Dynamic Optics in Economics: Quantitative, Experimental and Econometric Analyses

List of Authors

Akın Usupbeyli, Asst. Prof., PhD,

Ankara University

ORCID: 0000-0003-2792-8805

Ali İlhan, PhD,

Tekirdağ Namık Kemal University

ORCID: 0000-0001-6201-5353

Ayça Özkekin, Asst. Prof., PhD,

Bandırma Onyedi Eylül University

ORCID: 0000-0002-0566-7168

Aykut Karakaya, PhD,

Recep Tayyip Erdoğan University

ORCID: 0000-0001-6491-132X

Berna Tuncay, PhD,

Koç University

ORCID: 0000-0001-6398-1123

Bilgin Bari, Asst. Prof., PhD,

Anadolu University

ORCID: 0000-0001-7665-2740

Coşkun Akdeniz, PhD,

Tekirdağ Namık Kemal University

ORCID: 0000-0002-3973-754X

Erdal Alancıoğlu, PhD,

Harran University

ORCID: 0000-0002-5008-4957

Frank Windmeijer, Professorial Research Fellow,

University of Oxford

ORCID: 0000-0002-4232-2783

Hakan Ulucan, Asst. Prof., PhD,

Pamukkale University

ORCID: 0000-0002-1769-9407

Hale Kırer Silva Lecuna, Asst. Prof., PhD,

Bandırma Onyedi Eylül University

ORCID: 0000-0002-1879-3315

Melih Kutlu, PhD,

Samsun University

ORCID: 0000-0002-8634-6330

Merve Ertok Onurlu, Asst. Prof., PhD Çanakkale 18 Mart University

ORCID: 0000-0001-5187-1568

Metin Tetik, Asst. Prof. PhD,

Uşak University

ORCID: 0000-0003-2741-7175

Özlem Ayvaz Kızılgöl, Asst. Prof., PhD,

Bandırma Onyedi Eylül University

ORCID: 0000-0001-9996-7892

Özlem Kuvat, Asst. Prof., PhD,

Balıkesir University

ORCID: 0000-0001-7017-4557

Pınar Deniz, Asst. Prof., PhD,

Marmara University

ORCID: 0000-0001-5693-1144

Reşat Ceylan, Prof. Dr.,

Pamukkale University

ORCID: 0000-0003-3727-6644

Rüya Eser, PhD,

Mimar Sinan University

ORCID: 0000-0003-0763-5423

Chapter 1 A Nonlinear Approach for the Determinants of Exchange Rate Volatility: Evidence from Turkey

Abstract: Emerging markets are reflecting highly volatile exchange rates. Turkish economy, apart from being exposed to exchange rate fluctuations, is fragile due to its dependence on imports. The aim of this paper is to analyze the exchange rate fluctuations in Turkish economy under the following steps: (i) to run principal component-guided sparse regression (PC-LASSO) which is a method for large data to shrink predictions toward the most important variable for volatility out of many potential domestic and international macroeconomic indicators, (ii) to estimate a threshold level for the main determinant of the volatility and examine the impact of these macroeconomic indicators using structural threshold regression model. The empirical findings suggest the international reserves of the central bank as the most important variable for volatility and that the threshold regression reflects the nonlinearity of the model.

Keywords: Financial MarketsExchange RateInternational ReservesPC-LASSOThreshold Regression Models

