EQUILIBRIUM Quarerly Journal of Economics and Economic Policy 2017 VOLUME 12 ISSUE 2, June p-issn 1689-765X, e-issn 2353-3293 www.economic-policy.pl ORIGINAL PAPER Ciaion: Tura-Gawron, K. (2017). The forecass-based insrumen rule and decision making. How closely inerlinked? The case of Sweden. Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315. doi: 10.24136/eq.v12i2.16 Conac: karolina.ura@zie.pg.gda.pl, Gdansk Universiy of Technology, Naruowicza 11/12, 80-233 Gdansk, Poland Received: 13 December 2016; Revised: 12 April 2017; Acceped: 9 May 2017 Karolina Tura-Gawron Gdansk Universiy of Technology, Poland The forecass-based insrumen rule and decision making. How closely inerlinked? The case of Sweden JEL Classificaion: E52; E58; E61 Keywords: inflaion argeing regime; decision making; inflaion forecass; Taylor rule Absrac Research background: The Cenral Bank of Sweden declared in years 1999 2006 he implemenaion of he Svensson s concep of inflaion forecas argeing (IFT). I means ha he repo rae decision-making process depends on he inflaion fore-cass. The concep evolved from he sric IFT wih he decision-making algorihm called he rule of humb o he flexible IFT. Purpose of he aricle: The aim of he aricle is o: (1) analyze he influence of he inflaion rae and GDP growh rae on he repo rae decisions, (2) analyze he influence of he inflaion rae and GDP growh rae forecass (in wo year horizon) on he repo rae decisions in Sweden in years 1999 2006. Mehods: The analysis encompasses he repo raes decisions, CPI inflaion rae, GDP growh rae, cenral pahs of CPI inflaion forecass and cenral pahs of GDP growh rae forecass (he mode values) in he wo years horizon published by The Cenral Bank of Sweden in years 1999 2006. The sudies are based on he Taylor-ype insrumen rule and forecas-based Taylor-ype insrumen rule. The mehodology used is muliple linear regression models. Findings & Value added: The Cenral Bank of Sweden in years 1999 2006 implemened direc inflaion forecas argeing (DIFT) rule. The decision-making algorihm was based on he CPI inflaion forecass and he rule of he humb algorihm. The exac rule of he humb was as follow: if he inflaion forecas, in he wo year forecas s horizon exceeded he inflaion arge by 1 p.p., hen he cenral bank raised he repo rae by 0.4 p.p; if i was below i,
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 hen he cenral bank reduced he repo rae by 0.4 p.p. If he inflaion forecas was equal o he inflaion arge, hen he repo rae remained unchanged. The hisorical repo raes differ from he heoreical esimaed rule of he humb s repo raes by +/-0.28 p.p. Inroducion Inflaion argeing (IT) regime is nowadays one of he mos common moneary policy sraegy (i is used by 27 cenral banks). There are several cenral banks which are he pioneers in implemening he new ideas concerning his regime. To such pioneers surely belongs The Cenral Bank of Sweden (Tura, 2015, pp. 292). In his paper we analyse he repo rae decisions in The Cenral Bank of Sweden (Sveriges Riksbank, SR) in years 1999 2006. The sudy refers o he implemenaion of he decision-making algorihm called he rule of he humb. The Cenral Bank of Sweden and he research horizon has been chosen for his analysis due o five reasons: 1) The Cenral Bank of Sweden has a high level of ransparency according o he publicaion of implemened moneary policy rule, feaures of IT sraegy and feaures of forecasing model (Tura, 2015, pp. 292); 2) The Cenral Bank of Sweden is one of several cenral banks which published he weighs pu on he inflaion rae and GDP growh rae applied in he main forecasing RAMSES model; 3) he auhor of he concep of inflaion forecas argeing, L.E.O. Svensson, was acive as advisor o The Cenral Bank of Sweden during he years 1990-2007; 4) The Cenral Bank of Sweden officially declared in years 1999-2006 he implemenaion of inflaion forecas argeing rule, he rule of he humb decision-making algorihm, published he inflaion forecass and heir exac daa, and made he inflaion forecass based on he assumpion of consan insrumen rae during he forecas horizon (CIR); 5) cenral bank s inflaion forecass in Sweden had a large impac on consumers inflaion expecaions in Sweden (Szyszko, 2016, p. 9). The inflaion argeing regime may be perceived as a discreionary or based on a rules sraegy. In his paper we relae o he L.E.O. Svensson s rules approach (see: Svensson, 1997, pp. 1111 1146, 2005a, pp. 1 54, Svensson & Telow, 2005b, pp. 177 207). The researchers sill argue wheher he IT rule should be modelled as an insrumen or argeing rule (see: Svensson, 2002, pp. 771 780, 2003, pp. 426 477, McCallum & Nelson, 2005, pp. 597 611). I is difficul o achieve he consensus. In his paper we lean o he insrumen- based IT rules. Such a view comprises he reurn o Taylor rule and forecas-based Taylor rule. 296
