TAYLOR RULE: PRESENTATION, INTERPRETATION AND ESTIMATION THE CASE OF THE TUNISIAN CENTRAL BANK. MANSOUR Samia *

Transcription

1 Submission for he 3 rd Ialian Congress of Economerics And Empirical Economics (ICEEE 2009) TAYLOR RULE: PRESENTATION, INTERPRETATION AND ESTIMATION THE CASE OF THE TUNISIAN CENTRAL BANK MANSOUR Samia * * PhD Suden, Faculy of Economics Sciences and Managemen, Tunis, and Assisan Professor, Insiue of Higher Managemen Sousse Tunisia. mansoursamia@yahoo.fr, samia.mansour@ihec.rnu.n. 1

2 Taylor Rule: Presenaion, Inerpreaion and Esimaion The case of The Tunisian Cenral Bank By Mansour Samia ABSTRACT In his paper, I ry o verify if he shor erm ineres raes, as announced by he Tunisian Cenral Bank, follow he original form of The Taylor rule. The Taylor rule aims a giving a guideline for cenral bank o follow in he seing of heir moneary policy by ensuring macroeconomic sabiliy. This rule, in spie of is simple srucure, capures, also, he essenial of he moneary auhoriy s behavior. The iniial specificaion of he Taylor rule is o deermine he level of shor erm ineres raes supposed o be compaible wih he objecives of price sabiliy and he eliminaion of he oupu gap (keeping he oupu a is poenial level). The difference beween he Taylor rae (as calculaed by he rule) and he observed ineres rae is used as an indicaor of he appropriaeness of moneary policy wih respec o he wo key goals: deviaions of he observed inflaion rae from a pre-se arge rae and deviaions of he observed real oupu from is poenial level. In he firs par of his paper, I presen a brief survey of he lieraure on he characerisics of he original Taylor rule. I focus also on is advanages and limis. I review he modificaions made o is original form and heir jusificaions. I also presen a survey of is empirical uses.. The objecive of he second par of his paper is o esimae following he Taylor mehodology he Tunisian s shor erm ineres raes using quarerly daa during he period o and o verify if he Cenral Bank of Tunisia akes ino accoun he inflaion argeing sraegy and he sabiliy of he oupu. Keywords: Taylor s rule, inflaion, oupu gap. Any views expressed on his paper are hose of he auhor. John. B. T: «Discreion Versus Policy Rules in Pracice», Carnegie Rocheser Conference Series on public policy, N 39, pp

3 Conens Page 1. Inroducion Taylor Rule: Presenaion and Inerpreaion Limis of The Taylor Rule Survey on The Empirical Uses of The Taylor Rule Esimaing Original Taylor Rule for The Tunisian Cenral Bank Conclusion: References Figures: 1. The margin beween Taylor ineres rae for he Euro area (CPI vs. GDP deflaor) Oupu Gap Trend Variaion of Inflaion Difference Comparison beween he calculaed and announced ineres rae.19 Tables : Table 1: Esimaion resuls

4 1. Inroducion In recen years, boh academic economiss and cenral bankers gave an upsurge of ineres concerning he opic of simple and explici rules for conducing he moneary policy. Many ypes of moneary policy rules have been presened, discussed and become widespread in he economic lieraure 1. Some economiss believe ha economic growh can be enhanced and price sabiliy achieved by implemening a moneary policy rule. In fac, rules are argued o reduce policy misakes, improve he ransparency of policy and end poliical influence on policy making. By pursuing a moneary rule, cenral banks reinforce heir credibiliy, avoid inflaion bias and ensure he efficiency of heir moneary policy. In he pracice, moneary rules have wo ypes: arge s rules and insrumen s rules. We find numerous forms of moneary rules such as he money growh rae rule (Friedman 1968), he nominal GDP growh rule (McCallum 1993, 1995), he price arge rule, he arge exchange rae rule and he dual arge rule ha has boh an inflaion arge and a producion arge (Taylor s rule). The firs moneary rule is Milon Friedman s (1968, 1972) rule. This rule would require policymakers o conrol money growh raes by conrolling he growh in bank reserves. The immediae objecive of his rule is o limi he rae of growh of producion and o mainain price sabiliy. Bu, he rule ha seems o be he mos used and experimened wih in macroeconomic models in recen years and ha characerize he behavior of he moneary auhoriy is he Taylor rule proposed by J.B.Taylor in This rule has been used, for he firs ime, o approximae Federal Reserve policy and deerminae he American ineres rae during he period Taylor showed ha moneary policy in he Unied Saes could be usefully described in erms of an ineres rae simple rule ha described he moneary policy process in erms of he wo major operaional objecives of moneary policy: inflaion and economic growh. Taylor noes ha he Federal funds rae rises if inflaion rises above is arge value or if oupu rises above is arge level. When boh inflaion and oupu are equal o heir arge levels, he nominal Federal funds rae will be equal o he sum of he equilibrium real ineres rae and he inflaion arge s rae. 1 Clarida, Gali and Gerler (1999), McCallum (1999) and Taylor (1999a) provide surveys of he recen moneary policy rules lieraure. 4

