ACADEMY OF ECONOMIC STUDIES BUCHAREST DOCTORAL SCHOOL OF

ACADEMY OF ECONOMIC STUDIES, BUCHAREST DOCTORAL SCHOOL OF FINANCE AND BANKING DOFIN Dissertation Paper Real effective exchange rates and their influence on Romania’s trade with European Union Countries MSc Student : Grigorescu Madalina Supervisor : Professor Moisa Altar

Topics Objectives n Introduction n Review of the literature n Theoretical models & formulas n Empirical analysis n Conclusions n

Objectives n To determine REER based on CPI and PPI indices weighted by the export volume of Romania to European Union countries n To provide an empirical investigation on the Romania’s REER influence on its trade with European Union countries n Export n Import n Trade balance graph

Introduction: Real Effective Exchange Rate n n n Useful indicator of one country’s competitiveness The appropriate definition and calculation of REER depend upon the economic issue to be demonstrated and data availability The “effective” aspect of REER is referring to the weights to be put upon each interacting partner country § § n Import-weighted indices Exports-weighted indices Total direct trade (export and imports) Multilateral export-weight Indices to be included in REER’s measurement formula § § CPI PPI GDP deflators ULC each having its advantages and disadvantages

Theoretical models and formulas n RER = nominal exchange rate adjusted for price level differences between countries (domestic P and abroad P* ) REER is usually presented in several context including: n real exchange rate 1)REER= relatingmultilateral real exchange rates to productivity differencials 2) estimating the relative price responsiveness of the trade flow 3) assessing its impact on country’s competitiveness

Review of the literature: Studies on EU accession countries n Barell, Dawn, Smidkova (2002) „Estimates of Fundamental real effective exchange rate for the five EU preaccession countries” n Stability of REER will not automatically be in line with economic developments n De Broeck, Slok (2001) „Interpreting real exchange rate movements in transition countries” n EU accession countries can expect to experience further productivity –driven REER appreciations n Egart, Balasz (2002) „Investigating the Balassa-Samuelson hypothesis in transition : do we understand what we see? ” n Continuous capital inflows will upward pressure on nominal exchange rate and provoke exchange rate to appreciate to unsustainable levels n Egart, Balasz and Drine , Imed and Rault, Cristophe (2002), „On the Balassa-Samuleson effect in the transition countries : a panel study” n Evidence for Romania : cointegration very unstable n Stucka, Tihomi (2004) „The effect of exchange rate change in the trade balance in Croatia” n It is questionable weather permanent depreciation is desirable to improve the trade balance n Kim, Korhonen (2002), ”Equilibrium exchange rates in transition countries: evidence from dynamic panel models” n Serious challenges for the exchange rates policies in EU accession countries as joining Euro at the current level of exchange rate risks undermining exports to EU countries

Theoretical implications: n When REER rises (REER depreciates) -> each unit of domestic output purchases fewer units of foreign output; ¨ Foreign consumers demand more of our products-> the volume of exports will rise ¨ Domestic consumers purchase fewer units of expensive foreign products -> imports decreases measured in foreign output units but increases measured in domestic output units n When REER decrease (REER appreciates) -> the opposite situation n The evolution of the exports is obvious while the evolution of imports is ambiguous n All things equal, the volume effect of REER changes outweighs the value effect , and a depreciation of REER improves the trade balance and an appreciation worsens the trade balance

Empirical analysis n Data series n Results

Data series n n Period : 1990 -2003 Frequency : quarterly data n Log of REER_CPI index calculated as a geometric average using CPI index and weights as bilateral exports of Romania with EU countries n Log of REER_PPI index calculated as a geometric average using PPI index and weights as bilateral exports of Romania with EU countries n Log of Exports and Imports series of Romania with EU countries n Log of Trade Balance of Romania with EU countries back

Results n Unit n root tests on series Augmented Dickey Fuller tests: Given the I(1) nature of the series, the cointegration analysis is employed to explore the long-run relationship among the variables n Cointegration n Vector n n analysis Error Correction Models To observe short-run deviations of variables from long-run equilibrium path To see the speed of adjustment of the variables to shocks from long-run equilibrium