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 The main aim of he sudy is o analyse empirically he applicaion of he rule of he humb decision-making algorihm and inflaion forecas argeing (IFT) rule in The Cenral Bank of Sweden. The main research quesion is as follows: Did he Moneary Policy Commiee in Sweden beween 1999 2006 make he repo raes decisions on he forecass-based insrumen rule and he rule of he humb algorihm? This will be achieved in he framework of he hypohesis: If he cenral bank implemen he sric IFT wih he algorihm he rule of he humb, he Execuive Board s repo rae decisions depend on he inflaion forecass; if flexible IFT wih he algorihm he rule of he humb depend on inflaion rae and GDP growh rae forecass. According o his, he four subquesions have been posed: 1) Did The Cenral Bank of Sweden apply in years 1999 2006 he rule of he humb? 2) Wha were he weighs conferred on he inflaion rae and GDP growh rae in he Moneary Policy Commiee repo raes decisions? How flexible were hey? 3) Wha were he weighs conferred on he inflaion rae forecass and GDP growh rae forecass in he Moneary Policy Commiee repo raes decisions? How flexible were hey? 4) Were he repo raes decisions easy o predic by economic agens? The paper is organised as follows. I consiss of five pars. The auhors begin in secion 1 by providing some heoreical background abou insrumen Taylor rule, Svensson s concep of IFT rule and he Taylor-ype forecass-based insrumen rules. The nex hree secions include he descripion of he mehodology, he daa and he resuls of he research. The conclusions and implicaions for moneary policy are conained in he fifh secion. Theoreical background The sudy relaes o he wo similar and based on rules conceps on conducing he moneary policy. The firs one is he Taylor insrumen rule and he second one, he Svensson s rule of he humb. Boh conceps refer o seing he cenral bank s insrumen rae on he basis of he deviaions from he arge variables. The rule of he humb comprised, in addiion o he Taylor rule, he forward looking approach on moneary policy, which requires he forecas publicaion. 297
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 The original Taylor rule was esimaed for US economy for years 1987 1992. I showed he relaion beween he federal funds rae, inflaion and real GDP. The derived policy rule, is as follows (Taylor, 1993, p. 202): i = π + 0.5y + 0.5( π 2) + 2, (1) where: i is federal funds rae, πˆ is rae of inflaion (measured by GDP deflaor) over he previous four quarers, y is he percen deviaion of real GDP from he arge. The inflaion arge was se in his example on 2 percen and real GDP arge was explained as he real GDP rend. The equaion indicaes he moneary policy rule: he federal funds rae raises if inflaion increases above a arge of 2 percen or if real GDP raises above rend GDP. If boh, he inflaion rae and real GDP are on arge, hen he federal funds rae would equal 4 percen, or 2 in real erms (Taylor, 1993, p. 202). The Taylor rule was esimaed in 1993, since han a lo of new Taylor-ype rules have been rerieved and described. The one kind of his evoluion is Taylorype forecas-based insrumen rule. L.E.O. Svensson s concep of inflaion forecas argeing was inroduced in 1997. The ground of his idea is he forward looking aiude on conducing moneary policy. According o L. E. O. Svensson, he IT regime may be characerised by hree specific feaures: cenral bank commimen o mainain price sabiliy, explici inflaion arge and publicaion of cenral banks inflaion forecass, which play a role of inermediae arges. The rule of he humb implies ha condiional inflaion forecas should hi he inflaion arge in wo year horizon. If he inflaion forecas, in he chosen horizon, is above he inflaion arge, hen he cenral bank should raise he repo rae. If he inflaion forecas in he chosen horizon is lower han he inflaion arge, hen he cenral bank should reduce he repo rae. If he inflaion forecas is equal o he inflaion arge, hen he repo rae should remain unchanged (Svensson, 1997, pp. 1111 1146). The rule of he humb implemenaion indicaes he publicaion of inflaion forecass made for a wo year horizon and on he assumpion of consan insrumen rae during he enire forecas horizon (called CIR). The inflaion forecass may shape he economic agens inflaion expecaions and anchor hem on he inflaion arge. Inflaion forecas argeing (IFT) may be divided ino wo ypes. The firs one, called direc inflaion forecas argeing (DIT), assumes seing he cenral bank s ineres rae only on he basis of inflaion forecass. I is im- 298