5 The Taylor rule, as proposed by J.B.Taylor in 1993, is used as a guideline for cenral bank o follow especially for developed counries cenral banks. The srucure of his rule is simple and easy o pracice bu a he same ime i capures he essenial of he moneary auhoriy s behavior. Since he firs use of he original Taylor rule, many economiss criicized is simple srucure. Therefore, here have many modificaions o i in order o ake ino accoun ohers macroeconomic variables ha have influence on he moneary policy. The lieraure on he modificaions of he original Taylor rule is immense. These modificaions affec he weighs of he inflaion gap and he oupu gap and he inclusions of addiional variables ha allow racking observed ineres raes more closely o he Taylor rae. In he economic lieraure, we have observed many of experiences in esimaing and formulaing Taylor ype rules. Manu economiss used he Taylor rule in is original or modified version o esimae many cenral banks reacion funcion eiher for developed or emerging counries such as he Federal Reserve Bank, he European cenral Bank, he Luxembourg Cenral Bank. These differen sudies confirm ha boh he inflaion arge and he oupu arge are decisive for he level of ineres rae. However, he weighs on he inflaion gap and he ou pu gap differ beween cenral banks and depend on heir moneary policy prioriies. The purpose of his paper is o presen a brief survey of he lieraure on he characerisics of he original Taylor rule, o focus on he limis of his rule, o review he modificaions made o is original form, o review he empirical issues relaed o he use of he Taylor s rule and o ry o verify he applicaion of his rule o he Tunisian Cenral Bank. I will srucure my paper as follows. In secion 2, I presen an overview of he Taylor rule s lieraure, he characerisics and inerpreaions given o his rule. In secion 3, I focus on he limis and criics of he original version of Taylor rule and I presen he modificaions proposed by researchers. In secion 4, I review he lieraure on he empirical uses of his rule by differen cenral banks. In secion 5, I ry o esimae, following he Taylor mehodology (1993) he shor erm ineres rae of Tunisia, and hen I compare i wih he observed Tunisian Moneary Marke ineres rae (TMM) and I ry o verify if Tunisian Cenral Bank ake ino accoun he objecives of he price sabiliy and he economic growh. 5

6 2. Taylor Rule: Presenaion and Inerpreaion 2.1. The Original Taylor Rule John.B.Taylor 2 presened for he firs ime his moneary rule in This rule is a simple moneary rule ha shows how he cenral bank should adjus is nominal ineres rae in a sysemaic manner in response o he divergences of acual GDP from is poenial level and he divergences of acual inflaion raes from he inflaion arge rae. Taylor specified he rule as follows: ( π ) β y i = r + π + α π + ~ Where i is he nominal federal funds rae, r * is he equilibrium real federal funds rae, π is he inflaion rae over he previous four quarers, π * is he arge inflaion rae, ỹ is he deviaion of real GDP from is arge level expressed in percenage: 100*[(y - y * ) / y * ]. According o his rule in equaion (1), he federal funds rae is expeced o rise if inflaion rises above is arge value π * or if real GDP rise above is arge level. Taylor s rule suggess ha he fed increases ineres raes in imes of high inflaion or when employmen is above he full employmen level and decreases ineres raes in he opposie siuaions. This rule means ha when boh inflaion and GDP are on arge, he nominal federal funds rae will be equal o he sum of he equilibrium real rae r * and arge inflaion π *. Taylor (1993) ses he inflaion arge π * and he equilibrium real rae r * equal o 2 and aribues an equal weigh of 0.5 o he inflaion and oupu gaps erms. The rule is wrien as follows: ( 2) 0.5 y i = π + ~ (1) π (2) 2 For more deails see he paper of John. B. T: «Discreion Versus Policy Rules in Pracice», Carnegie Rocheser Conference Series on public policy, N 39, pp

7 Taylor (1993) showed ha he behavior of he US federal funds rae was well described by he simple formula in equaion (2). He proposed a paricular parameerizaion of his rule ha has araced considerable aenion (r * = π * =2 and α = β = 0.5). In fac, his parameerizaion appeared o describe Federal Reserve policy decisions in he periods ha Taylor had originally examined and o reflec exacly he preference of he American auhoriy for price sabiliy and economic growh. We can rewrie he rule in equaion (2) as follows: i = π 0.5 π ( * ) ~ y i = π ~ y (3) The equaion (3) shows ha he esimaed coefficien on inflaion mus be greaer han 1. I means ha if here is an increase in acual inflaion, he nominal ineres rae mus rise more proporionally o compensae he rise of inflaion and o limi he inflaionary pressure. Taylor rule offered a convenien ool for describing Federal Reserve s moneary policy and has had many follow-ups since I owes much of is success o is simpliciy and is efficiency. I raduces fairly well he Federal Reserve s policy during he period sudied by Taylor from hrough Inerpreaion of The original Taylor Rule The principal objecives of moneary policy are o dampen business cycle flucuaions, o mainain price sabiliy and o realize he maximum susainable growh. The rule suggesed by Taylor (1993) used he ineres raes as policy insrumen o achieve hese objecives. Taylor remarked in 1990 ha «he Federal Reserve generally increases ineres raes when inflaionary pressures appear o be rising and lowers ineres raes when inflaionary pressures are abaing and recession appears o be more of a hrea», (Council of Economic Advisers, 1990, p 85) and he proposed his rule o describe closely he Federal Reserve behavior in he nineies. According o his rule, boh α and β he weighs of price sabiliy and oupu gap- should be posiive and equal o 0.5. Tha is, he rule "recommends" a relaively high ineres rae when inflaion is above is arge or when he economy is above is full employmen level, and a relaively low ineres rae in he opposie siuaions. 7

