Oil Prices, Monetary Policy, and Counterfactual Experiments

Transcription

1 w o r k i n g p a p e r 05 0 Oil Prices, Moneary Policy, and Counerfacual Experimens by Charles T. Carlsrom and Timohy S. Fuers FEDERAL RESERVE BANK OF CLEVELAND

2 Working papers of he Federal Reserve Bank of Cleveland are preliminary maerials circulaed o simulae discussion and criical commen on research in progress. They may no have been subjec o he formal ediorial review accorded official Federal Reserve Bank of Cleveland publicaions. The views saed herein are hose of he auhors and are no necessarily hose of he Federal Reserve Bank of Cleveland or of he Board of Governors of he Federal Reserve Sysem. Working papers are now available elecronically hrough he Cleveland Fed s sie on he World Wide Web:

3 Working Paper 05-0 Ocober 2005 Oil Prices, Moneary Policy, and Counerfacual Experimens By Charles T. Carlsrom and Timohy S. Fuers Recessions are associaed wih boh rising oil prices and increases in he federal funds rae. Are recessions caused by he spikes in oil prices or by he sharp ighening of moneary policy? This paper discusses he difficulies in disenangling hese wo effecs. Charles T. Carlsrom is a he Federal Reserve Bank of Cleveland and may be reached a Charles.T.Carlsrom@clev.frb.org or (26) Timohy S. Fuers is a Bowling Green Sae Universiy.

4

5 . Inroducion. Oil price increases have preceded every recession since 97. Each of hese recessions has also been preceded by an increase in he federal funds rae. (See Figure.) Are hese recessions caused by he spikes in oil prices or by a sharp ighening of moneary policy? How can we disenangle hese wo effecs? Wha does i mean o disenangle he wo effecs? Bernanke, Gerler, and Wason (997, 2004), hereafer BGW, ried o answer hese quesions empirically using a VAR analysis. Using Hamilon s (996) measure of oil price shocks, BGW (2004) repor ha a 0% oil price increase is associaed wih a 50 basis poin increase in he funds rae and a peak oupu decline of 0.7%. Presumably his funds rae behavior reflecs he endogenous ighening of policy in response o such an inflaionary shock. BGW use his VAR analysis o answer he following counerfacual quesion: How much would oupu have declined if he funds rae had remained consan for, say, four quarers in he wake of he oil shock? BGW answer his quesion by adding unexpeced moneary policy innovaions o he VAR analysis of he exac magniude needed o keep he funds rae sable in he wake of an oil shock. Because his counerfacual experimen is relaed o previous work by oher auhors (Sims and Zha 996), BGW call i he Sims-Zha experimen. The resul of BGW s (2004) Sims-Zha experimen was ha if he Fed had kep he funds rae consan, oupu would have fallen by only abou half of is acual decline. BGW hus conclude ha he endogenous ighening of moneary policy accouned for a subsanial porion of he negaive impac of oil shocks on he economy.

6 One poenial problem wih BGW s Sims-Zha experimen is he Lucas criique: is he VAR sable under such changes in moneary policy? 2 BGW asser ha i seems plausible o us ha a purely ransiory deviaion from he usual policy rule would no significanly affec he srucure of he economy (ha is, he quaniaive effec of he Lucas criique should be small). This paper s firs conribuion is o use he sandard New-Keynesian model o assess he quaniaive relevance of he Lucas criique. Wha would happen if hese unanicipaed policy shocks were acually anicipaed? In conras o he asserion of BGW, wihin his model he Lucas criique problem is quie severe. In paricular, if he Sims-Zha experimen were anicipaed by he public, hen oupu would acually increase in he wake of an oil shock. Hence, an anicipaed version of Sims-Zha experimen would lead one o conclude ha oil price shocks acually increase oupu. This conclusion brings up a broader quesion. Wha does i mean o keep policy consan? The above resuls sugges ha BGW s experimen may no really be holding policy consan. The paper s second conribuion is herefore o expand his counerfacual quesion more broadly. To isolae he impac of an oil shock we need o ask wha effec would oil have on he economy if moneary policy were consan or neural. Bu wha exacly does neural mean? In addiion o he Sims-Zha and anicipaed Sims-Zha experimens, we consider several oher alernaives including a money growh peg, an ineres rae peg, and a Wicksellian ineres rae policy. The laer is a policy ha adjuss he funds rae so ha he real economy behaves as if here There are, of course, numerous sudies ha analyze he effec of moneary shocks and oil shocks in isolaion. See BGW (997) for references. 2 The heoreical relevance of he criique is noed by BGW (997, 2004) and Sims and Zha (996). 2

