DEVALUATION AND ITS IMPACT ON TRADE - USP...
Transcript of DEVALUATION AND ITS IMPACT ON TRADE - USP...
1
DEVALUATION AND ITS IMPACT ON TRADE
PERFORMANCES IN FIJI:
A SECTORAL AND BILATERAL LEVEL ANALYSIS
by
Kushneel Avneet Prakash
A thesis submitted in fulfillment of the
requirements for the degree of
Masters of Commerce
Copyright © 2015 by Kushneel Avneet Prakash
School of Economics
Faculty of Business and Economics
The University of the South Pacific
September, 2015
ii
DECLARATION
I, Kushneel Avneet Prakash, declare that this thesis is my own work and that, to the best
of my knowledge, it contains no material previously published, or substantially
overlapping with material submitted for the award of any other degree at any institution,
except where due acknowledgement is made in the text.
Signature: Student ID No: S11057839
Name: Kushneel Avneet Prakash Date:
Statement by Supervisor
The research in this thesis was performed under my supervision and to my knowledge is
the sole work of Mr. Kushneel Avneet Prakash.
iii
DEDICATION
This thesis is dedicated to my beloved parents;
Mr. Satendra Prakash and Mrs. Arun Lata Prakash
Mummy and Papa – this is for you.
iv
ACKNOWLEDGEMENTS I would like to acknowledge the following persons who have been instrumental in the
completion of my thesis:
1. My principal supervisor: Associate Professor Dr. Dibyendu Maiti, who helped me
understand the core concepts and principles of macroeconomics in order to frame my
thesis. He has provided a lot of useful suggestions and insights into many
complicated issues dealt with in this thesis. Alongside his teaching and research
commitments, he has always been able to take out time to assist me in my research
progress with regular discussions. The guidance and wisdom provided by him during
the course of the thesis have been indispensable.
2. Former Head of School of Economics: Professor Saqib Jafarey (now a Professor of
Economics at the City University, London) for his motivation and consideration in
my teaching commitments during his tenure to ensure smooth progress of my thesis.
3. Visiting Professor to School of Economics in August 2014: Professor Anindya
Banerjee (currently a Professor of Econometrics at the University of Birmingham)
for his suggestion on the econometric analysis in the thesis.
4. The participants at the 55th New Zealand Association of Economists (NZAE)
conference in July 2014 at AUT University, Auckland for their comments and
feedback on a paper originating from the work of this thesis.
5. The Head of Department of Economics at the AUT University: Professor Tim
Maloney for his approval to work on my thesis for an extended period as a Visiting
Scholar in his department while I was in Auckland for the 55th NZAE conference.
6. My parents: for their unconditional support and motivation during the progress of
this thesis. I am blessed to have them who continuously push me to strive for
excellence. I also thank my brother, Ronil Prakash, and my friend, Reema Kumar,
for their understanding and patience during the writing of my thesis.
7. The two external examiners who provided very useful and encouraging comments on
the thesis.
v
ABSTRACT Over centuries, international trade has allowed countries to expand their product markets
and to derive welfare gains; it also exposes them to shocks in the global economy. Small
island developing economies, particularly in the Pacific region in order to deal with
external shocks find that many a times fiscal policies are quite exhaustive due to high
debt and monetary policies have been used restrictively after the global financial crisis,
resort to maintaining the exchange rate system fixed with the major trading partners to
use it as a shock absorber. The strategy of devaluation has often been employed in an
effort to withstand declining trade performance and foreign reserves in the country. The
effect of devaluation could be unfavourable immediately in the short-run but seem to be
beneficial in the medium-run and demonstrated as the J-curve phenomenon in the
traditional literature. In order to understand the resultant impact of devaluation on trade
performance many scholars have very often attempted to validate the J-curve
phenomenon. Therefore, this thesis uses Fiji as a case study to explore the effectiveness
of devaluation on the improvement of trade balance not only at aggregate level but also
at sectoral and bilateral level in the short– and in the long–run along. The J-curve
phenomenon is also examined in response to devaluation at aggregate, sectoral and
bilateral level.
This thesis uses vector error correction methodology (VECM) to carry out empirical
analysis using annual data over the 1975–2012 periods. Using the widely applied
reduced form of the trade balance model of Rose and Yellen (1989), the study models
trade balance, categorised by goods and services sectors, bilateral trade relationship with
its ten trade partners along with trade relationships at sectoral level. The results on the
aggregate goods trade balance suggest that currency devaluation in Fiji significantly
improves trade balance in the goods sector but the services sector is adversely affected.