8 Someimes hese goals are in conflic: inflaion may be above is arge while he economy is below full employmen (as in he case of sagflaion). In such siuaions, he rule provides guidance o policy makers on how o ake ino accoun boh inflaion and economic growh for seing he appropriae level for he ineres rae. A crucial deail is ha Taylor rule says ha he nominal ineres rae should be increased by more han one poin for each poin increase in inflaion (equaion 3). Taylor s rule was he firs moneary rule consruced by arbiraion beween hese wo moneary policy goals. Many analyss show ha he rule does a fairly accurae job of describing US moneary policy and serves as an aid in assessing moneary policy for many cenral banks. Taylor s analysis in 1993 has had considerable influence on he way moneary economiss and praciioners hink abou he moneary policy and i has been widely discussed, criicized and reformulaed by many economiss. In he nex secion, I will presen he limis of he original Taylor rule and he modificaions made o i. 3. Limis of The Taylor Rule The greaes srengh and weakness of he Taylor rule is is simpliciy. Indeed, is simple srucure pushed many economiss o criicize i and leads many cenral bankers o conclude ha he coefficiens in he rule and he equilibrium ineres rae mus differ across counries and overime. The firs limi concerns he choice of he hree variables: inflaion, oupu gap and he neural (equilibrium) real ineres rae. We find differen ways o measure inflaion, o esimae he equilibrium real ineres rae and poenial oupu. The robusness of esimaion resuls can be sensiive o daa selecion and depend on he esimaion of rend GDP, he measure of inflaion... The second limi is he iming of he informaion used by he rule. Taylor used conemporaneous observaions of inflaion and oupu in his rule while in realiy cenral banks mus rely on lagged informaion. The srucure of he Taylor rule assumes ha policy makers consider only curren informaion when making policy decisions and his view is a odds wih he forward-looking naure of cenral banks. Finally, many economiss demonsraed ha he parameerizaion chosen by Taylor refleced exacly he preference of he American moneary auhoriy bu hey mus differ across counries. 8

9 Since he publicaion of he original Taylor rule, we observed a widespread modificaion o is iniial srucure by including addiional economic variables such as erms represening ineres rae smoohing behavior or exchange rae... In his secion, I will look a each limi in more deail The Difficulies in Measuring Variables: Measuring inflaion Despie is apparen simpliciy, implemenaion of he Taylor rule in pracice is no sraighforward 3 because implemenaion requires exac definiion of he inflaion and how o calculae i. In fac, here is no consensus neiher on which price variables should be used and measured nor on how long he inflaion period should be in Taylor rule. Taylor s original rule measured inflaion as he year-over-year change in he price deflaor for GDP (Taylor 1993). However, oher researchers have experimened wih alernaive price measures. We can measure inflaion wih he GDP price deflaor, The CPI, core CPI The ineres rae calculaed by he rule depends on he chosen measure of inflaion. Kozicki (1999), for example, compares he racabiliy of he Taylor rule wih four alernaive measures of inflaion for he U.S. These measures include year-over-year inflaion calculaed wih he CPI, core CPI, he GDP price deflaor, and a measure of expeced inflaion obained from an average of privae-secor forecass. Kozicki hen uses a Taylor rule o calculae ineres rae recommendaions ha would resul from hese alernaive inflaion measures. As expeced, he recommendaions differ significanly. Differences in ineres rae seings range from a minimum of 0.6 percenage poins o a maximum of 3.8 percenage poins. A firs glance, he mos appropriae measure of inflaion is he one ha he bank ges judged o be he bes measure and indicaor of fuure inflaionary pressure which mos moneary auhoriies seek o minimize. I is he indicaor used by he cenral bank o conduc is moneary policy and o deec inflaionary pressure 4. Alesina and al. (2001), proposed o use a hybrid inflaion rae equal o he average of he real and esimaed inflaion 5. 3 For more deails see Orphanides (1997, 1998). 4 Planier and Scrimgeour (2002). 5 Alesina and al (2001): «Defining a macroeconomic Framework for he Euro Area, Monioring he European Cenral Bank Series, N 3, CEPR. 9

