Estimating a Small Open-Economy Model for Egypt: Spillovers, Inflation Dynamics, and Implications for Monetary Policy

Transcription

1 WP/11/108 Esimaing a Small Open-Economy Model for Egyp: Spillovers, Inflaion Dynamics, and Implicaions for Moneary Policy Elif C. Arbali and Kenji Moriyama

2 2011 Inernaional Moneary Fund WP/ 11/108 IMF Working Paper Middle Eas and Cenral Asia Deparmen Esimaing a Small Open-Economy Model for Egyp: Spillovers, Inflaion Dynamics and Implicaions for Moneary Policy Prepared by Elif C. Arbali and Kenji Moriyama 1 Auhorized for disribuion by Ralph Chami May 2011 Absrac This paper esimaes a small open economy model for Egyp o analyze inflaion, oupu dynamics and moneary policy during The ineres rae channel is found o be relaively weak in Egyp, complicaing he use of ineres raes as he immediae arge of moneary policy. However, he paper also finds a significan level of persisence in he policy rae, making moneary policy pro-cyclical. More acive use of ineres rae policy, measures o improve domesic deb markes and a gradual move owards inflaion argeing can help suppor a successful disinflaion sraegy for Egyp. JEL Classificaion Numbers: E17, E37, E52, F41 Keywords: Small open economy model, moneary policy, ransmission mechanism Auhors Address:earbali@imf.org, kmoriyama@imf.org This Working Paper should no be repored as represening he views of he IMF. The views expressed in his Working Paper are hose of he auhor(s) and do no necessarily represen hose of he IMF or IMF policy. Working Papers describe research in progress by he auhor(s) and are published o elici commens and o furher debae. 1 The auhors are graeful for helpful commens from Alan MacArhur, Hassan Al-Arash, and paricipans in he seminar held a he Cenral Bank of Egyp in January 2011.

3 3 Conens Page I. Inroducion...5 II. Macroeconomic Developmens and Moneary Policy in Egyp...7 III. Descripion of he Model...9 IV. The Daa, The Selecion of Priors, and he Parameer Esimaes for Egyp...10 A. The Daa and Esimaion Mehod...10 B. The Selecion of Priors and Idenificaion...11 C. Esimaed Parameers...12 V. Oupu Gap, Inflaion and Ineres Rae Dynamics, Impulse Responses and Variance Decomposiion...13 A. Esimaed Dynamics of Main Variables: Oupu Gap, Inflaion, and Nominal Ineres Rae Implied by he Esimaed Taylor Rule...13 B. Impulse Responses...14 C. Variance Decomposiion...15 VI. Applicaions of he GPM+: Moneary Policy Implicaions...16 A. Opimal Moneary Policy...16 B. Policy Scenario Simulaions...18 VII. Conclusions and Policy Implicaions...19 Tables 1. Priors and Esimaes for Key Variables Calibraed Parameers a. Roo Mean Squared Errors b. Variance Decomposiion Esimaed Parameers of he Seleced Variables in he GPM+ for Egyp Esimaed and Opimal Taylor Rule...25 Figures 1. Expeced Inflaion and Acual Inflaion Effecive Exchange Rae and Exchange Rae vis-à-vis U.S. Dollar, TB Rae and Inerbank Rae, Oupu Gap Esimaed by he GPM Esimaed Oupu Gap and is Decomposiion by he GPM Esimaed Phillips Curve and Inflaion Decomposiion by GPM Esimaed Taylor Rule and Policy Rae Decomposiion by GPM Pass-hrough Rae of he Exchange Rae o Inflaion Relaive Size of he Sandard Deviaion of Real Ineres Rae o Real Exchange Rae wih Differen Transmission Level Impulse Responses o Foreign Oupu Gap Shock Impulse Responses o Foreign Real Ineres Rae...27

4 4 12. Impulse Responses o Nominal Ineres Rae Impulse Responses of Main Variables o Temporary Demand Shock Impulse Responses of Main Variables o Persisen Demand Shock Simulaed Responses of Main Variables wih Capial Inflows Derived from he Esimaed GPM Simulaed Responses of Main Variables wih Inflaion Targeing from he Esimaed GPM Simulaed Responses of Main Variables o Supply Shocks Derived from he Esimaed GPM Simulaed Responses o Capial Inflows and Delaying Tighening Derived from he Esimaed GPM (Par I) Simulaed Responses o Supply Shocks and Delaying Tighening Derived from he Esimaed GPM (Par II)...35 Appendixes I. Complee Model Equaions For GPM II. Daa Sources...38 III. Esimaion Mehod...39 IV. Shocks Added o he Esimaed GPM+ in he Simulaions...40 References...21

5 5 I. INTRODUCTION Since mid 2005, Egyp, like many oher emerging marke economies, wen hrough a significan economic reform process characerized by increasing real and financial linkages wih he res of he world and he ransiion owards inflaion argeing in he conduc of moneary policy. Sudying he ransmission of shocks in main rading parners economic aciviies and global ineres raes o he Egypian economy has been increasingly imporan from a policy perspecive. In addiion, Egyp s planned ransiion o an inflaion argeing regime implies ha i will be imporan o make inflaion forecass and policy simulaions wihin a consisen macroeconomic framework ha incorporaes he conduc of moneary policy and key moneary ransmission mechanisms ha are esimaed or calibraed for Egyp. Moivaed by he above wo key developmens, his paper incorporaes a small open economy model for Egyp ino he IMF s Global Projecion Model (GPM). This exended model, also called GPM+, is he firs sep in conducing forecass and policy analysis in an esimaed model for Egyp ha is also consisen wih he IMF s global economic projecions. This paper describes he applicaion of GPM+ o Egyp and draws ineresing conclusions regarding spillovers and he ransmission of global shocks and inflaion dynamics, as well as implicaions for moneary policy. The GPM is he main quarerly projecion model used in he research deparmen of he IMF ha encompasses 6 main blocks: Unied Saes, Euro Area, Japan, Emerging Asia, Lain America and oher counries. 2 The GPM is used o sudy he ransmission of counry-specific and global shocks and make projecions ha are based on a common global environmen, aking ino accoun spillovers among regions. I was recenly exended o include a canonical small open economy (GPM+) ha is assumed o ake as given he developmens in he differen regions of GPM and hence can be hough of as a saellie model ha is linked o he bigger GPM model. Alhough here are no feedback effecs from he small open economy o he res of GPM a reasonable assumpion as long as a small open economy is analyzed, his model offers a consisen framework o sudy he effecs of global economic developmens in relaively small open economies such as Egyp. The GPM+ was firs applied o he Indonesian economy as described in Andrle e. al. (2009), which also provides a summary of he greaer GPM projec and fuure planned exensions. From a modeling perspecive, he GPM+ feaures five reduced-form behavioral equaions ha describe movemens in he oupu gap, unemploymen, inflaion, nominal ineres rae and he exchange rae, in addiion o several ideniies ha describe he evoluion of equilibrium ineres rae, oupu and unemploymen. The equaions in he model are comparable o he reduced form versions of Dynamic Sochasic General Equilibrium (DSGE) models developed in he academic lieraure such as in Gali and Monacelli (2005) 3 and adoped in many cenral banks around he world for policy analysis and projecions. 2 See Carabenciov e. al. (2008) for a deailed summary of he evoluion of GPM, Canales-Krijenko e. al. (2009) for is exension o Lain America, and Andrle e al. (2009) for is exension o Indonesia. 3 The reduced form equaions are somewha differen in ha GPM incorporaes some lagged effecs, especially in he Philips Curve, o capure inflaion indexaion and backward looking expecaions.

