EUROPEAN ELSEVIER European Economic Review 46(2002)771-780 Inflation Targets an instrument rule or a targeting rule?r p Inflation targeting: Should it be modeled E O. Sy Department of Economics, Princeton Unicersity. Fisher Hall, Princeton, NJ08544-1021, US.A bCPER, London, UK eNBER, Cambridge, MA, USA The paper discusses how current infation targeting should be modeled, and argues that it is better represented as a commitment to a targeting rule (a rule specifying operational objectives for monetary policy or a condition for the target variables ), than as a commitment to a simple instrument rule(like a Taylor rule). 2002 Elsevier Science B V. All rights reserved. El classification: E42; E52: E58 Keywords: Inflation targeting: Monetary policy rules 1. Introduction Inflation targeting was introduced in New Zealand in 1990, so by now we have about 10 years of experience. It spread very quickly to an increasing number of countries Canada 1991, the UK 1992, Sweden, Finland and Australia 1993. The Czech Republic was the first transition economy to introduce inflation targeting, and Brazil was the first developing country to introduce full-fledged inflation targeting. Israel and Chile have gradually developed into inflation targeters. Recent additions in Europe include Switzerland, Norway and Iceland. f This paper is a shortened version of a presentation at the EEA 2001 Annual Congress, Lausanne, August Corresponding author. Department of Economics, Princeton University, Princeton, NJ 08544-1021, USA Tel:+1-609-258-0329 E-Jmail address: svensson(@princeton.edu(LEO. Svensson) Url.http:/ww However, stating reasons that I find contrived and unconvincing, the Swiss National Bank insists that it is not doing inflation targeting, in spite of it fulfilling the criteria specified below( Baltensperger et al. 2001). 0014-2921/02/S-see front matter 2002 Elsevier Science B V. All rights reserved. P:s0014-2921(01)00212-4
European Economic Review 46 (2002) 771 – 780 www.elsevier.com/locate/econbase In ation Targeting In ation targeting: Should it be modeled as an instrument rule or a targeting rule? Lars E.O. Svenssona;b; c; ∗ aDepartment of Economics, Princeton University, Fisher Hall, Princeton, NJ 08544-1021, USA bCPER, London, UK cNBER, Cambridge, MA, USA Abstract The paper discusses how current in ation targeting should be modeled, and argues that it is better represented as a commitment to a targeting rule (a rule specifying operational objectives for monetary policy or a condition for the target variables), than as a commitment to a simple instrument rule (like a Taylor rule). c 2002 Elsevier Science B.V. All rights reserved. JEL classi,cation: E42; E52; E58 Keywords: In ation targeting; Monetary policy rules 1. Introduction In ation targeting was introduced in New Zealand in 1990, so by now we have about 10 years of experience. It spread very quickly to an increasing number of countries: Canada 1991, the UK 1992, Sweden, Finland and Australia 1993. The Czech Republic was the @rst transition economy to introduce in ation targeting, and Brazil was the @rst developing country to introduce full- edged in ation targeting. Israel and Chile have gradually developed into in ation targeters. Recent additions in Europe include Switzerland, Norway and Iceland. 1 This paper is a shortened version of a presentation at the EEA 2001 Annual Congress, Lausanne, August 2001. ∗ Corresponding author. Department of Economics, Princeton University, Princeton, NJ 08544-1021, USA. Tel.: +1-609-258-0329. E-mail address: svensson@princeton.edu (L.E.O. Svensson). URL: http://www.princeton.edu/∼svensson/ 1 However, stating reasons that I @nd contrived and unconvincing, the Swiss National Bank insists that it is not doing in ation targeting, in spite of it ful@lling the criteria speci@ed below (Baltensperger et al. 2001). 0014-2921/02/$ - see front matter c 2002 Elsevier Science B.V. All rights reserved. PII: S0014-2921(01)00212-4
LEO. Svensson/ European Economic Review 46(2002)771-780 Except for research on the beneficial consequences of central-bank independence and low inflation, and academic advice by Charles Goodhart on the incentive structure for the Governor of the Reserve Bank of New Zealand, inflation targeting was introduced in early 1990s without any preceding specific academic research on inflation targeting The credit for the initial rise of inflation targeting goes mostly to insightful central-bank and finance-department officials rather than academics. Once inflation targeting was introduced in the mid 1990s, though, an increasing number of academics started to do research on the topic, and by now there is large volume of accumulated research, and the number of papers and books on inflation targeting is growing fast How should we define inflation targeting? I believe it is useful to emphasize three characteristics:(1) There is a numerical inflation target, in the form of either a point target (with or without a tolerance interval) or a target range. This numerical infla- tion target refers to a specific price index. Achieving the inflation target mary objective of monetary policy, although there is room for additional secondary objectives, as we shall see. There is no other nominal anchor, like an exchange-rate target or a money-growth target.