9 NEW KEYNESIAN MODELS OF AGGREGATE DEMAND

Transcription

1 Economics 314 Coursebook, 2019 Jeffrey Parker 9 NEW KEYNESIAN MODELS OF AGGREGATE DEMAND Chaper 9 Conens A. Topics and Tools... 2 B. Romer s New Keynesian IS/LM and IS/MP... 3 Uiliy maximizaion... 3 Consumpion behavior and he IS curve... 6 The LM curve in he new Keynesian model... 7 Romer s MP curve... 9 IS/LM, IS/MP, and aggregae demand C. Some Simple Aggregae-Supply Models Case 1: Nominal-wage sickiness Case 2: Inflaion sickiness wih a compeiive labor marke Case 3: Inflaion sickiness wih labor-marke imperfecions Case 4: Sicky wages wih imperfec compeiion D. The Open Economy Expendiures in an open economy Capial flows and ineres-rae pariy Floaing and fixed exchange raes The Mundell-Fleming model Imperfec capial mobiliy E. Unemploymen and he Phillips curve Oupu, employmen, and unemploymen The original Phillips curve The naural-rae hypohesis Modern challenges o he naural-rae hypohesis F. Suggesions for Furher Reading The IS/LM and Mundell-Fleming models The Phillips curve G. Works Cied in Tex... 25

2 A. Topics and Tools In many ways, life is easier now han i was for he previous generaion. No doub you have all been subjeced o your elders sories abou walking en miles o school every morning hrough driving snow sorms, uphill boh ways. However, when i comes o learning macroeconomics, hings have goen much harder raher han easier. When I ook macroeconomics in 1973, virually all of he semeser was devoed o he basic Keynesian IS/LM model and some close relaives. From he 1940s ino he 1970s, his modeling framework was hough o be an adequae descripion of he macroeconomy. I had successfully explained he major evens of macroeconomic hisory paricularly he Grea Depression and all ha seemed o remain for macroeconomic research was o esimae he parameers of he consumpion, invesmen, and money-demand funcions wih ever-greaer precision as he passage of ime provided us wih more daa poins. Subsequen evens have shown ha he basic Keynesian model by iself is far from an adequae represenaion of he macroeconomy. Neverheless, many of he predicions of he model sill hold rue when IS/LM or is varians are viewed as a componen of a more complex sysem involving boh aggregae demand and aggregae supply. Those who criicize he basic Keynesian model end o judge models by a sandard of how well he model is grounded in uiliy- and profi-maximizing microeconomic behavior of agens. By his sandard, he radiional IS/LM model seems inadequae even as a descripion of aggregae demand, when compared, for example, o he microbased models of consumpion and invesmen ha we will develop in laer chapers. The new Keynesian school of macroeconomics akes on he challenge of explaining how i migh be in he ineres of profi-maximizing firms o keep prices consan when demand changes. I also examines how somehing similar o he IS and LM curves may be derived from individual maximizing behavior of households and firms, wih he effecs of raionally formed expecaions of fuure variables being incorporaed ino he model. 9 2

3 B. Romer s New Keynesian IS/LM and IS/MP Romer begins Chaper 6 by developing he new Keynesian IS/LM model. This model ends up looking similar o he radiional IS/LM, bu is derived from an explici uiliy-maximizaion decision by households. Romer s version of his model ignores capial. While accumulaion of capial is a he cener of long-run economic growh, i plays a secondary role in shor-run business cycles. 1 We may jusify leaving capial ou of he model by reaing i as a fixed facor of producion in he shor run. This assumpion leads us o Romer s producion funcion (6.1) in which labor is he only (variable) inpu. Uiliy maximizaion In Coursebook Chaper 7, we described several ways in which a need for money can be inroduced. Alhough i is mos realisic o model he ransacion process direcly, he conclusions of hese models are similar o simpler ones in which moneary services are placed direcly in households uiliy funcions. Romer uses he money-inhe-uiliy-funcion approach in equaion (6.2): M U U C V L. 0 P The overall naure of his uiliy funcion should be familiar, bu several aspecs warran discussion. Addiiviy. The hree componens of uiliy ener he funcion addiively. This simplifies he analysis grealy because he marginal uiliy of consumpion a ime is U (C ), which depends only on C and no on M /P or L. Similarly, he marginal uiliy of money holding depends only on he curren real money balance and he marginal disuiliy of labor depends only on he curren level of labor. Discoun rae. Insead of e or 1/ 1, equaion 6.2 has as a discoun facor. This is jus a simplified, general noaion o represen exacly he same 1 Empirically, cyclical flucuaions in invesmen are boh large and imporan, bu we ignore hem here o keep he model simple. Our emphasis in mos of he analysis is on he shor-run behavior of aggregae supply, wih a simple quaniy heory specificaion siing in for aggregae demand. 9 3

4 kind of discouning we have used before. You can hink eiher of coninuously compounded discouning or 1/ 1 e for for annually compounded discouning o conver his noaion ino more familiar erms. Marginal uiliies. The derivaive condiions on he U funcion reflec he familiar assumpion ha he marginal uiliy of consumpion is posiive bu diminishing. Nohing is new here, and Romer specifies a familiar funcional form for U in (6.3). Similarly, he parial derivaion is he marginal uiliy of an addiional uni of real money held. ( is capial gamma.) The condiions > 0 and < 0 reflec he assumpion ha people ge posiive bu diminishing marginal uiliy from holding money. Labor is reaed differenly in (6.2) han in he real-business-cycle uiliy funcion of Chaper 5. Insead of adding posiive uiliy from leisure, equaion (6.2) subracs he disuiliy of labor. The condiion V > 0 coupled wih he minus sign in fron of V assures ha an increase in labor effor lowers uiliy. The second-derivaive condiion V > 0 appears asymmeric relaive o he oher second-derivaive condiions, bu i has he same implicaion: ha he marginal disuiliy of work increases wih more labor. This corresponds o a diminishing marginal uiliy of leisure, which would appear as a negaive second derivaive wih respec o leisure if we wroe he uiliy funcion in erms of leisure raher han labor. The budge consrain (6.5) differs from our previous analyses because i is wrien in nominal (dollar) raher han real erms. The variable A measures he household s dollar asses here and is defined so ha A is household nominal wealh a he beginning of period. Since here is no capial in he model, each dollar of his wealh is held in one of wo forms: as money or as bonds. To undersand equaion (6.5), consider he choices a household makes during period. I begins he period wih financial asses A, earns W L in labor income during period, and spends P C on consumpion goods. Thus, A W L PC is he amoun ha he household has lef o allocae during period beween is holdings of money M and is holdings of bonds, he laer of which are A W L PC M. The money holdings carry forward ino he nex period wih no ineres; he bond holdings earn ineres a rae i, which gives us he expression in (6.5) for nominal asses a he beginning of period There is an awkwardness inroduced ino he budge consrain by he discree-ime assumpion. Equaion (6.3) reas M as he amoun of asses allocaed o money ou of he wealh ha is available afer period income and consumpion have happened. The does no si comforably wih he idea ha money is held for he purpose of conducing hese ransacions and ha he 9 4

