Drumhead: In this paper, we try to analyse the cogency of the buying power para ( PPP ) in Turkey for monthly informations from 2003:1 to 2012:07. The grounds suggests that existent effectual exchange rate is stationary employing, conventional unit root trials including Augmented Dickey-Fuller and Phillips-Perron. However, both unit root trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee and Strazicich ( 2003 ) trials and conventional unit root trials including KPSS and Ng-Peron trials suggest that existent exchange rate series are integrated of order I ( 1 ) . Empirical trials show that PPP does non keep in the long tally in Turkey.

Keywords: Real exchange rate, Buying Power Parity, unit root trials, structural interruptions, Turkey

Jel Code: F31, C22

Introduction

The finding of equilibrium existent exchange rates is one of the cardinal economic constructs in an international economic system particularly during economic crisis periods. There are a figure of different methodological analysiss[ 3 ]to measure equilibrium exchange rates. And the hypothesis of buying power para ( PPP ) is one of the most studied methodological analysiss in finding existent exchange rates in economic literature. Although the subject of PPP ” was foremost used eight-five old ages ago by the Swedish economic expert Gustav Cassel ( Cassel, 1918 ) , it has received much great attractive force by economic experts since so. Theoretically, PPP has been accepted as the rising prices theory of exchange rate and besides is the get downing point of the exchange rates theories. It by and large states that the exchange rate alteration between two currencies over a period of clip is determined by the alteration in the two states ‘ comparative monetary value degrees ( Dornbusch, 1985:1 ) .

One manner to analyze the long-term cogency of PPP is to look into the stationarity belongingss of existent exchange rates. A necessary status for long-term PPP is that the existent exchange rate should be average backsliding Besides, predictability of existent exchange rates based on average reversion conditions provide empirical support for PPP. For illustration, Adler and Lehmann ( 1983 ) , Darby ( 1980 ) , Hakkio ( 1986 ) , and Roll ( 1979 ) note that the existent exchange rate can or should follow a random walk, theoretically[ 4 ].

ThereA exist twoA types of PPP which areA absolute and relative.A Harmonizing to Absolute PPP, exchange rates and monetary values should set such that goods in different states have the same monetary value when expressed in the same currency.

Relative PPP provinces that the per centum depreciation in the exchange rate between two currencies over any period equals the difference between rising prices rates in the place state and in the foreign state.

Relative PPP does non to keep in the short tally and in the intermediate tally. Failures of the short tally are due to monetary value rigidnesss, merchandise distinction and local production cost barriers to international trade, dii¬ˆerencies in gustatory sensations and A consumer behaviorA and monetary value index acrossA countries. , little grade of mobility of goods and engineering.

However, in the long tally, it is expected to more support forA the cogency of comparative PPP that some of those ineffciencies will vanish, connoting a convergence in the monetary value degrees across states.

Palatopharyngoplasty can besides be used to state if a currency is over or under-valued.A It measures the exchance misalignment by agencies of divergences of the nominal exchange rate from the PPP para with regard to a basal state.

Morever, monetary value stableness and the grade of capital flows are two of import factors that may ai¬ˆect the cogency of long-term PPP. Fluctuations inA existent exchangeA are affected chiefly by the motions A the relativeA priceA of tradable goods ( pass-through affect ) . And besides, the stableness of exchange rate and capital flows is of great concern for inflation-targeting states. Cardinal Bankss in these states use involvement ratesA to command rising prices that may take to

greater volatility in exchange rate.

The relationship between the grade of capital flows and the existent exchange rate volatility are besides indispensable factor in explicating the long tally behavior of existent exchange rate. As the states became more incorporate with the the rise of grade of capital flows to the remainder of the universe, it is impossible for the cardinal bank to accomplish free capital mobility, an independent pecuniary policy and fixed exchange rate at the same time, known as impossible three.

The intent of this paper is to look into long-term buying power para in Turkey for monthly informations from 2003:1 to 2012:07. The selected survey periods and state will give A significantA sum for following emprical perusals. Because, inflationary targating government has been A introduced and there have been a crisp addition in the capital flows in Turkey during the analyzing periods.

For this purpose, we employed unit root trials to analyze the the hypothesis of PPP for Turkey through conventional unit root trials including Augmented Dickey-Fuller and Phillips-Perron. However, both unit root trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee and Strazicich ( 2003 ) trials.

The paper is organized as follows. Section II briefly introduces informations and methodological analysis. Section III presents empirical consequences and subdivision IV concludes.

Literature Review

Several surveies re-examined the cogency of long-runA purchasingA powerA parityA ( PPP ) hypothesis for TurkeyA utilizing different technique.

By utilizing co-integration techniques, Telatar and KazdaglA± ( 1998 ) found noA grounds toA support the PPP hypothesis for Turkey and A chief trading spouses ( France, Germany, the UK, and the USA ) .

Sarno ( 2000 ) re-examineA thisA hypothesisA utilizing the exponential smooth passage autoregressive theoretical account ( ESTAR ) for Turkey and its major trading spouses ( the US, the UK, Germany and France ) and supported the cogency of long-term PPP.

