CAN OVERVALUATION PRELUDE TO CRISIS AND HARM GROWTH IN TURKEY? Murat Alper* and **İrfan Civcir *Central Bank of Turkey **Ankara University Abstract: This paper estimates the equilibrium real exchange rate of the Turkish lira to evaluate whether or not the overvaluation results in currency crisis and low growth rate over the period from the first quarter of 1987 to the fourth quarter of 2009. We follow the behavioural equilibrium exchange rate model proposed by Faruqee (1995) and Alberola et al. (1999), where the equilibrium real exchange rate depends on both the balance of payments approach and the BalassaSamuelson hypothesis. The results indicate that although large and persistent overvaluation contributes to a financial crisis in Turkey, a relatively small overvaluation, contrary to both the Washington Consensus view and the Rodrik view, promotes the growth of Turkish economy. Keyword: Equilibrium real exchange rate, behavioral equilibrium exchange rate, misalignment, Turkey I. Introduction In the last few years, the real exchange rate misalignment, which is defined as the difference between the actual real exchange rate and equilibrium real exchange rate, has gained great prominence in economic policy discussions because overvaluation is argued to have been the cause of the currency crises and the lower growth rate in the developing countries. Indeed, it is believed that the recent experience with economic crisis in Turkey, Mexico, East Asia, Brazil and Argentina is the result of the persistent overvaluation of the real exchange rate. In addition, the consistent avoidance of overvaluation has seen as a distinguishing feature of East and Southeast Asia countries’ success with sustainable growth in recent years. However, there are at least two difficulties in estimating real exchange rate misalignment since equilibrium real exchange rate is not only an unobservable variable, but also a dynamic indicator and moves over time as its fundamentals change. In the literature, various approaches dealing with the equilibrium exchange rate can be collected under two main groups, the traditional and modern equilibrium exchange rate models1. Traditional models can be listed as purchasing power parity, uncovered interest rate parity and monetary models. However, these models are found to be lacking in their explanations. The lack of explanatory power of the traditional models has facilitated the emergence of the modern equilibrium exchange rate models performing more sophisticated approaches in estimating the equilibrium exchange rate as a function of the fundamentals, 1 MacDonald (2000) and Driver and Westaway (2004) review the alternative approaches to modelling the equilibrium exchange rate. 1 such as FEER (Fundamental Equilibrium Exchange Rate) and BEER (Behavioral Equilibrium Exchange Rate). In the FEER approach advocated by Williamson (1983, 1994), the equilibrium real exchange rate is defined as the real exchange rate which is consistent with simultaneous internal and external balances. Internal balance is defined as achieving the level of output consistent with both full employment and low inflation. External balance is identified with the current account balance being at not only sustainable but also appropriate level when the economy is in internal balance. The macroeconomic equilibrium principles of the FEER approach are also strongly similar to the equilibrium concepts of the DEER (Desired Equilibrium Exchange Rate) approach suggested by Bayoumi et al. (1994) and Artis and Taylor (1995), the NATREX (Natural Real Exchange Rate) approach put forward by Stein (1990, 1994), the DARER (Debt Adjusted Real Exchange Rate) approach introduced by Fabella (1996) and Frait and Komárek (2002, 2008), and the FRER (Fundamental Real Exchange Rate) approach proposed by Bulíř and Šmídková (2005). The BEER approach as a second part of the family of the modern equilibrium exchange rate models was popularized by Clark and MacDonald (1998). It can be considered as a predominantly empirical approach estimating the equilibrium real exchange rate based on the econometric long-run relationship between the real exchange rate and its fundamentals. In this approach, the process of econonometric estimation is composed of two phases: At first phase, once a long-run relationship amongst the variables is identified, the equilibrium real exchange rate is estimated by substituting the actual or long-run values of the explanatory variables into that relationship. For the second phase, short-run and long-run misalignments of the real exchange rate are computed by the difference between the actual and fitted values of the real exchange rate. Short-run misalignment is derived on the basis of the actual values of the fundamentals rather than their long-run values as in the case of longrun misalignment. Employing the same econometric procedure but different theoretical frameworks, some alternative variations of BEER approach have recently been developed by Edwards (1989, 1994), Elbadawi (1994), Faruqee (1995), Alberola et al. (1999, 2002, 2004), Baffes et al. (1999), Montiel (1999), Wadhwani (1999), Alberola and López (2001), Alberola (2003), Rubaszek (2004), and Alberola and Naiva (2007). We considers the variant of BEER approach proposed by Faruqee (1995) and Alberola et al. (1999) using the time series data of Turkey to estimate the equilibrium real exchange rate and the associated of the short-run and the long-run real exchange misalignments from the first quarter of 1987 to the fourth quarter of 2009. Therefore, this paper differs from the earlier researches estimating the equilibrium exchange rate of the 2 Turkish lira in several aspects: Firstly, we rely on the BEER approach of Alberola et al. (1999) rather than other approaches. Secondly, we use a new data set including the recent quarterly data. Thirdly, we compute both short-run and long-run misalignments of the real exchange rate. Finally, we evaluate whether overvaluation contributes to the crisis and the lower growth rate. The remainder of the paper is organized as follows. Section II sets out the theoretical framework, Section III discusses the data set and the empirical results, Section IV presents the estimation results of real equilibrium exchange rates and analyzes the impact of long run misalignment on the financial crisis and the growth, and Section V bears the conclusions. II. Theoretical Framework The theoretical framework used in this study follows that advanced by Faruqee (1995) and extended by Alberola et al. (1999) and hinges on the conventional open economy macro model with a sectoral extention. The real exchange rate, defined as the relative price of domestic goods in terms of foreign goods, is given by qt pt et pt* (1) where qt is the logarithm of the real exchange rate, et is the logarithm of the nominal exchange rate, which represents the foreign price of domestic currency, and pt is the aggregate price levels, corresponding foreign price levels are denoted by an asterisk. Thus, an increase in qt denotes a real appreciation of the domestic currency. We assume that the aggregate price levels can be decomposed into the prices of traded and nontraded goods both at home and in the foreign country with weights α and 1 α , respectively 0 < α < 1 . For simplicity, let the weights be the same in both countries, such that pt αpT ,t 1 α pN ,t (2) pt* αpT* ,t 1 α p*N ,t where pT ,t and pN ,t are the logarithms of traded and nontraded goods price levels, respectively. By