5 In order o undersand he uiliy-maximizaion decision and he relaionship beween he nominal ineres rae i in (6.5) and he real ineres rae r in (6.6), i is useful o examine he budge consrain in real erms raher han nominal. The real value of household asses a ime is A /P. Dividing each erm on boh sides of (6.5) by P yields A M A W M L C i P P P P P 1 1 (1) Bu real asses a ime + 1 are A + 1/P + 1. Working wih he lef-hand side of (1), A 1 A 1 P 1 A 1 1 1, P P P P 1 1 P 1 P 1 P where 1 1 is he inflaion rae from o + 1. Subsiuing ino (1) P P yields A 1 1 M 1 i A W M L C. P P 1 1 P P P (2) Several aspecs of equaion (2) deserve special aenion. The real ineres rae r ha measures he amoun of real wealh in period + 1 ha one ges for each uni of real period asses held as bonds is defined by 1 i 1 r. (3) 1 1 We can derive he more familiar formula by noing ha 1 i 1 r 1 1 r r household s money balance should rise or fall hrough he period as i is earned or spen. (Consider, for example, he coninuous-ime Baumol-Tobin model discussed in Chaper 7.) To inroduce inra-period flucuaions in money balances would complicae he model wihou fundamenally changing he resuls. 9 5

6 The final erm in his expression is he produc of wo numbers ha are (when he inflaion rae is low) much smaller han one, so his erm is order-of-magniude smaller han he oher erms in he equaion. For example, if he real ineres rae is 0.02 and he inflaion rae is 0.03, he produc is If we neglec his erm and subrac one from boh sides, we ge he familiar condiion i, r 1 or r. i 1 The expression 1 + r given by equaion (3) is he amoun of real, period + 1 asses ha is obained if we reduce period consumpion by one uni. Thus, 1 + r is he price of period consumpion in erms of period + 1 consumpion i is, as before, he slope of he budge consrain beween C and C + 1. Each uni of money holding a ime is worh unis of real asses in period + 1. If he inflaion rae is posiive, hen his number is less han 1, which reflecs he real depreciaion of money hrough inflaion. By a similar 1 approximaion as above, 1 1. To see his, noe ha because for small raes of inflaion he square of inflaion is of a small order of magniude. So each uni of real money holdings a ime ranslaes o approximaely unis of real asses in period + 1, making he real rae of reurn on money approximaely equal o + 1: he rae of deflaion. Consumpion behavior and he IS curve The consumpion decision in his model is very similar o he analysis we performed in he Diamond model in Romer s Chaper 2. Romer s equaion (2.53) gives he Euler equaion in he Diamond model as 1 1 C 1 1 C 2, 1 1, r 1. (4) There are several small differences beween he models ha reconcile our equaion (4) wih Romer s equaion (6.6). Firs, he discoun facor in his model is raher han 1 1. Second, he real rae of reurn beween periods and + 1 is now called r raher han r + 1 as in he Diamond model. Third, he ime-of-life disincion in consumpion ha is necessary in he Diamond model is irrelevan here, so we drop he firs subscrip 9 6

7 on C. Making hese adjusmens o our Diamond-model equaion (4) gives us 1 C 1 C, or C 1 r C 1, which is Romer s equaion (6.6). 1 r Since we have no capial, no governmen, and no foreign secion, C = Y in his simple model, so we can rewrie he Euler equaion in erms of Y. Taking logs, his gives lny ln1 r ln lny 1. For small values of r, r ln 1 r, so we 1 1 can wrie his equaion as lny a ln Y 1 r, where a ln. The new Keynesian IS curve of equaion (6.8) shares a defining propery wih he Hicksian IS curve discussed in he previous secion: i exhibis a negaive relaionship beween desired curren spending and he real ineres rae. Equaion (6.8) explicily includes fuure income (presumably expeced fuure income in any realisic applicaion) as a deerminan of he new Keynesian IS curve. In he earlier analysis we discussed expeced fuure income as affecing he radiional IS curve hrough he consumpion funcion, so his also is similar. Perhaps he mos obvious difference beween he radiional and new Keynesian IS curves is he absence of any fiscal-policy variables from he laer. Operaionally, his happened rivially because we negleced governmen spending and axes in he consumpion model ha led o equaion (6.8). Bu he inroducion of governmen ino he opimal consumpion model is far from rivial, as we have seen. The effec of curren governmen spending and axes on curren consumpion is likely o depend on wheher he underlying condiions for Ricardian equivalence hold infinie lifeimes, lump-sum axes, perfec capial markes, ec. While mos new Keynesian economiss would probably argue ha increases in curren governmen spending (wihou changing axes) would increase spending, augmening equaion (6.8) o reflec his is a challenging enerprise. The LM curve in he new Keynesian model The budge consrain (6.5) shows us he radeoff ha households face beween holding higher money balances and consuming more. Re-arranging erms from equaion (1) above, A A W M 1 L C 1 i i P P P P (5) Equaion (5) shows ha he household rades off money balances agains consumpion (holding fuure asses consan) according o M, or 1 i C i P 9 7