The cogency of theA long-term PPP hypothesis is besides supported by Yazgan ( 2003 ) . He performed his analysis with standard multivariate co-integration techniques.

Erlat ( 2003 ) found grounds of buying power para by utilizing unit root trials and autoregressive fractionally incorporate traveling norm ( ARFIMA ) theoretical accounts. Erlat ( 2004 ) besides investigated PPP for the period 1984:1-2000:9 and found strong grounds of nonlinear stationary.

TaAYtan ( 2005 ) accepted the long-term buying power para ( PPP ) hypothesis holds for the UK and US $ based existent exchange rates for Turkish economic system during the period 1982M1-2003M12.A

Kalyoncu ( 2009 ) tested the cogency of buying power para ( PPP ) for Turkey and its trading spouses ( USA, Germany, Japan, France, Netherlands and UK ) . Harmonizing to trial consequences, PPP testing is sensitive to the pick of the basal state and can be influenced by the type of trial.

Kum ( 2012 ) examines the cogency of the buying power para ( PPP ) in Turkey for one-year informations from 1953 to 2009. He found the evidience of PPP for Turkey by utilizing Zivot and Andrews and Lagrange Multiplier unit root trials.

II. DATA AND METHODOLOGY

Formalized by Cassel ( 1918 ) , the existent Exchange ( RER ) rate can be defined in equation 1:

( 1 )

where E denotes the nominal exc hange rate ( measured as domestic currency monetary value of foreign currency ) , P* is an index of the foreign monetary value degree, and P is an index of the domestic monetary value degree.

Denoting logarithms in lower instance letters, we therefore have equation 2.

( 2 )

R defines logarithmic signifier of existent exchange and p* and p denote the logarithms of the domestic and foreign state monetary value index severally.

The cogency of long-term PPP requires the stationarity of existent exchange rates. If R is stationary so divergences from PPP are impermanent and will vanish with clip, and therefore PPP is likely to keep in the long tally. But if R, has a unit-root, so it implies that divergences from the para are cumulative and non finally self-reverting.

The existent exchange rate ( RER ) is calculated from the nominal effectual exchange rate ( NER ) and a step of the comparative monetary value or cost between the state under survey and its trading spouses. The most popular monetary value and costs steps are consumer monetary values ( CPI ) , manufacturer monetary values ( PPI ) , GDP deflator, unit labour costs ( ULC )[ 5 ].

In this survey, we calculate CPI-based RER for USA and the Euro Area. We employ monthly existent exchange rate informations between Turkish Lira and USA Dollar and Euro for the period 2003M1-2012M6. Nominal US dollar and Euro exchange rates and Turkish CPI series were obtained from Central Bank of Turkey, Electronic Data Delivery System. US CPI was obtained from US Department of Labor, Bureau of Labor Statistics and European CPI was obtained from Eurostat. Real exchange rate series for US Dollar and Euro are denoted as RERUS and REREU severally.

This survey aims to look into stationarity belongingss of the existent exchange rate of Turkey using both conventional unit root trials and unit root trials with structural interruptions. In order to look into stationary belongingss of the RER series, we employed conventional unit root trials including ADF ( Dickey and Fuller, 1981 ) , PP ( Phillips-Perron, 1987 ) , KPSS ( Kwiatkowski, Phillips, Schmidt, and Shin, 1992 ) and Ng-Perron trials ( 2001 ) and unit root trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee and Strazicich ( 2003 ) trials.

Empirical plants based on clip series informations assume that the implicit in clip series is stationary. A stochastic procedure is said to be stationary if its mean and discrepancy are changeless over clip and the value of the covariance between the two clip periods depends merely on the distance or spread or slowdown between the two clip periods and non the existent clip at which the covariance is computed ( Gujarati, 2004 ) . Stationary series exhibits mean reversion in that it fluctuates around a changeless long-term mean and has a finite discrepancy that is clip invariant. On the other manus, a nonstationary series has no long tally mean to which series returns and the discrepancy is clip dependent and goes eternity as clip attacks eternity ( Enders, 2004 ) .

The most normally used conventional unit root trials in empirical surveies are ADF, PP and KPSS trials. However, the ADF, PP and KPSS trials have some restrictions and may take to a deceptive decision ( Silvia and Iqbal, 2011 ) .

Ng and Perron ( 2001 ) introduced four unit root trial statistics that are calculated utilizing generalized least squares ( GLS ) de-trended informations for a clip series variable to cover with the restraints of ADF and PP unit root trials. Compared to widely used Dickey Fuller ( DF ) and Philips Perron ( PP ) unit root trials, that has better power and size belongingss ( Wickremasinghe, 2004 ) .