6 6 However, he fac ha is equaions are expressed in reduced form implies ha i is no possible o use GPM+ o sudy he effecs of srucural changes. Raher, he GPM+ should be inerpreed as a very useful ool o undersand pas developmens, assess he effecs of exernal shocks and make policy simulaions. Cenral banks in Canada, Norway and Chile use GPM ogeher wih oher ypes of models for projecions and policy analysis. Several counry eams a he IMF have also incorporaed GPM+ ino heir forecasing and policy analysis oolbox. Main findings of he paper are summarized as follows: The oupu gap in Egyp has increased during , reflecing a depreciaed exchange rae and a posiive exernal demand. The oupu gap shrank sharply in 2009, wih he global financial crisis as exernal demand declined and moneary policy remained igh. The esimaed gap has been somewha negaive bu no saisically significan since he middle of 2009; Response of he oupu gap in Egyp o he foreign oupu gap is relaively small, which is consisen wih he relaively srong growh performance of Egyp during he recen global crisis. This could reflec he srucure and composiion of Egyp s expors and also he lack of srong financial linkages wih advanced counries; The esimaed Phillips curve indicaes ha supply shocks and expeced inflaion (including inflaion ineria) componens are main deerminans of inflaion; Increased foreign real ineres rae has posiive effec on he oupu gap in Egyp, mainly hrough real depreciaion; Ineres rae channel of moneary policy ransmission mechanism appears weak which is in line wih Masha and Billmeier (2007); Egyp has high ineria in is nominal ineres (policy) rae which causes real ineres raes o be pro-cyclical, magnifying economic flucuaions; Increasing he response of nominal ineres rae o inflaion and he oupu gap is desirable for he CBE o reduce flucuaions in inflaion and he oupu gap; Afer a large devaluaion in 2003, he CBE has anchored inflaion expecaions unil he end of 2007 mainly hrough he exchange rae channel; The paper is organized as follows: he nex secion provides a brief descripion of relevan recen macroeconomic developmens in Egyp. The following secion describes he exension of GPM o a small open economy and in paricular o Egyp. Secion IV discusses he daa, he esimaion sraegy (a Bayesian-like mehod called regularized maximum likelihood) and he parameer esimaes. Secion V presens he esimaed main variables in he model he oupu gap, he Phillips curve, and he nominal ineres rae rule (he Taylor rule) and he esimaed impulse responses, concenraing in paricular on he response o exernal demand shocks o sudy spillovers and he response o domesic ineres rae shocks o sudy he moneary ransmission mechanism. Secion VI conducs some policy simulaions and drives conclusions for he conduc of moneary policy going forward. Conclusion will be briefly presened in he final secion.

7 7 II. MACROECONOMIC DEVELOPMENTS AND MONETARY POLICY IN EGYPT Egyp s long-run growh performance during was srong compared o he average of emerging marke economies, alhough i was well below he high-growh emerging marke economies in Asia (Enders 2007). Growh in Egyp has acceleraed in sep wih he implemenaion of a series of reforms since 2004, including ariff reducion, ax adminisraion and public expendiure managemen, exchange rae unificaion, and business environmen improvemens (e.g., lower fees and processing ime, cus in corporae ax raes and oher ax incenives for foreign direc invesmen). 4 However, high growh raes during his period had no resuled in a subsanial decline in unemploymen rae and povery. The decomposiion of oupu growh (producion side and expendiure side GDP daa) and he growh accouning exercise 5 sugges he following as he main drivers of Egyp s srong growh during FY2004/05 07/08: Srong exernal demand prior o he global financial crisis; Improved compeiiveness of Egyp suppored by he produciviy gains as a resul of a series of srucural reforms since 2004 and an undervalued exchange rae afer he significan devaluaion of he Egypian pound in 2003, 6 and Srong capial (FDI) inflows, due o improved reurn of capial in Egyp and favorable financial condiions in world markes. While growh in Egyp declined afer he Lehman shock o below 5 percen (year-over-year) from an average 7 percen prior o he crisis, Egyp weahered he global financial crisis relaively well compared o oher emerging marke economies wih similar income levels, due, in par, o relaively low credi growh in Egyp prior o he crisis a proxy for increased financial vulnerabiliies during he boom in he lieraure and less sensiiviy of Egyp s expors o global downurns (IMF 2010). Inflaion (12-monh) in Egyp since 2007 appears o have a gradual increasing rend wih several double digi inflaion episodes: for example, he firs spike in 2004 mainly refleced he pass-hrough effec of he huge devaluaion of he Egypian pound in 2003; he second spike in was due o an avian flu oubreak and Figure 1. Expeced Inflaion (Consensus Forecas) and Acual Inflaion (12-monh, percen) Acual inflaion Expeced inflaion in he consensus forecas 4 For deails, see IMF (2007) and Enders (2007). 5 In he growh accouning exercise, capial sock is esimaed following he Perpeual Invenory Mehod, wih 5 percen depreciaion rae and 33 percen of he share of capial. Using differen parameers, however, does no change he main findings of he analysis, where human capial is esimaed following Collins and Bosworh (1996). 6 See IMF (2010).