substituting (2) into (1), a general expression for real exchange rate with presence of traded and nontraded goods can be obtained as qt pT ,t et pT* ,t 1 α pN ,t pT ,t pN* ,t pT* ,t (3) Equation (3) states that the real exchange rate is expressed as the sum of two components: The first component, the relative price of traded goods between countries, is referred as external real exchange rate qT ,t , and the second one, proportional to the ratio 3 of the domestic to foreign relative price of nontraded goods, represents internal real exchange rate qN ,t . Thus we get from (3) qt qT ,t qN ,t (4) qT ,t pT ,t et pT* ,t (5) qN ,t 1 α pN ,t pT ,t α* pN* ,t pT* ,t (6) External real exchange rate is derived from the balance of payments approach in which current account CAt can be defined as the sum of the trade balance Bt and the net payments on net foreign assets Ft : CAt Bt rt .Ft (7)) where rt is the real international interest rate. A country with negative net foreign assets Ft 0 is net debtor while Ft 0 identifies the country as a net creditor. Given this expression, the current account in terms of gross domestic product (GDP) can be thought of as being determined in following way: cat bt rt gt ft (8) where cat , bt and f t stand for the relatives of respectively the current account, the trade balance and net foreign assets to GDP, and gt is the growth rate of domestic real GDP. Suppose now that the external real exchange rate has an adverse influence on the trade balance, and then current account can be rewritten as cat δqT ,t rt gt ft (9) where δ denotes the elasticity of trade balance with respect to the real exchange rate δ > 0 . When current account is zero in equilibrium 2 , the external real exchange rate is expressed as a function of net foreign assets: qT ,t rg . ft δ (10) where real international interest rate and the growth rate of domestic real GDP are assumed, for simplicity, to be constant. Equation (10) states that the effect of net foreign assets on the external real exchange rate is ambiguous: The sign of the difference between the real 2 Current account can also be defined as cat f ft where f is the desired level of net foreign assets and is the adjusment speed of the net foreign assets. When net foreign assets is below (above) its desired level, current account is in surplus (deficit). In equilibrium, ft f , and then cat 0 . 4 international interet rate and the growth rate of domestic real GDP is usually assumed to be positive. However, developing countries can grow above the real international interest rate due to using advanced technologies. The internal exchange rate is derived from the Balassa-Samuelson, or productivity bias hypothesis, which is based on a two country model with one factor assumed where labor is internationally immobile but perfectly mobile across sectors within the economy. Thus, nominal wages in both traded and nontraded sectors are equalized, and then the real wages paid by profit maximizing firms are adjusted for productivity (Strauss, 1995, 1996 and 1999): pT ,t wt aT ,t , pN ,t wt aN ,t , pT* ,t wt* aT* ,t , p*N ,t wt* aN* ,t (11) where wt denotes the logarithm of the nominal wage, and aT ,t and aN ,t are the indices of productivity in traded and nontraded sectors in logaritmic form. By substituting (11) into (6), an expression for the internal real exchange rate in terms of productivity differentials is obtained as qN ,t 1 α aT ,t aN ,t aT* ,t aN* ,t (12) Equation (12) implies that there is a connection between the internal real exchange rate and the relative sectoral productivity differential. According to Balassa-Samuelson hypothesis, a country with relatively higher productivity in the traded sector compared to nontraded sector has a lower price of traded to nontraded goods and experiences an internal real exchange rate appreciation. By combining (10) and (12) with (4), the real exchange rate can be expressed as a combination of the balance of payments approach and the BalassaSamuelson hypothesis in such a form: qt λf t θat (13) where the sign of λ is not clear-cut and θ is expected to be positive λ r g δ , θ 1 α and at aT ,t aN ,t aT* ,t aN* ,t . Thus, three well known special cases are nested within (13): In the balance of payments approach, θ is constrained to be zero because the Balassa-Samuelson effect is not considered as a determinant of the real exchange rate. If we assume that λ 0 , the equation (13) reduces further to Balassa-Samuelson approach. Purchasing power parity, which assumes that the real exchange rate is stationary, imposes the additional constraints λ θ 0 . This equation will motivate the empirical work. This variant of BEER approach is applied to industrialized countries by Alberola et al. (1999, 2002), and Alberola and López (2001), to transition economies from Central and 5 Eastern Europe by Rahn (2003), Babetskii and Égert (2005), and Alberola and Naiva (2007), to Latin American countries by Alberola (2003) and Alberola et al. (2004), to G20 countries except Russia and Saudi Arabia by Bénassy-Quéré et al. (2004, 2006) and to 35 countries including 15 industrialized OECD countries, 8 emerging countries of Asia and the Americas, 11 transition economies in Central and Eastern Europe and Cyprus by Égert et al. (2004). Table 1 shows that although the productivity improvements always causes the real exchange to appreciate, it is possible that an increase in the net foreign asset leads contrary to expectations to a depreciation of the real exchange rate. In fact, increasing net foreign assets are found to give rise to a depreciation of the real exchange rate in the transition economies from Central and Eastern Europe by Égert et al. (2004), in Canada, Japan, South Africa and the UK by Bénassy-Quéré et al. (2006), and in the Czech Republic by Alberola and Naiva (2007). Table 1 Signs of the Estimated Coefficients Authors Countries Time Periods Alberola et al. (1999) CA, DK, EU, FR, DE, GR, IT, JP, ES, SE, GB, US Alberola and López (2001) Fundamentals f a 1980-1998 + + ES 1975-1998 + + Alberola et al. (2002) CA, EU, JP, GB, US 1980-1999 + + Alberola (2003) AR, BR, CH, CO, MX, PE, VE 1960-2001 + + Rahn (2003) CZ, HU, EE, PL, SI 1990-2002 + + Alberola et al. (2004) AR 1960-2001 + + Bénassy-Quéré et al. (2004) AR, AU, BR, CA, CN, EU, GB, ID, IN, JP, KR, MX, TR, US, ZA 1980-2001 + + Égert et al. (2004) AT, AU, BE, BG, BR, CA, CL, CY, CZ, DK, EE, ES, FI, GR, HR, HU, ID, IE, KR, LT, LV, MX, MY, NL, NZ, PL, PT, RO, SE, SG, SI, SK, TH, TR, ZA 1970-2002 +/− + Babetskii and Égert (2005) CZ 1993-2004 + + Bénassy-Quéré et al. (2006) AR, AU, BR, CA, CN, EU, GB, ID, IN, JP, KR, MX, TR, US, ZA 1980-2004 +/− + Alberola and Naiva (2007) CZ, HU, PL 1993-2004 +/− + Notes: (1) f: net foreign assets, a: productivity (2) AR: Argentina, AT: Austria, AU: Australia, BE: Belgium, BG: Bulgaria, BR: Brazil, CA: Canada, CL: Chile, CN: China, CO: Colombia, CY: Cyprus, CZ: Czech Republic, DE: Germany, DK: Denmark, EE: Estonia, ES: Spain, EU: Eurozone, FI: Finland, FR: France, 6 GB: United Kingdom, GR: Greece, HR: Crotia, HU: Hungary, ID: Indonesia, IE: Ireland, IN: India, IT: Italy, JP: Japan, KR: Korea, LT: Lithuania, LV: Latvia, MX: Mexico, MY: Malaysia, NL: Netherlands, NZ: New Zealand, PE: Peru, PL: Poland, PT: Portugal, RO: Romania, SE: Sweden, SG: Singapore, SI: Slovenia, SK: Slovakia, TH: Thailand, TR: Turkey, US: United States, VE: Venezuela, ZA: South Africa (3) +(−) means that an increase in the given variables brings about an appreciation (depreciation) of the real exchange rate. Relatively few studies are also available that estimate the BEER of the Turkish lira compared to the currencies of the other emerging economies. In addition, almost all studies (Alper and Sağlam, 2000; Achy, 2001; Doroodian et al., 2002; Atasoy and Saxena, 2006; Dağdeviren et al., 2009) rely on the BEER approach put forward by Edwards (1989, 1994) instead of that introduced by Alberola et al. (1999). Table 2 provides an overview of these studies on the time periods, and the long-run and short-run fundamentals. Alper and Sağlam (2000) are among the first to apply the BEER approach to Turkey using Johansen cointegration technique and quarterly data for the period running from the first quarter of 1987 to the first quarter of 1999. In this study, comparing the fitted values of the estimated equation with the actual bilateral real exchange rate, the real exchange are found to have been undervalued before 1990 and following the crisis in 1994. Using yearly observations from 1970 to 1997, Achy (2001) employs Johansen cointegration method to detect possible long-run relationship between the real exchange rate and its determinants. Long-run values of the fundamentals are obtained employing the approach suggested by Cottani et al. (1990) and Baffes et al. (1999). The author comes to the conclusion that the Turkish lira was overvalued during the last years of his sample period. Doroodian et al. (2002) estimate a single equation over the period covering January 1987 to June 1996 and implement the moving average method to uncover the long-run fundamentals. The results show that the real effective exchange rate was undervalued prior to the second quarter of 1989 and mostly overvalued between the second quarter of 1989 and 1994, but these misalignments were disappeared in the long-run. Further, a study by Atasoy and Saxena (2006) estimate five reduced-form equations with the help of Johansen cointegration analysis over the period covering the first quarter of 1980 to the second quarter of 2003. Permanent component of the long-run fundamentals is estimated by the methodology proposed by Gonzalo and Granger (1995). They concludes that although the real exchange rate was overvalued on the eve of the crises in 1994 and 2001, this overvaluation was eliminated until the second quarter of 2003. The most recent study by Dağdeviren et al. (2009) employs two measures for the real exchange rate and applies Johansen cointegration test over the third quarter of 1998 to first quarter 2008. The misalignment determined based on Hodrick-Prescott filter indicates that the Turkish lira was overvalued before the 2001 crisis, but this misalignment has been 7 corrected in the post-crisis period. Kibritçioğlu and Kibritçioğlu (2004) adopt different BEER approach than the approach that was put forward by Edwards (1989, 1994). They estimate several models using different explanatory variables which capture fundamentals, the ratio of government consumption to GDP, terms of trade and openness ratio by using Johansen and Engle-Granger cointegration methods over the first quarter of 1987 to the third quarter of 2003. The authors also explore the sensitivity of estimation results to the alternative combinations of different real exchange rate and equilibrium real exchange rate. Altogether, 16 different specifications for the misalignment show that the degree of misalignmet is highly sensible to the combination of real exchange rate and its equilibrium level. Table 2 Studies Implementing the BEER Approach for Turkey Authors Time Periods Fundamentals Long-Run Short-Run Alper and Saglam (2000) 1987-1999 TOT, OPEN, INT TECHPROGRESS, KFLOWS Achy (2001) 1970-1997 TOT, OPEN, GOV, KFLOWS DEPRECIATION Doroodian et al. (2002) 1987-1999 TOT, GOV, INV, CAPCONTROL, EXCCONTROL, TECHPROGRESS DEFICIT, DEPRECIATION Kibritcioglu and Kibritcioglu (2004) 1987-2003 TOT, OPEN, GOV Atasoy and Saxena (2006) 1980-2003 TOT, OPEN, GOV, INV, CAPCONTROL, EXCCONTROL, TECHPROGRESS, DEFICIT, EXCREDIT, CONFIDENCE, CURRENT Dagdeviren et al. (2009) 1998-2008 TOT, OPEN, INV, INT TECHPROGRESS, KFLOWS, GOV Note: CAPCONTROL: capital controls proxied by the lagged ratio of capital flows to GDP, CONFIDENCE: a change in the composite confidence index, CURRENT: current account, DEFICIT: fiscal policy measure proxied by the ratio of fiscal deficit to lagged high powered money, DEPRECIATION: depreciation rate, EXCCONTROL: trade and exchange controls proxied by the ratio of custom revenues to import, EXCREDIT: monetary policy measure proxied by the rate of growth of domestic credit minus the lagged growth of GDP, GOV: the ratio of government consumption to GDP, INT: international real interest rate, INV: the ratio of investment to GDP, KFLOWS: capital flows, OPEN: openness ratio as a proxy for import tariffs, TECHPROGRESS: technological progress proxied by productivity levels, TOT: terms of trade 8 There are also three studies which used different approaches to estimate the equilibrium exchange rate of the Turkish lira. Civcir (2003a) investigates the validity of purchasing power parity to shed light on whether the real exchange rate was overvalued prior to the 2001 crisis. Using the Johansen cointegration approach and the monthly data stretching from January 1987 to December 2000, he finds that both the CPI-based bilateral real exchange rate and the WPI-based trade weighted real exchange were overvalued while the WPI-based bilateral real exchange rate was undervalued before the crisis in 2001. Employing the augmented monetary exchange rate model, Civcir (2003b) seeks to determine the nominal equilibrium exchange rate. Based on Johansen cointegration method and monthly data spanning from January 1987 to December 2000, a cointegration relationship is estimated including the nominal exchange rate, relative money supply, relative real income, relative nominal interest rate, relative inflation rate and relative price differential. He suggests that the nominal exchange rate was substantially overvalued prior to the 2001 crisis. Özlale and Yeldan (2004) use a time varying parameter model to estimate the real exchange rate in a single equation framework and monthly data for the period running from January 1992 to December 2001. The ingredients of the empirical relationship are the real exchange rate, exchange rate volatility, short-run capital movements, industrial production index, inflation based on consumer price index, budget balance of the public sector and openness ratio. The authors find that the real exchange rate was overvalued after the crisis in 1994 until 1998, and undervalued in the last eight months of 2000 contrary to the expectations because an overvaluation was targeted at the early stages of the 2000 disinflation program. III. Data and Estimation Results The empirical analysis of the real exchange rate of the Turkish lira makes use of a quarterly data spanning from the first quarter of 1987 to the fourth quarter of 2009. Times series are compilied from the Central Bank of Turkey’s Electronic Data Delivery System, the IMF’s International Financial Statistics and the OECD Statistics. There are three variables: the real exchange rate qt , the net foreign assets in ratio to GDP ft , and the relative sectoral productivity differential at . Taking 2005=100 as the base year, the real exchange rate is constructed from the nominal exchange rate expressed in units of US dollar per unit Turkish lira and the GDP deflators of Turkey and the