Exchange rate and trade balance a new approach using big mac index for the case of thailand

ii ABSTRACT This thesis focuses on achieving two main objectives: i an evaluation of the Thailand’s currency to confirm whether a currency is over- or undervalued during the period fro

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

EXCHANGE RATE & TRADE BALANCE:

A NEW APPROACH USING BIG MAC INDEX

FOR THE CASE OF THAILAND

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

i

ACKNOWLEDGEMENTS

It is assistance, guidance, and encouragement that contribute to the success of this thesis Therefore, I would like to express my gratitude to those who always stand by

me during periods of writing thesis

My profound appreciation and sincere thanks must first be expressed to my academic supervisor, Dr Vo Hong Duc, Director of Economic Regulation Authority in Perth, Australia, for his interest, encouragement, and guidance throughout thesis-writing process from initial themes to entire finish, especially during boring time of reading the final draft

I would like to express my grateful thanks to Assc Prof Dr Nguyen Trong Hoai,

Dr Truong Dang Thuy, Dr Phung Thanh Binh, and Dr Pham Khanh Nam, Lectures at Department of Economics Development at University of Economics (HCMC), for their helpful commands and valuable pieces of advice during the time I write the thesis Special thanks are extended to Dr Le Van Chon for his invaluable econometric guidance Thanks to such academic commands, advice, and guidance, I could finish the thesis on the best way

I am be grateful to not only staff members at Department of Economics Development at University of Economics (HCMC) but also my classmates at VNP-MDE 19 for their help and support and for offering such a warm and friendly studying atmosphere

Most importantly, I am indebted to my parents for financing my study as well as for their constant love and moral support This thesis could not be finished without their comprehension

ii

ABSTRACT

This thesis focuses on achieving two main objectives: (i) an evaluation of the

Thailand’s currency to confirm whether a currency is over- or undervalued during the period from 1980 to 2013 for the case of Thailand; and (ii) a consideration of the effects of a currency’s devaluation on trade balance for the case of Thailand In this study, these effects are considered in different aspects such as the determinants of a trade balance of Thailand; and the long run relationship between bilateral exchange rate and Thailand’s trade balance

In relation to the first objective, the prominent feature of the study is to use the

Big Mac Index (BMI), the first ever study of this kind conducted in Vietnam, for evaluating Thailand’s currency and to apply this evaluation in the context of the link between exchange rate and trade balance Nevertheless, the common-used index, the Consumer Price Index (CPI), is also utilized to compare the effectiveness of BMI with that of the CPI

Theoretical grounds for evaluating a currency in term of purchasing power parity (PPP) theory are first presented, together with the theory of the link between currency’s devaluation and trade balance Next, the empirical studies associated with such theory are discussed Empirical models and results are reported accordingly

To obtain the first objective, the PPP hypothesis is required to be satisfied in terms of panel cointegration tests The fully modified OLS (FMOLS) technique is adopted to determine the equilibrium exchange rate for evaluation process On the grounds of the tests of PPP, the panel-based unit root tests as developed by Breitung (2001) and the panel cointegration tests by Pedroni (1999, 2001, 2004) and Kao (1999)

are adopted The empirical results confirm a solid validity of PPP for the case of

CPI-based exchange rate and a weak evidence of the PPP for BMI-CPI-based exchange rate In

the relation to evaluate Thailand’s currency, the results illustrate that (i) the valuation

iii

of Thailand currency using CPI-based exchange rate is fairly consistent to that of

BMI-based exchange rate with an exception of the outcomes during the 1997 Asian

financial crisis; and (ii) the Big Mac exchange rate is better when bilaterally evaluating

the Thailand Baht to US dollar

The second main objective - a consideration of the effects of a currency’s devaluation on trade balance - could be achieved by the two procedures The first procedure is to analyze how changes on exchange rate policy, fiscal policy, and monetary policy affect Thailand’s trade balance In this task, effects of devaluation on trade balance are examined on various scenarios: (i) the entire sample of 62 countries who are trading partners with Thailand; (ii) different geography (between regions and regions of countries); (iii) different income levels; and (iv) during different periods in which Thailand’s currency is over- or under-valued Both OLS method and IV technique are employed because of an endogeneity problems as mentioned in previous studies The empirical findings indicate that the exchange rate policy plays a central role in explaining Thailand’s trade balance and the fiscal and monetary policies are beneficial in some cases

The second procedure is to examine the long run relationship between a devaluation of Thailand’s currency and trade balance with the applications of the panel-based co-integration tests by Pedroni (1999, 2001, 2004) and Kao (1999) and the FMOLS model The panel FMOLS estimations illustrate that a devaluation of Thailand Baht could provide positive effects on trade balance in the long run, especially for the groups of country with high income, upper middle income, in America, and Europe The individual FMOLS regressions between Thailand and each of her 62 trading partners indicate that the devaluation of Thailand’s currency would stimulate Thailand’s trade performance with over 20 trading partners, but hurt its performance with the other 10 countries and inconclusive conclusion for the others

