Máté Barnabás Tóth: Monetary policy rules and a normative approach to the central bank s objective function

Transcription

1 Máé Barnabás Tóh: Moneary policy rules and a normaive approach o he cenral bank s objecive funcion This sudy aemps o explain in an undersandable manner ha he cenral bank s effor o keep inflaion low is no an end in iself, bu ulimaely serves he ineress of social welfare. We aemp o subsaniae his argumen on he basis of economic heory, based on he logic of New Keynesian models, by describing loss funcions ha conain welfare relevan variables and ineres rules ha minimise hem. By using his framework, we poin ou ha aking ino accoun he limis of measurabiliy, learning and poenially non-raional expecaions decision-making rules ha give considerable weigh o a deparure from he inflaion arge and ake ino accoun real economy consideraions generally perform well in erms of welfare and may be considered robus in New Keynesian-ype models wih forward looking agens. Finally, we argue ha hrough he sraegy of inflaion argeing he normaive implicaions of he above framework can be pu ino pracice. INTRODUCTION I is a generally acceped view in economics and in he pracice of moneary policy ha he bes way for a cenral bank o conribue o he long-erm welfare of sociey is o creae a sable, predicable environmen for marke agens by mainaining price sabiliy and anchoring inflaion expecaions. However, his does no mean ha, in addiion o heir price sabiliy objecive, cenral banks do no or should no ake ino accoun real economy consideraions o some exen. In esablishing moneary policy objecives, i is of primary imporance o evaluae such objecives from a welfare perspecive. On he one hand, any moneary policy can be mainained legiimaely in keeping wih he norms of democracy if, by considering he srucure of he economy as a given, i leads o he maximum welfare of sociey for he longes possible period of ime. In order o provide a normaive definiion for he opimal moneary policy as described above, i is indispensable o ensure ha he decisions (regarding ineres raes) made by he cenral bank can be evaluaed in respec of heir impac on welfare. In order o do so wihin he framework of economic models we need o define an objecive or loss funcion ha is suiable for evaluaing he oucome of variables ha are relevan for welfare and for ranking, in keeping wih he above, moneary policies ha weigh diverse consideraions in differen ways. For a long ime, economics was no able o provide a solid welfare crierion which was in line wih he decision-making pracices of cenral banks and which could serve as a basis for classifying and prioriising various ypes of moneary policies. The primary reason for his was ha for a long ime here was no realisic analyical framework wih a microeconomic basis o model he impac of moneary policy on he real economy and he explici welfare coss of inflaion. Consequenly, comparison beween differen moneary policies happened in an ad-hoc manner and/or on he basis of loss funcions wih arbirarily chosen parameers. Naurally, an ad-hoc forma is also capable of describing a cenral bank s sysem of goals, bu a loss funcion which can be analyically deduced from a realisic model and which conains he so-called deep parameers of he model consiues an objecive basis for comparison, which helps formulae normaive implicaions for moneary policy. WHAT DOES ECONOMIC THEORY SAY: A NORMATIVE APPROACH In he pas decade, he so-called New Keynesian or neo- Wicksellian analyical framework has became widespread in boh academic research and he pracices of cenral banks (see Clarida e al., 1999; Woodford, 2003; Gali and Gerler, 2007). Compared o previously widespread models, he mos novel feaure of he New Keynesian framework is ha i deducs conclusions from firs principles, 1 wih a microeconomic basis and has an explicily modelled connecion beween moneary policy and he real economy. Due o he microeconomic basis, he aforemenioned framework has proven suiable for evaluaing he impac of moneary policy on welfare and for approaching moneary policy from a normaive perspecive (see Roemberg and Woodford, 1997). An addiional feaure of New Keynesian 1 Such is, for example, he maximisaion of uiliy by households and of profi by companies. MNB BULLETIN NOVEMBER