However, the combined goods and services trade balance is significantly and positively
affected in the long–run. The results also indicate that domestic income has an adverse
impact while foreign income positively influences trade balance in Fiji. The J-curve
vi
phenomenon as such is valid only for the goods sector and the combined goods and
services sector trade. The services sector, on the other hand, exhibits inverse J-curve
phenomenon implying that improvements in service trade balance are more pronounced
in the short–run only. By further disaggregating the data at sectoral level, evidence
suggests that the trade balance for food, travel and transportation sectors are adversely
affected by devaluation in the long–run. The additional three export sectors and another
ten import sectors being modeled show mixed responses. The tests on the J-curve
phenomenon appears to be valid only for the travel sector using the impulse response
analysis. Advancing the analysis further, the results show significant improvements in
the bilateral goods trade balance of Fiji only with New Zealand and the USA in the
long–run. On the other hand, it has adverse impact on the bilateral trade performance
with Australia, China, Hong Kong, India, Malaysia, Singapore and the UK.
Additionally, evidence of the J-curve phenomenon is obtained in four out of the ten
trading partners which includes Japan, New Zealand, the UK and the USA.
Hence, from a policy perspective, the comprehensive analysis carried out at the
aggregate, sectoral and bilateral levels of trade performance in response to devaluation
to the policy makers that devaluation has worked moderately for Fiji to boost overall
trade performance in the country. It is also important to highlight that results have been
more pronounced for the goods sector rather than the services sector. Even maintaining
international competitiveness of the domestic sectors, improving productivity and
diversification of exports is of paramount importance to ensure sustained benefits from
price adjustments due to deliberate exchange rate reductions. With the center of global
trade slowly shifting to Asian countries, Fiji is well situated to benefit from increased
goods trade and particularly by tapping into the Asian tourist market.
vii
ABBREVIATIONS ACP African, Caribbean and Pacific Group of States
ADF Augmented Dickey–Fuller
AIC Akaike’s Information Criteria
ANZ Australia and New Zealand
ARDL Autoregressive Distributed Lag
DOLS Dynamic Ordinary Least Square
EPA Economic Partnership Agreement
EU European Union
FBOS Fiji Bureau of Statistics
FICs Forum Island Countries
FIML Johansen-Juselius Full Information Maximum Likelihood
FJ$ Fijian Dollar
FMOLS Fully Modified Ordinary Least Square
GATT General Agreement on Tariffs and Trade
GDP Gross Domestic Product
GEC Global Economic and Financial crisis
GETS General to Specific Method
GNI Gross National Income
HDI Human Development Index
IFS International Financial Statistics
IMF International Monetary Fund
IRF Impulse Response Function
IS-LM Investment-Savings and Liquidity preference-Money supply
IT Information Technology
KOF Konjunkturforschungsstelle
LDC Least Developed Countries
LM Lagrange Multiplier
M-L Marshall-Lerner
MSGTA Melanesian Spearhead Group Trade Agreement
viii
NBER National Bureau of Economic Research
NZ New Zealand
OECD Organisation for Economic Cooperation and Development
PACER Pacific Agreement on Closer Economic Relations
PAM Partial Adjustment Model
PICs Pacific Island Countries
PICTA Pacific Island Countries Trade Agreement
PNG Papua New Guinea
RBF Reserve Bank of Fiji
RMB Chinese Renminbi
SDR Special Drawings Rate
SIC Schwarz Information Criteria
SPARTECA South Pacific Regional Trade and Economic Cooperation Agreement
SVAR Structural Vector Auto-regression
TMNP Temporary Movement of Natural Persons
UK The United Kingdom
US$ United States Dollar
USA The United States of America
VAR Vector Auto-regression
VECM Vector Error Correction Model
WDI World Development Indicators
WTO World Trade Organization
ix
TABLE OF CONTENTS
CHAPTER 1 INTRODUCTION 1 - 15
Section 1.1 Introduction 1
Section 1.2 Context and overview of Fiji’s economy 3
Section 1.3 Objective of the study 6
Section 1.4 Theoretical framework and hypotheses 7
1.4.1 The trade balance and the real exchange rate: A theoretical background
7
1.4.2 Theory of the J-curve phenomenon 9
1.4.3 Hypotheses 11