(2)The decision-making process can be described as nfiation-forecast targeting,, in the sense that the central banks inflation forecast has a prominent role and the instrument is set such that the inflation forecast conditional in the instrument setting is consistent with the target. This does not exclude that output and output-gap forecasts also enter in an essential way, as we shall see. (3) There is and accountability. The central bank is accountable for achieving the inflation target and provides transparent and explicit monetary-policy reports presenting its forecasts and explaining and motivating its policy However, these days many countries call themselves inflation targeters without seem- g to take the inflation targeting very seriously. Indeed, an institutional commitment to inflation targeting appears essential for inflation targeting to have much meaning Such an institutional commitment involves(1)a clear(preferably legislated) mandate for a monetary policy directed towards low inflation,(2) central-bank independence ("instrument independence", meaning independence in setting the monetary-policy in- strument, in some cases also independence in formulating an operational interpretation of the low-infiation mandate ), and (3)accountability of the central bank for achieving the mandate By now a large amount of research has been done on inflation targeting and re lated research topics. There has been work on institutions and political economics, for instance on governance, independence, delegation, accountability, decision-making in committees, and the role of transparency for incentives. There has been work on many aspects of the transmission mechanism, including aggregate supply/Phillips curves, ag gregate demand /IS curves, the credit channel and the financial accelerator, the impact of asset prices, the term structure of interest rates, the role of money, and the fiscal theory of the price level. There has been work on monetary-policy objectives, including the relation between inflation and economic growth, targeting inflation versus alterna- tives(the price level, nominal GDP, money growth, exchange rates, etc. ) flexible ver- sus strict inflation targeting(to what extent inflation targeting includes concerns about the real economy, the output gap, interest rates, and the exchange rate ), the relation
772 L.E.O. Svensson / European Economic Review 46 (2002) 771 – 780 Except for research on the bene@cial consequences of central-bank independence and low in ation, and academic advice by Charles Goodhart on the incentive structure for the Governor of the Reserve Bank of New Zealand, in ation targeting was introduced in early 1990s without any preceding speci@c academic research on in ation targeting. The credit for the initial rise of in ation targeting goes mostly to insightful central-bank and @nance-department oJcials rather than academics. Once in ation targeting was introduced in the mid 1990s, though, an increasing number of academics started to do research on the topic, and by now there is large volume of accumulated research, and the number of papers and books on in ation targeting is growing fast. How should we de@ne in ation targeting? I believe it is useful to emphasize three characteristics: (1) There is a numerical in ation target, in the form of either a point target (with or without a tolerance interval) or a target range. This numerical in ation target refers to a speci@c price index. Achieving the in ation target is the primary objective of monetary policy, although there is room for additional secondary objectives, as we shall see. There is no other nominal anchor, like an exchange-rate target or a money-growth target. (2) The decision-making process can be described as “in ation-forecast targeting”, in the sense that the central bank’s in ation forecast has a prominent role and the instrument is set such that the in ation forecast conditional in the instrument setting is consistent with the target. This does not exclude that output and output-gap forecasts also enter in an essential way, as we shall see. (3) There is a high degree (an exceptionally high degree, by historical standards) of transparency and accountability. The central bank is accountable for achieving the in ation target and provides transparent and explicit monetary-policy reports presenting its forecasts and explaining and motivating its policy. However, these days many countries call themselves in ation targeters without seeming to take the in ation targeting very seriously. Indeed, an institutional commitment to in ation targeting appears essential for in ation targeting to have much meaning. Such an institutional commitment involves (1) a clear (preferably legislated) mandate for a monetary policy directed towards low in ation, (2) central-bank independence (“instrument independence”, meaning independence in setting the monetary-policy instrument, in some cases also independence in formulating an operational