2 MAGYAR NEMZETI BANK models is ha heir basic versions can be described in a ransparen hree equaion sysem by using he firs order condiions arising from he opimisaion problems of households and firms and by adding a moneary policy rule. In he following we would like o provide a simplified descripion of he main feaures of he New Keynesian model ha can be reduced o hree equaions saring from he mos simple specificaion o he more complex and a he same ime more realisic ones. I is imporan o noe ha we do no believe ha he resuls generaed by he New Keynesian models can be direcly and uncondiionally ranslaed ino cenral bank pracice: heir mos imporan meri is ha hey provide a logically consisen, normaive framework for hinking abou moneary policy. New Keynesian models differ from similar, general equilibrium models (e.g. real business cycle models) in ha hey presuppose marke imperfecions ha impac nominal variables. In his framework, corporaions have marke power (which means ha hey are price-seers) and compee in an oligopolisic way while here is fricion in heir pricing decisions. 2 This fricion (or rigidiy) may arise, for example, if obaining informaion for a pricing decision or he very process of pricing is cosly for corporaions. The simples New Keynesian model approaches his fricion wih he socalled Calvo pricing (see Calvo, 1983). 3 Due o hese fricions, pricing is no insan and is no synchronised beween firms. Therefore, in any given ime period non-zero inflaion leads o uninended changes in relaive prices (in oher words, causes a real effec) because here will be firms ha are able o esablish an opimal price, while ohers are unable o change heir prices o heir opimal level. Due o his fricion, he acual oupu will flucuae around he socalled naural level (which would occur in he absence of nominal rigidiies). of a change in heir relaive prices, by way of heir oupu. Therefore, companies ha are in he process of making heir pricing decisions will deermine heir curren prices as a funcion of he prices and marginal coss which hey expec in he coming periods, in oher words, he expeced profi margin. The former is refleced in he real marginal cos indicaor which is he inverse of he above-menioned profi margin, ha is, he raio of he nominal marginal cos and he price. The so-called New Keynesian Phillips curve, which is an approximaion of he seady sae supply side of he model, is a resul of he above-described pricing behaviour of profimaximising companies. The New Keynesian Phillips curve is forward-looking in which inflaion a any given ime depends on he oupu gap and he inflaion expeced in he nex period. = β + { π 1} κy π E + ~ Where π is he inflaion a ime, y ~ is he oupu gap which, in his framework, is he difference beween he acual (y ) and he naural level (y n ) of oupu. E is an operaor ha indicaes (raional) expecaions which are based on informaion ha is available a ime, while β and κ are coefficiens which are derived from he srucural or deep parameers of he model. The oupu gap in he case of cerain assumpions concerning preferences, echnology and labour marke srucure (see Clarida e al., 1999) flucuaes in proporion o he deviaion of real marginal cos from is seady sae value. However, as a resul of various nominal (or real) fricions, he proporionaliy beween he real marginal cos and he oupu gap ceases o exis and a radeoff emerges beween he sabilisaion of inflaion and he oupu gap (we will discuss he role of rade-offs below in more deail). (1) In he absence of nominal rigidiies, he seady sae price se by firms would equal he sum of heir nominal marginal cos and a desired price mark-up; however, due o he sickiness of prices he acual and desired price mark-up will differ as a resul of various shocks. Firms ha are unable o change he price of heir produc a a given momen in ime will have o adap o he change in profi margin, which occurs as a resul Anoher imporan, demand-side consequence of assuming no insanly adjusing (or sicky ) prices is ha moneary policy is capable of having a shor-erm impac on he forward-looking real ineres rae by changing he nominal ineres rae, which hus affecs he disribuion of household consumpion over ime periods. If ime preference remains he same, 4 an increase in he real ineres rae compared o is 2 There may be several reasons for his: prices fixed in a long-erm conrac, coss relaed o re-pricing a produc (menu cos), ec. 3 In he Calvo ype pricing, firms can change heir prices a any given ime only as a funcion of an exogenous process and only wih a cerain degree of probabiliy, regardless of wheher hey changed heir prices in he previous periods. As a resul of his, he ime periods during which he prices of various producs remain fixed overlap, ha is, he changes in prices will no be synchronised. The Calvo ype ime-dependen pricing is of an ad-hoc naure; however, if used in sandard models, i yields resuls ha are similar o hose produced by he more realisic bu mahemaically more complex sae-dependen pricing (see e.g. Klenow and Kryvsov, 2005). A he same ime, however, Lombardo and Vesin (2007) poin ou ha alhough ha he wo mos popular ime and sae-dependen pricing ypes (Calvo- and Roemberg pricing) resul in idenical aggregae behaviour of firms (Phillips curve), hey may have differen welfare implicaions if he disorion ha appears as a resul of he oligopolisic compeiion canno be perfecly offse by he governmen. 