iv

ABBREVIATIONS

FMOLS Fully Modified Ordinary Least Squares

OECD Organization for Economic Cooperation and Development

v

Contents

LISTS OF FIGURES viii

CHAPTER 1 1

INTRODUCTION 1

1.1 Problem Statement 1

1.2 Research Objectives 3

1.3 The Structure of the Thesis 3

CHAPTER 2 5

LITERATURE REVIEW 5

2.1 Theoretical Grounds 5

2.1.1 PPP Theory 5

2.1.2 Alternative approaches to devaluation theory 6

2.2 Conceptual framework 12

2.3 Empirical reviews 13

2.3.1 Test of PPP Theory 13

2.3.2 Evaluation of a currency 15

2.3.3 Determinants of a Trade Balance 16

2.3.4 Devaluation and Trade Balance 17

CHAPTER 3 25

THAILAND’S TRADE BALANCE AND 25

vi

ITS BIG MAC INDEX 25

3.1 Thailand’s Trade Balance 25

3.2 Big Mac Index 27

CHAPTER 4 31

RESEARCH METHODOLOGY 31

4.1 Panel unit root test 31

4.2 Panel cointegration test 32

4.2.1 Kao’s cointegration test 32

4.2.2 Pedroni cointegration test 33

4.3 Fully modified OLS approach 36

4.4 Empirical models 37

4.4.1 Test of PPP theory 37

4.4.2 Evaluation of Thailand’s currency 39

4.4.3 Determinants of Thailand’s Trade Balance 39

4.4.4 Devaluation and Thailand’s Trade Balance 45

CHAPTER 5 47

DATA AND RESEARCH FINDINGS 47

5.1 Data description 47

5.2 Empirical results 49

5.2.1 Test of PPP theory 49

5.2.2 Evaluation of Thailand’s currency 52

vii

5.2.3 Determinants of Thailand’s trade balance 56

5.2.4 Devaluation and Thailand’s Trade Balance 63

CHAPTER 6 73

CONCLUSION 73

6.1 Conclusions 73

6.2 Implications for research and policies 76

6.3 Limitations 77

REFERENCE 79

APPENDIX 86

viii

LISTS OF FIGURES

Figure 2.1: Conceptual framework for exchange rate effects on trade balance 12

Figure 3.1: Thailand’s trade balance during 1995-2013 periods 26

Figure 5.1: Misalignments of nominal exchange rate based on CPI and BMI 53

LISTS OF TABLES Table 3.1: The structure of Thailand’ trade classified by currency 27

Table 5.1: Data Description 48

Table 5.2: Unit root test for PPP testing 50

Table 5.3: Panel cointegration test for PPP 51

Table 5.4: Results of OLS and IV estimation for determinants of Thai trade balance 57

Table 5.5: Estimation results for undervaluation and overvaluation periods 59

Table 5.6: Estimation results for countries within each of the seven sub-samples 61

Table 5.7: Results of Breitung (2001) unit root test - level and first difference 64

Table 5.8: Cointegration test 65

Table 5.9: Panel results of FMOLS estimation 67

Table 5.10: Individual results of FMOLS estimation 70

1

CHAPTER 1 INTRODUCTION 1.1 Problem Statement

Exchange rate is possibly one of the most concerned subjects among academics, exporters, importers, investors as well as policy-makers because its vitally important roles played in the international economics While academics have concerned and developed theories of disequilibrium and equilibrium real exchange rate, the policy-makers concentrate more on exchange rate adjustment in order to examine its effects

on the economy Additionally, exchange rate risk is a key element related directly to the costs and profits for importers, exporters as well as foreign investors Furthermore,

it is argued that developing countries have tendency to devaluate their currency in order to gain the relative competition Given the importance of exchange rate in the economy, issues in relation to exchange rate attract my interest and that this a key reason for my choice to make a topic for the thesis

In February 2014, with the opening of the first ever branch of McDonald’s, a giant US fast-food company in Ho Chi Minh City, Viet Nam, the Economists Magazine have added the Vietnam Dong (VND) to its collection of Big Mac Index (BMI) This index provides McDonald’s burger prices in all different parts of the world

in terms of local and foreign currencies The index is considered as a lighthearted guide

in order to evaluate whether a currency of a particular country of interest is at its correct level The index has been increasingly well-recognized as a global standard with its existence in several international economics books and many academic studies The thesis aims to evaluate the real value of Thailand Baht and to consider the link between Thailand Baht depreciation and Thailand’s trade balance Thailand is

opted to study due to several reasons First, according to Bahmani-Oskooee and

2

Kantipong (2001), Thailand has an intense motivation to devaluate her currency owing

to the fact that after Asian currency crisis, Thailand was one of the most suffered countries in comparison with other nations in the Asian region Consequently, it lost market shares of many export products and services to China and other ASEAN countries As a consequence, the country suffered a severe deficit in its trade of balance Therefore, the strategy of devaluation would allow Thailand increasing her regional competitiveness, recovering her lost market shares, and improving the trade

balance (Bahmani-Oskooee and Kantipong, 2001) Second, earlier researchers cast

doubt on whether the Consumer Price Index (CPI) reflects the real exchange rate in the economy In this study, the Big Mac Index (BMI) is used as an appropriate substitute in lieu of the CPI to re-examine the roles of the exchange rate to the Thailand economy The use of the BMI, the first attempt of this kind in emerging country, is argued to be

an advance From that, a comparison between BMI-based exchange rate and

CPI-based exchange rate in regardless of Thailand’s trade balance is made

This thesis provides key advantages compared to previous studies First, earlier

works investigated the link between real devaluation and trade balance without testing whether the currency used in the study is over- or under-value, so their results are ambiguous In this paper, the valuation of a currency is conducted before it is used to

consider the influences of devaluation on trade balances for a particular nation Second,

the Big Mac index is initially applied in the contexts of real exchange rates and trade balances although a majority of authors have exploited this index for testing purchasing power parity (PPP), forecasting the trend of exchange rate (Ong, 1997; Yang, 2004; Cumby,1996; Clements & Lan, 2010) Clements and Lan (2010) and Ong (1997) argued that Big Mac index offers a sense of simplicity and creates a surprisingly

accurate result Third, to control the fact that exchange rates are endogenous to trade

balances, this study applied panel data estimation using the fully modified ordinary least squares (FMOLS) method offered by Phillips and Hansen (1990) and expanded

on BMI and CPI is made The second aim of this thesis is to consider the link between

a currency’s devaluation on trade balance in Thailand

To achieve the above-mentioned objectives, specific research questions are raised, including:

i Does PPP hold in the case of Thailand currency?

ii Is the Thailand Baht under- or over-evaluated against other currencies?

iii Are there any differences between BMI-formed and CPI-based real exchange rates in the test of PPP and the evaluation of Thailand Baht?

iv What factors determine the Thailand trade balance?

v Is there any long run relationship between real bilateral exchange rate and Thailand’s trade performance?