8 8 world commodiy prices increase; and he hird one (2008) was caused by he increase in world commodiy prices. The expeced inflaion proxied by he inflaion projecions in he Consensus Forecas afer 2007 has been subsanially higher han ha before 2007 (Figure 1), implying ha recen spikes in inflaion migh have been feeding ino a higher expeced inflaion, due parly o rigidiies and disorions in price and wage seings. Since he large devaluaion of he Egypian pound in early 2003, Egyp s exchange rae boh effecive erm and vis-à-vis he U.S. dollar has been sable (Figure 2) Figure 2. Effecive Exchange Rae and Exchange Rae vis-a-vis US Dollar, (Index values: 2005 =100 for NEER, Naional currency per USD) Unil 2005, he CBE s operaional 120 arge for is moneary policy was excess reserves of banks, and 80 growh in M2 was he inermediae 60 arge. Since hen, he CBE began 40 o move oward inflaion argeing (Al-Masha and Billmeier 2007), NEER EGP vis-à-vis USD (RHS, reversed) 6 7 alhough i has no officially announced o adop i. Insiuional measures which have been implemened by he CBE include esablishing he Coordinaing Council on Moneary Policy headed by he Prime Miniser, he Moneary Policy Commiee (MPC), and he Moneary Policy Uni wihin he CBE, and releasing a Moneary Policy Saemen on he CBE s exernal web-sie afer each MPC meeing. The operaional measures include inroducing an ineres rae corridor by he CBE and cenral bank noes as he primary insrumen for liquidiy managemen hrough open marke operaions, and developing models o forecas inflaion. The CBE launched is core inflaion index in Ocober 2009, o capure he more persisen rend of underlying inflaion by separaing he noise and shor-run flucuaions Overnigh inerbank rae on average appears o move more smoohly since 2005, consisen wih he change in he CBE s policy framework, relying more on open marke operaions (Figure 3). Three monh reasury bills (T-bills) rae seems o have racked he overnigh inerbank rae, alhough when he economy faced srong capial inflows, i has ended o decline relaive o he overnigh inerbank rae (for example, Figure 3: TB Rae and Inerbank Rae (Overnigh), (Percen) TB rae (3M) Inerbank rae (ON) 7 See The Core inflaion Measure, a CBE s brochure on he core inflaion.

9 9 and early 2008, and afer July 2010). The laer observaion may imply ha mos capial inflows have so far gone ino shor-erm fixed income securiies. III. DESCRIPTION OF THE MODEL This secion highlighs key componens of he GPM+: oupu gap evoluion (IS curve), real ineres rae pariy wih risk premium, inflaion dynamics (he Phillips curve) and moneary policy (he Taylor rule), while he deails of he model are oulined in Appendix I. The GPM+ a small open economy model is linked o he res of he word hrough hree key exogenous variables: exernal demand capured by he foreign oupu gap, real world ineres rae and real world equilibrium ineres rae. All of hese variables are given as a weighed average of he variables esimaed for he six main blocks in GPM, where Egyp s rade shares are used as weighs. The domesic oupu gap is assumed o have forward and backward looking elemens reflecing habi formaion in he household s expendiure and depend on he gap in he shor-erm ineres rae ( r 1 ), he real effecive exchange rae gap ( z 1 ) 8 and he foreign oupu gap ( y, ). 9 The evoluion of he oupu gap IS curve is given by: f y y 1 y r z y f, 1 The real effecive exchange rae (an increase in he exchange rae means depreciaion) is deermined endogenously via a version of he real ineres pariy condiion wih risk premium, where he fuure expeced exchange rae has boh forward and backward looking componens: e e Z Z 4 * ( Z 1 Z ) ( R R f, ) ( R R f, ), where e Z 1 Z 1 ( 1 ) Z 1. Z Z Change in he real exchange rae depends on he real ineres rae gap differenials and which is inerpreed as he risk premium for financial asses in Egyp. Inflaion dynamics in GPM+ are governed by a sandard hybrid New Keynesian Philips curve equaion ha links curren inflaion o fuure expeced inflaion ( 1 ), he lagged inflaion ( 1 ) capuring boh he indexaion of price seing o pas inflaion and he backward looking componen of inflaion expecaions, he lagged oupu gap ( y 1 ) and he change in he real effecive exchange rae ( Z ) o reflec pass-hrough effecs. 1 1(1 1) 12y13Z, y e 8 The real effecive exchange rae gap is given by, where are he weighs of differen bilaeral exchange raes. These weighs are calculaed using he rade shares of Egyp wih he differen regions ha exis in GPM. Currenly he shares are: Unied Saes (0.14), EU (0.33), Japan (0.05), Emerging Asia (0.13), LA (0.02), ohers (0.33). 9 The foreign oupu gap is a weighed average of he oupu gaps of he small open economy s main rading parners (,, ).

10 10 The change in he real exchange rae is deermined by he change in nominal exchange rae * and he inflaion differenial beween Egyp and is rade parners: Z S. A higher esimae of he parameer for he lag of inflaion ( 1 1 ) implies higher inflaion persisence and will be one of he key parameers of ineres, since when 1 1 is large, inflaion is more persisen and disinflaion requires higher cos of increased unemploymen (i.e., he posiive sacrifice raio). While Egyp currenly does no have an announced inflaion arge, he cenral bank communicaes in is moneary policy announcemens ha i has a comfor zone ha covers he 6 8 percen range. 10 Therefore, he model can include shor erm nominal ineres raes ( I ) given by a moneary policy rule ha incorporaes ineres rae smoohing, and responds Tar o he oupu gap and he deviaion of expeced inflaion from is arge ( ) similar o a Taylor ype rule, where he arge is assumed o follow an exogenous random walk process and is simulaneously esimaed wihin he model. Tar I I 1 I 1 ( 1 1)[ R 43 2 ( 43 3 ) 3 y ] where 4 is he model consisen expeced inflaion derived from he Kalman filer, R is he naural rae of ineres and he policy rae used in he paper is he cenral bank s deposi rae. In GPM+, he cenral bank can affec inflaion hrough wo channels: ineres rae channel and exchange rae channel. The ineres rae channel is capured by he coefficien of he real ineres rae gap in he evoluion of he oupu gap (he IS curve). The exchange rae channel operaes hrough is direc effec on he oupu gap bu also hrough he pass-hrough effec in he Phillips curve. IV. THE DATA, THE SELECTION OF PRIORS, AND THE PARAMETER ESTIMATES FOR EGYPT A. The Daa and Esimaion Mehod The paper uses daa on unemploymen, seasonally adjused real GDP, he overnigh inerbank rae, he nominal and real effecive exchange raes and finally he CPI index (The daa source and availabiliy over he sample are described in more deail in Appendix II). The benchmark sample period is 2005: Q3 hrough 2010: Q2 couning ha he CBE decided o adop he ineres rae corridor as is operaional arge in June 2005 and announced is inenion o move oward an inflaion argeing regime. Change in moneary policy framework suggess a srucural shif in he parameers of he model. Also, focusing on his period should give he addiional benefi of excluding he effecs of he large devaluaion of he pound in early 2003 from he sample. The GPM+ is esimaed using regularized maximum likelihood which is a simple Bayesianlike echnique ha combines he priors of he economerician wih he likelihood of he model based on daa. The relaive weighs of he priors versus he daa can be adjused for 10 This model assumes ha he bank s arge and policy rule is known and ha i is credible.

