Intermediate Macroeconomics: New Keynesian Model

Transcription

1 Intrmdiat Macroconomics: Nw Kynsian Modl Eric Sims Univrsity of Notr Dam Fall 23 Introduction Among mainstram acadmic conomists and policymakrs, th lading altrnativ to th ral businss cycl thory is th Nw Kynsian modl. Whras th ral businss cycl modl faturs montary nutrality and mphasizs that thr should b no activ stabilization policy by govrnmnts, th Nw Kynsian modl builds in a friction that gnrats montary non-nutrality and givs ris to a wlfar justification for activist conomic policis. Nw Kynsian conomics is somtims caricaturd as bing radically diffrnt than ral businss cycl thory. This caricatur is unfair. Th Nw Kynsian modl is built from xactly th sam cor that our bnchmark modl is thr ar optimizing housholds and firms, who intract in markts and whos intractions giv ris to quilibrium prics and allocations. Thr is rally only on fundamntal diffrnc in th Nw Kynsian modl rlativ to th ral businss cycl modl nominal prics ar assumd to b sticky. By sticky I simply man that thr xists som friction that prvnts, th mony pric of goods, from adjusting quickly to changing conditions. This friction givs ris to montary non-nutrality and mans that th comptitiv quilibrium outcom of th conomy will, in gnral, b infficint. By infficint w man that th quilibrium allocations in th sticky pric conomy ar diffrnt than th fictitious social plannr would choos. Nw Kynsian conomics is to b diffrntiatd from old Kynsian conomics. Old Kynsian conomics aros out of th Grat Dprssion, adopting its nam from John Maynard Kyns. Old Kynsian modls wr typically much mor ad hoc than th optimizing modls with which w work and did not fatur vry srious dynamics. Thy also tndd to mphasiz nominal wag as opposd to pric stickinss. Wag and pric stickinss both accomplish som of th sam things in th modl thy man that th quilibrium is infficint and that mony is non-nutral. But nominal wag stickinss implis that ral wags may b countrcyclical, which is inconsistnt with th data. Fohis and othr rasons, Nw Kynsian modls tnd to mphasiz pric stickinss (though many of ths modls also fatur wag stickinss, too). Th Nw Kynsian framwork is th dominant paradigm fohinking about fluctuations and policy. A nic aspct of th modl is that, at its cor, it is th sam as th ral businss cycl

2 modl. Though th graphs that w ll us to analyz th modl look diffrnt, w could us th nw st of graphs that w r about to us to think about th ral businss cycl modl as wll (with a paramtric rstriction that would mak th Phillips Curv, to b introducd blow, vrtical). In that sns, what w r about to do is quit gnral. 2 Houshold Th houshold sid of th modl is xactly th sam as w v had bfor. Thr ar many, idntical housholds. W can thrfor modl it as thr bing a singl houshold that taks prics as givn. Housholds gt utility from consumption, labor, and holdings of ral mony balancs. Thy fac th sam budgt constraints that w v alrady sn. Th solution of th houshold problm implis xactly th sam consumption function, labor supply function, and mony dmand function (whr th mony dmand function maks us of th Fishr rlationship, with says that i t = + πt+, whr w tak xpctd inflation to b an xognous constant). Ths ar: N t = N s (w t, ) C t = C( G t, + G t+, ) M t = M d ( + π t+, ) Labor supply is assumd to b incrasing in th wag (substitution ffct dominats incom ffct), and incrasing in th ral intrst rat (whn gos up, th houshold would lik to sav mor, so it wants to arn mor incom as wll as consum lss). Housholds again bhav as though th govrnmnt balancs its budgt ach priod in othr words, Ricardian Equivalnc holds. Consumption is incrasing in currnt prcivd nt incom, with < MP C < dnoting th partial drivativ with rspct to currnt nt incom. Consumption is also incrasing in prcivd futur nt incom, whr, as bfor, w trat xpctd futur incom as ffctivly xognous (.g. w avoid th complications associatd with fdback ffcts into futur incom). Finally, consumption is assumd to b dcrasing in th ral intrst rat. Mony dmand is (i) proportional to th pric lvl, (ii) dcrasing in th currnt nominal intrst rat (so dcrasing in th ral intrst rat and th rat of xpctd inflation btwn t and t + ), and (iii) incrasing in th lvl of currnt incom. Ths conditions ar th sam as w had bfor, so w will not discuss thm in any furthr dpth. 3 Firms and Pric-Stting Th firm sid of th modl is diffrnt than w had bfor. Suppos that thr ar a bunch of diffrnt firms, indxd by i =,..., L. Thy ach produc output according to Y i,t = A t F (K i,t, N i,t ), 2

3 whr A t is common across all firms. Diffrntly than our prvious stup, w assum that firms hav som pric-stting powr suppos that th goods thy produc ar sufficintly diffrnt that firms ar not pric-takrs in thir output markt. Nvrthlss, w assum that thy do not bhav stratgically though firms might b abl to adjust thir own pric, thy don t act as though thir pric-stting (or production) dcisions hav any ffct on th aggrgat pric lvl or output (put diffrntly, th total numbr of ths firms, L, is assumd to b sufficintly larg. Formally, ths firms ar monopolistically comptitiv. W can think about aggrgat output and prics as ssntially wightd-avrags of individual firm output and prics. Ultimatly, sinc this is macro, w r rally only intrstd in th bhavior of aggrgats. W introduc firm htrognity bcaus w nd som pric-stting powr for pric-stickinss to mak any sns. Suppos that th dmand for ach firm s good is a dcrasing function of its rlativ pric: Y i,t = f ( ) Pi,t, X Hr X dnots othr stuff (lik tasts, aggrgat incom, tc). Th important and rlvant assumption is that f < : dmand foh good is dcrasing in th rlativ pric. Now, if all firms could frly adjust prics priod-by-priod, th rlativ prics of goods, P i,t, would b dtrmind by tasts and tchnologis (.g. diffrnt kinds of foods, or lctronics bcoming chapo produc, whatvr). Movmnts in th aggrgat pric lvl,, would hav no ffct on dmand for products if doubld but nothing ls changd, all firms would just doubl thir prics, P i,t. This wouldn t chang rlativ prics, so thr would b no chang in th dmand for goods, and hnc thr would b no ffct of a chang in on total output mony would b nutral. Suppos instad that firms hav to st thir prics in advanc basd on what thy xpct th aggrgat pric lvl to b. Dnot th aggrgat xpctd pric lvl as P t. Each firm individual sts its own pric to targt an optimal rlativ pric basd on othr conditions spcific to its product. Suppos that som fraction of firms cannot adjust thir pric within priod to changs in th aggrgat pric lvl, say bcaus of mnu costs or informational frictions. This mans that an incras in th aggrgat pric lvl,, ovr and abov what was xpctd, P t, will lad ths firms to hav rlativ prics that ar too low (whil firms that can updat thir prics will hav thiargt rlativ prics). With a lowr rlativ pric, thr will b mor dmand fohs goods. Th ruls of th gam ar that firms must produc howvr much output is dmandd at its pric th rational for which could b that rfusing to produc so as to mt dmand would lad to a loss in customr loyalty (or somthing similar). Thrfor, having a suboptimally low rlativ pric mans that ths firms must produc mor. With som firms producing mor than thy would lik to, aggrgat output will ris. Thus, thr will b a positiv rlationship btwn surpris changs in th pric lvl and th lvl of conomic activity. Lt Y f t dnot th hypothtical amount of output that would b producd in our standard, flxibl pric modl. This is unaffctd by pric rigiditis it would b th quilibrium lvl of output givn th ral xognous variabls (A t, A t+, G t, G t+, K t, and q) in th modl whr thr 3