1.3 The Structure of the Thesis

This thesis concludes six main chapters Following the Introduction chapter, the rest of the thesis is organized as follows

 Chapter II presents literature review with both theoretical and empirical considerations The chapter begins with the discussion on a theory of PPP and the devaluation theory on trade balance Then, the framework of

4

determination of trade balance is depicted Next, the relationship between bilateral exchange rate and trade performance is analyzed Finally, empirical studies in relation with these theories are also presented

 Chapter III provides a brief overview of trade performance in Thailand and the development of Big Mac Index in recently decades

 Chapter IV discusses the methodology adopted in the thesis This chapter focuses on the research methodology, which leads to mathematical equations and empirical models

 Chapter V presents description of data and research findings Sources of data are acknowledged, together with description of statistics Empirical findings from this study are then presented following a comparison between the findings from this study and those from previous studies

 Chapter VI provides a summary of conclusions, which are drawn from this study for Thailand Limitations of this study and ways for future research are added at the end of the thesis

5

CHAPTER 2 LITERATURE REVIEW

This chapter presents theoretical backgrounds and empirical studies associated with purchasing power parity (PPP) theory, determinants of trade performance, and the relationship between a currency devaluation and trade balance

2.1 Theoretical Grounds

2.1.1 PPP Theory

This section provides an overview of the PPP theory by presenting two versions

of the theory, known as the absolute PPP and the relative PPP

The absolute PPP relates to the relationship between nominal exchange rate and the corresponding to relative price level in terms of level form Meanwhile, the relative PPP expresses this relationship as the first difference form, meaning that percentage change in the exchange rate between two currencies over any period approximately equals the difference between the percentage changes in the national price levels (Krugman et al 2012, p.387)

In term of the absolute PPP, where 𝑞𝑖𝑡 represent the real exchange rate of country

i in year t; 𝐶𝑃𝐼𝑖𝑡 and 𝐵𝑀𝐼𝑖𝑡 are the Consumer Price Index and the Big Mac Index in

country i in year t 𝐶𝑃𝐼𝑡𝑇ℎ𝑎𝑖 and 𝐵𝑀𝐼𝑡𝑇ℎ𝑎𝑖 are the Consumer Price Index and Big Mac Index in Thailand in year t The ratio of 𝐶𝑃𝐼𝑖𝑡 to 𝐶𝑃𝐼𝑡𝑇ℎ𝑎𝑖 and the proportion of 𝐵𝑀𝐼𝑡𝑇ℎ𝑎𝑖 to 𝐵𝑀𝐼𝑖𝑡 illustrate the relative prices between a foreign country and Thailand

in term of CPI and BMI ERit is the bilateral nominal exchange rate, defined as a

number of Thailand Baht per unit of partner i’s currency

Where 𝑞_𝑐𝑝𝑖𝑖𝑡 and 𝑞_𝑏𝑚𝑖𝑖𝑡 respectively represent the error terms of CPI and BMI

in terms of natural logarithms

A currency is considered to be overvalued if 𝑞𝑖𝑡 is positive, undervalued if 𝑞𝑖𝑡 is negative, and at parity if 𝑞𝑖𝑡 equals zero

When it comes to the relative PPP, equation (2.1) and (2.2) are transformed as follows:

2.1.2 Alternative Approaches to Devaluation Theory

On the theoretical ground, the effect of currency devaluation could be established

on three approaches including elasticity, income-absorption, and monetary Each of these approaches is briefly discussed in turn below

7

2.1.2.1 The Elasticity Approach

The traditional exchange rate model is the elasticity approach, which puts the emphasis on the condition of supply and demand sides to exports and imports On the export side, it is assumed that the depreciation has a tendency to initially decrease foreign prices of the export county so that the price reduction would increase foreign demand for the devaluating nation How much the export would adjust depends upon the foreign demand and domestic supply elasticity for exports Concomitantly, on the import side, when a devaluation of domestic currency causes a rise in the domestic price of imports Thus, this leads to a reduction in a country’s demand for imports, which in turn may seem to decrease the foreign price of imported products How much the import would alter relies on the domestic demand and foreign supply for imports Hence, in the context of the elasticity approach, the responds of trade flows to exchange rate adjustment depend on the elasticity of demand and supply both of domestic and of foreign country The key issue of this method is postulated as Marshall-Lerner (ML) condition This widely-recognized condition states that when the current account begins at an equilibrium position, the devaluation will increase trade flows in the case of the total of (i) price elasticity of domestic demands for imports and (ii) price elasticity of foreign demand for exports in absolute terms exceeds unity In such situation, the greater of the total of the two elasticities is, the more significant the exchange rate depreciation influences on the trade balance

In contrast, the devaluation would deteriorate the balance of trade when the total

of these two elasticities is smaller than unity (Salvatore 2012 p.516) After depreciation, it is observed that the balance of trade initially worsens before achieving

an improvement This is known as the so-called J-curve phenomenon This phenomenon happens because of time lags in recognition, decision, delivery, replacement, and production (Junz & Rhomberg, 1973) The J-curve phenomenon also

8

results from currency pass-through, meaning that the depreciation’s influence on trade balances might rely on how quickly manufacturers transfer changes in costs to their consumers (Salvatore 2012 p.524)

On the ground of the mathematical model (Johnson, 1975; Salvartore, 2012 p.524), the effect of devaluation on the trade balance will be modelled in term of the elasticity approach

𝐵𝑓 =1

𝑟𝑝𝑥 𝑋 {𝑋𝑠(𝑝𝑥) = 𝑋𝑑(𝑝𝑥

𝑟)} − 𝑝𝑚 𝑀{𝑀𝑑(𝑟𝑝𝑚) = 𝑀𝑠𝑝𝑚)} (2.5) where 𝐵𝑓 is the trade balance expressed in foreign currency, r is the foreign

exchange rate defined as number of units of domestic currency per unit of foreign currency, 𝑝𝑥 and 𝑝𝑚 are respectively import and export prices in term of domestic

currency, and X and M are correspondingly import and export volume, d and s

represent for the demand and supply function

Differentiating equation (2.5) by r, the result is as below:

𝑑𝐵𝑓

where 𝐸𝑑𝑓 , 𝐸𝑑𝑑 are respectively the elasticity of foreign and domestic demands

To derive the Mashall-Lerner condition mathematically, equation (2.5) is differentiated in term of individual good markets to quantity The elasticity of the domestic price of export with respect of the exchange rate is:

9

where 𝑛𝑥 , 𝑒𝑥 are the price elasticity of demand and supply for exports, measured

by the percentage change in export quantity for a given percentage change in export price

𝑛𝑥 = −

𝑝 𝑥 𝑟 𝑋

𝑑𝑋𝑑𝑑(𝑝𝑥𝑟) 𝑎𝑛𝑑 𝑒𝑥 = −𝑝𝑥

Substituting equation (2.11) and (2.12) into equation (2.6) with the presumption

of 𝑟 = 𝑝𝑥 = 𝑝𝑚 = 1 and X=M, it could be obtained that:

1 𝑋

𝑑𝐵𝑓

𝑑𝑟 =𝑒𝑥 𝑒𝑚(𝑛𝑥+𝑛𝑚−1)+𝑛𝑥𝑛𝑚(𝑒𝑥+𝑒𝑚+1)

10

From equation (2.14), the trade balance enjoys an improvement when 𝑛𝑥𝑛𝑚 >

𝑒𝑥𝑒𝑚 and the Marshall-Lerner stability condition is satisfied In the case of the sum of the elasticities of demand for imports and demand for exports exceed than 1 (𝑛𝑥+

𝑛𝑚 > 1), the effectiveness of devaluation is experienced when elasticities of supply for imports and supply for exports are infinitely elastic (𝑒𝑥 = 𝑒𝑚 = ∞)

If the depreciation takes place from the condition of trade deficit (X>M), the trade balance would improve as if the elasticity of demand for imports should be given a proportionally greater weight than the elasticity of demand for exports

𝑛𝑥 +𝑀𝑝𝑚

2.1.2.2 The Absorption Approach

The second approach explaining influences of devaluation on trade balance is the absorption approach, which stresses how the devaluation affects expenditure behaviors

of domestic nation, and thus affects trade balances The trade balance is estimated by taking the difference between total aggregate demands and domestic absorption, measured by consumption, investment, and government purchases Therefore, if total output exceeds internal absorption, the balance trade is positive and vice versa The absorption approach states that the devaluation increases the trade balance when the economy is performing under its capacity However, the economy produces more domestic output than its capacity; devaluation would increase not only trade balance but also inflation

The absorption approach is proposed by Alexander (1952) and further developed

in Alexander (1959) The trade balance is generally defined as the difference between a total domestic production and a sum of goods and services domestically consumed On the light of devaluation effects on trade balance, Alexander (1952) proposed two channels The first is an indirect channel, in which a currency’s devaluation leads to a

11

change in the production of goods and services before altering the absorption of goods and services, and the second is a direct path, in which the devaluation may straightforwardly change the amount of real absorption at any real income level On the direct channel, the principle influence of devaluation on output is related to an increase

in exports, and thus resulting in an induced stimulation of internal demand, together with the terms of trade effects under which presumed that devaluation causes a decrease in export prices greater than a fall in import prices in term of foreign currency On the direct channel of devaluation to absorption, devaluation has generally resulted in rising price level to discourage consumption or investment at a given real output level The most generally recognized impact associated with direct absorption is the cash balance effect When price rises and money supply is inflexible, individuals have to accumulate cash to satisfy their desire to maintain their wealth at certain real level, consequently leading to a reduction in their real spending relative to their real income The cash balance effect may either influence directly the income-spending association through a reduction of spending in order to build up cash, or indirectly through the interest rate on the purpose of transferring assets into cash (Alexander, 1952)

2.1.2.3 The Monetary Approach to Devaluation Theory

The most third common-used approach is the monetary approach studying in a broaden concept of balance of payments This approach had been emerged during 1960s by Robert Mundell and Johnson and fully developed during 1970s The monetary approach considers balance of payments as an essentially monetary phenomenon on the view of international economics (Salvatore, 2012 p.509)

This approach supposes that devaluation may lead to a temporary increase in the balance of payments rather than a permanent improvement An explanation is that devaluation would push up the domestic consumer price, raise the demand for money,

12

and eventually lead to an increase in interest rate As a result, this draws the capital inflows from abroad However, in the long run, capital inflows would result in an increase in domestic expenditure and imports of goods and services, and this makes balance of payments return to its initial level Hence, depreciation has effects on balance of payments temporarily, and it causes the higher consumer prices permanently

Price index Interest rate

Term of Trade

PPP theory: (i) panel unit root tests and (ii) panel co-integration tests

On the first school, Oh (1996) made an argument that unit root tests using the panel data were more powerful in comparison with that of conventional univariate time series The author explicitly found the evidence in favor with PPP for G-6 and OECD nations during the period of flexible exchange rate Wu (1996) offered further support for Oh’s (1996) argument by demonstrating that the failure to prove an evidence of long-run PPP as reported by early researcher may result from the low power of standard univariate unit root tests With the same technique as Oh (1996) and Wu (1996), Papell (1997) examined the long-run PPP for industrial nations over the period

of floating exchange rate regimes The findings showed that the null hypothesis of unit roots had a stronger tendency to be rejected for large than small panels, for monthly than for quarterly data, and for German mark using as a base currency rather than for the US dollar Similarly, Connell (1998) was in support of the stationary of unit root with the quarterly panel data of 64 countries The author found the similar results when divided the whole sample into four subsamples including Europe, Asia, South

14

America, and Africa In comparison with earlier studies, his method had an advantage

of controlling cross-sectional dependences in the data, which raised significance level

of tests with a normal size of 5 percent up to 50 percent To sum up, on the first school, the majority of studies found the supportive proof of PPP thanks to the power of panel-based unit root tests although the results tend be sensitive to inclusion of different subsamples

Another school of examining long run PPP has been employed the panel cointegration test, developed and published in Pedroni (2000, 2001, 2004) There are a numerous researchers of this school, representative are Cazoneri et al (1999), and Pedroni (2001, 2004) Such researchers found valid evidences to reject the null hypothesis of no cointegration, thus reinforcing the weaker version of PPP with heterogeneous slope coefficients