11 11 differen parameers o reflec he differen levels of confidence ha one may have for he priors. A deailed descripion of he esimaion mehodology is presened in Appendix III. B. The Selecion of Priors and Idenificaion As in many Dynamic Sochasic General Equilibrium (DSGE) models wih a large number of esimaed parameers, idenifying parameers is an imporan ask. The regularized maximum likelihood mehod allows one o specify priors for all esimaed parameers which ensures ha he likelihood is smooh, grealy improving he numerical soluion algorihm. These priors are cerainly imporan for final esimaes, especially when he idenificaion is weak due o he srucure of he model or when he daa is no as informaive abou a cerain se of parameers. In GPM+, he key behavioral equaions are based on gaps (of oupu, unemploymen, real ineres raes ec.), however, in conras o many DSGE models, a sochasic law of moion for equilibrium oupu, unemploymen, real ineres rae, real exchange rae and inflaion (or inflaion arge) are also specified and esimaed wihin GPM+. Alhough his gives an imporan flexibiliy o GPM+, i also requires a more serious consideraion of idenificaion. For example, he parameer ha deermines he effec of he exchange rae gap on he oupu gap canno be idenified if he equilibrium exchange rae (and hence he exchange rae gap) is no pinned down. Therefore, i is necessary o impose igh priors on cerain parameers (or use direc calibraion) o achieve idenificaion. Under sandard mehods for esimaing DSGE models, he daa would be firs filered and hen fed ino he model, effecively idenifying equilibrium values of key variables by choosing an appropriae filering echnique (sor of a hard prior). Our choice of priors, presened in Table 1, is based on previous applicaions of he GPM+ o oher emerging marke counries, he resuls of a previous sudy by Moursi and Mossallamy (2009) which esimaes a DSGE model for Egyp, as well as basic calibraion exercises based on daa for Egyp. For insance, he prior means for he unemploymen block parameers, parameers in he oupu gap equaion and he Philips Curve are mainly adoped from he priors and esimaes in Kriljenko el. al. (2009). We used he esimaes in Moursi and Mossallamy (2009) o guide our priors for he parameers in he moneary policy rule. Some parameers such as he seady-sae inflaion arge and he seady-sae real ineres raes are no esimaed bu simply calibraed using sample averages. The paricular assumpions made abou hese parameers are repored in Table 2. The srucural shocks in he model are assumed o be disribued independenly and have a normal disribuion. They are also assumed o be uncorrelaed wih each oher and across ime. The poserior esimaes of he sandard deviaions of srucural shocks have for he mos par moved subsanially away from he iniial priors owards similar poserior modes under differen iniial priors. We choose small and igh priors for he sandard deviaions of he wo shocks o poenial oupu o reflec our noion of a relaively smooh poenial oupu.

12 12 C. Esimaed Parameers The esimaes of key parameers (priors and poseriors) are presened in Table 1: he evoluion of he oupu gap (IS curve), he Phillips curve, and ineres rae rule (Taylor rule), all of which are criical for he dynamics of inflaion and he oupu gap. Comparison of he IS curve and ineres rae rule beween Egyp and several emerging marke economies, aken from pas sudies using GPM, is presened in Table 4. Firs, he esimaed parameers in he oupu gap evoluion (IS curve) sugges ha he relaive size of he backward looking componen o he forward looking componen is abou 3, implying ha he habi formaion of expendiure (especially, household consumpion) has significan effecs on he oupu gap dynamics. While he relaive weigh of he backward looking componen, he coefficien for real exchange rae and for foreign oupu gap in Egyp are similar o oher counries, he esimaed coefficien for he real ineres rae gap in Egyp is significanly smaller less han one half on average han oher counries. The smaller coefficien for he real ineres rae gap in he IS curve indicaes weaker ineres rae channel in he moneary policy ransmission mechanism, since he cenral bank needs o move is policy rae more acively o achieve required oupu gap change o affec inflaion in he Phillips curve. This observaion is in line wih he findings in Al-Masha and Billmeier (2007). Second, he esimaed Phillips curve suggess ha he forward looking componen is slighly more dominan han he backward looking par in Egyp compared o oher emerging marke economies: he weigh of he forward looking componen is abou wo-hird for Egyp, whereas he median is abou A more forward looking Phillips curve is in fac desirable for moneary policy o achieve is inflaion arge wih a smaller impac on oupu. However; if moneary policy is no very responsive o inflaionary shocks (as in he case of Egyp); a more forward looking Phillips curve can exaggerae he inflaionary impac. A large coefficien for he oupu gap in he Phillips curve for Egyp compared o oher counries appears somewha reflecing larger volailiy of inflaion in Egyp han in oher counries. Third, he esimaed nominal ineres rae rule indicaes ha he coefficien for he lag of nominal ineres rae in Egyp is higher han in oher counries. This implies ha he nominal ineres rae in Egyp has responded less o inflaion han oher counries, a leas in he shorrun, leading he real ineres rae o be pro-cyclical. Therefore, ineres rae policy has no helped sabilize inflaion. Raher, non-ineres rae channel have been mainly used o miigae inflaionary pressures and anchor inflaion expecaions. Table 3a repors he roo mean squared forecas errors for real GDP growh, nominal ineres rae and inflaion. The one year ahead forecas error for annual growh is slighly higher han one percen, whereas for annual inflaion i is slighly higher han 5.5 percen. The model forecas errors for growh and ineres raes are comparable o previous applicaions of GPM (e.g. o Lain America (Kriljenko e. al., 2009) and Indonesia (Andrle e. al., 2009), bu for inflaion i is significanly higher. One facor ha can explain he difference a leas for Indonesia is he fac ha Andrle e. al. (2009) uses he core inflaion measure which implies ha he volailiy of inflaion is probably lower han in he oher sudies. Neverheless, hese

