The Relationship between Wage and Inflation: Case of Slovenia and Selected Central and Eastern European Countries

Transcription

1 The Relaionship beween Wage and Inflaion: Case of Slovenia and Seleced Cenral and Easern European Counries Simona Bovha Padilla, Helios Padilla Mayer January, Absrac: In he paper we es he new Phillips curve for Cenral and Easern European EU accession counries for he period from 99 o and use i o compare he efficiency of he radiional Phillips curve. More specifically, we wan o see wheher real marginal cos, which includes labor produciviy and real wage componens, can accoun for inflaion dynamics in he observed sample. Surprisingly, when observing all eigh seleced counries, he relaion beween real marginal cos and inflaion is opposie han expeced. On he oher hand, inflaion in Balic Saes and Slovenia seems o be influenced by real marginal cos. The elasiciy coefficien of real wages on inflaion for Slovenia shows ha inflaion was quie responsive o movemen in wages during he oal period, however, inflaion became quie inelasic wih respec o wages afer. Thus, economic policies ha were inroduced in Slovenia afer were quie efficien in wage regulaion, alhough he real effec will be observed in a more advanced period. Simona Bovha Padilla, IEDC Bled School of Managemen, Prešernova c., 6 Kranj, Slovenia, e- mail:simona.bovha@iedc.si, Helios Padilla Mayer, BP&Mayer Consulans, Vojkova 77, Ljubljana, Slovenia, hpmayer@bpmayer.com

2 . Inroducion One of he main asks of he Cenral Banks of he laes EU accession counries is o achieve price sabiliy and herefore i is really imporan o deermine sources and naure of inflaion in his area. Our research arges effecs of wages on inflaion. Or, more precisely, we ry o see wheher here is an evidence of wage pressures on inflaion. The mos convenional model used for esing his relaion is he sandard Phillips curve approach. However, we follow Gali, Gerler and Lopez-Salido () and es relaion beween wages and inflaion on he basis of so called new Phillips curve lieraure. As he sandard Phillips curve, also his one assumes ha inflaion varies posiively wih real secor economic aciviy in he shor run. However, he difference is ha he new Phillips curve relaes inflaion o movemens in real marginal cos. Thus, real marginal cos is he heoreically appropriae measure of real secor inflaionary pressures, as opposed o he cyclical measures used in radiional Phillips curve analysis, such as derended oupu and unemploymen. One of he advanages of he real marginal cos measure (which in our analysis corresponds o real uni labor cos) is ha i direcly accouns for he influence of boh produciviy and wage pressures on inflaion. Thus, according o he new Phillips curve heory, produciviy, wages and inflaion follow he similar pah. We wan o see wheher his relaion holds recen new EU member counries from Cenral and Easern Europe (Czech Republic, Hungary, Esonia, Lavia, Lihuania, Poland, Slovakia and Slovenia). More specifically, we wan o es wheher wages did have inflaionary pressures in Slovenia and if so, wha policy measures can be implemened o avoid his effec. In secion we firs presen analyze he connecion beween inflaion and wages on he basis of radiional Phillips curve approach. Then, in secion we discuss he new Phillips curve. Secion develops he heoreical model used for esimaion. Secion 5 presens empirical resuls and draws comparisons beween he full counry sample and wo sub-samples (Balic Saes including Slovenia and Cenral European Counries). Secion 6 concludes.