4 The uiliy funcion of households in a given ime period conains consumpion and leisure. The ime preference measures he amoun of fuure uiliy a household is willing o sacrifice for he sake of curren uiliy. 40 MNB BULLETIN NOVEMBER 2007

3 MONETARY POLICY RULES AND A NORMATIVE APPROACH TO THE CENTRAL BANK S... naural level moivaes households o reallocae heir consumpion from he presen o he fuure and he oher way round when real ineres raes decrease. In keeping wih he above, he demand side of he sandard New Keynesian model presupposing a closed economy and no aking ino accoun invesmens can be described in he form of he IS curve below which is derived from he behaviour of a uiliy maximising household: where i is he nominal ineres rae which can be affeced by he cenral bank, i E {π +1 } is he forward-looking real ineres rae, and r n he naural level of he real ineres rae (which prevails in an equilibrium wihou nominal rigidiies). The wo equaions described above also clearly show he basic ransmission channel of moneary policy: by changing he ineres rae he cenral bank divers he real ineres rae from is naural level and, by way of is impac on aggregae demand, opens he oupu gap which affecs inflaion hrough he New Keynesian Phillips curve. Since boh inflaion and he oupu gap are forward-looking variables, i is clear ha hey are affeced no only by he curren ineres rae bu by is fuure pah as well. As we have indicaed before, in he New Keynesian framework he welfare loss of households resuling from a deviaion from flexible price equilibrium 5 allocaion can be approached wih a loss funcion 6 which is direcly suiable for examining he impac of moneary policy on welfare (see Roemberg and Woodford, 1997; Woodford, 2003). W ~ y ~ 1 E (i r ) = { y + } { π } n 1 E σ + 1 E0 = 0 n U U 2 2 = E 0 + επ β ( UC C ( where U n is he (hypoheical) uiliy of households in he given period if prices were perfecly flexible and U designaes uiliy under he acual imperfecions. U c is he marginal uiliy of consumpion, C is he level of consumpion in a sable equilibrium, and he produc of he wo yields a uiliy measure. The quadraic loss funcion above resuls from a second order approximaion o his welfare loss which, in addiion o is microeconomic 1 λ β (κ y~ ) = 0 (2) (3) foundaions, is also inuiive because i is similar in form o objecive funcions ha are widely used in he lieraure, 7 bu which conain parameers and variables in an ad-hoc way. MONETARY POLICY IN THE NEW KEYNESIAN MODEL The New Keynesian models are closed wih a moneary policy block ha deermines he nominal ineres rae. If we describe moneary policy as a feedback rule ha reacs o endogenous variables, he inflaion parameer mus be greaer han one if we wan sable equilibrium, ha is, he forward ineres mus also increase if inflaion increases. For moneary policy 8 o be opimal, he cenral bank mus deermine he ineres rae in a way ha minimises social loss on he basis of he welfare crierion defined above. i = r +φ π φ ~ π + y n In he simples New Keynesian model, moneary policy has an easy ask: i follows a rule of acion ha always closes he oupu gap and herefore also brings he rae of inflaion o zero 9 by simulaneously minimising social loss; his is also rue he oher way around. I is imporan o underline ha in he framework of he New Keynesian model moneary policy reduces social loss no when i uncondiionally smoohes flucuaions in oupu, bu raher when i aemps o approximae i o is naural level which, in he case of echnology shocks, may be volaile. Therefore, he approach of he New Keynesian model is fundamenally differen from he approach ha defines he oupu gap as a difference beween acual and long-erm rend-oupu. Moneary policy also has an easy ask when only demand shocks (hose ha are relaed o he IS equaion ) can diver he economy from a sae of sable equilibrium. The demand shocks ha are assumed o be exogenous in he above framework change he oupu gap and inflaion in he same direcion, herefore he opimal reacion of moneary policy which minimises welfare loss is rivial. Because arge variables are forward looking, ha is, hey also depend on he values ha hey will ake on in he fuure, a any given ime hey will be deermined no only by curren y (4) 5 In he sandard New Keynesian model he Pareo-opimal or efficien (lacking any kind of disorions) and he naural oupu are he same. Alhough he laer conains disorions (similar o dead weigh losses) ha arise from he presence of monopolisic compeiion; however, based on an implici assumpion, fiscal policy is capable of providing a counerweigh in he form of lump sum (non-disoring) axes. 6 Wih a second order Taylor series approximaion around he flexible price equilibrium allocaion. 