With the development of panel econometrics technique, a number of empirical papers provide a supportive confirmation that the PPP holds in the long run A common explanation for the support of PPP is the power of panel tests (Oh, 1996; Wu 1996; & Connell 1998) Papell (2002) offered a different reason why PPP holds with panel data; the great appreciation and depreciation of currencies adhere to typical patterns of the dollar’s fall and rise However, it should be noted other issues hurting the validity of PPP tests Benergiee et al (2005) put an emphasis on an important underlining of cross-unit cointegrating relationship with panel-based unit root tests so that the use of panel methods for testing unit roots should be cautious Papell and Theodoridis (2002) pointed out that the choice of a currency in panel methods significantly influenced results of PPP testing, together with distances between nations and the volatility of exchange rate Taylor and Taylor (2004) presented a summary of debate associated with PPP that have been challenged academic researchers during three recent decades

15

2.3.2 Evaluation of a Currency

The basic idea of evaluating a currency is to find out the difference between the equilibrium exchange rates and its actual value or “misalignment” The currency is called overvaluation if the equilibrium levels exceed its actual values and undervaluation if the equilibrium levels are smaller than actual values Based on the principle, researchers attempt to identify the equilibrium of exchange rates with a variety of approaches

Chinn (2000) implemented three major procedures including (i) long-run PPP, (ii)

a productivity-based model, and (iii) a monetary model to estimate the equilibrium exchange rate of East Asian currency The authors attempted to prove the East Asian financial crisis in 1997 owing to currency overvaluation The results found that most selected currencies had been overvalued before the 1997 financial crisis happened The PPP and productivity-based approaches indicated closely conclusions in which currency of Hong Kong, Thailand, Malaysia, Indonesia, and Taiwan were overvalued and the Korea’s Won was undervalued The overvaluation of Malaysia’s Ringgit and Thailand’s Bath were small The study of Jonwanich’s (2008) for evaluating of Thailand Bath provides consistent findings of Chinn (2000)

Unlike Chinn (2000) studies, Rajan et al (2004) identified misalignments of Thailand’s currency in terms of economic fundamental model and examined effects of misalignments on trade performance during 1980s and 1990s The authors used both real effective exchange rate and bilateral real exchange rate against US dollar and Japanese Yen The findings indicated that the real exchange rate and bilateral exchange rate against US dollar had a moderate misalignments while the misalignments was significant for bilateral exchange rate against Japanese Yen, thus leading to a widening deficit in trade between Thailand and Japan during 1990-1996 periods

16

On the base of equilibrium exchange rate theory, Zhang (2001) estimated behavior of China’s currency, Renminbi, and its misalignments The evidence illustrated that the Renminbi was overvalued during central planning periods in China and converged to the equilibrium level after China’s economic reforms Yang (2004) first applied BMI to analyze of China’s currency and showed that the Renminbi was undervalued However, Cheung et al (2007) casted doubt on the conventional wisdom

of undervaluation of Renmini, asserting that when the uncertainty and serial correlation

in data accounted for, there was little evidence that China’s currency was undervalued

To summarize, there are numerous approach for evaluation of a currency Depending on valuation methods and periods, results have been mixed

2.3.3 Determinants of a Trade Balance

As previously discussed in the theoretical sections, the trade balance is determined on the ground of three well-known approaches including elasticity, absorption and monetary The elasticity approach indicates that exchange rate serves as

a main factor of trade balance and proposes depreciation as an effective way to deal with trade deficit The absorption approach provides income as a major element in explaining trade performance and suggests that any income-related policy like contractionary fiscal policy could cope with trade deficit (Alexander, 1952) Meanwhile, monetary approach takes money supply into consideration of trade disequilibrium and favors the use of monetary policy to correct deficit of balance of payments Therefore, exchange rate, income, money supply are such fundamental determinants of trade balance Additionally, it is worth noting that factors associated with fiscal and monetary policy that have been used to correct trade deficit are vitally important

17

To test empirically factors which determine a country’s trade performances, these factors have been divided into external and internal determinants On the internal category, Furstenberg (1983) had an attempt to examine such domestic factors that significantly influenced the US current account and to suggest policies associated with the national saving and internal investment rates Miles (1979) established direct relationship between exchange rate and trade balance, in addition to income, monetary supply, and the ratio of government consumption to output The author investigated the experience of 14 nations using annual data over 1956-1972 periods and concluded that exchange rate devaluations did not improve trade balance of most cases but they increased balance of payments through the capital account Although Himarios (1985) based on Miles’ framework, the author provided critiques of Miles’ findings, asserting

that (i) the findings have problems of unit measurement, (ii) the effects of foreign

variables and the real terms on trade balance may be differential from that of domestic

and nominal ones, and (iii) the longer lagged structure of exchange rate plays a

significant role Himarios (1985) found that in most countries, the cumulative exchange rate coefficient was statistically significant at 5% level, meaning that depreciations increased trade balance Another Himarious’ research in 1989 was in favor of his 1985 findings, showing that devaluations had been a helpful tool for adjusting trade balance

2.3.4 Devaluation and Trade Balance

When it comes to the empirical studies of investigating the relationship between devaluation and the trade balance, using different kinds of trade data, a vast number of researchers attempt to examine the J-curve phenomenon This theory states that when a real exchange rate depreciates, this initially makes the trade balances deteriorated, but enjoys an improvement permanently While some have used the aggregated trade data

at country levels to test this hypothesis, others have applied the disaggregate data or

18

bilateral trade data, meaning that the bilateral trade and exchange rate is used between

a nation and her trading partners

On account of aggregated trade data, Himarior (1985, 1989) and Miles (1979) are representative authors using this type of data Bahmani-Oskooee (1985) introduced an approach of Alan lag structure on the purpose of testing the J-curve phenomenon Using the quarter data of four countries including Greece, India, Korea, and Thailand, which have differently exchange rate regimes, the author found that the J-curve exists

in three former nations (Greece, India, and Korea), except for the case of Thailand The devaluation of Thailand Baht increases the trade balance in five quarters before the situation changes