Section 1.5 Potential contribution of this research 13
Section 1.6 Significance of the study 14
Section 1.7 Thesis organisation 15
Section 1.8 Concluding comments 15
CHAPTER 2 METHODOLOGY AND DATABASE 16 - 41
Section 2.1 Introduction 16
Section 2.2 The evolution of methodological approaches 16
Section 2.3 Testing the J-curve phenomenon on the trade balance 25
Section 2.4 The trade balance model 27
Section 2.5 The empirical model 30
Section 2.6 The VECM technique 34
Section 2.7 Sources of data 36
Section 2.8 Concluding comments 41
CHAPTER 3 FIJI ECONOMY OVERVIEW 42 - 62
Section 3.1 Introduction 42
Section 3.2 Overview of Fiji’s economic performance 42
3.2.1 Economic growth performance 43
3.2.2 Key sectoral performances 46
x
3.2.3 Investment and Savings 50
3.2.4 Monetary and Fiscal policies 51
Section 3.3 Trade Policies 54
3.3.1 Devaluation in Fiji 57
3.3.1.1 Exchange rate adjustment in Fiji – A brief history
57
3.3.1.2 Devaluation episodes in Fiji 58
3.3.1.3 Applicability of fixed exchange regime in Fiji
61
Section 3.4 Concluding comments 62
CHAPTER 4 AGGREGATE TRADE FLOWS IN RESPONSE TO CURRENCY DEVALUATION
63 - 109
Section 4.1 Introduction 63
Section 4.2 Literature review: Aggregate trade flows 64
4.2.1 On the aggregate trade flows and the J-curve phenomenon
65
4.2.2 On the impact of devaluation in the context of Fiji’s economy
69
4.2.3 Findings from the Literature review : Aggregate trade flows
78
Section 4.3 Trade in Fiji: Pace and Patterns 80
4.3.1 Trends in Fiji’s trade balance 80
4.3.2 Relative contribution of goods and services trade in Fiji
82
4.3.3 Trade response to devaluation in Fiji: A simple ‘before-after’ approach
84
Section 4.4 Empirical Analysis: Impact of currency devaluation on the performance of the aggregate trade balance in Fiji
89
4.4.1 Results of the unit root tests 90
4.4.2 Results of the cointegration tests 91
4.4.3 Estimates of the long–run elasticities 91
4.4.4 Estimates of the short run elasticities: Results from error correction models
98
4.4.5 Testing for the aggregate J-curve phenomenon 102
4.4.6 Highlights of the empirical analysis at aggregate level trade balances
107
Section 4.5 Concluding comments 108
xi
CHAPTER 5 SECTORAL AND BILATERAL LEVEL TRADE FLOWS IN RESPONSE TO CURRENCY DEVALUATION
110 - 183
Section 5.1 Introduction 110
Section 5.2 Literature review: Sectoral and Bilateral trade flows 111
5.2.1 On the sectoral trade flows and the J-curve phenomenon
111
5.2.2 On the bilateral trade flows and the J-curve phenomenon
117
5.2.3 On the impact of devaluation on sectoral and bilateral trade flows in Fiji
123
5.2.4 Findings from the Literature review : Sectoral and Bilateral trade flows
126
Section 5.3 Pace and Patterns of Sectoral and Bilateral trade flows in Fiji
128
5.3.1 Sectoral trade trend and patterns 129
5.3.1.1 Sectoral Performance 129
5.3.1.2 Sectoral trade response to devaluation in Fiji: A simple ‘before-after’ approach
134
5.3.2 Bilateral trade trend and patterns 139
5.3.2.1 With whom does Fiji trade? 139
5.3.2.2 Bilateral trade performance: Trade with major and emerging Asian trade partner countries
140
5.3.2.3 Bilateral trade response to devaluation in Fiji: A simple ‘before-after’ approach
142
5.3.2 Highlights from sectoral and bilateral trade patterns
146
Section 5.4 Empirical analysis: Sectoral and Bilateral trade 147
5.4.1 Results of the unit root tests 149
5.4.2 Results of the cointegration tests 149
5.4.3 Long–run elasticities 149
5.4.3.1 Sectoral trade analysis 150
5.4.3.2 Bilateral trade analysis 157
5.4.4 Short run elasticities – Results from error correction models
164
5.4.4.1 Sectoral trade analysis 164
5.4.4.2 Bilateral trade analysis 167
xii
5.4.5 Testing for the J-curve phenomenon 170
5.4.5.1 Sectoral J-curve phenomenon 170
5.4.5.2 Bilateral J-curve phenomenon 174
5.4.6 Highlights from the empirical analysis at sectoral and bilateral levels
181
Section 5.5 Concluding comments 182
CHAPTER 6 CONCLUSION AND POLICY RECOMMENDATIONS
184 - 191
Section 6.1 Introduction 184
Section 6.2 Major Findings 184
Section 6.3 Recommendations and policy implications 189
Section 6.4 Limitations 190
REFERENCES 192 - 203
APPENDICES 204 - 311
A Visitor arrivals in Fiji by country of residence, 1975–2012
204
B Percentage share of visitor arrivals in Fiji by country of residence, 1975–2012
205
C Variable definitions 208
D Results of unit root test for individual variables 213
E Results of cointegration test for various equations 218
