The impact of exchange rate on bilateral trade between Vietnam and China

304 | ICUEH2017 The impact of exchange rate on bilateral trade between Vietnam and China VO THANH THU University of Economics HCMC – vothanhthu@ueh.edu.vn TRAN QUOC KHANH CUONG Van Hien University – cuong.tqk@vnp.edu.vn Abstract The goal of this paper is to investigate the short term and long term relationship between exchange rate and trade balance on the trade balance between Vietnam and China Differing multivariate methods such as Johansen Cointegraion test, Vector Error Correction Model and monthly time series data spans from 2011:1 to 2015:12 have been employed The results prove the relationship among trade balance, real exchange rate, domestic and foreign output or ML condition hold The positive coefficient implies that the devaluation of Vietnamese currency will improve the bilateral trade between Vietnam and China Last but not least, the prominent aim of this paper is to prove the existence of Purchasing Power Parity between Vietnam and China hold before examining the effect of real exchange rate to trade balance Keywords: PPP; VECM; exchange rate; trade balance; ML condition Introduction Özkan (2013) states that real exchange rate plays an important role in the macroeconomic such as economic development, sustainable growth, especially in the trade balance The trade balance is a component of Gross Domestic Product (GDP) GDP which will increase if trade of balance is surplus, and decrease if trade of balance is deficit In recent years, Vietnam almost has a trade deficit with China The exchange rate between Vietnam and China has been considering as a reason for trade deficit Vietnam is in need to depreciate its currency because exchange rate plays a crucial role in the economy This problem is very important because exchange rate plays an important role in the economy, especially in emerging market such as Vietnam (Stone et al., 2009) It will effect on inflation targeting which is one of the urgent issues in Vietnam in recent years Vo Thanh Thu & Tran Quoc Khanh Cuong | 305 According to Marshall-Learner (ML) condition or J Curve, the exchange rate is vital in the trade balance Both of them state that if the currency is depreciation, with the condition that adding the absolute value of the elasticity of export and import more than one, the trade balance will increase in the long run The authors often use the formula to check the relationship of exchange rate and trade balance as follow: TB = f(RER, Y, Y*) where RER is real exchange rate and is calculated with the assumption Purchasing Power Parity between two countries is hold However, in the reality, many studies have proven PPP between two countries does not hold such as: Assaf (2006), Doğanlar (2006) This is the limitation of papers when the authors not prove PPP hold Therefore, in this paper, PPP is proved hold in the long term before calculating the RER The next part of this study include: section briefly discusses the literature review; section presents the model specification, section presents the results and discussion Literature review 2.1 The purchasing power parity approach (PPP) The PPP was first studied by the Salamanca School in Span in the 16th century At that time, PPP was basically means that when we changed to the common currency, the price level of every country should be the same (Rogoff, 1996) Cassel introduced the term purchasing power parity (PPP) in 1918 After that, PPP became the benchmark for a center bank to build up the exchange rate and for scholars to study exchange rate determinant The model of PPP of Cassel became the inspiration for Balassa and Samuelson to set up their models in 1964 They independently worked and gave the final explanation why absolute PPP became the good theory of exchange rate (Asea and Corden, 1994) The reason is that the relative price of each good in different countries should be equal to the same price when changing into the same currency The PPP has two versions, including absolute and relative PPP (Balassa, 1964) According to the