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 possible o implemen such an approach exacly in cenral banking pracice. The flexible inflaion forecas argeing (or forecass argeing, Svensson, 2005a, pp.1 54) preconceived ha insrumen rae decisions depend on wo arge variables, inflaion forecas and oupu gap forecas, and are made on he basis of is deviaions from he inflaion arge and poenial oupu gap (respecively). In such a case he inflaion arge may be achieved in a longer horizon. The weigh which is pu on he oupu gap forecas may deermine how quickly he inflaion forecas is adjused owards he inflaion arge (Svensson, 2009, pp. 1 9). The forecass argeing concep evolved in L.E.O. Svensson s sudies ino he opimal moneary policy plan. I includes he publicaion of macroeconomic forecass which conain he group of arge variables (forecass of inflaion and oupu gap/gdp, ec.) and ineres rae pah forecass (called as forward guidance (see: Svensson, 2015, pp. 19 64). The forecass-based arge variables are convergen wih he ineres rae forecas. The inflaion forecas a he end of he longer (usually hree years forecas) horizon achieve or is very close o he inflaion arge. Such an approach includes seing he insrumen rae accordingly o he ineres rae forecas (Svensson & Telow, 2005b, pp. 177 207). L.E.O. Svensson persised on modelling he IT sraegy as a kind of argeing rule. I is conneced wih minimalizaion of cenral bank loss funcion which consiss of deviaion of he arge variables from he arge level (deviaion of inflaion forecas from he inflaion arge and oupu gap forecas from he poenial oupu gap). According o Svensson and Rudebush (1999, p. 211) cenral bank loss funcion implies an implici insrumen rule. The Taylor rule is a ypical explici insrumen rule. The difference concerned he background of arge variables. In he original Taylor rule he arge variables were exogenous. In he Svensson s rule of he humb he arge variables (forecass) are endogenous. To simplify, he simple insrumen rule and he model are creaing he implici insrumen rule (Svensson & Rudebush, 1999, pp. 203 262). There are pleny of sudies which concerned he esimaion of simple Taylor rule for specific economies. In our paper we refer o he concep, which posed he consensus beween he simple original Taylor rule and L.E.O. Svensson s forecass argeing rule. In his poin we refer o he Taylor-ype forecass-based insrumen rules, which are he simple cenral bank implici reacion funcions, where he forecass of inflaion rae and oupu gap play a role of inermediae arge variables. These forecass are model consisen. 299
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 The general specificaion of forecas based insrumen rules is as follows (Levin e al, 2003, p. 625): i (2) = α ii 1 + ( 1 αi )( i * + E ˆ π + θ ) + απ ( E ˆ π + θ π*) + α ye y+ κ, where: i is shor-erm nominal ineres rae, πˆ is four quarer average inflaion rae, y is oupu gap (he deviaion of oupu gap from poenial), i * is uncondiional mean of he shor-erm ineres rae, E is operaor of he forecas of inflaion or oupu gap using informaion available in period, -years, is inflaion arge, π * θ is forecas horizon for inflaion forecas, κ is forecas horizon for oupu gap forecas. The Taylor-ype forecass-based insrumen rules esimaed and derived by he researchers differ in four main assumpions: he use of ineres rae smoohing, he chosen forecass horizon, oupu gap forecass encompassing and assumed poenial oupu. Mos of he rules include he ineres rae smoohing. The chosen forecas horizons oscillae beween wo and fifeen quarers. The poenial oupu gap may be derived from he model, as an oupu rend or be explicily arranged. The feaures of he chosen Taylorype forecass-based insrumen rules are shown in Table 1. In our sudies we are referring o he rules wih he wo year forecas horizon. Our choice was caused by hree reasons. Firsly, he original L.E.O. Svensson s rule of he humb assumed he wo year inflaion forecas horizon (see: Svensson, 1997, pp. 1111 1146). Secondly, Baini & Nelson (2001, p. 910) were analysing he opimal policy horizon for a se of forecas-based arge variables as a par of flexible inflaion argeing framework. They found ha i is opimal o remove he effecs of he various shock considered over a period of 8 o 19 quarers (Baini & Nelson, 2001, p. 910). Finally, The Cenral Bank of Sweden officially declared he use of he rule of he humb wihin wo year ime lags (see: Rosenberg, 2006, pp. 1 8). According o his, he similar rules were analysed by Rudebush & Svensson (1999, pp. 203 262). 300