interpretation of the low-in ation mandate), and (3) accountability of the central bank for achieving the mandate. By now a large amount of research has been done on in ation targeting and related research topics. There has been work on institutions and political economics, for instance on governance, independence, delegation, accountability, decision-making in committees, and the role of transparency for incentives. There has been work on many aspects of the transmission mechanism, including aggregate supply=Phillips curves, aggregate demand=IS curves, the credit channel and the @nancial accelerator, the impact of asset prices, the term structure of interest rates, the role of money, and the @scal theory of the price level. There has been work on monetary-policy objectives, including the relation between in ation and economic growth, targeting in ation versus alternatives (the price level, nominal GDP, money growth, exchange rates, etc.), exible versus strict in ation targeting (to what extent in ation targeting includes concerns about the real economy, the output gap, interest rates, and the exchange rate), the relation
LEO. Svensson/ European Economic Reuiew 46(2002)771-780 between social welfare(utility of the representative consumer)and the frequently used quadratic loss functions representing inflation targeting, and the role of interest-rate stabilization/smoothing. a good part of the research has been on monetary-policy rules, including whether commitment or discretion is the realistic assumption, prop- erties of simple instrument rules, especially Taylor-type rules, and the relation between instrument rules and targeting rules (to be defined). There has been work on mone- tary policy under uncertainty, for instance, on partial information about state of the world, certainty-equivalence, optimal estimation and control; on model uncertainty, Brainard-type multiplicative uncertainty, robust control versus Bayesian control; and on whether uncertainty implies more cautious or more aggressive policy. There has been work on the role of transparency and credibility for successful inflation targeting A research area of enormous practical importance, given the deteriorating economic situation in Japan, concerns avoiding and escaping from liquidity traps and deflation, and handling the zero bound on nominal interest rates. Much recent work has focused explicitly on inflation targeting in open economies, including the choice of the optimal target index the role of the exchange-rate. The pros and cons of inflation targeting in developing economies and transition economies is an increasingly important research area(Amato and Gerlach, 2002) There is no way I can cover this massive amount of work in a short paper. In- tead I will focus on one controversial issue, where I have done some research of my own(especially in Svensson, 2001b), namely how to model inflation targeting: more specifically, what the policy rule is, and whether this policy rule is best seen as instrument rule or a targeting rule. 2 2. How to model inflation targeting? There is by now widespread agreement among central bankers and academics that inflation targeting in practice is"flexible""inflation targeting(see, for instance, several contributions in Federal Reserve Bank of Kansas City (1996, 1999)): The objective is to stabilize inflation around the inflation target, but also to put some weight on stabilizing th potential output (the level of output that would result with flexible prices ) Such an objective can be described by a quadratic intertemporal loss function in period t, =(1-E∑δx+-x2)2+x2+1 where 8(0<8<1)is a discount factor, E, denotes expectations conditional on infor- mation available in period t, I and x denote inflation and the output gap in period t, respectively, T' is the inflation target, and i>0 is the relative weight on output-gap stabilization. Thus, inflation and the output gap are the"target variables", that is, the 2 My presentation at the EEA 2001 Annual Congress also covered the issues of commitment versus discretion in inflation targeting, the relation between inflation targeting and welfare-optimizing policy, and the zero bound for interest rates and liquidity traps. Because of space constraints, these issues are not discussed her
L.E.O. Svensson / European Economic Review 46 (2002) 771 – 780 773 between social welfare (utility of the representative consumer) and the frequently used quadratic loss functions representing in ation targeting, and the role of interest-rate stabilization=smoothing. A good part of the research has been on monetary-policy rules, including whether commitment or discretion is the realistic assumption, properties of simple instrument rules, especially Taylor-type rules, and the relation between instrument rules and targeting rules (to be de@ned). There has been work on monetary policy under uncertainty, for instance, on partial information about state of the world, certainty-equivalence, optimal estimation and control; on model uncertainty, Brainard-type multiplicative uncertainty, robust control versus Bayesian control; and on whether uncertainty