first version, Krugman et al (2012), define that absolute PPP implies that the exchange rate of pair countries equal to the ratio of the price level of these countries This mean: st = pt/𝑝"∗ (1) 306 | ICUEH2017 Shapiro (1983) states that the relative PPP implies the ratio of domestic to foreign prices would equal the ratio change in the equilibrium exchange rate This states that there is a constant k which has the relationship between price level and the equilibrium exchange rate, st = k*pt/𝑝"∗ (2) In the empirical studies, checking the validity of PPP by unit root test was popular in 1980s based on Dickey and Fuller approach, nevertheless, this approach has low power (Ender and Granger, 1998) After Johansen (1988) develops a method of conducting VECM, which becomes the benchmark model for many authors test PPP approach There are some papers have proved PPP hold such as Yazgan (2003), Doğanlar et al.(2009), Kim (2011), Kim & Jei (2013), Jovita Gudan (2016), and some papers not such as Basher et al (2004), Doğanlar (2006) 2.2 Marshall – Lerner condition According to Bahmani-Oskooee (1991), Marshall – Lerner was that if the currency depreciated, the trade balance would improve in the long run with the condition that the sum of the absolute value of the elasticity of export and import is more than one Assuming that the condition is satisfied, the appreciation of home country makes imported goods from abroad cheaper, thus, the consumers of Home country buy more Foreign consumer finds imported goods are more expensive and they buy less than before appreciation As Consequence, Home country has trade deficit Nevertheless, if Home country is depreciated, an opposite effect would be taken place and Home country would enjoy trade surplus with the rest of the world Furthermore, the condition states that devaluation of currency has positive effects on the trade balance if the total of the absolute value of the elasticity of export and import is larger than one For that reason, the effects on trade balance rely upon the elasticity of price However, the disadvantage of the Marshall – Lerner condition is that it can not explain why the trade deficit occurs in the short run after currency depreciation The reason why trade is deficit, which happens in the short run, was explained by Akbostanci in 2004 Most exporters and importers have signed the contract before depreciation In the short Vo Thanh Thu & Tran Quoc Khanh Cuong | 307 run, the quantity of export and import does not change much; nevertheless, the depreciation makes the imported goods cost more in domestic currency Therefore, the value of imported goods rises while exported products not change a lot As a result, trade balance becomes deficit Seeking for the relationship between trade balance and exchange rate is one of the most interested topics in international trade There have been many authors such as Rose, Yellen, Bahmani-Oskooee, Arize, Shahbaz have conducted researches in many countries from developing to developed one The common techniques, which are applied, are Johnsen cointegration (1988) technique such as Bahmani-Oskooee (1991), Onafowora (2003) etc., and Autoregressive Distributed Lags (ARDL) such as Narayan (2006), Shahbaz et al., (2012) etc Matesanz and Fugarolas (2009) examined the relationship between trade balance and exchange rate in Argentina from 1962 to 2005 The results indicated that the most important thing is that the trade balance of Argentina was supported by Argentina currency undervaluation As a result, M-L condition took place in Argentina Last but not least, in their paper, they argued that the coefficient of real exchange rate can be positive or negative If the coefficient is positive, the M-L condition will hold, this means that when a currency is devaluated, trade balance will increase, or vice versa The conclusions for J curve or M-L Condition are very different While any researchers such as Bahmani-Oskooee (1991), Shirvani and Wilbratte (1997), Narayan (2006) claim that that there is only the long run relationship between trade balance