7 were no nominal rigidiies. 3 The behavior of oupu and inflaion are quie differen under all of hese possible versions of neural policy. Leduc and Sill (2004) conduc a relaed analysis of sysemaic moneary policy and oil price shocks. There are several relevan differences beween heir work and he curren paper. Firs, hey do no consider he Sims-Zha experimen conduced by BGW, nor he quaniaive significance of he Lucas Criique. Insead, heir focus is enirely on sysemaic moneary policy. Second, while our analysis is buil around a model wih nominal rigidiies, heir principle focus is on a flexible price model wih a limied paricipaion consrain. They do repor resuls wih small nominal rigidiies, an average conrac duraion of one quarer. We assume a longer conrac duraion, and also assume ha capial is immobile across firms. Taken ogeher, he nominal rigidiies in our model are sixy imes larger han in Leduc and Sill (2004). Third, Leduc and Sill (2004) also consider he case of an ineres rae peg, bu deal wih he equilibrium deerminacy problem in a way quie differen han our approach, and his difference leads o quaniaively differen conclusions. Finally hey never consider a Wicksellian moneary policy. The nex secion oulines he basic model and clarifies he naure of he moneary experimens. Secion 3 and 4 presen he principle resuls. Secion 5 concludes. 2. The Model. The heoreical model is a marriage of he now-sandard Dynamic New Keynesian models of moneary policy (eg. Woodord (2003), Walsh (2003)), and he earlier real 3 The use of he modifier Wicksellian is suggesed by Woodford (2003). 3

8 business cycle models ha explicily included oil prices (eg., Kim and Loungani (992)). In his secion, we will skech he basic framework. The heoreical model consiss of households and firms. We presen he decision problems of each in urn. Households. Households are infiniely lived, discouning he fuure a rae β. Their period-byperiod uiliy funcion is given by ε γ σ ε γ σ ) / ( ),, ( P M L C P M L C U, () where σ > 0, γ > 0, ε > 0, L denoes labor, P M + denoes real cash balances ha can faciliae ime- ransacions and C denoes he consumpion aggregaor where ( ) ( ) 0 = p p p p p b dj j C b C θ θ θ θ θ. (2) The household begins period wih M cash balances and B - one-period nominal bonds ha pay R - gross ineres. Wih w denoing he real wage, P he price level, and X he ime- moneary injecion, he household s ineremporal budge consrain is given by X Pw L B R M M B C P (3) The household s porfolio choice is given by R R C P M ) / ( = + σ ε (4) / + + = C E R C π β σ σ. (5) From equaion (4), we have ha he money demand curve is given by 4

9 M P + ( σ / ε ) R ) (/ ε ) = C ( R, (6) so ha he ransacions elasiciy is (σ/ε) and he ineres elasiciy is /ε. Following Erceg, Henderson, and Levin (2000), we assume ha households are monopolisic suppliers of labor and ha firms employ a CES aggregaor of household labor wih an elasiciy of subsiuion equal o θ w >. In paricular, he labor aggregaor is symmeric wih (2): L d θ w = d 0 ( L ( j) ) θ w θ w dj θ w θ w. (7) Nominal wages are adjused as in Calvo (983). In his case labor supply behavior is given by C L = Zh W. σ γ For a given level of Zh, he Frisch labor supply elasiciy is /γ. The variable Zh is he monopoly disorion as i measures how far he household s marginal rae of subsiuion is from he real wage. In he case of perfecly flexible bu monopolisic wages, Zh = Zh is consan and less han uniy. The smaller is Zh, he greaer is he monopoly power. In he case of sicky nominal wages, Zh is variable and moves in response o he real and nominal shocks hiing he economy. Erceg e al. (2000) demonsrae ha in log deviaions nominal wage adjusmen is given by: π, W W W = λ zh + βeπ + 5