8 i M C. 1 i P (6) Equaion (6) shows ha each uni of real money balances held coss he household i 1 i unis of consumpion goods. If he household is maximizing uiliy, hen he marginal uiliy of i 1 i unis of consumpion mus equal he marginal uiliy of one uni of real money balances, so i M U 1 i P C. U C C M M and P P Given he funcions forms shown in Romer s (6.3) and (6.4), i. 3 M Subsiuing, C, which can 1 i P be re-arranged (noing ha Y = C) ino Romer s money-demand funcion (6.11). Equaion (6.11) serves he role of he LM curve. Noe ha we can rewrie he i erm o ge so or M P / 1 Y 1 i 1 M 1 Y i P M i Y P 1/ 1, 1. (7) Equaion (7) is he new Keynesian LM curve solved for i. I slopes upward in Y and shifs down o he righ when here is an increase in M/P, jus like he radiional LM curve. In Romer s Figure 6.1, he plos his LM curve ogeher wih he new Keynesian IS curve. However, in order o do his, we mus ranslae he LM curve o a form ha uses he real ineres rae r raher han he nominal rae i. This is mos easily done using he approximaion r i 1 as discussed above. Applying his o (7) yields 3 The parameer χ is Greek leer chi, no Roman leer x. They can be difficul o disinguish. 9 8

9 1 M r Y 1 1. P As Romer shows in Figures 6.1 and 6.2, we can perform comparaive saic analysis using he new Keynesian IS/LM model. The resuls are basically idenical o hose examined in Chaper 7 for he radiional, Hicksian model. Romer s MP curve The LM curve assumes ha he moneary-policy auhoriy (he cenral bank) follows a policy of seing he money supply a a fixed level and allowing ineres raes o be deermined enirely by marke forces. This kind of policy was ypical in he goldsandard world in which Hicks (and Keynes) lived, bu i has been abandoned by mos modern cenral banks in favor of policies ha se a key benchmark ineres rae (he federal funds rae in he Unied Saes) in response o he levels of oupu and inflaion. Romer reflecs his change by replacing he radiional LM curve wih an MP curve ha reflecs he more responsive moneary-policy sraegy. If, insead of fixing he money supply, cenral banks follow he now-common policy of argeing ineres raes, hen i is more naural o express he asse-marke equilibrium condiion by he MP curve, which makes he ineres rae depend on he policy rule esablished by he cenral bank. Whereas he exogenous level of M is he main deerminan of he posiion of he LM curve, he posiion of he MP curve will depend on he facors ha influence he cenral bank in seing is arge ineres rae. The supply of money is endogenously deermined as he amoun ha is required o achieve he cenral bank s ineres-rae arge. The mechanism of moneary policy is he same under eiher sysem: he cenral bank buys or sells governmen securiies o expand or conrac he moneary base and he supply of bank reserves. If i arges he money supply as in he model of he LM curve, hen i adds o or subracs from he moneary base unil he money supply is a he argeed level. If i arges he ineres rae, hen i expands or conracs he supply of reserves unil he ineres rae is a is arge level. In he Unied Saes, he Federal Reserve arges he federal-funds ineres rae. This is he rae on overnigh loans of reserves beween large banks, so i is very sensiive o he supply of and demand for reserves. By adding o he supply of reserves, he Fed can easily push he federal-funds rae down; by draining reserves from he banking sysem i can drive he rae upward. Replacing he LM curve wih he MP curve changes he moneary side of he model in several imporan ways. The quaniy of money is now an endogenous variable raher han being exogenously deermined by he cenral bank, bu i no longer plays a key role in he deerminaion of oupu. Given he level of oupu and he ineres rae 9 9

10 se by he cenral bank s policy rule, he quaniy of money is jus whaever amoun saisfies he public s demand for money. The quaniy of money essenially becomes an aferhough and disappears from he model. A second imporan change inroduced by he MP curve is ha we now incorporae ino he model he policy rule by which he cenral bank ses is policy insrumen. In he LM framework, M is exogenous: we do no build a model of how he cenral bank chooses M, we jus accep is choice as given. In he MP seup, raher han simply aking he cenral bank s ineres-rae arge as exogenously give, we build he policy response o economic condiions ino he model endogenously. Therefore he MP curve, raher han jus being a horizonal line a a given, policy-deermined r, represens he policymaker s response o curren economic condiions in paricular, o he curren level of real oupu Y. Modeling he MP curve hus requires ha we model how he cenral bank s ineres-rae arge depends on oher variables. Mos modern cenral banks worry abou wo economic oucomes: he rae of inflaion and he level of economic aciviy. In he long run, he value of a currency (he inverse of he price level) depends on is supply relaive o he demand for i. Thus, as we saw in he previous chaper, susained money-supply growh in excess of he growh in money demand will cause inflaion. Mos cenral banks respond o rising inflaion by raising ineres raes o curail aggregae demand and pu downward pressure on inflaion. Cenral banks ofen also pursue counercyclical moneary policy, decreasing ineres raes during recessions o ry o increase spending. This would mean ha arge ineres raes would be low when real oupu falls below he naural, full-employmen level and high when oupu rises above he naural level. This leads o a posiive relaionship beween he ineres rae and real oupu, which Romer represens by he monearypolicy, or MP, funcion of equaion (6.27), where r r lny ln Y,, wih r 1 > 0 and r 2 > 0. 4 A rule of his kind has come o be called a Taylor Rule afer Taylor (1993). If we plo he resuling equilibria in (Y, r) space, we ge an upward-sloping curve whose posiion depends on he rae of inflaion. An increase in would shif he MP curve upward. The operaion of he IS/MP model is similar o he IS/LM model. 4 The Federal Reserve, like mos cenral banks, ses is official ineres-rae arge in nominal raher han real erms. If he nominal rae is he real rae plus he rae of inflaion (ignoring he difference beween acual and expeced inflaion) hen his possibiliy can be easily accommodaed wihin Romer s equaion (6.26). The real-ineres-rae rule is jus he nominal-ineresrae rule minus he inflaion rae. In order for real ineres raes o rise o counerac an increase in inflaion, he cenral bank s nominal ineres rae arge mus go up more han one-for-one wih an increase in inflaion. Clarida, Galí, and Gerler (2000) show ha U.S. Federal Reserve policy failed his condiion prior o 1979, which presumably helped fuel rising inflaion. 9 10