10 We can see in he char 1, he imporance of variaion beween Taylor s ineres rae calculaed using wo differen measures of inflaion (GDP deflaor and CPI) in he case of he Euro area. Char 1. The margin beween Taylor ineres rae for he Euro area (CPI vs. GDP deflaor) Poins Quarer Source : ECB Clarida and al. (1998) noed ha he ineres raes calculaed by he Taylor rule are more sensiive o he measuremen of inflaion han he measuremen of he ou pu gap. We will see in he nex paragraph he differen mehods o calculae poenial oupu Measuring poenial ou pu The oupu gap is an economic measure of he difference beween he acual oupu of an economy and he poenial oupu. This laer is he oupu ha an economy could achieve when i is mos efficien, or a full capaciy. The calculaion for he poenial oupu presens some difficulies 6. Orphanides (2001) demonsraed ha here are differen mehods o measure he poenial oupu. These mehods are of a subjecive naure and can cause uncerainy surrounding he efficiency and accuracy of Taylor s rule. Taylor (1993) measures poenial oupu by fiing a ime rend o acual oupu. In he lieraure, we find oher measures. The mos used include regressing acual oupu on segmened linear rends and quadraic rends, he Hodrick-Presco filers (HP), and he mehod ha consiss in esimaing he producion funcion. These four measures have cerain limis. In fac, he mehods of linear and quadraic rends are inaccurae. The measure using 6 Orphanides ( ) 10

11 he HP filer is very sensiive o exreme dae of he sample and he one using he producion funcion approach needs abundan daa ha policymakers don necessarily have a heir disposal. The analysis of Kozicki (1999) shows ha ineres rae recommendaions using alernaive measures of poenial oupu differ by a minimum of 0.9 poins o a maximum of 2.4 poins Measuring equilibrium real ineres rae The mos difficul variable o esimae is he equilibrium real ineres rae. Taylor supposed ha his ineres rae is consan and considered i as he average of economic growh rae for he economy. Kozicki (1999), Clarida, Galί and Gerler (2000) and Judd and Rudebusch (1998) calculaed he equilibrium real ineres rae as he difference beween he average Federal funds rae and he average inflaion rae. Sachs (1999) used he average of he real ineres raes during he en las years o esimae he neural real ineres rae. Rudebusch (2001) esimaed he equilibrium real ineres rae from an IS equaion and found ha i is equal o 2.2 per cen nearly similar o he original Taylor rule. The assumpion of he consan equilibrium real ineres rae causes problems and many economiss showed ha his rae isn consan. Kozicki (1999) esimaed he equilibrium real ineres rae for he U.S. and showed ha i was varying wih he sample period. He suggesed ha equilibrium real ineres raes may no be consan. Rapach and Weber (2001) have also found he same resuls wih a number of oher counries, including Canada The Timing of Informaion: McCallum (1997) noed ha Taylor s formulaion wasn operaional because i required informaion ha policymakers did no necessarily have a heir disposal. The unavailabiliy of his curren informaion o policy makers a he ime when decisions are made has led o a debae abou wheher o use curren or lagged daa in he Taylor rule. In his original rule, Taylor supposed ha cenral banks had a heir disposal he curren quarer value of inflaion and oupu immediaely when he quarer was finished. Bu in realiy, hese indicaors aren available before several monhs. In he Unied Saes, for example, he firs release of real GDP daa for each quarer is no available unil roughly a monh afer he end of ha quarer. In addiion, we know ha hisorical daa may be revised annually. In fac, he economic daa go hrough numerous revisions afer heir original release, some of hese revisions occurring even years laer. 11

12 Taylor (1999b) uses curren-ime daa o demonsrae ha if he Federal Reserve had se he Federal funds rae as recommended by he Taylor rule, he inflaionary episode of he 1970s migh have been avoided. However, Orphanides (2001) couners his argumen by calculaing he recommendaions of Taylor s rule using he daa available o decision makers a he ime and shows ha policy makers following he original Taylor rule would no have been able o avoid he inflaionary episode The Weighs on Inflaion and Oupu Gap: The way o measure he weighs on inflaion and on he oupu gap are anoher largely discussed issue. In his 1993 paper, Taylor already menioned ha here was no a consensus on he coefficiens. He didn give any heoreical or empirical jusificaion of his parameerizaion. Normally, he weighs represen he responsiveness of moneary policy o deviaions of inflaion from he inflaion arge and deviaions of oupu from poenial oupu. In he Taylor rule, he oupu and inflaion gaps are each muliplied by a weigh of 0.5. The parameerizaion proposed by Taylor araced considerable aenion because i described exacly he American moneary policy, bu Taylor noed a lack of consensus abou he size of he weighs in policy rules. The original Taylor rule is no applicable o all cenral banks wih he same weighs as he iniial formulaion of he rule (0.5 for inflaion and oupu gap). The coefficiens differ across counries and over ime and mus be esimaed for each counry. The iniial parameers mus be adjused o much he srucure of he economy and he objecives of is cenral bank These limis involved immense modificaions o he original formulaion of he Taylor rule. These modificaions affec he weighs of inflaion and oupu gap and concern he inclusion of addiional variables o he iniial Taylor rule The Modificaions Given o he Taylor Rule: Since he firs implemenaion of he original Taylor rule, we have seen many modificaions and exensions o is original formulaion. The firs modificaion is concerning he iniial coefficien of inflaion and oupu gap. In fac, many experiences showed ha he sandard coefficiens wouldn be he same for all counries and involved he subsiuion of he weighs given by Taylor s original rule. 12