10 π W where is ime- ne nominal wage growh, zh denoes he log deviaion from seadysae, and λ w ( ηw)( ηwβ ), wih η w denoing he fracion of households ha η ( + γθ ) w w canno adjus heir nominal wages in he curren quarer. 4 Firms. There are a coninuum of firms producing differen varieies of consumpion goods (see he consumpion aggregaor (2)). The ypical firm uilizes labor services,, L from households, and energy, En, from exernal sources o produce is unique final good using he CES echnology: α α ρ ρ ( ) [( )( ) ( ) ] a K L + a ρ Y = f ( K, L, En) En. The ypical firm has a fixed and immobile level of capial given by K =. Labor inpu and energy is perfecly mobile across firms. The real energy price is equal o e P so ha a firm s nominal profis are given by e profis = P Y w L P En ). ( The firm is a monopolisic producer of hese goods, implying ha labor will be paid below is marginal produc. Le Z denoe marginal cos so ha we have w = Z f L () e P = Z f E (). The variable Z is he monopoly disorion as i measures how far he firm s marginal producs differ from he real facor prices. In he case of perfecly flexible bu 4 See page 224 of Woodford (2003) for deails. 6

11 monopolisic prices, Z = Z is consan and less han uniy. The smaller is Z, he greaer is he monopoly power. In he case of sicky prices, Z is variable and moves in response o he real and nominal shocks hiing he economy. Yun (996) demonsraes ha in log deviaions nominal price adjusmen is given by: π = λz + βeπ + whereπ is ime- nominal price growh (as a deviaion from seady-sae nominal price growh) lower case z denoes he log deviaion from seady-sae, and ( η p )( η pβ ) λ, wih η p denoing he fracion of firms ha canno adjus heir η ( + ϖθ ) p p nominal prices in he curren quarer, and ϖ denoing he firm s elasiciy of marginal cos wih respec o firm-level oupu. 5,6 Moneary policy. For our firs hree experimens (baseline, unanicipaed and anicipaed Sims-Zha) we use a Taylor-ype ineres rae rule of he form: i = τπ + τ y + η. y 5 See page 224 of Woodford (2003) for deails. 6 Le V(Y,w,P e ) denoe he firm s cos funcion where he capial sock is fixed a uniy. ϖ is he elasiciy of V Y wih respec o Y, for a fixed level of wages and energy prices. 7

12 where y denoes log deviaions in real oupu, π is he linear deviaion of inflaion from is seady sae, and η is an i.i.d. policy shock. Equilibrium. There are four markes in his heoreical model: he labor marke, he goods marke, he bond marke, and he money marke. The respecive marke-clearing condiions include: C e = Y P En and B = 0. The money marke clears wih he household holding he per capia money supply ineremporally. Calibraion. Before proceeding wih he analysis, we need o se parameer values a levels consisen wih empirical esimaes for a quarerly model. Preference parameers are given by β =.99 (implying a 4% annual seady-sae real rae of reurn), σ = 2, and γ = 3. The laer values are consisen wih micro evidence of fairly inelasic savings and labor supply behavior. We se ε = 2 implying a uni ransacions elasiciy for money demand and an ineres elasiciy of We se θ w = 8 implying a seady-sae mark-up of wages of 4%, andη = 0.5 implying ha wages are fixed on average for wo quarers. These w choices imply λ w = As for firms, he elasiciy of subsiuion beween oil and he capial-labor inpu is equal o /ρ. Consisen wih empirical esimaes, we se his elasiciy o.59, or ρ =.7 (Kim and Loungani (992)). The share parameer a is se o a =.02. This implies a share of energy in oal oupu of 4%. The capial parameer in he producion funcion is 8