The chief criticismA ofA proving PPP is based on the lack of conventional unit root trials that they are assumed that impermanent dazes have no long-term consequence on a variable. It is widely recognized that the outliers and interruptions in informations may take down the power of unit root trials and lead toward over-acceptance of the unit-root hypothesis ( Zhou and Kutan, 2011 ) . EmployingA unit root testsA withA structural interruptions for RER is one manner ofA deciding conventional unit rootA testsA failure. In that regard, we employ Zivot-Andrews ( 1992 ) trial with one-break and Lee and Strazicich ( 2003 ) trials with two-breaks for the being of unit roots and place the order of integrating for each variable.

The endogenous structural interruption trial of Zivot and Andrews ( 1992 ) is a consecutive trial which utilizes the full sample and uses a different silent person variable for each possible interruption day of the month ( Byrne and Perman, 2006 ) . They consider all three possible ways that a structural interruption can look in a clip series as Perron 1989. Zivot and Andrews ( 1992 ) consider three different theoretical accounts ( Models A, B and C ) for the unit root trial equation. Null hypothesis of unit root for these three theoretical accounts is the same and shown in equation 3.

( 3 )

Alternate hypothesis against void hypothesis of unit root is shown for every theoretical account in equation 4-6.

Model A

( 4 )

Model B

( 5 )

Model C

( 6 )

where is dummy variable and shows interruption in changeless and =1 if I, 0 otherwise ; is dummy variable and shows interruption in tendency and = if and 0 otherwise. Besides ; shows interruption clip and, shows interruption point.

The critical values in Zivot and Andrews ( 1992 ) are different to the critical values in Perron ( 1989 ) . The difference is due to that the selecting of the clip of the interruption is treated as the result of an appraisal process, instead than predetermined exogenously ( Glynn, Perera and Verma, 2007 ) . Harmonizing to Zivot-Andrews ( 1992 ) trials, the void hypothesis shows that the series have unit root. If the deliberate T statistics for variables are greater than the critical values in their degree signifiers, we reject void hypothesis of unit root and we say the variables are tendency stationary. Otherwise we could n’t reject void hypothesis of unit root.

Lee and Strazicich ‘s ( 2003 ) theoretical account allows for two endogenous interruptions both under the nothing and the alternate hypothesis. They show that the two-break LM unit root trial statistic which is estimated by the arrested development harmonizing to the LM rule will non spuriously reject the void hypothesis of a unit root. ( Glynn, Perera and Verma, 2007 ) .

Unlike from Zivot and Andrews ( 1992 ) , Lee-Strazicich ( 2003 ) trial uses Model A and Model C. For Lee-Strazicich ( 2003 ) trial, Model A shows two interruptions in changeless and is shown in equation 7.

,

( 7 )

Where is exogenic variables vector and demo break day of the month. if and 0 otherwise.

Model C shows two interruptions in changeless and tendency and shown in equation 8 ;

,

( 8 )

In equation ( 8 ) if and 0 otherwise ( Lee and Strazicich, 2003 ) . Critical values are computed by Lee-Strazicich ( 2003 ) .

III. EMPRICAL RESULTS

The consequences of conventional stationary trials are presented in Table 1.

Table 1: Conventional Unit Root Test Results

ADF Test Results

RERUS

-3.655*

REREU

-3.182**

ADF critical values for RERUS and REREU

% 1=-3.490 and % 5=-2.888

PP Test Results

RERUS

-3.375**

REREU

-2.911**

PP critical values for RERUS and REREU

% 1=-3.489 and % 5=-2.887

KPSS Test Results

RERUS

1.004

I”RERUS

0.306*

REREU

0.799

I”REREU

0.096*

KPSS critical values for RERUS and REREU

% 1=0.739 and % 5=0.463

KPSS critical values for I”RERUS and I”REREU

% 1=0.739 and % 5=0.463

Ng-Perron Test Results

Mutual savings bank

MPT

RERUS

-0.200

-0.165

0.824

38.352

I”RERUS

-25.979

-3.568

0.137

1.063

REREU

-1.523

-0.706

0.463

12.917

I”REREU

-50.445

-5.021

0.100

0.489

Ng-Peron critical values for RERUS, REREU, I”RERUS and I”REREU ; MZa, MZt, MSB, MPT severally ;

% 1 significance degree -13.800, -2.580, 0.174 and 1.780

% 5 significance degree for -8.100, -1.980, 0.233 and 3.170

* denote % 1 significance degree, ** denote % 5 significance degree

Harmonizing to Table 1,

For ADF and PP trials, the void hypothesis suggests that the series include unit root. The deliberate T statistics for RERUS and REREU are greater than the critical values in its degree signifiers for ADF and PP. Thus, the void hypothesis can be rejected, proposing that RERUS and REREU are stationary in the degree signifier harmonizing to both ADF ( at % 1 significance degree for RERUS and % 5 significance degree for REREU ) and PP trials ( at % 5 significance degree for both series ) .

For KPSS trial, the void hypothesis shows that investigated series is stationary. The deliberate T statistics for RERUS and REREU are greater than the critical values in their degree signifiers and the consequences of the first differenced RERUS and REREU variables are less than critical values at 1 % degrees. KPSS trial consequences suggest that RERUS and REREU series are I ( 1 ) .