F Results from Error Correction model (short-run dynamics)
223
xiii
List of Tables
3.1 Real GDP growth rates in the PICs, 1990–2015(f) 45
3.2 Sectoral contribution to real GDP in Fiji, 1970–2013 46
3.3 Exports of Fiji’s key activitities 47
3.4 Visitor arrivals in Fiji by country, 1970–2014(%) 50
3.5 Level of Investment and Savings in Fiji as a % of GDP, 1970–2014 51
3.6 Interest rates (%) in Fiji, 1995–2013 52
3.7 Inflation rates (%) and level of foreign reserves in Fiji, 1980–2014 53
3.8 Budget deficits and debt levels in Fiji, 2000–2014(p) 54
3.9 Devaluation episodes in Fiji (1987–2009) 58
4.1 Short–, medium– and long–run response of major trade indicators to 1% devaluation in Fiji
87
4.2 Estimates of long–run coefficients of the trade balance models 92
4.3 Estimates of long–run coefficients of the export and import models 93
4.4 Short–run coefficient estimates of real exchange rates in the trade balance models
99
4.5 Short run coefficient estimates of real exchange rates in the export and import models
100
4.6 Summary of the assessment on the J-curve assessment in Fiji on the aggregate trade performance
107
5.1 Response of major goods sectors trade performance to 1% devaluation in Fiji
136
5.2 Response of major services sectors trade performance to 1% devaluation in Fiji
137
5.3 Fiji’s trading partners’ composition in 2012 139
5.4 Response of Bilateral trade performance with major trade partner countries to 1% devaluation in Fiji
143
5.5 Response of bilateral trade performance with emerging Asian trade partner countries to 1% devaluation in Fiji
145
5.6 Estimates of long–run coefficients of major sectoral trade balance models
150
5.7 Estimates of long–run coefficients of major services sector export and import models
152
5.8 Estimates of long–run coefficients of goods sector export and import models
155
5.9 Estimates of long–run coefficients of trade balance models with major bilateral trade partners
157
5.10 Estimates of long–run coefficients of export and import models 159
xiv
with major bilateral trade partners 5.11 Estimates of long–run coefficients of trade balance model with
emerging Asian trade partners 161
5.12 Estimates of long–run coefficients of export and import models with emerging Asian trade partners
162
5.13 Short–run coefficient estimates of real exchange rates in the sectoral trade balance models
165
5.14 Short–run coefficient estimates of real exchange rates in the major trade partners’ trade balance models
168
5.15 Short–run coefficient estimates of real exchange rates in emerging Asian trade partners’ trade balance models
169
5.16 Results on the sectoral level J-curve phenomenon 170
5.17 Results on the bilateral level J-curve phenomenon 174
Appendix
A Visitor arrivals in Fiji by country of residence, 1975–2014 204
SR1 Short–run coefficient estimates of real exchange rates in Fiji’s sectoral exports model
224
SR2 Short–run coefficient estimates of real exchange rates in Fiji’s sectoral imports model
225
SR3 Short–run coefficient estimates of real exchange rates in Fiji’s sectoral imports model
226
SR4 Short–run coefficient estimates of real exchange rates in Fiji’s major bilateral trade partners export model
227
SR5 Short–run coefficient estimates of real exchange rates in Fiji’s major bilateral trade partners import model
228
SR6 Short–run coefficient estimates of real exchange rates in Fiji’s emerging Asian bilateral trade partners export model
229
SR7 Short–run coefficient estimates of real exchange rates in Fiji’s emerging Asian bilateral trade partners import model
230
SA1 Short–run coefficient estimates of Fiji’s goods trade balance model, 1975–2012
231
SA2 Short–run coefficient estimates of Fiji’s services trade balance model, 1975–2012
233
SA3 Short–run coefficient estimates of Fiji’s goods and services trade balance model, 1975–2012
234
SA4 Short–run coefficient estimates of Fiji’s food sector trade balance model, 1975–2012
235
SA5 Short–run coefficient estimates of Fiji’s food sector trade balance model, 1975–2012
236
SA6 Short–run coefficient estimates of Fiji’s transport services sector trade balance model, 1975–2012
238
SA7 Short–run coefficient estimates of Fiji’s Goods Trade Balance model with Australia, 1975–2012
239
SA8 Short–run coefficient estimates of Fiji’s goods trade balance model 240
xv
with New Zealand, 1975–2012 SA9 Short–run coefficient estimates of Fiji’s goods trade balance model