13 13 resuls sugges ha here is room for improving he performance of he model in capuring inflaion dynamics which can be addressed in fuure work. 11 V. OUTPUT GAP, INFLATION AND INTEREST RATE DYNAMICS, IMPULSE RESPONSES AND VARIANCE DECOMPOSITION This secion presens hisorical developmens of he esimaed main macroeconomic variables he oupu gap, inflaion decomposiion, and nominal ineres rae and heir impulse responses o exogenously given shocks: foreign oupu gap shock; foreign ineres rae shock; moneary policy shock; and he pass-hrough effec of he exchange rae o inflaion, based on he esimaed GPM+ in he previous secion. I also repors he variance decomposiion of key variables abased on parameer esimaes and he esimaed variances of srucural shocks in he model. A. Esimaed Dynamics of Main Variables: Oupu Gap, Inflaion, and Nominal Ineres Rae Implied by he Esimaed Taylor Rule The esimaed oupu gap in Egyp has been somewha negaive since he middle of 2009, alhough i is no saisically differen from zero (Figure 4). The posiive oupu gap widened since 2005, driven by he undervalued exchange rae as well as a favorable exernal environmen (Figure 5). Since he ouse of he global crisis, however, he gap declined sharply and urned ino negaive due o he weak exernal demand and he increased real ineres raes. Recen sharp real appreciaion began o negaively affec he oupu gap since he beginning of 2009, in line wih he exchange rae assessmen in IMF (2010, 2011) Figure 4. Oupu Gap Esimaed by he GPM+ (Deviaion from he poenial oupu, percen, confidence inerval = wo sandard deviaion) Oupu gap by GPM Upper limi of Confidence inerval of GPM Lower limi of Confidence inerval of GPM Figure 5. Esimaed Oupu Gap and is Decomposiion by he GPM+ (Deviaion from he poenial oupu, percen) Lagged Oupu Gap Fuure Oupu Gap Real Ineres Rae Gap Foreign Oupu Gap Real Exchange Rae Gap Esimaed Oupu Gap Supply shocks and inflaion ineria have been he main deerminans of inflaion dynamics in Egyp. The esimaed Phillips curve appears on average explaining inflaion developmens in Egyp well, while large supply shocks have played a significan role, especially during he surge in 2008 and 2009 (Figure 6). Inflaion ineria also plays an imporan role and may make he cos of disinflaion bigger in Egyp. Exchange rae appreciaion has somewha helped miigae inflaionary pressures, consisen wih he pass-hrough effec. 11 In fuure work, we would like o incorporae he effecs of adminisered prices and commodiy prices (especially food) ino he model.

14 14 Figure 6. Esimaed Phillips Curve and Inflaion Decomposiion by GPM+ (Qoq inflaion, annualized, percen) Figure 7. Esimaed Taylor Rule and Policy Rae Decomposiion by GPM+ (Discoun Rae, annualized, percen) Lagged Inflaion Oupu Gap Real Exchange Rae Depreciaion Inflaion Prediced Acual Inflaion Naural Rae of Ineres Oupu Gap Deviaion from Inflaion Targe Nominal Ineres Rae Prediced Acual Nominal Ineres Rae Ineria is he main deerminan of he nominal ineres rae, while he oupu gap and inflaion (deviaion from he arge) has played relaively limied role o affec nominal ineres rae (Figure 7). Ineria has prevened moneary policy from responding imely o inflaionary pressures: for example, he responses of moneary policy o a spike in commodiy prices in early 2008 and he global financial crisis were even more delayed han ha implied by he esimaed Taylor rule, which already incorporaes a significan level of persisence. B. Impulse Responses Figure 10 plos he impulse responses of main variables o a foreign oupu gap shock, inending o capure spillovers of exernal shocks, in paricular, shocks in Europe which is Egyp s main rading parner. Increased exernal demand widens he oupu gap in Egyp. Despie he posiive oupu gap, curren and fuure expeced exchange rae appreciaion makes inflaion negaive during he firs couple of periods by he pass-hrough effec. Nominal ineres rae gradually increases and his is peak abou 7 periods afer he shock, reflecing he posiive oupu gap and increased inflaion afer period 2. The ineria of he nominal ineres rae, however, makes he real ineres rae negaive in he shor-run (from period 2 o 5) bu posiive in he medium erm. While he negaive ineres rae makes moneary policy pro-cyclical, he laer posiive real ineres rae increases he size of appreciaion of he exchange rae, since he real ineres rae pariy indicaes ha he curren real exchange rae is given as he presen value of he real ineres rae gap differenials beween Egyp and is rade parners. Impulse responses o a change in foreign ineres rae inend o capure ineres rae shocks in advanced counries (Figure 11). Since he ineres rae pariy depends on he foreign ineres rae gap, a foreign real ineres rae increase should have impac similar o a decline in he equilibrium foreign real ineres rae which should mainly be affeced by oupu growh, fiscal balance and public deb in he advanced economies. In he conex of he curren economic condiions in he advanced economies, here are wo poenial ineres rae shocks: addiional moneary policy simulus like addiional quaniaive easing and increased equilibrium real ineres rae due o deerioraed fiscal posiion. A posiive shock in foreign real ineres rae would depreciae he Egypian pound. The depreciaed exchange rae increases he oupu gap and inflaion hrough he pass-hrough effec. While nominal ineres rae reacs o he increased oupu gap and inflaion, is ineria

