Macroeconomic Perspectives on Inflation and Unemployment

Ni zan, Jona han Wo king Pape Mac oeconomic Pe spec i es on In la ion and Unemploymen Discussion Pape s, Depa men o Economics, McGill Uni e si y P o ided in Coope a ion wi h: The Bichle & Ni zan A chi es Sugges ed Ci a ion: Ni zan, Jona han (1990) : Mac oeconomic Pe spec i es on In la ion and Unemploymen , Discussion Pape s, Depa men o Economics, McGill Uni e si y, The Bichle and Ni zan A chi es, To on o, h p://bna chi es.yo ku.ca/159/ This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/157850 S anda d-Nu zungsbedingungen: Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen Zwecken und zum P i a geb auch gespeiche und kopie we den. Sie dü en die Dokumen e nich ü ö en liche ode komme zielle Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich machen, e eiben ode ande wei ig nu zen. So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen (insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en, gel en abweichend on diesen Nu zungsbedingungen die in de do genann en Lizenz gewäh en Nu zungs ech e. Te ms o use: Documen s in EconS o may be sa ed and copied o you pe sonal and schola ly pu poses. You a e no o copy documen s o public o comme cial pu poses, o exhibi he documen s publicly, o make hem publicly a ailable on he in e ne , o o dis ibu e o o he wise use he documen s in public. I he documen s ha e been made a ailable unde an Open Con en Licence (especially C ea i e Commons Licences), you may exe cise u he usage igh s as speci ied in he indica ed licence. h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ Depa men o Economics Discussion Pape 1990 MACROECONOMIC PERSPECTIVES ON INFLATION AND UNEMPLOYMENT Jona han Ni zan This pape is a d a o wo k in p og ess Depa men o Economics McGill Uni e si y 855 She b ooke S . Wes Mon eal, Quebec H3A 2T7 Con en s Abs ac 1 In oduc ion 2 1. The O iginal Philips Cu e 3 2. F om Disequilib ium o Equilib ium 5 3. Pe ec Compe i ion? 8 4. An Agg ega e View o Ma ke ‘Impe ec ions’ 11 5. Expec a ions: Economic Agen s S ike Back 16 6. In Ques o In o ma ion: The Unemployed as an In es o 22 7. A Ra ional Expec a ions ‘Re olu ion’? 28 8. ‘Ins i u ional Ins abili y’ and S ag la ion 36 9. The S ag la iona y Menace o ‘Exogenous Fo ces’ 37 10. Concluding Commen s 43 Re e ences 44 Abs ac The p esen pape is he i s in a se ies o h ee essays in which we examine he mac oeconomic and s uc u al app oaches o in la ion. In his pape we explo e some o he key con ibu ions o he mac oeconomic li e a u e which appea ed since he la e 1950s. Much o his li e a u e e ol ed in a dual lo e-ha e ela ionship wi h he Phillips Cu e. Schola s who endo sed he Phillips Cu e on he basis o his o ical e idence we e su p ised when i s a ed o c umble as soon as hey assimila ed i in o hei mac oeconomic models. The g adual eme gence o s ag la ion and he p og essi e b eakdown o he Phillips Cu e p esen ed mains eam mac oeconomics wi h he mos se ious challenge since he Second Wo ld Wa . Mac oeconomis s a acked he Phillips Cu e bu hei c i icisms sough o modi y, no nulli y. The idea ha in la ion and unemploymen we e in e sely ela ed was appa en ly oo signi ican o disca d so he no ional ela ionship was simply ‘augmen ed’ by auxilia y ac o s. The cos o sa ing he Phillips Cu e was subs an ial. To explain s ag la ion, mac oeconomis s eso ed o ‘disequilib ia,’ ‘ igidi ies’ and ‘exogenous shocks’ and hey abandoned, a leas empo a ily, he ideal o mula ion o he neoclassical syn hesis. 1 In oduc ion Mode n mac oeconomic heo ies o in la ion and unemploymen ha e e ol ed in a dual lo e-ha e ela ionship wi h he Phillips Cu e. The no ion ha he e exis s a s able in e se ela ionship be ween in la ion and he a e o unemploymen – dubbed as he Phillips Cu e a e he o iginal wo k by Phillips (1957) – was assimila ed in o mac oeconomic models du ing he 1960s. The heo e ical ela ionship was suppo ed by obse a ions s e ching o e close o a cen u y, ye , as soon as mac oeconomis s pu hei new disco e y in o use, he Phillips Cu e seemed o b eak down! Du ing he la e 1950s, when Phillips published his o iginal a icle on he B i ish expe ience, many mac oeconomis s in he Uni ed S a es we e pe plexed by he pe sis ence o in la ion in he mids o ecession. La e , du ing he 1960s, in la ion in mos ad anced capi alis economies accele a ed wi h no appa en decline in he a e o unemploymen . Finally, since he mid 1970s, a e a d ama ic ise in bo h in la ion and unemploymen , he wo a iables began o mo ed oge he , in an open de iance o he Phillips Cu e. The g adual eme gence o s ag la ion and he p og essi e b eakdown o he Phillips-Cu e ela ionship p esen ed mains eam mac oeconomics wi h he mos se ious challenge since he Second Wo ld Wa . Mac oeconomis s launched a se ies o bi e a acks on he Phillips Cu e, ye hei c i icism sough o modi y, no nulli y. Behind he heo e ical Phillips Cu e lay s ong neoclassical con ic ions ega ding he wo king o supply and demand. Al hough mac oeconomics abs ac ed om he s uc u e o unde lying ma ke s, he nega i e associa ion be ween in la ion and unemploymen seemed o indica e ha pe ec compe i ion was a use ul assump ion in he s udy o b oad agg ega es. The basic ela ionship be ween in la ion and unemploymen was simply oo signi ican o disca d. As a esul , mos mac oeconomic challenges o he Phillips Cu e ha e been hal -hea ed: hey ‘augmen ed’ he elemen a y ela ionship wi h auxilia y ac o s. By he ea ly 1990s, a e h ee decades o heo e ical challenges, mac oeconomic heo ies o in la ion and unemploymen s ill domina e he collec i e consciousness o economis s and policy make s alike. In his sense, he s uggle o sa e he Phillips Cu e has been success ul. Ye he achie emen came a conside able cos . Amendmen s o he Phillips Cu e we e ne e qui e su icien and addi ional modi ica ions we e cons an ly called o in o de o accommoda e changing eali ies. This epea ed ‘augmen a ion’ o he Phillips Cu e inju ed he appa en in eg i y o mac oeconomics. The mos se ious damage, howe e , was caused by he na u e o modi ica ions. In o de o explain he b eakdown o he Phillips Cu e, mac oeconomis s eso ed o ad e si ies such as ‘disequilib ia,’ s uc u al and in o ma ional ‘impe ec ions,’ and ex e nal ‘shocks’ deli e ed om ou side he mac oeconomic sys em. In o he wo ds, hey abandoned he ca dinal belie in equilib ium and pe ec compe i ion which p e iously cha ac e ized he ‘neoclassical syn hesis.’ 2 In his essay we deal wi h some o he key con ibu ions o he mac oeconomic li e a u e on in la ion and unemploymen . Ou aim is no o p o ide a comp ehensi e o e en a pa ial su ey. Ins ead, we ocus ou a en ion on undamen al me hodological issues which a ise as mac oeconomis s lea e he ideal neoclassical domain o pe ec compe i ion and equilib ium and en u e in o al e na i e e ain. The i s and second sec ions deal wi h he o iginal Phillips Cu e and i s heo e ical ounda ions. In he hi d sec ion, we mo e om he labou ma ke in o he mac oeconomic a ena. The ou h sec ion deals wi h he no ion o s uc u al impe ec ions. The i h and six h sec ions examine he in eg a ion o expec a ions and he na u al a e o unemploymen in o he Phillips-Cu e amewo k. In he se en h sec ion, we app aise he a ional-expec a ions amewo k. The eigh sec ion e alua es he e ec o ins i u ional ins abili y on s ag la ion and, in he nin h sec ion, we explo e he no ion o supply shocks. We conclude wi h se e al commen s. 1. The O iginal Phillips Cu e In 1958, A.W. Phillips published a ca e ul empi ical s udy examining he ela ion be ween unemploymen and wage in la ion in he Uni ed Kingdom o e a pe iod ex ending om 1861 o 1957. Fi s he i ed a nonlinea unc ion, nega i ely ela ing wage in la ion o he a e o unemploymen be ween 1861 and 1913, and hen he demons a ed how his unc ion could explain he ela ionship o he subsequen pe iod be ween 1913 and 1957. The s ylized, s able ela ionship sugges ed ha a 5.5 pe cen o unemploymen was associa ed wi h ze o wage in la ion. When unemploymen was abo e his h eshold, he e was a modes decline in nominal wages. On he o he hand, when unemploymen was below 5.5 pe cen he a e o wage in la ion inc eased apidly.1 Phillips also iden i ied coun e -clockwise ‘loops’ o da a obse a ions a ound he s ylized i ed unc ion. These loops indica ed ha when he a e o unemploymen was alling, wage in la ion exceeded he alue gi en by he unc ion and when unemploymen was g owing, he a e o change o wages was lowe han alues p edic ed by he unc ion. Phillips’ esul s we e assimila ed quickly, pa ly because hey p o ided s ong con i ma ion o he wo king o compe i i e ma ke o ces, pa icula ly o he way p ices adjus ed o ‘excess demand’ o ‘excess supply.’ The en a i e heo e ical 1 Phillips (1958, p. 290) i ed he ollowing unc ion o his da a: log (w + a) = log b + c log U , whe e w deno ed he a e o change o wage a es and U measu ed he pe cen age unemploymen . The es ima ed alues o he pa ame e s we e 0.9 o a, 9.638 o b and 1.394 o c. 3 hypo hesis o his adjus men p ocess is s a ed explici ly in Phillips’ opening passage (1958, p. 283): When he demand o a commodi y o se ice is high ela i e o he supply o i we expec he p ice o ise, he a e o he ise being g ea e he g ea e he excess demand. Con e sely when he demand is low ela i ely o he supply we expec he p ice o all, he a e o he all being g ea e he g ea e he de iciency o demand. I seems plausible ha his p inciple should ope a e as one o he ac o s de e mining he a e o change o money wage a es, which a e he p ice o labou se ices. Hence, i ollows ha i he a e o unemploymen and i s i s de i a i e a e aken as wo independen p oxies o ‘excess supply’ in he labou ma ke , bo h should be nega i ely ela ed o he a e o change in money wages. The a e o unemploymen could explain wage in la ion along he nega i ely-sloped Phillips Cu e and he a e o change in unemploymen would accoun o he coun e -clockwise loops a ound i . Mos o he ea ly li e a u e ha ollowed Phillips’ o iginal s udy emphasized his s ylized ela ionship be ween wage in la ion and unemploymen bu he Phillips Cu e was signi ican also o wha i ailed o explain. In ac , Phillips ook g ea pain o explain e e y de ia ion om he s ylized loop. His explana ions a e in e es ing because hey poin o s uc u al elemen s ha a e inconsis en wi h he assump ion o pe ec compe i ion in labou and commodi y ma ke s. Se e al examples could be ci ed o illus a e his poin . In he upswing be ween 1893 and 1896, o ins ance, wage a es ose mo e slowly han usual, a de elopmen ha Phillips (p. 292) a ibu ed o he apid g ow h o employe s’ ede a ions and he consequen ise in employe s’ esis ance o ade-union demands. Simila ly, he egula ela ionship was again dis u bed in 1912, p esumably by s ike ac i i y o union membe s in he coal-mining indus y (ibid.). Ano he obse a ion was he p og essi e na owing o he cyclical loops be ween 1861 and 1909. Phillips (pp. 292-93) explained his in wo ways; i s , by he p oli e a ion o wage-indexa ion and, second, by inc easing ime lags in he esponse o wage changes o changes in he le el o unemploymen . The signi icance o hese lags, he a gued, inc eased wi h he his o ical ex ension o collec i e ba gaining and a bi a ion. Ano he illus a ion (pp. 293-94) poin s o he d ama ic decline o wages in 1921 and 1922 (22.2 and 19.1 pe cen , espec i ely) which exceeded by a he mode a e dec eases sugges ed by he i ed cu e. Phillips a ibu ed much o hese declines o au oma ic cos -o -li ing adjus men s igge ed by subs an ial dec eases o impo p ices in hose yea s. Finally, he obse a ions o he 1948-1957 pe iod appea ed o gene a e a e e se loop, which Phillips (pp. 297-98) again explained by a lagged adjus men o wage a es o unemploymen . The e is a common ea u e in hese ealis ic supplemen a y explana ions. Employe s’ ede a ions, ade unions, collec i e ba gaining, a bi a ions, 4 wage-indexa ion and lagged adjus men s can be pe cei ed as ‘ins i u ional igidi ies’ ha dis o he unc ioning o a laissez ai e ma ke sys em. In his sense, by ecognizing such ins i u ional eali ies, Phillips an icipa ed he subsequen dilemma ha la e mac oeconomis s o en aced when hey ied o ela e he Phillips Cu e o a changing wo ld. The cos o being able o explain ising in la ion (and, subsequen ly, s ag la ion) in ol ed sac i icing he heo e ical ‘ideal’ o pe ec compe i ion as i s p is ine simplici y was inc easingly ain ed by a ious ealis ic social and ins i u ional ‘dis o ions.’ 2. F om Disequilib ium o Equilib ium The heo e ical unde pinning o Phillips’ empi ical indings was de eloped by Lipsey (1960). ‘The usual a gumen ,’ w i es Lipsey (p. 13), ‘me ely s a es ha when he e is excess demand . . . wage a es will ise, while when he e is excess supply . . . wages will all. No hing is said abou he speed a which he adjus men akes place.’ In o he wo ds, a heo e ical amewo k whe e disequilib ium gene a es equilib a ing o ces is incomple e unless we speci y a dynamic ‘adjus men mechanism’ o explain he speed a which he sys em mo es owa d equilib ium. Phillips indeed sugges ed ha wage in la ion was posi i ely co ela ed wi h he magni ude o excess demand bu , acco ding o Lipsey (p. 2), he had no p o ided a ‘model o ma ke beha iou ’ ha explained his ela ionship. Hence, in o de o elimina e he po en ial o se ious misin e p e a ion, he model unde lying he Phillips Cu e mus be ‘ ully speci ied’ (Lipsey, p. 12). In ligh o his emphasis on igo ous speci ica ion i is in e es ing o no e ha Lipsey does no speci y he unde lying ma ke s uc u e o his own model. Ins ead, he w i es: We shall conside his ela ionship, i s , o a single ma ke , and hen o he whole economy. . . . We migh analyze he ma ke o any commodi y since he a gumen a his s age is qui e gene al. Since, howe e , he subjec o he p esen a icle is he labou ma ke we shall use he e minology app op ia e o ha ma ke . (pp. 12-3, emphases added) The use o such ambiguous language is unhelp ul o i is ha d o imagine a ‘gene al’ model o p ice adjus men ha can be applied o ‘any’ ma ke s uc u e. The emphasis Lipsey pu s on he ole o ‘excess supply’ and ‘excess demand’ sugges s ha his own model may be applicable o pe ec compe i ion bu is p obably inadequa e o o he s uc u es.2 2 In monopoly and monopolis ic compe i ion he e is no unique supply cu e (supply depends on demand condi ions) and in oligopoly he meaning o bo h supply and demand cu es is ambiguous. Unde hese condi ions he e is no clea de ini ion o excess supply o demand. 5 The model o he single ma ke con ains h ee basic ela ions. One is he ‘adjus men mechanism’ which speci ies he a e o change o wages as a linea unc ion o he ela i e excess demand o labou : (1) w = α [(d - s) ÷ s] , whe e w deno es he a e o change o wages, α is a ixed coe icien , d is he demand o labou and s is he supply o labou . The second ela ion is a cu ilinea , nega i e unc ion linking he a e o unemploymen wi h he ela i e excess demand: (2) U = 1 [(d - s) ÷ s] , Equa ion (2) me ely desc ibes he ela ionship be ween he a e o unemploymen and ela i e excess demand, and i has no causal implica ions. When he ma ke is in equilib ium (no excess demand o supply), he e is only ‘ ic ional unemploymen ,’ wi h numbe o acancies being jus equal o he numbe o unemployed wo ke s. When excess supply de elops, unemploymen inc eases linea ly, while an inc ease in excess demand is associa ed wi h a cu ilinea all in unemploymen (as excess demand inc eases, he all in unemploymen becomes p og essi ely smalle because unemploymen canno become nega i e). The hi d ela ionship is he ‘adjus men unc ion’ which is de i ed by combining equa ions (1) and (2): (3) w = α 2 (U) . This las equa ion is, o cou se, he s anda d Phillips Cu e. I is in e es ing o examine he me hodology employed in de eloping his model because some o i s ea u es eappea in subsequen mac oeconomic heo ies o in la ion and s ag la ion. Two aspec s a e wo h no ing: he cen al ole assigned o non-obse able a iables, and he iew ha ma ke s con inuously mo e owa d equilib ium. We conside each o hese elemen s in u n. Acco ding o Lipsey, he i s p oblem o analysis s ems om he con inuous shi s o demand and supply cu es, mo emen s which make di icul he iden i ica ion o hese indi idual cu es. Fo una ely, he a gues, his is no an insu moun able obs acle o , in o de o ob ain Equa ion (1), i is ‘only necessa y o know demand and supply’ a he exis ing ma ke p ice and o he poin s on he cu es can be igno ed (p. 13, emphasis added). No e ha e en wi h his quali ica ion, one may s ill ask he p ac ical ques ion as o how we could disco e hese wo magni udes. The heo e ical analysis is cas in e ms o supply and demand; ha is, in e ms o desi es, o plans o sell and buy labou se ices. These a e psychological 6 endencies, no obse able ma ke ou comes. In his ligh i is unclea how could we sol e he p oblem by limi ing ou sel es o he exis ing ma ke p ice.3 Reliance on non-obse able magni udes in oduces a s ong axioma ic elemen in o he analysis. Lipsey (p. 13) asse s ha in o de o obse e he linea ela ion illus a ed in Equa ion (1), ‘i is necessa y only ha he e be an unchanging adjus men mechanism in he ma ke .’ Un o una ely, e en wi hin Lipsey’s own amewo k, his is only a necessa y and no a su icien condi ion, o in o de o obse e his ela ion we mus i s be able o obse e he ‘excess demand’ a iable. This is no always possible, admi s Lipsey, bu o p ac ical no concep ual easons. In his opinion, he di e ence be ween he numbe o un illed acancies and he numbe o unemployed wo ke s could p o ide a ‘ easonable di ec measu emen o excess demand’ bu , un o una ely, acancy da a a e seldom a ailable and e en when hey a e a ailable hese da a migh be se iously lawed. As a p ac ical solu ion, Lipsey sugges s we ela e excess demand only o unemploymen ( a he han o he di e ence be ween acancies and unemploymen ). The solu ion is no e y help ul, howe e , because Lipsey igno es he concep ual challenge al oge he . His ‘empi ical’ de ini ion o supply and demand in he labou ma ke is speci ied in e ms o ac ual ma ke ou comes a he han in e ms o hypo he ical desi es. The quan i y supplied is assumed o be equal o he sum o eco ded employmen and unemploymen , while he quan i y demanded is assumed o be equal o he sum o ac ual employmen and acancies. Hence, he di e ence be ween obse ed unemploymen and acancies is equal o ‘excess demand’ by de ini ion. In his ligh , eliance on unemploymen igu es o wan o acancies da a does no sol e any hing o i me ely inse s an axioma ic link – a nega i e cu ilinea unc ion be ween obse able unemploymen and non-obse able excess demand – in he heo e ical chain. The exis ence o hese axioma ic elemen s weakens he scien i ic s a us o Lipsey’s heo y. One eason o de eloping his heo y in he i s place is ha i he ela ion ceases o hold, o changes, and we ha e no model o explain i , we can only say ‘ he ela ion has ceased o hold’ o ‘ he ela ion has changed’ and we will ha e lea ned no hing mo e han his. I we ha e a model explaining he ela ionship, we will know he condi ions unde which he ela ion is expec ed o emain unchanged. Then, i a change occu s, he model will p edic why his has happened and his p edic ion will gi e ise o u he es s om which we can lea n. (Lipsey, p. 12) 3 One could a gue ha ‘quan i y supplied’ and ‘quan i y demanded’ a e obse able when he exis ing ma ke p ice is an equilib ium one because, when we de ine equilib ium as a ‘chosen posi ion,’ we asse ha he ac ual ou come is iden ical o he desi ed one. (See Asimakopulos, 1978, p. 43.) This easoning, howe e , is qui e misleading. Equilib ium he e is de ined in e e ence o desi es and no he o he way a ound, and unless we could i s obse e hose desi es we could no know ha he ma ke is indeed in equilib ium! 7 The analysis indica es ha policy make s may ha e mo e lexibili y han ini ially assumed by Samuelson and Solow (1960). They can be sa is ied wi h an exis ing in la ion-unemploymen adeo bu hey can also a emp o imp o e i . Acco ding o Pe y (ch. 5), his can be done by a ec ing