4 wr no pricing rigiditis. Lt dnot th actual amount producd. Our story abov says that whn th aggrgat pric lvl incrass, output incrass, bcaus som firms cannot/don t adjust thir own pric, and hnc nd up producing mor than thy find optimal. Th story from th abov paragraph suggsts that thr ought to b a positiv rlationship btwn th gap btwn and Y f t and th gap btwn th actual and xpctd pric lvl th actual pric lvl bing highhan xpctd lads to mor production than would tak plac without pric rigidity, whras th pric lvl bing lowhan xpctd lads to lss production. W thrfor suppos that th aggrgat pric-output dynamics oby th following Phillips Curv (or somtims calld an AS (for aggrgat supply ) rlationship): = P t + γ( Y f t ) γ is a paramtlls us how sticky prics ar (in ssnc th fraction of firms than ar unabl to adjust thir pric). If γ, thn prics ar prfctly flxibl: w d hav = Y f t, vn if P t. If th aggrgat pric lvl diffrs from what was xpctd, but if all firms can adjust prics frly, thn all will do so with no chang in rlativ prics at th micro lvl. contrast, if γ, thn this conforms with all firms having sticky prics if no firms can adjust thir pric within priod, thn th aggrgat pric lvl will b qual to what it was xpctd to qual (th aggrgat pric lvl cannot chang within priod it is fixd if all firms ar unabl to adjust thir pric). For intrmdiat cass btwn γ and γ, th Phillips Curv will b upward sloping in a graph with on th horizontal axis and on th vrtical axis. Whn = Y f t, w will hav = Pt (xcpt in th cas in which th Phillips Curv is prfctly vrtical, with γ ): In 4

5 PC f Th valu of γ dtrmins th slop of th Phillips Curv (PC). Whn γ is larg, th curv is stp. This mans that it would tak a larg chang in th aggrgat pric lvl to gnrat a givn chang in output whn γ is larg, not many firms hav sticky prics, so it would tak a larg chang in th aggrgat pric lvl to gnrat much chang in aggrgat output (bcaus only a fw firms nd up with suboptimally low rlativ prics). In contrast, if γ is small, many firms hav sticky prics. It thrfor taks a smallr chang in th aggrgat pric lvl to gnrat a givn chang in output th curv is flattr. Two diffrnt Phillips Curvs ar shown blow. In ithr cas, whn = Y f t, w hav = P t. 5

6 γ larg γ small f Espcially whn thinking about dynamics, it is hlpful to think about th long run vrsus short run Phillips Curvs. In th long run, prics ar flxibl givn sufficint tim, all firms will b abl to adjust thir postd prics rlativ to xpctation and will achiv thir dsird rlativ pric. Hnc, in th long run, w will hav = Y f t rgardlss of th pric lvl. Th basic ida hr is that, ovr a sufficintly long horizon, thr will b no pric stickinss. Hnc, th long run Phillips Curv (LRPC) will b a vrtical lin at Y f t, which, to ritrat, is th hypothtical lvl of output that would b producd if prics wr flxibl. Th short run Phillips Curv (PC) must cross th LRPC whn = P t. If prics wr prfctly flxibl, with γ, thn th short run and long run Phillips Curvs would coincid at Y f t. 6

7 LRPC PC f What will shift th Phillips Curv? Mathmatically, th Phillips Curv plots a rlationship btwn and, holding Pt and Y f t fixd. Hnc, changs in Pt or Y f t will rsult in shifts of th Phillips Curv. An incras in Y f t will shift both th PC and th LRPC horizontally to th right, by th sam amount (such that th short run and long run curvs would cross at = Pt ). 7

8 LRPC LRPC PC PC Y,t f Y,t f An incras in th xpctd pric lvl, in contrast, would ffctivly shift th short run Phillips Curv up (with no chang in th LRPC). Holding fixd, highr Pt (quivalntly, you could think about an inward horizontal shift). translats into highr 8

9 LRPC PC PC P,t P,t f In summary:. Firms produc diffrntiatd goods. Th dmand fohir good is a dcrasing function of its rlativ pric, P i,t. 2. If firms could frly adjust prics, thn thir rlativ pric would b dtrmind by things lik tasts and tchnology. Any chang in th aggrgat pric lvl would hav no ffct on rlativ prics if doubld, all firms would just doubl P i,t. With no chang in rlativ prics, thr would b no rlationship btwn th aggrgat pric lvl and th amount of output producd. Mony would b nutral. 3. Suppos that firms post prics in advanc basd on what thy xpct th aggrgat pric lvl to b, P t. 4. Only a subst of firms can adjust thir pr-st prics in rspons to changing conomic conditions. This mans that prics ar sticky. This could b du to mnu costs (it is costly/impossibl to chang a postd pric) or informational frictions (individual firms don t obsrv th aggrgat pric lvl prfctly). 5. Suppos that th aggrgat pric lvl gos up rlativ to what was xpctd. Thos firms that can adjust thir prics do so proportionally to th chang in th aggrgat pric lvl. But at last som firms cannot. Sinc ths firms cannot adjust thir P i,t, thir ffctiv rlativ 9