11 Prices affec IS/MP equilibrium hrough he effec of he inflaion rae on he cenral bank s real-ineres-rae arge. IS/LM, IS/MP, and aggregae demand We can derive an aggregae-demand curve by exploring he effecs of a change in he price level on he IS and LM curves eiher radiional or new Keynesian. An increase in he price level reduces he real value of he exising supply of money, so if M says consan, M/P mus fall. This leaves households and firms wih less money han hey wan given he curren (and so far unchanged) values of income and he ineres rae. To re-esablish asse equilibrium, people will aemp o acquire addiional money by selling ineres-bearing asses. Bu since everyone is rying o sell bonds and no one is buying hem, somehing mus change in order o make bonds more aracive. The obvious oucome is ha he ineres rae mus rise, making bonds more aracive relaive o money and reversing he desire o exchange bonds for money. Thus, he LM curve mus shif up and o he lef in response o an increase in prices, which lowers he equilibrium quaniy of oupu. This negaive relaionship beween he quaniy of oupu demanded and he price level (for given values of G, T, and M) can be drawn in (Y, P) space as a downwardsloping aggregae-demand curve. For he IS/MP model, he verical axis of he AD curve mus be he inflaion rae raher han he price level. An increase in he inflaion rae causes he cenral bank o raise is real-ineres-rae arge, pushing he MP curve up o he lef and lowering he demand for goods and services. Thus, he AD curve based on he IS/MP model also slopes downward. C. Some Simple Aggregae-Supply Models Romer s Chaper 6 and much of he remainder of his course are devoed o models in which he microfoundaions of aggregae supply are carefully specified. In Secion 6.2, Romer presens four ses of assumpions and he aggregae-supply curves ha would resul. We consider he inuiion of hese cases, along wih anoher imporan reference case, in his secion. Before discussing Romer s four cases we begin wih one ha I will call Case 0. This is he case where, as in he real-business-cycle model and our growh models, here are no imperfecions in he adjusmen of wages or prices. In his case, he level of real oupu Y is deermined solely by applying he aggregae producion funcion o 9 11

12 he equilibrium amouns of labor and capial in he economy. In Case 0, he AS curve is verical a he naural level of oupu, as shown in Figure 1. In Case 0, a change in moneary policy ha shifs he MP and AD curves simply resuls in a change in he rae of inflaion: money is neural and oupu is unaffeced. Similarly, a change in expendiures due o fiscal policy ha shifs he IS curve and he AD curve would leave oupu unchanged and affec only inflaion. In order for aggregae demand o have any effec on real oupu, we mus inroduce some imperfecion ino he price/wage adjusmen process. This is wha Romer does in an ad-hoc way in Secion 6.2 and more rigorously in Chaper 6 Par B. Case 1: Nominal-wage sickiness Keynes clearly believed in he sickiness of nominal, bu no real, wages. He argues ha a worker would accep a reducion in her real wage hrough an increase in prices, bu no hrough a decline in her nominal wage. 5 Modern Keynesians bring his kind of wage sickiness ino he model hrough nominal-wage conracs. 6 As Romer discusses on page 245, an increase in inflaion for any given nominal wage leads o a lower real wage: he more P goes up for given W, he lower W/P is. This lower real wage causes firms o hire addiional labor, increasing heir oupu via he producion funcion. Thus, nominal-wage sickiness can provide a raionale for an upward-sloping aggregae-supply curve. Increases in inflaion lead o increases in oupu by lowering he real wage. Sicky-wage models have one serious counerfacual implicaion. They work precisely because oupu moves in he opposie direcion as real wages. Firms produce a lo when real wages are low. This implies ha real wages should be srongly counercyclical. However, mos evidence suggess ha real wages are mildly procyclical exacly opposie o he predicions of he sicky-wage model. This empirical conradicion has eroded he suppor for his class of aggregae-supply models. 5 This clearly conflics wih radiional noions of economic raionaliy, in which only he purchasing power of he wage should maer. Keynes anicipaes some modern heories of fairness, envy, and alruism by suggesing ha any single worker would resis a reducion in her nominal wage because ha would imply a lowering (a leas for a ime) in her wage relaive o ohers. In conras, a decline in real wages due o an increase in prices affecs all workers symmerically, so here is no change in her relaive wage. 6 Examples of modern Keynesian models buil around conracs are Fischer (1977) and Taylor (1979). In Romer s Chaper 7, we sudy varians of hese models in which prices raher han wages are assumed o be sicky. 9 12

13 AS * AD Y* Y Figure 1. Aggregae supply wih no marke imperfecions. Case 2: Inflaion sickiness wih a compeiive labor marke Romer s second case assumes ha inflaion is sicky bu he wage adjuss perfecly o equae supply and demand in he labor marke. One raionale for such a model would be if firms se prices in advance and commi o hem hrough conracs wih buyers. In ha case, he curren price level or inflaion rae would be predeermined and unresponsive o changes in curren oupu Y. The behavior of he labor marke in his model is shown in Romer s Figure 6.4. The verical segmen of he kinked labor-demand curve shows ha for mos levels of he real wage, firms level of employmen is independen of he wage. They hire he amoun of labor needed o produce he level of oupu demanded by heir cusomers, F 1 (Y), where F 1 () is he inverse of he producion funcion and ells how much labor is required o produce a given Y. However, if he real wage ges high enough, firms are unwilling o produce even ha much oupu and hey are on he downward-sloping par of heir labor-demand curves. An increase in aggregae demand shifs he labordemand curve as shown in Figure 6.4, leading o a srongly procyclical real wage and a counercyclical markup of prices over marginal cos. Case 3: Inflaion sickiness wih labor-marke imperfecions Romer s Case 3 differs lile from Case 2. In fac, from he sandpoin of he deerminaion of oupu and inflaion, here is no essenial difference. The disincion lies 9 13