7 The so-called loss funcions are widespread in he lieraure and conain he sum of he deviaion of inflaion from is arge value and he square of he oupu gap in respec of a paricular period. The populariy of he square loss funcion can be explained by he fac ha i is a good represenaion of he symmerical naure of moneary policy: decision makers do no wan high inflaion, deflaion, oupu below or in excess of is poenial. On he oher hand, a quadraic arge funcion yields a mahemaically easily manageable sysem coupled wih equaions ha describe he demand and supply in he economy in a linear (or log-linear) form (see, for example, Benigno and Woodford, 2006). 8 I is imporan o noe ha normaive conclusions depend on he assumpion ha expecaions are raional and moneary policy is fully credible. 9 Blanchard and Gali (2005) calls his phenomenon divine coincidence. MNB BULLETIN NOVEMBER

4 MAGYAR NEMZETI BANK moneary policy bu also by he expecaions relaed o i. The so-called discreionary moneary policy does no recognise ha expecaions can be influenced and reopimises in every period. A he same ime, nondiscreionary moneary policy recognises ha if here is a shor-erm rade-off beween inflaion and he oupu gap (see below) welfare loss can be reduced by using he expecaions channel. Moneary policy can influence expecaions if i is credibly commied o an opimal ineres rule in he long erm (indefiniely). In order o be able o benefi from such a commimen, he cenral bank mus be fully credible, ha is, i mus sick o is decisions made in he pas even if hey did no have he opimal resul in respec of a given ime period ( hisory dependence, see Woodford, 2003). Naurally, in real life cenral banks are unable o commi hemselves for an indefinie period of ime; however, he above serves as an imporan lesson in he pracice of moneary policy. Esablishing credibiliy and managing and/or anchoring expecaions resul in welfare gain, and expecaions for he fuure are imporan for effecively achieving he sabilisaion goals of moneary policy. I is imporan o explain he difference beween opimal and simple ineres rules. The opimal rules minimises he loss funcion derived from he uiliy of he represenaive agen; is coefficiens resul from he deep parameers of he model, and i assumes a knowledge of he naural level of oupu and real ineres raes. Because he laer are variables ha are no direcly observable, herefore he opimal rules ha conain hem are no necessarily suiable for direc use in moneary policy pracice. On he oher hand, he so-called simple rules are based on observable variables and are no derived from he opimisaion of a concree model (an example is he classic Taylor-rule 10 ). In an ideal case, simple rules may have he advanage of performing well in several models wih differen specificaions, in oher words, hey resul in a social welfare loss ha is close o one ha can be achieved wih a moneary policy ha is considered opimal in he given models. are ofen accused of leaving all facors, oher han he inflaion process, ou of consideraion, he simplified approach above is no characerisic of hem eiher because he shor-erm rade-off beween oupu gap and inflaion sabilisaion is an empirically well-esablished fac. TRADE-OFF BETWEEN TARGETS The rade-off beween inflaion and oupu gap can be illusraed by making a small change o he sandard models described above. The simples way o inroduce he rade-off is o supplemen he New Keynesian curve above wih an adhoc cos shock. As a resul of he cos shock, he oupu gap and inflaion sar moving in opposie direcions, herefore heir simulaneous sabilisaion is no longer possible. In such cases, he ask of moneary policy is o disribue he effec of he shock in an opimal manner in an aemp o minimise social loss beween oupu gap and inflaion. π β 1 { π } + y u = E + κ + Alhough he inroducion of an ad-hoc cos shock (u ) makes he implicaions of he New Keynesian model regarding moneary policy more realisic, i is no in line wih he principle of building from microeconomic foundaions. However, he rade-off can also be inroduced from a deeper