Real Exchange Rate Adjustment In and Out of the Eurozone. Martin Berka Michael B. Devereux Charles Engel

Transcription

1 Real Exchange Rae Adjusmen In and Ou of he Eurozone Marin Berka Michael B. Devereux Charles Engel 5 h Bi-Annual Bank of Canada/European Cenral Bank conference on Exchange Raes: A Global Perspecive, ECB, Frankfur, Germany, June,

2 Real Exchange Rae Adjusmen in Eurozone Popular view is real exchange raes do no adjus in Eurozone o Counries would be beer off wih floaing raes Logic is based on Friedman s (1953) Case for Flexible Exchange Rae o Flexible exchange raes subsiue for sicky nominal expor prices o Achieve efficien erms of rade adjusmen This paper examines daa on real exchange raes in Europe, for counries in and no in he eurozone Also examines simple open-economy New Keynesian model Imporan cavea: we do no consider crisis period 2

3 Real Exchange Raes under Balassa-Samuelson The Balassa-Samuelson model relaes real exchange raes o changes in produciviy of raded goods An increase in Home raded produciviy will lead o a real Home appreciaion This resul holds as long as home bias in preferences is no oo grea A simple exension of Balassa-Samuelson is o allow for changes in produciviy of nonraded goods An increase in Home nonraded produciviy will lead o a real Home depreciaion Again, assuming home bias no oo grea. We also consider he effec of differences in economies (in our model deermined by differences in leisure preferences) ha lead o differences in uni labor coss. Counry wih higher ULC has higher price level 3

4 Inuiion of Friedman argumen P H - Home-currency price of Home good * P F - Foreign-currency price of Foreign good Friedman (and modern analyss) assume exporers se he price in heir currency * P H and P F adjus slowly If exchange raes are flexible, he erms of rade, * SP / P, adjus freely and efficienly H F Friedman s world assumes Prices are se in exporer s currency, even for consumers No capial flows. Nominal exchange raes clear goods marke. o This laer assumpion is clearly no applicable oday 4

5 Modern Keynesian models Some prices for final purchasers are se in buyer s currency and adjus slowly * o Relaive prices PF / P H and P * F / P H do no adjus auomaically as S changes. Oher prices se in producers currencies, so SP * F / P H adjuss o Bu S is responding o moneary and financial shocks Exchange raes are deermined in asse markes o Influenced by expecaions of fuure, risk premiums, moneary shocks, ec. o In LCP case, inefficien deviaions from LOOP o In PCP case, inefficien TOT movemens o Leads o inefficien wealh changes no based on produciviy changes 5

6 Our ack Build a prey sandard open-economy New Keynesian model o Compare he implied behavior of real exchange raes under Flexible prices (exchange rae regime doesn maer) Fixed exchange raes, sicky prices Floaing exchange raes, sicky prices Compare o behavior of real exchange raes for EU counries in and ou of eurozone o How hey correspond o relaive prices of nonraded o raded goods o How hey correlae wih relaive produciviy in raded and non-raded secors This is no a welfare analysis 6

7 Model Feaures Two counry model, complee asse markes Households ge uiliy from consumpion and leisure Firms produce oupu using labor and fixed capial Households ge uiliy from Home and Foreign goods Each good produced by a monopolis All goods use non-raded disribuion services Price sickiness Calvo pricing Under floaing exchange raes, we consider LCP, PCP, and a mix of firms pricing in LCP and PCP. Produciviy shocks in Home and Foreign, in boh raded and non-raded goods producion Leisure demand shocks ha may proxy for changes in labor laws or union negoiaions Moneary policy se by Taylor rule, wih shocks In currency union, a common moneary policy Under floaing raes, separae moneary policies and separae moneary policy shocks 7

8 Model Feaures 1 σ 1+ φ C N Households: U = E 0 β χ = 0 1 σ 1+ φ θ θ θ θ θ θ = γ T + (1 γ) N C C C λ λ λ = ω + λ ( 1 ω) λ T H F C C C φ φ φ φ = κ + ( κ) φ H H 1 H C I V φ φ φ φ = κ + ( κ) φ F F 1 F C I V 8