10 pric falls whn gos up. A lowr rlativ pric mans mor dmand fohir good. Th ruls of th gam ar such that thy will produc nough to mt dmand. Sinc som firms nd up producing mor bcaus thir rlativ prics ar suboptimally low, aggrgat output gos up whn th aggrgat pric lvl gos up rlativ to whr firms xpctd it. 6. Th Phillips Curv is = P t + γ( Y f t ). It is upward-sloping in a graph with on th vrtical axis and on th horizontal axis. γ is a masur of how sticky prics ar and govrns th slop if prics ar not vry sticky (most firms can adjust pric within priod), th Phillips Curv is stp. If prics ar vry sticky, thn γ is small and th Phillips Curv is flat. 7. Th Phillips Curv shifts out whnvr Y f t incrass or P t dcrass. 8. Th long run Phillips Curv (LRPC) is a vrtical lin at = Y f t. In th long run, all firms can adjust thir prics, so output is indpndnt of th pric lvl. Now, w havn t mntiond labor dmand yt, and with good rason. As notd abov, th ruls of th gam ar that firms must produc howvr much output is as dmand at thir pric. Effctivly, whn firms hav pricing powr, thy don t choos laboo maximiz profit dirctly. Rathr, thy choos thir pric to maximiz profit givn th dmand fohir good, and thn hir sufficint labor (givn thir inhritd stock of capital and xognous lvl of productivity) to produc th output dmandd at thir rlativ pric. If prics wr flxibl, you could think about th problm ithr way hiring laboo maximiz profit would imply an optimal pric, whras choosing an optimal pric (what I dscribd abov) would imply an optimal choic of labor. If prics ar sticky, thr is no optimization going on firms just hav to hir as much labor as is ncssary to produc th output that is dmandd at thir pric. Effctivly, this is a long-windd way of saying that, to th xtnt to which prics ar sticky, output is dmand-dtrmind. Output will b dtrmind by th intrsction of th Phillips Curv with a nw curv calld th AD curv dscribd blow. Givn that lvl of output, firms will hav to hir as much labor as ncssary to produc that lvl of output. This mans that labor dmand is now longo hir labor up until th point at which th marginal product of labor quals th ral wag. Rathr, firms just hir nough laboo mt dmand, rgardlss of th wag. This mans that th aggrgat labor dmand curv is a vrtical lin and dpnds on th lvl of aggrgat output (as wll as th lvl of A t and K t ). Th aggrgat production function is th sam as w v had bfor: = A t F (K t, N t ). Givn, which will b dtrmind blow, as wll as K t and A t, which ar xognous, this quation implicitly dtrmins N t indpndntly of th wag. Hnc, labor dmand is vrtical in a graph with N t on th horizontal axis and w t on th vrtical. It shifts out whnvr (i) gos up, (ii) A t gos down, or (iii) K t gos down. Th lattr two ffcts rsult bcaus, if A t or K t go down, but is unaffctd, you nd mor N t to produc a givn lvl of. Givn individual production functions of this typ, you can actually show this mathmatically that it aggrgats up, at last undr crtain conditions

11 N d ( ) w t N t Th pictur abov shows labor dmand in th sticky pric modl. Th salint point hr is that, on prics ar sticky, w think about labor dmand as bing dtrmind by output, not by th marginal product of hiring an additional unit of labor. 4 Th IS-LM-AD Curvs Th Phillips Curv charactrizs th supply sid of th conomy. To charactriz th dmand sid, w r going to introduc a nw curv and rlabl an old on. 4. Th LM Curv In th ral businss cycl modl, mony was nutral, and w dtrmind th pric lvl last aftr all ral variabls wr dtrmind. Bcaus of pric stickinss, mony will no longr b nutral, and w dtrmin nominal prics simultanously with ral variabls. W ar going to introduc a nw curv, which is calld th LM curv. Th L stands for liquidity (or mony dmand) and th M stands for mony (or mony supply). Th LM curv is a plot showing all combinations of and for which th mony markt is in quilibrium (mony dmand quals mony supply), for givn lvls of mony supply, M t, and th pric lvl,. This dfinition of th LM curv dos not rly upon any notion of pric stickinss. W could hav dfind and usd this curv arlir in th flxibl pric modl.

12 Th figur blow dscribs th graphical drivation of th LM curv. Procd as follows. Start with som hypothtical intrst rat-output combination, (r t, Y t ), in th right panl in a graph with on th vrtical axis and on th horizontal axis. This dtrmins a position of th mony dmand curv in a graph with on th vrtical axis and M t on th horizontal axis. Suppos that th pric lvl, P t, and th mony supply, M t, ar initially such that th mony markt is in quilibrium (dmand = supply). Call this point (a). Now, suppos that w kp th ral intrst rat at r t, but incras output to Y t. This would caus th mony dmand curv to pivot to th right, whr at a givn pric lvl thr would now b mor dmand for mony. Call this point (b). For a givn and M t, th mony markt is not in quilibrium at point (b). Holding th mony supply and pric lvl fixd, with this nw highr lvl of, th only way th mony markt can rmain in quilibrium is if gos up (which would hav th ffct of pivoting in th mony dmand curv). Call this point (c), which, in th mony-dmand supply graph, is th sam position of mony dmand as at th original output and intrst rat combination. Hnc, if is highr, thn must b highr foh mony markt to b in quilibrium holding and M t fixd. This mans that th LM curv is upward-sloping, as shown blow. M s M d (, )=M d (, ) LM(M t, ) (a)=(c) (b) M d (, ) (a) (c) (b) M t M t As you can s from th labling abov, th LM curv is drawn holding fixd lvls of th pric lvl,, and th mony supply, M t. Hnc, changs in ithr of ths will caus th LM curv to shift. Start from an initial point (a), with rt, Yt, Pt, and Mt. First, suppos that thr is an incras in th mony supply, from Mt to Mt. This shifts th mony supply curv to th right. 2

13 Now, holding fixd at P t, if nithr nor wr to chang, th mony markt would not b in quilibrium. For a givn pric lvl, to b in quilibrium in th mony markt givn th nw highr mony supply, w nd th mony dmand curv to pivot out. This could occur if ithr (i) gos up or (ii) gos down (or som combination of th two). I lik to think about this as a horizontal curv shift, so suppos that Yt is th nw lvl of that would caus th mony markt to b in quilibrium for fixd at rt (i.. this is th nw lvl of that would caus th mony dmand curv to pivot so that th mony markt is in quilibrium with Mt at Pt ). Call this point (b). This mans that th LM curv shifts out whnvr M t incrass. W s this blow. M s M d (, ) LM(M t, ) (a) (b) M d (, ) (a) (b) LM(M t, ) M t M t M t Now, a chang in th pric lvl (holding M t fixd) will also shift th LM curv. Suppos that th pric lvl dcrass, from P t to P t. For givn valus of and (.g. r t and Y t ), mony supply would xcd dmand at a lowr pric lvl. To rstor quilibrium in th mony markt, w d nd th mony dmand curv to pivot out. This could occur if ithr incrass or dcrass, or som combination of th two. Sinc I lik to think about horizontal shifts of curvs, I pick out th nw lvl of, call it Y t, that would lad th mony dmand curv to pivot sufficintly fao th right to rstor quilibrium in th mony markt at th nw lowr pric lvl. Call this point (b). Similarly to th cas of an incras in th mony supply, w s that a dcras in th pric lvl also causs th LM curv to shift out. 3