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 Daa The Cenral Bank of Sweden (Sveriges Riskbank, SR) has been implemened IT sraegy since 1993 and has deermined he inflaion arge as 2% measured by CPI index. During he years 1993 2016 i declared wo ypes of IT rules: he rule of he humb and opimal moneary policy algorihm. In his connecion, he cenral bank published he inflaion forecass condiioned by he consan insrumen rae during he enire forecas horizon (called CIR) and he se of macroeconomic forecass condiioned by he ineres rae pah forecas. The forecas horizon depends on he chosen rule ype. The daa are analysed quarerly. The poenial GDP growh rae was esimaed and declared o be as a desirable value in a range 2 2.5% in Sweden (Heikensen, 2000; 2003). A he end of 1999 The Cenral Bank of Sweden has sared o publish he forecass daa. The analysis of he rule of he humb includes he years 1999 2006. The main informaion on he forecas-based moneary policy in Sweden is presened in he Table 2. In his paper, he cenral pahs of he cenral bank s forecass are analysed a he wo year prognosic momen of he forecass horizon. This is due o he rule of he humb assumpion: he cenral bank should be forward-looking and reac o on he deviaions of he forecass from he arge in wo year forecass horizon. The forecass cenral pahs have been downloaded from he swedish cenral bank websie (inflaion repors boxes) and Inflaion/Moneary Policy Repors. The repo raes daa were colleced from The Cenral Bank of Sweden websie. Inflaion forecass cenral pahs published in years 1999 2006 by The Cenral Bank of Sweden are presened in Figure 1. The repo rae in Sweden in years 1999-2006 is shown in Figure 2. The CPI inflaion rae and GDP growh rae daa were colleced from he Eurosa daabase. During he years 1999 2006 he forecass were made on he basis of he DSGE RAMSES model. The model applicaion assumed seing he insrumen rae on he rule of he humb algorihm. The enforced in he model insrumen rule has he following form (Adolfson e al., 2007b, p. 21): = ( π π π + ε, (3) i f *; ; y ; y ; i 1 ; x ) i, where: i is policy rae; π is underlying inflaion rae; π is change in he rae of underlying inflaion; π * is inflaion arge; 301
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 y is GDP gap; is change in he GDP gap; y i, x is exchange rae gap; ε is called as a as a measure of he elemen of moneary policy surprises; means years, {1,2,...}. According o he model of Adolfson e al. (2007a, pp. 481 511) he real exchange rae gap is measured as he percenage deviaion of he acual real exchange rae from an assumed equilibrium level ha is consan. The model implemened also he ineres rae smoohing. Research mehodology The research includes he esimaions of differen ype-taylor insrumen rules for he Sweden economy based on hisorical daa. The esimaions differ in he chosen arges variables and assumpions. The main mehod used is muliple linear regression models. The sudies conduced have been divided ino wo pars. The firs par of he sudy consiss of wo sages. A he firs sage, we assume ha he moneary policy reacion funcion is linear funcion of he arge variables and lagged insrumen rae. To he arge variables belong he CPI inflaion rae and GDP growh rae gap. According o his, he simple Taylor-ype insrumen rae rule has a form (Svensson, 2003, p. 426, Taylor, 1999, p. 5) : where: i is policy rae; i α + α π π π *) + α ( y y*) + α i +, (4) = 0 ( y i 1 ε π is CPI inflaion rae; π * is CPI inflaion arge seled a 2%; y is GDP growh rae, y * means years, {1,2,...}. is poenial GDP growh rae; 302
