The effects of real exchange rate volatility on sectoral export flows under intermediate and flexible exchange rate regimes : Empirical evidence from Turkey

The aim of this paper is to analyse empirically the effects of real exchange rate volatility on sectoral exports in Turkey under intermediate and flexible exchange rate regimes. The cointegration test and error correction models are used to test the long-run relationship and short-run effects, respectively. The estimation results show that the real exchange rate volatility has negative and significant effects on sectoral exports in both intermediate and flexible exchange rate regimes. These empirical results are consistent with the theory. However, the impact of real exchange rate and foreign income appeared to be quite different for the two exchange rate regimes. Further, research is required to analyse the impacts of real exchange rate and foreign income on sectoral exports.


Introduction
The adoption of the flexible exchange rate regime in 1973 by many developed and emerging market economies put into agenda the exchange rate volatility or uncertainty and its effects on international trade flows.It is argued that exchange rate volatility creates an uncertain environment for international trade flows and this may reduce trade flows.Both theoretical and empirical literature gives mixed results about the effects of exchange rate volatility on international trade flows.While some of the studies find negative effects of exchange rate volatility on international trade flows, some of them find positive or statistically insignificant effects of exchange rate volatility on international trade flows.In this study, the effects of real exchange rate volatility on Turkish sectoral export flows with the rest of the world under intermediate and flexible exchange rate regimes are analysed.In other words, we investigate long-run and short-run effects of real exchange rate volatility on Turkish sectoral export flows under intermediate and flexible exchange rate regimes.The rest of this study is organised as follows: In Section 2, a brief literature review is presented.In Section 3, theoretical framework of the study is explained.In Section 4, variable definitions and data sources are explained.In Section 5, empirical result are presented.Section 6 concludes the paper.

A brief literature reviews
There are both theoretical and empirical studies about the effects of exchange rate volatility on international trade flows.However, both theoretical and empirical studies give mixed results about the effects of volatility of exchange rates on international trade flows.While some of the studies find negative effects of exchange rate volatility on international trade flows, some of them find positive or statistically insignificant effects of exchange rate volatility on trade flows.McKenzie (1999) and Auboin and Ruta (2013) give theoretical and empirical literature surveys about exchange rate volatility and international trade flows.The empirical results that examine the relationship between exchange rate volatility and international trade flows in Turkey are few.Vergil (2002) found a negative relationship between real exchange rate volatility and export flows in Turkey.Kasman and Kasman (2005) found a positive relationship between exchange rate volatility and export flows in Turkey.
The existing empirical studies about Turkey as well as other countries used aggregate trade flows of countries with the rest of the world or with their major trading partners.However, the current debate about this issue is that sect oral data can be helpful to distangle the linkages between the exchange rate volatility and trade flows that may exist across sectors but not in total trade flows (Auboin & Ruta, 2013;Bahmani-Oskooee & Durmaz, 2016).In this framework, Caglayan and Di (2010) investigated empirically the effects of real exchange rate volatility on sectoral bilateral trade flows between the United States and its top 13 trading partners.They found little effect of exchange rate volatility on sectoral trade flows of advanced and emerging economies.Bahmani-Oskooee, Hegerty and Satawatananon (2015) examined the effect of exchange rate risk on Japan-Thailand trade using data from 117 Japanese exporting and 54 importing industries.They found that in the short run, slightly more than half of 117 exporting industries and 54 importing industries are affected by exchange rate volatility.In the long-run, 6 exporting and 2 importing industries are affected positively and 22 exporting and 9 importing industries are affected negatively.Besides, they also found the evidence that small exporting industries and exports of manufacturing and certain machinery and transport equipment industries might be relatively more affected by exchange rate risk.
Regarding Turkey, Caglayan, Dahi and Demir (2013) examined the effects of exchange rate uncertainty on manufacturing goods exports of 28 emerging economies, including Turkey.They find that exchange rate uncertainty affects trade flows of 24 of the 28 emerging economies, including Turkey.Bahmani-Oskee and Durmaz (2016) investigated the short-run and long-run effects of exchange rate volatility on exports of 23 industries and imports of 39 industries including one-digit and two-digit industries in Turkey.They found significant short-run effects of exchange rate volatility in many industries, but its long-run effects are significant on 24 Turkish importing industries and 12 Turkish exporting industries.

Theoretical framework
The traditional long-run export demand function is as follows: where EXPt is the volume of a country's real export goods at time t, FXt is the bilateral real exchange rate at time t, Yt is the real foreign economic activity at time t, SDVFXt is the standard deviation of real exhange rate that measures exchange rate volatility and as proxy to risk at time t and ut is the error term.The expected signs of the coefficients are as follows: B0 = The sign of the coefficient is expected to be positive.An increase in exchange rate shows depreciation of domestic currency and export volume should increase.B1 = The sign of the coefficient is expected to be positive.An increase in foreign income or foreign demand should increase export volume.B2 = The sign of the coefficient is expected to be negative.An increase in exchange rate volatility should decrease export volume.But, empirical studies give ambiguous results about the sign of the coefficient.So, this may be an empirical issue.