3 . Tradiional Phillips Curve The radiional Phillips curve relaes inflaion o some cyclical indicaor (eiher level of unemploymen of real GDP) and lagged values of inflaion. In our analysis, we are going o use he real GDP as he cyclical indicaor. Thus, if ð denoes inflaion and ŷ he log deviaion of GDP from is long run rend, hen he common specificaion of he Phillips curve is: h = γiπ + δ ˆ, () i= π y + ε where ε is a random disurbance. This equaion can be undersood as a demand-side explanaion of inflaion; aggregae demand increases following a rise in money supply. I is someimes assumed ha he sum of weighs on lagged inflaion equals o one so ha here is no long-run radeoff beween inflaion and oupu. We did no impose his assumpion in our analysis because coefficiens on lagged inflaion variables in he regression equaion added up close o one. We run he generalized leas squares (GLS) regression conrolling for fixed effecs based on equaion () for he full counry sample, and hen (due o heir differences) wo subsamples, (a) Cenral and Easern European counries, and (b) Balic Saes and Slovenia. We use quarerly daa from 99 o. 5 We run several regressions and he one where we use four lags for inflaion fis bes he quarerly daa. To measure inflaion we use log difference of he CPI index, which we obain from naional saisic daabases on a quarerly basis. The oupu erm is HP filered real GDP on a quarerly basis. We also obain daa on quarerly real GDP from naional saisic daabases. Figure. shows inflaion rae for Slovenia and Cenral and Easern European EU candidae counries on a For example, an increase in price level can be caused by a rise in aggregae demand, which is due o an expansionary fiscal policy. Due o heeroskedasiciy of daa and auocorrelaion in he daa series, ordinary leas square (OLS) esimaes are no mos efficien, so insead we choose o run he GLS regression and obain unbiased as well as mos efficien coefficiens. We also run he regression for Slovenian daa only, bu since he size and sign of coefficiens do no change significanly, we only repor regression resuls on he wo sub-samples. 5 We do no include daa for firs wo quarers of because i is no available for he full counry sample.

4 yearly basis (average inflaion rae). 6 same counry sample on a yearly basis. Figure. plos he real GDP growh raes for he The corresponding equaion for he full daa se is: π =.567π +.9π.5π +.π -.9ŷ (.7) (.) (.95) (.95) (.7) F saisic = 6.7, p- value =. rho =.(fracion of variance due o u_i) - where sandard errors are shown in parenheses. The esimaes for he wo sub-samples are as follows: π () Balic Saes wih Slovenia =.7π +.π +.5π.57π -.76ŷ (.9) (.9) (.) (.7) (.8) F saisic = 5.8, p- value =. rho =.98(fracion of variance due o u_i) - π () Cenral European Counries =.68π +.π.6π +.6π -.6ŷ (.5) (.6) (.5) (.9) (.667) F saisic =.8,p - value =. rho =.9(fracion of variance due o u_i) - Surprisingly, in all hree cases, here is a negaive relaion beween he HP filered oupu and curren inflaion. Coefficiens on oupu for he full sample and a sub-sample for Balic Saes and Slovenia are significan a 5 percen level, bu he coefficien on Cenral European counries is no significan. Gali, Gerler and Lopez-Salido () on he oher hand show ha here was a posiive relaion beween he derended oupu and inflaion, 6 For he purpose of presenaion, inflaion raes are expressed in logarihmic form.

5 which acually corresponds o he real essence of he radiional Phillips curve. Thus, he radiional Phillips curve is no really able o explain he sabiliy of relaion beween inflaion and oupu over he observed period nor accoun for he policy reforms ha were inroduced in hese counries in order o suppress inflaion. Moreover, coefficiens on lagged inflaion may very well embed expecaions of fuure inflaion. Hence, i is necessary o remodel he radiional Phillips curve in order o observe he wheher inflaion can be beer explained hrough he supply side of he economy.. The New Phillips Curve The new Phillips curve is based on saggered nominal price seing (Taylor, 98). The price seing behavior is based on opimizaion by monopolisically compeiive firms. The basic equaion ha we are going o es is he one ha relaes inflaion ð o he anicipaed fuure inflaion ð + and real marginal cos: ( π ) Mˆ λ π + = βe + () where Mˆ is he average real marginal cos, in percen deviaion from is seady sae level, â is subjecive discoun facor, and ë is he slope coefficien ha depends on he parameers of he model. This equaion shows direc effecs of he fricions in he price adjusmen process ha are he key aspecs of he heory (Gali,Gerler,Lopez-Salido, ). Neverheless, he new Phillips curve lieraure (Goodfriend and King, 997) relaes inflaion o an oupu gap variable. 7 Assuming resricions on echnology and labor marke srucure, real marginal cos are proporionally relaed o he oupu gap (Roemberg and Woodford, 997), Mˆ ( * = δ y y ), () 5