foundaion: he possibiliy of sabilising inflaion and oupu gap simulaneously depends primarily on he fricions implied in he model. One example is when here is a ime-varying difference beween effecive (compleely fricionless) and naural (absen of nominal fricions) level of oupu. A rade-off resuls also when price rigidiies are coupled wih a rigid labour marke causing he real wage of he represenaive household deviaing from is naural level. This is anoher facor ha, by disoring he real marginal coss of companies and he choice of households beween labour or leisure, resuls in a non-opimal allocaion and herefore leads o social loss. (5) In he above cases moneary policy could minimise he social loss funcion wihou confroning a rade-off. Therefore, his feaure of he sandard New Keynesian model implies a lossminimising moneary policy which immediaely reurns inflaion o zero in he case of exogenous shocks. Mainaining zero inflaion eliminaes welfare loss arising from he presence of nominal price rigidiies; in he New Keynesian framework his is he mos ha moneary policy can do. Alhough cenral banks ha follow inflaion argeing Erceg e al. (2000) inroduces nominal wage rigidiies ino he sandard New Keynesian model analogously o Calvo pricing: he nominal wage of households remains fixed o a cerain degree of probabiliy in every ime period, regardless wheher he wage changed in he previous period or no. As a resul of he fricions affecing prices and nominal wages, wage inflaion (changes in nominal wages, π w ) is also included in he social loss funcion; however, he simulaneous sabilisaion of inflaion, wages and he oupu 10 The classic Taylor rule (Taylor, 1993) is an esimaed feedback rule which ies he level of he cenral bank s prime rae o one consan (or he deferred prime rae) o he deviaion from he arge value of inflaion and a variable ha measures ension in he real economy. 42 MNB BULLETIN NOVEMBER 2007

5 MONETARY POLICY RULES AND A NORMATIVE APPROACH TO THE CENTRAL BANK S... gap will no longer be feasible and here will be a rade-off beween hese goals. 11 E 2 W 0 β y y~ = (φ + φp π + φ w π w, ) In his case, an opimal moneary policy reacs o price and wage inflaion as well as he oupu gap by giving more weigh o he sickier nominal variable. In he previous framework Erceg e al. (2000) examined he performance of several simple ineres rules in addiion o ha of he opimal one. The resuls show ha a welfare loss generaed by he opimal rule can also be achieved by a rule ha reacs o price and wage inflaion, or price inflaion and he oupu gap. Besides labour marke rigidiies, he rade-off beween inflaion and oupu gap may also emerge when he New Keynesian model is exended o an open economy, and we assume ha here is fricion concerning he pass-hrough of exchange rae movemens ino prices. Gali and Monacelli (2005), by placing boh foreign and domesic goods in he represenaive consumer baske and assuming ha he passhrough of exchange rae changes ino consumer prices is immediae and complee, concluded ha social loss depends on domesic consumer price inflaion and on he oupu gap. In his approach foreign moneary policy is assumed o be opimal (which means ha i reached he flexible price equilibrium allocaion) and he law of one price holds in he case of impored consumer goods (which means ha he domesic and foreign prices of hese producs, if calculaed in he same currency, will be he same a all imes) Based on he above, his open economy model approach is analogous o he sandard New Keynesian model, ha is, here is no radeoff beween he oupu gap and he sabilisaion of domesic inflaion, herefore he opimal moneary policy can be approximaed wih an ineres rae rule ha reacs o he inflaion of domesic consumer goods in addiion o he naural level of he real ineres rae. Having a moneary policy ha reacs o he enire price index or fixes he nominal exchange rae may resul in greaer welfare loss because, due o he parial or full sabilisaion of he exchange rae, he relaive price adjusmens of foreign and domesic consumer goods will occur, parially or fully, hrough he nominally rigid domesic prices. Monacelli (2005) proved in a similar open economy framework ha by relaxing he implausible assumpion of complee and immediae pass-hrough (see Campa and (6) Goldberg 2005) of exchange rae movemens ino prices here will be a rade-off beween he sabilisaion goals of moneary policy. Due o he incomplee pass-hrough in he shor erm, he law of one price does no hold in he case of foreign consumer goods, and exchange rae changes direcly affec he oupu gap and inflaion. In his case moneary policy could reach he