Cointegration analysis For the obtained number of lags I found cointegration equation for Export and REER and for Import and REER both for the 5% level of significance

Export and REER_CPI and REER_PPI Lags interval (in first differences): 1 to 5 Unrestricted Cointegration Rank Test Import and REER_CPI and REER_PPI

Pairwise Granger Causality Tests Sample: 1990: 1 2003: 4 Lags: 1 Null Hypothesis: REER_CPI does not Granger Cause EXPORT Obs F-Statistic Probability 55 12. 7740 0. 00077 1. 37356 0. 24654 30. 6393 2. 3 E-09 1. 05514 0. 35592 3. 82998 0. 01571 0. 79001 0. 50570 3. 96543 0. 00795 1. 92690 0. 12323 2. 03730 0. 09400 1. 74625 0. 14634 2. 45737 0. 04225 1. 28232 0. 28922 EXPORT does not Granger Cause REER_CPI Lags: 2 REER_CPI does not Granger Cause EXPORT 54 EXPORT does not Granger Cause REER_CPI Lags: 3 REER_CPI does not Granger Cause EXPORT 53 EXPORT does not Granger Cause REER_CPI Lags: 4 REER_CPI does not Granger Cause EXPORT 52 EXPORT does not Granger Cause REER_CPI Lags: 5 REER_CPI does not Granger Cause EXPORT 51 EXPORT does not Granger Cause REER_CPI Lags: 6 REER_CPI does not Granger Cause EXPORT does not Granger Cause REER_CPI 50 ü The hypothesis that REER_CPI and REER_PPI do not Granger cause the volume of export are rejected while the hypothesis that EXPORT do not Granger cause REER_CPI and REER_PPI are not rejected

Pairwise Granger Causality Tests Sample: 1990: 1 2003: 4 Lags: 1 Null Hypothesis: REER_CPI does not Granger Cause IMPORT Obs F-Statistic Probability 55 6. 71508 0. 01238 3. 99534 0. 05087 6. 02671 0. 00457 1. 28449 0. 28595 3. 03152 0. 03866 0. 72388 0. 54292 Obs F-Statistic Probability 52 2. 33387 0. 07077 0. 37474 0. 82536 Obs F-Statistic Probability 51 0. 97039 0. 44750 1. 43388 0. 23320 IMPORT does not Granger Cause REER_CPI Lags: 2 REER_CPI does not Granger Cause IMPORT 54 IMPORT does not Granger Cause REER_CPI Lags: 3 REER_CPI does not Granger Cause IMPORT 53 IMPORT does not Granger Cause REER_CPI Lags: 4 Null Hypothesis: REER_CPI does not Granger Cause IMPORT does not Granger Cause REER_CPI Lags: 5 Null Hypothesis: REER_CPI does not Granger Cause IMPORT does not Granger Cause REER_CPI üThe hypothesis that REER_CPI and REER_PPI do not Granger cause the volume of Import are rejected while the hypothesis that IMPORT do not Granger cause REER_CPI and REER_PPI are not rejected

Responses of Export and Import to REER_CPI and REER_PPI impulses

Results of regression for the two types of REER Newey-West HAC Standard Errors & Covariance (lag truncation=3) Export= REER_CPI*2. 714627 -13. 4857 R-squared 0. 735833 D-W=0. 25 [7. 37] [-7. 15] Export= REER_PPI*3. 058773 -15. 33536 R-squared 0. 677775 D-W=0. 24 [6. 98] [-6. 83] Import =REER_CPI*2. 726184 -13. 44575 R-squared 0. 863549 D-W=0. 47 [10. 88] [-10. 57] Import =REER_PPI*3. 121839 -15. 55607 R-squared 0. 821542 D-W=0. 45 [10. 22] [-9. 97] n 1. 072. 714627 =1. 2016 ≈20. 16% and n 1. 072. 726184 =1. 2025 ≈ 20. 25 % n 0. 932. 714627 =0. 8211 ≈ 17% and n 0. 932. 726184 = 0. 82050≈ 18% 1. 043. 058773 =1. 1274 ≈12, 74 % respectively the volume of Export and 1. 043. 121839 =1. 13025≈13% the volume of Import 0. 963. 058773 =0. 8826 ≈11% respectively the volume of Export 0. 963. 121839 = 0. 8803≈ 12% the volume of Import back