and exchange rate, Rose and Yellen (1988) and Rose (1990), Rahman et al.,(1997) insist that there is no relationship between the two factors Some papers indicate that if real exchange rate is depreciation, it will lead to trade balance deterioration such as Arora et al., (2003) and Shahbaz et al., (2012) Model specification Model for purchasing power parity approach Take log from the equation (1) we have: Log(st) = log(pt) – log(p∗% ) (3) 308 | ICUEH2017 where: s is natural log exchange rate of VN, pt and p∗% is the natural log CPI of VN and CPI of China respectively If all variables mean reverting, the real exchange rate is calculated by the PPP approach following the equation: qt = st + pt* - pt (4) where all variables are log form, s is log exchange rate of VN, pt and p∗% is the log CPI of VN and CPI of China respectively Model for trade balance Determining the relationship of exchange rate, GDP Vietnam and GDP China with trade balance follow the equation: Ln(TB) = α0 + α1ln(Q) + α2ln(Yt) + α3ln(Y%∗ ) + εt (5) where TB is trade balance and is defined as the ratio of export to import, Q is the real exchange rate defined as the relative price of domestic to foreign goods or RER = P/ (SP*) whe5re: S is nominal exchange rate, P is domestic price index and P* is foreign price index Yt and Y%∗ are Vietnam real income and China real income respectively εt is white noise process The expected signs of the coefficient of real exchange rate can be positive or negative α1 is positive if the real exchange rate of VND depreciation encourages exporting goods and reduces imported goods, then increase trade balance, or Marshall Lerner condition holds If α1 is negative, the M-L does not hold or depreciate of VND which makes trade balance worsen The expected sign of the coefficient of Vietnam real income will be negative When Vietnam income increases, Vietnamese imports more goods and services from China, therefore, trade balance will decrease or α2 < Similarly, the expected sign of the coefficient of China real income will be positive When China income increases, Chinese imports more goods and services from Vietnam, therefore, trade balance of Vietnam will increase or α3 > Because of the limitation of data, the index of industrial production (IIP) will be used to replace for real income as a proxy Vo Thanh Thu & Tran Quoc Khanh Cuong | 309 Doing VEC Model follows the steps: Step 1: Testing unit root test based on ADF test VECM requires variables are stationary and therefore all-time series express no spurious regression Augmented Dickey-Fuller test whether each variable stationary or not Step 2: Using Johansen (1988) test to test the cointegration of these variables Using Johansen (1988) technique with maximum likelihood to test cointegration The precondition before doing this test is determine lag order p Optimal lag must be chosen before conducting Johansen (1988) procedure In view package, there are five lags length criteria which have the same power Therefore, if there are more than lag criteria, every lag is used for every case in VECM Step 3: Running VECM VECM can be written as: ∆𝑙𝑛𝑇𝐵, = 𝛿/ + 68, µ6 ∆ ln 𝑄"56 𝜀, 68, 𝜃2 ∆ ln 𝑇𝐵"56 + 68, 𝛾2 ∆ ln 𝑌"56 + 𝜆 ln 𝑇𝐵"5, − 𝛼/ − 𝛼, ln 𝑄"5, − ∗ 68, 𝜑2 ∆ ln 𝑌"56 + ∗ 𝛼A ln 𝑌"5, − 𝛽C ln 𝑌"5, + + (6) where ∆ is the first difference operator, 𝜆 is the speed of adjustment coefficient of long run, TB is trade balance, Q is the real exchange rate, Y is IIP of Vietnam, Y* is IIP of China and 𝜀, is white noise The estimating coingtegration relationship of the function reflects the long run relationship of parameters And the equation yields a short run and long run relationship among variables (Matesanz and Fugarolas, 2009) 𝜆 is the speed of adjustment coefficient of long run If 𝜆 is negative in sign and signification, there is long run causality running from dependent variables to the independent variable In other words, the real exchange rate, Vietnam income and China income impact on bilateral trade between Vietnam and China Wald test is used to test whether short run causality between exchange