1. Introduction

After the breakdown of Bretton Woods, countries adopted various exchange rate policies depending on their conditions. Exchange rate is a critical indicator for open economy transactions as import and export levels are dependent on their relevant prices in the international markets. Leaving all other links to a side, stable exchange rate is crucial to enhance trade volume which is – from the perspective of neoclassical theories – argued to be beneficial for consumers and producers together with a rise in the variety of goods and services and a trigger for innovation due to the rise in the level of competition. Prices of domestic products in the international market are theoretically one of the determinants of exports of a domestic country and prices of foreign products in a domestic market are one the determinants of imports the country. Hence, determination of the exchange rate is important to specify the trade balance, which is one of the determinants of the GDP growth as it is a component of GDP. When it comes to developing countries, adjustments in the exchange rate are way more critical due to their fragilities. Regarding the export levels, relatively lower value added ←11 | 12→and lack of competitiveness, and regarding the import levels, dependence on foreign products highlights the importance of developing an effective exchange regime. Emerging economies carry the vulnerabilities of developing countries however their growth potential is very high compared to all others. Relatively higher returns attract global capital searching for best options regarding risk-return rates. Thereby, these countries have become the short-term harbors of the global liquidity. During relatively higher risks or lower returns, capital outflows from one harbor and inflows to another one with lower risks and/or higher returns. Capital inflows bring in international currency triggering an appreciation whereas capital outflow lowers international currency triggering a depreciation of the local currency. Being open to short term capital in and outflows, emerging countries are the countries most exposed to exchange rate volatilities.

The determinants of exchange rate volatility are examined for Turkey using key indicators highlighted in the empirical and theoretical literature. The dataset covers the flexible exchange rate regime (de jure) period using monthly data between February 2006 and February 2020. All domestic variables are obtained from Central Bank of Republic of Turkey (CBRT) database. Dependent variable, exchange rate volatility is the standard deviation of daily nominal exchange rate which is measured as Turkish Lira per US Dollar. Domestic determinants are as follows: inflation rate (INF) is CPI annual inflation; domestic interest rate (INT) is the policy rate of the central bank; budget balance (BB) is in domestic currency; trade balance (TB) is in US dollar; financial openness (FO) is measured as the summation of net portfolio investment and net foreign direct investment (assets minus liabilities); industrial production index (IPI) is used for a monthly proxy for total production; foreign reserves (RES) is the foreign assets in US dollar terms. External determinants are as follows: oil price (OIL) is the daily average of closing spot Brent oil prices from US Energy Information Administration (EIA); federal funds rate (FFR) (effective) is from Federal Reserve Bank of Saint Louis (FRED); global economic policy uncertainty index¹ (EPU) is retrieved from and the “Economic Policy Uncertainty” webpage² which is based on PPP-adjusted GDP measured by Davis (2016). The regression for the determination of exchange rate volatility is represented with the following model specification:

where the is the standard deviation of the exchange rate as a proxy for volatility, is a vector of domestic monetary macroeconomic indicators (INF, INT, FO, RES), is a vector of real macroeconomic indicators (BB, TB, IPI) and is a vector of external shocks (FFR, OIL, EPU).

The regression model allows for an asymmetric approach to examine the effect of potential macroeconomic indicator taking a threshold variable () into account. After the determination of the threshold variable, Kourtellos et al. (2016) structural regression model is employed to estimate the threshold parameter () and the parameters of the coefficients () under different regimes. Higher regime (H) is reflecting the period when the threshold variable is higher than estimate of the threshold parameter, and low regime (L) is reflecting the period when the threshold variable is lower than estimate of the threshold parameter. is the indicator function for the presentation of the subsamples obtained from structural break via Kourtellos et al. (2016). The methodology also allows for the endogeneity of the slope variables using instrumental variables³. For the determination of the threshold variable, principal component-guided sparse regression (pcLasso⁴) model of Tay et al. (2018) is employed which is the combination of lasso and principle components (pc) that shrinks the coefficient vector to the leading principal components and hence provides the most important variable within a group of variables.

3. Empirical Findings

Historically, Turkey has adopted fixed exchange rate regime until 1980. After that period, several different regimes were adopted such as crawling peg, crawling band, managed floating, real exchange rate targeting until one of the deepest economic downturns of Turkish history in February 2001. Following the crisis, inflation targeting regime with floating exchange rate regime was adopted motivated by the success stories of several other inflation targeters. However, as an emerging economy, Turkey is subjected to capital flows and exchange rate volatilities are very high accordingly. Fig. 1.1 depicts the rise in the volatility of exchange rate around the neighborhood of global financial crisis and after coup attempt. The figures of all variables vis-à-vis exchange rate volatility are available in the Appendix.