US. Thus, an increase in the real exchange signifies an appreciation. We use the net foreign assets of the Central Bank of Turkey obtained by the difference the foreign assets and the liabilities to non-residents. This series also is expressed as a share of GDP. Regarding the relative sectoral productivity, we employ average labor productivity as a proxy, due to the lack of availability of sectoral data, by expressing it as a ratio of GDP to the the total employment. We also a set of dummy 9 variables to the currency crisis in the first quarter of the year 1994 (D94), the two-tier crisis in the fourth quarter of 2000 and the first quarter of 2001 (D00), and the harsh impact of the global financial turmoil in Turkish economy in the first quarter of 2009 (D09). All these dummies take the value of one on the defined quarters and zero otherwise. Figure 1 illustrates the real exchange rate and explanatory variables. Before moving from the theoretical model to the empirical implementation, the relationships between the real exchange rate and its fundamentals can be determined as follows: There is a close link between the real exchange and the relative labor productivity although the effect of net foreign assets on the real exchange rate is unclear. However, it depends on a very simple analysis. The statistical analysis is presented below. Figure 1 The Real Exchange Rate and Fundamentals 4.9 4.8 .6 q .32 f a .28 .4 4.7 4.6 .24 .2 4.5 .20 4.4 .0 4.3 .16 4.2 -.2 .12 4.1 -.4 4.0 88 90 92 94 96 98 00 02 04 06 08 .08 88 90 92 94 96 98 00 02 04 06 08 88 90 92 94 96 98 00 02 04 06 08 Prior to attempting to estimate the long-run relationship between the real exchange rate and the two fundamentals, we apply the standard unit root tests, that is, the augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests to investigate the integrated order of the time series. The results are reported in Table 3, which clearly indicates that both the ADF and the PP unit root tests fail to reject the presence of a unit root for all series in the level forms, but not in the first differences. Therefore, all series in our sample are integrated order one. Table 3 Tests of Integrated Order Levels First differences Constant Constant & Trend Constant Constant & Trend -2.004 (0) -2.158 (0) -8.152 (0)* -8.107 (0)* * -7.285 (0)* -8.104 (0)* ADF (k) test statistics q f -1.658 (1) -3.691 (1) -7.307 (0) a -1.475 (0) -1.953 (0) -8.147 (0)* PP (k) test statistics q -2.004 (0) -2.158 (0) -8.152 (0)* -8.107 (0)* f -1.661 (1) -3.104 (1) -7.307 (0)* -7.285 (0)* a -1.475 (0) -1.953 (0) -8.147 (0)* -8.104 (0)* Notes: (1) The ADF test (Dickey and Fuller, 1979, 1981) is based on estimating the test regression 10 xt α βt δxt 1 i 1 γi xt 1 ut where represents the first difference operator, α is k constant, β is the coefficient of a time trend t , k is the lag order of the autoregressive process, and ut is the white noise error term. Under the null hypothesis that xt has a unit root, δ = 0 . The PP test (Phillips and Perron, 1988) estimates the non-augmented form of this regression and directly modifies the t-ratio of the δ coefficient to correct for any serial correlation and heteroscedasticity in the error term. (2) The superscript * signifies the rejection of the null hypothesis at 1 percent critical values. Critical values are taken from the tables compiled by MacKinnon (1996). (3) Numbers in parantheses denote the lag length and are determined by using the minimum value of Schwarz information criterion. The maximum lag is taken as 8. (4) Sample period is 1987Q1−2009Q4. The first step in investigating the long-run relationship amongst the variables integrated order one is to decide the appropriate lag length of the unrestricted VAR system. We choose the proper lag length as 6 because it is the minimum lag sufficient to eliminate serial correlation, heteroscedasticity and non-normality in the residuals. In order to test the absence of serial correlation, heteroscedasticity and non-normality in the residuals of VAR (6), the Breusch-Godfrey LM test (Breusch, 1978; Godfrey, 1978), the Ljung-Box Q test (Ljung and Box, 1979), the White heteroscedasticity test (White, 1980) without cross terms and the Jarque-Bera normality test (Jarque and Bera, 1987) are implemented. The results of Table 4 reveals that the residuals do not display any serial correlation, and are homoscedastic and multivariate normal. Table 4 Residual Misspecification Tests for Unrestricted VAR (6) Tests Test Statistics p-value Serial Correlation Breusch-Godfrey Test LM(1) 13.73 0.13 LM(4) 14.68 0.10 42.60 0.21 Ljung-Box Test LB(10) Heteroscedasticy White Test 240.8 0.37 Normality Jarque-Bera Test 4.761 0.57 Skewness 0.795 0.85 Kurtosis 3.966 0.27 Notes: (1) The null hypotheses of residuals tests are that the residuals do not display any serial correlation, and are homoscedastic and multivariate normal. The multivarite tests are based on Cholesky decomposition of the covariance matrix. (2) The estimation period is 1987Q1−2009Q4. 11 (3) Unrestricted VAR includes six lags on each variable, a constant, and D94, D00 and D09 dummy variables while there is only a separate drift in the cointegrating vector. As a second step for our econometric analysis, we examine whether the variables used are cointegrated with each other on the basis of the trace and maximal eigenvalue tests. The results of both test statistics presented in Table 5 allow us to reject the null hypothesis of no cointegration in favor of one cointegration relationship at a 95 percent significance level. Therefore, we have a strong support that there is just one cointegrating relationship in the chosen set of variables. Table 5 Johansen Cointegration Test Trace test Rank Eigenvalue r 0 Maximal eigenvalue test λtrace p-value λmax p-value 0.258 30.62 0.04** 25.32 0.01** r 1 0.052 5.300 0.78 4.549 0.80 r2 Notes: 0.009 0.751 0.39 0.751 0.39 (1) The methodology of Johansen (1988, 1991, 1995) estimates the vector error correction model which can be specified as xt xt 1 + i 1 i xt i ut where xt is an n x 1 vector of I(1) k variables. If has less than full rank, but the rank of is not equal to zero, then can be written as αβ where α is an n x r matrix of weights interpreted as a speed of adjustment towards equilibrium, and β is an n x r matrix of parameters determining the cointegrating relationships. The numbers of cointegrating vector are determined by the trace and maximal eigenvalue tests. (2) The statistics λtrace and λmax are the trace and maximal eigenvalue statistics, respectively. (3) VAR includes six lags on each variable, a constant, and D94, D00 and D09 dummy variables while cointegrating vector has only an intercept. (4) The superscript ** signifies the rejection of the null hypothesis at 5 percent critical values. Critical values are extracted from MacKinnon et al. (1999). (5) The estimation period is 1987Q1-2009Q4. Table 6 reports the cointegrating coefficients with associated t-statistics corresponding to (14). All coefficients in the cointegrating vector are found to be statistically significant at 1 percent level. Thus, we can express the cointegration relationship with a constant term as follows: qt 3.756 0.424 ft 5.181at (14) The normalized cointegrating equation exhibits that the relative labor productivity is correctly signed, but the sign of net foreign assets is not predicted but expected because the 12 result of negative relationship between net foreign assets and the real exchange rate is consistent with