,alternatively, putting the value of the domestic currency on a predetermined path vis-a-visthe foreign currency in a variant of this fixed exchange rate regime known as a crawling pegThe exchange rate anchor has the advantage of avoiding the time-inconsistency problem by5

precommitting a country's central bank so that it cannot pursue an overly expansionarymonetary policy that would lead to a devaluation of the exchange rate. In addition, anexchange rate anchor helps reduce expectations that inflation will approach that of the countryto which its currency is pegged. Perhaps most important, an exchange rate anchor is a monetarypolicy strategy that is easily understood by the public.As forcefully argued in Obstfeld and Rogoff (1995), however, a fixed exchange rateregime is not without its costs and limitations. With a fixed exchange rate regime, acountry no longer exercises control over its own monetary policy. Not only is the countryunable to use monetary policy to respond to domestic shocks, but it is also vulnerable toshocks emanating from the country to which its currency is pegged. Furthermore, in thecurrent environment of open, global capital markets, fixed exchange rate regimes aresubject to breakdowns that may entail sharp changes in exchange rates. Such developmentscan be very disruptive to a country's economy, as recent events in Mexico havedemonstrated. Defending the domestic currency when it is under pressure may requiresubstantial increases in interest rates that directly cause a contraction in consumer andinvestment spending, and the contraction in turn may lead to a recession. In addition, aspointed out in Mishkin (1996), a sharp depreciation of the domestic currency can producea full-scale banking and financial crisis that can tip a country's economy into a severedepression.An inflation target (or its variant,a price-level target) clearly provides a nominalanchor for the path of the price level, and, like a fixed exchange rate anchor, has theimportant advantage of being easily understood by the public.The resulting transparencyincreases the potential for promoting low inflation expectations, which helps to produce adesirable inflation ourcome. Also, like a fixed exchange rate or a monetary targetingstrategy, inflation targeting reduces the pressure on the monetary authorities to pursueshort-run output gains that would lead to the time-inconsistency problem. An inflation-targeting strategy also avoids several of the problems arising from monetary targeting orfixed exchange rate strategies.For example, in contrast to a fixed exchange rate system,inflation targeting can preserve a country's independent monetary policy so that themonetary authorities can cope with domestic shocks and help insulate the domesticeconomy from foreign shocks. In addition, inflation targeting can avoid the problempresented by velocity shocks because it eliminates the need to focus on the link between amonetary aggregate and nominal income; instead, all relevant information may be broughtto bear on forecasting inflation and choosing a policy response to achieve a desirableinflationoutcome.Inflation targeting does have some disadvantages. Because of the uncertain effects ofmonetary policy on inflation, monetary authorities cannot easily control inflation.Thus, itis far harder for policymakers to hit an inflation target with precision than it is for them tofix the exchange rate or achieve a monetary aggregate target. Furthermore, because the lagsof the effectof monetary policy on inflation arevery longtypical estimates are in excess oftwo years in industrialized countries--much time must pass before a country can evaluate6