14 M s M d (, ) LM(M t, ) LM(M t, ) (a) (b) M d (, ) (a) (b) M t M t In summary:. Th LM curv shows th combinations of (, ) whr th mony markt is in quilibrium, holding fixd th quantity of mony supplid, M t, and th pric lvl,. It is upward-sloping. 2. Th LM curv shifts out if M t incrass or dcrass. 3. An asy way to rmmbhis is that th LM curv shifts out of if th quantity of ral mony balancs, Mt incrass. 4. Th dfinition and graphical drivation of th LM curv in no way dpnd upon pric stickinss. modl. 4.2 Th IS Curv W could hav drivd th LM curv in xactly th sam way in th flxibl pric Th IS curv stands for Invstmnt = Saving. It is xactly th sam thing as what w hav bn calling th Y d curv. It shows th st of (, ) pairs for which xpnditur quals incom, givn optimizing bhavior by housholds and firms. Total dsird xpnditur in th conomy, Yt d is th sum of dsird xpnditur by housholds (consumption), firms (invstmnt), and th govrnmnt (govrnmnt spnding). Th functions 4

15 dscribing optimal bhavior ar as w hav sn. incom. Mathmatically: Dsird consumption is a function of dsird Y d t = C( G t, + G t+, ) + I(, A t+, q) + G t Hr I hav imposd Ricardian Equivalnc, whrby th houshold bhavs as though th govrnmnt balancs its budgt ach priod. Th marginal propnsity to consum, MPC or C Y, is btwn and. This mans that graphing dsird xpnditur against incom (xpnditur is Y d t, and incom is, w gt an upward-sloping lin with slop lss than on. W assum that dsird xpnditur is positiv vn at zro currnt incom (bcaus of th componnts of xpnditur which do not dpnd on incom, as wll as from xpctd futur incom). This mans that th dsird xpnditur lin must cross a 45 dgr lin along which Yt d quilibrium xpnditur must qual incom. = xactly onc. In any To driv th IS curv (which is th sam thing as what w hav bn calling th Y d curv, not that th position of th dsird xpnditur lin in (, Yt d spac dpnds on th ral intrst rat. Whn gos up, dsird xpnditur is smallr at vry lvl of incom th dsird xpnditur lin shifts down, and th point whr Yt d = is lowr. Th convrs is tru whn gos down. Conncting th dots, w gt a downward-sloping curv, just as w hav sn bfor. This is shown blow: d d = d (2 ) d ( ) d ( ) 2 Y d /IS Th IS/Y d curv will shift whnvr anything othhan lads to a chang in dsird xpn- 5

16 ditur. This includs changs in G t (incrass in which caus th IS curv to shift out on-for-on, as pr arlir argumnts), G t+ (incrass in which caus th IS curv to shift in), or incrass in q or A t+ (which caus th IS curv to shift out to th right by stimulating dsird invstmnt). 4.3 Th AD Curv Th AD, or Aggrgat Dmand, curv shows combinations of (, ) for which th conomy is both on th LM curv (mony markt clars) and th IS/Y d curv (goods markt clars). In othr words, th IS-LM curvs ar th building blocks of th AD curv. What conncts th IS-LM curvs (which show combinations of (, ) for which th goods and mony markts clar) is th pric lvl. As w showd abov, th pric lvl affcts th position of th LM curv. Changs in th pric lvl lads to shifts in th LM curv. Shifts of th LM curv combind with th IS curv imply diffrnt lvls of associatd with diffrnt lvls of. W can graphically driv th AD curv. Draw two graphs on top of on othr, whr th uppr graph has on th vrtical axis and on th horizontal axis, and with th lowr plot having on th vrtical axis and on th horizontal axis. Suppos that w initially start with a pric lvl, P t, and all othr dtrminants of th IS and LM curvs ar hld fixd. This pric lvl dtrmins a position of th LM curv. Th intrsction of th LM and IS curvs dtrmins a combination of and, call it r t and Y t. Bringing this point down, w gt a point (P t, Y t ). Now, suppos that w hav a highr pric lvl, P t. This causs th LM curv to shift in, as w saw abov. This mans that th whr th nw LM and IS curvs cross is lowr, Y t. Considhn a lowr pric lvl, P 2 t. This shifts th LM curv out, maning that th whr th IS and LM curvs cross is highr, Y 2 t. Conncting th dots, w gt a downward-sloping curv, which w call th AD curv. 6

17 LM( ) LM( ) LM(2 ) Y d /IS 2 AD What will shift th AD curv? Th AD curv will shift if anything othhan causs ithr th LM curv oh IS curv to shift. This includs M t (which shifts th LM curv), and G t, G t+, q, and A t+ (which shift th IS curv). Blow I show how th AD curv shifts to th right if th mony supply incrass: 7

18 LM(M t, ) LM(M t, ) Y d /IS AD AD In this pictur, th LM curv shifts right for a givn pric lvl, maning that th, combination whr th IS and LM curvs intrsct is down and to th right. Th highr for a givn mans that th ntir AD curv shifts out to th right. Nxt, considr an incras in G t, dcras in G t+, incras in q, or incras in A t+ (or a dcras in uncrtainty). Any of ths would caus th IS curv to shift right. This mans that th lvl of whr ths curvs intrsct (for a givn ) is highr, maning that th AD curv shifts to th right. 8

19 LM( ) r t r t IS IS AD AD In summary:. Th AD curv shows (, ) pairs whr both th mony markt (LM) and goods markt (IS/Y d ) clar. 2. A highr shifts th LM curv in, lading to a lvl of whr th IS and LM curvs intrsct that is lowr. Thrfor, th AD curv slops down. 3. Th AD curv shifts whnvr anything othhan shifts th IS or LM curvs. This includs M t (LM shift), as wll as G t, G t+, q, or A t+ (IS shifts). 4. Nothing about th IS, LM, or AD curvs dpnds on pric stickinss. W could hav usd ths xact sam curvs to summariz our arlir modl whr prics wr flxibl. It s just a diffrnt graphical way of charactrizing goods dmand, mony dmand, and mony supply. 5 Short Run Equilibrium Equilibrium in th conomy occurs whn w r on both th AD and th PC curvs. Bing on th AD curv mans that th mony and goods markts both clar. Bing on th PC summarizs th rvisd firm sid of th modl with pric rigidity. Bing on both th PC and AD curvs jointly dtrmins th pric lvl and output. Onc w know th lvl of output, w know th position of th labor dmand curv, and w can thn dtrmin th quilibrium quantity of mploymnt and th ral wag. In that sns, th labor markt is th last thing w look at, whras with flxibl prics it was th first. Th position of th IS and LM curvs, and hnc th position of th AD curv, is dtrmind by th xognous variabls rlatd to th dmand sid of th modl: A t+, G t, G t+, q, K t, and 9