Tab. 1.1 presents the descriptive statistics for all variables and Tab. 1.2 shows ADF unit root test results for different periods. Unit root variables are transformed into stationary form considering the period. Variables in rate or index form are differenced (monthly) whereas variables in level form are ←15 | 16→logarithmic differenced (monthly). All variables are standardized for a comparison of the coefficients⁵.

Fig. 1.1: Exchange Rate and Volatility (US Dollar/Turkish Lira), Source: Author’s own construction. Note: Global financial crisis hit Turkish economy within the period December 2008 – September 2009. The date of the coup attempt is July 15, 2016. Exchange rate volatility is depicted in the right axis.

Tab. 1.1: Descriptive Statistics, Source: Author’s own calculations

Tab. 1.3 presents pcLasso coefficients for full period and subsamples regarding dramatic changes within the period such as global financial crisis that affected Turkish economy in the period December 2008 – September 2009⁶ and coup attempt on June 15th, 2016. Hence, besides the full period, the following subsamples are investigated: pre global financial crisis period (February ←16 | 17→←17 | 18→2006 – December 2008); post global financial crisis period (October 2009 – February 2020); pre coup attempt period (February 2006 – June 2016); post-coup attempt period (July 2016 – February 2020). Out of 10 domestic and external macroeconomic variables, international reserves of the central bank (RES) is observed to be the most important variable for exchange rate volatility and has a negative association. This negative link is observed in the Turkish economic history during the depletion of international reserves – following crises – leading to devaluations (Kasman and Ayhan, 2008). In the similar manner, Parlaktuna (2005) explains that exchange market pressure in Turkey is reduced by adjustments in the foreign reserves. In the post-crisis subperiod, EPU is very close to RES and is positively associated to exchange rate volatility. This finding is in line with the recent empirical literature regarding the positive and significant impact of economic policy uncertainty index on exchange rate volatility for different country groups (Krol, 2014; Abid, 2019; Bartsch, 2019; Chen et al., 2020).

Tab. 1.2: ADF Unit Root Test (Constant and Trend Case), Source: Author’s own calculations

Tab. 1.3: PC-LASSO Coefficients, Source: Author’s own calculations

After the specification of the most relevant variable for the exchange rate volatility, structural threshold model is investigated with the international reserve as the threshold variable. In order to control for multicollinearity, correlation matrix is examined, given in Tab. 1.A1. The threshold model estimations are presented in Tab. 1.4 with low and high regime distinction for the overall period of February 2006 – February 2020. Low regime reflects the period when the threshold variable is below the threshold level, that is, the estimated threshold parameter; whereas high regime reflects the period when the threshold variable is above the threshold level. Hence, low (high) regime provides the effects of 10 internal and external macroeconomic variables on exchange rate volatility when the international reserves are below (above) a threshold level.

During low regime, that is, when international reserves are lower than an estimated threshold level, trade balance (TB), international reserves (RES) and policy rate (INT) reflect negative and significant impact on exchange rate volatility. These findings reflect that as trade deficit deepens, due to the rise in the necessity of foreign currency, exchange rate volatility rises; and, lower international reserves lead to higher volatility which can be attributed to capital flight due the rise in risk perceptions of the foreign investors; and as the policy rate declines, due to decline in the returns of domestic assets together with arbitrage, exchange volatility increases. During high regime, that is, when international reserves are higher than an estimated threshold level, policy uncertainty and budget balance are observed to have positive and significant impact on volatility, whereas the impact of trade balance is negative, as observed in the low regime period. The positive impact of EPU reflects that the rise in global political uncertainties increases exchange rate volatility due to capital flight in ←18 | 19→the times of rising risks. The positive impact of BB reflects that as budget balance improves, exchange rate volatility rises which can be linked to an indirect transmission mechanism: as budget balance improves, the return from holding domestic government bond declines which renders domestic treasury securities less attractive for the foreign investors and lead to capital outflow.