9 Model Feaures 1 1 θ 1 θ 1 θ Price Indexes: P = ( γ + (1 γ) ) PT P N 1 λ ( ω) λ = λ T H F ( ω ) P P P 1 1 φ 1 ( κ) φ 1 = + 1 φ H H N P P P ( κ ) 1 1 φ 1 ( κ) φ 1 = + 1 φ F F N P P P ( κ ) Real Exchange Rae: RER = SP P * 9

10 Model Feaures Household FOCs: W C C I I = χ PC N σ φ T N H F P T = γ P θ C P N = (1 γ ) P θ C φ P P = κω H H PH PT λ C T φ P F PF = κ ( 1 ω) PF PT λ C T 10

11 Model Feaures Firms: Y () i = A N () i H H H Y () i = A N () i N N N α α Relaive produciviy shocks are key variables ha move relaive prices and real exchange raes in long run Flexible prices: Serves as a benchmark. Wih opimal subsidy in place, prices are efficien Wihou subsidy, relaion beween relaive prices and relaive produciviy mimics efficien prices However, he shocks o leisure demand migh have an inerpreaion as an inefficien resricion in labor supply arising from unions or governmen regulaions. 11

12 Model Feaures Calvo pricing We compare he flexible price model o Calvo pricing. Under fixed exchange raes, he currency of pricing is irrelevan. Under flexible exchange raes, we consider LCP, PCP, and a version in which some firms price LCP and he res PCP. Moneary policy r = ρ + σπ p + σq( q u) + σs( s s 1) Under fixed exchange raes, σ Under floaing exchange raes, σ = 0 s s 12

13 Model Feaures Equilibrium Financial Markes (complee) C P C σ * σ = * SP Goods Markes Y = + * H IH IH Y = I + I * * F M M YN = CN + IHN + I MN Y = C + I + I * * * * N N HN MN Labor Markes N = NN + N H N = N + N * * * N H 13

14 Inuiion from decomposiions q = (1 γ ) qn + q T, where 1 γ is non-raded share in C q = p p ( p p ) - relaive price of nonraded goods * * N N T N T 1 κ qt = qn + (2ω 1) τ + Δ, κ where 1 κ is share of cos due o disribuion, * * τ = pf ph = pf p H - erms of rade Δ = p + s p = p + s p F - deviaions from LOOP * * H H F q Since effecively 2ω 1, if no deviaions from LOOP, hen, q and q should all be highly correlaed. N T 14

15 Inuiion from decomposiions All should be driven by produciviy shocks and labor supply shocks: An increase in a a should lead o a decline in q, q, H F N q T An increase in a * N an should lead o an increase in q, qn, qt * An increase in ln( χ) ln( χ ) should lead o a decline in, q, q N qt 15

16 Price daa Price daa are from Euorsa, based on Eurosa PPP projec Annual, , for 146 consumer goods These are price level daa. We consruc raded and non-raded indexes, using expendiure weighs for each counry 12 Eurozone counries, 6 non-eurozone core European counries, as well as 13 non-eurozone periphery European counries 16

17 Produciviy and Uni Labor Cos Daa We consruc secoral level TFP daa We comine he Groningen Grwoh and Developmen Cener 1997 TFP level daabase wih he KLEMS daabase on secoral produciviy indexes We use measures of radeabiliy o aggregae he indexes ino a radable and non-radable TFP index for each counry. See Appendix for deails Uni Labor Cos daa are consruced from he OECD STAN daabase TFP daa for 9 Eurozone counries (no Greece, Porugal or Luxembourg) TFP daa for 6 non-ez counries (Sweden, Denmark, UK, Hungary, Slovenia, Czech Republic) 17