with Japan, 1975–2012 241
SA10 Short–run coefficient estimates of Fiji’s goods trade balance model with the USA, 1975–2012
242
SA11 Short–run coefficient estimates of Fiji’s goods trade balance model with the UK, 1975–2012
243
SA12 Short–run coefficient estimates of Fiji’s goods trade balance model with Singapore, 1975–2012
245
SA13 Short–run coefficient estimates of Fiji’s goods trade balance model with China, 1975–2012
246
SA14 Short–run coefficient estimates of Fiji’s goods trade balance model with Malaysia, 1975–2012
247
SA15 Short–run coefficient estimates of Fiji’s goods trade balance model with India, 1975–2012
248
SA16 Short–run coefficient estimates of Fiji’s goods trade balance model with Hong Kong, 1975–2012
250
SA17 Short–run coefficient estimates of Fiji’s goods export model, 1975–2012
252
SA18 Short–run coefficient estimates of Fiji’s domestic export model, 1975–2012
253
SA19 Short–run coefficient estimates of Fiji’s services export model, 1975–2012
254
SA20 Short–run coefficient estimates of Fiji’s goods and services export model, 1975–2012
255
SA21 Short–run coefficient estimates of Fiji’s export of food model, 1975–2012
256
SA22 Short–run coefficient estimates of Fiji’s export of sugar model, 1975–2012
258
SA23 Short–run coefficient estimates of Fiji’s export of fish model, 1975–2012
259
SA24 Short–run coefficient estimates of Fiji’s export of gold model, 1975–2012
260
SA25 Short–run coefficient estimates of Fiji’s export of travel services model, 1975–2012
262
SA26 Short–run coefficient estimates of Fiji’s export of transportation services model, 1975–2012
264
SA27 Short–run coefficient estimates of Fiji’s goods export model with Australia, 1975–2012
265
SA28 Short–run coefficient estimates of Fiji’s goods export model with New Zealand, 1975–2012
266
SA29 Short–run coefficient estimates of Fiji’s goods export model with Japan, 1975–2012
267
SA30 Short–run coefficient estimates of Fiji’s goods export model with the USA, 1975–2012
268
SA31 Short–run coefficient estimates of Fiji’s goods export model with 269
xvi
the UK, 1975–2012 SA32 Short–run coefficient estimates of Fiji’s goods export model with
Singapore, 1975–2012 271
SA33 Short–run coefficient estimates of Fiji’s goods export model with China, 1975–2012
272
SA34 Short–run coefficient estimates of Fiji’s goods export model with Malaysia, 1975–2012
274
SA35 Short–run coefficient estimates of Fiji’s goods export model with India, 1975–2012
275
SA36 Short–run coefficient estimates of Fiji’s goods export model with Hong Kong, 1975–2012
276
SA37 Short–run coefficient estimates of Fiji’s goods import model, 1975–2012
277
SA38 Short–run coefficient estimates of Fiji’s services import model, 1975–2012
278
SA39 Short–run coefficient estimates of Fiji’s goods and services import model, 1975–2012
279
SA40 Short–run coefficient estimates of Fiji’s import of food model, 1975–2012
280
SA41 Short–run coefficient estimates of Fiji’s import of fuel model, 1975–2012
282
SA42 Short–run coefficient estimates of Fiji’s import of manufactured goods model, 1975–2012
284
SA43 Short–run coefficient estimates of Fiji’s import of crude oil model, 1975–2012
285
SA44 Short–run coefficient estimates of Fiji’s import of textile model, 1975–2012
286
SA45 Short–run coefficient estimates of Fiji’s import of machinery and transport equipment model, 1975–2012
287
SA46 Short–run coefficient estimates of Fiji’s import of tobacco and beverage model, 1975–2012
288
SA47 Short–run coefficient estimates of Fiji’s import of chemical model, 1975–2012
290
SA48 Short–run coefficient estimates of Fiji’s import of oil and fats model, 1975–2012
291
SA49 Short–run coefficient estimates of Fiji’s import of miscellaneous manufactured goods model, 1975–2012
292
SA50 Short–run coefficient estimates of Fiji’s import of travel services model, 1975–2012
293
SA51 Short–run coefficient estimates of Fiji’s import of transportation services model, 1975–2012
295
SA52 Short–run coefficient estimates of Fiji’s goods import model with Australia, 1975–2012
296
SA53 Short–run coefficient estimates of Fiji’s goods import model with New Zealand, 1975–2012
298
SA54 Short–run coefficient estimates of Fiji’s goods import model with 300
xvii
Japan, 1975–2012 SA55 Short–run coefficient estimates of Fiji’s goods import model with
the USA, 1975–2012 301
SA56 Short–run coefficient estimates of Fiji’s goods import model with the UK, 1975–2012