14 in he assumpion of some kind of labor-marke imperfecion ha leads o non-zero unemploymen. There are many reasons why firms migh pay wages in excess of he compeiiveequilibrium real wage. We discuss some in he chaper on unemploymen laer in he course. If firms pay an efficiency wage in order o reduce urnover, moivae workers, or for some oher reason, here will be a general excess supply of labor in he marke. Incorporaing labor-marke imperfecions ino he model allows he flexible-wage model of Case 2 o be reconciled wih he exisence of counercyclical unemploymen in he labor marke. Case 4: Sicky wages wih imperfec compeiion Case 4 exends Case 1, he basic wage-sickiness model. I allows for imperfec compeiion in he produc marke, so ha firms prices are higher han marginal cos. This markup rae is assumed o depend posiively on employmen L so ha firms reduce markups in recessions and increase hem in booms. 7 The chief benefi of his model is ha i rescues he sicky-wage model from he counerfacual predicion ha real wages are srongly counercyclical. D. The Open Economy [Noe: Romer has eliminaed he open-economy analysis of he Mundell-Fleming and relaed models from he fourh and fifh ediions of his ex. I am reaining his secion in he coursebook for he use of any sudens who wan o go back o he hird ediion and sudy his maerial. All references in his secion are o Romer s hird ediion.] Exending he basic Keynesian model o he open economy involves modeling wo kinds of inernaional connecions. Firs, we mus bring ne expors (expors minus impors) ino aggregae demand. Second, we mus ake accoun of inernaional flows of borrowing and lending. 7 In sandard models of monopoly, he markup depends on he elasiciy of demand. The more elasic demand is, he smaller he markup. Thus, he assumpion being made here follows if cusomers demand curves become more elasic in recessions and less elasic in booms. 9 14

15 Boh he flow of goods and he flow of capial 8 beween economies are ofen consrained. Some governmens impose high ariffs or resricive quoas on impors (or, rarely, expors) ha preven inernaional ransacions. Someimes here are resricions placed on he abiliy of domesic residens o hold foreign asses (i.e., o lend abroad) or o borrow from foreigners. The workhorse model of inernaional macroeconomics, analogous o he IS/LM model of he closed economy, is he Mundell-Fleming model. This model corresponds closely o he model wih floaing exchange raes and perfec capial mobiliy in Romer s Secion 5.2 (3 rd ed.). Expendiures in an open economy In an economy wih inernaional rade, we mus accoun for he desired expendiures of foreigners on domesic goods (and for he desired impors of foreign goods by domesic buyers). Boh domesic and foreign buyers of radable goods face a choice beween goods produced in he home counry (America, for convenience) and goods produced abroad. There are many facors ha will deermine he amoun someone spends on domesic vs. foreign goods. Preferences, differences in qualiy, and oher facors will cerainly play a par, bu heir choice should also depend parially on he relaive price of American goods and foreign goods. This relaive price is he real exchange rae. A numerical example should clarify he definiion of he real exchange rae. The relaive price of foreign goods (in erms of American goods) is he amoun of American goods you mus give up in order o ge one uni of foreign goods. To buy one foreign good you need P* unis of foreign currency, where P* is he foreign price level. The nominal exchange rae e measures he price of foreign currency (in erms of dollars), so each uni of foreign currency coss e dollars. Thus, you need ep* dollars in order o buy one uni of foreign goods. Since each American good coss P dollars, you need o give up 1/P American goods o obain a dollar. Therefore, o ge ep* dollars you have o give up ep*/p unis of American goods; he real and nominal exchange raes are linked by he formula = ep*/p. The higher is he real exchange rae, he more expensive foreign goods are relaive o domesic ones, so he more inclined boh foreign and domesic buyers are o buy domesic goods. Thus, desired expendiures on domesic goods should depend posiively on, as shown by Romer s modified expendiure funcion (5.14). 8 We are alking abou financial capial here, no fixed capial goods. In he inernaional macroeconomic lieraure, flows of borrowing and lending across borders are called inernaional capial flows. 9 15

16 Capial flows and ineres-rae pariy The oher major form of macroeconomic ineracion beween counries is in he capial marke, where domesic residens may lend o foreigners (a capial ouflow) or foreigners may lend o domesic households, firms, and governmens (capial inflow). There are many consideraions ha go ino choosing wheher o lend money a home or abroad. Boh he expeced rae of reurn on he loan and he risk involved will generally be imporan. A special case ha has araced he aenion of macroeconomiss is he case of perfec capial mobiliy. This is a siuaion in which risk is eiher symmeric across counries or unimporan o lenders, so all wealh-holders choose o lend in he counry ha offers he highes rae of reurn. If expeced raes of reurn are higher in he Unied Saes han in Europe, hen everyone will wan o lend in he Unied Saes and financial capial will flow rapidly from Europe o he U.S. If reurns are higher in Europe hen capial will flow he oher way. Only when he expeced raes of reurn on American and European asses are equal will here be no endency for capial o flood one way or he oher. So when here is perfec capial mobiliy, equilibrium in he inernaional asse marke requires expeced raes of reurn o be equal. The expeced real reurn for an American lender on an American asse such as a bond is jus he real ineres rae r. Buying a European bond is more complicaed for he American lender because i involves firs exchanging dollars for euros, hen buying he European bond, hen exchanging he euros back for dollars when he bond maures. The real rae of reurn on he European bond for an American invesor is given by Romer s equaion (5.20). I is equal o he European real rae of ineres plus he expeced rae of appreciaion of he real exchange rae over he period of he bond. Thus, if here is no expeced change in he real exchange rae, perfec capial mobiliy will lead o he equaliy of real ineres raes across counries: r = r*. This is known as he ineres-rae pariy condiion in erms of real ineres raes. As Romer shows a he op of page 235, he ineres-rae pariy condiion can be wrien eiher in erms of he real ineres rae as above or in erms of nominal ineres raes. The simples case is one in which he nominal exchange rae is expeced o remain unchanged. In his case, he nominal ineres raes (as well as real raes) in he wo counries mus be equal under perfec capial mobiliy, i = i*. However, suppose ha he exchange rae is expeced o increase a rae x, so ha each euro will be worh more dollars when he bonds maure han i is worh now. If he euro is expeced o appreciae agains he dollar, hen he U.S. bond will have o pay higher ineres han he euro bond in order o compensae for he expeced depreciaion in he dollars o be received a mauriy. In nominal erms, i = i* + x, which is approximaely equivalen o Romer s equaion (5.21). In real erms, he ineres-rae pariy condiion becomes Romer s equaion (5.20) when we consider he possibiliy of expeced change in he real exchange rae. 9 16