6 where y and y* are he logarihms of real oupu and he naural level of real oupu, respecively. Combining equaions () and () resuls in he sandard oupu gap based for relaions of he new Phillips curve. ( π ) + k( y y* ) π, () = βe + where k = λδ. However, in he lieraure equaion () has been he subjec of a considerable discussion. According o he radiional Phillips curve, inflaion varies posiively wih he oupu gap. On he oher hand, in he new Phillips curve conex, inflaion is an enirely forwardlooking phenomenon. By ieraing equaion () forward we ge: k * π = k β E ( y+ k y+ k ), (5) k = This equaion shows ha if a cenral bank can commi o sabilizing he oupu gap ( y * + k y + k ), i can reach price sabiliy. On he oher side his equaion also sugges ha inflaion should anicipae movemens in he oupu gap. So far, as he approximaion of he radiional Phillips curve recommend, he oupu gap ends o lead inflaion. Gali, Gerler, and Lopez-Salido,, Sbordone, 999, and Gali and Gerler, 999 have shown ha his relaion holds for he US as well as for he EU counries. Thus, he oupu gap formulaion of he new Phillips curve canno accoun for he persisence of inflaion eiher for he US or he EU counries. Neverheless, above menion auhors find ha he relaion beween inflaion and real marginal cos given by equaion () is some how consisen wih he daa. Thus, he link beween he real marginal cos and he oupu gap corresponds wih daa. I seems ha 7 For a deailed discussion on oupu gap refer o Bovha Padilla and Padilla Mayer,. 6

7 real marginal cos response sluggishly o oupu gap movemens, he same way as inflaion does. The firs possible explanaion for ha is ha convenional measures of he oupu gap are no adequae. I may no be appropriae o use derended oupu o esimae he oupu gap if here are significan real shocks o he economy. The second explanaion could be ha here are rigidiies in he labor marke (eiher in he form of real or nominal wage rigidiies), which break down he relaion beween marginal cos and oupu gap (equaion ). These rigidiies offer an explanaion for inacive behavior of real marginal cos.. The Baseline Model In his secion we presen relaion beween inflaion and real marginal cos across firms. As Gali, Gerler and Lopez-Salido,, we assume ha firms face increasing real marginal cos. 8 We assume ha here is a coninuum of firms indexed by j [, ]. Each firm is a monopolisic compeior, i produces a differeniaed good Y ( j ), and sells i a nominal price P ( j ). The demand curve ha each firm faces is hen Y ( j) Y and P P ( j ) ε Y =, where P are aggregae oupu and aggregae price, respecively. The producion funcion for firm j is given by α ( ) ( ) Y j = A L j, where A is a echnological parameer, L ( j) is employmen, and α is he share of labor in income. Firms se nominal prices on a reel basis as in Calvo, 98. Each firm reses is price wih probabiliy θ each period, while a fracion θ of producers leave heir prices unchanged. Thus, he expeced ime ha he price remains fixed is parameer θ hen shows he degree of price rigidiy. θ and he 7

8 The price level p can be expressed as a funcion of he newly se price p* and he lagged price p. 9 p = ( θ ) p* + θp (6) All firms ha change price in period choose he same value of p *. A firm ha is able o rese price in period chooses price o maximize expeced discouned profis given echnology, facor prices and he consrain on price adjusmen (defined by he rese probabiliy rule: θ). An opimizing firm will se p * according o he following log-linear k n ( βθ ) ( βθ ) E ( M ) * p = log +, + k k = µ, (7) where β is a discoun facor, ( n ) period M + k, is he logarihm of nominal marginal cos in + k of a firm ha las rese is price in period, and desired gross markup. ε µ = is he firm s ε We need o find an expression for inflaion in erms of an observable measure of aggregae marginal cos. By cos minimizing rule he firm s real marginal cos equals he real wage divided by he marginal produc of labor. Given he Cobb-Douglass producion funcion, he real marginal cos is given by M, + k = ( W / P ) + k ( α )( Y / L ), + k + k, + k, where Y + k, and L, + k are oupu and employmen for a firm ha has se price in a he 8 Allowing marginal cos o vary across firms gives more reliable esimaes of he degree of price rigidiy in he Euro area as showing in Gali, Gerler and Lopez-Salido,. 9 This expression is obained afer log-linearizing he price index around zero inflaion seady sae and i is aken in a log form. 8