For Ng-Peron trial, harmonizing to, tests the void hypothesis shows that the series have unit root and harmonizing to MSB and MPT tests the void hypothesis shows that the series are stationary. For, trials, the deliberate T statistics RERUS and REREU are less and for MSB and MPT tests the deliberate T statistics for RERUS and REREU are greater than the critical values proposing that RERUS and REREU are non-stationary in their degree signifiers. For the first difference of series, harmonizing to, trials, the deliberate T statistics for RERUS and REREU are greater and for MSB and MPT tests the deliberate T statistics for RERUS and REREU are less than the critical values at 1 % degrees, proposing that RERUS and REREU become stationary after differencing so that the series are I ( 1 ) harmonizing to Ng-Peron trials.

The consequences of unit root trials with structural interruptions are presented in Table 2.

Table 2: Unit of measurement Root Tests with Structural Breaks

Zivot-Andrews ( 1992 ) Trial

Degree

First Difference

Model A

Model C

Model A

Model C

RERUS

-4.44

-4.41

-6.56*

-6.52*

REREU

-4.43

-4.42

-6.19*

-6.46*

Critical values ( % 5 )

-4.80

-5.08

-4.80

-5.08

Lee-Strazitch ( 2003 ) Trial

Degree

First Difference

Model A

Model C

Model A

Model C

RERUS

-2.89

-5.18

-6.28*

-8.94*

REREU

-3.59

-4.67

-8.43*

-8.57*

Critical values ( % 5 )

-3.84

-5.71

-3.84

-5.71

Harmonizing to both Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) trials, the void hypothesis shows that the series have unit root. For both Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) trials, the deliberate T statistics for RERUS and REREU variables are less than the critical values in their degree signifiers and greater than the critical values in their first difference at 5 % significance degrees. Furthermore both Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) tests suggest that RERUS and REREU variables become stationary after differencing.

In sum-up ; harmonizing to ADF and PP trials, RERUS and REREU series are stationary in their degree signifiers and harmonizing to KPSS trial, Ng-Perron trials and unit root trials with structural interruptions, RERUS and REREU series become stationary after differencing so that RERUS and REREU series are I ( 1 ) .

IV. Decision

This survey aims to look into the issue of whether existent exchange rates are mean-reverting in Turkey. During the sample covered by this survey ( 2003-2012 ) , the Turkish economic system has sustained stable economic growing in combination with prudent financial direction, higher volume of international fiscal minutess with flexible exchange rates and lower rising prices degree. Better macroeconomic indexs have accelerated the procedure of integrating into the universe economic system, while transforming the state into one of the major receivers of capital flows in its part.

But, capital influxs may drive bad growing, inflationary force per unit areas and existent exchange rate grasp. During the recent drifting exchange-rate period, capital flows, equilibrium existent exchange rate and current history shortages are of particular involvement for Turkey and this type of issues leads to analyze the cogency of the PPP in Turkey.

The Purchasing Power Parity ” theory has received particular accent by economic experts in the unfastened economic system and macroeconomic theoretical accounts. This theory briefly states one of the oldest and most reputed relationships between exchange rate and monetary values. The theory of PPP is used to find the general tendency of exchange rates and monetary values and shows the transmittal mechanisms through which domestic monetary values alteration besides likely consequence the exchange rates and frailty versa. PPP besides plays an of import function in calculating existent exchange rates on the long-run footing. While it is accepted that PPP fails to keep in the short tally in empirical perusal, long tally PPP is still under probe ( see, Froot and Rogoff, 1995, Rogoff, 1996 ) .

Testing of stationary ( or mean-reverting belongings ) of existent exchange rates is one of the most popular methods to look into the cogency of long tally PPP. Normally, if the unit root ( non-stationary ) premise for existent exchange rates can be rejected, so the grounds is in favour of PPP.

In this paper, we investigated the cogency of PPP for Turkey for 2003M1-2012M6 periods by using both conventional unit root trials including the ADF, PP, KPSS and Ng-Perron and unit roots trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) for RERUS and REREU series. Our consequences show that, RERUS and REREU series become stationary after differencing harmonizing to all unit root trials except ADF and PP so that RERUS and REREU series are I ( 1 ) . These findings indicate that PPP does non keep in the long tally for Turkey in our analysis. It means that nominal dazes have lasting effects on existent exchange rates in the long tally. If PPP is improbable to keep, there is rather strong empirical grounds that pass-through from exchange rates to monetary values is uncomplete ( see, Osbat, 2005:3 ) . Therefore, the grounds besides shows that comparative monetary values does non set when there are strong depreciations or grasps of the nominal exchange rate, so we may reason that A consumerA pricesA areA notA much affected byA nominal exchange rates. The effectA ofA changesA inA exchange rates depends on the degreeA ofA exchange-rate pass-through ( ERPT ) . A low ERPT implies that authorities need non worry about monetary value instability or rising prices when seting exchange rate policy ( Jin, 2012:135 ) .