17 Floaing and fixed exchange raes The cenral bank in an open economy has an addiional opion for is policy arge beyond he simple money-supply and ineres-rae arges discussed above. Many cenral banks choose o arge he exchange rae wih moneary policy. This leads o fixed exchange raes, in which he exchange rae is decided upon by he cenral bank. Romer chooses no o analyze fixed exchange raes wih perfec capial mobiliy, bu insead pospones he analysis o he imperfec-capial-mobiliy case. The Mundell-Fleming model Wih he opening of he economy, we now have hree key endogenous variables in play: oupu, he real ineres rae, and he real exchange rae. We can only work in wo dimensions a a ime, so Romer chooses o analyze he economy in (Y, ) space, wih he exchange-rae equivalens of he IS and MP curves designaed as IS* and MP*. One can equally well o he analysis in (Y,r) space corresponding o he IS/MP model. Under perfec capial mobiliy wih no expeced real currency depreciaion, he domesic real ineres rae mus be equal o he foreign (world) rae. Thus, he MP* curve is given by Romer s equaion (5.16), which does no involve and hus is verical in (Y, ) space. The IS* curve slopes upward in (Y,) space because an increase in (a real depreciaion of he dollar) causes an increase in ne expors and hus in aggregae demand. Romer considers he case of an increase in governmen spending, which shifs he IS* curve o he righ. Wih he verical MP* curve, his leads o an appreciaion of he domesic currency (a decrease in ) wih no change in oupu demanded. The inuiion of his resul is somewha opaque in (Y, ) space, so le s hink abou i in (Y, r) space. Figure 2 shows he resuling equilibrium. The economy sars a poin e. Then governmen spending increases, which shifs he IS curve o he righ o IS. In a closed economy, his increase in demand and oupu would cause he cenral bank o raise real ineres raes o r, esablishing a new equilibrium a poin c. However, in an open economy wih perfec capial mobiliy, he real exchange rae canno say above he world rae r*. As he domesic ineres rae begins o increase, capial will flood in from he res of he world, pushing up he value of he domesic currency (reducing ) and causing ne expors o decrease. This decrease in ne expors drives he IS curve back o where i sared and oupu reurns o Y 0 wih a lower exchange rae. Wih perfec capial mobiliy, fiscal policy has no effec on aggregae demand under floaing exchange raes. Any increase in governmen spending compleely crowds ou an equal amoun of ne expors. 9 17

18 r MP r r* e c x IS IS Y 0 Y Y Y Figure 2. Expansionary fiscal policy in an open economy I is worh noing ha he moneary auhoriy could have decided o keep he exchange rae fixed, buying up all he foreign currency ha people waned o sell in order o ge dollars. Had hey done so, he MP curve would have shifed o he righ and equilibrium could have been resored a x. If he cenral bank follows a fixed-exchangerae policy hen fiscal policy has a very powerful effec on aggregae demand. This is one reason why some analyss have argued ha fiscal policy is especially imporan in a moneary union such as he euro-area. Imperfec capial mobiliy Capial is highly mobile among he advanced counries of he world, bu even here i is unlikely ha i is perfecly mobile. We noed above ha he dollar reurn ha an American earns on a foreign bond depends on he fuure exchange rae. Because fuure exchange raes are no known wih cerainy, his makes holding foreign bonds riskier han holding American bonds for someone who wans o earn a reurn in erms of dollars. 9 If wealh-holders have some degree of preference for one currency over anoher, hen domesic and foreign bonds are less-han-perfec subsiues. 9 I is worh noing however ha he real reurn on he foreign bond could be less risky if he domesic inflaion rae is highly uncerain while he foreign inflaion rae is sable. 9 18

19 Romer models imperfec capial mobiliy wih a capial-flow funcion given by his equaion (5.22). Capial will end o flow ino he domesic counry when ineres raes are high relaive o foreign raes and flow ou when domesic ineres raes are low. You can hink of he CF funcion as a ne demand curve for he domesic counry s asses. The case of perfec capial mobiliy is he special case in which CF a r = r*. In his case, he domesic and foreign bonds are perfec subsiues so he demand curve is perfecly elasic. E. Unemploymen and he Phillips curve The aggregae-supply models surveyed above ypically imply ha aggregae demand shocks lead o a posiive shor-run relaionship beween prices (or inflaion) and real oupu. Does his imply a negaive relaionship beween inflaion and unemploymen? Convenional wisdom suggess ha unemploymen is srongly and inversely correlaed wih oupu over he business cycle. I is emping, hen, o simply leap from an upward-sloping shor-run aggregae supply curve o a downward-sloping shor-run Phillips curve relaing unemploymen and inflaion. In his secion, we consider wheher his is a reasonable inference and review some of he hisory of he Phillips curve. Oupu, employmen, and unemploymen We ypically model firms as being on heir producion funcions, meaning ha hey are producing he maximum oupu ha hey can, given he inpus hey are using and he echnology ha hey have. If echnology (he producion funcion iself plus any A parameer we migh inroduce o represen echnology) and he capial sock are fixed, his implies a direc, one-o-one relaionship beween oupu and employmen given byy F L. 10 This means ha employmen and oupu mus move ogeher over he cycle, so any change in oupu is accompanied by a change in employmen. However, his change in employmen need no imply a change in he unemploymen rae. Changes in he number of workers employed could reflec movemens ino and ou of he labor force raher han movemens beween employmen and unemploymen. Only if he labor force is held consan hen a change in employmen implies a 10 In Chaper 13 of he coursebook, we consider he issue of varying uilizaion of labor and capial. In paricular, we examine evidence suggesing ha firms hold ono, or hoard, labor when hey reduce oupu during a recession. Such behavior would break he igh, producionfuncion relaionship beween oupu and labor inpu over he business cycle. 9 19