13 The second modificaion is concerning he inclusion of addiional economic variables. Many economiss suggesed adding he lagged ineres rae. I means o add he ineres rae of he previous period because, normally, he majoriy of cenral banks ake ino accoun he previous ineres rae when hey are deermining he acual rae: Ineres rae pahs se by cenral banks end o be fairly smooh, moving slowly in he same direcion raher han frequenly reversing he direcion of change. By pracicing he ineres rae smoohing, cenral banks avoid he erraic flucuaions of ineres raes 7. Williams (1999) suggess ha frequen reversals in he direcion of ineres rae movemens may appear as misakes o he public, hus mainaining momenum in ineres rae movemens will mainain confidence in he cenral bank 8. The rule is became wrien as follows: i ( ϕ )[ r + α ( π π ) + β ( y y )] = ϕ r (4) Wih φ is he smoohing degree. A sudy done by Clarida, Gali and Gerler in 1998, finds ha for a big number of counries he φ coefficien of smoohing is relaively high. I proves ha, in pracice, cenral banks smooh heir ineres raes. A hird modificaion was made o he original Taylor s rule concerning he inroducing of he exchange rae as an imporan economic variable ha can influence he level of inflaion and by he same ime he level of ineres rae. Ball (1999) and Svensson (1998) suggesed adding his variable, especially afer finding beer resuls when hey applied i o he Canadian s cenral bank. The rule hey suggesed is as follows: i ( π ) + g ( y y ) + h ( e e ) + h ( e e ) = r + f π (5) Wih e is he observed exchange rae and e * is he equilibrium exchange rae. Greiber and Herz (2000) approved he rule proposed by Ball and demonsraed ha i gave beer resuls han he original Taylor s rule because in pracice he moneary policy affecs he economy via boh he ineres rae channel and he exchange rae channel. 7 Clarida. R and Gerler. M (1996). 8 See Nell Hamalainen (2004). 13

14 In spie of hese limis and modificaions, he original Taylor s rule is he mos used moneary rule in he pracice and he one ha gives he bes resuls in guiding he moneary policy decisions. In he nex secion, we will see he lieraure on he empirical use of his rule. 4. Survey on The Empirical Uses of The Taylor Rule The lieraure on he empirical esimaion of Taylor ype rules is exensive. In fac, we find differen esimaion of he original rule in is iniial form and afer modificaions. However, empirical resuls differ subsanially beween sudies depending on he ype of daa or model used or he srucure of he economy. Drumez and Vendelhan (1997) esimaed he original Taylor rule (1993) for he case of he French cenral bank. Bu he resuls of heir esimaion were non conclusive. Also, for he same cenral bank, Morin and Thibaul (1998) implemened he original Taylor s rule and hey concluded ha he French ineres raes are sysemaically superior han he raes calculae by he rule. Sibi (2000) focused on he case of he European Cenral Bank (ECB) and esimaed he original Taylor s rule during he period 1990 o This sudy concluded ha he ECB followed effecively he simple rule of Taylor based on he wo economic aggregaes: inflaion and oupu bu wih coefficiens differen han hose used by Taylor (1993). Clarida, Gali and Gerler (1998) reached he same conclusion for he German Cenral Bank: he esimaed coefficiens are 0.47 for inflaion and 0.55 for oupu gap. In he same way, Peno, Pollin, and Selz (2000), and Clausen and Hayo (2002) showed ha he weighs on inflaion and on he oupu gap differ across European counries. Furher (1994), Clark Laxon and Rose (1997) worked on differen models of he American economy and concluded ha he iniial weighs aribued o inflaion and oupu gap (0.5) are very weak o realize he sabiliy of he economy. They suggesed ha he coefficiens mus be more imporan. Judd and Rudebusch (1998) used, also, American daa incorporaing ineres rae smoohing in a modified Taylor rule. Their research led o finding ha he esimaed coefficiens on inflaion, α, and on oupu gap, β, are greaer han 2. Levin (1996) 9, wih American daa, compared he efficiency of he original Taylor rule wih he modified Taylor rule proposed by Henderson and Mckibbin (1993) using coefficiens on 9 For more deails see Levin (1996) 14