20 uncrtainty. In dtrmining th position of th PC curv, w nd to know (i) Pt and (ii) Y f t. W tak Pt, th xpctd pric lvl, to b xognous. Th lvl of Y f t would b dtrmind by th intrsctions of th Y d and Y s curvs in a hypothtical flxibl pric conomy, givn th valus of th xognous variabls. Mathmatically, th conditions charactrizing th quilibrium ar: N t = N s (w t, ) C t = C( G t, + G t+, ) I t = I(, A t+, q, K t ) = A t F (K t, N t ) = C t + I t + G t = P t + γ( Y f t ) M t = M d ( + π t+, ) = i t π t+ As in th ral businss cycl modl, thr ar 8 quations. Thr ar 8 ndognous variabls:, C t, I t, N t,, w t, i t, and. Svn of th ight quations ar idntical to what w had bfor. Th nw quation is th Phillips Curv xprssion, which rplacs th labor dmand condition that firms should hir labor up until th point at which th wag quals th marginal product. W can d facto think about Y f t as bing xognous hr (in rality, Y f t would b dtrmind givn th valus of othr xognous variabls in th flxibl pric vrsion of th modl). Th xognous variabls of th modl ar th sam as bfor: A t, G t, G t+, q, K t, M t, πt+, and now a nw on, Pt. Th IS curv summarizs consumption dmand, invstmnt dmand, and total goods dmand ( = C t + I t + G t ). Th LM curv summarizs mony dmand, th Fishr rlationship, and th xognous mony supply. Th AD curv summarizs both th IS and LM curvs. Th Phillips Curv summarizs th pricing sid of th modl. Th Phillips curv plus th AD curv dtrmin and. Givn and othr xognous variabls, w know th positions of th IS and LM curvs, which dtrmins th composition of output and. Givn and th xognous valus of A t and K t, w dtrmin mploymnt from th production function. Givn that lvl of mploymnt, w thn dtrmin th ral wag from th labor supply xprssion. In contrast to th RBC modl, whr w lookd at th labor markt first, in th NK modl w figur out output first, and thn find th lvl of N t consistnt with that and th w t consistnt with th houshold bing on its labor supply curv. W run into an irritating complication. With th xcption of M t, all of th xognous variabls which affct th AD curv (A t+, G t, G t+, q, K t, and any uncrtainty about th futur) would also affct th hypothtical Y f t. This mans that, othhan M t, any of th xognous variabls which affct th AD curv should also affct th PC curv. This mans that thr isn t a clar dichotomy btwn dmand shocks and supply shocks (othhan M t, which only affcts th 2

21 AD, and A t, which only affcts th PC). W r hrtofor going to ignor ths ffcts. In othr words, from hr on out, assum that xognous variabls which affct th IS curv (A t+, q, G t, and G t+ ) do not impact Y f t. This mans that changs in ths xognous variabls will not lad to shifts in th PC. W could gt this litrally if w assumd that labor supply wr not snsitiv to : this would man that th Y s curv was vrtical, so changs in ths xognous variabls would not impact Y f t. Mor gnrally, w can think of this as an approximation whrin th ffcts of changs in ths xognous variabls on Y f t ar sufficintly small that w can safly ignor th ffcts on th PC and focus on ths xognous variabls as dmand shocks which only impact th AD. W will split xognous variabls into thr camps fohis modl. Th first camp is montary. Changs in M t will affct th AD curv. Th scond is supply think changs in A t. This will lad to changs in Y f t and hnc shifts of th PC and th LRPC. Th third camp ar IS shocks: as notd abov, ths ar changs in G t, G t+, A t+, q, or uncrtainty. Changs in ths variabls impact th IS curv and lad to shifts in th AD curv. W assum that thy do not lad to shifts in th PC, as in th paragraph abov. LM IS LRPC PC = AD = f Th pictur abov shows th IS-LM and AD-PC quilibrium. Jointly, th intrsction of ths curvs dtrmin,,, and th componnts of. Onc w know and, w know th positions of th labor supply curv (sam as it vr was) and th modifid, vrtical labor dmand curv. This allows us to comput th quilibrium lvls of N t and w t, as shown blow. 2

22 N d ( ) w t N s ( ) w t N t N t An quilibrium is a st of prics (, w t, and ) and quantitis (, C t, I t, and N t ), givn M t, A t, A t+, q, G t+, G t, Pt, and πt+, such that th mony, goods, and labor markts simultanously clar. W hav thr sub-markts that must clar, so w hav thr prics that do th claring. Nxt, lt s go through th ffcts of changs in th xognous variabls. I ll split ths into thr catgoris, th sam as abov. First, w ll considr a montary shock. Scond, a supply shock. Third, an IS shock. 5. Montary Shock First, considr an xognous incras in M t. This has th ffct of shifting th LM curv to th right, and hnc also th AD curv to th right. Thr is no ffct on Y f t,, or any of th componnts of th IS curv, so only th LM and AD curvs shift. Ths ar shown by th blu lins in th pictur blow. For a givn pric lvl, th nw quantity of goods dmandd xcds th quantity supplid from th Phillips Curv. Hnc, th pric lvl,, must ris to th point whr th nw AD curv intrscts th PC curv. Call ths nw points Yt and Pt. Thr is now an indirct ffct on th LM curv in that th highr pric lvl causs it to shift back in part-way. This maks th quantity of output in th IS-LM graph coincid with th quantity from th intrsction of th AD-PC curvs. As long as th PC is not vrtical, output incrass, so th LM curv dos not shift back in all th way. Put diffrntly, on nt an incras in M t lads to an incras in Mt, ral mony balancs, so on nt th LM curv has shiftd out. Hnc, on nt, 22