302
SA57 Short–run coefficient estimates of Fiji’s goods import model with Singapore, 1975–2012
304
SA58 Short–run coefficient estimates of Fiji’s goods import model with China, 1975–2012
306
SA59 Short–run coefficient estimates of Fiji’s goods import model with Malaysia, 1975–2012
308
SA60 Short–run coefficient estimates of Fiji’s goods import model with India, 1975–2012
310
SA61 Short–run coefficient estimates of Fiji’s goods import model with Hong Kong, 1975–2012
311
xviii
List of Figures
1.1 The J-curve 10
3.1 Fiji’s real GDP growth rates, 1975–2015(p) 44
3.2 Fiji’s KOF Economic and Trade index, 1970–2011 56
3.3 Real effective exchange rate ( rE ) index in Fiji, 1975–2012 60
4.1 Fiji’s exports and imports of goods and services, 1975–2012 81
4.2 Goods and Services trade balance in Fiji, 1975–2012 82
4.3 Evolution of goods and services export shares in Fiji, 1975–2012 83
4.4 Evolution of goods and services import shares in Fiji, 1975–2012 84
4.5 Response of the goods trade balance to generalised one standard deviation innovation in the real exchange rate
103
4.6 Response of the services trade balance to generalised one standard deviation innovation in the real exchange rate
104
4.7 Response of the goods and services trade balance to generalised one standard deviation innovation in the real exchange rate
105
5.1 Composition of Fiji’s goods exports sectors in 2012 129
5.2 Fiji’s trade trend of her major goods exports sectors, 1975–2012 130
5.3 Composition of goods imports sectors in 2012 131
5.4 Fiji’s trade trend of her major goods imports sectors, 1975–2012 132
5.5 Composition of services exports sectors in 2012 132
5.6 Composition of services imports sectors in 2012 133
5.7 Fiji’s trade trend of her major services exports and imports sectors, 1975–2012
134
5.8 Fiji’s trend of trade balance with her major trade partner countries, 1975–2012
141
5.9 Fiji’s trend of trade balance with her emerging Asian trade partner countries, 1975–2012
142
5.10 Response of the food sector trade balance to generalised one standard deviation innovation in the real exchange rate
171
5.11 Response of the travel services sector trade balance to generalised one standard deviation innovation in the real exchange rate
173
5.12 Response of the transportation services sector trade balance to generalised one standard deviation innovation in the real exchange rate
173
5.13 Response of the trade balance between Fiji and Australia to generalised one standard deviation innovation in the real bilateral
175
xix
exchange rate 5.14 Response of the trade balance between Fiji and New Zealand to
generalised one standard deviation innovation in the real bilateral exchange rate
176
5.15 Response of the trade balance between Fiji and Japan to generalised one standard deviation innovation in the real bilateral exchange rate
177
5.16 Response of the trade balance between Fiji and the UK to generalised one standard deviation innovation in the real exchange rate
177
5.17 Response of the trade balance between Fiji and the USA to generalised one standard deviation innovation in the real bilateral exchange rate
178
5.18 Response of the trade balance between Fiji and Singapore to generalised one standard deviation innovation in the real exchange rate
179
5.19 Response of the trade balance between Fiji and Malaysia to generalised one standard deviation innovation in the real exchange rate
179
5.20 Response of the trade balance between Fiji and India to generalised one standard deviation innovation in the real exchange rate
180
5.21 Response of the trade balance between Fiji and Hong Kong to generalised one standard deviation innovation in the real exchange rate
180
5.22 Response of the trade balance between Fiji and China to generalised one standard deviation innovation in the real exchange rate
180
1
CHAPTER 1
INTRODUCTION
1.1 Introduction In the globalised era, where economies are opening themselves to a wider range of
international markets, a country’s trading with the rest of the world plays an integral part
in its development. As goods and services leave national boundaries, the issue of
exchange rates becomes of paramount importance. A number of countries in the world
still use fixed exchange rates deliberately so that it can be used strategically to improve
the trade and current account balances and to derive its favourable impact on the
economic growth when needed.