With the advent of cointegration technique by Engle-Granger (1987) and developed by Johansen and Juselius (1990), the technique allows regressing two I(1) variables without spurious regression Bahmani-Oskooee and Alse (1994) pointed out that the first differenced data in Miles (1979) research was stationary Committedly, the data used in previous studies (Bahmani-Oskooee 1985; Himarios 1985, 1989) was not stationary As a consequence, standard critical value for determination of the significance of estimated coefficients are not reliable After taking into consideration most problems related to early research, Bahmani-Oskooee and Alse (1994) reexamined the effects of devaluation on trade balance with error-correction modeling and Engle-Granger cointegration approach Using quarter data over the 1971-1990 periods from 19 developed countries and 22 less developed nations, they indicated that devaluation positively influence trade balance of Costa Rica, Brazil, as well as Turkey and negatively impact that of Ireland in the long-run There is no long-run effect between devaluation and trade balance for Canada, Denmark, Germany, India, Mexico, Portugal, Spain, Sri Lanka, U.K and U.S, but he found the J-curve phenomenon for Costa Rica, Ireland, Netherlands, and Turkey Bahmani-Oskooee (1998) conducted a

19

research with the aim at estimating multilateral with the application of Juselius cointegration technique His results revealed that the devaluation could stimulate the trade balance of most surveyed nations, and the Marshall-Lerner (ML) condition is satisfied for the case of Korea, South Africa, and Greece

Johansen-In light of aggregate data of eight OECD countries, Boyd et al (2001) investigated the long-run effects of real exchange rate on the trade balance The authors combined vector autoregressive (VAR) cointegration with autoregressive distributed lag (ARDL)

to form structural cointegrating vector autoregressive distributed lag (VARDL) models Their results presented that the J-curve held for six out of eight countries with the exception of France and Germany while the Marshall-Lerner (ML) condition was satisfied for five nations including France, German, Japan, the Netherland, and the US Meanwhile, Lal and Lowinger (2002) utilized Johansen- Juselius cointegration technique, error correction model (ECM) by Engel and Granger (1987), and impulse response function to assess the short-run and long-run the connection between trade balance and nominal exchange rates with the case of five South Asian countries (Bangladesh, India, Nepal, Pakistan, and Sri Lanka) The study indicated that currency depreciations immediately worsen a country’s trade balance, but the situation may reverse in the coming quarters Employing the same approach as that of Lal and Lowinger (2002) with annual data during the period of 1960-1995, Singh (2002) concluded that real exchange rate was statistically related to the balance of trade in India

Some authors (Brissimis & Leventankis 1989; Rose 1990) noted that estimating the link between trade balance and exchange rate might face econometric problems of simultaneity and dynamics, and thus ordinary least square (OLS) estimates lack consistency They employed instrumental variable (IV) technique to correct both mentioned problems Rose (1990) found no relationship between the exchange rate and

20

the trade balance when applying annual data of 30 developing nations over the period

of 1970-1988 He obtained virtually same results when replacing annual by quarter data Rose (1991) examined this relationship for five main OECD nations over the period of post-Bretton Woods era using both parametric and non-parametric methods The results pointed out that exchange rates played no role in improving trade balances, which is consistent with findings of Rose (1990) Therefore, although such authors (Bahmani-Oskooee 1985, 1998, Boyd et al 2001, Lal and Lowinger 2002, Rose 1990,

1991, Singh 2002) used a variety of approaches with aggregate data, the results for effects of exchange rate changes on trade balance are mixed

The studies reviewed above raised tremendous concerns when using the effective exchange rate as an aggregate proxy for exchange rate Bahmani-Oskooee and Brooks (1999) supposed that this method is highly likely to face a serious problem that a nation’s currency might appreciate with some currencies, but depreciate with other currencies Additionally, taking weighted averaging estimate of the exchange rate would smooth out the fluctuation of real effective exchange rate, leading to an unsustainable connection between the effective exchange rate and total trade balance

To cope with the problem of effective exchange rate, a large body of studies have employed the disaggregate data or bilateral exchange rate to investigate the link between trade balances and exchange rate changes

Theoretically, devaluation of nominal exchange rate worsens the trade balance in the short-run, but it improves the balance of trade in the long run However, existing empirical works provides ambiguous results For examples, Rose and Yellen (1989) found no relationship between devaluation and the trade balance using America data over the period from 1960 to 1985 between the US and six trading partners Halicioglu (2008) analyzed the bilateral J-curve dynamics of Turkey and found that in the short run, there is no specific influence of exchange adjustment on trade balance while in the

al (2001) together with error correction model (ECM) by Engel - Granger (1987) Researchers (Bahmani-Oskooee & Brooks, 1999; Arora et al, 2003; Bahmani-Oskooee

& Harvery, 2010; Bahmani-Oskooee et al, 2006; Bahmani-Oskooee et al, 2005) tested

a variety of countries including America, India, Malaysia, the UK, and Canada, respectively Particularly, Bahmani-Oskooee and Brooks (1999) use quarter US’ bilateral trading data with six developed countries comprising Canada, France, Germany, Italy, Japan, and the UK over the period between 1973 and 1996 The results indicated that there is no specific influence of exchange adjustment on trade balances

in the short run while the real exchange rates are positively related to the trade balances

in most cases, except for the UK in the long run

Additionally, investigating India’s bilateral trade with seven developed nations, Arora et al (2003) illustrated that although there is no support for J-curve phenomenon, the devaluation of Indian rupee would stimulate its trade balance with at least with four countries (Australia, Germany, Italy and Japan) The similar findings were found when Bahmani-Oskooee et al (2006) used quarter UK data for more than 20 countries over the period from 1973 to 2001 They found no J-cuvre effect on short run and in the long run, solely five cases of Australia, Austria, Greece, South Africa, Singapore and Spain that real exchange rates positively affected trade balances Instead of testing the Marshall-Lerner condition by net trade balance value, Bahmani-Oskooee et al (2005), Bahmani-Oskooee and Harvery (2006) used in-payments and out-payments, correspondingly representing for nominal export values and nominal import values to