Export and REER

Import and REER

Error correction equations: Estimation Method: Least Squares Sample: 1991: 3 2003: 4 Included observations: 50 Total system (balanced) observations 100 Equation: D(EXPORT) = C(1)*( EXPORT(-1) - 3. 505269075*REER_CPI(-1) + 17. 52164625 ) + C(2)*D(EXPORT(-1)) + C(3)*D(EXPORT(2))+ C(4)*D(EXPORT(-3)) + C(5)*D(EXPORT(-4)) + C(6)*D(EXPORT(-5)) + C(7)*D(REER_CPI(-1)) + C(8)*D(REER_CPI(-2)) + C(9)*D(REER_CPI(-3)) + C(10)*D(REER_CPI(-4)) + C(11) *D(REER_CPI(-5)) + C(12) Observations: 50 C(1)=-0. 026831 t-Statistic =-3. 37139 Prob =0. 0012 R-squared 0. 977714 Mean dependent var 0. 032526 Adjusted R-squared 0. 971262 S. D. dependent var 0. 052673 S. E. of regression 0. 008929 Sum squared resid 0. 003030 Durbin-Watson stat 2. 013430 Equation: D(EXPORT) = C(1)*( EXPORT(-1) - 4. 165968926*REER_PPI( -1) + 21. 01019822 ) + C(2)*D(EXPORT(-1)) + C(3)*D(EXPORT(-2)) + C(4)*D(EXPORT(-3)) + C(5)*D(EXPORT(-4)) + C(6)*D(EXPORT(-5)) + C(7)*D(REER_PPI(-1)) + C(8)*D(REER_PPI(-2)) + C(9) *D(REER_PPI(-3)) + C(10)*D(REER_PPI(-4)) + C(11) *D(REER_PPI(-5)) + C(12) Observations: 50 C(1)=-0. 028198 t-Statistic =-3. 767567 Prob =0. 0003 R-squared 0. 979022 Mean dependent var 0. 032526 Adjusted R-squared 0. 972949 S. D. dependent var 0. 052673 S. E. of regression 0. 008663 Sum squared resid 0. 002852 Durbin-Watson stat 2. 047369

Equation: D(IMPORT) = C(1)*( IMPORT(-1) - 1. 568281763*REER_CPI(-1) + 7. 625304795 ) + C(2)*D(IMPORT(-1)) + C(3)*D(IMPORT(-2)) + C(4)*D(IMPORT(-3)) + C(5)*D(IMPORT(-4)) + C(6)*D(IMPORT( -5)) + C(7)*D(REER_CPI(-1)) + C(8)*D(REER_CPI(-2)) + C(9) *D(REER_CPI(-3)) + C(10)*D(REER_CPI(-4)) + C(11) *D(REER_CPI(-5)) + C(12) Observations: 50 C(1)=-0. 026887 t-Statistic =-3. 289858 Prob =0. 0015 R-squared Adjusted R-squared S. E. of regression Durbin-Watson stat 0. 878063 0. 842766 0. 016188 1. 664252 Mean dependent var S. D. dependent var Sum squared resid 0. 035058 0. 040824 0. 009958 Equation: D(IMPORT) = C(1)*( IMPORT(-1) - 1. 300769017*REER_PPI(-1) + 6. 323013095 ) + C(2)*D(IMPORT(-1)) + C(3)*D(IMPORT(-2)) C(4)*D(IMPORT(-3)) + C(5)*D(IMPORT(-4)) + C(6)*D(IMPORT(-5)) + C(7)*D(REER_PPI(-1)) + C(8)*D(REER_PPI(-2)) + C(9) C(10)*D(REER_PPI(-4)) + C(11) *D(REER_PPI(-5)) + C(12) Observations: 50 C(1)=-0. 032755 t-Statistic =-3. 185857 Prob =0. 0021 R-squared Adjusted R-squared S. E. of regression Durbin-Watson stat 0. 876274 0. 840458 0. 016306 1. 675513 Mean dependent var S. D. dependent var Sum squared resid 0. 035058 0. 040824 0. 010104 + *D(REER_PPI(-3)) +