15 15 makes he real ineres rae negaive during he firs couple of periods afer he shock and moneary policy pro-cyclical, furher increasing he oupu gap and inflaion. Impulse responses o a posiive moneary policy shock are highlighed in Figure 12. One sandard deviaion increase in nominal ineres rae (abou one percen) would appreciae he exchange rae of he Egypian pound by 2 3 percenage poins, and creae negaive oupu gap of abou 0.4 percenage poins in he shor run. Exchange rae appreciaion and negaive oupu gap would cause a decline in he price level (negaive inflaion) which lass nearly 10 periods. Negaive inflaion would increase real ineres rae and cause furher exchange rae appreciaion and decline in oupu gap. The pass-hrough effec of he exchange rae on inflaion is given by he impulse response of inflaion o a one sandard deviaion shock o he equilibrium level of real exchange rae, which also implies a shock o he level of he nominal exchange rae assuming no change in foreign inflaion. The esimaed pass-hrough in Egyp is relaively low compared o esimaes for oher EMEs. The esimaed elasiciy is 0.23 a is peak wihin one year (Figure 8), consisen wih he resuls by Rabanal (2005) which follows he mehod in McCarhy (1999) and found ha he pass-hrough is somewhere beween 6 27 percen depending on he model specificaion. Rabanal (2005) suggess ha he low passhrough could be due o he high share of adminisered prices in Egyp s CPI index. C. Variance Decomposiion Figure 8. Pass-hrough Rae of he Exchange Rae o Inflaion (Elasiciy, percen) (Periods, quarerly) We compue he variance decomposiion for key real variables o assess he relaive conribuion of differen shocks (Table 3b). According o he model esimaes, variaion in oupu growh in Egyp is driven mainly by demand shocks (26 percen), moneary policy shocks (27 percen) and shocks o he equilibrium real ineres rae (28 percen). This cerainly reflecs he esimaed variances of he srucural shocks in he model. Shocks o he nominal ineres rae (surprise policy shocks) and he equilibrium real ineres rae are large; reflecing o some exen differences beween model implied ineres rae pah, based on he esimaed Taylor rule, and he acual pah of he policy rae. Alhough he ineres rae channel is weak, he magniude of he policy rae shocks lead o he allocaion of a significan par of oupu growh variaion o ineres rae shocks. This is also apparen during , when he model based Taylor rule calls for a faser reducion in policy raes han wha acually ook place, suggesing a posiive (i.e. pro-cyclical) policy rae shock, reinforcing he decline in oupu growh. Moneary policy shocks also explain a significan share of variaions in real ineres rae, real exchange rae appreciaion and he real exchange rae gap. Exernal shocks (foreign real ineres raes and foreign demand shocks) have a relaively

16 16 more significan conribuion o real exchange rae movemens and only explain 3 4 percen of he variaion in oupu growh or he oupu gap. VI. APPLICATIONS OF THE GPM+: MONETARY POLICY IMPLICATIONS This secion presens an applicaion of GPM+ o moneary policy analyses. 12 In paricular, given he esimaed parameers of GPM+, he secion firs compues he opimal moneary policy (nominal ineres rae) responses o shocks, where opimal policy reacion means minimizing he loss funcion (presen value of he weighed sum of he variance of inflaion and he oupu gap), and hen simulaes he pah of main macroeconomic variables wih differen moneary policy reacion funcions o he shocks added o he sysem. While, as menioned in he inroducion, he IS curve and he Phillips curve in he GPM are reduced form insead of derived from opimizaion, he specificaion of he nominal ineres rae rule (he Taylor rule) is a ypical one in he lieraure of DSGE models and is independen from he opimizaion of households and firms. Therefore, sensiiviy analysis of he parameers in he Taylor rule should no affec he parameers in he IS curve and he Phillips curve. A. Opimal Moneary Policy There are wo ypes of opimal moneary policy, boh of which minimize he objecive funcion defined as he weighed average of he quadraic form of inflaion and he oupu gap subjec o he esimaed model 13 : The Ramsey opimal policy (he social planner s soluion): he policy should minimize he presen value of he weighed average of he quadraic form of inflaion and he oupu gap, given he esimaed behavioral equaions oher han he Taylor rule in he GPM+; and The opimal policy rule: he rule should minimize he weighed average of he variance of inflaion and he oupu gap, given all of he esimaed parameers oher han coefficiens for inflaion and he oupu gap in he Taylor rule in he GPM+. By is consrucion, he opimal moneary policy rule is a special case of he Ramsey opimal policy wih furher assumpions: (i) he discoun facor is one (i.e., no discoun rae), and (ii) he social planner mus change ineres rae following he Taylor rule. Alhough he opimal policy is a heoreical concep and hus i is no realisic ha he cenral bank can follow i in he real world, he opimal policy analysis sill gives useful insighs on how he cenral bank should reac o shocks added o an economy. The esimaed Ramsey opimal policy indicaes ha he cenral bank needs o ake quick and fron-loaded responses o shocks. Panels 4 and 5 highligh he impulse responses of main variables wih he Ramsey opimal policy agains emporary and persisen demand shocks, 12 In his secion, GPM+ is rewrien so ha all variables are defined as deviaion from he seady sae because of compuaional capaciy of Dynare. 13 See Gali (2008) for an idea of he opimal moneary policy.

17 17 where he discoun facor is 0.98 (8 percen of he annual discoun rae) and he weigh of he oupu gap is 6 in he objecive funcion, where he model consisen variance of inflaion is nine imes as large as ha of he oupu gap. The Ramsey policy raises ineres rae quickly and significanly and hus makes he exchange rae appreciae (negaive sign indicaes appreciaion in he GPM+) righ afer he shock in order o compleely remove he inflaionary demand pressures, sharp conras o he responses of he esimaed Taylor rule. The opimal policy rule also requess much larger response of ineres rae o inflaion and he oupu gap han he esimaed Taylor rule. Table 5 presens he esimaed opimal moneary policy rule (response of nominal ineres rae o inflaion and he oupu gap) wih differen weighs for he oupu gap in he objecive funcion. As expeced, he response of nominal ineres rae wih he opimal rule is subsanially larger han ha implied by he esimaed Taylor rule, similar o he Ramsey opimal policy The opimal moneary policy rule could provide anoher insigh on he allocaion of moneary policy ransmission over he ineres rae and he exchange rae channels. If he cenral bank follows he opimal policy rule, such policy should efficienly combine he ineres rae and he exchange rae channels o minimize he weighed sum of he variance of inflaion and he oupu gap, given he esimaed behavioral equaions in he GPM+. For example, if he increased ineres rae can reduce inflaion hrough he smaller oupu gap more han hrough he pass-hrough effec and he price effec of he exernal demand by exchange rae appreciaion, he cenral bank should use he ineres rae channel more acively, implying more volailiy in he ineres rae han he exchange rae. Figure 9 plo he relaive size of he sandard deviaion of he real ineres rae o he real exchange rae wih differen coefficiens of he real ineres rae in he IS curve, inending o capure he effeciveness of he ineres rae channel ( wo in he horizonal axis in he figures indicaes ha he coefficien is wice as large as he esimaed one). Sronger ineres rae channel proxied by he larger coefficien for he real ineres rae in he IS curve should make he Figure 9. Relaive Size of he Sandard Deviaion of Real Ineres Rae o Real Excange Rae wihdifferen Transmission Level (Proxied by he coefficien of Real Ineres Rae in IS curve, Based on he Esimaed GPM) Esimaed Taylor rule Opimal rule wih coefficien 2 for he oupu gap Opimal rule wih coefficien 9 for he oupu gap (Level of ransimission compared o he esimaed GPM, raio) cenral bank use he ineres rae channel more exensively han he exchange rae, leading o he larger volailiy of he real ineres rae relaive o he real exchange rae. The relaive size of volailiy wih he opimal policies is subsanially larger han ha wih he esimaed Taylor rule for all range of he coefficien of real ineres rae (a proxy for ineres rae channel), suggesing ha he CBE has used he exchange rae channel o anchor inflaion expecaion more acively han he ineres rae channel and herefore here is much room for he CBE o use he ineres rae channel in he fuure.