previous findings of Égert et al. (2004), Bénassy-Quéré et al. (2006) and Alberola and Naiva (2007). The underlying cause of this result is probably the difference between the real international interest rate and the growth rate of domestic real GDP. In fact, average long-term real interest rate of the US is calculated to be 3.42 percent and average real growth rate of Turkey is computed as 3.75 percent throughout the period under investigation. Table 6 also illustrates the speed of adjustment parameters. The adjustment coefficient on the real exchange rate is highly significant and negatively signed. It indicates that the real exchange rate moves to close the gap of a disequilibrium. The percentage of the total adjustment offset in each successive quarter is 50 percent for the real exchange rate. However, we find the adjustment coefficient of the net foreign assets is modestly significant and that of the relative labor productivity is insignificant. Tablo 6 Estimation of Cointegrating and Adjustment Coefficients Normalized cointegrating coefficients Coef. Std. Err. Adjustment coefficients t-Stat Coef. Std. Err. t-Stat Δq −0.501 0.175 −2.862* q 1.000 f 0.424 0.072 5.848* Δf −0.218 0.112 −1.950*** a −5.181 0.397 −13.06* Δa −0.032 0.035 0.907 * Note: The superscripts and values, respectively. *** signify the rejection of the null hypothesis at 1 and 10 percent critical We also implement exclusion, weak exogeneity and multivariate stationarity tests. All test results are summarized in Table 7. Recall that in Balassa-Samuelson hypothesis, net foreign assets are not considered as one of the fundamentals while the balance of payment approach ignores the effect of the relative labor productivity on the real exchange rate. Purchasing power parity entails that both net foreign assets and the relative labor productivity are excluded from the cointegrating vector. The exclusion test results show that net foreign assets and the relative labor productivity are not only individually but also jointly significantly different from zero. This result demostrates that purchasing power parity does not hold, and the real exchange rate is determined by both the balance of payment approach and Balassa-Samuelson hypothesis. Johansen cointegration analysis permits us to conduct tests for weak exogeneity of the variables with respect to the parameters of cointegrating relationship. The weak exogeneity test results notes that the weak exogeneity of all variables, apart form relative labor productivity, can be rejected at least at 5 percent significance level. The weak exogeneity of relative labor productivity is not a surprising result because it is mainly 13 determined outside the system by the conditions of labor market in Turkey and the US. However, the weak exogeneity of net foreign assets can not be rejected at 1 percent significance level. It is conceivable that the net foreign assets of the Central Bank of Turkey may, independently from the exchange market conditions, be based on its monetary policy programming. Furhermore, the joint weak exogeneity test indicates net foreign assets and the relative labor productivity are weakly exogenous at 1 percent significance level. Finally, we employ the multivariate stationarity test proposed by Johansen and Juselius (1990) to determine the order of integration of the variables within the multivariate context because there is only one cointegrating relationship. The multivariate stationarity test result indicates the nonstationarity of all the variables, confirming the univariate unit root test results. Tablo 7 Exclusion, Weak Exogeneity and Multivariate Stationarity Tests Exclusion Test Weak Exogeneity Test Multivariate Stationarity Test χ2 Stat p-value Χ2 Stat p-value χ2 Stat p-value q 19.70 0.00 9.091 0.00 19.18 0.00 f 17.26 0.00 4.432 0.04 21.22 0.00 a 19.11 0.00 1.000 0.32 22.00 0.00 f&a 19.18 0.00 6.621 0.04 Note: The likelihood ratio statistic for the exclusion, weak exogeneity and multivariate stationary tests has an asymptotic χ2 distribution. Finally, we estimate the error correction model using the result that the relative labor productivity is a only weak exogenous variable at 5 percent level and deleting step by step insignificant lags of the variables. Thus, the parsimonious model for short-run real exchange rate is given as follows: qt 0.0005 0.258ect 1 0.257 qt 3 0.089qt 5 4.397 at 1.604at 3 0.095D94 0.030 D00 0.005D09 (15) where ect 1 denotes the lagged error correction term. The negative and significant coefficient of the lagged error correction term indicates how quickly variables return to equilibrium when an exogenous shock distrubs the equilibrium condition; as well R2 is 0.86. The magnitude of this coefficient shows that approximately 26 percent of the adjustment towards the equilibrium takes place per quarter, which implying that, in the absence of further shocks, 50 percent of the gap would be eliminated within about 10 months. In the short-run, the real exchange rate is affected by its third and fifth lags. The relative labor productivity still plays an important role on the movements of the real exchange rate, but its effect decreases. As 14 opposed to the long-run relationship, the impact of net foreign assets on the real exchange rate is eradicated. If we consider both net foreign assets and the relative labor productivity as weakly exogenous variables at 1 percent level of significance, the short-run dynamics of the real exchange rate can also be written as qt 0.002 0.239ect 1 0.258qt 3 0.100qt 5 4.469at 1.585at 3 0.154ft 0.078D94 0.035D00 0.006 D09 (16) The short-run parameters now estimated are roughly the same as that of the first error correction model with R2 is 0.87. However, net foreign assets enters with positive sign in the new error correction model on the contrary to the long-run relationship, but consistent with the theoretical prediction. IV. Estimation of the Equilibrium Exchange Rates and Misalignments In the BEER appoach, there are two basic equilibrium real exchange rate concepts, short-run equilibrium real exchange rate and long-run equilibrium real exchange rate. The first one is the real exchange rate given by the current values of the fundamentals, and the second one is defined as the real exchange rate determined by the long-run values of the fundamentals. It is highly possible that these two equilibrium concepts are differentiated from each other because the current values of fundamentals can depart from their long-run levels. Therefore, we also have two different real exchange rate misalignment concepts, namely short-run misalignment and long-run misalignment if the misalignment is defined as the sustained deviation of the actual value of the real exchange rate from its equilibrium level. Following Clark and MacDonald (1998), the real exchange rate can simply be characterized as qt αZt βTt ut (17) where Z t is a set of fundamentals which have persistent effects on the real exchange rate, such as the relative labor productivity and net foreign assets, Tt is a set of transitory factors which have a short-run effect on the real exchange rate and include current and lagged variables as well as dynamic effects from the fundamentals, and u t is unexpected shocks. Both the short-run and the long-run misalignments, which are denoted by mistSR and mistLR , are obtained as mistSR βTt ut (18) mistLR mistSR α Zt Z (19) 15 where Z represents the long-run values of the fundamentals. It is obviously seen from (18)−(19) that short-run