rate and trade balance takes place 310 | ICUEH2017 3.1 Data for PPP approach Nominal exchange rate (is defined as the number of units of domestic currency per unit of foreign currency), consumer price index of Vietnam and consumer price index of China are in terms of logarithm form and data spans from 2011:1 to 2015:12 Nominal exchange rate was collected from Thomson Reuter, Consumer Price Index of Vietnam, base year = 2010, was collected from IFS and Consumer price index of China, base year = 201, was collected from OECD 3.2 Data for trade balance This procedure needs real exchange rate, trade balance between Vietnam and China, industrial production index (as a proxy of GDP) of Vietnam and China All variables are natural logarithm Data was collected from IFS (export, import) and ERIC (IIPVN, IIPCN, base year: 2010 and are calculated by the authors) indicator system spans from 2011:1 to 2015:12 Results and discussion 4.1 PPP approach Unit root test The Augmented Dickey Fuller test is used to check the stationary of consumer price index of Vietnam (CPIVN), consumer price index of China (CPICN) and nominal exchange rate (S) between Vietnam and China All variables have log form Mackinon (1996) which is available in Eview package software is used as the critical value Table Unit root test for PPP approach ADF Variables At level At first different t – statistic p-value t – statistic p-value CPICN 0.411854 0.9819 -6.646516 0** CPIVN 3.090598 -2.943013 0.0466* S 0.293415 0.9760 -6.902586 0** Note: *, ** indicate significant at 5% and 1% levels respectively Vo Thanh Thu & Tran Quoc Khanh Cuong | 311 Table reports the results of unit root test for time series of consumer price index of Vietnam (CPIVN), consumer price index of China (CPICN) and nominal exchange rate (S) between Vietnam and China As the table suggests, all variables have t-statistic greater than the critical value at level As the consequence, all variables have unit root or nonstationary at level On the contrary, at the first difference of S and CPICN have the t-statistic smaller than critical value at 1% Therefore, they not have unit root or stationary at the first difference Similarly, CPIVN has t-statistic smaller than the critical value at 5% so that it is stationary at the first difference As being analyzed above, all variables are nonstationary at level and stationary at first difference, therefore they cointegrated at I(1) or same order As a result, Johansen (1988) procedure is examined to investigate the cointegration among these time series Optimal lags for VECM Table Lag criteria for PPP approach Criterion LR Lag FPE AIC SC HQ 1 1 LR: sequential modified LR test statistic FPE: Final prediction error AIC: Akaike information criterion SC: Schwarz information criterion HQ: Hannan-Quinn information criterion Table illustrates criterion for choosing lag In other words, 1-lag was chosen for conducting Johansen (1988) procedure or testing cointegration of three variables Johansen (1988) procedure for cointegration test For the reason all variables are cointegrated at the first order I(1), Johansen (1988) cointegration with lag is conducted to test the long run relationship among variables 312 | ICUEH2017 Table Johansen cointegration test with lags Cointegration equation Number of Ces Trace test Eigenvalue test Statistic Critical value (5%) Statistic Critical value (5%) 29.81360 29.79707 16.95114 21.13162 At most 12.86246 15.49471 9.705488 14.2646 At most 3.156967 3.841466 3.156967 3.841466 None * Table presents the Johansen (1988) cointegration test The results indicate Trace test statistically significant at 5% because the statistic is greater than the critical value of 5% As consequence, the null hypothesis of r = is rejected R = implies one cointegration equation in the long run That is the reason why VECM can be used to find out the long run relationship of three variables Table The speed of adjustment coefficient of long run λ Coefficient Std Error t-Statistic Prob -0.479592 0.116593 -4.113400 0.0001 Table 𝜆 equals -0.479592and probability equals 0.001 