22

directly access the exchange rate depreciation on exports, imports as well as trade balances In terms of Canada, Bahmani-Oskooee et al (2005) reported that the real exchange rate depreciation reduces Canada’s out-payments to at least 50% of her trading partners, but merely raises her in-payments from two (Italy and Spain) in the total of 20 nations In other words, real devaluation of Canadian dollar has influenced more in-payment than out-payment Closely empirical results were discovered in the research of Bahmani-Oskooee and Harvery (2006) for the case of Malaysia In the long run, although real depreciation of Malaysian ringgits created positively significant influences on its in-payments for five cases over a total of 14 surveyed countries, it had

no impacts on her out-payments

Another common approach that researchers use to investigate the link between real devaluations and trade balances is Johansen’s co-intergration test and error correction model (ECM) together with impulse response function (IRF) (Bahmani-Oskooee & Wang, 2007; Thorbecke 2006) On the one side, Bahmani-Oskooee and Wang (2007) checked a hypothesis of whether real renminbi devaluation was a factor leading to a deterioration in the trade balances between China and American from a deficit of 600 million dollars in 1978, the year China reforms her economy, to a huge surplus of 120 billion dollars in 2002 The findings demonstrated that real China’s Renminbi devaluation made a substantial contribution to the changes in China-US trade balance, in which China’s in-payments of 40 industries had experienced a considerable improvement while solely 18 over 88 industries surveyed suffered a reduction in their out-payments On the other side, Thorbecke (2006) found that a renminbi appreciation would help reduce the US trade deficit with China since it could lead to a generalized appreciation of Asian currencies and eventually impact China’s exports

23

However, the results of earlier research may suffer the bias and ineffectiveness due to the problems of simultaneity and endogeneity among each variable (Chiu et al 2010) Rose and Yellen (1989) indicated that the trade balance model, containing output and exchange rate, may raise the problem of potential simultaneity bias and thus, instrumental variable (IV) method is more appropriate than ordinary least square (OLS) Yol and Baharumshah (2007) indicated that it is highly likely to be a direct and indirect causal relationship between trade balances and such macroeconomic as output, exchange rate, and money supply and some other indicators To cope with the problem

of endogenity, some studies use the IV method (Brissimis & Leventankis 1989; Rose and Yellen 1989, Rose 1990), whereas other utilize the fully modified ordinary least squares (FMOLS) approach (Yol and Baharumshah 2007, Chiu et al 2010) Unlike previous studies, panel data estimations allow taking such heterogeneity explicitly into account by controlling individual variances Additionally, panel data provides more informative and less collinearity among the variables, more degree of freedom, and more efficiency (Gujarati & Porter 2009)

As far as Thailand is concerned, the link between Thailand’s trade balance and bilateral exchange rates have been investigated For example, Bahmani-Oskooee and Kantipong (2001) used quarter data over the period between 1973 and 1997 on the behave of testing the J-curve phenomenon between Thailand and her five major trading partners The findings revealed that there was at least existence of J-curve phenomenon for two cases of Japan and America

Baharumshah (2001) used real effective exchange rate to investigate the short-run and long-run effects of real exchange rate changes on the bilateral trade balances of Thailand and Malaysia to the US and Japan over the 1980-1996 period The results indicated that there is no J-curve phenomenon in the short run and the real exchange rate devaluation would affect the trade balance after eight or nine quarters Instead of

24

using real effective exchange rate like Baharumshah (2001), Onafowara (2003) applied bilateral exchange rates and extended the period from 1973 to 1996 for three Asian countries (Thailand, Malaysia, and Indonesia) with the US and Japan The findings showed that while the J-curve exist for Thailand’s bilateral trade balance with the US

in the short run, Thailand’s bilateral trade balance with Japan is more consistent with S-curve than J-curve.On one hand, Bahmani-Oskooee and Harvery (2010) pointed out that the real exchange rate changes have no effects on Malaysia’s bilateral trade balance to Thailand On the other hand, Chiu et al (2010) presented that the devaluation of the US dollar would improve her trade balance to Thailand

25

CHAPTER 3 THAILAND’S TRADE BALANCE AND

ITS BIG MAC INDEX

In this chapter, an overview of Thailand’s trade balance during 1980-2013 periods is initially provided Then, Big Mac Index (BMI) is brief reviewed in terms of the development, its advantage and disadvantage, as well as its appliance in academic research

3.1 Thailand’s Trade Balance

On the Figure 3.1, the Thailand’s trade balance is defined as the ratio of exports

to imports, so this figure solely shows where Thailand’s trade balance is surplus or deficits without considering real numbers As depicted in the figure, Thailand trade had suffered a huge deficit before 1997, the year marked Asian crisis However, when the Thailand currency considerably devalued in 1997, Thailand’s trade immediately reacted from a big deficit to a significant surplus in 1998 The Thailand trade had a tendency to converge to its balance level in next five years from 1989 to 2004 During the global financial crisis, Thailand had experienced a large fluctuation in her trade balance In two recent years, Thailand had suffered a deficit in her trade balance The value of exports was somewhat higher 90% of imports value

26

Figure 3.1: Thailand’s trade balance during 1995-2013 periods

Sources: Author’s calculation on data retrieved from Bank of Thailand

Presented in Table 3.1 is the structure of trade in Thailand classified by different currency over two decades from 1993 to 2013 Generally, the most striking feature from the table is the dominance of US dollar in Thailand trade with a large majority of total trade value Additionally, the structure does not much change in a popularity of

US dollar The percentage of US dollar in total trade value consisted of 83% in 1993 and somewhat reduced to nearly 80% over two decades The Thailand Bath has experienced a substantial surge in currency structure from 0.75% in 1993 to 8.55 in

2013, an increase higher than 10 times The Deutsche mark is replaced by Euro, and keeps remained the proportion of structure currency Other currencies, Pound sterling, and Singaporean dollar have been reduced the role played in Thailand’s trading