18 18 B. Policy Scenario Simulaions This subsecion quaniaively invesigaes he impac of shocks o main macroeconomic variables wih he esimaed Taylor rule and he opimal ineres rae rule (wih coefficien 6 for he volailiy of he oupu gap derived in he previous subsecion), where all variables are re-defined as deviaion from he seady sae. The shocks added o he model economy are summarized in Appendix IV. Capial inflows shock (emporary), proxied by combinaion of a posiive demand shock and a decline in he risk premium in he real ineres rae pariy (Figure 15). Increased capial inflows would resul in real appreciaion wih he esimaed Taylor rule. If he cenral bank follows policies resising o nominal appreciaion o avoid furher deerioraion in compeiiveness, such policies would resul in higher inflaion. This observaion is consisen wih he hypohesis derived in he previous secion: because of he weak ransmission mechanism of he ineres rae channel, he cenral bank needs o rely more on he exchange rae channel o conain inflaionary pressures caused by increased capial inflows. In his conex, moneary policy response o increased capial inflows wihou allowing exchange rae appreciaion would send a very conradicing message o he privae secor. Adoping inflaion argeing moneary policy framework Adoping inflaion argeing appears o help anchor inflaion expecaions, where he cenral bank announces he adopion of inflaion argeing in period 1, assuming ha is policy commimen is credible, where he pah of he assumed inflaion arge is presened in he righ op char in Figure 16. Wih inflaion argeing, inflaion declines o he arge prior o he changes in he arge level, suggesing ha inflaion expecaions are well anchored in he forward looking moneary policy framework (Figure 16). The real ineres rae is negaive during he firs couple of periods o creae exchange rae appreciaion expecaion. The real exchange rae depreciaes firs before i appreciaes. Similar sory holds for he periods prior o he reducion in he arge inflaion (period 25 30). This observaion is also consisen wih he hypohesis ha he cenral bank uses exchange rae channel more acively o conain inflaionary pressures, reflecing he weak ineres rae channel of moneary policy ransmission. Supply shocks o he Phillips curve Figure 17 highlighs he responses of main variables o supply shocks. The cenral bank sill relies on he exchange rae channel more han he ineres rae channel, as indicaed by he iniial depreciaion wih he esimaed Taylor rule wih shocks (green line). On he oher hand, real exchange rae does no depreciae firs wih he opimal policy rule wih inflaion argeing, reflecing is larger response of ineres rae o inflaion. The real ineres rae wih he opimal policy rule urns o posiive afer period wo, responding o he increased inflaion. The posiive real ineres rae makes he oupu gap and he need for he cenral bank o use he exchange rae channel o reduce inflaion smaller han in oher scenarios.

19 19 This observaion implies ha increased response of ineres rae o inflaion and oupu gap i.e., more counercyclical moneary policy helps he cenral bank reduce is reliance on he exchange rae channel o anchor inflaion expecaions. Delay in ighening moneary policy o capial inflows and supply Panels 9 and 10 highligh he effecs of delaying ighening moneary condiion when an economy faces capial inflows (Figure 18) or supply shocks (Figure 19). The panels clearly show he coss of delaying moneary policy reacion o inflaionary pressures. Inuiively speaking, since he cenral bank needs o creae more appreciaion expecaion o conain addiional inflaionary pressures due o he delay, exchange rae needs o depreciae more, leading o higher inflaion hrough he pass-hrough effec, he price effec of foreign demand, and he negaive real ineres rae. Such mechanism is sronger wih supply shocks in which he cenral bank canno exploi rae appreciaion pressures as in a capial inflows case. Similar o he supply shocks scenario, he spike in inflaion wih he opimal policy rule is smaller han he esimaed Taylor rule, reflecing less required appreciaion o conain he addiional inflaionary pressures. This is mirror image of more acive response of ineres rae o inflaionary pressures, conribuing o anchor inflaion expecaions less relying on he exchange rae channel. VII. CONCLUSIONS AND POLICY IMPLICATIONS This paper ried o invesigae spillovers, he oupu gap and inflaion dynamics, and implicaions for moneary policy in Egyp based on he esimaed small macroeconomic model. Firs, he paper applied he exended version of he Global Projecion Model (GPM) developed by he Research Deparmen a he IMF and hen esimaed he developmens of he oupu gap, and he decomposiion of he Phillips curve and nominal ineres rae based on he esimaed model. Second, he paper presened he responses of he model economy o exogenously given shocks such as he foreign oupu gap shock, foreign real ineres rae shock, and nominal ineres shocks by he cenral bank. The firs wo impulse responses should help capure he spillovers of he shocks in foreign counries, in paricular he advanced counries. Third, he paper conduced simulaions based on he esimaed model o sudy economy s response o shocks under differen policy frameworks and in paricular, he paper ries o highligh he impac of he adopion of inflaion argeing framework. The main findings of he paper are summarized as follows: Afer posiive oupu gap during , reflecing a depreciaed exchange rae and a posiive exernal demand, he gap shrank sharply in 2009, wih he global financial crisis. The esimaed gap has been somewha negaive bu no saisically significan since he middle of 2009;