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 GDP growh rae gap is calculaed as he difference beween real GDP growh rae and he poenial GDP growh rae (which was seled by he auhors a 2.25% (as he midpoin of he range 2 2.5%)). The second sage is similar o he previous one, bu assumed he linear Taylor-ype reacion funcion in he form which was applied in he RAMSES model (see: Adolfson e al., 2007, pp. 481 511). In he esimaion we assumed he consan exchange rae gap. In his sep we esimaed he equaion for he arge variables: deviaions of CPI inflaion rae from he inflaion arge, change in he CPI inflaion rae, change in he GDP growh rae and GDP growh rae gap, following he form (Adolfson e al., 2007, pp. 5 40): i α + α π π*) + α π + α ( y y*) + α y + α i, (5) = 0 π ( π y y i 1 + ε where: is change in he rae of underlying inflaion; π y is change in he rae of GDP growh rae; means years, {1,2,...}. The purpose of hese wo sages is o calculae he empirical weighs which are pu on he deviaions of CPI inflaion rae from he inflaion arge and GDP growh rae oupu gap in seing he insrumen rae in simple insrumen Taylor rule and Taylor-ype insrumen rule derived from RAMSES model. A he end of his par we compare he regression resuls wih he proposed exac values of coefficiens for arge variables in he main forecasing RAMSES model. The second par of he sudy is similar o he previous one. We also assumed ha reacion funcion is linear funcion of he arge variables, bu insead of he inflaion rae and GDP growh rae we placed he inermediaes arges: inflaion forecas and GDP growh rae forecass wo years ahead. Such a view is coinciden wih he forecas-based insrumen argeing rule proposed by Svensson (1997, pp. 1111 1146). According o his, he simplified version of Taylor- ype forecas- based insrumen rae rule may have a form: 303
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 i = 0 + 2 ( + 2 i 1 y + 2 + 2 i 1 i 1 α + α π π π *) + α ( y y*) + α i + ε. (6) + 2 i 1 where: π is CPI inflaion forecas in eigh quarer horizon made on he assumpion of consan insrumen rae i 1 over he forecas horizon, y + 2 i 1 is GDP growh rae forecas in eigh quarer horizon made on he assumpion of consan insrumen rae i 1 over he forecas horizon. A he nex sage we also assumed ha reacion funcion is linear funcion of he arge variables and we placed he inermediaes arges: inflaion forecas and GDP growh rae forecass wo years ahead. The Taylor- ype forecas- based insrumen rule has a form rerieved from RAMSES model: i = α + α ( π + α y y i 1 + ε. π*) + α π + α ( y 0 π + 2 i 1 π y + 2 i 1 + α i y*) + (7) The purpose of his par is o calculae he empirical weighs which are pu on he deviaions of he CPI inflaion forecass from he inflaion arge and deviaions of GDP growh rae forecass from he previously assumed poenial GDP growh rae in seing he insrumen rae. This sep may show wheher SR implemened he rule of he humb and wha was he degree of is flexibiliy. A he end of his par we compare he resuls wih he weighs suggesed in he RAMSES model. The RAMSES model assumed he following weighs: 1.7 for he inflaion deviaions from he inflaion arge, 0.3 for inflaion changes, 0.04 for GDP gap and 0.1 for GDP changes (Adolfson e al., 2007b, p. 21). The whole research plan is presened in Table 3. Resuls Inflaion forecass cenral pahs a he wo year prognosic momen of he forecass horizon and he repo raes changes in Sweden in years 1999 2006 are shown in Figure 3. Firsly we esimaed he simple linear Taylor-ype insrumen rule wih arge variables: deviaions from he CPI inflaion rae and inflaion arge, ad GDP growh rae gap. Afer ha we esimaed he Taylor-ype insrumen rule wih he form downloaded from he RAMSES model. In boh 304
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 α y cases, he derived arge variables coefficiens have significan, posiive and similar influence on insrumen rae ( α π =.14 and, α y =0.1; α π =0.14 and =0.11) and indicae he flexible ype of implemened IT regime. In Figure 4 here are he variaions of hese formulaions wih differenial responses o inflaion and oupu, following he forms (Orphanides, 2003, p. 985; Adolfson e al., 2007b, p. 21): i i i i 1 y y = α π ( π π*) + α ( y *), (8) = α π ( π π *) + α π + α y ( y y*) + α y. (9) 1 y On he graph we compare hree he repo rae pahs: heoreical pahs derived from he simple Taylor rule, he form from RAMSES and he heoreical pah calculaed from he declared in RAMSES coefficiens. The accomplished repo raes from he RAMSES model(wih he declared weighs coefficien) differ from he hisorical repo raes by +/- 1.65. I means ha he repo raes decisions may no be prediced on he basis of his equaion. Secondly we esimaed he simple linear