he a iables o coe icien s in equa ions (2) and (3). Fo example, go e nmen s can educe co po a e ax- a es o accele a e dep ecia ion schedules in o de o main ain exis ing cash- lows wi h a lowe p e- ax a e o p o i , o hey can y o encou age p oduc i i y g ow h. They can also change he ins i u ional s uc u e o wage and p ice de e mina ion by educing he monopoly powe o unions and i ms, o by ying o pe suade he gene al public owa d a g ea e es ain . Pe y’s model su e s om se e al sho comings which a ise because he acknowledges he signi icance o economic s uc u e bu hen ails o deal wi h i e ec i ely. Fi s , he wage equa ions does no seem o e lec ma ke ‘impe ec ions’ in any clea way. As we a gued ea lie , he obse ed a e o unemploymen is no necessa ily equi alen o he non-obse able alues o excess demand and, hence, he e is oom o o he a iables in explaining wage in la ion e en unde pe ec compe i ion. Second, Pe y’s explana ion o p ice in la ion is no cons uc ed as a es able hypo hesis bu is a he based on he simple assump ion ha he agg ega e p ice le el is de e mined as a ixed ma kup o e cos . Un o una ely, his assump ion seems unwa an ed o bo h heo e ical and empi ical easons. The s anda d heo y o he i m usually emphasizes he ul ima e goal o maximizing e u n on in es men . In his con ex , he ma kup is ei he an insigni ican co olla y o p o i maximiza ion o a means owa d his end, bu he e is no eason o assume i is cons an .6 I seems only plausible o changes in he a e o p o i o a ec he ma kup. Indeed, why should i ms be willing o g an la ge wage inc eases ha lowe hei ma kup when hei a e o p o i inc eases, bu no a emp o aise hei ma kup a e he a e o p o i alls? The empi ical da a o mos capi alis economies clea ly indica e ha ma kups o p ice o e p ime-cos luc ua e h ough ime. Unde hese condi ions, why would one s ill insis on a ixed-ma kup assump ion? Pe y p o ides no explici answe o his ques ion bu no es ha i is he ‘neu al s anda d’ (p. 64). In o he wo ds, by assuming a ixed ma kup we imply ha in la ion has no e ec on income dis ibu ion (i is ‘neu al’ in his sense) and a se ious complica ion is esol ed be o e i e en 6 In he model o pe ec compe i ion, i ms a e p ice ake s no p ice make s. When he ma ke p ice changes hey al e hei ou pu in o de o equa e he new p ice wi h hei ma ginal cos , bu his also causes he a e age ma kup o change. In he long un, pe ec ly compe i i e i ms ealloca e hei capi al and p oduc ion o ollow he highes a e o p o i and his o en implies changes in he a e age ma kup. The s anda d model o monopoly also sugges s ha he ma kup changes wi h demand condi ions when he monopolis equa es ma ginal e enue and cos . Fo oligopolies, he esul s a e mo e ambiguous; when oligopolies compe e, in e dependency be ween hem may lead o any one o an in ini e numbe o possible ma kup le els, whe eas when hey coope a e, hey may se and al e he ma kup acco ding o some a bi a y ‘ a ge ’ a e o e u n. 14 a ises. Un o una ely, hese me hodological manoeu es a e qui e cos ly because hey in alida e mos o Pe y’s conclusions abou he Phillips Cu e adeo . His model indica es ha he oo o p ice in la ion is in he wage de e mina ion p ocess, bu ha may be ue only i we accep his assump ion o ixed ma kup p icing. O he wise, in he absence o a es able hypo hesis abou he ma kup, he p icing equa ion is incomple e and, hence, he Phillips-Cu e adeo be ween unemploymen and p ice in la ion is uns able.7 A hi d p oblem conce ns Pe y’s assump ion ha ‘agg ega e ela ionships exis ’ (p. 57). He ag ees ha wage de e mina ion in di e en indus ies may ely on di e en ac o s linked by di e en unc ional ela ionships, bu a gues ha hey can be sa ely igno ed om a mac oeconomic pe spec i e. This assump ion is unwa an ed and may lead o misleading empi ical esul s. Fo example, Pe y (pp. 30-31) s ipula es ha 1/4 h o all wage con ac s a e nego ia ed in each qua e , so he annual a i hme ic a e age o wage in la ion is a unc ion o annual a i hme ic a e ages o he ca ie s in Equa ion (1). This asse ion has no empi ical basis and, as Rowley and Wil on (1974) demons a e, he pa icula dis ibu ion o wage se lemen s h ough he yea has a d ama ic e ec on he sign o es ima ed coe icien s, hei magni ude and hei associa ed le els o signi icance. Accoun ing o o he aspec s o he e ogenei y (such as ypes o indus ial ac i i y o co po a e size) will only in oduce u he ins abili y in o Pe y’s model. The ou h p oblem we deal wi h is he assump ion ha unde lying ela ionships be ween he a iables a e s able. Pe y begins his disse a ion by disassocia ing himsel om he s able model o pe ec compe i ion and en u es owa d a g ea e ecogni ion o s uc u al ‘impe ec ions.’ He concludes his analysis by a guing ha he go e nmen can y o a ec he Phillips-Cu e adeo by al e ing he unde lying economic s uc u e. Howe e , i he go e nmen can a ec ins i u ional pa e ns o wages, p ices and p o i s why should we assume ha hese pa e ns a e s able o begin wi h? Fo example, o ha e a s able Phillips-Cu e ela ionship we need o ha e a s able a e o p o i and Pe y’s use o only ou di e en a es (10.0, 10.8, 11.8 and 12.5 pe cen ) may gi e he inco ec imp ession ha his a e is indeed s able. Acco ding o Figu e 3.7 (p. 48) howe e , he a e o p o i du ing he 1948-62 pe iod luc ua ed be ween 8 and 16 pe cen ! Un o una ely, he a e o p o i in Pe y’s model is ‘exogenously gi en’ and, hence, such empo al luc ua ions make i ha d o p edic in la ion and unemploymen , o design policy o imp o e he adeo 7 This po en ial ins abili y is heigh ened when Pe y (p. 64) ag ees ha ‘[a]c ual p ice beha io may no con o m o his s anda d’ and discusses he possible implica ions o de ia ions om a ixed ma kup. Fo ins ance, when hal o all p ices inc ease ‘au onomously’ by 2 pe cen (independen ly o changes in cos and p oduc i i y), he posi ion and slope o he Phillips-Cu e be ween unemploymen and in la ion a e al e ed. The p oblem, as Pe y (p. 68) admi s, is ha his esul is only hypo he ical and ‘has no empi ical ounda ion.’ 15 be ween hem.8 The sou ce o ins abili y is no limi ed o he a e o p o i . The pa ame e s in equa ion (2) and (3) a e also de e mined exogenously by he unde lying ins i u ional s uc u e and Pe y does no explain why hey should emain s able o e ime. These commen s indica e ha in o de o analyze he e ec s o ins i u ional s uc u es on agg ega e unemploymen and in la ion, we mus i s ca e ully analyze hese s uc u es, some hing ha Pe y ailed o do. The s udy by Pe y sugges ed ha he e was no one bu many po en ial ‘Phillips Cu es,’ each co esponding o a pa icula se o ins i u ional pa ame e s. These unde lying pa ame e s we e p esumed o be ela i ely s able and, unless he go e nmen a ec ed hei alues, he adeo be ween in la ion and unemploymen could emain s able o e a subs an ial pe iod o ime. This idea o s abili y was no unanimously accep ed. Se e al mains eam mac oeconomis s a gued ha indeed he e we e many po en ial Phillips Cu e, hough he eason o his mul iplici y was o be ound in in o ma ional, no s uc u al ‘impe ec ions.’ Fu he mo e, while he e we e many possible Phillips Cu es, all o hem we e inhe en ly uns able. 5. Expec a ions: Economic Agen s S ike Back F om he ea ly 1960s, many de eloped capi alis economies began o expe ience ising a es o in la ion wi h li le o no decline in he a e o unemploymen . This was a signi ican de elopmen because i pu in o ques ion he ime-honou ed link be ween sca ci y and p ice mo emen s. Was i possible o p ices and wages o be independen o excess supply o demand? Acco ding o F iedman (1968) and Phelps (1968) he answe was nega i e bu he eason was no ‘s uc u al impe ec ions.’ Phelps (p. 678) a gued ha mos exis ing explana ions o wage mo emen s (like ha o Pe y) con ained coun less independen a iables in nume ous combina ions and i was di icul o choose among he di e en models because hey o en lacked any clea a ionale. Ins ead, he sugges ed we mo e owa d a ‘uni ied and empi ically applicable heo y o money-wage dynamics,’ whe e indi idual ma ke s we e compe i i e bu economic ou comes we e s ill ‘dis o ed’ because he low o in o ma ion was impe ec . Acco ding o F iedman and Phelps, he e ical Phillips Cu e did no cons i u e an anomaly in economic heo y simply because he e y cons uc ion o his cu e in ol ed a basic con usion: unemploymen depended on eal, no nominal wages and p ices. The nuisance o economic heo y, w o e F iedman in his Noble Lec u e (1977, p. 12), was ha nominal and eal alues need no mo e oge he : Low unemploymen would, indeed, mean p essu e o a highe eal wage – bu eal wages could be highe e en i nominal wages we e lowe , p o ided 8 In his discussion o he dynamic p ope ies o his model, Pe y speci ies an equa ion o changes in he a e o p o i bu does no explain he a e o p o i i sel (pp. 90-2). 