9 opimal value P *. Since we canno observe individual firm marginal cos (absence of firm level daa), we define average marginal cos, which depends only on aggregaes: M ( W / P ) ( α )( Y / L ) = (8) By exploiing he assumpions of Cobb-Douglas producion funcion and isoelasic demand curve, we obain he following log- linear relaion beween M + k, and M : ˆ ˆ εα * M, + k = M + k ( p p + k ) (9) α where ˆ and M ˆ + k are he log deviaions of M, + k and M from heir respecive M, + k seady sae values. We obain he formulaion of he new Phillips curve ha relaes inflaion o he real marginal cos by combining equaions (6), (7), and (9), ( π ) Mˆ λ π = βe + +, where () ( θ)( βθ)( α) λ = () θ [ + α( ε ) ] Slope of he coefficien λ depends of parameers of he model. λ is decreasing in he degree of price rigidiy, θ, which shows he fracion of firms ha keep heir prices consan. Thus, he smaller he fracion of firms ha adjus prices, he less sensiive inflaion will be o movemens in marginal cos. Moreover, λ is also decreasing in he slope of producion funcion, measured by α, and in he elasiciy of demand, ε. The larger α and ε, he more sensiive is he marginal cos of an individual firm o deviaions of is price from he average price level. Finally, he model suggess ha inflaion should equal a discouned sream of expeced fuure average real marginal coss. 9

10 5. Empirical Resuls In his secion we presen esimaes of he baseline model (equaion ) for he sample of seleced new EU member counries. Since we are ineresed o observe wheher Slovenia follows he full sample, we also perform analysis of Slovenia inflaion dynamics separaely. We use quarerly daa from 99 o. Daa for inflaion and real GDP are obained from he Counries in Transiion CD-Rom (on a yearly basis), WIIW, and quarerly series are obained from he naional saisical offices. Inflaion is given as a percenage change in prices wih respec o he previous year, and he real GDP daa are given in 995 prices. We calculae average real marginal cos (equaion 8) in he following way. Firs, we obain daa on gross monhly real wages per capia, quarerly daa on real GDP and labor force. Since we calculae quarerly marginal cos, we se up he nominaor as a sum of he hree monh gross real wages per capia. We also need daa on labor income share, which we calculae according o Dobrinsky (). He presened a measure of capial income share for counries in ransiion. According o him, he capial income share can be calculaed as he share of gross operaing surplus and gross mixed income compared o oal facor incomes. Thus, he labor income share is simply calculaed as a share of compensaion of employees compared o oal facor incomes. Since we canno obain quarerly daa on compensaion of employees and oal facor income, we use yearly daa and herefore calculae labor income share on a yearly basis. However, we believe ha his share is no varying significanly during he year and we use i for calculaing quarerly real average marginal cos. Thus, he denominaor is calculaed as a produc of annual labor income share and raio of a real GDP and labor force on a quarerly basis. Then, we use he log deviaion of real marginal cos from is mean as a measure of Mˆ. We do no include daa for firs wo quarers of because i is no available for he full counry sample. Toal facor incomes can be calculaed as he difference beween he nominal GDP and ne indirec axes.

11 Figures 5. and 5. show our measure of real marginal coss along wih inflaion for new EU member counries. Surprisingly, when we ake ino accoun he full sample, variables do no move closely ogeher. Alhough correlaion beween inflaion and real marginal coss is posiive, i is exremely small and non-significan. Thus, i is hard o say ha inflaion is really relaed o movemens in marginal coss as suggesed by he heory explained in Secion. Moreover, Table 5. shows correlaion beween inflaion and real marginal cos for all counries ha are in he sample. Inflaion could be bes explained by variaion in real marginal cos in Balic region (Esonia, Lavia and Lihuania). Correlaion beween boh variables is relaively high also for Slovenia. Moreover, elasiciy coefficien for Slovenia shows ha inflaion is relaively responsive on real wages. Increasing wages by percen would increase inflaion by.76 percen. Thus, according o he real daa from he observed period, wages did have inflaionary pressures in Slovenia. On he oher hand, here is srong negaive correlaion beween inflaion and real marginal cos in Poland, Czech Republic, Slovakia and Hungary. Table 5.: Correlaion beween Real Marginal Cos and Inflaion Counry Correlaion Czech Republic -.7 Esonia.7 Hungary -.6 Lavia.66 Lihuania.6 Poland -.8 Slovakia -.55 Slovenia.5 Average. We hen presen empirical esimaes of coefficiens in equaion (9). Firs, we presen esimaes of he baseline model for he full sample, and hen for wo sub-samples: ()