Taylor-ype forecas-based insrumen rule wih arge variables: deviaions from he inflaion forecas and inflaion arge, and deviaions from GDP growh rae forecas and poenial GDP growh rae. Afer ha we esimaed he Taylor-ype insrumen rule form from he RAMSES model. The resuls are similar in boh cases. Only he deviaions of inflaion forecas from he inflaion arge have significan, posiive influence on insrumen rae (0.4). I indicaes he implemenaion of sric inflaion forecas argeing and he original rule of he humb. The repo raes from he model esimaed differ from he hisorical by +/- 0.29. In Figure 5 here are he variaions of hese formulaions wih differenial responses o inflaion forecas and oupu forecas, following he form: i i = απ ( π π*) + α π π + α y ( y y*) + α y. (10) 1 + 2 y i 1 + 2 i 1 The exac heoreical repo raes derived from he original model s RAMSES forecas-based Taylor rule differ (absolue average) from he exac repo raes by +/-0.4 p.p. I also means ha he repo raes decisions may no be prediced on he basis of his equaion. In Table 4 here are he differences beween he exac hisorical repo raes and he heoreical repo raes derived from he calculaion of weighs from he RAMSES insrumen 305
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 equaion pu on he arge variables. The whole research esimaion resuls are shown in Table 5. Conclusions In he years 1999 2006 he Swedish Cenral Bank declared he implemenaion of inflaion argeing sraegy. According o he esimaed simple Taylor-ype rule, we may sae ha he cenral bank applied inflaion argeing flexible ype, wih he weighs pu on he CPI inflaion rae and GDP growh rae. The esimaions resuls for he simple Taylor-ype rule and he form of his rule from he RAMSES model did no differ significanly. The Cenral Bank of Sweden in years 1999 2006 also declared he use of he concep of inflaion forecas argeing and he rule of he humb decision-making algorihm. In his case he deviaions of CPI inflaion forecass from he inflaion arge and he deviaions of GDP growh rae forecass from he poenial GDP growh rae were our arge variables in Taylor-ype forecas-based insrumen rules. The esimaion resuls describe he implemened sraegy as a direc inflaion forecas argeing (DIFT), wih he weigh pu on he CPI inflaion forecas. The GDP growh rae forecass ranspired o be no significan in seing he repo raes. The weigh pu on he inflaion forecass is posiive, consisen wih he rule of he humb. The exac rule of he humb for Sweden in years 1999 2006 was as follows: if he inflaion forecas, in he wo year horizon exceeded he inflaion arge by 1 p.p., hen he cenral bank raised he repo rae by 0.4 p.p. If he inflaion forecas in he wo year forecas horizon was lower by 1 p.p. han he inflaion arge, hen he cenral bank reduced he repo rae by 0.4 p.p. If he inflaion forecas was equal o he inflaion arge, hen he repo rae remained unchanged. The hisorical repo raes differ from he heoreical rule of he humb repo raes by +/-0.28 p.p. Wha is more, here were large differences beween he exac hisorical repo raes and he heoreical repo raes calculaed from he exac insrumen equaion from forecasing RAMSES model. I means ha he economic agens migh no predic he repo raes changes on he basis of declared weighs pu on arge variables from he model. 306
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 References Adolfson, M., Laseen, S., Linde, J., & Villani, M. (2007a). Bayesian esimaion of an open economy DSGE model wih incomplee pass-hrough. Journal of Inernaional Economics, 72(2). doi: 10.1016/j.jineco.2007.01.003. Adolfson, M., Laseen, S., Linde, J., & Villani, M. (2007b). RAMSES- a new general equilibrium model for moneary policy analysis. Economic Review, Sveriges Riksbank, 2. Baini, N., & Nelson, E. (2001). Opimal horizons for inflaion argeing. Journal of Economic Dynamics and Conrol, 25(6-7). doi: 10.1016/S0165-1889(00)00060-9. Clarida, R., Gali, J., & Gerler, M. (2000). Moneary policy rules and macroeconomic Sabiliy: evidence and some heory. Quarerly Journal of Economics, 115(1). doi: 10.1162/003355300554692. de Brouwer, G., & Ellis, L. (1998). Forward-looking behaviour and credibiliy: some evidence and implicaions for policy. Reserve Bank of Ausralia Research Discussion Paper, 9803. Heikensen, L. (2000). Moneary policy and wage formaion. Speech a he Office of Labour Marke Policy Evaluaion (IFAU). Sockholm.05.05.2000. Heikensen, L. (2003). Moneary policy and poenial growh. Speech