23 w hav highr, highr, and lowr. lowr, plus highr, mans that C t and I t both incras. Hr th montary transmission mchanism is that th montary xpansion lowrs ral intrst rats, which stimulats consumption and invstmnt. LM(M t, ) LM(M t, ) LM(M t, ) IS = LRPC PC AD AD In th pictur abov, not that if th Phillips Curv wr vrtical (such as would b th cas in th long run ), thr would b no ral ffct of th montary xpansion. Th LM curv would still shift right, causing th AD curv to shift right, but with a vrtical PC, th pric lvl would ris such that thr would b no ffct on ral balancs, Mt, and thrfor no nt ffct on th position of th LM curv. Thr would b no chang in th ral intrst rat, no chang in, and no chang in C t or I t. Just as w saw bfor, th only ffct of an incras in M t would b a proportional incras in th pric lvl. Hr you can s that this nw graphical stup can also b usd to study th flxibl pric modl w had alrady considrd. Now, back to th cas in which th PC is upward-sloping, not vrtical, what happns in th labor markt? Highr mans that aggrgat labor dmand incrass th vrtical labor dmand curv shifts to th right. A lowr ral intrst rat lads to th labor supply actually shifting in. An inward shift of labor supply, and an outward shift of th vrtical labor dmand curv, mans that N t and w t ar both highr in th nw quilibrium. 23

24 w t N d ( ) N d ( ) N s ( ) w t N s ( ) w t N t N t N t In summary, thn, a incras in M t lads to a dcras in and incrass in w t,,, C t, N t, and I t. With th xcption of, ths movmnts will all b largh flattr is th Phillips Curv (.g. th stickir ar prics). In th xtrm cas in which th Phillips Curv is prfctly vrtical, nothing but th pric lvl would ract. 5.2 Supply Shock Nxt, suppos that w hav a supply shock that incrass Y f t but has no ffct on th IS, LM, or AD curvs (.g. an incras in A t ). Thr is no dirct ffct (holding th pric lvl fixd) on th LM or IS curvs, and hnc no ffct on th AD curv. Th Phillips Curv shifts out to th right bcaus of th incras in Y f f t, by an amount horizontally qual to th incras in Yt (which is th amount by which th LRPC shifts out to th right). This rsults in th pric lvl fall, from Pt to Pt. Output incrass, from Yt to Yt. Not that th incras in is lss than th horizontal shift of th PC (in othr words, incrass by lss than Y f t ). Th lowr pric lvl translats into an outward shift of th LM curv. Th LM curv intrscts th unaffctd IS curv at a lowr ral intrst rat, r t, and a lvl of consistnt with th intrsction of th PC and AD curvs. Th lowr ral intrst rat, plus highr output, man that C t and I t ar highr. 24

25 LM( ) LM( ) IS LRPC PC PC = AD It is difficult to dtrmin what happns in th labor markt. Th lowr ral intrst rat causs labor supply to shift to th lft. Othhings bing qual, this would work to rais th ral wag. Th ffct on labor dmand is unclar. On th on hand, is highr, but on th othr, so is A t, which mans that firms nd lss laboo produc a givn amount of output. Hnc, w cannot sign th shift in th N d curv it could shift out or it could shift in. Hnc, w cannot dtrmin what happns to w t or N t with any crtainty. W can say th following, howvr. W know that N t gos up by lss (or down by mor) aftr an incras in A t than it would if prics wr flxibl. If prics wr flxibl, th PC would b vrtical, thr would b a biggr chang in, a biggr drop in, and a biggr drop in (and hnc biggr incrass in C t and I t ). Anothr way to put this is that output dos not ris by nough rlativ to what it would if prics wr flxibl. Th rason is that highr A t puts downward prssur on prics. Sinc som firms cannot adjust thir prics, thy nd up with rlativ prics that ar too high, and consquntly produc lss than thy would othrwis lik. In th xtrm cas of prics bing prfctly rigid (γ ), thr would b no ffct of th supply shock on or any othr ral variabls. With not rising, and A t highr, w would know that th vrtical labor dmand curv would shift in (you d nd lss laboo produc a givn amount of output with productivity highr). 25

26 5.3 Dmand Shock Nxt, considr som xognous chang which affcts th IS curv, but not Y f t. As ph discussion abov, this could rprsnt xognous changs in A t+, G t, G t+, q, or uncrtainty, assuming th Y s curv is vrtical, or at last sufficintly clos to vrtical so that th changs in Y f t ar sufficintly small so as to b abl to safly ignor th ffcts on th PC. Th ffcts on th IS-LM and AD-PC curvs ar shown blow: LM( ) LM( ) IS IS LRPC PC = AD AD = f Th IS curv shifts to th right. This would rais for a givn : in othr words, th AD curv shifts horizontally to th right. Th AD curv shifting right, with no shift in th PC, mans that w nd up with highr and highr. Th incras in is smallhan th horizontal shift in th AD (unlss th PC is prfctly flat). Th highr pric lvl lads to an inward shift of th LM curv, so that th whr th IS-LM curvs intrsct is th sam as whr th AD-PC curvs intrsct. W nd up with,, and all highr. How C t and I t ar affctd dpnds on what shock is driving th IS shift if q is going up, I t would b highr; whras if it is G t, thn C t and I t will both b lowr. Hnc, all w can say gnrically is that th ffcts on C t and I t of an IS shock ar ambiguous. Lastly, go to th labor markt. Sinc output is highr, firms nd to hir mor labor, so th vrtical labor dmand curv shifts right. Th highr ral intrst rat stimulats labor supply. Hnc, w know that N t will go up, but it is unclar what will happn to th ral wag. It could go up or go down. 26

27 Finally, sinc w hav implicitly assumd that th Y s curv is vrtical (so that things which shift th IS curv hav no ffct on Y f t, and hnc th position of th PC curv), not that ths IS shocks hav largr ffcts on output th flattr is th Phillips Curv. Wr th Phillips Curv vrtical (prics flxibl), thr would b no ffct. Hnc, thr is a crtain symmtry with th cas of a supply shock (whr w argud that th ffct on was smallh stickir wr prics). Th stickir ar prics (th flatth PC), th biggh ffcts of IS shocks on. Th intuition fohis is that ths shocks lad to incrass in th pric lvl, which causs som firms to gt stuck with rlativ prics that ar too low, and hnc ths firms nd up producing mor than thy would othrwis find optimal. In othr words, with pric stickinss, dmand sid shocks hav largr ffcts on output than thy would if prics wr flxibl, with th opposit th cas for supply shocks. Th tabl blow summarizs th ffcts on th ndognous variabls for (i) a montary shock, (ii) a supply shock, and (iii) a positiv shock to th IS curv. Variabl: M t A t (Supply) IS Shock (positiv) C t + +? I t + +? N t +? + w t +?? Lt s quickly compar ths qualitativ co-movmnts with what w obsrv in th data (s th last st of nots, on ral businss cycl thory). Th basic facts ar: quantitis tnd to comov positivly with output, th ral wag is procyclical but not as strongly as quantitis lik consumption or invstmnt, th ral intrst rat is ssntially uncorrlatd with output, and th pric lvl if mildly countrcyclical. W obsrv kind of a mixd bag hr. A chang in A t (supply shock) producs th right co-movmnts in trms of,, C t, and I t. That said, pric stickinss maks it lss likly that N t and w t incras whn A t gos up; and with a sufficint amount of pric-stickinss N t is in fact likly to go down. Montary shocks do prtty wll in producing movmnts in ndognous variabls that look lik what w obsrv in th data, with th xcption of. IS shocks gt and N t to mov togthr, but produc movmnts in and that arn t prfctly consistnt with th data. Positiv IS shocks may lad to incrass in w t though, which is mor consistnt with th data than w obsrv in th flxibl pric modl. Whil non of ths shocks dos a prfct job of accounting for movmnts in ndognous variabls rlativ to what w s in th ral world, w can draw th following conclusion. Pric stickinss improvs th ability of dmand shocks to account for fluctuations ( dmand maning ithr montary shocks or IS shocks), and maks it hardr for 27