implies more cautious or more aggressive policy. There has been work on the role of transparency and credibility for successful in ation targeting. A research area of enormous practical importance, given the deteriorating economic situation in Japan, concerns avoiding and escaping from liquidity traps and de ation, and handling the zero bound on nominal interest rates. Much recent work has focused explicitly on in ation targeting in open economies, including the choice of the optimal target index the role of the exchange-rate. The pros and cons of in ation targeting in developing economies and transition economies is an increasingly important research area (Amato and Gerlach, 2002). There is no way I can cover this massive amount of work in a short paper. Instead I will focus on one controversial issue, where I have done some research of my own (especially in Svensson, 2001b), namely how to model in ation targeting: more speci@cally, what the policy rule is, and whether this policy rule is best seen as an instrument rule or a targeting rule. 2 2. How to model ination targeting? There is by now widespread agreement among central bankers and academics that in ation targeting in practice is “ exible” in ation targeting (see, for instance, several contributions in Federal Reserve Bank of Kansas City (1996, 1999)): The objective is to stabilize in ation around the in ation target, but also to put some weight on stabilizing the output gap, the diNerence between actual output and the “natural” output level, potential output (the level of output that would result with exible prices). Such an objective can be described by a quadratic intertemporal loss function in period t, Lt = (1 − )Et ∞ =0 [(t+ − ∗) 2 + x2 t+]; (2.1) where (0 ¡¡1) is a discount factor, Et denotes expectations conditional on information available in period t; t and xt denote in ation and the output gap in period t, respectively, ∗ is the in ation target, and ¿0 is the relative weight on output-gap stabilization. Thus, in ation and the output gap are the “target variables”, that is, the 2 My presentation at the EEA 2001 Annual Congress also covered the issues of commitment versus discretion in in ation targeting, the relation between in ation targeting and welfare-optimizing policy, and the zero bound for interest rates and liquidity traps. Because of space constraints, these issues are not discussed here
LEO. Svensson/ European Economic Review 46(2002)771-780 variables that enter the loss function. The corresponding"target levels"are I'and zero. The zero target level for the output gap corresponds to an output target equal to potential output. There is general agreement that inflation-targeting central banks do normally not have overambitious output targets, that is, exceeding potential output. Thus, discretionary optimization does not result in average inflation bias, counter to the case in the standard Kydland-Prescott-Barro-Gordon setup Since the inflation target is subject to choice but not the output target, there is an asymmetry between the inflation and output target, consistent with the inflation target being the "primary objective". 3 Regarding the two parameters, 8 and 4, the discount factor is for all practical pur- poses likely to be very close to one, especially whe model is use estingly, when the discount factor approaches one, the limit of the intertemporal loss function is the weighted sum of the unconditional variances of inflation and the output lim t= varn]+4. Var[x,] (when the unconditional mean of infation and the output gap equal the inflation target and zero, respectively, E[]= I' and ELx]=0).As mentioned, flexible inflation targeting corresponds to i>0.Strict"inflation targeting would be the unrealistic case of i=0 2.1. Commitment to a simple instrument rule How do and should inflation-targeting central banks achieve the inflation target and minimize the loss function? most of the literature discusses this in terms of a com- mitment to a simple reaction function, a simple"instrument rule", where the central bank mechanically sets its instrument rate (usually a short interest rate like a one- or two-week repurchase rate), i,, as a given simple function of a small subset of the nformation available to the central bank. Although several different simple instrument rules have been discussed since the 1970s the best known and most discussed is the Taylor(1993)rule, a frequent variant of which can be written i=(1-f1)[F+元1+fn(π1-π*)+fxx]+f-1, where the response coefficients fx, fx and fi fulfill fx>0,fr>0 and0<fi<1 (although cases with fi l have also been discussed) and r is the average real interest rate. A large volume of research, for instance in Taylor (1999), has examined the papers have also estimated empirical reaction functions of this type terminacy properties of (2.3)and its variants in different models, with respect to determinacy of equilibria, performance measured by (2.2 ), robustness to different models, etc. Several The advantages of a simple instrument rule like(2.3)are:(1) The rule can easily be verified by outside observers and a commitment to the rule would therefore be techni- cally feasible.(2)Variants of the Taylor rule have been found to be relatively robust 3 An interesting and important research area, discussed in my presentation at the EEA 2001 Annual Congress, concerns to what extent inflation targeting as represented by (2.1)corresponds to maximizing the welfare of the representative consumer. Another research area is to what extent a quadratic loss function is sufficient, or if higher-order terms corresponding to asymmetric preferences are needed