15 inflaion and oupu gap equal o 2. He found ha he las rule is more efficien for he price sabiliy and economic growh. In he pracice, we find many empirical esimaions of he original and modified Taylor rule. Unforunaely, hese empirical resuls differ beween sudies and counries depending on he model of he rule used, srucure of he economy and he objecives of cenral banks. These divergence beween resuls, make he conclusion concerning he weighs of he inflaion and oupu gap very difficul. However, hese sudies come o he conclusion ha he parameers mus be differen han he Taylor s original rule. Beween he empirical esimaion, we remark ha he economiss implemened he original or modified Taylor s rule for developed counries and ignored he developing counries. In he nex secion, we ried o esimae following he original Taylor s rule (1993) he shor ineres rae of a Medierranean emerging counry: Tunisia. 5. Esimaing Original Taylor Rule for The Tunisian Cenral Bank The aim of his paper is o empirically esimae he original Taylor moneary rule for a small developing economy: Tunisia during he period o Then, we compare he calculaed ineres raes wih he observed raes for he same period. This comparison allows us o deermine he opimal Taylor rule and he weighs of coefficiens for he Tunisian Cenral Bank (TCB). In he las years, he ineres rae became he preferred moneary policy insrumen for he TCB concerning he liquidiy regulaion and resource allocaion. Since 1986, he moneary auhoriies renounced progressively o he idea of he adminisraive fixing of ineres raes. The moneary policy as announced by he Tunisian Cenral Bank akes ino accoun an inflaion argeing sraegy bu many heoreical elemens and empirical ess le us believe ha, a leas implicily, he Tunisian cenral bank akes ino consideraion he oupu gap. These argumens pushed us o check if he TCB follows he Taylor s rule and aims he price sabiliy and he economic growh. Then we will esimae he appropriae coefficiens for he Tunisian economy. 15

16 5.1. Model Specificaions and Variables Definiions: In he firs sep, we will use he original Taylor s rule (1993) as presened in he equaion (6): ( π π ) 0.5 y i = r ~ π (6) We use quarerly daa during he period o The daa are colleced from he Tunisian insiue of he saisics and he Tunisian Cenral Bank financial saisics. The used variables are: he ineres rae (Taylor rae), neural ineres rae, inflaion and ou pu gap. Ineres raes: are he raes calculaed following he iniial rule of Taylor. In our case, we will compare hese raes wih he Tunisian Moneary Marke (TMM) ineres raes. If he TCB respecs he Taylor rule, he TMM will be equal o he calculaed raes. Neural (Equilibrium) ineres rae: This ineres rae is calculaed wih differen mehods depending on he srucure of he economy. Sibi (2002) 10 demonsraed using «a gold rule» ha he neural ineres rae is equal o he growh rae of he economy and Sachs (1999) suggesed ha his rae is equal o he average of he previous shor erm ineres raes. In he case of he Tunisian economy, we use he second approximaion because if we use he firs approximaion (average growh rae), we find ha he neural ineres rae is equal o 4.38% and his isn reliable because he TMM never reached his value ye. We calculae he equilibrium ineres rae during he period o using he second approximaion and we find ha i is equal o 7.25%. Oupu Gap: The difference beween he acual GDP and he poenial GDP. I was calculaed by Taylor (1993) as follows: Υ = ( GDPacual GDPpoenial ) 100 GDP poeniel In our sudy, we faced wo difficulies. The firs difficuly is he deerminaion of he poenial GDP. The second difficuly is he lack of quarerly daa concerning he real GDP. In fac, he TCB gives his informaion only per year. 10 Sibi.F (2000) 16

17 To resolve he firs problem, we used he Hodrick-Presco filer o calculae he poenial GDP. The HP filer 11 is a simple saisical rend exracion echnique easy o implemen and i is widely used. For he second difficuly, we used he indusrial producion index o approximae he quarerly real GDP. We chose his index because i refleced more han 70% of he real GDP. The oupu gap is calculaed following Taylor (1993). We represened i in he following char: Char 2: Oupu Gap Trend Oupu Gap We can see in he char very imporan flucuaions of he oupu gap especially during he period 1994 o Consequenly, hese flucuaions mus have an impac when we will calculae he ineres raes following Taylor s rule. The calculaed ineres rae mus follow he movemens of he oupu gap because when he oupu gap is flucuan, normally he ineres rae mus be flucuan oo. 11 The degree of smoohing is 1600 because we used quarerly daa. 17

18 Inflaion: The inflaion is measured by he variaion of he consumer price index (CPI). This daa is available in he financial saisics of he TCB. However, we find difficuly o calculae he poenial inflaion because he Tunisian Cenral Bank don express an explici objecive of inflaion and don announce he arge inflaion. We used also in his case, he HP filer o calculae he poenial inflaion. The difference beween he real and poenial inflaion rae is represened in his Char: Char 3: Variaion of Inflaion Difference Inflaion Difference The variaion of he inflaion difference shows an erraic movemen ha mus affec he calculaed ineres rae. 5.2.: Esimaion resuls: When we calculaed he Taylor s ineres raes for he Tunisian economy, we noiced ha here is an imporan difference beween he announced ineres rae by he TCB and he calculaed rae. The calculaion of he ineres rae, aking ino accoun he price sabiliy and he economic growh, wih he original weighs showed clearly ha hese parameers can be hose fixed by he Tunisian economic auhoriy. We can see he difference beween he announced and calculaed ineres rae in he following char: 18