Therefore, this consequence besides can be interpreted that Central BankA of the Republic ofA Turkey can supply greater freedom for prosecuting an independent pecuniary policy and to do it easier to implement ini¬‚ation aiming.

Keywords: Real exchange rate, Buying Power Parity, unit root trials, structural interruptions, Turkey

Jel Code: F31, C22

Introduction

The finding of equilibrium existent exchange rates is one of the cardinal economic constructs in an international economic system particularly during economic crisis periods. There are a figure of different methodological analysiss[ 3 ]to measure equilibrium exchange rates. And the hypothesis of buying power para ( PPP ) is one of the most studied methodological analysiss in finding existent exchange rates in economic literature. Although the subject of PPP ” was foremost used eight-five old ages ago by the Swedish economic expert Gustav Cassel ( Cassel, 1918 ) , it has received much great attractive force by economic experts since so. Theoretically, PPP has been accepted as the rising prices theory of exchange rate and besides is the get downing point of the exchange rates theories. It by and large states that the exchange rate alteration between two currencies over a period of clip is determined by the alteration in the two states ‘ comparative monetary value degrees ( Dornbusch, 1985:1 ) .

One manner to analyze the long-term cogency of PPP is to look into the stationarity belongingss of existent exchange rates. A necessary status for long-term PPP is that the existent exchange rate should be average backsliding Besides, predictability of existent exchange rates based on average reversion conditions provide empirical support for PPP. For illustration, Adler and Lehmann ( 1983 ) , Darby ( 1980 ) , Hakkio ( 1986 ) , and Roll ( 1979 ) note that the existent exchange rate can or should follow a random walk, theoretically[ 4 ].

ThereA exist twoA types of PPP which areA absolute and relative.A Harmonizing to Absolute PPP, exchange rates and monetary values should set such that goods in different states have the same monetary value when expressed in the same currency.

Relative PPP provinces that the per centum depreciation in the exchange rate between two currencies over any period equals the difference between rising prices rates in the place state and in the foreign state.

Relative PPP does non to keep in the short tally and in the intermediate tally. Failures of the short tally are due to monetary value rigidnesss, merchandise distinction and local production cost barriers to international trade, dii¬ˆerencies in gustatory sensations and A consumer behaviorA and monetary value index acrossA countries. , little grade of mobility of goods and engineering.

However, in the long tally, it is expected to more support forA the cogency of comparative PPP that some of those ineffciencies will vanish, connoting a convergence in the monetary value degrees across states.

Palatopharyngoplasty can besides be used to state if a currency is over or under-valued.A It measures the exchance misalignment by agencies of divergences of the nominal exchange rate from the PPP para with regard to a basal state.

Morever, monetary value stableness and the grade of capital flows are two of import factors that may ai¬ˆect the cogency of long-term PPP. Fluctuations inA existent exchangeA are affected chiefly by the motions A the relativeA priceA of tradable goods ( pass-through affect ) . And besides, the stableness of exchange rate and capital flows is of great concern for inflation-targeting states. Cardinal Bankss in these states use involvement ratesA to command rising prices that may take to

greater volatility in exchange rate.

The relationship between the grade of capital flows and the existent exchange rate volatility are besides indispensable factor in explicating the long tally behavior of existent exchange rate. As the states became more incorporate with the the rise of grade of capital flows to the remainder of the universe, it is impossible for the cardinal bank to accomplish free capital mobility, an independent pecuniary policy and fixed exchange rate at the same time, known as impossible three.

The intent of this paper is to look into long-term buying power para in Turkey for monthly informations from 2003:1 to 2012:07. The selected survey periods and state will give A significantA sum for following emprical perusals. Because, inflationary targating government has been A introduced and there have been a crisp addition in the capital flows in Turkey during the analyzing periods.

For this purpose, we employed unit root trials to analyze the the hypothesis of PPP for Turkey through conventional unit root trials including Augmented Dickey-Fuller and Phillips-Perron. However, both unit root trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee and Strazicich ( 2003 ) trials.

The paper is organized as follows. Section II briefly introduces informations and methodological analysis. Section III presents empirical consequences and subdivision IV concludes.

Literature Review

Several surveies re-examined the cogency of long-runA purchasingA powerA parityA ( PPP ) hypothesis for TurkeyA utilizing different technique.

By utilizing co-integration techniques, Telatar and KazdaglA± ( 1998 ) found noA grounds toA support the PPP hypothesis for Turkey and A chief trading spouses ( France, Germany, the UK, and the USA ) .

Sarno ( 2000 ) re-examineA thisA hypothesisA utilizing the exponential smooth passage autoregressive theoretical account ( ESTAR ) for Turkey and its major trading spouses ( the US, the UK, Germany and France ) and supported the cogency of long-term PPP.

The cogency of theA long-term PPP hypothesis is besides supported by Yazgan ( 2003 ) . He performed his analysis with standard multivariate co-integration techniques.