12 Balic Saes wih Slovenia, and () Cenral European counries. We show ha in conras wih findings for he Euro area and he USA (Gali, Gerler, Lopez-Salido, ) his model does no explain he dynamics of inflaion in he Cenral and Easern European EU candidae counries in expeced way. 5.. Baseline Model Esimaes We esimae he coefficiens in equaion (9) by using generalized mehod of momens (GMM). Under raional expecaions, equaion (9) defines he se of orhogonaliy condiions: (( βπ ) z ) E π λmc, + = where z denoes a vecor of variables observed in ime, ha is, insrumenal variables. The opimal choice of insrumenal variables are lags of explanaory variables already presened in he model: lags of inflaion, lags of real marginal cos and derended oupu. Since he daa series consiss only from 5 observaions (per counry), we choose relaively small number of lags for insrumens in order o minimize he poenial esimaion bias ha can arise in small samples. As Gali, Gerler and Lopez-Salido,, we use four lags of inflaion, wo lags of he real marginal coss, and derended oupu. The corresponding equaion for he full daa se is: π =.5 E ( ).7 ˆ π + M F saisic = 6., p - value =. (.7) (.) J saisic = 5., p - value =. where sandard errors are shown in parenheses. The esimaes for he wo sub-samples are as follows: We ried o esimae coefficiens wih using more han four lags for inflaion and more han wo lags for he real marginal coss, bu resuls were no affeced.

13 () Balic Saes wih Slovenia π =. E ˆ ( π + ) +.69 M F saisic =.7, p - value =. (.) (.78) J saisic =., p - value =. π () Cenral European Counries =.78E ˆ ( π + ).695M Fsaisic = 8.65,p - value =. (.8) (.77) J saisic =.65,p - value =. In all hree equaions, sandard errors are modified using a Newey-Wes correcion o accoun for serial correlaion in he error erms. Also, we ake care of fixed effecs in all hree regressions. In order o check for poenial weakness of he insrumens, we perform an F-es applied o he firs sage regression. Resuls show ha insrumens ha we use are relevan (see F saisic in above regressions). We also es for overidenifying resricions of he model. On he basis of Hansen es, we rejec overidenifying resricions for all hree regressions (see J saisic in above regression). In general, he empirical model does no work as expeced in he full sample and in he sub-sample of Cenral European counries (Czech Republic, Slovakia, Poland). The coefficiens on marginal cos are negaive and significan a 5 percen level in boh cases. The only case ha coefficien is posiive and significan as implied by he heory is in he case of Balic Saes including Slovenia. This resul is in conras wih findings of Gali, Gerler, and Lopez-Salido,, where hey esimae posiive and highly significan effec of marginal cos on inflaion in he Euro area and he USA. 5 Again, we run an individual regression analysis for Slovenia, bu since he size and significance of coefficiens do no change, we repor resuls only on he wo sub-samples. The model is overidenified is here is more han one way o calculae is parameers from he reduced form parameers, leading o resricions on he reduced form parameers. 5 However, we need o ake ino he accoun ha he findings for he Euro are and he USA are based on he quarerly daa se from 97 o 998. Afer ha ime here has been a decline in unemploymen and a

14 6. Final Remarks Our resuls indicae ha neiher radiional nor a marginal cos-based Phillips curves do no provide a saisfacory explanaion of Cenral and Easern European area inflaion from 99 o. Firs, due o policy reforms ha were oriened oward reducion of inflaion in he pos ransiion period, he radiional Phillips curve did no show he posiive relaion beween he oupu and price level neiher in he full sample nor in he wo sub-samples. The new Phillips curve model suggess ha marginal cos is negaively relaed o inflaion for he full sample. On he oher hand, resuls for Balic Saes and Slovenia are in accordance wih he model predicion. According o his informaion, wages in Slovenia did have inflaionary pressures. However, one should be aware ha his resul is obained on he basis of quarerly daa from 99 o. Since here have been wage policies inroduced which allowed he increase of wages only wih respec o an increase in produciviy and hus should no have inflaionary pressures. Indeed, if one checks he elasiciy of wages wih respec o inflaion on he basis of daa from on, he coefficien drops o.67 percen (wih respec o.76 percen when full period daa is aken ino accoun). However, if one runs he regression on he basis of he new Phillips curve wih he daa from on, resuls do no change: here is sill posiive and significan relaion beween real marginal cos and inflaion. Therefore, a his poin i is hard o say how marginal cos really affecs inflaion dynamics and hus give more precise policy recommendaions. I would be appropriae o decompose he movemen in real marginal cos in order o isolae he facors ha drive his variable. One possible decomposiion is ino a componen ha accouns for labor marke fricions (measured as he wage markup) and he oher one ha would correspond o he fricionless compeiive equilibrium. ris e in he oupu growh in he Euro area, wihou any corresponding rise in inflaion. Thus, including more recen period in he analysis of Euro inflaion dynamics may aler resuls.