Very often, the resultant impact in the economy has been found to be adverse and
minimal, mainly on the trade or current account and on fiscal balances. If so, one round
of currency devaluation1 could compel the following of the same in the immediate future
to deal with these balances. According to the contemporary research, since the
devaluation has a tendency to push up the inflationary pressure, it might bring an
‘exchange rate management crisis’ to the economy. As a result, many countries in Asia,
Europe and Latin America have changed their exchange rate system from fixed to
flexible regimes. Some have even adopted the use of a single currency, as in the case of
the Euro zone, to better manage their exchange rate based stabilization policies.
In spite of the plethora and increasing number of studies on the relationship between
exchange rate and trade balance2, the ultimate impact of changes in exchange rate on
trade performance is still an open and controversial issue. The dominant view up until
the late 1970s was that the devaluation improves trade balance and consequently
enhances employment and output. However, Acar (2000) argues that this conventional 1 It is the official lowering of the value of a country’s currency within a fixed exchange rate system, by which the monetary authority formally sets a new fixed rate with respect to a foreign reference currency. 2 See Bahmani-Oskooee and Ratha (2004) for a detailed review of the literature on the relationship between exchange rate and trade performance.
2
idea has been challenged by the end of 1970s and an alternative approach has emerged
which contends that devaluation could be contractionary, especially in developing
countries. This has been argued largely on the grounds that developing countries depend
heavily on import of capital and intermediate goods for domestic production.
For the last few decades, numerous studies in the context of developing and developed
countries have investigated the relationship among currency devaluation or depreciation,
trade balance and economic growth. Recent research also focuses on how the sectoral
and bilateral level trade flows along with the aggregate trade flows respond to currency
devaluation and depreciation in an economy. Given the fact that countries differ with
respect to population size, geography, resource endowment, economic structures and
social structures along with other factors, there is no general consensus in the existing
literature on the effectiveness of devaluation on aggregate, sectoral and bilateral level
trade flows.
Currency devaluation, which is often implemented as a policy instrument in many
developing economies to improve trade balance, by and large follows a time delay to
portray signs of improvements, if any. Bahmani-Oskooee and Ratha (2004) note that it is
due to time lag structure; devaluation initially worsens trade balance before improving it
at a later stage. This results in a pattern of trade balance that is similar to the letter J.
This is known in the literature as the J-curve3 phenomenon. The empirical literature on
the J-curve phenomenon is vast; however, the general consensus in the literature is that
the response of trade balance to currency devaluation or depreciation does not always
follow any specific pattern.
There have been experiences which note that currency devaluation or depreciation leads
to an improvement in the trade balances, some noted that it actually worsens trade
balance while some have not been able to establish any relationship between the trade
balance and exchange rate in an economy. The results, however, are country specific;
differ due to differences in the study period and due to different methodology being used 3 Further detailed discussion on the J-curve phenomenon is done in Section 1.3.3.
3
to investigate the relationship between exchange rate and trade balance. There are many
studies in the literature that have found the J-curve phenomenon of trade balance after
devaluation or depreciation (such as Noland (1989), Gupta-Kapoor and Ramakrishnan
(1999), Arora et al. (2003), Yusoff (2010), Bahmani-Oskooee and Hegerty (2011) and
Bahmani-Oskooee et al. (2013)) while some studies failed to find evidence of the J-
curve phenomenon (such as Bahmani-Oskooee and Ratha (2004), Yazici (2010), Yusoff
(2010), Bahmani-Oskooee and Gelan (2012)). Additionally, there are some studies
which have found mixed results (such as Lal and Lowinger (2002), Bahmani-Oskooee
and Kutan (2009), Kalyoncu et al. (2009)).
This critical issue of currency devaluation has also long been a major discussion on the
economic and political agendas of Less Developed Countries (LDCs, hereafter). The
Pacific region as well is no exception as most of the countries4 in the region, by and
large still follow the fixed exchange rate system. Hence, these countries very often
devalue their currency in order to improve its international competitive position and to
derive its resultant impact on the economy. The recent Global Economic and Financial
Crisis (GEC, hereafter) has added to their existing economic problems and has situated
them to devalue once again.
1.2 Context and overview5 of Fiji’s Economy Fiji’s four devaluations within the last three decades make it the classic example for
studying the impact of devaluation on trade performance in the Pacific islands context.
The two devaluations in 1987 totaled 33%, that in 1998 a further 20% and in 2009 a
further 20%.
Fiji’ economy, since independence in 1970, has experienced episodes of low economic
growth, increasing debt levels and annual budget deficits against a background of
4 The South Pacific countries with independent currencies that still follow fixed exchange rate regimes are Fiji, Samoa, Tonga, Solomon Islands and Vanuatu. 5 A detailed discussion on Fiji’s economic overview is carried out in Chapter 3.