Erlat ( 2003 ) found grounds of buying power para by utilizing unit root trials and autoregressive fractionally incorporate traveling norm ( ARFIMA ) theoretical accounts. Erlat ( 2004 ) besides investigated PPP for the period 1984:1-2000:9 and found strong grounds of nonlinear stationary.

TaAYtan ( 2005 ) accepted the long-term buying power para ( PPP ) hypothesis holds for the UK and US $ based existent exchange rates for Turkish economic system during the period 1982M1-2003M12.A

Kalyoncu ( 2009 ) tested the cogency of buying power para ( PPP ) for Turkey and its trading spouses ( USA, Germany, Japan, France, Netherlands and UK ) . Harmonizing to trial consequences, PPP testing is sensitive to the pick of the basal state and can be influenced by the type of trial.

Kum ( 2012 ) examines the cogency of the buying power para ( PPP ) in Turkey for one-year informations from 1953 to 2009. He found the evidience of PPP for Turkey by utilizing Zivot and Andrews and Lagrange Multiplier unit root trials.

II. DATA AND METHODOLOGY

Formalized by Cassel ( 1918 ) , the existent Exchange ( RER ) rate can be defined in equation 1:

( 1 )

where E denotes the nominal exc hange rate ( measured as domestic currency monetary value of foreign currency ) , P* is an index of the foreign monetary value degree, and P is an index of the domestic monetary value degree.

Denoting logarithms in lower instance letters, we therefore have equation 2.

( 2 )

R defines logarithmic signifier of existent exchange and p* and p denote the logarithms of the domestic and foreign state monetary value index severally.

The cogency of long-term PPP requires the stationarity of existent exchange rates. If R is stationary so divergences from PPP are impermanent and will vanish with clip, and therefore PPP is likely to keep in the long tally. But if R, has a unit-root, so it implies that divergences from the para are cumulative and non finally self-reverting.

The existent exchange rate ( RER ) is calculated from the nominal effectual exchange rate ( NER ) and a step of the comparative monetary value or cost between the state under survey and its trading spouses. The most popular monetary value and costs steps are consumer monetary values ( CPI ) , manufacturer monetary values ( PPI ) , GDP deflator, unit labour costs ( ULC )[ 5 ].

In this survey, we calculate CPI-based RER for USA and the Euro Area. We employ monthly existent exchange rate informations between Turkish Lira and USA Dollar and Euro for the period 2003M1-2012M6. Nominal US dollar and Euro exchange rates and Turkish CPI series were obtained from Central Bank of Turkey, Electronic Data Delivery System. US CPI was obtained from US Department of Labor, Bureau of Labor Statistics and European CPI was obtained from Eurostat. Real exchange rate series for US Dollar and Euro are denoted as RERUS and REREU severally.

This survey aims to look into stationarity belongingss of the existent exchange rate of Turkey using both conventional unit root trials and unit root trials with structural interruptions. In order to look into stationary belongingss of the RER series, we employed conventional unit root trials including ADF ( Dickey and Fuller, 1981 ) , PP ( Phillips-Perron, 1987 ) , KPSS ( Kwiatkowski, Phillips, Schmidt, and Shin, 1992 ) and Ng-Perron trials ( 2001 ) and unit root trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee and Strazicich ( 2003 ) trials.

Empirical plants based on clip series informations assume that the implicit in clip series is stationary. A stochastic procedure is said to be stationary if its mean and discrepancy are changeless over clip and the value of the covariance between the two clip periods depends merely on the distance or spread or slowdown between the two clip periods and non the existent clip at which the covariance is computed ( Gujarati, 2004 ) . Stationary series exhibits mean reversion in that it fluctuates around a changeless long-term mean and has a finite discrepancy that is clip invariant. On the other manus, a nonstationary series has no long tally mean to which series returns and the discrepancy is clip dependent and goes eternity as clip attacks eternity ( Enders, 2004 ) .

The most normally used conventional unit root trials in empirical surveies are ADF, PP and KPSS trials. However, the ADF, PP and KPSS trials have some restrictions and may take to a deceptive decision ( Silvia and Iqbal, 2011 ) .

Ng and Perron ( 2001 ) introduced four unit root trial statistics that are calculated utilizing generalized least squares ( GLS ) de-trended informations for a clip series variable to cover with the restraints of ADF and PP unit root trials. Compared to widely used Dickey Fuller ( DF ) and Philips Perron ( PP ) unit root trials, that has better power and size belongingss ( Wickremasinghe, 2004 ) .

The chief criticismA ofA proving PPP is based on the lack of conventional unit root trials that they are assumed that impermanent dazes have no long-term consequence on a variable. It is widely recognized that the outliers and interruptions in informations may take down the power of unit root trials and lead toward over-acceptance of the unit-root hypothesis ( Zhou and Kutan, 2011 ) . EmployingA unit root testsA withA structural interruptions for RER is one manner ofA deciding conventional unit rootA testsA failure. In that regard, we employ Zivot-Andrews ( 1992 ) trial with one-break and Lee and Strazicich ( 2003 ) trials with two-breaks for the being of unit roots and place the order of integrating for each variable.