misalignment is simply the sum of the transitory factors and unexpected shocks while the long-run misalignment is composed of the short-run misalignment and the extent to which the fundamentals are away from their long-run values. Thus, the differences between these misalignments arise actually from the second component of the long-run misalignment. Figure 2 shows both the equilibrium real exchange rates and the real exchange rate misalignments for the whole period of study. The left-hand side illustrates the short-run and long-run equilibrium real exchange rates by along with the actual real exchange rate while the right-hand side plots the short-run and the long-run misalignments in the real exchange rate. Positive value of the misalignment means that the real exchange is overvalued. Figure 2 Equilibrium Real Exchange Rates and Real Exchange Rate Misalignments 5.2 .2 q q_SR q_LR 5.0 .1 4.8 .0 4.6 -.1 4.4 -.2 4.2 -.3 4.0 -.4 mis_SR 88 90 92 94 96 98 00 02 04 06 08 88 (a) Equilibrium Real Exchange Rates 90 92 94 96 98 00 02 04 mis_LR 06 08 (b) Real Exchange Rate Misalignments Note: q_SR and q_LR denote the short-run and long-run equilibrium real exchange rates, and mis_SR and mis_LR represent the short-run and long-run misalignments in the real exchange rate, respectively. Looking first at the equilibrium real exchange rates, the real exchange rate follows more or less the same path of its equilibrium levels although it moves closer to the short-run equilibrium real exchange rate. However, the long-run equilibrium real exchange is more stable than short-run equilibrium real exchange rate because the former is derived from the long-run values of the fundamentals estimated using Hodrick-Prescott filter. The long-run equilibrium real exchange rate can be analyzed in three sub-periods: a appreciation from the first quarter of 1987 to the second quarter of 1992, a depreciation from the third quarter of 1993 to the fourth quarter of 2000 and a appreciation from the first quarter of 2001 to the fourth quarter of 2009.It is noteworthy to point out that the fundamentals account for most of the long-run equilibrium real exchange rate. The appreciations at the first and the end of the sample period appear to be associated only with upward trend in the relative labor productivity whereas a rise in net foreign assets may in part account for the depreciation period in addition to declining relative labor productivity. 16 Turning to the real exchange rate misalignments, we can say that there is a connection between the short-run and the long-run misalignments until the early 2006 except for 1993. At the beginning of 1987, the real exchange rate was undervalued by about 10 percent and this initial undervaluation lasted until the third quarter of 1989 owing to unsuccessful disinflationary efforts and debt financing policies. With the advent of capital account liberalization in August 1989, the period of overvaluation started at the end of 1989 and lasted until the second quarter of 1993. Although short-run misalignment was inferior to zero in the rest of 1993 as opposed to long-run misalignment, the monetization of fiscal deficits in the last months of 1993 because of rapidly rising public sector borrowing requirement during 1992-1993 led eventually to the currency crisis in 1994. In the aftermath of the crisis, the Turkish lira was sharply undervalued since it was devalued more than 80 percent against the US dollar in the second quarter of 1994. In the start of 1995, the real exchange rate converged back to its equilibrium levels and was mildly overvalued until the end of this year. However, the currency in real terms was slighty undervalued in 1996 and 1997, this resulted from political uncertainties arising from the early elections held in December 1995 and the contagion effect of the Asian crisis broke out in 1997. Disinflation programs of 1998 and 2000, which were conducted under the supervision and technical support of the IMF, decreased the level of anxiety about Turkish economy and overvaluation occurred from 1998 to 2000 although neither of the programs lived long and Turkey witnessed a serious capital outflow after the Russian crisis of 1998. In the run up to the twotier crisis in December 2000 and February 2001, the real exchange rate was overvalued by 11 percent in 1999 and 7 percent in 2000. Following the twin crises, the real exchange rate was undervalued for two consecutive years because no sooner had Turkey switched from the crawling peg to a floating exchange rate regime in February 2001 than the currency depreciated massively. With the favorable economic environment provided by the political stability following the November 2002 elections, overvaluation started in second quarter of 2003 and came to an end as of the second quarter of 2006 although the Federal Reserve started to increase policy rates in the second half of 2004. After 2006, it is worth noting that not only has the link between the short-run and the long-run misalignments been broken down for the last three years, the short-run misalignments have also started to move in the reverse direction from the long-run misalignments because the relative labor productivity as a main fundamental of the real exchange rate has significantly been departed from its long-run levels. In fact, the positive difference between the relative labor productivity and its long-run level has triggered off undervaluation of the real exchange rate in the short-run and overvaluation in the long-run 17 while negative difference has brought about overvaluation in the short-run and undervaluation in the long-run since 2006. Considering only the long-run misalignment, we can see that undervaluation in 2006 was a result of the deterioration in global risk perception in the post-May period of the year. Since this shock was short-lived, the real exchange rate was overvalued in the second quarter of 2007 and remained overvalued until the last quarter of 2008. However, undervaluation appeared once again in the last year of the sample with the global financial turmoil spreading to Turkey after the last quarter of 2008. Together with annual growth rate, the magnitudes of the misalignments reported in Table 8, where the long-run misalignments indicate that the real exchange rate was overvalued by more than 7 percent on average for at least three years prior to the crises of 1994 and 2000/01. Therefore, large and persistent overvaluation can serve as an early warning indicator for potential crisis in Turkey. Tablo 8 Short-Run and Long-Run Misalignments and Growth Rate (annual averages of quarterly values, in percent) Year Misalignment Growth Rate Year Misalignment Short-Run Long-Run 1987 −11.7 −9.24 1988 −10.8 −16.2 2.98 2000 7.22 6.88 7.16 1989 −3.13 −10.1 −0.02 2001 −5.53 −15.4 −6.95 1990 2.73 11.4 10.3 2002 −3.70 −10.4 7.57 1991 5.60 6.91 0.36 2003 0.74 1.93 6.07 1992 3.05 3.44 5.09 2004 1.78 5.62 8.87 1993 −2.76 8.12 7.64 2005 2.25 5.45 7.54 1994 −7.16 −18.2 −4.19 2006 4.16 −3.54 6.14 1995 1.45 1.74 8.27 2007 −5.51 2.87 5.33 1996 −0.97 −1.84 7.30 2008 −8.66 6.75 1.74 1997 −2.83 −0.72 7.42 2009 5.19 −9.35 −4.88 1998 1.62 7.01 3.56 1999 Short-Run 10.6 Long-Run Growth Rate 10.6 −4.74 We also compare the long-run misalignment with the annual growth rate of real GDP. In the literature, there are two different views related to the role of misalignment on the growth; one view is the Washington Consensus view and the other is the Rodrik view (Berg and Miao, 2010). The first view