15 Moreover, o deermine wheher wages really have inflaionary pressures on seleced counries in he observed sample, i is necessary o undersand he deerminans of he wage markup. Neverheless, i will be necessary o include in he analysis addiional facors, such as union pressures, ha would accoun for variaion in he wage markup and hus explain is effec on inflaion dynamics of Cenral and Easern European EU candidae counries. 7. References. Bovha Padilla, S. and H. Padilla Mayer,. Sources of GDP Growh, Poenial Oupu and Oupu Gap for Slovenia: Mid-Term Analysis, IB Review, Insiue for Macroeconomic Analysis and Developmen, Ljubljana, Slovenia, June.. Bovha Padilla, S. and H. Padilla Mayer,. Sources of Growh in Seleced Cenral and Easern European Counries, Counries in Transiion Inernaional Conference, Kranjska Gora, Slovenia, June.. Calvo, G., 98. Saggered Prices in a Uiliy Maximizing Framework, Journal of Moneary Economics,, Eurosa, Saisical Informaion, various issues. 5. Eurosa, Saisical Yearbook for Cenral and Easern European Counries, various issues. 6. Gali, J. and M. Gerler, 999. Inflaion Dynamics: A Srucural Economeric Analysis, Journal of Moneary Economics,, Gali, J., M. Gerler and J. Lopez-Salido,. European Inflaion Dynamics, NBER Working Paper Series, Working Paper 88, 9 pp. 8. Goodfriend, M. and R. King, 997. The New Neoclassical Synhesis, in NBER Macroeconomic Annual, Ben Bernanke and Julio Roemberg, eds., MIT Press. 9. Inriligaor, M., R. Bodkin, and C, Hsiao, 978. Economeric Models, Techniques and Applicaions, Prenice Hall, Upper Saddle River, NJ.. Roemberg, J. and M. Woodford, 997. An Opimizaion-based Economic Framework for he Evaluaion of Moneary Policy, in NBER Macroeconomic Annual, Ben Bernanke and Julio Roemberg, eds., MIT Press.. Sbordone, A., 999. Prices and Uni Labor Coss: A New Tes of Price Sickiness, mimeo, Rugers Universiy.. Taylor, J., 98. Aggregae Dynamics and Saggered Conracs, Journal of Moneary Economics,, WIIW: Counries in Transiion, CD-Rom. 5

16 Figure.: Inflaion in Slovenia and Cenral and Easern European EU Candidae Counries from 99 o (log form) 6.5 ln Inflaion Average Slovenia Source: Counries in Transi ion, WIIW, Eurosa, seleced publicaions. Figure.: Real GDP Growh Raes in Slovenia and Cenral and Easern European EU Candidae Counries from 99 o, in percen in % Slovenia Average Source: Counries in Transiion, WIIW, Eurosa, seleced publicaions. 6

17 Figure 5.: Inflaion and Real Marginal Cos in Cenral and Easern European EU Candidae Counries* Inflaion Real Marginal Cos * Daa are in logarihmic forms. Lef-side scale measures inflaion and he righ-side scale measures real marginal coss as explained in Secions and. Source: Auhors calculaions. Figure 5.: Inflaion and Real Marginal Cos in Cenral and Easern European EU Candidae Counries* a) Czech Republic Inflaion Real Marginal Cos

18 b) Esonia 5..5 Inflaion Real Marginal Cos c) Hungary.5 Inflaion Real Marginal Cos

19 d) Lavia 6 5 Inflaion Real Marginal Cos e) Lihuania Inflaion Real Marginal Cos

20 f) Poland 7 6 Inflaion Real Marginal Cos g) Slovak Republic Inflaion Real Marginal Cos

21 h) Slovenia Inflaion Real Marginal Cos * Daa are in logarihmic forms. Lef-side scale measures inflaion and he righ side scale measures real marginal coss as explained in Secions and. Source: Auhors calculaions.