19 Char 4: Comparison beween he calculaed and announced ineres rae Taylor rae TMM During he sudied period ( ), he announced ineres raes (TMM) were sable whereas he ineres raes calculaed following Taylor s rule were very flucuan. This flucuaion can be explicaed by he erraic movemens of boh inflaion difference and he oupu gap. We have seen above ha he oupu gap showed a very imporan movemen and he same hing for he difference of inflaion. The flucuaion of oupu gap can be responsible in he majoriy of he flucuaion of he ineres rae. The flucuaion of he ineres rae can be also explicaed by he lack of he lagged ineres rae in he model used by Taylor especially because we know ha he TCB, like many cenral banks, ake ino accoun he lagged ineres rae o avoid he erraic movemen of ineres raes and o srenghen is credibiliy. This resul demonsraes ha original Taylor rule don reflec he accurae evoluion of real ineres rae announced by he TCB and ha original coefficiens can describe he TCB behavior. The coefficiens mus be adjused o mach he srucure of he economy and he objecives of he TCB. In he nex paragraph of his paper, I esimaed he original Taylor rule o deermine he adequae coefficiens on he inflaion and oupu gap. 19

20 I used he equaion (7) o deermine α and β for Tunisian economy: TMM = r ( π π ) + β y + + α ~ π + ε (7) Before he esimaion, we used he Uni Roo Tes o verify if he daa are saionary. We used he Dickey-Fuller es for his objecive. This sep is very imporan because he esimaion of he coefficiens require ha he daa are saionary. The es concerned he used variables: TMM, inflaion and oupu gap. The difference of inflaion, inflaion, and oupu gap are saionary bu he TMM is no saionary. We differeniaed he series of TMM o make i saionary. The equaion o esimae is wrien as follow: ( π π ) + β ~ y TMM = c+ α + ε (8) The resuls of his esimaion 12 are summarized in he able (1). Table 1: Esimaion resuls Esimaed Variable Coefficien (Suden) c ( ) α β The esimaed coefficiens are saisically significan. The esimaion demonsraes ha he TCB ake ino accoun he price sabiliy and he oupu gap in is behavior wih weighs differen from hose of he Taylor original rule (0.5). However, we conclude ha he Tunisian auhoriies ake ino consideraion he inflaion (0.212) more han he economic growh (0.015): he esimaion shows he chronology beween he objecives of he Tunisian moneary auhoriies: hey prefer o reduce inflaion before he realizaion of he economic growh. Our empirical es on he Tunisian economy shows ha he original Taylor rule can be a perfec rule for all counries. This rule can be used as a guideline for he Tunisian Cenral Bank o follow bu wih differen weighs o he objecives of price sabiliy and economic growh. The use of his rule can also improve he credibiliy of he TCB and give more 12 We used he E-Views sofware. 20

21 efficiency o is decisions. Finally we mus indicae ha he ineres moneary policy is a recen policy ha Tunisian auhoriies ry o reinforce. Consequenly, i means ha oher economic variables can affec he ineres rae and can be added o he original Taylor rule wih differen weighs such as: lagged ineres rae, exchange rae, he economic siuaion Conclusion: The original Taylor moneary policy rule sipulaes ha he insrumen of he moneary auhoriy (usually a shor-erm ineres rae) reacs o wo key goal variables: deviaions of conemporaneous inflaion from a pre-se arge rae and deviaions of conemporaneous real oupu from is poenial level. The exen o which he ineres rae reacs o shocks will depend on he relaive weighs assigned o he inflaion gap and he oupu gap. The Taylor ype rules are simple, racable, bu a he same ime capure he essence of he behavior of he moneary auhoriy. Taylor rule sill performs well in many differen models because i responds direcly o he cenral banks behavior. This aricle has focused on he usefulness of Taylor rule o policymakers as hey decide how o adjus heir shor erm ineres raes. The aricle presens he lieraure on he proprieies of he original Taylor rule; he major limis and modificaions aribued o i and focus on he experiences of some counries wih he original or modified Taylor rule. Taylor ype rule is very useful o policymakers. Firs, because Taylor-ype rules reain a simple srucure and embed crucial aspecs of moneary policy. I serves as a simple and easily undersood saring poin for hinking abou moneary policy 13. Second, i provides a convenien communicaion ool for focusing policy discussions and for educaing he public abou some of he issues of concern o Cenral Banks. And finally, i uses he wo imporan objecives of all cenral bank: price sabiliy and economic growh. Bu, he simple srucure of Taylor rules clearly has disadvanages, hiding he fac ha many aspecs of rule specificaion are subjec o considerable uncerainy and ignoring same ohers economic variables. Also, his paper aims o verify he usefulness of he Taylor rule for he Tunisian Cenral Bank: I esimae empirically a simple, Taylor moneary rule for he case of he Tunisian economy and deermine he relaive weighs on he inflaion gap and he oupu gap. The experience shows ha Tunisian cenral bank uses Taylor rule bu wih differen weighs. 13 Meyer (1999) and Blinder (1998). 21