Results of regressions: EXPORT =REER_CPI *0. 565837+GDP_EU*0. 390866 -0. 971709 R-squared 0. 691239 , D-W=0. 54 [3. 57] [2. 71] [- 1. 26] EXPORT =REER_PPI *0. 441380+GDP_EU*0. 507131 -0. 887198 R-squared 0. 608194 , D-W=0. 38 [3. 16] [3. 26] [-1. 11] IMPORT=REER_CPI*-0. 095769+EXPORT*0. 802969+AGR_DEMAND*0. 048147+ 1. 078879 R-squared 0. 961766 , D-W=0. 28 [-1. 59] [16. 92] [1. 33] [3. 66] IMPORT=REER_PPI*-0. 007240+EXPORT*0. 793771+AGR_DEMAND*0. 023037+ 1. 078879 R-squared 0. 969995 , D-W=0. 25 [-0. 137] [15. 98] [0. 48] [2. 42]

REER influence on Trade Balance Romania has negative Trade Balance (TB) with EU countries VAR lag length criteria : 7 lags for both REER_CPI and REER_PPI relationship with TB

REER influence on Trade Balance n cointegration equation for 5% level of significance for the two cases TB and REER_CPI and TB and REER_PPI Lags interval (in first differences): 1 to 7 Unrestricted Cointegration Rank Test

Pairwise Granger Causality Tests: Sample: 1990: 1 2003: 4 Lags: 1 Null Hypothesis: REER_CPI does not Granger Cause TB TB does not Granger Cause REER_CPI Obs 55 F-Statistic 9. 52595 0. 02620 Probability 0. 00324 0. 87203 Lags: 2 Null Hypothesis: REER_CPI does not Granger Cause TB TB does not Granger Cause REER_CPI Obs 54 F-Statistic 2. 32283 0. 02812 Probability 0. 10869 0. 97229 Lags: 1 Null Hypothesis: REER_PPI does not Granger Cause TB TB does not Granger Cause REER_PPI Obs 55 F-Statistic 9. 19004 0. 01979 Probability 0. 00379 0. 88866 Lags: 2 Null Hypothesis: REER_PPI does not Granger Cause TB TB does not Granger Cause REER_PPI Obs 54 F-Statistic 2. 31398 0. 02818 Probability 0. 10958 0. 97223

Results of regressions for the two types of REER TB=REER_CPI*1. 65779 -8. 692956 R-squared 0. 441621 , D-W=0. 79 [3. 6841] [-3. 8573] TB=REER_PPI*1. 92424 -9. 293298 R-squared 0. 431312 , D-W=0. 78 [3. 6981] [-3. 8518] n n n 1. 071. 65 =1. 118 ≈11. 8 % and 1. 041. 92 =1. 078 ≈7. 8 % 0. 931. 65 =0. 887 ≈ 12 % and 0. 961. 92 =0. 92 ≈8 % TB does not have the expected sign and consequently it initially worsens at REER depreciations and then it improves (starting with lag 4 it has the expected negative sign)