20 20 The esimaed Phillips curve indicaes ha supply shocks and expeced inflaion (including inflaion ineria) componens are main deerminans of inflaion; Increased foreign real ineres rae has posiive effec on he oupu gap in Egyp, mainly hrough real depreciaion; Ineres rae channel of moneary policy ransmission mechanism appears weak; Egyp has high ineria in is nominal ineres (policy) rae which causes real ineres raes o be pro-cyclical, magnifying economic flucuaions; Increasing he response of nominal ineres rae o inflaion and he oupu gap is desirable for he CBE o reduce flucuaions in inflaion and he oupu gap; Needless o say, here is much room in he paper o be improved furher. For example, he paper focuses only on he period from Q o Q2 2010, afer he implemenaion of new moneary policy framework, suggesing ha he esimaed model may be disored by he global financial crisis and he policy responses o i, since GPM+ does no include fiscal and financial secors. Analysis based on he more deailed model wih addiional sample period afer Q is a very promising opic for fuure sudy.

21 21 REFERENCES Al-Masha, Rania and Andreas Billmeier (2007): Moneary Transmission Mechanism in Egyp, IMF Working Paper 07/285. Andrle. Michal, Charles Freedman, Rovero Garcia-Salos, Danny Hermawan, Douglas Laxon and Haris Munandar (2009): Adding Indonesia o he Global Projecion Model, IMF Working Paper 09/253. Canales-Kriljenko, J., C. Freedman, R. Garcia-Salos, M. Johnson, and D. Laxon (2009): Adding Lain America o he Global Projecion Model, IMF Working Paper 09/85. Carabenciov, I., I. Ermolaev, C. Freedman, M. Juillard, O. Kamenik, D. Korshunov and D. Laxon (2008a): A Small Quarerly Projecion Model of he U.S. Economy, IMF Working Paper No. 08/278. Carabenciov, I., I. Ermolaev, C. Freedman, M. Juillard, O. Kamenik, D. Korshunov, D. Laxon, and J. Laxon (2008b): A Small Quarerly Muli-Counry Projecion Model, IMF Working Paper No. 08/279. Carabenciov, I., I. Ermolaev, C. Freedman, M. Juillard, O. Kamenik, D. Korshunov, D. Laxon, and J. Laxon (2008c): A Small Quarerly Muli-Counry Projecion Model wih Financial-Real Linkages and Oil Prices, IMF Working Paper No. 08/280. Collins, S. and B. Bosworh (1996): Economic Growh in Eas Asia: Accumulaion versus Assimilaion, Brookings Papers on Economic Aciviy 2: Enders, Klaus-Sefan (2007): Egyp Searching for Binding Consrains on Growh, IMF Working Paper No. 07/57. Gali, Jordi (2008): Moneary Policy, Inflaion, and he Business Cycle, Princeon Universiy Press, Princeon, New Jersey. Gali, Jordi and Tommaso Monacelli (2005): Moneary Policy and Exchange Rae Volailiy in a Small Open Economy, Review of Economic Sudies, 72, Inernaional Moneary Fund (2007): Arab Republic of Egyp, 2007 Aricle IV Consulaion Saff Repor, Washingon, D.C. Inernaional Moneary Fund (2010): Arab Republic of Egyp, 2010 Aricle IV Consulaion Saff Repor, Washingon, D.C. Inernaional Moneary Fund (2011, forhcoming): Arab Republic of Egyp, 2011 Aricle IV Consulaion Saff Repor, Washingon, D.C.

22 22 McCarhy, J., (1999): Pass-Through of Exchange Rae and Impor Prices o Domesic Inflaion in Some Indusrialized Economies, BIS Working Paper No. 79. Moursi, Tarek Abdelfaah and Mai El Mossallamy (2009): Moneary Policy Analysis wih New Keynesian Models for Egyp, Mimeograph, IDSC. Rabanal, Pau (2005): Exchange Rae Pass-Through, Chaper I in Seleced Issues for Egyp, IMF Counry Repor 05/179.

23 23 Table 1. Priors and Esimaes for Key Variables Parameer Equaion: Descripion Prior Poserior Mode Dispersion Mode Dispersion alpha1 Unemploymen gap: lag of unemploymen gap alpha2 Unemploymen gap: oupu gap bea1 Oupu gap: lag of oupu gap bea2 Oupu gap: lead of oupu gap bea3 Oupu gap: real ineres rae gap lagged bea4 Oupu gap: real exchange rae gap lagged bea5 Oupu gap: real foreign oupu gap lagged gamma1 Nominal Ineres Rae: lag of nominal ineres rae gamma2 Nominal Ineres Rae: deviaion of inflaion from arge gamma3 Nominal Ineres Rae: oupu gap lambda1 Inflaion: expeced inflaion lambda2 Inflaion: lag of oupu gap lambda3 Inflaion: real exchange rae depreciaion phi Expeced real exchange rae: weigh of forward looking componen rho Equilibrium Real Ineres Rae: seady sae real ineres rae au Growh Rae of Equilibrium Oupu: seady sae growh rae beaf Foreign oupu gap: lag of foreign oupu gap rhobarf Real equilibrium ineres rae (foreign): seady sae real ineres rae rhof Real ineres rae (foreign): seady sae real ineres rae pief_ss Seady sae foreign inflaion (level) rhopief Foreign Inflaion: seady sae foreign inflaion growh_ss Seady sae growh rae of poenial oupu Sandard Deviaion of Shocks sd_res_lgdp_bar Level shock o poenial oupu sd_res_y Shock o oupu gap sd_res_g Growh shock o poenial oupu sd_res_unr_gap Shock o unemploymen gap sd_res_unr_bar Shock o equilibrium level of unemploymen sd_res_pie Shock o inflaion sd_res_rr_bar Shock o real equilibirum ineres rae sd_res_rs Shock o nominal ineres rae sd_res_pietar Shock o inflaion arge sd_res_rr_diff Shock o uncovered ineres pariy (risk premium shock) sd_res_lz_bar Shock o equilibirum level of real exchange rae sd_res_yf Shock o foregin oupu gap sd_res_rrf_bar Shock o foreign equilibrium real ineres rae sd_res_rrf Shock o foreign real ineres rae sd_res_piestr Shock o foreign inflaion rae