Variable definitions and data sources
In the empirical part of the study, the effects of real exchange rate volatility on sectoral export data are examined for Turkey under intermediate (January 1991-February 2001) and flexible exchange rate (March 2001-June 2013) regimes.To do that, firstly, the augmented dickey fuller (ADF) test is done if the variables have a unit root.Then, cointegration analysis is conducted and error correction models are estimated.The following export demand equation is estimated for the intermediate and flexible exchange rate regimes: where EXPt is the real sectoral export of Turkey with the rest of the world, FXt is the real exchange rate, the amount of Turkish lira per unit of U.S. dollar, Yt is the real foreign income or foreign demand, SDVFXt is the standard deviation of real exchange rate.All the variables are in the logarithmic forms.
The variables are constructed as follows: EXPt Since, there are not fully time-series data for Sector 9 (i.e., commodities and transactions), it is not included in the empirical part of the study.

Procedure of study and empirical results
Firstly, each of the variable was tested using ADF test whether the variable has a unit root.The ADF test consists of regressing each series on its lagged value and lagged difference terms.The ADF test results are shown in Table 1.The results suggest that all variables used in export demand function are nonstationary in their levels and they are integrated of order one.Therefore, their first differences are used in the estimation of regressions.In order to analyse the long-run and short-run effects of real exchange rate volatility on sectoral trade flows, cointegration analysis and error correction models are used.

Cointegration analysis
The Johansen's test statistics (trace and maximum eigenvalue) are used.The cointegration test results for real export volume, real exchange rate, real foreign income, standard deviation of real exchange rate for the periods of intermediate (January 1994-February 2001) and flexible exchange rate (March 2001-September 2012) regimes are presented in Tables 2 and 3, respectively.The existence of cointegration between variables means that there is a long-run equilibrium among real exports, real exchange rate, real exchange rate volatility and foreign income.The estimation of cointegrating relationship for intermediate and flexible exchange rate regimes are given in Tables 4 and 5, respectively.As can be seen in Table 4, the cointegration test results show that under the intermediate exchange rate regime, the signs of the explanatory variables are as expected as a whole.The sign of the real exchange rate volatility coefficient is negative and statistically significant as expected.The impacts of real exchange rate and foreign income are positive and significant.However, as can be seen in Table 5, the estimation results are not the same under the flexible exchange rate regime.The impact of exchange rate volatility is negative and statistically significant for total sector and all sub-sectors except Sector 1.However, the real exchange rate and foreign income appear to be statistically insignificant or negatively significant contrary to expected.The summary of the estimation of cointegrating relationships for intermediate and flexible exchange rate regimes are presented in Table 6.As can be seen in Table 6, under the flexible exchange rate regime, the volatility of real exchange rate is negative and statistically significant, except for beverages and tobacco (Sector 1).It is significantly negative also under the intermediate exchange rate regime; except, for crude materials, inedible, except fuels (Sector 2); animal and vegetable oils fats and waxes (Sector 4); machinery and transport equipment (Sector 7).However, the impacts of real exchange rate and foreign income or foreign demand seem to be quite different for two different exchange rate regimes.Under the intermediate exchange rate regime, the impact of real exchange rate appeared to be positive and statistically significant for almost all the subsectors except, mineral fuels, lubricants and related materials (Sector 3); animal and vegetable oils fats and waxes (Sector 4), which are negative and statistically significant.On the other hand, under the flexible exchange rate regime, the coefficient of real exchange rate is negative and statistically significant for total sector exports and five of the nine sub-sectors, that is, food and live animals (Sector 0); beverages and tobacco (Sector 1); mineral fuels, lubricants and related materials (Sector 3); manufactured goods (Sector 6); machinery and transport equipment (Sector 7); miscellaneous manufactured articles (Sector 8).The coefficient of real exchange rate is statistically insignificant for four sub-sectors, that is, food and live animals (Sector 0); crude materials, inedible, except fuels (Sector 2); animal and vegetable oils, fats and waxes (Sector 4); chemical and related products (Sector 5).
As can be seen in Graphs 1 and 2, the highest share of total exports belongs to manufactured goods (Sector 6); machinery and transport equipment (Sector 7) and miscellannous manufactured articles (Sector 8).As we give special attention to these sectors, we could see that real exchange rate is positive and significant for these sectors under intermediate exchange rate regime and negative and significant under flexible exchange rate regime.Kizildere, Kabadayi, and Emsen (2014) also found that the depreciation of Turkish lira decreased exports.Hepaktan, Cinar and Dundar (2011) also found weak effects of real exchange rate on exports.Similar estimation results also appeared for foreign income or foreign demand.Under the intermediate exchange rate regime, while it is positive and significant for total sector exports and four of the nine sub-sectors, that is, mineral fuels, lubricants and related materials (Sector 3); chemical and related products (Sector 5); manufactured goods (Sector 6); machinery and transport equipment (Sector 7).It is statistically insignificant for five sub-sectors, that is, food and live animals (Sector 0); beverages and tobacco (Sector 1); crude materials, inedible, except fuels (Sector 2); animal and vegetable oils, fats and waxes (Sector 4); miscellanous manufactured articles (Sector 8).
Under the flexible exchange rate regime, the coefficient of foreign income is only positive and significant for machinery and transport equipment (Sector 7).The coefficient of foreign income is negative and significant for food and live animals (Sector 0); crude materials, inedible, except fuels (Sector 2); chemical and related products (Sector 5).It is statistically insignificant for total sector exports; beverages and tobacco (Sector 0); mineral fuels, lubricants and related materials (Sector 3); animal and vegetable oils, fats and waxes (Sector 4); manufactured goods (Sector 6); miscellanous manufactured articles (Sector 8).