22 References - Ball. L (1997), «Efficien Moneary Policy rules», NBER working paper N Benhabib. J, Schmi. S and Uribe. M (2001), «The Perils of Taylor Rules», Journal of Economic Theory N 96, pp Cecchei. S.G (1998), «Policy Rules and Targes: Framing he Cenral Banker s Problem», Federal Reserve Bank of New York, Economic Policy Review, Volume 4, N 2, pp Clarida. R and Gerler. M (1996), «How he Bundesbank Conducs Moneary Policy», NBER working paper N Clarida. R, Gali. J and Gerler. M (1998), «Moneary Policy Rules and Macroeconomic Sabiliy: Some Inernaional Evidence», European Economic Review, Volume 42, pp Clarida. R, Gali. J and Gerler. M (1998), «Moneary Policy Rules and Macroeconomic Sabiliy: Evidence and Some Theory, NBER Working Paper, N De Masi. P.R (1997), «IMF Esimaes of Poenial Oupu: Theory and Pracice», Working Paper of he Inernaional Moneary Fund, WP/97/117, December. - Drew, A and B Hun (2000), «Efficien Simple Policy Rules and he Implicaions of Poenial Oupu Uncerainy», Journal of Economics and Business Special ediion on Money and Moneary Policy in a Changing World, N 52, pp Goodfriend, Marvin. (1991), «Ineres Raes and he Conduc of Moneary Policy», Carnegie-Rocheser Conference Series on Public Policy, Volume 34, pp Gujarai Damodar. N (1988), «Basic Economerics», Inernaional Ediion. - Hezel, R (2000), «The Taylor Rule: Is i a useful guide o undersanding moneary policy?» Federal Reserve Bank of Richmond, Economic Quarerly, Volume 86, pp Judd. J and Rudebusch.D (1998), «Taylor s Rule and he FED: , Federal Reserve Bank of San Francisco», Economic Review, N 3, pp Kozicki. S (1999), «How useful are Taylor rules for Moneary Policy», Federal Reserve Bank of Kansas, Economic Review, Volume 84, Second quarer, pp Kydland. E and Presco. E (1977), «Rules Raher han Discreion: he Inconsisency of Opimal Plans», Journal of Poliical Economy, Volume 85, pp

23 - McCallum, B.T. (1999), «Recen Developmens in he Analysis of Moneary Policy Rules», Federal Reserve Bank of S. Louis Review, Volume 81, N 6, pp McCullum. B.T (2000), «The Presen and The Fuure of Moneary Policy Rules», NBER working paper N Moniel. P, (1991), «The Transmission Mechanism for Moneary Policy in Developing Counries», IMF Saff Papers, N 38, pp Orphanides. A (2001), «Moneary policy rules based on real-ime daa», American Economic Review, Volume 91, pp Orphanides. A (2003), «Hisorical Moneary Analysis and Taylor Rule», working paper of he Federal Reserve Bank. - Razzak. W.A (2001), «Is he Taylor Rule Differen From he McCallum Rule?», Discussion Paper Series of he Reserve Bank of New Zealand. N 7. - Rudebush. G and Svensson. L.E.O (1998), «Policy Rules for Inflaion Targeing», NBER working paper N Sibi. F (2000), «Règle de Taylor e Applicaion à la Zone Euro», paper presened a he GDR conference, 18 h Days of Moneary and Banking Economy, Pau 21 e 22 June Söderlind. P, Södersröm. U and Verdin. A (2003), «Taylor Rules and he Predicabiliy of Ineres Raes», Sveriges Riksbank working paper N Svensson. L.E.O (1998), «Inflaion argeing as a Moneary Policy Rule», NBER working paper N Svensson. L.E.O (2002), «Wha is wrong wih Taylor Rules? Using Judgmen in Moneary Policy hrough Targeing Rules», NBER working paper N Taylor. J.B (1993), «Discreion Versus Policy Rules in Pracice», Carnegie Rocheser Conference Series on public policy, N 39, pp Taylor. J.B (1998), «An Hisorical Analysis of Moneary Policy Rules», NBER working paper N Taylor. J.B (1999), «The Robusness and Efficiency of Moneary Policy Rules as Guidelines of Ineres Rae Seing By European Cenral Bank», Journal of Moneary Economics, Volume 43, pp

24 - Ulrich. K (2003), «A Comparison beween he FED and he ECB: Taylor Rules», Cener for European Economic Research, working paper. - Verdelhan. A (1999), «Taux de Taylor e aux de marché de la zone Euro», Bullein of he French Cenral Bank, N 61, pp Weymark. D (2001), «Using Taylor Rules as Efficien Benchmarks», NBER working paper N Young, P, Dunn. D, Giusiniani. A, Nadal De-Simone.F, Tanner.E, and McHugh.J (1999), «Dominican Republic: seleced issues», IMF Saff Repor N.99/117, Inernaional Moneary Fund, Washingon, D.C., Ocober. 24