774 L.E.O. Svensson / European Economic Review 46 (2002) 771 – 780 variables that enter the loss function. The corresponding “target levels” are ∗ and zero. The zero target level for the output gap corresponds to an output target equal to potential output. There is general agreement that in ation-targeting central banks do normally not have overambitious output targets, that is, exceeding potential output. Thus, discretionary optimization does not result in average in ation bias, counter to the case in the standard Kydland–Prescott–Barro–Gordon setup. Since the in ation target is subject to choice but not the output target, there is an asymmetry between the in ation and output target, consistent with the in ation target being the “primary objective”. 3 Regarding the two parameters, and , the discount factor is for all practical purposes likely to be very close to one, especially when a quarterly model is used. Interestingly, when the discount factor approaches one, the limit of the intertemporal loss function is the weighted sum of the unconditional variances of in ation and the output gap: lim →1 Lt = Var[t] + Var[xt] (2.2) (when the unconditional mean of in ation and the output gap equal the in ation target and zero, respectively; E[t] = ∗ and E[xt] = 0). As mentioned, exible in ation targeting corresponds to ¿0. “Strict” in ation targeting would be the unrealistic case of = 0. 2.1. Commitment to a simple instrument rule How do and should in ation-targeting central banks achieve the in ation target and minimize the loss function? Most of the literature discusses this in terms of a commitment to a simple reaction function, a simple “instrument rule”, where the central bank mechanically sets its instrument rate (usually a short interest rate like a oneor two-week repurchase rate), it, as a given simple function of a small subset of the information available to the central bank. Although several diNerent simple instrument rules have been discussed since the 1970s, the best known and most discussed is the Taylor (1993) rule, a frequent variant of which can be written it = (1 − fi)[ Rr + t + f(t − ∗) + fxxt] + fiit−1; (2.3) where the response coeJcients f; fx and fi ful@ll f ¿ 0; fx ¿ 0 and 0 6 fi 6 1 (although cases with fi ¿ 1 have also been discussed) and Rr is the average real interest rate. A large volume of research, for instance in Taylor (1999), has examined the properties of (2.3) and its variants in diNerent models, with respect to determinacy of equilibria, performance measured by (2.2), robustness to diNerent models, etc. Several papers have also estimated empirical reaction functions of this type. The advantages of a simple instrument rule like (2.3) are: (1) The rule can easily be veri@ed by outside observers and a commitment to the rule would therefore be technically feasible. (2) Variants of the Taylor rule have been found to be relatively robust 3 An interesting and important research area, discussed in my presentation at the EEA 2001 Annual Congress, concerns to what extent in ation targeting as represented by (2.1) corresponds to maximizing the welfare of the representative consumer. Another research area is to what extent a quadratic loss function is suJcient, or if higher-order terms corresponding to asymmetric preferences are needed
LEO. Svensson/ European Economic Review 46(2002)771-780 to different models. in the that they perform reasonably well (in terms of (2.2)) in simulations with different models and rarely result in very bad outcomes(although these simulations have disregarded the important role of central-bank "judgment",to be discussed below) The disadvantages are: (1) The rule will not result in an optimal outcome, for sev eral reasons. For instance, it responds only to a small subset of the information about the economy available to the central bank, and it only imperfectly allows for"history dependence".*(2)More specifically, because of the simplicity of the rule, there will be many situations when good judgment calls for deviations from the rule. Indeed, Taylor (1993) to a large extent discusses sensible discretionary departures from the rule. But then the rule is incomplete, since it provides no rule for departures from the rule. (3) In spite of all the research and discussion of the Taylor rule, no central bank has made a commitment to follow it. Instead of mechanical instrument-rate decision inflation-targeting central banks have set up very elaborate decision-making processes, re huge amounts of information are collected and processed, and conditional fore casts of inflation and output or the output-gap are constructed and contemplated, before an instrument-rate decision is made (Brash, 2001; Svensson, 2001a).