Error correction models (ECMs)
As a third step, ECMs are estimated.To do that, three-period lags of the independent variables are included in the regressions, and they are estimated for intermediate and flexible exchange rate regimes.Then, the statistically insignificant variables are dropped from the regressions and the statistically significant ones are kept in the regressions and they are reestimated.These estimation results are presented in Tables 7 and 8, respectively.The cointegration will be supported if ECMt−1 carries a negative and statistically significant coefficient.Besides, the coefficient of ECMt−1 represents the proportion of disequilibrium in long-run values in one period corrected in the next period.As can be seen in Tables 7 and 8, the ECMt−1 coefficients for all sectors have a negative sign and statististically significant, which confirm all the variables are cointegrated.The coefficients of ECMt−1t also show that about half of the deviations from the long-run values are corrected in the following period for the total sector and all other sub-sectors.The first difference of foreign income appeared to be positive and significant also for almost all sub-sectors.(1.73) (0.17) (−9.71)Note: '**' shows that the variable is significant at 5% level.'∆' shows the first difference of the variable.

Summary and conclusions
This paper analysed empirically the effects of real exchange rate volatility on sectoral level export data in Turkey under intermediate intermediate (January 1991-February 2001) and flexible exchange rate (March 2001-June 2013) regimes.The empirical findings show that real exchange rate volatility have negative and significant effects on sectoral level export data in both intermediate and flexible exchange rate regimes.However, the impact of the real exchange rate and foreign income must be scrutinised in more details.The model works for intermediate exchange rate regime, where as the results vary for the flexible exchange rate regimes.Further research is required for the sub-sectors.One point may be the dependency of a particular sector on import in terms of intermediate goods.A second point may be the competitiveness of Turkey in a particular sector.
(real sectoral exports) = Nominal sectoral exports/U.S. consumer price index (CPI) FXt (real exchange rate) = Nominal exchange rate* (U.S. CPI/Turkish CPI) Yt (foreign income or demand) = Organisation for Economic Cooperation and Development (OECD) index of industrial production as a proxy to foreign real GDP SDVFXt (real exchange rate volatility) = standard deviation of real exchange rate.One year's worth of monthly data were used.The data are monthly and data sources are as follows: the nominal exchange rate (i.e., Turkish lira per U.S. dollar, period average rate) and consumer price indices (2010 = 100) are taken from the International Monetary Fund's International Financial Statistics.The export data (millions, U.S. dollars) are taken from the OECD's Monthly Statistics of International Trade.The index of industrial production of OECD countries is used a proxy to foreign real GDP and are taken form the OECD's main economic indicators (2010 = 100).The sectors are classified by sections of Standard Industrial Trade Classification at one-digit level.The codes and names of sectors are given as follows: 05 level.(**) Trace test indicates 1 cointegrating equation at the.05 level.(***) denotes rejection of null hypothesis at the.05 level.(****) MacKinnon-Haug-Michelis (1999) p values.(*****) Trace test indicates two cointegrating equations at the.05 level.(******) Trace test indicates three cointegrating equations at the.05 level.(*******) Trace test indicates 4 cointegrating equations at the.05 level.

Table 1 . Augmented dickey fuller unit root test results for the intermediate and flexible exchange rate regimes (including intercept) Variable Intermediate exchange rate regime
Note: '**' shows the rejection of null hypothesis of a unit root at the 1% level and "*" shows the rejection of the null hypothesis at the 5% level.The McKinnon critical values for intermediate exchange rate regime period −3.48 at the 1% level and −2.88 for the 5% level.The McKinnon critical values for flexible exchange rate regime period −3.47 at the 1% level and −2.88 for the 5% level."∆" shows the first difference of the variable.

Table 3 . Cointegration test results for the flexible exchange rate regime Sector Eigenvalue Trace statistic 0.05 Critical calue
*) Trace test indicates no cointegrating equation at the.