13 se o α = /3. We se θ p = 8 implying a seady-sae mark-up of 4%, and η p = 0.5 implying ha firms re-se prices on average every wo quarers. These choices imply ϖ = 0.46 and λ p = Noice ha he assumpion of capial immobiliy leads o a relaively small value for λ p. Leduc and Sill (2004) model he nominal rigidiies wih a convex adjusmen cos o nominal prices and wages. Since all firms face symmeric adjusmen coss, he issue of capial mobiliy is irrelevan as here is essenially a represenaive firm. Leduc and Sill repor ha for echnical reasons (see heir foonoe 5), hey calibrae heir model o imply very frequen price and wage adjusmen, an average duraion of.4 quarers for prices,.0 quarers for wages. This shor conrac duraion as well as he represenaive firm implies very lile nominal rigidiy. For example, in our Calvo environmen, he Leduc and Sill (2004) calibraion corresponds o a price adjusmen parameer of λ p = 6.2, 60 imes larger han our calibraed value of λ p =.07. Their value of λ p = 6.2 is quie close quaniaively o a flexible price model. The logged real price of oil is given by an exogenous AR(2) process: p e e e = a p + a2 p 2 + ν. e e P Where p ln( = ) and P e is he mean real price of oil since 974. Esimaing his P e process since 974 yields a =.2 and a 2 = -.5. In all he experimens below we repor impulse response funcions for a one-ime, exogenous 0% increase in he price of oil (ν =.0). 7 This calibraion is only relevan for he consan money growh experimen. 9

14 The calibraion of he Taylor rule comes from Kozicki (2002) who suggess ha since 983 he coefficiens in his moneary policy rule are τ =.53 and τ y = Sims-Zha Experimens and he Lucas Criique. Recall ha he Taylor-ype ineres rae rule is of he form: i = τπ + τ y + η. y For he baseline experimen, we se η =0. In he (unanicipaed) Sims-Zha experimen ineres raes are held consan for 4 quarers or equivalenly, η = τπ τ y, for = o 4. y Noe he sysemaic surprises here: The Sims-Zha experimen assumes ha households anicipae η o be whie noise when, in fac, i is a funcion of inflaion and oupu. In he anicipaed Sims-Zha experimen he policy rule is given by i = τπ + τ y + η y where η j (for j = o 4) are chosen o zero ou he ineres rae for four periods, bu hese values are forecasable by he public for j = 2 o 4. The public undersands ha for four quarers ha moneary auhoriy is going o deviae from he endogenous ighening under he Taylor rule and insead keep ineres raes consan. Figure 2 repors he impulse response funcions for a 0% increase in he price of oil (ν =.0). I is insrucive o compare our baseline and unanicipaed Sims-Zha numbers o hose obained by BGW in heir economeric esimaion. BGW esimaed ha a 0 percen oil price shock is associaed wih a 50 basis poin increase in he funds rae and a peak oupu decline of 0.7%. Our model suggess ha ineres raes would increase 0

15 by around 2 basis poins and oupu would decline by 0.3%. Boh esimaes are somewha smaller han esimaed by BGW. However, he model predics ha his decline would essenially be cu in half under he Sims-Zha experimen (0.7% versus 0.3%). This is essenially he conclusion of BGW s (2004) VAR analysis. In all hree scenarios here is a susained decline in consumpion and he real wage. This reflecs he negaive welfare consequences of he increase in he price of oil. The consumpion decline is significanly miigaed for he firs four quarers in he unanicipaed and anicipaed Sims-Zha experimens. This reflecs he simulaive effec on oupu of he consan ineres rae for hese firs four quarers. This is especially pronounced in he anicipaed Sims-Zha experimen. Bu he magniude of he Lucas Criique is quie clear. In sharp conras o BGW s hunch, he Lucas effec is quaniaively relevan. If he sable ineres rae had been anicipaed, oupu would have increased by 0.23%! Compared o he baseline ineres rae movemen, he anicipaed decline in he ineres rae (in he firs four quarers) is much more simulaive han he unanicipaed decline in he ineres rae. This arises because he anicipaed ineres rae sabiliy leads o a much larger effec on inflaion, and hus a larger decline in he real rae of ineres. When anicipaed, he simulaive impac of lower ineres raes are brough forward in comparison o he unanicipaed Sims-Zha experimen. This implies ha he level of consumpion during he four quarers ha ineres raes are held consan is always higher for he anicipaed experimen. The increase in consumpion for he anicipaed experimen implies ha money growh mus be higher o keep ineres raes pegged during he firs four quarers.