Error Correction Term is negative and significant in sign For this reason, variables take place relationship in the long run or all variables come back to mean As a result, real exchange rate is calculated followed this formula: q = s + p* - p with all variables have logarithm 4.2 The impact of the exchange rate on trade balance Unit root test Similarity of PPP approach, the Augmented Dickey Fuller test is used to check the stationary of real exchange rate (Q), industrial production index of China (IIP CN), industrial production index of Vietnam (IIP VN) and trade balance (TB) All variables have the natural log form Mackinon (1996) which is available in Eview package software, is used as the critical value as well Vo Thanh Thu & Tran Quoc Khanh Cuong | 313 Table Unit root test on time series ADF At level Variables At first different t – statistic p-value t - statistic p-value TB -1.657948 0.7561 -10.25748 0.0000 ** Q -2.556931 0.3010 -7.147927 0.0000 ** IIP VN -0.255281 0.9242 -6.986130 0.0000 ** IIP CN -1.986125 0.2918 -12.51762 0.0000 ** Note: ** indicate significant at 1% levels TB and Q test unit root with “Exogenous: Constant, Linear Trend” The table reports the results of unit root test for time series of trade balance (TB), real exchange rate between Vietnam and China, industrial production index of Vietnam (IIP VN) and industrial production index of China (IIPCN) The table illustrates that all variables have a t-statistic greater than the critical value at level As the consequence, all variables have unit root or nonstationary at level On the contrary, at the first difference, all variables have the t-statistic smaller than the critical value of 1% Therefore, all variables have not unit root or stationary at the first difference To sum up, all variables nonstationary at level and stationary at the first difference; therefore, they cointegrated at I(1) This is the reason why Johansen (1988) procedure can be used to examine the cointegration among these time series 4.3 Optimal lags for VECM Table Lag selection of VECM Criteria SC FPE LR AIC HQ Lags 8 Table suggests that five criteria choose two lags differently, and Therefore, in this paper, all of lags were used for conducting VECM Choosing the best model is based on the diagnostic check after running VECM 314 | ICUEH2017 Table Diagnostic check for every lag Lags Serial Correlation Heteroskedasticity Normality Test Cusum Cusum of Square + + + + + - + + + - +: good -: not good Table presents the diagnostic check and normality test of lags after conducting VECM The most important of diagnostic check is serial correlation As table illustrates, there is only lag satisfied serial correlation and other tests as well That is the reason why conducting VECM with lag is the best model Johansen procedure for cointegration test for the impact of exchange rate on trade balance Because all variables are cointegrated at the first order I(1), Johansen (1988) cointegration with lag can be used to seek for the long run relationship among variables Table Johansen cointegration test with lags Cointegration equation Trace test Number of Ces Eigenvalue test Statistic Critical value (5%) Statistic Critical value (5%) None 68.77004 47.85613 30.02283 27.58434 At most 38.74721 29.79707 19.10663 21.13162 At most 19.64058 15.49471 13.84609 14.2646 At most 5.794494 3.841466 5.794494 3.841466 Table suggests the Johansen (1988) cointegration test The results indicate both Trace test and Eigenvalue test statistically significant at 5% because the statistic is greater than the critical value of 5% As a result, the null hypothesis of r = and r = is rejected This implies at least cointegration equations in the long run or all variables have long run relationship This model concentrates on factors which effect on trade balance, for Vo Thanh Thu & Tran Quoc Khanh Cuong | 315 that reason first equation is focused on investigating the role of the exchange rate in the trade balance: Table VECM test with lags D(TB) = C(1)*( TB(-1) + 0.628543515991*RER(-1) - 1.40013175523*IIPVN(-1) + 