che success of monetary policy in achieving its inflation target. This problem does not arisewith either a fixed exchange rate regime or a monetary aggregate target.Another potential disadvantage of an inflation target is that it may be takenliterally as a rule that precludes any concern with output stabilization. As we will see in thecases later in our study, this has not occurred in practice. An inflation target, if rigidlyinterpreted, might lead to greater output variability, although it could lead to tightercontrol over the inflation rate. For example, a negative supply shock that raises the inflationrate and lowers output would induce a tightening of monetary policy to achieve a rigidlyenforced inflation target, The resule, however, would add insult to injury because outputwould decline even further. By contrast, in the absence of velocity shocks, a monetaryaggregate target is equivalent to a target for nominal income growth, which is the sum ofreal output growth and inflation. Because the negative supply shock reduces real output aswell as raises the price level, its effect on nominal income growth would be less than oninflation, thus requiring less tightening of monetary policy.The potential disadvantage of an inflation-targeting regime that ignores outputstabilization has led some economists to advocate the use of a nominal income growthtarget instead (for example, see McCallum [1995a] and Taylor [1995]). A nominal incomegrowth target shares many characteristics with an inflation target; it also has many of thesame advantages and disadvantages. On the positive side, it avoids the problems of velocityshocks and the time-inconsistency problem and allows a country to maintain anindependent monetary policy. On the negative side, nominal income is not easilycontrollable by the monetary authorities, and much time must pass before assessment ofmonetary policy's success in achieving the nominal income target is possible. Still, anominal growth target is advantageous in that it explicitly includes some weight on a realoutput objective and thus may lead to smaller fluctuations in real output.9Nonetheless, nominal income targets have two very important disadvantagesrelative to inflation targets. First, a nominal GDP target forces the central bank or thegovernment to announce a number for potential GDP growth. Such an announcement ishighly problematic becauseestimates of potential GDPgrowthare farfromprecise andthey change over time. Announcing a specific number for potential GDP growth may thusindicate a certainty that policymakers may not have and may also cause the public tomistakenly believe that this estimate is actually a fixed target. Announcing a potentialGDP growth number is, therefore, likely to create an extra layer of political complicationit opens policymakers to the criticism that they are willing to settle for growth rates thatare too low. Indeed, it may lead to the accusation that the central bank or the targetingregime is antigrowth, when the opposite is truethat is, a low inflation rate is a means topromote a healchy economy that can experience high growth. In addition, if the estimate forpotential GDP growth is too high and it becomes embedded in the public mind as a target,the classic time-inconsistency problemand a positive inflation bias-will arise.7

The second disadvantage of a nominal GDP target relative to an inflation target isthat the concept of nominal GDP is not readily understood by the public, thus making itless transparent than an inflation target.No one speaks of"headline nominal GDP growth'when discussing labor contracts.In addition,because nominal and real GDP can be easilyconfused, a nominal GDP target may lead the public to believe that a central bank istargeting real GDP growth, with the attendant problems mentioned above.8

Part Il.Design Issues in the Implementationof Inflation TargetsPart I has outlined the reasons why several countries have chosen to base their monetarystrategies on the targeting of inflation. It also raises a set of issues about the design of aninflation-targeting regime.Before examining in detail how inflation targeting has worked inthe countries we examine here, we briefly outline the choices policymakers face in designingan inflation-targeting strategy. The fundamental question is how best to balancetransparency with flexibility in operation, given the uncertainties of monetary policy andthe economic environment. The simpler and tighter the constraints on policy, the easierit is for the public to understand and hold policy accountable, but the harder it is for policyto respond to events and maintain credible performance.Choices about target design aretherefore critical in setting this balance appropriately.In the case studies that follow, we will see that the design choices for an inflationtargeting regime fall into four basic categories: definition and measurement of the target,transparency, flexibility, and timing.DEFINITIONANDMEASUREMENTOFTHETARGETBecause inflation targeting by its very nature requires a numerical value for the target,setting such a target requires explicit answers to several questions about how the target isdefined and measured.Wbat does price stability mean in practice? Inflation targeting requires a quantitative statementas to what inflation rate is consistent with the pursuit of price stability in the nextfewyears.Because of innovation and changing tastes,all inflation measures have a net positivebias. For example, measurement error for consumer price index (CPI) inflation in theUnited States has been estimated to be in the range of 0.5 to 2.0 percent at an annual rate(Shapiro and Wilcox 1996; Advisory Commission to Study the Consumer Price Index1996). Anocher factor to be taken into account in setting the target level of inflation is theasymmetricdangersfrom deflation.That is,through financial and otherchannels,the coststo the real economy from undershooting zero inflation outweigh the direct costs to theeconomy from overshooting zero inflation by a similar amount.These potential costs mightwarrant a price stability objective in which the inflation rate, corrected for any measurementerror, might be set slightly above zero.What inflation series should be targeted and wbo sbould measure it? A target series must bedefined and measured. The series needs to be considered accurate, timely, and readilyunderstandable by the public, but it may also need to exclude from its definition individualprice shocks or onetime shifts that do not affect trend inflation, which is what monetarypolicy can influence.9