a he Swedish Economics Associaion, Sockholm, 28.03.2003. Levin. A, Wieland, V., & Williams, J. C. (2003). The performance of forecasbased moneary policy rules under model uncerainy. American Economic Review, 93(3). doi: 10.1257/000282803322157016. McCallum, B., & Nelson, E. (2005). Targeing vs. insrumen rules for moneary policy. Federal Reserve Bank of S. Louis Review, 87(5). Orphanides, A. (2001). Moneary policy rules based on real-ime daa. American Economic Review, 91(4). doi: 10.1257/aer.91.4.964. Orphanides, A. (2003). Hisorical moneary policy analysis and Taylor rule. Journal of Moneary Economics, 50. doi: 10.1016/S0304-3932(03)00065-5. Rosenberg, I. (2006). The Riksbank s inflaion argeing policy he significance of he new ineres rae assumpion. Speech a Swedbank markes, Sockholm, 19.04.2006. Svensson, L. E. O. (1997). Inflaion forecas argeing: implemening and monioring inflaions arges. European Economic Review, 41(6), doi: 10.1016/S0014-2921(96)00055-4 Svensson, L. E. O., & Rudebush, G. (1999). Policy rules for inflaion argeing. In J. T. Taylor (Ed.). Moneary policy rules. Chicago: Universiy of Chicago Press. Svensson, L. E. O. (2002). Inflaion argeing: Should i be modeled as an insrumen rule or a argeing rule?. European Economic Review, 46(4-5). doi: 10.1016/S0014-2921(01)00212-4. Svensson, L. E. O. (2003). Wha is wrong wih Taylor rules? Using judgemen moneary policy hrough argeing rules. Journal of Economic Lieraure, 41(2). doi: 10.1257/002205103765762734. 307
Equilibrium. Quarerly Journal of Economics and Economic Policy, 12(2), 295 315 Svensson, L. E. O. (2005a). Moneary policy wih judgmen: forecas argeing. Inernaional Journal of Cenral Banking, 1(1). Svensson, L. E. O., & Telow, R. (2005b). Opimal policy projecions. Inernaional Journal of Cenral Banking, 1(3). doi: 10.2139/ssrn.813284. Svensson, L. E. O. (2009). Flexible inflaion argeing lessons from he financial crisis. Speech a he workshop: Towards a new framework for moneary policy? Lessons from he crisis. Neherlands Bank. Amserdam. 21 Sepember. Svensson, L. E. O. (2015). Forward guidance. Inernaional Journal of Cenral Banking, 11(4). doi: 10.3386/w20796. Szyszko, M. (2017). Cenral bank s inflaion forecass and expecaions. A comparaive analysis. Pargue Economic Papers. 26(3). doi: 10.18267/j.pep.614. Taylor, J. B. (1993). Discreion versus policy rules in pracice. Carnegie-Rocheser Conference Series on Public Policy, 39. doi: 10.1016/0167-2231(93)90009-L. Taylor, J. B. (1999). Inroducion. In J. T. Taylor (Ed.). Moneary policy rules. Chicago: Universiy of Chicago Press. Tura, K. (2015). Cenral banks inflaion projecions. On he edge of echnical and sraegical approach. Warszaw: Difin. Acknowledgemens The sudy was performed during he Dekaban-Liddle Fellowship a Universiy of Glasgow in 2017. 308
Annex Table 1. The overview of he chosen Taylor-ype forecass-based insrumen rules Sudies Ineres rae smoohing Inflaion forecas Inflaion forecas horizon (quarers) Oupu gap forecas Clarida e al. (2000, pp. 147 180) Yes Yes 4 Yes 0 Orphanides (2001, pp. 964 985) Yes Yes 4 Yes 4 De Brouwer & Ellis (1998) No Yes 4 Yes 4 Baini & Nelson (2001, pp. 891 910) Yes Yes 2 and 15 No - Rudebush & Svensson (1999, pp. 203 262) Yes Yes 8 and 12 No - Oupu gap forecass horizon (quarers) Source: Levin e al. (2003, p. 625).
Table 2. The main informaion on he forecas based moneary policy in Sweden 1999 2000 2001 2002 2003 2004 Year The forecas based rule Rule of he humb 2005 CIR 2006 and ME 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Opimal moneary policy The insrumen rae assumpion in he forecas Published forecass CIR Inflaion forecass, GDP forecass* E Inflaion forecass, GDP forecass, Insrumen rae forecass 4 3 Forecass published per year 8 Q 12 Q Forecass horizon No Forward Guidance Moneary Policy Commiees Meeings CIR-Consan insrumen rae during he forecas horizon ME- Marke expecaions insrumen rae during he forecas horizon E- Endogenous insrumen rae doveailed wih insrumen rae forecas pah *In hese years GDP growh forecass were no published in he form of chars bu were described in he inflaion repors wih all necessary cenral pahs daa Each monh No Moneary Policy Trade of Descripion Source: own elaboraion based on he Inflaion Repors published by The Cenral Bank of Sweden beween 1999 2016.