flexible price equilibrium allocaion if i could simulaneously sabilise domesic inflaion and he deviaion from he law of one price, which is infeasible in his case. Wih respec o social welfare loss, a moneary policy wih commimen ha reacs o he whole consumer price index appears o be he leas cosly on he basis of a loss funcion, used by Monacelli (2005), which conains he oupu gap and oal consumer price inflaion. However, i is imporan o noe ha he parial sabilisaion of exchange rae changes is implici in his moneary policy. THE LIMITS OF IMPLEMENTING OPTIMAL MONETARY POLICIES In his sub-secion we describe he problems which are relaed o he basic assumpions of he heoreical framework described above ha modify some of he normaive implicaions semming from he simpler New Keynesian models. One such problem may arise from model-uncerainy and non-raional expecaions of marke paricipans. Orphanides and Williams (2007) assume ha he cenral bank and economic agens, insead of raional expecaions, have imperfec informaion abou economic srucures, especially concerning naural raes (oupu, unemploymen, real ineres rae). Agens form heir expecaions on he basis of a learning process, in he course of which hey use a simple forecas model which is coninuously re-esimaed as new daa come in. These circumsances considerably change loss minimising moneary policy compared o models ha consider known economic srucure and rely on raional expecaions. On he basis of he objecive funcion used by Orphanides and Williams (2007) which is similar o hose described above, bu is no deduced from he uiliy of he represenaive household a loss-minimising moneary policy reacs sronger o inflaion and o a lesser exen and more gradually (by ineres rae smoohing) o real economy variables, which are esimaed wih uncerainy. In connecion wih he above, i is imporan o noe ha he real ime esimaes of he oupu gap may be raher inaccurae due o various mehodological problems and frequen daa revisions and may no even have he appropriae sign (see Orphanides e al., 2000). I has been demonsraed wihin a 11 We have menioned above ha moneary policy is an economic policy ool ha is suiable primarily for eliminaing disorions ha resul from nominal rigidiies. If in he above-menioned case only wage rigidiies prevailed, moneary policy would be able o eliminae hem similarly o he sandard New Keynesian model. However, he simulaneous prevalence of price and wage rigidiies direcly affecs he accommodaion of real wages and herefore resuls in a real economy disorion ha moneary policy is unable o counerbalance. MNB BULLETIN NOVEMBER

6 MAGYAR NEMZETI BANK model framework by Clarida and ohers (2000) ha a moneary policy, which gives considerable weigh o real ime oupu gap esimaes, may lead o srongly subopimal oucomes. The problem of aking real economy consideraions ino accoun explicily is apparen when a cenral bank s loss funcion or policy rule conains deviaions from he esimaed long-erm oupu rend (ha is, from an aheoreic measure generaed by a simple ime series mehod) as a variable ha measures real economy ensions. Alhough a firs sigh a moneary policy ha aemps o smooh real economy flucuaions appears o be jusifiable from a welfare perspecive, oupu should flucuae as an effecive response o echnological shocks ha canno be observed in real ime. If in his case moneary policy aemps o reurn oupu o is longerm rend, i may generae inflaion or deflaion, conrary o is original inenion. On he basis of he above, an explici reacion o he oupu gap esimae ha is available for decision makers in real ime poses a risk which is also refleced in he pracice of cenral banks: ypically, hey ake ino accoun real economy consideraions wihin he olerance band surrounding he inflaion arge, in he ime-horizon of moneary policy reacion, or by argeing a cerain core inflaion indicaor (see Palmquis, 2007). Targeing core inflaion means ha moneary policy hardly, or no a all, akes ino accoun changes in price index componens ha are frequenly affeced by cos/supply shocks (his is he moneary policy ha is carried on, for example, by he US Fed). The olerance band around he inflaion arge also allows for such impac. The New Keynesian models described above also assume an absolue credibiliy of moneary policy; however, his is no always he case in real life. If marke agens do no find he cenral bank s commimen o fighing inflaion credible, aking ino accoun and emphasising real economy consideraions may be perceived as an inflaion bias. In order o avoid his, clear communicaion, ransparency and accounabiliy by he cenral bank are of key imporance. WHAT DOES GOOD