The endogenous structural interruption trial of Zivot and Andrews ( 1992 ) is a consecutive trial which utilizes the full sample and uses a different silent person variable for each possible interruption day of the month ( Byrne and Perman, 2006 ) . They consider all three possible ways that a structural interruption can look in a clip series as Perron 1989. Zivot and Andrews ( 1992 ) consider three different theoretical accounts ( Models A, B and C ) for the unit root trial equation. Null hypothesis of unit root for these three theoretical accounts is the same and shown in equation 3.

( 3 )

Alternate hypothesis against void hypothesis of unit root is shown for every theoretical account in equation 4-6.

Model A

( 4 )

Model B

( 5 )

Model C

( 6 )

where is dummy variable and shows interruption in changeless and =1 if I, 0 otherwise ; is dummy variable and shows interruption in tendency and = if and 0 otherwise. Besides ; shows interruption clip and, shows interruption point.

The critical values in Zivot and Andrews ( 1992 ) are different to the critical values in Perron ( 1989 ) . The difference is due to that the selecting of the clip of the interruption is treated as the result of an appraisal process, instead than predetermined exogenously ( Glynn, Perera and Verma, 2007 ) . Harmonizing to Zivot-Andrews ( 1992 ) trials, the void hypothesis shows that the series have unit root. If the deliberate T statistics for variables are greater than the critical values in their degree signifiers, we reject void hypothesis of unit root and we say the variables are tendency stationary. Otherwise we could n’t reject void hypothesis of unit root.

Lee and Strazicich ‘s ( 2003 ) theoretical account allows for two endogenous interruptions both under the nothing and the alternate hypothesis. They show that the two-break LM unit root trial statistic which is estimated by the arrested development harmonizing to the LM rule will non spuriously reject the void hypothesis of a unit root. ( Glynn, Perera and Verma, 2007 ) .

Unlike from Zivot and Andrews ( 1992 ) , Lee-Strazicich ( 2003 ) trial uses Model A and Model C. For Lee-Strazicich ( 2003 ) trial, Model A shows two interruptions in changeless and is shown in equation 7.

,

( 7 )

Where is exogenic variables vector and demo break day of the month. if and 0 otherwise.

Model C shows two interruptions in changeless and tendency and shown in equation 8 ;

,

( 8 )

In equation ( 8 ) if and 0 otherwise ( Lee and Strazicich, 2003 ) . Critical values are computed by Lee-Strazicich ( 2003 ) .

III. EMPRICAL RESULTS

The consequences of conventional stationary trials are presented in Table 1.

Table 1: Conventional Unit Root Test Results

ADF Test Results

RERUS

-3.655*

REREU

-3.182**

ADF critical values for RERUS and REREU

% 1=-3.490 and % 5=-2.888

PP Test Results

RERUS

-3.375**

REREU

-2.911**

PP critical values for RERUS and REREU

% 1=-3.489 and % 5=-2.887

KPSS Test Results

RERUS

1.004

I”RERUS

0.306*

REREU

0.799

I”REREU

0.096*

KPSS critical values for RERUS and REREU

% 1=0.739 and % 5=0.463

KPSS critical values for I”RERUS and I”REREU

% 1=0.739 and % 5=0.463

Ng-Perron Test Results

Mutual savings bank

MPT

RERUS

-0.200

-0.165

0.824

38.352

I”RERUS

-25.979

-3.568

0.137

1.063

REREU

-1.523

-0.706

0.463

12.917

I”REREU

-50.445

-5.021

0.100

0.489

Ng-Peron critical values for RERUS, REREU, I”RERUS and I”REREU ; MZa, MZt, MSB, MPT severally ;

% 1 significance degree -13.800, -2.580, 0.174 and 1.780

% 5 significance degree for -8.100, -1.980, 0.233 and 3.170

* denote % 1 significance degree, ** denote % 5 significance degree

Harmonizing to Table 1,

For ADF and PP trials, the void hypothesis suggests that the series include unit root. The deliberate T statistics for RERUS and REREU are greater than the critical values in its degree signifiers for ADF and PP. Thus, the void hypothesis can be rejected, proposing that RERUS and REREU are stationary in the degree signifier harmonizing to both ADF ( at % 1 significance degree for RERUS and % 5 significance degree for REREU ) and PP trials ( at % 5 significance degree for both series ) .

For KPSS trial, the void hypothesis shows that investigated series is stationary. The deliberate T statistics for RERUS and REREU are greater than the critical values in their degree signifiers and the consequences of the first differenced RERUS and REREU variables are less than critical values at 1 % degrees. KPSS trial consequences suggest that RERUS and REREU series are I ( 1 ) .