argues that any misalignment in the real exchange rate has adverse effects on the growth rate while the second view asserts that only overvaluation hurts the growth, but undervaluation facilitates the growth. As evident from Figure 3, however; it is found that, contrary to both views, there is a mildly positive relationship between the long-run misalignment and the growth; the more overvalued the Turkish lira, the 18 higher the growth rates and the more undervalued the Turkish lira, the lower the growth rates3. In fact, overvaluation led to the economic contraction only in 1999 while economic growth was associated with undervaluation only in the years of 1988, 1996, 1997, 2002 and 2006, as can be seen in Table 8. This result emanates from the strategy of the import-led growth: Overvaluation cheapens the costs of imported machinery and equipment and intermediate products which are vastly necessary for Turkey to stimulate higher growth before completing the process of economic development. However, this strategy does not seem to be sustainable because, again, the real exchange rate is undervalued and economic performance is flawed when a financial crisis bursts out after overvaluation reaches a dangerous level. Figure 3 Growth Rate and Long-Run Misalignment 20 10 0 -10 -20 -30 g mis_LR -40 88 90 92 94 96 98 00 02 04 06 08 Note: g is the annual growth rate of real GDP and mis_LR is the long-run misalignment in the real exchange rate. Finally, we attempt to compute the equilibrium nominal exchange rates because the nominal exchange rate is more easily controlled by policymakers and more observable to others than the real exchange rate. Using a highly simplified approach, equilibrium nominal exchange rates are calculated by dividing an index of the ratio of Turkish prices to US prices by the real equilibrium exchange rates, and the results are given in Table 9. Tablo 9 Short-Run and Long-Run Nominal Equilibrium Exchange Rates (annual averages of quarterly values, in percent) Nominal Exchange Rate Year Nominal Exchange Rate Equilibrium Actual ShortRun LongRun Year Equilibrium Actual ShortRun LongRun 1987 0.0009 0.0008 0.0008 1999 0.4188 0.4665 0.4653 1988 0.0014 0.0013 0.0012 2000 0.6252 0.6713 0.6693 3 Growth equation estimation results can be obtained from authors upon request. 19 1989 0.0021 0.0021 0.0019 2001 1.2256 1.1405 1.0311 1990 0.0026 0.0027 0.0029 2002 1.5072 1.4528 1.3555 1991 0.0042 0.0044 0.0044 2003 1.5009 1.5087 1.5248 1992 0.0069 0.0071 0.0071 2004 1.4255 1.4505 1.5056 1993 0.0110 0.0107 0.0119 2005 1.3436 1.3742 1.4191 1994 0.0296 0.0277 0.0245 2006 1.4285 1.4883 1.3770 1995 0.0458 0.0465 0.0468 2007 1.3029 1.2368 1.3373 1996 0.0814 0.0806 0.0799 2008 1.3015 1.2076 1.3829 1997 0.1519 0.1475 0.1511 2009 1.5500 1.6343 1.4106 1998 0.2607 0.2659 0.2813 V. Conclusion In this paper, we aim to investigate the impact of overvaluation on the financial crisis and the growth rate for Turkey using the quarterly data from 1987 to 2009. For that reason, we estimate the short-run and the long-run equilibrium real exchange rates along the lines of Faruqee (1995) and Alberola et al. (1999). Our empirical results indicate that the equilibrium real exchange rate appreciates with positive shocks to the relative labor productivity whereas an increase in net foreign assets yields a depreciation, which is contrary to what the model predicts. However, this result is expected because there is a negative difference between average long-term real interest rate of the US and average real growth rate of Turkey during the period under investigation. It is also found that purchasing power parity does not hold, and the real exchange rate is determined by both the balance of payment approach and Balassa-Samuelson hypothesis in Turkey. Comparing the real exchange rate with the equilibrium exchange rates, we can draw two important policy implications: First, large and persistent overvaluation, which is more than 7 percent on average for at least three consecutive years, can serve as an early warning indicator for potential crisis in Turkey. Second, a relatively small overvaluation, contrary to both the Washington Consensus view and the Rodrik view, stimulate the growth of Turkish economy due to the strategy of import-led growth. 20 References Achy, L., (2001), “Equilibrium Exchange Rate and Misalignment in Selected MENA Countries”, European University Institute, Working Paper No. 2001/42. Alberola, E., (2003), “Misalignment, Liabilities Dolarization and Exchange Rate Adjustment in Latin America”, Bank of Spain, Working Paper No. 0309. Alberola, E. and H. López, (2001), “Internal and External Exchange Rate Equilibrium in a Cointegration Framework: An Application to the Spanish Peseta”, Spanish Economic Review, 3, 23-40. Alberola, E. and D. Naiva, (2007), “Equilibrium Exchange Rates in the New EU Members: External Imbalances vs. Real Convergence”, Bank of Spain, Working Paper No. 0708. Alberola, E., López, H. and L. Servén, (2004), “Tango with the Gringo: The Hard Peg and Real Misalignment in Argentina”, Bank of Spain, Working Paper No. 0405. Alberola, E., Garcia-Cervero, S., López, H. and A. Ubide, (1999), “Global Equilibrium Exchange Rates: Euro, Dollar, “Ins”, “Outs”, and Other Major Currencies in a Panel Cointegration Framework”, International Monetary Fund (IMF), Working Paper No. 99/175. Alberola, E., Garcia-Cervero, S., López, H. and A. Ubide, (2002), “Quo vadis Euro?”, European Journal of Finance, 8, 352-370. Alper, C.E. and I. Sağlam, (2000), “The Equilibrium Real Exchange Rate: Evidence form Turkey”, Topics in Middle Eastern and North African Economies, 2, (Online Journal available at http://www.luc.edu/orgs/meea/volume2/alper.html). 21 Artis, M. and M. Taylor, (1995), “The Effect of Misalignment on Desired Exchange Rates: Some Analytical Results”, in: C. Bordes, E. Girardin and J. Mélitz (eds.), European Currency Crises and After, Manchester: Manchester University Press. Atasoy, D. and S.C. Saxena, (2006), “Misaligned? Overvalued? The Untold Story of the Turkish Lira”, Emerging Markets Finance and Trade, 42, 29–45. Babetskii, I. and B. Égert, (2005), “Equilibrium Exchange Rate in the Czech Republic: How Good is the Czech BEER?”, Center for Economic Research and Graduate Education-Economic Institute (CERGE-IE), Working Paper No. 267. Baffes, J., Elbadawi, İ.A. ve S.A. O’Connell, (1999), “Single Equation Estimation of the Equilibrium Real Exchange Rate”, in: L. Hinkle and P. Montiel (ed.), Exchange Rate Misalignment: Concepts and Measurement for Developing Countries, Washington: World Bank. Bayoumi, T., Clark, P., Symansky, S. And M. Taylor, (1994), “The Robustness of Equilibrium Exchange Rate Calculations to Alternative Assumptions and Methodologies”, in: J. Williamsom (ed.), Estimating Equilibrium Exchange Rates, Washington: Institute for International Economics. Bénassy-Quéré, A., Lahrèche-Révil, A. and V. Mignon, (2006), “World Consistent Equilibrium Exchange Rates”, Centre d’Etudes Prospectives et d’Informations Internationales (CEPII), Working Paper No. 2006/20. Bénassy-Quéré, A., Duran-Vigneron, P., Lahrèche-Révil, A. and V. Mignon, (2004), “Burden Sharing and Exchange Rate Misalignments within the Group of Twenty”, Centre d’Etudes Prospectives et d’Informations Internationales (CEPII), Working Paper No. 2004/13. Berg, A. and Y. Miao, (2010), “The Real Exchange Rate and Growth Revisited: The Washington Consensus Strikes Back?”, International Monetary Fund, Working Paper No. 10/58. Breusch, T.S., (1978), “Testing for Autocorrelation in Dynamic Linear Models”, Australian Economic Papers, 17, 334–355. Bulíř, A. ve K. Šmídková, (2005), “Exchange Rates in the New EU Accession Countries: What Have We Learned from the Forerunners”, Economic Systems, 29, 163–186. Civcir, İ., (2003a), “Before the Fall, Was the Turkish Lira Overvalued?”, Eastern European Economics, 41, 69–99. Civcir, İ., (2003b), “The Monetary Model of the Exchange Rate under High Inflation: The Case of the Turkish Lira/US Dollar”, Finance a úvĕr – Czech Journal of of Economics and Finance, 53, 113–129. 