MONETARY POLICY LOOK LIKE ON THE BASIS OF THE LOGIC OF NEW KEYNESIAN MODELS? On he basis of he logic of he New Keynesian model framework described above, we can formulae he following normaive implicaions in respec of moneary policy: Moneary policy mus ake ino accoun ha heoreically, all i can do is o eliminae he impac on welfare of disorions ha resul from nominal rigidiies; in oher words, i canno aemp o reach an oupu ha is permanenly higher han he naural level. I has a decision-making rule ha focuses on sabilising hose price and/or wage inflaion indices ha exhibi nominal rigidiies. The weigh(s) given o sabilising individual price and/or wage indexes is in proporion o he exen of heir sickiness. I is especially imporan for he cenral bank s ineres rae o change in he same direcion as he inflaion bu o a greaer exen, ha is, for example, in he case of a demand shock ha increases inflaion, he real ineres rae mus increase as well. A deviaion from he naural level of oupu is included in he loss funcion of he represenaive marke player, and moneary policy also aemps o minimise i. If he economy suffers a shock ha changes inflaion and he oupu gap in opposie direcions, an opimal moneary policy would disribue is effec beween he wo variables. Furhermore, when aking ino accoun real economy consideraions, i is imporan o ensure ha he price sabiliy goals of he cenral bank s moneary policy is credible. I is aware ha expecaions of fuure changes in moneary policy, when marke players are forward looking, are jus as imporan in respec of welfare-relevan variables as curren ineres rae changes. A loss-minimising moneary policy benefis from he welfare gains semming from he managemen of expecaions and creaes a high degree of credibiliy in order o do so. Credibiliy depends primarily on wheher previously declared goals and commimens have been me. If here is considerable uncerainy abou he models ha describe he economy or if assumpions abou being perfecly informed and abou expecaions being raional are no realised, a loss-minimising moneary policy will reac o a considerably greaer degree o inflaion and o a lesser exen o variables which reflec he slack in he real economy bu canno be direcly observed. Furhermore, under he condiions described above a lower welfare loss may resul if moneary policy changes he ineres rae only gradually (smoohing over he ineres rae) in response o new informaion. CONCLUSIONS: IS INFLATION TARGETING A GOOD MONETARY POLICY? We argue ha inflaion argeing provides a sraegic framework wihin which he implicaions formulaed above can be pu ino pracice. Cenral banks ha conduc he bes pracice of inflaion argeing (hereinafer IT cenral 12 Such is he Briish, Canadian, Norwegian, Swedish and New Zealand cenral bank. 44 MNB BULLETIN NOVEMBER 2007

7 MONETARY POLICY RULES AND A NORMATIVE APPROACH TO THE CENTRAL BANK S... banks) 12 ypically do no have a well-defined real economy arge in addiion o heir explici numerical inflaion arge. This, however, does no mean ha hey do no ake ino accoun real economy consideraions. This may no be eviden, bu by aemping o reach heir price sabiliy goals for a horizon of 1-2 years, IT cenral banks implicily declare ha hey are unwilling o ake any magniude of real economy cos in order o sabilise inflaion in he shor erm when shocks ha move inflaion and he oupu gap in opposie direcions occur. However, making decisions 1-2 years ahead requires he cenral bank o prepare forecass and formulae is moneary policy accordingly. This pracice ( inflaion forecas argeing see Svensson 1997) also allows real economy consideraions o be aken ino accoun even hough he cenral bank only reacs if he inflaion forecas deviaes from he arge. This is because he inflaion forecas by definiion includes he impac of all variables ha may have a significan impac on inflaion in he fuure. They ypically include he curren values of variables ha may be considered relevan in respec of welfare on he basis of he models ha are described above (e.g. wage and oupu gap esimaes). As Svensson (2007) poins ou, mos modern cenral banks follow flexible inflaion argeing (which means ha hey also ake ino accoun real economy consideraions). Sric IT (which is only consrucion wih inflaion), on he oher hand, can be considered more of a heoreical consrucion wih he possible excepion of periods when moneary policy has credibiliy problems. IT cenral banks also fi ino he described heoreical framework because hey ypically use a shor-erm ineres rae as an insrumen, and, furhermore, heir behaviour ex pos can be well approximaed wih an esimaed Taylor rule ha gives considerable (greaer han uniy) weigh o deviaion form he inflaion arge. 13 This, however, does no mean ha in pracice moneary policy follows a pre-defined ineres rae rule ha is effecive a any poin in ime. There is a commimen ha is in line wih he heoreical implicaions; however, i is no a commimen o an ineres rule bu o a medium-erm arge crierion ha consiues he inflaion arge (see Svensson and Woodford, 2003). This pracice if i appears o be credible for economic agens creaes inflaion expecaions ha correspond o price sabiliy and as we have menioned above provides adequae flexibiliy in he shor run in adaping o supply or cos shocks. In order o use shor-erm flexibiliy effecively and/or o creae credibiliy ha is necessary for anchoring mediumerm expecaions, IT cenral banks focus on ransparency and he accounabiliy of decision makers. Furhermore, commimen, ransparency and he accounabiliy of decision-makers also ensure ha moneary policy does no sysemaically aemp o reach an oupu ha is higher han he naural level. Consequenly, he curren pracice of inflaion argeing places considerable weigh on reaching price sabiliy while also aking ino accoun real economy consideraions, mosly by avoiding poenial risks arising from explici reacions o variables ha are hard o observe or measure in real ime. Furhermore, inflaion argeing akes advanage of managing expecaions in he form of a commimen o a medium-erm arge crierion. Based on he above, we can claim ha he sraegy of inflaion argeing saisfies he mos imporan normaive implicaions ha can be derived from he heoreical framework inroduced above. REFERENCES BENIGNO, P. AND WOODFORD, M. (2006): Linear-Quadraic Approximaion of Opimal Policy Problems, CEPR Discussion Paper No CALVO, G. (1983): Saggered Prices in a Uiliy-Maximizing Framework, Journal of Moneary Economics, 1983, 12(3). CAMPA, J. M. AND GOLDBERG, L. S. (2005): Exchange Rae Pass-hrough ino Impor Prices, Review of Economics and Saisics, 87(4). CLARIDA, R., GALÍ, J. AND GERTLER, M. (1999): The Science of Moneary Policy: A New Keynesian Perspecive, Journal of Economic Lieraure, 37(4). CLARIDA R., GALÍ, J. AND GERTLER, M. (2000): Moneary Policy Rules and Macroeconomic Sabiliy: Evidence and Some Theory, The Quarerly Journal of Economics, January GALÍ, J. AND MONACELLI, T. (2005): Moneary Policy and Exchange Rae Volailiy in a Small Open Economy, Review of Economic Sudies, 72(3). GALÍ, J. AND GERTLER, M. (2007): Macroeconomic Modeling for Moneary Policy Evaluaion (in publicaion, Journal of Economic Perspecives). HIDI, J. (2006): A magyar moneáris poliikai reakciófüggvény becslése, Közgazdasági Szemle, LIII. évf., December Similar feedback rules in respec of Hungary were esimaed by Hidi (2006). MNB BULLETIN NOVEMBER

8 MAGYAR NEMZETI BANK KLENOW, P. J. AND KRYVTSOV, O. (2005): Sae-Dependen or Time-Dependen Pricing: Does i Maer for Recen U.S. Inflaion?, NBER Working Paper No LOMBARDO G. AND VESTIN, D. (2007): Welfare implicaions of Calvo vs. Roemberg pricing assumpions, Working Paper Series 770, European Cenral Bank. MONACELLI, T. (2005): Moneary Policy in a Low Pass- Through Environmen, Journal of Money Credi and Banking, 37(6). ORPHANIDES, A., PORTER, R. D., REIFSCHNEIDER, D.,TETLOW, R. AND FINAN, F. (2000): Errors in he measuremen of he oupu gap and he design of moneary policy, Journal of Economics and Business, Elsevier, vol. 52(1-2). ORPHANIDES, A AND WILLIAMS, J. C. (2007): Robus Moneary Policy wih Imperfec Knowledge, Working Paper Series 764, European Cenral Bank. PALMQVIST, S. (2007): Flexible inflaion argeing how should cenral banks ake he real economy ino consideraion?, Sveriges Riksbank Economic Review, 2007/2. ROTEMBERG J. AND WOODFORD, M. (1997): An Opimizaion-Based Economeric Framework for he Evaluaion of Moneary Policy, NBER Macroeconomics Annual SVENSSON, L. E. O. (1997): Inflaion forecas argeing: Implemening and monioring inflaion arges, European Economic Review, Elsevier, 41(6). SVENSSON, L. E. O. AND WOODFORD, M. (2005): Implemening Opimal Policy hrough Inflaion-Forecas Targeing. In: Bernanke, B.S. and Woodford, M. (ed.), Inflaion Targeing, Universiy of Chicago Press, SVENSSON, L. E. O. (2007): Opimal Inflaion Targeing: Furher Developmens of Inflaion Targeing. In Mishkin, F. Schmid-Hebbel, K. (eds.), Moneary Policy under Inflaion Targeing, Banco Cenral de Chile, TAYLOR, J. B. (1993): Discreion versus Policy Rules in Pracice, Carnegie-Rocheser Conference Series on Public Policy, Elsevier, vol. 39. WOODFORD, M. (2003): Ineres and Prices Foundaions of a Theory of Moneary Policy, Princeon Universiy Press, MNB BULLETIN NOVEMBER 2007