16 ha p ices we e s ill lowe . Simila ly, high unemploymen would, indeed, mean p essu e o a lowe eal wage – bu eal wages could be lowe , e en i nominal wages we e highe , p o ided p ices we e s ill highe . So why did ea lie obse a ions indica e ha he Phillips Cu e was nega i ely sloped? F iedman and Phelps answe ed his ques ion by making he cu e a special case wi hin a b oade heo e ical amewo k. The a gumen o he wo heo is s was simila and we ocus mainly on he wo k by F iedman (1968; 1977). Because in o ma ion ega ding employmen oppo uni ies and he a ailabili y o wo ke s is cos ly (S igle , 1961; 1962) and because wo ke s possess speci ic human capi al (Becke , 1964), employees and employe s en e in o explici o implici long- e m con ac s. Al hough bo h sides seek o denomina e hei ag eemen in eal e ms, mos labou con ac s a e signed in nominal dolla s.9 Consequen ly, he eal wage o e he li e o he con ac depends on an unknown u u e p ice le el. Unde hese condi ions, he desi ed nominal wage a e is se equal o he p oduc o he desi ed eal wage and he expec ed p ice index. The hallma k o he new heo y, hen, is his emphasis on p ice expec a ions o med by economic agen s. I agen s a e always success ul in co ec ly an icipa ing u u e p ices, he ealized eal wage is always equal he desi ed one. Since in his case he eal wage is independen o in la ion, i ollows ha he a e o unemploymen – which esponds only o he eal wage – is also independen o in la ion. In la ion ceases o be neu al, howe e , when economic agen s e in hei p edic ions. When p ice changes a e unan icipa ed, he ealized eal wage di e s om he desi ed eal wage ha is embodied in labou con ac s, and un il hese long- e m con ac s expi e, employmen and unemploymen de ia e om hei equilib ium ela ionship wi h eal wages. Why do e o s in expec a ions lead o a adeo be ween in la ion and unemploymen ? Acco ding o F iedman (1977, p. 13), he answe could be ound by examining how bo h wo ke s and employe s misin e p e he e ec o an unan icipa ed change in ma ke condi ions. Fo example, when he g ow h a e o nominal agg ega e demand inc eases unexpec edly, each p oduce eels his inc ease p ima ily h ough ising demand o his own commodi y. Al hough he e is an o e all expansion, he single p oduce myopically misin e p e s i as an imp o emen in his own ela i e posi ion. He belie es ha his own p ices will be ising as e han he o e all p ice le el and, hence, is willing o aise he wage a e o a ac addi ional wo ke s. Wo ke s all in a simila ap when hey belie e ha hei wages inc ease as e han p ices in gene al. As a esul , a ise in nominal wages may be pe cei ed by wo ke s as a ise in eal wages and hence call o h an inc eased supply, a he same ime ha i is pe cei ed 9 Some collec i e ag eemen s inco po a e a COLA clause bu he ela i e signi icance o such con ac s has o en been limi ed. Con ac s can also be ‘ eopened’ in special ci cums ances. 17 by employe s as a all in eal wages and hence calls o h an inc eased o e o jobs. (ibid.) Fo he economy as a whole he ne esul is a new posi ion wi h lowe unemploymen and highe wages and p ices. In o he wo ds, he economy mo es up and o he le on he g aph o he Phillips Cu e. Bu his new si ua ion is inhe en ly uns able because i is based on an open ‘lie.’ I nominal demand con inues o g ow a i s new highe pace, p oduce s and wo ke s will e en ually ealize hey ha e been ooled by he ma ke . The p ice o hei own commodi y is indeed ising bu so oo a e all o he p ices and, hence, he eal p ice o hei commodi y may no change a all! Wi h his new, co ec in o ma ion, unemploymen becomes a i icially low. As agen s adjus hei expec a ions and e ise hei con ac s o e lec he new a e o in la ion, he Phillips Cu e i sel mo es upwa d. The cu e will s abilize in i s new highe posi ion when all con ac s embody he new a e o in la ion. When his happens, he economy will e u n o i s o iginal, ‘na u al a e o unemploymen ’: A any momen o ime, he e is some le el o unemploymen which has he p ope y ha i is consis en wi h equilib ium in he s uc u e o eal wage a es . . . The ‘na u al a e o unemploymen ,’ in o he wo ds, is he le el ha would be g ound ou by he Wal asian sys em o gene al equilib ium equa ions, p o ided he e is embedded in hem he ac ual s uc u al cha ac e is ics o he labo and commodi y ma ke s, including ma ke impe ec ions, s ochas ic a iabili y in demands and supplies, he cos o ga he ing in o ma ion abou job acancies and labo a ailabili y, he cos s o mobili y, and so on. (F iedman, 1968, p. 8) Hence, he Phillips-Cu e adeo is only a empo a y ela ion based on he elemen o su p ise. The au ho i ies can use his adeo o educe unemploymen below i s ‘na u al a e’ only because hey can ool all he people some o he ime. Bu such e o s a e sel -de ea ing because no one can ool all he people all o he ime. E en ually, economic agen s will s ike back, o cing policy make s o cope wi h he o iginal le el o unemploymen coupled wi h a highe a e o in la ion. In he long un, he e is no adeo and he Phillips Cu e is e ical. Unemploymen can hus be kep below i s na u al a e only a he cos o accele a ing in la ion. The oo s o in la ion, hen, a e no ‘impe ec ions’ in he ma ke s uc u e. Such impe ec ions, o he ex en hey exis , a ec mainly he na u al a e o unemploymen and beyond his in luence, he ma ke ope a es la gely as a pe ec ly compe i i e sys em. Va ia ions in demand and supply o ac o s o p oduc s can change only ela i e p ices, so he sou ce o o e all p ice inc eases mus be exogenous inc eases in a ailable means o paymen . In la ion is caused by expansiona y demand policies when go e nmen s y o keep unemploymen a an a i icially low le el, bu 18 i is pe pe ua ed h ough expec a ions. In o he wo ds, in la ion pe sis s because agen s expec i o pe sis . Se e al ea u es in his expec a ion-adjus ed Phillips Cu e a e wo h no ing and dese e close examina ion. Fi s , F iedman (1977, p. 12) emphasizes ha ‘only su p ises ma e .’ I is he su p ise o unan icipa ed in la ion which con uses economic agen s and causes hem o misin e p e hei ela i e si ua ion. F iedman (1977, p. 13) explains ha bo h wo ke s and hei employe s ‘a e likely o adjus mo e slowly hei pe cep ion o p ices in gene al – because i is mo e cos ly o acqui e in o ma ion abou ha – han hei pe cep ion o he p ice o he pa icula good hey p oduce.’ Bu his appea s con a y o common expe ience. In p ac ice, o e all p ice indexes a e published mon hly and announced in he p in ed and elec onic media. The cos o inding ou wha is he o e all a e o in la ion is su ely edundan , especially conside ing he c ucial signi icance o his in o ma ion o he o ma ion o long- e m con ac s. Fu he mo e, in a pe ec ly compe i i e ma ke , selle s and buye s a e assumed o be ‘p ice ake s,’ so why should hey belie e ha hei own p ice ises as e han p ices o o he ma ke pa icipan s? Clea ly, such a collec i e e o canno s em om a pe ec ly compe i i e amewo k. Second, i is no clea why a ise (decline) in he a e o expansion o nominal agg ega e demand mus lead o an inc ease (dec ease) in he a e o in la ion. F iedman (1977, p. 13) asse s ha when agg ega e demand inc eases commodi y p ices ise (o a e expec ed o ise) and p oduce s aise hei wage o e s o wo ke s. This may be a likely ou come i we assume ha commodi y and labou ma ke s ope a e a ‘ ull’ capaci y and employmen o begin wi h, bu he e is e y li le eason o expec p ices and wages o ise when capaci y u iliza ion is ‘ e y’ low and unemploymen is ‘excessi ely’ high. O cou se, i he inc ease in demand g ow h is su icien ly la ge, bo lenecks may e en ually be eached and, as we app oach he ‘na u al’ a e o unemploymen , p ices and wages may s a o ise. In his ligh , he expec a ion-adjus ed Phillips Cu e in ol es a ci cula a gumen : an inc ease in he g ow h o nominal agg ega e demand canno cause a pe manen educ ion in unemploymen because unemploymen is al eady a i s pe manen ‘na u al’ a e! I he economy indeed ope a es as a pe ec ly compe i i e Wal asian sys em, hen excess nominal demand could lead only o ising p ices as he neoclassical dicho omy asse s. Bu unde such assump ions, he expec a ion-augmen ed Phillips Cu e canno be used o p o e ha he e is no long- un adeo since his was al eady assumed. F iedman’s expec a ion heo y hen me ely asse s how he ‘ eal’ economy supposedly shields i sel om he in luence o ‘mone a y’ o ces. Thi d, he asse ion by F iedman (1977, p. 12, emphases added) ha i ‘e e yone an icipa ed ha p ices would ise a , say 20 pe cen a yea , hen his an icipa ion would be embodied in u u e wage (and o he ) con ac s,’ is impossible o p o e. F iedman a gues ha ‘ eal wages would hen beha e p ecisely as hey would i e e yone an icipa ed no p ice ise, and he e would be no eason o he 20 pe cen a e o in la ion o be associa ed wi h a di e en le el o unemploymen han a ze o 19 a e.’ This could be a meaning ul asse ion o a hypo he ical economy whe e he eal wage is equal, by de ini ion, o bo h he ma ginal p oduc o labou and he ma ginal disu ili y o wo k. In such an economy he eal wage is clea ly independen o he o e all a e o in la ion, bu eali y is sligh ly mo e complica ed han his ic i ious wo ld. In p ac ice, he ma ginal alues o p oduc i i y and u ili y a e no obse able and he e is a con inuous dispu e be ween employe s and employees on he ‘app op ia e’ le el o eal ac o p ices. The de e minan s o eal wages a e qui e ‘a bi a y’ and may in ol e elemen s o ‘powe .’ The e is no basis o an a p io i assump ion ha ac o s such as ‘ba gaining s eng h’ a e independen o in la ion, e en when his in la ion is ully an icipa ed by all sides. Fu he mo e, e en i we igno e hese di icul ies, he s a emen by F iedman is s ill i e u able because in p ac ice we canno dis inguish be ween an icipa ed and unan icipa ed in la ion. To illus a e hese p edicamen s, conside he ollowing hypo he ical example. Suppose Gene al Mo o s and he Uni ed Au o Wo ke s’ union ag eed o a nominal wage inc ease o 25 pe cen o e he e m o he con ac and, suppose u he , ha he ac ual a e o in la ion o e ha pe iod was 20 pe cen . Could we es he p oposi ion ha his a e o in la ion was in ac ‘ ully embodied’ in he con ac ? To do ha we mus know whe he o no bo h sides had he same an icipa ion o in la ion, whe he o no hey expec ed his a e o be 20 pe cen , and whe he o no he nego ia ions p oceeded in ‘ eal e ms,’ independen ly o hese expec a ions. Unless we ha e all o his in o ma ion, he neu ali y p oposi ion canno be p o en. Fou h, he in oduc ion o addi ional non-obse able a iables u he diminishes he scien i ic cha ac e o he Phillips Cu e amewo k. F iedman (1968, p. 10) qui e openly admi s ha we canno know wha he na u al a e is. ‘Un o una ely,’ he w i es ‘we ha e as ye de ised no me hod o es ima e accu a ely and eadily he na u al a e o ei he in e es o unemploymen .’ A u he complica ion is in oduced when F iedman asse s ha he na u al a e o unemploymen is no ixed and ‘will i sel change om ime o ime.’ Unde hese assump ions, whe e he ‘na u al a e’ is an in isible mo ing a ge , he hypo hesis o a e ical long un Phillips Cu e canno be e u ed. Fo ins ance, suppose ha he go e nmen inc eases he pace o g ow h o nominal agg ega e demand and, some ime la e , unemploymen declines and in la ion ises. P oponen s o he na u al- a e hypo hesis can a gue ha he all in unemploymen was in ac a educ ion in he na u al a e i sel and, hence, go e nmen policy was me ely in la iona y, p ecisely as p edic ed by he heo y. This easoning aises one simple bu dis u bing ques ion: wha empi ical obse a ion will be inconsis en wi h he na u al- a e heo y? The a gumen is ‘ lawless’ simply because i canno be empi ically e u ed! So unless we can speci y he condi ions unde which his hypo hesis ails, he na u al- a e amewo k mus be iewed as a me e au ology. Some mac oeconomis s such as Go don (1985) and Fo in (1989), o ins ance, ha e a emp ed o es ima e he na u al a e o unemploymen om eg ession analyses based on he expec a ion-adjus ed Phillips Cu e. Such es ima es canno be used o es he 20 na u al- a e hypo hesis o in la ion because he la e was al eady assumed o be alid when he es ima es we e de i ed. Expec a ions c ea e ano he se ious p oblem o measu emen because, like demand and supply, hey also canno be obse ed di ec ly. I expec a ions and, hence, changes in expec a ions canno be obse ed, how could we es he hypo hesis ha such adjus men s cause he Phillips Cu e o shi ? Many economis s a emp ed o ackle he p oblem by simply subs i u ing speci ica ions o obse a ions bu , un o una ely, hey only eplaced one p oblem wi h ano he . Fo example, suppose we impose an adap i e expec a ion mechanism on ma ke p ices and disco e i has a subs an ial explana o y powe . Can we conclude on he basis o such e idence ha p ices a e de e mined by adap i e expec a ions o ma ke pa icipan s? The answe o his ques ion is nega i e because he s a is ical amewo k con ains obse a ions on p ices bu no on expec a ions. In ac , we ne e demons a ed ha economic agen s o m adap i e expec a ions (o any o he expec a ions), o ha hey ac on he basis o such expec a ions. Fo ha ma e , cu en p ices a e ‘de e mined’ by pas p ices and ma ke pa icipan s play no explici ole in he model! The i h and inal issue conce ns he ‘neu ali y’ p oposi ion associa ed wi h he expec a ion-augmen ed Phillips Cu e. The s a emen by F iedman ha demand policy canno ha e a pe manen ‘ eal’ e ec on he economy has been challenged by se e al mac oeconomis s,10 bu hei c i icism e e s mainly o he inal impac and igno es he ini ial na u e o he policy i sel . Conside wha happens when he go e nmen inc eases i s demand o goods and se ices by aising mili a y spending, o ins ance. Mos o he new o de s will ypically go o a g oup o 50 o 100 co po a ions which, in u n, will subcon ac some o he wo k o a ew hund ed addi ional i ms. The emaining companies in he economy will be excluded om his ini ial injec ion o spending. O conside he di ec e ec o open ma ke ope a ions by he cen al bank. In he Uni ed S a es, go e nmen bonds a e no e enly dis ibu ed be ween households in he economy bu a he a e concen a ed mainly in he hands o la ge ins i u ional in es o s. An a emp o he cen al bank o inc ease he money supply by buying bonds equi es ha he bank bid up hei p ices. So he immedia e bene icia ies o his mone a y expansion a e he la ge ins i u ional in es o s while o he economic agen s emain una ec ed. Clea ly, he di ec e ec o such mac oeconomic policies is o al e s he exis ing dis ibu ion o income, asse s, p oduc ion and ela i e p ices be ween ma ke pa icipan s. In ac , i is ha d o hink o a single mac oeconomic policy which does no ha e such ini ial ‘ eal’ e ec s on he economy. To summa ize, he expec a ion-adjus ed Phillips Cu e and he ela ed na u al- a e hypo hesis a e based on some igid explici o implici assump ions ega ding economic s uc u e and scien i ic me hodology. The economy is assumed o ope a e ‘as i ’ i was a Wal asian compe i i e sys em whe e agen s espond o 10 See Bui e (1980, pp. 39-40) o a summa y s a emen on such c i icism. 21 ‘ eal’ s imulus and a e impa ial o ‘nominal’ ones. When mac oeconomic demand policies a e execu ed, hei ini ial e ec is assumed o be e enly dis ibu ed among all economic agen s, so as no o upse he o iginal ‘ eal’ s uc u e o he ma ke . An inc ease in he pace o agg ega e demand g ow h causes in la ion o accele a e because ma ke s al eady ope a e a ull capaci y and employmen . In o ma ion abou agg ega e p ice and in la ion indices is a ailable a no cos , bu p ice- aking selle s and buye s a e ne e heless con used by his ini ial u n o e en s and ail o ealize ha p ices a ound hem ise as as as he p ice o hei own commodi y. As a esul , hey inc ease hei supply and demand o p oduc s and ac o s and cause he o e all le el o unemploymen o all below i s na u al a e. Ul ima ely, agen s disco e hei collec i e e o and seek o educe demand and supply as soon as hei long e m-con ac s expi e. This causes a g adual upwa d shi in he Phillips Cu e e lec ing he adjus men o expec a ions and con ac s o he new le el o in la ion. When he adjus men is comple e, he economy e u ns o i s o iginal, ‘ eal’ Wal asian equilib ium bu wi h a highe a e o in la ion. Un o una ely, his p ocess o adjus men canno be acked down because bo h expec a ions and he na u al a e o unemploymen a e no obse able. 6. In Ques o In o ma ion: The Unemployed as an In es o Al hough he ‘na u al’ a e o unemploymen could no be obse ed empi ically, many economis s s ill el i was a c ucial concep which dese ed igo ous heo e ical elabo a ion. The i s sys ema ic discussion on he opic appea ed in an impo an collec ion o a icles edi ed by Phelps in 1970 and i led Mic oeconomic Founda ions o Employmen and In la ion Theo y. In he in oduc ion, Phelps (1970, pp. 4-5) alked abou a majo heo e ical b eak h ough in he making: The heo e ical depa u e ha is common o hese o he wise neoclassical pape s is hei emo al o he Wal asian pos ula e o comple e in o ma ion, . . . [and] . . . Wi h he pos ula e o pe ec in o ma ion emo ed, he way is a las open o o mal s udy o gene al disequilib ium. Wi h his mino ‘in o ma ional’ amendmen , he pe sis ence o unemploymen was no longe o be pe cei ed as a condemna ion o capi alism and an emba assmen o neoclassical heo y. Ins ead, unemploymen became a desi able aspec o economic ac i i y and an in eg al pa o con en ional heo y. To se a amewo k o he new mic o ounda ions, Phelps (1970, p. 6) desc ibes ou economy as a collec ion o islands. Compe i ion on each indi idual island is impeccable: [L]abo is echnically homogeneous in p oduc ion unc ions and indi e en among he many he e ogeneous jobs o p oducing a a ie y o p oduc s. 22 P oduce s on each island a e in pu e compe i ion in he labo and p oduc ma ke s. Each mo ning, on each island, wo ke s ‘shape up’ o an auc ion ha de e mines he ma ke -clea ing money wage and employmen le el. Un o una ely, he i ues o such sys em a e dis o ed because he low o in o ma ion be ween di e en islands in he a chipelago is no ee. Acco ding o Alchian (1970, p. 29), in o ma ion is a commodi y like any o he and, as such, i is subjec o s anda d economic laws o p oduc ion and cos : Dissemina ion and acquisi ion (i.e., he p oduc ion) o in o ma ion con o ms o he o dina y laws o cos s o p oduc ion: as e dissemina ion, o acquisi ion cos s mo e . . . [and] . . . Like any o he p oduc ion ac i i y, specializa ion in in o ma ion is e icien . Ga he ing and dissemina ion in o ma ion abou goods o abou onesel is in some ci cums ances mo e e icien ly done while he good o pe son is no employed, and hus able o specialize (i.e., while specializing) in he p oduc ion o in o ma ion. Phelps’ a chipelago economy p esen s no excep ion o hese pos ula ed ules. P esumably he e a e no mode n means o communica ion (such as elephone, newspape s o elex) be ween he islands and, hence, wo ke s who wan o know mo e abou job o e s mus ‘specialize’ in ga he ing his in o ma ion by owing om island o island: To lea n he wage paid on an adjacen island, he wo ke mus spend he day a elling o ha island o sample i s wage ins ead o spending he day a wo k. (Phelps, 1970, p. 6) In his con ex , unemployed wo ke s owing be ween he islands a e no seeking ‘jobs’ bu ‘job in o ma ion.’ Acco ding o Alchian (1970, p. 30): Jobs a e always easily a ailable. Timely in o ma ion abou he pay, wo king condi ions, and li e expec ancy o all a ailable jobs is no cheap. In a sense, his kind o unemploymen is sel -employmen in in o ma ion collec ion. Since jobs a e always a ailable, wo ke a e unde no p essu e o accep any pa icula o e . Ins ead, he choice o employmen is based on a ca e ul op imiza ion s a egy. Like any o he in es o , a ypical wo ke in he a chipelago ies o maximize he p esen alue o his in es men , namely, o his labou powe . Unde ce ain ci cums ances, his wo ke may ind i highly ad an ageous o wi hd aw he se ices o his commodi y. Such unemploymen hen cons i u es a o m o in es men ac i i y. In his axonomy o di e en ypes o unemploymen , Phelps (1972, p. 3) candidly sugges s o ca ego ize he mo i es o unemploymen ‘much as 23 and hey ‘was ed’ non-p ice in o ma ion ha could ha e been used o imp o e hei p ice o ecas . The c i iques poin ed ou ha i indi iduals we e indeed a ional decision make s, hey should also o mula e ‘ a ional expec a ions.’ The a ional-expec a ion hypo hesis was i s sugges ed by Mu h (1961, p. 316) who a gued ha ‘expec a ions, since hey a e in o med p edic ions o u u e e en s, a e essen ially he same as he p edic ions o he ele an economic heo y.’ In his opinion, his mean ha expec a ions o i ms (o , mo e gene ally, he subjec i e p obabili y dis ibu ion o ou comes) end o be dis ibu ed, o he same in o ma ion se , abou he p edic ion o he heo y (o he ‘objec i e’ p obabili y dis ibu ions o ou comes). Mu h’s emphasis on a ional expec a ions was igno ed by mac oeconomis s o o e a decade un il i was picked up by Lucas, Sa gen and o he s in he ea ly 1970s. Sa gen (1973, p. 431) o ins ance, asse ed ha expec a ions o in la ion a e assumed o be endogenous o he sys em in a e y pa icula way: hey a e assumed o be ‘ a ional’ in Mu h’s sense – which is o say ha he public’s expec a ions a e no sys ema ically wo se ha he p edic ions o economic models. This amoun s o supposing ha he public expec a ions depend, in he p ope way, on he hings ha economic heo y says hey ough o. Since hese ea ly o mula ions, he idea o a ional expec a ion depended c i ically on wo key concep s: he ‘objec i e dis ibu ion o ou comes’ and he ‘ ele an economic heo y’ associa ed wi h i . Despi e close o wo decades o heo izing, hese concep s emained su p isingly enigma ic. The language used in he a ional-expec a ion li e a u e is o en c yp ic and he emphasis on ma hema ical symbolism helps o u he cloak subs an i e issues. In ou examina ion we p esen some o he basic claims ad anced in he a ional-expec a ions li e a u e and assess hei me i s. A simple desc ip ion o he a ional-expec a ion amewo k could un as ollows: The economy is a closed sys em wi h i s own ‘laws o mo ion.’ These laws o mo ion de e mine how he endogenous a iables o he sys em in e ac wi h he exogenous and p ede e mined ones (in o he wo ds, hese laws de e mine he educed o m o he simul aneous equa ion sys em).13 The economic sys em in e ac s wi h o he sys ems like ‘na u e’ and ‘poli ics.’ These sys ems de e mine alues o he exogenous a iables. Some o hese exogenous a iables ollow sys ema ic pa e ns while o he s a e andom a iables wi h gi en dis ibu ions. I his o y could ha e been 13 The e m ‘laws o mo ion’ (as used by Sa gen , 1986, p. 3, o ins ance) e e s o a desc ip ion o a s a iona y p ocess and has no hing o do wi h Ma x’s o iginal e e ence o p inciples go e ning he dynamic ans o ma ion o socie y. 30 ‘ e- un’ wi h gi en laws o mo ion, gi en alues o he sys ema ic exogenous a iables and gi en alues o he p ede e mined a iables, i would ha e gene a ed an ‘objec i e dis ibu ion o ou comes.’ The mean o his dis ibu ion would e lec he impac o p ede e mined and sys ema ic exogenous a iables, and he dispe sion would be a ec ed by he dis ibu ion o exogenous dis u bances.14 In accep ing such a se up, he a ional-expec a ions heo is s me ely ollow he s anda d app oach owa d mac oeconomic modelling. The di e ence be ween s anda d mac oeconomic models and ones based on a ional expec a ions s ems om assump ions ega ding wha people know abou he economic sys em. In a wo ld o a ional expec a ions, people possess conside able knowledge abou he sys em. They unde s and he sys em’s laws o mo ion (in o he wo ds, hey know he ‘ ele an heo y’ and he alues o i s pa ame e s). They also know all abou he pas his o y o he sys em ( hey know he alues o he p ede e mined a iables). They u he know he alues o hose exogenous a iables which ollow a sys ema ic pa e n. They do no know he alues o he andom exogenous a iables bu hey know he dis ibu ion om which hese a iables a e d awn. Unde hese condi ions, a simple a ional expec a ions hypo hesis o in la ion can be summa ized by he ollowing equa ions: (1) e = E(p | I -1) , (2) p = E(p | I -1) + u . In hese equa ions he expec ed a e o in la ion is deno ed by e and he ac ual a e by p; E is he condi ional expec a ion ope a o and I is he ‘se o ele an in o ma ion a ailable’ (a ca ch-all e m o he ‘ ele an heo y’ ega ding he ‘laws o mo ion,’ he pa ame e s o ha heo y and he alues o p ede e mined and sys ema ic exogenous a iables); inally, u is he p ice e ec on in la ion o andom exogenous shocks. Unde hese condi ions, expec a ional e o s s em only om hese unp edic able shocks and ha e no sys ema ic componen . Wha happens in a Wal asian, ic ionless communi y o a ional economic agen s when, s a ing om equilib ium, he go e nmen a emp s o inc ease he pace o g ow h o agg ega e demand? I hese in en ions become known be o e hey a e execu ed ( o example when he go e nmen ollows a ‘policy ule’), he e ec s o he policy a e immedia ely neu alized by he coun e ac ions o p i a e economic agen s. The eason o his ‘policy-ine ec i eness’ is s aigh o wa d. Acco ding o he neoclassical dicho omy be ween he ‘ eal’ and ‘nominal’ domains, he ul ima e e ec o demand policy is on he p ice le el. Since his demand policy is pa o I when expec a ions a e o med, he impac o such policy on nex pe iod’s p ices can be 14 The se up o his dis ibu ion is based on he con en ional assump ion ha he mean impac o andom shocks on endogenous a iables is ze o. 31 accu a ely p edic ed by Equa ion (1). When agen s adjus hei ‘ eal’ supply and demand schedules in an icipa ion o he new policy measu es, hey a he same ime make hese measu es ine ec i e (in o he wo ds, by al e ing hei ‘decision ules’ hey also al e he sys em’s ‘laws o mo ion’). This ins an aneous adjus men means ha policy does no in lic e en a sho - un disequilib ium and he economy shi s smoo hly om one long- e m equilib ium in o he nex . No e ha ‘su p ise’ policy can a ec he eal economy. In he absence o a ‘policy ule’ o ins ance, policy changes cons i u e a andom shock o he sys em and a ec p ices h ough u. Since a ional expec a ions do no accoun o such unp edic able jol s, he eal economy is dis o ed by he nominal impe us. Fo una ely, his e ec is e y sho -li ed because he execu ion o he policy makes i pa o he sys em’s laws o mo ion and, hence, an ing edien o he ‘ ele an heo y.’ The conclusion o his new-classical scheme esembles he amous Ca ch-22: in o de o s abilize he economy, policy mus be ela ed o e en s in some sys ema ic way. Bu a sys ema ic policy is p edic able and a p edic able policy is neu al. To pu i somewha di e en ly, in o de o s abilize he economy he go e nmen mus be able o a ec i , bu his calls o an e a ic, unp edic able policy which can only des abilize he economy! The ci cle is closed and he case agains demand managemen is comple e. The a ional expec a ion hypo hesis has been o en hailed as a ‘ e olu ion’ in mac oeconomic hinking. Many o i s leading ligh s downplay hei con ibu ion, howe e , s a ing i is me ely a na u al e olu ion owa d a g ea e consis ency o mac oeconomic models wi h basic mic oeconomic ene s. Taylo (1985, p. 393) asse s ha mac o-models wi h a ional expec a ions a e now he ‘ ule a he han he excep ion,’ ye se e al key ea u es sugges ha emb acing he new classical amewo k may in ac hinde a he han enhance ou unde s anding o how a mode n economy wo ks. These aspec s dese e some close examina ion and we conside hem now. The i s ques ion conce ns he ‘ ele an heo y.’ In a Lucas- ype abs ac economy, he p oblem does no e en a ise simply because he economy is de ined by he heo y, bu in a complex, mode n economy like ha o he Uni ed S a es, he ques ion can no longe be igno ed. Reali y has no enclosed se o bluep in s and, indeed, economis s a ely ag ee abou i . The e exis s a ich menu o di e en heo ies and i is no clea which heo y (i any) p o ides an accu a e desc ip ion o he economy’s alleged ‘laws o mo ion.’ A second ques ion ega ds he assimila ion o a ‘ ele an heo y.’ E en i a ‘co ec ’ heo y does exis , why should i become common knowledge? Again, in a Lucas- ype economy, agen s a e simply assumed o possess all he necessa y in o ma ion abou he economy’s bluep in s and i s his o ical e olu ion, bu wha occu s in a eal economy? Mu h (1961, p. 330) s a ed ha expec a ions mus be a leas ‘mode a ely a ional’ o o he wise ‘ he e would be oppo uni ies o economis s o make p o i s in commodi y specula ion, unning a i m, o selling he in o ma ion o p esen owne s’[?] In o he wo ds, by aking ad an age o hei supe io 32 unde s anding, economis s u n hei p i a e co ec heo y in o common knowledge and he ‘ ele an ’ heo y is assimila ed.15 The e a e wo di icul ies wi h his logic. One, when he economy is changing, he ele an heo y o oday need no be he ele an one o omo ow and, hence, his p ocess may mislead economic agen s o adop ou da ed iews.16 Two, he assimila ion o heo ies has a e y ‘ eal’ e ec on he economy because i p esumably edis ibu es income (pa icula ly p o i ) om hose who canno ead he ma ke o hose who can. Mu h (p. 316, emphasis added) a gues ha a ‘public p edic ion’ has no subs an ial e ec on he ope a ion o he economic sys em ‘unless i is based on inside in o ma ion’ bu his own iew on he assimila ion o ma ke knowledge sugges s ha e e y ele an heo y g ows om ‘inside in o ma ion.’ Hence, whe he assimila ed o no , ‘ ele an ’ heo ies mus ha e a subs an ial impac on he economy.17 Thi d, he a ional-expec a ions hypo hesis asse s ha people’s expec a ions cons i u e pa o he sys em’s laws o mo ion. This implies ha in e dependency be ween he ‘objec i e dis ibu ion o ou comes’ and he ‘ ele an heo y’ is po en ially des abilizing. F ydman and Phelps (1983) a gued ha he ‘a e age opinion’ o economic agen s is one o he exogenous a iable in he economic sys em, so when agen s a emp o de e mine his ‘a e age opinion,’ hey ge en angled in an in ini e- eg ess p oblem and may d i e he sys em owa d a pe manen s a e o disequilib ium. Cagan (1983, p. 45), commen ing on he same poin , w o e ha Maximizing beha io equi es ha economic agen s can in ac ind he maximum posi ion on hei own. I ha posi ion is a ec ed by he 15 In a guing ha i a ional expec a ions a e necessa ily sho -li ed, Maddock and Ca e (1982, p. 45) in oke he au ho i y o Keynes (1930, p. 160) who w o e ha ‘ac ions based on inaccu a e an icipa ions will no long su i e expe iences o a con a y cha ac e , so ha ac will soon o e ide an icipa ion excep whey hey ag ee.’ Howe e , his me ely sugges s ha people may ealize hey we e w ong, no ha hey will necessa ily lea n om hei mis akes. As a gued bellow, he con e gence o expec a ions owa d a ional expec a ion is no ine i able. 16 Economis s ha e been con inuously al e ing hei models ye hei p edic ions published in he popula and scien i ic media do no seem o con e ge owa d any single, ‘co ec ’ ec o . Fo ins ance, a ecen su ey o ‘Wha Economis s a e P edic ing o 1990’ published in Business Week o Decembe , 25, 1989, epo s 25 p edic ions o eal-GNP g ow h anging be ween a high o 5.1 pe cen and a low o –3.2 pe cen . P edic ions o in la ion ange be ween 2.5 o 6.3 pe cen , p edic ions o he in e es a e a y be ween 6.5 and 12.5 pe cen and p edic ion o he a e o unemploymen un be ween 4.4 and 8.8 pe cen . No e ha hese p edic ions we e no made by ‘i o y owe ’ economis s bu by business economis s wo king o la ge companies who s ood o lose om e oneous o ecas s. 17 ‘Inside in o ma ion’ on he s ock ma ke gene a ed and con inues o gene a e subs an ial p o i s bu a e such in o ma ion is used, i becomes useless a he han ele an public knowledge. 