Sources: Author’s calculation on data retrieved from Bank of Thailand

3.2 Big Mac Index

Big Mac index is the reflection of McDonald hamburger’s prices all over the globe that the world-wide magazine, the Economists, has been published regularly since 1986 The index is considered as a lighthearted guide in order to evaluate whether

a currency is at its correct level The index has been increasingly well-recognized as a global standard with its existence in several international economics books and many academic studies

The advantage of Big Mac index in comparison with other aggregate kinds of indies like consumer price index (CPI), wholesale price index (WPI), etc is that its virtually exact composition In other words, with few exceptions, ingredients of a burger seem to be the same in everywhere it is sold Thus, the use of Big Mac standard satisfy at least a requirement of testing for PPP, so its contributions to real exchange rate estimation could be relatively precise However, like CPI, the BMI cannot overcome other strict conditions for testing PPP theory Pakko and Polard (2003)

28

provided explanations which contributed to the failure of the PPP tests The first factor was trade barriers including transaction costs, trade restrictions, and taxes The second element resulting in the PPP fail was non-traded goods in relation with productivity bias, government spending, and current account deficits The final determinant of PPP ineffectiveness was connected with the pricing to market

Since 1986, the Economist magazine has annually published the Big Mac Index, recording the Mc’s Donald bugger prices among different countries around the world The index has been generally applied by many researchers such as Cumby (1996); Parley & Wei (2007); and Ong (1997) as a rough-and-ready guide to whether a national currency is trading at the right exchange rate Although academics such as Clements, Lan and Seah (2012) provide a solid evidence to show the bias of BMI, various authors have also used the index to evaluate the exchange rate Yang (2004) and Ong (1997) adjusted BMI and applied to evaluate Chinese currency and testing exchange rates among nations in the long run Cumby (1996), Clements and Lan (2010) employed the original index for testing purchasing power parity (PPP) and for forecasting exchange rates Clements et al (2010) strongly believes that Bugernomics is

an emerging stream to examine whether the BMI can be used to predict exchange rates because of its simplicity (Clements & Lan, 2010) and a surprisingly accurate result (Ong 1997) Parsley and Wei (2007) employed Big Mac prices with costs of ingredients to study real exchange rate and the speed of convergence of the real exchange rate adjusted by Big Mac index

Contemporary studies have shifted attention to using another index for academic purposes Cumby (1996) used Big Mac index with aims at predicting exchange rate He stated that though average deviations from absolute Big Mac parity are large for a variety of currencies, the convergence of relative Big Mac parity was rapid Therefore, deviations from relative Big Mac parity provide useful information for forecasting

29

exchange rate; a 10-percent undervaluation according to hamburger standard would lead to a 3.5 percent appreciation in the following year Click (1996) concluded that Big Mac prices are in favor with PPP, especially in time-series dimension and the deviations from PPP could be explained by the Balassa-Samuelson effects, which states that non-traded products would be more expensive in high-productive or high-income countries

Ong (1997) argued that the validity of PPP testing resulted from the selection of price index for making purchasing power comparisons The author pointed out that it was the difference of productivity in tradable and non-tradable products across countries that raised concerns of PPP findings It is believed that the use of BMI could

be considered as a palatable alternative for examining PPP Therefore, Ong (1997) derived No-Frills index excluding non-tradable elements from BMI for taking into consideration of the productivity differentials across countries Surprisingly, the findings based on the BMI were accurate in tracking exchange rate in the long run and consistent with previous PPP testing results

Caetano et al (2004) reexamined the PPP with the use of Big Mac index They claimed that Click’ results had not been robust to adjusting estimation approach, country data, and time periods This problem could be solved by grouping countries in terms of income The results showed that the PPP from BMI had not held and the reasons for PPP failures were income and trade effects Similarly, Chen and Wang (2007) applied the panel co-integration approach on the purpose of comparing the validity of PPP using CPI and BMI Amazingly, Big Mac price was more supportive to the validity of PPP than the CPI price

Yang (2004) made an analysis of China’s currency, Renminbi, with the application of the BMI as a guide The findings supported the statement that the Renminbi was undervalued in comparison with the US dollar over the surveyed period

30

The author argued that the undervaluation of Renminbi resulted from two natural

elements: (i) the non-traded ingredients of the products and (ii) the wage differences

between in China and in American A policy recommendation was implied that the Renminbi should revalue when the gaps in productivity and labor compensation between two nations narrow

Parsley and Wei (2007) approached the Big Mac index in terms of its ingredients with the aim of studying real exchange rate They proposed that this approach provides

a number of advantages, including (i) the levels of Big Mac real exchange rate could be estimated meaningfully, (ii) the estimation of Big Mac’ s tradable and non-tradable

components could be relatively precise owing to its exactly-known composition

31

CHAPTER 4 RESEARCH METHODOLOGY

This chapter will present the research methodology for this thesis First, the panel

unit root test and co-integration test will be used with the aim of testing the validity of

PPP theory Second, the group-mean fully modified ordinary least square (FMOLS) is

applied to estimate the coefficients of long –run exchange rate as a proxy for equilibrium level Based on these results, the exchange rate valuation is taken place under condition that the currency is called “undervalued” when the actual exchange rate exceeds its equilibrium level, and “overvalued” if the actual one is smaller than its

equilibrium level Third, factors influencing trade balance are tested with the use of

panel econometrics technique The ordinary least square (OLS) and instrument variable (IV) method is applied The serial correlation, heteroskedasticity, and endogeneity tests

are also conducted at this step Fourth, the long-run relationship between trade balance,

bilateral real exchange rate, and the income level is examined with the given techniques

4.1 Panel Unit Root Test

To avoid spurious regressions associated with time series data, the panel data is examined whether it is stationary or not with the use of Breitung’s (2001) panel-based unit root test It is argued that this test’s performance is more powerful than unit root tests employed in individual time series data Unlike panel-based unit root tests provided by Im et al (2003) and Maddala and Wu (1999), the approach of Breitung (2001) allowed individual process to have a common unit root which is similar to that

of Levin et al (2002) (hereafter LLC) A common unit root indicates that the tests are estimates assuming a common autoregressive (AR) structure for all the series