20 one-o-one change in he opposie direcion in unemploymen. In his case, i is reasonable o hink of he aggregae supply curve and he Phillips curve as differen ways o elling he same sory. However, we mus be careful in carrying his sory oo far. I reas workers as mere pawns of employers and assumes ha hey make no real decisions for hemselves. When laid off, hey jus si around being involunarily unemployed and waiing o be rehired. There are cerainly examples of his kind of unemploymen, bu i is less common han you migh hink. Mos unemployed workers, including hose who have been laid off, make imporan decisions ha affec heir job saus. They may drop ou of he labor force (early reiremen, reurning o school, engaging in home producion or family aciviies) or aggressively pursue oher employmen opions. Macroeconomiss and labor economiss have developed a rich se of heories abou he behavior of unemployed workers. These heories, a few of which will occupy our aenion in Romer s Chaper 11, sugges ha we mus be cauious in approaching he relaionship beween flucuaions in oupu and hose in unemploymen. The original Phillips curve The Phillips curve was originally proposed as an empirical regulariy by A.W. Phillips, an Ausralian economis. Phillips (1958) ploed nearly a cenury of daa on he unemploymen rae and he rae of wage inflaion for Briain and found ha he daa poins raced ou a downward-sloping curve ha appeared o be sable over his very long sample period. Over he nex decade, economiss examined he Phillips curve on boh heoreical and empirical levels. Empirically, Phillips s curve was found o be robus o a number of changes: a similar curve held for he Unied Saes and he same kind of relaionship held beween price inflaion and unemploymen. Theoreically, a simple explanaion for he Phillips curve was quickly devised. I was assumed o be he resul of parial adjusmen of wages oward equilibrium in response o excess demand or supply in he labor marke. When unemploymen was high, here was excess labor supply and wages would fall (or rise less quickly); low unemploymen indicaed excess demand for labor, which would drive wages upward. The naural-rae hypohesis This is an enirely reasonable heory, as long as cerain oher facors are held consan. Milon Friedman, in his now-famous presidenial address o he American Economic Associaion in December 1967, prediced ha he Phillips relaionship could 9 20

21 no be relied upon o remain sable because i confused nominal and real wages. 11 Low unemploymen lower han wha Friedman defined as he naural rae should lead o an increase in real wages. If he general level of inflaion in he economy is zero, hen an increase in real wages is equivalen o an increase in nominal wages. However, in an economy wih, say, 10 percen general inflaion, a nominal wage rise of more han 10 percen would be required o raise real wages. 12 Friedman hus argued ha unemploymen should be relaed no o he rae of (wage) inflaion in nominal erms, bu o inflaion relaive o people s expecaions. If people expec 10 percen inflaion, hen low unemploymen should lead o inflaion greaer han 10 percen and high unemploymen should lead o inflaion below 10 percen. This heory, which is ofen called he naural-rae hypohesis, implies ha here is no sable relaionship beween inflaion and unemploymen in a world where expecaions abou inflaion are no anchored a zero. Raher, here is a relaionship, which may be sable, beween unexpeced inflaion and unemploymen relaive o is naural rae. The Phillips curve beween acual inflaion and acual unemploymen should shif whenever here is a change eiher in he expeced rae of inflaion or in he microeconomic facors ha deermine he naural rae of unemploymen. Moreover, he naural-rae hypohesis implies no radeoff beween inflaion and unemploymen a all in he long run. No maer how high he inflaion rae, if i persiss for a while people in he economy will evenually adjus heir expecaions in order o anicipae i correcly. A ha poin, here will be no gap beween acual and expeced inflaion and he unemploymen rae should reurn o he naural rae. Thus, any rae of inflaion is consisen wih he naural rae of unemploymen in he long run. A look a Romer s Figure 6.7 shows ha Friedman s predicion abou he breakdown of he empirical Phillips curve came rue shorly afer his wriing. The daa poins beween 1961 and 1969 line up wih he sable, downward-sloping curve ha Phillips found for However, in 1970 he economy moved direcly o he righ and i did no reurn o he same neighborhood of low inflaion and low unemploymen unil he lae 1990s. From 1970 o 1995, he Unied Saes (and mos oher major economies) suffered hrough sagflaion he simulaneous occurrence of high unemploymen and high inflaion. The naural-rae hypohesis inerpres he swirling paern in Figure 6.7 as resuling from shifs in expeced inflaion and in he naural rae of unemploymen. As people 11 This paper was published as Friedman (1968). Anoher imporan se of early papers on he modern heory of he Phillips curve is Phelps (1970). 12 The naural rae of unemploymen is someimes called he non-acceleraing inflaion rae of unemploymen, or NAIRU. 9 21