33 expec a ions o o he s, I do no see ha maximizing beha io unde such ci cums ances, e en wi h Bayesian lea ning, is any longe well de ined. E en Taylo , an o hodox adhe en o a ional expec a ions, admi ed ha ‘[b]ecause o he sel - ul illing ea u e o a ional expec a ions, he e is gene ally a con inuum o solu ions o a ional expec a ions models’ (1985, p. 419). A ou h p oblem a ises when we examine how he p i a e sec o esponds o public-sec o ini ia i es in a ‘game heo e ic’ s uc u e. Fo ins ance, i he go e nmen can e oke i s policy commi men s (when i ollows an uncons ained a he han cons ained ‘policy ule’), he neu ali y p oposi ion ails. Kydland and P esco (1977) a gue ha in a dynamic game be ween wo agen s ( he p i a e sec o agains he go e nmen a he ha agains ‘na u e’), a ional expec a ions may lead o ‘inconsis ency o op imal plans.’ Bui e (1980, p. 36) concludes ha adi ional op imal con ol echniques ‘ ail o ake accoun o he impac o u u e policy measu es on cu en e en s h ough he changes in cu en beha iou induced by an icipa ion o hese u u e policy measu es.’ This cul i a ed language con eys a simple message: when human beings a e allowed disc e ion and he e is some in e dependency be ween hei economic decisions, he e may be no ‘objec i e dis ibu ion o economic ou comes.’ Fi h, he a ional-expec a ions amewo k ocuses on how p i a e-sec o agen s espond o public-sec o ini ia i es, while li le o no a en ion is paid o dynamic ini ia i es in he p i a e sec o i sel . This choice o emphasis is common in much o he mac oeconomic li e a u e on expec a ions bu i is s iking in he new classical w i ings. In i s c ude o mula ion, he a ional-expec a ions hypo hesis examines only one ype o ini ia i e: go e nmen a emp s o change agg ega e demand. E e y o he economic ac ion is ‘au oma ic.’ P i a e agen s wi h a ixed se o p e e ences a e locked in hei uncomp omising d i e o maximize u ili y. To achie e his goal unde p e ec compe i ion hey mus ollow one p e-de e mined cou se o op imal ac ion. Thei s is a ‘game o man agains na u e’ whe e na u e changes ‘ echnology’ and man esponds ollowing ixed, known ules o conduc . I we disca d his pe e ed animism and ecognize ha ini ia i e, disc e ion and in e dependency exis in he ela ion be ween agen s such as i ms, consume s, wo ke s and in es o s, we open a Pando a box o dis u bing ques ions. Fo example, wha a ional expec a ions can agen s o mula e on a wo ld domina ed by oligopolies wi h complex business ies? Wha p ices should we expec o see when manage s ell us hey ollow a ule-o - humb in se ing p o i ma kups? Wha a e he expec ed ‘objec i e ou comes’ om a emp s by p i a e agen s o o m coali ions o o in luence he go e nmen owa d a edis ibu ion o income? To ou knowledge he e is no de ini ion o he ‘objec i e dis ibu ion o ou comes’ in he a ional-expec a ions li e a u e. The idea seems o imply ha he expe ience o ou economy in any ‘sample pe iod’ is gene a ed by some speci ied ‘laws o mo ion,’ and ha his ac ual ‘his o y’ is me ely one obse a ion d awn om 34 a in ini e sample o po en ial ou comes, wi h a s able mean and a gi en dispe sion.18 This amewo k becomes meaningless when we iew he economic p ocess as a quali a i e ans o ma ion o e olu ion a he han a ‘d aw’ om a s a iona y p ocess. When he e is human ini ia i e, his o ical change has ew i any ‘de e minis ic’ componen s and e en a ional agen s canno ‘jump o e Rhodes’ o disco e he u u e. ‘Abou hese ma e s,’ a gued Keynes (1937, p. 185) ‘ he e is no scien i ic basis o which o o m any capable p obabili y wha soe e . We simply do no know.’19 These c i icisms should no be in e p e ed as sugges ions owa ds imp o emen s o he a ional-expec a ions amewo k. We belie e ha new classical economics is ba en and misleading, and ha heo e ical ‘imp o emen s’ o his app oach a e simply u he s eps in he w ong di ec ion. The dange o accep ing he legi imacy o such ‘imp o emen s’ is illus a ed by ecen a emp s o inco po a e seemingly ‘ ealis ic’ ea u es in o a a ional-expec a ions heo y. The p ominence o new classical ideas also b ough hem unde he magni ying glass o mac oeconomis s. Schola s like Tobin (1980), Bui e (1980), F ydman (1981) and Go don (1981) a gued ha he policy-ine ec i eness conclusion depended no only on he assump ion o a ional expec a ions bu also on he exis ence o a Wal asian, ma ke -clea ing sys em o p ices. When a sys em wi h sluggish wage o p ice adjus men was subs i u ed o he Wal asian cons uc , he sho - un Phillips Cu e eappea ed e en unde a ional expec a ions ( ecall ha ins i u ional igidi ies we e one o he elemen s in F iedman’s model). Fische (1977), o ins ance, in oduced mul ipe iod con ac s in he labou ma ke and concluded ha he au ho i ies could a ec eal a iable p o ided he policy du a ion was sho e ha he leng h o con ac s. Phelps and Taylo (1977) eached a simila conclusion when hey examined he consequences o p ices and wages being se one pe iod in ad ance. In hese models u u e p ices a e se o clea he ma ke on he basis o cu en in o ma ion bu when new in o ma ion abou policy a i es, p ices a e oo ‘s icky’ o adjus immedia ely and he policy becomes e ec i e. Talyo (1979) in oduced o e lapping, o s agge ed wage con ac s in o he a ional-expec a ions amewo k and concluded ha policy can be e ec i e e en i i s announced lead- ime is longe han he du a ion o he longes con ac (Taylo , 1985, p. 414). 18 No e ha hese p esump ions unde lie he no ion o ‘ unc ional ela ionships’ in he social sciences and a e common in con en ional econome ic app oaches o es ima ion, es ing and p edic ions o mac oeconomic models. The signi icance o he a ional-expec a ions amewo k is in making hese p esump ions explici . 19 Acco ding o Geo gescu-Roegen (1979, p. 322), he mos no able ea u e o he economic p ocess is he con inuous eme gence o no el y, o quali a i e change. Un o una ely, he a gues, ‘no analy ical model can deal wi h he eme gence o no el y, o e e y hing ha can be de i ed om such a model can only conce n quan i a i e a ia ions . . . no ing can be de i ed om an analy ical model ha is no logically con ained in i s axioma ic basis.’ Con a y o he new-classical eupho ia, Geo gescu-Roegen concludes ha ‘we canno possibly ha e a bi d’s eye iew o he u u e e olu ion o mankind’ (p. 325). 35 Acco ding o Taylo (1985, pp. 411), he algeb a o hese models e ains he long- un neu ali y o policy bu allows he same policy o be e ec i e in he sho un. Hence, such models can be iewed as a emp s o esol e wha Go don (1981, p. 509) labelled he ‘pe sis ence dilemma’ o he a ional-expec a ions hypo hesis. The acknowledgmen o con ac s and p ice s ickiness may appea o econcile he a ional-expec a ions hypo hesis wi h pe sis en de ia ions o ac ual unemploymen om i s end. Un o una ely, his au a o ealism is a a he decep i e deco a ion o a ba en axioma ic model ha has e y li le o do wi h dynamics o complex ma ke s uc u es. Taylo mus be awa e ha eal-li e con ac s ha e nume ous ins i u ional and dynamic aspec s which canno allow s able ARMA ep esen a ions. Ye , inding such ime-in a ian ep esen a ions a e c ucial o his model so eal con ac s mus gi e way o axioma ic ones, whe e all dange ous ac uali y has been con enien ly emo ed. The model appa en ly d esses in ‘ ealism’ while, in ac , i is shallow. 8. ‘Ins i u ional Ins abili y’ and S ag la ion The his o y o he Phillips Cu e could be desc ibed as an ongoing duel be ween eali y and heo y, in which he cunning o his o y has p o en o be no ma ch o he ingenui y o mac oeconomis s. When, du ing he la e 1960s and ea ly 1970s, in la ion accele a ed wi h no appa en decline in unemploymen , mac oeconomis s esponded by modi ying he downwa d-sloping Phillips Cu e in o a e ical one. To do so, hey in oduced expec a ions and he ‘na u al a e’ axiom in o he amewo k. Subsequen ly, when his o y s aged a combina ion o ising in la ion and ising unemploymen , mac oeconomis s esponded by ying o bend he Phillips Cu e in o an upwa d-sloping posi ion using concep s such as ‘ins i u ional ins abili y’ and ‘exogenous shocks.’ We conside hese la e modi ica ions in his and he ollowing sec ion. In his 1977 Nobel lec u e, F iedman asse ed ha he e ical cu e could su i e he new eali y o s ag la ion wi h only a ‘modes elabo a ion o he na u al- a e hypo hesis.’ The elemen missing om his own o iginal o mula ion was he equi emen o he a e o in la ion i sel be s able. When he same a e o in la ion p e ails o ‘many decades,’ w o e F iedman (p. 24), we could expec ha p ices be ully an icipa ed and ully adjus ed. These condi ions o a e ical Phillips Cu e a e likely o be me in wha F iedman calls he ‘long-long un’ bu he in e im phase o ansi ion owa d in la ion s abili y may in ol e some unpleasan complica ions. The inc ease in he a e o in la ion du ing he pos -wa pe iod in Eu ope and he Uni ed S a es also b ough wi h i inc eased luc ua ions in ha a e. F iedman specula es ha his inc ease in in la ion ins abili y led o ising ins i u ional ins abili y, whe eby he op imum leng h o unindexed commi men s was sho ened, he e iciency o he p ice sys em in coo dina ing economic ac i i y was educed, public policies became inc easingly con used, and he ex en o go e nmen in e en ion in ee ma ke s was g ea ly inc eased. F iedman a gues ha such 36 de elopmen s had ad e se consequences o economic e iciency, bu he admi s ha hey do no eally explain he appa en d i o unemploymen .20 In o he wo ds, accep ing he p oposi ion ha he Phillips Cu e is e ical in he long-long un does no help us esol e he puzzle o con empo a y s ag la ion. In ou opinion, he weakness o F iedman’s analysis s ems no om his ailu e o u he amend he Phillips Cu e amewo k bu a he om his e y a emp o do so. F iedman (1977, pp. 7-8) implies ha his heo e ical manoeu es a e cons uc i e s eps in scien i ic p og ess bu i seems ha , ins ead o di ec ing us owa ds be e unde s anding o s ag la ion, his ‘pa ching-up’ leads us in o a heo e ical acuum. Each successi e in e p e a ion o he Phillips Cu e u ns he exis ing cons uc in o a ‘special case’ o a ‘mo e gene al’ amewo k. The ‘sho un’ ha ex ended om he la e nine een h cen u y and un il he middle o his cen u y became a special case o a ‘long un’ ha emb aced us be ween he la e 1960s and ea ly 1970s, bu e en his ‘long un’ was me ely a subse o a ‘long-long un’ phase which we en e ed in he mid-1970s. The i s shi was c ea ed when economis s disco e ed ha in o ma ion was ‘impe ec .’ The second ans o ma ion was ins i u ed when economis s ealized ha ins i u ions we e slow o ‘adjus .’ This leads us o pose one simple ques ion: i economic li e amoun s o a con inuous and p og essi e ‘depa u e’ om some enigma ic equilib ium ela ionships and i hese ela ionships will be alid only in some imagina y u u e when s a iona i y eplaces his o y, why should such equilib ium ela ionships be use ul in explaining eal phenomena? The p edicamen is well illus a ed in F iedman’s own w i ings. On he one hand he pain ully acknowledges he ‘ eal’ consequences o a high, a iable in la ion: . . . some g oups gain . . . o he lose. . . . The socie y is pola ized; one g oup is se agains ano he . Poli ical un es inc eases. The capaci y o any go e nmen o go e n is educed a he same ime ha he p essu e o s ong ac ion g ows. On he o he hand, he has e y li le o say on hese issues, since con lic o in e es s and con inuous edis ibu ion canno be in eg a ed in o a amewo k which pa ien ly looks o wa d owa d some long-long un s a e o bliss, when ull ‘adjus men s’ and es o ed social ha mony eins a e he neoclassical dicho omy be ween in la ion and unemploymen . 9. The S ag la iona y Menace o ‘Exogenous Fo ces’ Al hough mac oeconomics was c i icized du ing he 1970s o i s ailu e o e ec i ely deal wi h s ag la ion, mac oeconomis s we e no eady o ake he blame. 20 Some au ho s (like Fische , 1981) es ed and ejec ed he p esumed link be ween in la ion, in la ion ins abili y and unemploymen . 37 Blinde (1979, pp. 3 and 5-6) o example, insis s ha he e is no hing w ong wi h mac oeconomics o , by using he e y udimen a y agg ega e demand and supply cu es, one can p o ide a ‘ ai ly simple and gene al heo y o s ag la ion’ ha ‘can indeed explain wha has happened.’ In his opinion, c i iques ha e o en e ed by con using he p oblem wi h i s solu ion: s ag la ion could be easily explained bu i could no be easily cu ed. Mo eo e , poli icians gene ally ailed o unde s and his and made a di icul si ua ion e en wo se. Wi hin wha has now become he s anda d mac oeconomic model, s ag la ion a ises ei he as an adjus men p ocess ollowing an ea lie expansion o agg ega e demand, o as a esul o ad e se con ac ion in agg ega e supply. Bo h cases begin and end in a long- un mac oeconomic equilib ium, bu hey di e in he sou ce o dis up ion and in he way he economy esponds o i . Conside he i s case, whe e he ini ial equilib ium is upse by an ‘au onomous’ expansion o agg ega e demand. In he sho - un, he expansiona y demand ‘shock’ causes ou pu o ise beyond i s ‘na u al’ a e wi h no pa allel inc ease in p ices. This is a alse anquilli y, howe e . As ime passes, he economy mo es in o an ‘in e media e un’ and inpu p ices begin o ise, pulling ou pu p ices wi h hem. Un o una ely, his is no he end o he s o y. E en ually, ou pu s a s o all because, by de ini ion, he economy mus con e ge o i s ‘po en ial,’ o ‘na u al’ a e o ou pu . Hence, we mo e h ough h ee phases in he ollowing o de : g ow h wi hou in la ion, g ow h wi h in la ion and, inally, s agna ion wi h in la ion, o s ag la ion. ‘Because wages and p ices mo e sluggishly,’ asse s Blinde (1979, p. 14, emphasis added), ‘ eal ou pu mus o e shoo i s e en ual posi ion’ and s ag la ion is me ely he ine i able p ocess by which his ‘e en ual’ posi ion is eached. The explana ion also sugges s ha s ag la ion is in ac implici in he augmen ed Phillips-Cu e amewo k: an expansiona y demand-policy causes he economy o climb up he sho - un Phillips Cu e, bu he subsequen shi o he cu e causes unemploymen o inc ease back o i s ‘na u al’ a e in he mids o ising in la ion. While mos mac oeconomis s accep ed he heo e ical alidi y o demand-induced s ag la ion, i was he ‘supply-shock’ a ionale ha cap u ed hei imagina ion. Why complica e he analysis, many asked, when he ‘laws o supply and demand’ o e ed he mos simple solu ion o he s ag la ion iddle? I p ices and ou pu mo e in opposi e di ec ions, i was only na u al o associa e his ou come wi h changes in supply, no in demand. A ‘supply shock’ which shi ed he in e media e and long- un agg ega e supply cu es o he le , would cause s ag la ion wi h ising p ices and alling ou pu . Mo eo e , since we assume ha he na u al a e o ou pu i sel is educed, he si ua ion is o en i e e sible and he ad e se e ec s o he o iginal shock may be wi h us ‘ o e e ’ (Blinde , 1979, p. 16). Finally, a supply shock may c ea e a leng hy wage-p ice spi al ha will u he agg a a e he ini ial e ec s o ha shock. All o his means ha when supply-shocks hi he economy, poli icians a e co ne ed in o a policy nigh ma e: 38 The limi ed capabili y o policy o in luence supply poses a pa icula ly exing p oblem in a s ag la iona y wo ld since any s abiliza ion policy adop ed in esponse o s ag la ion is bound o agg a a e one o he p oblems [in la ion o unemploymen ] e en as i helps cu e he o he . Such is he policy dilemma o s ag la ion. (Blinde , 1979, pp. 20-1) Many mac oeconomis s we e exci ed by his al e na i e heo e ical a enue hough only ew we e ully awa e o i s wide me hodological implica ions. Bo h agg ega e demand and agg ega e supply a e ‘con enien ’ ools o analyzing he neoclassical syn hesis. In his amewo k we can always a gue ha agg ega e demand inc eased o ha agg ega e supply dec eased, and i is p ac ically impossible o e u e such asse ions since ‘desi ed’ magni udes o spending o p oduc ion a e no obse able. Ye beyond his con enience, he e lies a dis u bing asymme y be ween he wo concep s o supply and demand. I seems ha agg ega e demand can shi o a hos o ‘subjec i e’ easons; o ins ance when consume s change hei ‘p e e ences’ o ‘p opensi ies,’ when in es o s expe ience a bu s o ‘animal spi i s,’ o when poli icians make an ‘au onomous’ policy mo e. Thus, since he wo ld o demand is o en a he me cy o human impulse, i can be easily blamed o much o ou ins abili y. A simila hypo hesis o supply is no e y con incing, howe e . The agg ega e supply cu e supposedly eme ges om a a ional, e icien sphe e o ac i i y wi h no oom o des abilizing elemen s o human ancy. Shi s in he cu e occu o ‘objec i e’ easons, such as changes in he p oduc ion unc ion o he a ailabili y o ac o s o p oduc ion. This asymme y poses an obs acle o a supply-based heo y o s ag la ion o how could he u moil o s ag la ion, o igina e om his s able domain o ac i i y? Dis u bing as his ques ion migh ha e been, ew mac oeconomis s we e discou aged by i s implica ions. Fo decades, mac oeconomics made an e icien use o asso ed ‘impe ec ions’ o pa ch up he heo y o agg ega e demand, and he e was e y li le eason no o use his e y app oach in making necessa y adjus men s o he heo y o agg ega e supply. Supply p ices depend on ac o cos s. In an ideal neoclassical wo ld, such inpu cos s a e ‘endogenous’ o he sys em o hey eme ge as simple de i a i es om he p oduc ion unc ion: he wage a e is equal o he ma ginal p oduc o labou , he a e o p o i is equal o he ma ginal p oduc o capi al, and so on. Un o una ely, no ed many mac oeconomis s, ou own ma ke sys em was a om his ideal because some ac o s had he powe o se hei p ices highe han hei co esponding ma ginal p oduc s. In p inciple, such impe ec ions could dis o he p icing o e e y ac o ye p ac ical mac oeconomis s p e e o emphasize he pi o al ole o aw ma e ials and labou . B uno and Sachs (1985, p. 7) a e ypical when hey poin hei i s ‘blaming inge ’ a he wea he and he oil sheiks: 39 Phelps, E.S. and J.B. Taylo (1977) ‘S abilizing Powe s o Mone a y Policy Unde Ra ional Expec a ions,’ Jou nal o Poli ical Economy, Vol. 85, No. 1, Feb ua y, pp. 163-90. Phillips, A.W. (1958) ‘The Rela ion Be ween Unemploymen and he Ra e o Change o Money Wages in he Uni ed Kingdom, 1861-1957,’ Economica, New Se ies, Vol. 25, Nos. 97-100, pp. 283-99. Rowley, J.C.R. and D.A. Wil on (1974) ‘The Sensi i i y o Qua e ly Models o Wage De e mina ion o Agg ega e Assump ions,’ The Qua e ly Jou nal o Economics, Vol 88, No embe , pp. 671-80. Samuelson, P.A. and R.M. Solow (1960) ‘The P oblem o Achie ing and Main aining a S able P ice Le el: Analy ical Aspec s o An i-In la iona y Policy,’ Ame ican Economic Re iew, Vol. 50, No. 2, May, pp. 177-94. Sa gen . T.J. (1973) ‘Ra ional Expec a ions, he Real Ra e o In e es , and he Na u al Ra e o Unemploymen ,’ B ookings Pape s on Economic Ac i i y, No. 2, pp. 429-72. S igle , G.J. (1961) ‘The Economics o In o ma ion,’ Jou nal o Poli ical Economy, Vol. 69, June, pp. 213-25. —. (1962) ‘In o ma ion in he Labo Ma ke ,’ Jou nal o Poli ical Economy, Supplemen , No. 5, Pa 2, Oc obe , pp. 94-105. Taylo , J.B. (1975) ‘Mone a y Policy Du ing a T ansi ion o Ra ional Expec a ions,’ Jou nal o Poli ical Economy, Vol. 83, Oc obe , pp. 1009-21. —. (1979) ‘S agge ed Wage Se ing in a Mac oeconomic Model,’ Ame ican Economic Re iew, Pape s and P oceedings, Vol. 69, No. 2, pp. 108-13. —. (1985) ‘Ra ional Expec a ions Models in Mac oeconomics,’ in A ow and Honkapohja (eds.), pp. 391-425. Thu ow, L.C. (1969) Po e y and Disc imina ion (Washing on D.C.: The B ookings Ins i u ion). Tobin, J. (1972) ‘In la ion and Unemploymen ,’ Ame ican Economic Re iew, Vol. 62, No. 1, May, pp. 26-37. —. (1980) Asse Accumula ion and Economic Ac i i y (Ox o d: Blackwell). 46