28 supply shocks (changs in A t ) to account for mpirically obsrvd businss cycls. In trms of IS shocks, w gt smallr incrass in and at last hav th possibility that w t riss aftr a positiv shock which raiss, unlik in th flxibl pric modl. Montary shocks can hav ral ffcts hr, whras thy hav no ral ffct in th flxibl pric modl. Finally, supply shocks hav smallr ffcts on th stickir ar prics, and thrfor ar lss likly to lad to positiv co-movmnt btwn and N t. Blow is a summary about how to work through xognous changs in th NK modl:. Start in th IS-LM diagram. Figur out if th IS or LM curvs shift, holding th pric lvl fixd. This will tll you if th AD curv shifts or not, and in which dirction. 2. Thn ask yourslf whthh PC shifts. This will occur if Y f t xtnsivly abov) or if Pt wr to chang. changs (rcall cavat discussd 3. Combin th AD and/or PC shifts to dtrmin a nw quilibrium lvl of and. 4. Tak th nw and adjust th position of th LM curv in such a way that intrsction lins up. Dtrmin th nw at th point at which th IS and LM curvs intrsct. Combin th changs in and, along with th chang in th xognous variabl, to try to figur out how C t and I t ar affctd (you may not b abl to). 5. Finally, go to th labor markt. Givn th you found, dtrmin what N t nds to b. This dtrmins th position of th vrtical labor dmand curv. Thn givn, dtrmin th position of labor supply. Combin labor supply and dmand to dtrmin w t. 6 Dynamics Th xrciss abov ar awfully static in th sns that no thought is givn to pric adjustmnt. In th short run firms may hav to kp thir prst pric, but aftr a whil pricing rigiditis should b irrlvant. In this sction w think about th dynamic rsponss to th sam shocks w lookd at in th prvious sction. Th basic ida of dynamics is that Pt (i.. firms bing surprisd by th aggrgat pric lvl) will lad to subsqunt adjustmnt of pric xpctations that will shift th Phillips Curv. In particular, it sms natural that if firms wr surprisd by an aggrgat pric lvl that is highhan xpctd, > Pt, thn, whn givn th opportunity, firms will updat thir pric xpctations. This would lad to th Phillips Curv shifting in/up. In contrast, if firms ar surprisd with lowr than xpctd prics, < Pt, thy will do th opposit: xpctd prics will fall, which will shift th Phillips Curv out. Now in trms of thinking of ths dynamics, a complication ariss in th sns that w rally ought to b formal about th numbr of tim priods. Instad of doing that, w r going to ngag in a bit of a hand-waiving xrcis. Think about a priod, t, as having multipl parts (.g. a day has morning, aftrnoon, and night; a yar has 2 months or four quartrs). To b as spcific 28

29 as possibl, lt s divid a priod, t, into two parts: call thm th short run and th mdium run. In th short run, Pt is fixd. In th mdium run, xpctd prics will adjust basd on movmnts in actual prics, in such a way = Pt, so that w r always at Y f t. This is anothr way of saying that mony can only b non-nutral for a short priod of tim. Th way w ll go through ths dynamic xrciss is as follows:. Go through and do th xrcis as abov. Figur out how and will chang basd on shifts of th AD or PC curvs. Think about this as happning in th first half of priod t. 2. If < Pt, assum that Pt adjusts up in such a way that th PC shifts in/up in such a way that th quilibrium lvl of is qual to Y f t. Think about this is occurring in th scond half of priod t. W r going to go through and do ths xrciss for a montary shock, a supply shock, and an IS/dmand shock. I us black lins to dnot th original curv positions, blu lins to dnot th short run ffcts of th shock (i.. what happns in th first half of t), rd lins to dnot th intrmdiat or indirct ffcts of pric changs on th position of th LM curv, and grn lins to dnot what happns in th mdium run (i.. th scond half of priod t). 6. Montary Shock Suppos that thr is an xognous incras in M t. From our arlir analysis, w know that this lads to an outward shift of th LM curv for a givn pric lvl, as indicatd by th blu LM curv blow. This rsults in an outward shift of th AD curv. This rsults in an incras in th pric lvl to P t, and an incras in output to Y t. Th incras in output is smallhan th horizontal shift of th AD curv, so that th incras in th pric lvl has a scondary or indirct ffct on th LM curv, pushing it back in slightly so that th quantitis of align at Y t. This is no diffrnt than that which w hav alrady don. 29

30 LM(M t, )=LM(M t,2 ) LM(M t, ) LM(M t, ) IS P 2 t P t = LRPC PC PC AD AD What s nw hr is th grn part. W s in th pictur that > P t : put diffrntly, firms hav bn foold. Fool m onc, sham on you. Fool m twic, sham on m. Firms ract by raising thir xpctd pric, and hnc thir postd prics, in th scond half of th day. This rsults in an inward shift of th PC. Th magnitud of th inward shift is such that th PC intrscts th nw AD curv at th original quilibrium lvl of Yt (which w implicitly assum is qual to Y f t, which is unaffctd by M t ). This incras in th xpctd pric lvl, lading to th inward shift of th PC, rsults (rlativ to th first half of priod t) in a rduction in output and an incras in. Th incras in rsults in an inward shift of th LM curv, in such a way that th outputs align in both picturs. Sinc w rturn to th original lvl of output, th LM curv rturns to its initial position (albit with diffrnt lvls of M t and ). W s hr that, in th mdium run, mony is nutral. Th chang in M t has no ffct on, no ffct on, and hnc no ffct on C t or I t. With no ffct on or, thr is no mdium run impact on th positions of th labor supply oh vrtical labor dmand curvs, and hnc no chang in N t or w t. In othr words, w r back to th montary nutrality cas w saw bfor th only ffct of a chang in M t is to chang, with no ral consquncs. In th short run (.g. th first half of th priod t), mony is not nutral bcaus prics ar imprfctly flxibl (th Phillips Curv is no vrtical): this mans that th incras in M t rsults in a tmporary rduction in, which stimulats consumption and invstmnt and hnc output. 3