3.74657529993*IIPCN(-1) - 5.24314357486 ) + C(2)*D(TB(-1)) + C(3)*D(RER(-1)) + C(4)*D(IIPVN(-1)) + C(5)*D(IIPCN(-1)) + C(6) Coefficient Std Error t-Statistic Prob C(1) -0.220293 0.082286 -2.677177 0.0099 C(2) -0.102943 0.128296 -0.802388 0.4260 C(3) 1.353124 1.338220 1.011137 0.3166 C(4) -0.891131 0.233924 -3.809485 0.0004 C(5) 1.546537 0.334740 4.620118 0.0000 C(6) -0.005079 0.017578 -0.288939 0.7738 Table suggests the normalized cointegration coefficients TB = - 5.24 + 0.63*RER - 1.4*IIPVN + 3.75*IIPCN The result indicates Q and IIPCN have positive sign, IIPVN has the negative sign These signs imply that: The coefficient of real exchange rate is positive, for that reason, trade balance increases 0.63 % when the depreciation of Vietnam’s currency is 1% The coefficient of IIPVN is negative, thus trade balance would decrease 1.4% when GDP Vietnam increase 1% Finally, when GDP of China increases 1%, Vietnam will take trade surplus 3.75% The impact of exchange rate on trade balance in long run Table 10 The speed of adjustment coefficient of long run λ Coefficient Std Error t-Statistic Prob -0.220293 0.082286 -2.677177 0.0099 Table 10 suggests the impact of exchange rate on trade balance in the long run The result indicates 𝜆 equals -0.22 and probability equals 0.001 As a result, the relationship between exchange rate and trade balance occurs in the long run, as well as IIPVN and IIP China 316 | ICUEH2017 The impact of independent variales in short run Using Wald test to test the short run relationship between the real exchange rate and trade balance: Null hypothesis: µj = 0, (j = 1,p) Alternative hypothesis: At least one µj ≠ Table 11 Wald test for the short run relationship Variables Coefficient Value Probability RER 1.353124 1.022398 0.3120 IIPVN -0.891131 14.51217 0.0001 IIPCN 1.546537 21.34549 0.0000 Table 11 presents the probability of chi square for three independent variables The probability of IIPVN and IIPCN are smaller than 5% Therefore, the null hypothesis can be rejected, which implies the relationship among the IIPVN, IIPCN and trade balance occur in the short run The coefficients state that when IIPVN grows 1%, TB declines 0.89% while IIPCN grows 1%, TB increases 1.55% On the contrast, the probability of Q is 0.31 and larger than 5% Therefore, the null hypothesis cannot be rejected, which implies the relationship between the real exchange rate and trade balance does not occur in the short run In short, the robustness check of VECM with lag is the best model The results indicate PPP hold between Vietnam and China, based on this conclusion; Q is calculated through PPP approach There is the existence of the long run relationship between Vietnam and China in the trade balance and the relationship does not happen in the short run Hence, ML condition holds and J curve does not exist between two countries Besides that, IIPVN and IIPCN affect on trade balance in the short term and long term as well Finally, because ML condition holds between two countries, Vietnam should devaluate its currency to cure the trade deficit Vo Thanh Thu & Tran Quoc Khanh Cuong | 317 References Akbostanci, E (2004) Dynamics of the Trade Balance The Turkish J-Curve Emerging Market Finance and Trade, 40(5), 57 – 73 Retrieved from: http://mesharpe.metapress.com/app/home/contribution.asp? referrer=parent&backto=issue,4,5;journal,74,84;linkingpublicationresults,1:111024,1 Arize, C A (1994) Cointegration Test of a Long-Run Relation between the Real Effective Exchange Rate and the Trade Balance International Economic Journal, 8(3), 1-9 doi: 10.1080/10168739400000001 Arora, S., Bahmani-Oskooee, M & Goswami, G (2003) Bilateral J-curve between India and her trading partners Applied Economics, 35(9), 1037-1041 doi:10.1080/0003684032000102172 Asea, P K & Corden, W M (1994) The Balassa-Samuelson model: an overview Review of international economics, 2(3), 191 – 200 doi: 10.1111/j.1467-9396.1994.tb00040.x Assaf, A (2006) Nonstationarity in real exchange rates using unit root tests with a level shift at unknown time International Review of Economics and Finance, 