Error Correction Model TB and REER_CPI (7 lags): D(TB) = 0. 1370008082*( TB(-1) + 0. 01767896916*REER_CPI_LOG(-1) ) + 0. 7865814588*D(TB(-1)) 0. 3968784339*D(TB(-2)) + 0. 03360259529*D(TB(-3)) - 0. 2447805494*D(TB(-4)) -0. 04381380141*D(TB(-5)) 0. 04652583436*D(TB(-6)) - 0. 1803369447*D(TB(-7)) +3. 425845879*D(REER_CPI (-1)) – 0. 8003956003*D(REER_CPI (-2)) +1. 207371803*D(REER_CPI (-3)) +1. 756795848*D(REER_CPI (-4)) 3. 157573105*D(REER_CPI (-5)) + 2. 403071583*D(REER_CPI (-6)) - 0. 01985208971*D(REER_CPI (-7)) D(REER_CPI) = - 0. 06954231854*( TB(-1) + 0. 01767896916*REER_CPI(-1) ) – 0. 07424238375*D(TB(-1)) + 0. 1036032247*D(TB(-2)) +0. 005701677302*D(TB(-3)) +0. 03426812401*D(TB(-4)) + 0. 01956357912*D(TB(-5)) + 0. 05240118994*D(TB(-6)) +0. 04620054128*D(TB(-7)) - 0. 5301569997*D(REER_CPI(-1)) + 0. 02040601877*D(REER_CPI(-2)) – 0. 4077126554*D(REER_CPI(-3)) + 0. 3907634519*D(REER_CPI(-4)) +0. 1055090966*D(REER_CPI(-5)) - 0. 5415890667*D(REER_CPI(-6)) +0. 03241797129*D(REER_CPI(-7)) TB and REER_PPI (7 lags): D(TB) = 0. 1453655075*( TB(-1) + 0. 2806808741*REER_PPI(-1) - 1. 039744678 ) + 0. 7616830328*D(TB(-1)) 0. 5830842106*D(TB(-2)) + 0. 1383830284*D(TB(-3)) - 0. 2598415963*D(TB(-4)) - 0. 006246235075*D(TB(-5)) 0. 08143625724*D(TB(-6)) - 0. 1648990411*D(TB(-7)) + 3. 078886096*D(REER_PPI(-1)) - 1. 223630924*D(REER_PPI(-2)) + 1. 706903517*D(REER_PPI(-3)) + 1. 682785129*D(REER_PPI(-4)) - 2. 927038695*D(REER_PPI(-5)) + 2. 73469155*D(REER_PPI(-6)) - 0. 8225060185*D(REER_PPI(-7)) - 0. 00491755967 D(REER_PPI) = - 0. 07763000086*( TB(-1) + 0. 2806808741*REER_PPI(-1) - 1. 039744678 ) - 0. 04419584655*D(TB(-1)) + 0. 1514702121*D(TB(-2)) - 0. 01653847494*D(TB(-3)) + 0. 03061622078*D(TB(-4)) + 0. 01204721762*D(TB(-5)) + 0. 05570758654*D(TB(-6)) + 0. 04820989632*D(TB(-7)) - 0. 4477123584*D(REER_PPI(-1)) + 0. 03722262183*D(REER_PPI(2)) - 0. 5756806286*D(REER_PPI(-3)) + 0. 2824755348*D(REER_PPI(-4)) - 0. 0615261533*D(REER_PPI(-5)) 0. 6419286947*D(REER_PPI(-6)) + 0. 161200314*D(REER_PPI(-7)) + 0. 01206296165

Dependent Variable: D(TB) Method: Least Squares Sample(adjusted): 1992: 1 2003: 4 Included observations: 48 after adjusting endpoints D(TB) = C(1)*( TB(-1) + 0. 01069773101*REER_CPI(-1) + 0. 2191682532 ) + C(2)*D(TB(-1)) + C(3)*D(TB(-2)) + C(4)*D(TB(-3)) + C(5)*D(TB( -4)) + C(6)*D(TB(-5)) + C(7)*D(TB(-6)) + C(8)*D(TB(-7)) + C(9) *D(REER_CPI(-1)) + C(10)*D(REER_CPI(-2)) + C(11)*D(REER_CPI(-3)) + C(12)*D(REER_CPI(-4)) + C(13) *D(REER_CPI(-5)) + C(14)*D(REER_CPI(-6)) + C(15) *D(REER_CPI(-7)) + C(16) C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) Coefficient Std. Error t-Statistic Prob. 0. 154977 0. 772608 -0. 430887 0. 042718 -0. 252806 -0. 052952 -0. 063642 -0. 191026 3. 421184 -0. 841963 1. 268492 1. 716229 -3. 198078 2. 399025 -0. 156624 -0. 016135 0. 087303 0. 225150 0. 253643 0. 140261 0. 076088 0. 085477 0. 080023 0. 071940 0. 646882 0. 769861 0. 664479 0. 449494 0. 704544 0. 873477 0. 839850 0. 026257 1. 775161 3. 431524 -1. 698796 0. 304563 -3. 322545 -0. 619482 -0. 795298 -2. 655359 5. 288729 -1. 093657 1. 909003 3. 818133 -4. 539218 2. 746522 -0. 186491 -0. 614526 0. 0854 0. 0017 0. 0991 0. 7627 0. 0022 0. 5400 0. 4323 0. 0122 0. 0000 0. 2823 0. 0653 0. 0006 0. 0001 0. 0098 0. 8532 0. 5432 0. 705076 0. 566830 0. 111433 0. 397356 46. 94995 Mean dependent var S. D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat R-squared Adjusted R-squared S. E. of regression Sum squared resid Log likelihood White Heteroskedasticity Test: Jarque-Bera normality Test: F-statistic Statistic 1. 788021 2. 391790 Probability 0. 022440 0. 169311 -1. 289581 -0. 665848 2. 087974 0. 116205 0. 302433