31 6.2 Supply Shock Nxt, considr a positiv supply shock that raiss Y f t but has no ffct on th AD curv (.g. an incras in A t ). This has no dirct impact on th IS or LM curvs. It rsults in an outward shift of both th PC and th LRPC, by an amount horizontally qual to th chang in Y f t. This is shown in th graph by th blu lins, with P C dnoting th nw Phillips Curv. Bcaus th AD curv is not prfctly horizontal, in th nw quilibrium, actual output incrass by lss than th horizontal shift in th PC, to Y t. Th pric lvl falls to P t, which is lowhan P t. Th lowr pric lvl works to shift th LM curv out (rd lin), so that th ral intrst rat is lowr and th quantity of output aligns with th AD-PC intrsction. This is idntical to what w saw abov. r t r t r 2 t LM( ) LM( ) LM(2 ) IS LRPC PC PC = PC 2 2 AD 2 Now, lt s think about th dynamics. Th pric lvl is lowhan what is xpctd, but w also hav quilibrium output lowhan th nw Y f t (in th pictur, Yt < Yt 2. Firms bing surprisd on th downsid rsults in thm lowring Pt. This rsults in an outward shift of th PC, by an amount such that it intrscts th AD curv at Y 2 t, which is th nw flxibl pric lvl of output. This rsults in a furthr dcras in th pric lvl, which rsults in an outward shift of th LM curv so that th quantitis of output align. Put diffrntly, in th short run (th first half of th priod), output undr-racts to th supply shock. This happns bcaus th pric lvl is unabl to fully ract bcaus of pric stickinss. In th mdium run, firms adjust pric xpctations, in such a way that w hav = Y f t. 3

32 6.3 IS Shock Finally, lt s considr a shock to th IS curv (chang in A t+, G t, G t+, or q) which has no ffct on Y f t, and hnc no ffct on th PC (s discussion abov). From our arlir discussion, w know how things will play out in th short run. Th IS curv shifts out to IS. This raiss th quantitis of goods dmand at a givn pric lvl, rsulting in an outward shift of th AD curv. Th outward shift of th AD curv, along with an upward-sloping (but not vrtical) PC mans that and both go up. Th incras in is smallhan th outward shift of th AD (unlss th PC is prfctly horizontal). Th incras in th pric lvl has an indirct ffct whrin it rsults in an inward shift of th LM curv, so that th quantitis of output in th uppr and lowr graphs align. Th ral intrst rat riss. r 2 t r t LM(2 ) LM( ) LM( ) IS IS P 2 t P t = LRPC PC PC AD AD = f Lt s think about th dynamics at work. In th nw short run quilibrium, w hav Yt > Y f t, and > Pt. Firms hav bn surprisd by highhan xpctd prics. Thy ract in th scond half of th day by incrasing thir pric xpctation, which has th ffct of shifting th Phillips Curv in, to P C. Th magnitud of this shift is such that th PC intrscts th nw AD at Y f t, which has not changd. Th highr pric lvl rsulting from this inward PC shift works to shift th LM curv in vn mor, so that th quantitis of output in th uppr and lowr diagrams align. Th ral intrst rat nds up highr. What s going on hr is that output xpands by too much (rlativ to th flxibl pric lvl of output, Y f t ) bcaus of pric rigidity in th short run, th pric lvl cannot ris by nough, so 32

33 th ral intrst rat riss by too littl, and output xpands by mor rlativ to Y f t (which dos not chang, by construction). This is th mirror imag of what happns in rspons to a supply shock output rsponds by too littl to a supply shock, and by too much aftr a dmand shock, whn prics ar sticky (th Phillips Curv is not vrtical). But in th mdium run, (th scond half of th priod), prics ar ffctivly not rigid firms can updat thir xpctd prics in such a way that th actual pric lvl will b what thy xpct, and w ll hav = Y f t. Hnc, pric stickinss is a short run friction that maks th quilibrium lvl of output diffr (at last foh first half of th priod) from Y f t. 6.4 Th Limits of Montary Expansion Th introduction and study of ths dynamics (simpl and contrivd nough as thy ar) allows us to addrss som potntially intrsting policy qustions. Can th cntral bank gt th conomy to produc mor than Y f t? As w hav sn abov, th answr is ys, at last in th short run. Whn th cntral bank incrass M t, incrass, but not by nough bcaus som firms ar stuck with thir old pric. In ssnc, montary nutrality works by fooling firms into producing too much. But w hav sn that thr is a limit to this procss. Whil th cntral bank can tmporarily boost output through montary policy, in th mdium run all that happns is that prics ris. What would happn if th cntral bank trid to prmanntly kp output high (say bcaus of political prssur)? It would hav to continually incras th mony supply: in our graphs, this would rprsnt continual outr shifts of th AD. But ths continual outr shifts of th AD would b mt by continual inward shifts of th PC as firms adjust thir pric xpctations. Hnc, th cost of trying to xpand output abov its potntial lvl (Y f t ) is highr prics. Th cost of trying to continually do this is continually highr prics (.g. high and rising inflation). Evntually, if th cntral bank trid to do this nough, firms would catch on, and would build ths pric incras xpctations into thir pric-stting, and montary xpansion would hav no ral ffcts it would just rsult in inflation. Bottom lin: th cntral bank cannot prmanntly stimulat output by printing mor mony. To th xtnt to which it can do so in th short run (bcaus of pric rigidity), th cost is highr prics. Trying to continually incras th mony supply in such a way as to furthr stimulat output would rsult in furthr pric lvl incrass, which would vntually b factord into pric-stting xpctations, making montary policy inffctiv in stimulating output. 6.5 Costly Disinflation and Crdibility Suppos that, for whatvr rason, a cntral bank would lik to bring th pric lvl down (or mor gnrally, crat a disinflation a slow down in th rat of growth in prics). As w hav sn abov, th cntral bank can do this by rducing th mony supply. This rsults in an inward shift of th LM curv, and a rsulting inward shift of th AD curv. Th inward shift of th AD curv along an upward-sloping PC curv mans that both and fall. In othr words, th cost of bringing prics down is lowr output. W can s this in th pictur blow: 33