16 This resul is reminiscen of recen Federal Reserve policy decisions. Afer decreasing he funds rae o an unprecedened percen in June 2003, he FOMC inroduced a dramaic change in language saring wih he Augus 2003 meeing: he Commiee believes ha policy accommodaion can be mainained for a considerable period. The goal of his language was o condiion expecaions ha he funds rae would say unusually accommodaive. I was believed ha his would lead o higher inflaion and oupu han if he same sequence of ineres raes occurred, bu were unanicipaed. Here a series of announced shocks o keep ineres raes from rising in response o oil price increases has a much bigger impac han a series of unanicipaed shocks. 4. Oher Neural Policies. While he previous experimens provided poenial answers o he quesion, how would he economy have behaved if ineres raes were kep consan in he wake of an oil shock,, hese experimens do no seem o answer he quesion originally posed by BGW: how much of oupu s decline in response o an oil price shock is due o oil and how much is due o moneary policy? In fac he anicipaed scenario suggess ha oil s impac on he economy would be posiive! The nex series of experimens ry o answer BGW s counerfacual quesion by asking how would he economy behaved in response o an oil price shock if moneary policy were neural. The quesion hen is wha does i mean o keep moneary policy neural? The remaining moneary policy experimens do no have he form of a simple Taylor rule bu are all plausible versions of neural moneary policy. One idea of neural is a moneary policy rule ha holds he labor marke disorion (he oupu gap ) 2

17 consan, z zh = 0. We call his he Wicksellian policy. 8 In his case, he impulse + response funcions are idenical o he corresponding real business cycle model, i.e., a model wih no nominal sickiness. These resuls are hus analogous o Kim and Loungani (992). Anoher candidae for neural policy is where he money growh rae is held consan. In his case, he money demand curve is used o deermine he endogenous behavior of he nominal ineres rae. This is Leduk and Sill s (2004) definiion of neural. The final saemen of neural policy is an ineres rae peg. In conras o he previous experimens ha held ineres raes consan for four quarers, his is a rule in which ineres raes are always consan and his behavior is anicipaed. As is wellknown, here is real indeerminacy in his case. The decision rules can be expressed as funcions of wo lags in he real wage, wo lags of he exogenous oil price, and mean-zero sunspo shocks. For he simulaions below we eliminae he sunspo shock so ha he impulse response funcions represen fundamenal behavior only, i.e., his is he minimum sae vecor (MSV) equilibrium. Below we will discuss Leduk and Sill s (2004) alernaive way of supporing he ineres rae peg. Figure 3 repors he impulse response funcions for a 0% increase in he price of oil (ν =.0). Since he Wicksellian policy causes real behavior o mimic he real business cycle model i is a naural place o sar. The hump-shaped behavior in oupu and consumpion reflec he hump-shape in oil prices. These dynamics in consumpion 8 The Wicksellian policy is equivalen o a Taylor-ype rule wih a very large coefficien on he oupu gap. In his linearized model, his policy mimics he real behavior of he RBC model. However, his policy is no opimal as i ignores he higher order losses due o Calvo pricing. For he purposes of his paper, he naure of opimal policy is irrelevan. 3