For Ng-Peron trial, harmonizing to, tests the void hypothesis shows that the series have unit root and harmonizing to MSB and MPT tests the void hypothesis shows that the series are stationary. For, trials, the deliberate T statistics RERUS and REREU are less and for MSB and MPT tests the deliberate T statistics for RERUS and REREU are greater than the critical values proposing that RERUS and REREU are non-stationary in their degree signifiers. For the first difference of series, harmonizing to, trials, the deliberate T statistics for RERUS and REREU are greater and for MSB and MPT tests the deliberate T statistics for RERUS and REREU are less than the critical values at 1 % degrees, proposing that RERUS and REREU become stationary after differencing so that the series are I ( 1 ) harmonizing to Ng-Peron trials.

The consequences of unit root trials with structural interruptions are presented in Table 2.

Table 2: Unit of measurement Root Tests with Structural Breaks

Zivot-Andrews ( 1992 ) Trial

Degree

First Difference

Model A

Model C

Model A

Model C

RERUS

-4.44

-4.41

-6.56*

-6.52*

REREU

-4.43

-4.42

-6.19*

-6.46*

Critical values ( % 5 )

-4.80

-5.08

-4.80

-5.08

Lee-Strazitch ( 2003 ) Trial

Degree

First Difference

Model A

Model C

Model A

Model C

RERUS

-2.89

-5.18

-6.28*

-8.94*

REREU

-3.59

-4.67

-8.43*

-8.57*

Critical values ( % 5 )

-3.84

-5.71

-3.84

-5.71

Harmonizing to both Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) trials, the void hypothesis shows that the series have unit root. For both Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) trials, the deliberate T statistics for RERUS and REREU variables are less than the critical values in their degree signifiers and greater than the critical values in their first difference at 5 % significance degrees. Furthermore both Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) tests suggest that RERUS and REREU variables become stationary after differencing.

In sum-up ; harmonizing to ADF and PP trials, RERUS and REREU series are stationary in their degree signifiers and harmonizing to KPSS trial, Ng-Perron trials and unit root trials with structural interruptions, RERUS and REREU series become stationary after differencing so that RERUS and REREU series are I ( 1 ) .

IV. Decision

This survey aims to look into the issue of whether existent exchange rates are mean-reverting in Turkey. During the sample covered by this survey ( 2003-2012 ) , the Turkish economic system has sustained stable economic growing in combination with prudent financial direction, higher volume of international fiscal minutess with flexible exchange rates and lower rising prices degree. Better macroeconomic indexs have accelerated the procedure of integrating into the universe economic system, while transforming the state into one of the major receivers of capital flows in its part.

But, capital influxs may drive bad growing, inflationary force per unit areas and existent exchange rate grasp. During the recent drifting exchange-rate period, capital flows, equilibrium existent exchange rate and current history shortages are of particular involvement for Turkey and this type of issues leads to analyze the cogency of the PPP in Turkey.

The Purchasing Power Parity ” theory has received particular accent by economic experts in the unfastened economic system and macroeconomic theoretical accounts. This theory briefly states one of the oldest and most reputed relationships between exchange rate and monetary values. The theory of PPP is used to find the general tendency of exchange rates and monetary values and shows the transmittal mechanisms through which domestic monetary values alteration besides likely consequence the exchange rates and frailty versa. PPP besides plays an of import function in calculating existent exchange rates on the long-run footing. While it is accepted that PPP fails to keep in the short tally in empirical perusal, long tally PPP is still under probe ( see, Froot and Rogoff, 1995, Rogoff, 1996 ) .

Testing of stationary ( or mean-reverting belongings ) of existent exchange rates is one of the most popular methods to look into the cogency of long tally PPP. Normally, if the unit root ( non-stationary ) premise for existent exchange rates can be rejected, so the grounds is in favour of PPP.

In this paper, we investigated the cogency of PPP for Turkey for 2003M1-2012M6 periods by using both conventional unit root trials including the ADF, PP, KPSS and Ng-Perron and unit roots trials with structural interruptions including Zivot-Andrews ( 1992 ) and Lee-Strazicich ( 2003 ) for RERUS and REREU series. Our consequences show that, RERUS and REREU series become stationary after differencing harmonizing to all unit root trials except ADF and PP so that RERUS and REREU series are I ( 1 ) . These findings indicate that PPP does non keep in the long tally for Turkey in our analysis. It means that nominal dazes have lasting effects on existent exchange rates in the long tally. If PPP is improbable to keep, there is rather strong empirical grounds that pass-through from exchange rates to monetary values is uncomplete ( see, Osbat, 2005:3 ) . Therefore, the grounds besides shows that comparative monetary values does non set when there are strong depreciations or grasps of the nominal exchange rate, so we may reason that A consumerA pricesA areA notA much affected byA nominal exchange rates. The effectA ofA changesA inA exchange rates depends on the degreeA ofA exchange-rate pass-through ( ERPT ) . A low ERPT implies that authorities need non worry about monetary value instability or rising prices when seting exchange rate policy ( Jin, 2012:135 ) .

Therefore, this consequence besides can be interpreted that Central BankA of the Republic ofA Turkey can supply greater freedom for prosecuting an independent pecuniary policy and to do it easier to implement ini¬‚ation aiming.