(4) Empirical estimates of Taylor-type reaction functions show that they at most explain 3 of the empirical instrument-rate changes ( Judd and Rudebusch, 1998). Thus, central banks in practice deviate substantially from such a reaction function. Furt rmore. a empirical fit would still be consistent with the reaction function being the endo- genous reduced-form outcome of a complex central-bank decision process rather than a commitment that is the beginning and the end of the process 2.2. Commitment to a targeting I find a commitment to a simple instrument rule unsuitable both as description of and prescription for inflation targeting. Instead, I believe inflation targeting is better described and prescribed as a commitment to a"targeting rule". I find it practical to distinguish between"general"and"specific"targeting rules. A general targeting ru specifies the objectives for monetary policy in an operational way, that is, specifies operational loss function. A commitment to a general targeting rule is hence a com- mitment to minimize such a loss function. Such a commitment is often taken as giv by researchers modeling optimal policy, for instance in the Kydland-Prescott-Barro- Gordon-Rogoff tradition. However, in practical monetary policy, making the objectives operational and explicit, is a large step compared to previous ad hoc policies. Getting the objectives straight and creating an institutional commitment to those objectives are rucial contributions of inflation target 4 Optimal instru dence” in the ten rules hay in forward-lookins lagged shocks, "history depen- vension a Svensson(1997,1999,200 (1998, 2000), Clarida et al. (1999)and Svensson and Woodford(1999)
L.E.O. Svensson / European Economic Review 46 (2002) 771 – 780 775 to diNerent models, in the sense that they perform reasonably well (in terms of (2.2)) in simulations with diNerent models and rarely result in very bad outcomes (although these simulations have disregarded the important role of central-bank “judgment”, to be discussed below). The disadvantages are: (1) The rule will not result in an optimal outcome, for several reasons. For instance, it responds only to a small subset of the information about the economy available to the central bank, and it only imperfectly allows for “history dependence”. 4 (2) More speci@cally, because of the simplicity of the rule, there will be many situations when good judgment calls for deviations from the rule. Indeed, Taylor (1993) to a large extent discusses sensible discretionary departures from the rule. But then the rule is incomplete, since it provides no rule for departures from the rule. (3) In spite of all the research and discussion of the Taylor rule, no central bank has made a commitment to follow it. Instead of mechanical instrument-rate decisions, in ation-targeting central banks have set up very elaborate decision-making processes, where huge amounts of information are collected and processed, and conditional forecasts of in ation and output or the output-gap are constructed and contemplated, before an instrument-rate decision is made (Brash, 2001; Svensson, 2001a). (4) Empirical estimates of Taylor-type reaction functions show that they at most explain 2 3 of the empirical instrument-rate changes (Judd and Rudebusch, 1998). Thus, central banks in practice deviate substantially from such a reaction function. Furthermore, a good empirical @t would still be consistent with the reaction function being the endogenous reduced-form outcome of a complex central-bank decision process rather than a commitment that is the beginning and the end of the process. 2.2. Commitment to a targeting rule I @nd a commitment to a simple instrument rule unsuitable both as description of and prescription for in ation targeting. Instead, I believe in ation targeting is better described and prescribed as a commitment to a “targeting rule”. 5 I @nd it practical to distinguish between “general” and “speci@c” targeting rules. A general targeting rule speci@es the objectives for monetary policy in an operational way, that is, speci@es an operational loss function. A commitment to a general targeting rule is hence a commitment to minimize such a loss function. Such a commitment is often taken as given by researchers modeling optimal policy, for instance in the Kydland–Prescott–Barro– Gordon–RogoN tradition. However, in practical monetary policy, making the objectives operational and explicit, is a large step compared to previous ad hoc policies. Getting the objectives straight and creating an institutional commitment to those objectives are crucial contributions of in ation targeting. 4 Optimal instrument rules in forward-looking models include responses to lagged shocks, “history dependence” in the terminology of Woodford (1999) and Svensson and Woodford (1999). 5 Target(ing) rules have previously been discussed by Sims (1980), RogoN (1985), Walsh (1998), Svensson (1997, 1999, 2001b), Rudebusch and Svensson (1999), Cecchetti (1998, 2000), Clarida et al. (1999) and Svensson and Woodford (1999)