18 correspond o an iniial decline in he real rae (-38 basis poins), followed by a jump in he real rae above seady-sae. Noe ha oupu and consumpion behavior for he Wicksellian policy are quie comparable o he benchmark Taylor-rule model. While he real behavior of he Wicksellian and benchmark Taylor-rule model are similar, he benchmark model delivers subsanially more inflaion han he Wicksellian rule. This occurs even hough he spike in nominal ineres raes is much greaer under he benchmark Taylor-rule. The key o his puzzle lies in he real ineres rae. In he earlier periods i is much lower (or more expansionary) for he Wicksellian policy bu here is a very long-period of ime in which i is slighly less expansionary han he benchmark Taylor-rule. Surprisingly his disan behavior is enough o drive he higher inflaion raes observed under he benchmark rule. If we accep he Wicksellian policy as neural hen he answer o he original counerfacual quesion is quie surprising. Despie he fac ha ineres raes in he baseline model increase over 00 basis poins wih respec o a 0 percen oil price shock, he real oupu response is very similar o Wicksell. This suggess ha essenially all of oupu s decline in he baseline Taylor rule scenario is due o oil. Anoher inerpreaion of neural policy is an ineres rae peg. An ineres rae peg leads o a sharp increase in he inflaion rae and hus a decline in he real ineres rae (42 basis poins). The low real rae implies a surge in oupu, consumpion, and inflaion. The nominal rae peg is anicipaed by he public so hese resuls are a naural exension of he anicipaed Sims-Zha experimen in which he ineres rae is held consan for four quarers. The longer period of ime (forever!) in which ineres raes are held consan resuls in larger oupu gains for he ineres rae peg (0.6%) compared o 4

19 he anicipaed Sims-Zha (0.22%) experimen. This version of neural suggess ha oil price increases have a simulaive impac on he economy. Finally our las version of neural is a consan money growh peg. The key observaion wih he money growh peg is ha he decline in consumpion leads o a decline in real money demand. Simulaneously he increased price level lowers he real money supply. The consumpion effec ends o lower ineres raes, he price effec ends o increase ineres raes. For his calibraion he consumpion effec dominaes so ha he oil shock leads o an endogenous decline in he nominal ineres rae. Despie his decline in ineres raes, oupu and inflaion are lower han wih an ineres rae peg. The reason is ha while nominal ineres raes are lower for a money growh peg, real ineres raes are higher. Boh he ineres rae peg and he money growh peg are more expansionary and lead o higher levels of oupu and consumpion (relaive o he baseline scenario) in he shor run. A money growh peg suggess ha a 0 percen oil shock would iniially increase oupu slighly bu hen evenually lead o a 0.22 percen decline in GDP, slighly less han he immediae 0.30 percen decline prediced by he benchmark. Table summarizes he resuls of his secion. We repor he eigh-quarer cumulaive oupu decline from an oil price shock under alernaive moneary policies. The second row of Table measures he oupu decline relaive o he oupu decline under he baseline Taylor rule. This percenage can be viewed as he fracion of he oupu decline due o oil under various definiions of neural moneary policy. For example, if he money growh peg is considered neural,.23 percenage poins of he.6 percenage poin decline in oupu is aribued o oil (76%), he remainder o non- 5