Dependent Variable: D(TB) Method: Least Squares Sample(adjusted): 1992: 1 2003: 4 Included observations: 48 after adjusting endpoints D(TB) = C(1)*( TB(-1) + 0. 3370609842*REER_PPI(-1) - 1. 444550447 ) + C(2)*D(TB(-1)) + C(3)*D(TB(-2)) + C(4)*D(TB(-3)) + C(5)*D(TB(-4)+ C(6)*D(TB(-5)) + C(7)*D(TB(-6)) + C(8)*D(TB(-7)) + C(9) *D(REER_PPI(-1)) + C(10)*D(REER_PPI(-2)) + C(11 ) *D(REER_PPI(-3)) + C(12)*D(REER_PPI(-4)) + C(13) *D(REER_PPI(-5)) + C(14)*D(REER_PPI(-6)) + C(15) *D(REER_PPI(-7)) + C(16) C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) Coefficient Std. Error t-Statistic Prob. 0. 146502 0. 645211 -0. 472682 0. 098739 -0. 265859 -0. 022232 -0. 076169 -0. 156350 2. 884267 -0. 906927 1. 541538 1. 775626 -2. 656261 2. 369020 -0. 509762 -0. 006860 0. 074730 0. 208309 0. 226722 0. 133076 0. 073777 0. 084235 0. 077014 0. 068061 0. 560499 0. 690607 0. 586483 0. 459297 0. 619329 0. 781475 0. 766876 0. 024101 1. 960429 3. 097381 -2. 084853 0. 741979 -3. 603558 -0. 263930 -0. 989016 -2. 297198 5. 145893 -1. 313232 2. 628445 3. 865960 -4. 288936 3. 031472 -0. 664725 -0. 284647 0. 0587 0. 0040 0. 0451 0. 4635 0. 0011 0. 7935 0. 3301 0. 0283 0. 0000 0. 1984 0. 0131 0. 0005 0. 0002 0. 0048 0. 5110 0. 7777 0. 704920 0. 566601 0. 111463 0. 397566 46. 93725 Mean dependent var S. D. dependent var Akaike info criterion Schwarz criterion Durbin-Watson stat R-squared Adjusted R-squared S. E. of regression Sum squared resid Log likelihood White Heteroskedasticity Test : Jarque-Bera normality Test: F-statistic Statistic 1. 687595 Probability 6. 482801 Probability 0. 022440 0. 169311 -1. 289052 -0. 665318 2. 215315 0. 141207 0. 039109

Conclusions n Results show that is possible to start building a quantitative background for discussion about REER in Romania during the accession process n REER is a useful summary indicator of essential economic information n REER can be a good indicator for monetary and exchange rate policies in order to forecast trade balance in a country (R-squared ≈ 70%) n Exports and Imports have the expected reaction to REER movements n Trade Balance initially worsens after a REER depreciation and then it improves n It is questionable whether permanent depreciation is desirable to improve trade balance

Romanian “ Trade Openness” to GDP ratio 86. 0% 84. 0% 82. 0% 80. 0% 78. 0% 76. 0% 74. 0% 72. 0% 70. 0% 68. 0% Weight in GDP mil USD 2001 2002 period 2003 Romanian Trade volumes export with EU export with Europe period back 20 02 20 00 19 98 19 96 19 94 Total export 19 92 19 90 14000 12000 10000 8000 6000 4000 2000 0 Source: Romanian External Trade Department