17(2), 269–278 doi:10.1016/j.iref.2006.08.008 Bahmani-Oskooee, M (1991) Is there a long run relation between trade balance and real effective exchange rate of LDCs? Economics Letters, 36(4), 403 – 407 doi:10.1016/0165-1765(91)90206-Z Balassa, B (1964) The Purchasing-Power Parity doctrine: a reappraisal Journal of Political Economy, 72(6), 584 – 596 Retrieved from: http://www.jstor.org/stable/1829464 Basher, S A & M Mohsin., 2004 PPP Tests in Cointegrated Panels: Evidence from Asian Developing Countries Applied Economics Letters, 11 (3), pp.163-166 Doğanlar, M (2006) Long-run validity of purchasing power parity and cointegration analysis for Central Asian countries Applied Economics Letters, 13, 457 – 461 Doğanlar, M., H Bal & M Özmen (2009) Testing Long-Run Validity of Purchasing Power Parity for Selected Emerging Market Economies Applied Economics Letters, 16 (14), 1443-1448 Enders, W., Granger, C.W.J (1998) Unit Root Tests and Asymmetric Adjustmentwith an Example Using the Term Structure of Interest Rates Journal of Business and Economics Statistics, 16, 304–311 Gomes, F A R and L S Paz (2005) Can real exchange rate devaluation improve trade balance? The 19901998 Brazilian case Applied Economics Letters, 12, 525-528 Johansen, S (1988) Statistical Analysis of Cointegrated Vectors Journal of Economic Dynamics and Control, 12(2-3), 231 -254 doi:10.1016/0165-1889(88)90041-3 Jovita G., 2016 Modelling and forecasting exchange rate Lithuanian Journal of Statistics, 55(1), pp.19 – 30 Kim, H-G., 2011 VECM Estimations of the PPP Reversion Rate Revisited Journal of Macroeconomics doi: 10.1016/j.jmacro.2011.10.004 Kim, H-G & S Y Jei, 2013 Empirical Test for Purchasing Power Parity Using a Time-Varying Parameter Model: Japan and Korea Cases Applied Economics Letters, 20 (6), pp.525-529 Krugman, R P., & Obstfeld, M., and Melitz, J M (2012) Price levels and the exchange rate in the long run In S Yagan (Ed.), International Economics theory and policy, (pp 385 – 386) Pearson Education 318 | ICUEH2017 Matesanz, D & Fugarolas, G (2009) Exchange rate policy and trade balance: a cointegration analysis of the Argentine experience since 1962 Applied Economics, 41(20), 2571-2582 doi: 10.1080/00036840701222660 Narayan, K P (2006) Examining the relationship between trade balance and exchange rate: the case of China's trade with the USA Applied Economics Letters 13(8), 507-510 doi: 10.1080/13504850500400488 Onafowara, O (2003) Exchange rate and trade balance in East Asia: is there a J curve? Economics Bulletin, 5(18), – 13 Retrieved from: http://www.accessecon.com/pubs/EB/2003/Volume5/EB03E00003A.pdf Özkan, F (2013) Comparing the forecasting performance of neural network and purchasing power parity: The case of Turkey Economic Modelling, 31, 752–758 doi:10.1016/j.econmod.2013.01.010 Rogoff, K (1996) The Purchasing Parity Puzzle Journal of Economic Literature, 34, 647 – 668 Retrieved from: http://scholar.harvard.edu/rogoff/publications/purchasing-power-parity-puzzle Rose, K A (1990) The role of exchange rates in a popular model of international trade does the ‘MarshallLerner’ condition hold? Journal of International Economics, 30(34), 301 – 316 doi:10.1016/00221996(91)90024-Z Rose, K A, & Yellen, L J (1998) Is there a J curve? Journal of Monetary Economics, 24(1), 53 – 68 doi: 10.1016/0304-3932(89)90016-0 Stone, M (2009) The role of exchange rate in inflation-targeting emerging economies IMF Occasional Paper, 267 Retrieved from: http://www.imfbookstore.org/pdf/s267eaexcerpt.pdf Shahbaz, M., Jalil, A & Islam, F (2012) Real Exchange Rate Changes and the Trade Balance: The Evidence from Pakistan The International Trade Journal, 26(2), 139-153 doi: 10.1080/08853908.2012.657588 Shirvani, H & Wilbratte, B (1997) The Relationship between the Real Exchange Rate and the Trade Balance: An Empirical Reassessment International Economic Journal 11(1), 39-50 doi:10.1080/10168739700000003 Yazgan, E (2003) The Purchasing Power Parity Hypothesis for a High Inflation Country: A Re-Examination of the Case of Turkey, Applied Economics Letters, 10 (3), 143-147