20 neural moneary policy (24%). However, if we define neural as he Wicksellian policy, hen he baseline moneary policy is simulaive so ha over 00% of he oupu decline is a resul of he oil shock. This implies ha he baseline Taylor Rule formulaion of moneary policy slighly off-ses oil price flucuaions in oupu compared o a RBC economy. We repor wo versions of he ineres rae peg. The firs is he peg suppored by he R = 0 policy rule in which we pick he MSV soluion. This corresponds o he impulse response funcions repored above. In his case he oil shock causes an oupu boom. If we rea his policy as neural, he decline in oupu following an oil shock is enirely caused by he moneary ighening under he Taylor rule. In conras, Leduc and Sill (2004) he ineres rae peg is suppored by posiing he following ineres rae rule: i = ϕ i + τπ + τ y, y where τ and τ y are very small and φ =.000 (see heir foonoe 9). This policy rule implies equilibrium deerminacy and near consancy of he ineres rae in heir model. (This is also he case in he model of his paper.) If we model he ineres rae peg as do Leduc and Sill (2004), hen oupu falls wih he oil shock. Because of space consideraions we do no repor hese impulse response funcions bu do summarize he oupu behavior in Table. There are wo peculiar characerisics of his rule. Firs, he decline in oupu is surprising given he resuls in he previous secion for he anicipaed Sims-Zha experimen. One inerpreaion of an ineres rae peg is an infiniely-long anicipaed Sims-Zha experimen. Second, alhough he realized ineres rae process is consan, equilibrium deerminacy resuls because he ou-of-equilibrium behavior is 6

21 explosive, φ >. I is no clear ha agens could learn φ since ineres raes do no move in equilibrium. 5. Conclusion. In wo influenial papers BGW (997, 2004), ried o answer he quesion: how much of GDPs decline wih respec o oil price increases is due o oil, and how much is due o he fac ha ineres raes also end o rise sharply as well. They repored ha a 0% oil price increase is associaed wih a 50 basis poin increase in he funds rae and a peak oupu decline of 0.7%. BGW hen use his VAR analysis o conclude ha approximaely half of his decline is due o oil and approximaely half is due o he increase in he funds rae. The firs conribuion of his paper is o use he sandard New Keynesian model o assess he accuracy of heir hunch ha he Lucas criique is no quaniaively relevan. We show ha if ineres raes were expeced o be kep consan for four quarers ha oupu would acually increase in he wake of an oil shock. Hence, wihin his heoreical model he Lucas criique is quaniaively relevan. While BGW designed a sensible experimen i is no clear ha i really answers he posed quesion: holding moneary policy consan (or neural) wha impac would an oil price shock have on he economy? The paper s second conribuion is herefore o expand his counerfacual quesion more broadly. In addiion o he Sims-Zha and anicipaed Sims-Zha experimens, we consider several oher versions of neural policy including a money growh peg, an ineres rae peg, and a Wicksellian ineres rae 7

22 policy. The laer is a policy ha adjuss he funds rae so ha he real economy behaves as if here were no nominal rigidiies. The behavior of oupu and inflaion are quie differen under all of hese possible versions of neural policy. Bu arguably he Wicksellian policy corresponds mos closely wih wha is ypically mean by a neural policy as real behavior mimics he real business cycle model. In conras o BGWs conclusion, his version of neural suggess ha all of he oupu decline associaed wih oil prices is due o oil, and none of he decline is aribuable o moneary policy. 8

23 References. Bernanke, Ben S., Mark Gerler, and Mark Wason, Sysemaic Moneary Policy and he Effecs of Oil price Shocks, Brookings Papers on Economic Aciviy, 9-42, Bernanke, Ben S., Mark Gerler, and Mark Wason, Oil Shocks and Aggregae Macroeconomic Behavior: The Role of Moneary Policy A Reply, Journal of Money Credi and Banking, V. 36, No. 2, April 2004, Calvo, G., Saggered Prices in a Uiliy-Maximizing Framework, Journal of Moneary Economics 2: , Carlsrom, C.T, and Timohy S. Fuers, Oil Prices, Monearry Policy, and Expecaions, manuscrip, Erceg, C.J., D.W. Henderson, A.T. Levin, Opimal Moneary Policy wih Saggered Wage and Price Conracs, Journal of Moneary Economics 46, Hamilon, James D., This is Wha Happened o he Oil Price-Macroeconomy Relaionship, Journal of Moneary Economics 38(2), 996, Hamilon, James D. and Ana Maria Herrera, Oil Shocks and Aggregae Macroeconomic Behavior: The Role of Moneary Policy, Journal of Money Credi and Banking, V. 36, No. 2, April 2004, Kozicki, S. How Useful Are Taylor Rules for Moneary Policy? Federal Reserve Bank of Kansas Ciy Economic Review, Second Quarer 999, Kim, I., and P. Loungani, The Role of Energy in Real Business Cycle Models, Journal of Moneary Economics 29, Leduc, S, and K. Sill, A Quaniaive Analysis of Oil-Price Shocks, Sysemaic Moneary Policy, and Economic Downurns, Journal of Moneary Economics 5 (2004), Lucas, R.E., Jr., Economeric Policy Evaluaion: A Criique, Carnegie- Rocheser Conference Series on Public Policy :9-46, Sims, Chrisopher A. and Tao Zha, Does Moneary Policy Generae Recessions?, Princeon Universiy Working Paper, Walsh, Carl, Moneary Theory and Policy, MIT Press: Woodford, Michael, Ineres and Prices, Princeon Universiy Press:

24 5. Yun, Tack, Nominal Price Rigidiy, Money Supply Endogeneiy, and Business Cycles, Journal of Moneary Economics 37(2), April 996,

25 Table : Eigh quarer cumulaive oupu decline Moneary Policy Baseline Taylor Rule Wicksell M Peg R Peg* R peg** 8 q cum oupu decline % oupu decline relaive o Taylor rule 00% 08.79% 76.33% NA 72.8% Row one is he eigh quarer cumulaive oupu decline under he corresponding policy rule. The second row is he raio of he oupu decline under a given policy rule relaive o he oupu decline under he baseline Taylor rule. *The ineres rae peg is defined as he policy rule R = 0 and he MSV soluion ha suppors his peg. **The ineres rae peg is defined as in Leduc and Sill (2004):, wih φ=.000, where τ and τ y are small. i = i ϕ + τπ + τ y y 2

26 FIGURE Percen, quarerly average Q:2005 Dollars per barrel OIL PRICES AND EFFECTIVE FEDERAL FUNDS RATE Wes Texas inermediae crude oil price Effecive Fed Funds rae :I 976:I 98:I 986:I 99:I 996:I 200:I 2006:I

27 FIGURE 2: Impulse Response o a 0% Oil Price Shock Percen.4 NOMINAL INTEREST RATE Percen.5 INFLATION.2 Baseline.2 Anicipaed SZ Anicipaed SZ SZ experimen Baseline SZ experimen Quarers Percen 0.3 OUTPUT Quarers Percen 0.4 REAL INTEREST RATE Anicipaed SZ 0 Baseline SZ experimen -0. SZ experimen Baseline -0.8 Anicipaed SZ Quarers Percen 0. CONSUMPTION Quarers Percen 0. REAL WAGE 0-0. Anicipaed SZ SZ experimen -0.2 Anicipaed SZ Baseline -0.3 Baseline SZ experimen Quarers Quarers

28 FIGURE 3: Impulse Response o a 0% Oil Price Shock Percen.5 NOMINAL INTEREST RATE Percen 2.0 INFLATION.0.5 Rpeg 0.5 Baseline.0 Wicksell 0.5 Baseline 0.0 Mgrowh Rpeg 0.0 Wicksell Mgrowh Quarers Percen 0.8 OUTPUT Quarers Percen 0.5 REAL INTEREST RATE 0.6 Baseline Mgrowh Rpeg 0.2 Rpeg -0.5 Wicksell 0.0 Mgrowh Baseline Wicksell Quarers Percen 0.6 CONSUMPTION Quarers Percen 0. REAL WAGE Rpeg -0. Rpeg Mgrowh Baseline -0.3 Baseline -0.4 Wicksell -0.4 Wicksell Mgrowh Quarers Quarers

29 Federal Reserve Bank of Cleveland Research Deparmen P.O. Box 6387 Cleveland, OH 440 PRST STD U.S. Posage Paid Cleveland, OH Permi No. 385 Address Correcion Requesed: Please send correced mailing label o he Federal Reserve Bank of Cleveland Research Deparmen P.O. Box 6387 Cleveland, OH 440