22 began o cach on o he presence of inflaion, he Phillips curve shifed verically upward, implying a higher rae of inflaion for any level of unemploymen. As he babyboom generaion flooded he labor marke wih young, inexperienced, and ofen ransien workers, he number of workers searching for beer jobs naurally increased. 13 This increase in he naural rae of unemploymen shifed he Phillips curve o he righ. One quesion ha can fairly be posed o advocaes of he naural-rae hypohesis is how he Phillips curve could have remained sable for one hundred years. Is i really credible ha expeced inflaion and naural unemploymen anchored a sable Phillips curve for a cenury, only o sar wandering all over he map in 1970? The sabiliy of inflaionary expecaions is acually quie plausible. For mos of he cenury ha Phillips sudied, England and he major economies of he world were on some form of he gold sandard. This mainained a long-run link beween he value of he currency and he price of gold, which prevened seady, ongoing inflaion from occurring. Indeed, he consumer price index in he Unied Saes was a abou he same level afer World War II as i was in 1800! Ongoing, and herefore expeced, inflaion did no arise in he Unied Saes or Briain unil he 1960s, which explains why he Phillips curve did no begin o shif unil abou Modern challenges o he naural-rae hypohesis By he end of he 1970s, he naural-rae hypohesis had become he new macroeconomic orhodoxy. All of he inermediae macro exs and mos of he inroducory exs had been rewrien o reflec he new heory. Theoreical developmen and empirical esing proceeded apace and generally suppored he hypohesis. While few, if any, macroeconomiss believe in he old Phillips curve, mild challenges o he nauralrae hypohesis have come from several direcions. You may noice ha Romer is very careful on page 259 o disinguish beween his concep of core inflaion or underlying inflaion and a sric noion of expeced inflaion. Early heories of he expecaions-augmened Phillips curve based on he naural-rae hypohesis sressed he imporance of expecaions errors resuling from imperfec informaion. 14 According o hese heories, all deviaions of oupu and unemploymen from heir naural levels could be eliminaed if households and firms could correcly 13 The baby boom is only one explanaion for he increase in naural unemploymen in he 1970s and 1980s. We shall sudy ohers in due course. 14 The crucial role of expecaions and, in paricular, he developmen of he heory under condiions of raional expecaions follows from he work of Rober Lucas. We shall sudy a varian of Lucas s imperfec informaion model in Romer s Chaper 6 Par B and in Chaper 10 of he coursebook. 9 22

23 perceive and forecas he inflaion rae. This came o be known as he new classical macroeconomics. In conras, new Keynesian macroeconomiss, of whom David Romer is one, have developed models wih aribues ha are similar in many ways o he expecaionsbased naural-rae hypohesis, bu ha are based on sickiness of prices or wages raher han on imperfec informaion. Romer s core inflaion reflecs he ineria ha inflaion may acquire when i is embodied in cos of living adjusmens in wage conracs and in he insiuional process of price seing. Expecaions are cerainly a major par of core inflaion, bu here may be elemens o core inflaion ha are more difficul o change han expecaions, which can adap very quickly once people obain credible new informaion. Thus, he firs challenge akes he form of a broadening of he benchmark inflaion rae ha moves he Phillips curve from a sric expecaion o a more inclusive core inflaion rae. The cenral premise of he naural-rae hypohesis is ha he naural rae iself is independen of inflaion. I may shif due o microeconomic facors such as changes in he skill-composiion of he job pool and he labor force, changes in policies such as minimum wages or unemploymen insurance, or changes in he srengh and behavior of labor unions, bu i assumes ha changes in inflaion have no effec on he naural rae. However, wo heories have recenly suggesed ways ha inflaion could affec he naural rae of unemploymen. The firs is he hypohesis of hyseresis in unemploymen. According o he hyseresis heory, periods of high unemploymen, such as would resul from prolonged disinflaions, would cause he naural rae iself o increase. Among he reasons why his migh occur are deerioraion of relevan job skills by he long-erm unemployed and disenfranchisemen of unemployed ousiders in he process of negoiaing wages and employmen levels. 15 Hyseresis is a possible explanaion for he experience of coninenal Europe since 1980, where unemploymen has been well above hisorical levels for more han hree decades. While inflaion has been quie low, i seems implausible ha core inflaion would no have adjused o he lower inflaion rae wihin a few years. Thus, i is unlikely ha unemploymen has been above he naural rae all his ime. Insead, i seems probable ha he naural rae iself is higher. The hyseresis heory proposes ha high unemploymen iself has caused he naural rae o rise. A more recen challenge o he naural-rae hypohesis proposes ha here may be a lefward bulge a low inflaion raes in he oherwise verical long-run Phillips curve. The raionale behind he bulge heory is ha when inflaion is low, people may ignore 15 The hyseresis hypohesis is discussed briefly in Romer s Chaper 10 and in he unemploymen chaper of he coursebook. A readable exposiion from he research lieraure is Blanchard and Summers (1986). 9 23

24 i alogeher and behave as hough core or expeced inflaion is zero. 16 If inflaionary expecaions do no adjus o permanen changes in inflaion ha say near zero, hen he downward-sloping original Phillips may be valid in ha range. Inflaion of, say, 2 percen migh lead o permanenly lower unemploymen han zero inflaion if people ignore he inflaion. This heory could explain how he Unied Saes was able o achieve and susain remarkably low unemploymen raes in he lae 1990s wih seady bu low inflaion. F. Suggesions for Furher Reading The IS/LM and Mundell-Fleming models Mos inermediae macroeconomics exs have basic descripions of he IS/LM model. Some of he beer ones are lised below. No ediion numbers or publicaion daes are given because hey change very frequenly and almos any ediion will be suiable. Blanchard, Olivier, Macroeconomics, (Upper Saddle River, New Jersey: Prenice-Hall). Abel, Andrew, and Ben Bernanke, Macroeconomics, (Reading, Mass: Addison Wesley Longman). Hall, Rober E., and John B. Taylor, Macroeconomics, (New York, W.W. Noron). Dornbusch, Rudiger, and Sanley Fischer, Macroeconomics, (New York, McGraw- Hill). Burda, Michael, and Charles Wyplosz, Macroeconomics: A European Tex, (Oxford: Oxford Universiy Press). (A less Ameri-cenric presenaion wih good mahemaical appendices.) Sachs, Jeffrey, and Felipe Larrain, Macroeconomics in he Global Economy, (Upper Saddle River, New Jersey: Prenice-Hall). (A more open-economy presenaion.) The original presenaions of hese models are: Hicks, John R., Mr. Keynes and he Classics : A Suggesed Inerpreaion, Economerica 5(2), April 1937, (The original exposiion of he IS/LM model.) Hansen, Alvin H., A Guide o Keynes (New York: McGraw-Hill, 1953). (An early full exposiion of he IS/LM framework as an inerpreaion of Keynes.) Mundell, Rober A., Inernaional Economics (New York: Macmillan, 1968). (A seminal exposiion of he Mundell-Fleming model.) 16 See Akerlof, Dickens, and Perry (2000). 9 24