22 Clark, P.B. and R. MacDonald, (1998), “Exchange Rates and Economic Fundamentals: A Methodological Comparison of BEERs and FEErs”, International Monetary Fund, Working Paper No. 98/67. Cottani, J.A., D.F. Cavallo, M.S. Khan, (1990), “Real Exchange Rate Behavior and Economic Performance in LDCs”, Economic Development and Cultural Change, 39, 61–76. Dağdeviren, S., Binatlı, A.O. and N. Sohrabji, (2009), “Real Exchange Rate Misalignment: The Turkish Experience”, Social Science Research Network, Working Paper. Dickey, D.A. and W.A. Fuller, (1979), “Distribution of the Estimators for Autoregressive Time Series with a Unit Root”, Journal of the American Statistical Association, 74, 427–431. Dickey, D.A. and W.A. Fuller, (1981), “Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root”, Econometrica, 49, 1057–1072. Doroodian, K., Jung, C. and A. Yücel, (2002), “Estimating the Equilibrium Real Exchange Rate: The Case of Turkey”, Applied Economics, 14, 1807–1812. Driver, R.L. and P.F. Westaway, (2004), “Concepts of Equilibrium Exchange Rates”, Bank of England, Working Paper No. 248. Edwards, S., (1989), “Real Exchange Rates, Devaluation and Adjustment: Exchange Rate Policy in Developing Countries”, Cambridge: MIT Press. Edwards, S., (1994), “Real and Monetary Determinants of Real Exchange Rate Behavior: Theory and Evidence from Developing Countries”, In Estimating Equilibrium Exchange Rates, ed. J. Williamson, pp. 61–91, Washington: Institute for International Economics. Égert, B., Lahrèche-Révil, A. and K. Lommatzsch, (2004), “The Stock-Flow Approach to the Real Exchange Rate of CEE Transition Economies”, Centre d’Etudes Prospectives et d’Informations Internationales (CEPII), Working Paper No. 2004/15. Elbadawi, I., (1994), “Estimating Long-Run Equilibrium Real Exchange Rates”, in: J. Williamson (ed.), Estimating Equilibrium Exchange Rates, Washington: Institute for International Economics. Fabella, R.V., (1996), “The Debt-Adjusted Real Exchange Rate”, Journal of International Money and Finance, 15, 475–484. Faruqee, H., (1995), “Long-Run Determinants of the Real Exchange Rate: A Stock-Flow Perspective”, International Monetary Fund Staff Papers, 42, 80–107. 23 Frait, J. and L. Komárek, (2002), “Theoretical and Empirical Analysis of the Debt Adjusted Real Exchange Rate in Selected Transition Economies During 1994–2001”, University of Warwick, Department of Economics, Research Paper No. 646. Frait, J. and L. Komárek, (2008), “The Debt Adjusted Real Exchange Rate for China”, University of Warwick, Department of Economics, Research Paper No. 850. Godfrey, L.G., (1978), “Testing for Higher Order Serial Correlation in Regression Equations When the Regressors Include Lagged Dependent Variables”, Econometrica, 46, 1303–1310. Gonzalo, J. and C. Granger, (1995), “Estimation of Common Long-Memory Components in Cointegrated Systems”, Journal of Business and Economic Statistics, 13, 27–35. Jarque, C.M. and A.K. Bera, (1987), “A Test for Normality of Observations and Regression Residuals”, International Statistical Review, 55, 163–172. Johansen, S., (1988), “Statistical Analysis of Cointegrating Vectors”, Journal of Economics and Dynamic Control, 12, 231–254. Johansen, S., (1991), “Estimation and Hypothesis Testing of Cointegration: Vectors in Gaussian Vector Autoregressive Models”, Econometrica, 59, 1551–1580. Johansen, S., (1995), Likelihood-Based Inference in Cointegrated Vector Autoregressive Models, Oxford: Oxford University Press. Johansen, S. and K. Juselius (1990), “Maximum Likelihood Estimation and Inference on Cointegration-With Applications to the Demand for Money”, Oxford Bulletin of Economics and Statistics, 52, 169–210. Kibritçioğlu, A. And B. Kibritçioğlu, (2004), “Real Exchange Rate Misalignment in Turkey, 1987-2003 (in Turkish)”, Ankara University Faculty of Political Sciences, Discussion Paper No. 68. Ljung, G. And G. Box, (1979), “On a Measure of Lack of Fit in Time Series Models”, Biometrika, 66, 265–270. Loayza, N., Fajnzylber, P. and C. Calderón, (2005), “Economic Growth in Latin America and the Caribbean: Stylized Facts, Explanations, and Forecasts”, Central Bank of Chile, Working Paper No. 265. MacDonald, R., (2000), “Concepts to Calculate Equilibrium Exchange Rates: An Overview”, Deutsche Bundesbank, Discussion Paper No. 3/00. MacKinnon, J.G., (1996), “Numerical Distribution Functions for Unit Root and Cointegration Tests”, Journal of Applied Econometrics, 11, 601–618. 24 MacKinnon, J.G., Haug, A.A. and L. Michelis, (1999), “Numerical Distribution Functions of Likelihood Ratio Tests for Cointegration”, Journal of Applied Econometrics, 14, 563–577. Montiel, P.J., (1999b), “Determinants of the Long-Run Equilibrium Real Exchange Rate: An Analytical Model”, in: L. Hinkle and P. Montiel (ed.), Exchange Rate Misalignment: Concepts and Measurement for Developing Countries, Washington: World Bank. Phillips, P.C.B. and P. Perron, (1988), “Testing for a Unit Root in Time Series Regression”, Biometrika, 75, 335–346. Rahn, J., (2003), “Bilateral Equilibrium Exchange Rates of EU Accession Countries against the Euro”, Bank of Finland-Institute for Economies in Transition (BOFIT), Working Paper No. 2003/11. Rubaszek, M., (2004), “A Model of Balance of Payments Equilibrium Exchange Rate: Application to the Zloty”, Eastern European Economics, 42, 5–22. MacKinnon, J.G., Haug, A.A. and L. Michelis, (1999), “Numerical Distribution Functions of Likelihood Ratio Tests for Cointegration”, Journal of Applied Econometrics, 14, 563–577. Özlale, Ü. And E. Yeldan, (2004), “Measuring Exchange Rate Misalignment in Turkey”, Applied Economics, 36, 1839–1849. Stein, J.L., (1990), “The Real Exchange Rate”, Journal of Banking anf Finance, 14, 1045– 1078. Stein, J.L., (1994), “The Natural Real Exchange Rate of the US Dollar and Determinants of Capital Flows”, in: J. Williamson (ed.), Estimating Equilibrium Exchange Rates, Washington: Institute for International Economics. Strauss, J., (1995), “Real Exchange Rates, PPP and the Relative Prices of Nontraded Goods”, Southern Economic Journal, 61, 991–1005. Strauss, J., (1996), “The Cointegrating Relationship between Productivity, Real Exchange Rates and Purchasing Power Parity”, Journal of Macroeconomics, 18, 299–313. Strauss, J., (1999), “Productivity Differentials, the Relative Price of Nontradables and Real Exchange Rates”, Journal of International Money and Finance, 18, 383–409. Wadhwani, S.B., (1999), “Sterling’s Puzzling Behaviour”, Bank of England Quarterly Bulletin, 39, 416–427. White, H., (1980), “A Heteroskedasticity-Consistent Covariance Matrix and a Direct Test for Heteroskedasticity”, Econometrica, 48, 817–838. Williamson, J., (1985), The Exchange Rate System, Washington: Institute for International Economics. 25 Williamson, J., (1994), “Estimates of FEERs”, in: J. Williamson (ed.), Estimating Equilibrium Exchange Rates, Washington: Institute for International Economics. 26
© Copyright 2026 Paperzz