18 Some Summary Saisics 18

19 Price Regressions Eurozone: q = a+ b* qn Mehod: Pooled FE RE XS ˆb 0.70** 0.60** 0.61** 0.71** s.e. (0.06) (0.08) (0.07) (0.25) Floaing: q = a+ b* qn Mehod: Pooled FE RE XS ˆb 0.26** 0.79** 0.72** 0.17 s.e. (0.10) (0.15) (0.14) (0.14) 19

20 Simulaions Eurozone: q = a + b * Traded + c * NonTraded + d * ULC Mehod: Pooled FE RE XS ˆb s.e ** (0.06) -0.18** (0.09) -0.26** (0.08) -0.93** (0.19) ĉ s.e. 0.29** (0.08) 0.36** (0.18) 0.36** (0.13) 0.27 (0.22) ˆd s.e ** (0.08) -0.46** (0.07) -0.46** (0.08) -0.43** (0.20) Floaing: q = a + b * Traded + c * NonTraded + d * ULC Mehod: Pooled FE RE XS ˆb s.e (0.12) 0.21 (0.08) 0.28 (0.09) 0.32 (0.33) ĉ s.e (0.07) (0.24) (0.12) (0.16) ˆd s.e ** (0.09) -0.57** (0.05) -0.53** (0.06) -1.53** (0.26) 20

21 Model Calibraion We assume sandard parameers for elasiciies: γ and κ - non-radable shares = 0.5 λ - elasiciy of home/foreign goods = 8.0 ω - assume no home bias = 0.5 α - producion funcion curvaure = 1.0 σ - relaive risk aversion = 2.0 ψ - Frisch elasiciy of labor supply = 1.0 φ - elas. of subs. beween good and dis. service = 0.25 θ - elas. of subs. beween raded and nonraded = 0.7 β - se = 0.99 for quarerly daa δt, δ = 0.10 N 10% price adjusmen each quarer produciviy processes AR1: se o mach daa labor supply shock process: se equal o produciviy process 21

22 Model Calibraion Sd dev (ime series) Sd dev (crosssecion) Fixed - Sicky Flex Price Daa (0.030,0.042) (0.071,0.125) Serial correlaion (0.720,0.880) (0.036,0.050) (0.085,0.131) (0.570,0.759) Regression of Real Ex. Rae on Relaive Nonraded Price Time series (1.567,1.628) Cross-secion (0.791,1.052) (1.558,1.617) (0.877,1.068)

23 Discussion Regression of Real Ex. Rae on produciviy and ULC Fixed - Sicky Flex Price Daa Time series Traded Prod (-0.162,-0.065) (-0.201,-0.169) Time series Nonraded Prod (0.423,0.580) (0.155,0.218) Time series ULC (-0.580,-0.284) (-1.470,-1.320) Cross-secion Traded Prod (-0.662,-0.498) (-0.654,-0.545) Cross-secion Nonraded Prod (0.015,1.150) (0.143,0.955) Cross-secion ULC (-1.608,0.364) (-1.471,0.128)

24 Discussion Real exchange raes are slighly less volaile and somewha more persisen under sicky prices/ fixed exchange raes han under flexible prices Why does he real exchange rae behavior looks so good under fixed exchange raes? Nominal prices do adjus. We have assumed fairly sicky prices (half life of over 1 year), bu wih regressions on annual daa, he price adjusmen is relaively large. Required real exchange rae adjusmen is no large in he efficien economy. Price adjusmen always goes in he righ direcion. This is in conras o flexible exchange raes, where relaive prices and real exchange raes respond o moneary and risk premium shocks 24

25 Conclusions No a welfare analysis of currency unions Would need o consider, e.g. Loss of independence of moneary policy Gains from policy credibiliy May allow counries o overcome orginal sin May spur greaer fiscal and poliical cooperaion Our daa mosly are during normal imes, and we have no considered eiher acual or opimal real exchange rae adjusmen in crisis imes. Our poin here is simply ha real exchange raes have adjused well in he Eurozone. 25