Table 3. Research plan Par Sage Research quesion? Descripion Par I Par I Sage I Sage II Sage I Sage II How flexible is SR in his ineres rae decisions? How flexible is SR in his forecas-based ineres rae decisions? Calculaion of weighs imposed on he deviaions of inflaion rae from he inflaion arge and GDP growh gap in simple Taylor-ype insrumen rule Calculaion of weighs imposed on he deviaions of inflaion rae from he inflaion arge, GDP growh gap, change in inflaion and change in GDP gap in Taylor-ype insrumen rule prosposed in RAMSES Calculaion of weighs imposed on he deviaions of inflaion forecass from he inflaion arge and GDP growh forecas from poenial GDP growh Calculaion of weighs imposed on he deviaions of inflaion forecass from he inflaion arge and GDP growh forecas from poenial GDP growh, change in inflaion and change in GDP gap Table 4. Differences beween he exac hisorical repo raes and he heoreical repo raes derived from he exac RAMSES equaion Rule Targe variables Diference (absolue average) Simple Taylor-ype insrumen rule CPI inflaion, GDP growh rae +/-1.65 Forecas-based Taylor-ype CPI inflaion, GDP growh rae insrumen rule forecass +/-0.4
Table 5. Insrumen rules esimaion resuls Simple Taylor-Type Insrumen Rule Taylor-Type Insrumen Rule form from RAMSES model Weighs declared in RAMSES (Adolfson e al., 2007b, p.21) Forecas-Based Simple Insrumen Taylor Rule Direc Inflaion Forecas-Based Insrumen Taylor Rule Forecas-Based Insrumen Taylor-ype Rule form from RAMSES model Weighs declared in RAMSES model- forecass based arge variables (Adolfson e al., 2007b, p.21) Coefficien α 0 α i α π α y α π α y 0.25 [0.15] R square 0.91 ε 0.27 Coefficien 0.22 [0.16] R square 0.92 ε 0.27 0.91*** [0.04] 0.92*** [0.04] 0.14*** [0.04] 0.14** [0.05] 0.1*** [0.04] 0.11*** [0.03] α π + 2 i 1 - - - - -0.1 [0.09] -0.1* [0.05] - - Coefficien - - 1.7 0.004 0.3 0.1 - - Coefficien 0.07 [0.12] R square 0.91 ε 0.29 Coefficien 0.07 [0.11] R square 0.9 ε 0.29 Coefficien 0.08 [0.12] R square 0.9 ε 0.29 0.96*** [0.03] 0.91*** [0.03] 0.96*** [0.04] ***Significan a 0.001**Significan a 0.01*Significan a 0.05. Robus sandard errors in parenheses - - - - - - - - - - 0.004 [0.09] -0.04 [0.05] 0.4* [0.17] 0.4* [0.16] 0.4* [0.17] α y -0.01 [0.09] - -0.01 [0.09] - - - - - 0.3 0.1 1.7 0.004 + 2 i 1
Figure 1. Inflaion forecass cenral pahs published in years 1999-2006 by The Cenral Bank of Sweden 3,50 3,00 2,50 2,00 1,50 1,00 0,50 0,00-0,50 01.08.1999 01.01.2000 01.06.2000 01.11.2000 01.04.2001 01.09.2001 01.02.2002 01.07.2002 01.12.2002 01.05.2003 01.10.2003 01.03.2004 01.08.2004 01.01.2005 01.06.2005 01.11.2005 01.04.2006 01.09.2006 01.02.2007 01.07.2007 01.12.2007 01.05.2008 01.10.2008 01.03.2009 01.08.2009 Source: own elaboraion based on he Inflaion Repors published by The Cenral Bank of Sweden beween 1999 2006. Figure 2. The repo rae in Sweden in years 1999 2006 4,50 4,00 3,50 3,00 2,50 2,00 1,50 1,00 0,50 0,00 01.01.1999 01.05.1999 01.09.1999 01.01.2000 01.05.2000 01.09.2000 01.01.2001 01.05.2001 01.09.2001 01.01.2002 01.05.2002 01.09.2002 01.01.2003 01.05.2003 01.09.2003 01.01.2004 01.05.2004 01.09.2004 01.01.2005 01.05.2005 01.09.2005 01.01.2006 01.05.2006 01.09.2006 Source: own elaboraion based on he Inflaion Repors published by The Cenral Bank of Sweden beween 1999 2006.
Figure 3. Inflaion forecass cenral pahs a he wo year prognosic momen 3,00 Repo rae change Inflaion forecas +2 2,50 2,00 1,50 1,00 0,50 0,35 0,50 0,25 0,25 0,25 0,25 0,00-0,50-0,25-0,25-0,25-0,50-0,50-0,50-0,50-1,00 1999.08.262000.06.072001.05.302002.04.252003.03.172004.03.312005.03.142006.04.27 Figure 4. The variaions of formulaions wih differenial responses o inflaion and oupu 3,00 2,00 1,00 0,00-1,00-2,00-3,00-4,00-5,00 01.08.1999 01.12.1999 01.04.2000 01.08.2000 01.12.2000 01.04.2001 01.08.2001 01.12.2001 01.04.2002 01.08.2002 01.12.2002 01.04.2003 01.08.2003 01.12.2003 01.04.2004 01.08.2004 01.12.2004 01.04.2005 01.08.2005 01.12.2005 01.04.2006 01.08.2006 01.12.2006 Simple Taylor-ype insrumen rule Taylor-ype insrumen rule form from RAMSES Wieighs declared in RAMSES i-i-1
Figure 5. The variaions of formulaions wih differenial responses o inflaion and oupu forecass 1,00 0,80 0,60 0,40 0,20 0,00-0,20-0,40-0,60-0,80-1,00 01.08.1999 01.12.1999 01.04.2000 01.08.2000 01.12.2000 01.04.2001 01.08.2001 01.12.2001 01.04.2002 01.08.2002 01.12.2002 01.04.2003 01.08.2003 01.12.2003 01.04.2004 01.08.2004 01.12.2004 01.04.2005 01.08.2005 01.12.2005 01.04.2006 01.08.2006 01.12.2006 i-i-1 DIT Wieighs declared in RAMSES