4
political instability and natural disasters that has severely affected its growth trend. The
record presents a mixed pattern of growth with episodes of high and low growth rates
and at times a negative growth in output. The economy has traditionally been dominated
by agriculture particularly sugarcane, mining and fisheries. In the recent past, the service
sector, most notably the tourism industry, has become the largest foreign income earner.
Nevertheless, the sugar industry, along with garments and fisheries, continue to be the
important industries in the country providing export earnings and livelihood for many in
the country. On the monetary policies, it is noted that monetary discipline has been well
maintained by the Reserve Bank of Fiji (RBF, hereafter). In particular, the foreign
reserves and inflation rates in recent years have been maintained at acceptable levels
despite political instability and expansionary fiscal policies.
Additionally, in an effort to foster trade integration into the global economy, Fiji since
1989 has become party to several bilateral, regional and international trade agreements.
Trade liberalisation has continued to be an important agenda for Fiji’s economic reforms
and policies aimed at improving trade performance is gaining dominance. However,
Fiji’s overall trade balance on goods and services has been consistently in deficit for a
relatively long period of time. The goods sector has continuously faced an increasing
trade deficit while the services trade performance has remained in surplus balance since
independence. By and large, the services sector performance has remained stable with
some signs of modest improvements in the recent past.
Despite the trade surplus created by the services sector, it has still not been favourable
enough to offset the deficit created by goods sector in the economy. This resulted in the
overall trade deficit in Fiji. This has often put pressure on the economy’s level of foreign
reserves to meet the ever-increasing import bills. Hence, at times when monetary
policies using interest rates have been difficult to reduce further and the fiscal position
has been tight to maintain international competitiveness, the RBF has used currency
devaluation as the last option to gain competitiveness in the global market.
5
The various devaluation episodes in Fiji have prompted scholarly debate on the
appropriateness of devaluation and investigation of its consequences for trade and output
in the country (see Reddy (1997), Chand (1998), Jayaraman (1999), Narayan and
Narayan (2004a, 2004b, 2005), Singh (2006), Narayan and Narayan (2007), Narayan
(2013)). The literature as such on this issue in the context of Fiji’s economy is still small
and narrow. Though there are relatively few studies analysing the impact of exchange
rate on aggregate trade balance in Fiji, the general consensus regarding the impact of
devaluation has been that devaluation improves exports and reduces imports in both
short and long run. Nonetheless, the one study analysing the impact on the trade balance
in Fiji has found evidence of the J-curve phenomenon (see Narayan and Narayan
(2007)).
It has been noted that none of the studies in the context of Fiji has empirically
investigated the relationship between devaluation and trade performance incorporating a
dataset that extends beyond the last, 2009, devaluation in Fiji. Bahmani-Oskooee and
Brooks (1999) and Bahmani-Oskooee and Xu (2012) also point out that using aggregate
trade data to investigate the relationship with exchange rate leads to biased results. They
argue that this aggregation biasness results when significant trade flows of a devaluing
country with its one partner could be offset by an insignificant relation that might exist
with another trading partner country. Therefore, it is argued that the difference in the
impact of devaluation on aggregate trade performance would not necessarily be the same
for different commodities and trade partners of a country.
More than often, the consensus in the literature regarding the impact of devaluation on
Fiji’s economy is that devaluation has been effective. Nonetheless, the question still
remains that if it is effective then why still there has been a number of devaluation in the
last three decades. This is not clear in the literature. Explanations for recurring
devaluations that currency devaluation raises inflation in the economy might not be
sufficient.
6
Consequently, this thesis will make a modest attempt to investigate empirically the
impact of devaluation on trade balance disaggregated into sectoral and bilateral trade
performance. Using recent dataset that extends beyond the 2009 devaluation, analysis to
understand the short– and long–run dynamics of devaluation in Fiji is undertaken.
Though devaluation can lead to increase in exports and consequently income, increase in
income will result in an increase in imports for an import dependent economy, Fiji.
Hence, in the relatively long-run, devaluation may not be effective.
Although there might be skepticism surrounding the study by analysing Fiji’s
experience, it is ascertained that despite its size and economic structure, some useful
lessons can still be learnt from its experience. It is also well recognised in academic
research that the validity of any theory or concept achieves greater acceptance
universally if it is empirically tested in economies of various structure and sizes. This
explains the interest in studying the Fiji’s economy.
1.3 Objective of the study The study has two broad objectives:
(i) to investigate the effectiveness of devaluation on the improvement of trade balances
not only at aggregate level but also at sectoral and bilateral levels in the short to long
run.
(ii) to investigate the existence of the J-curve phenomenon of the trade balance in
response to devaluation both at the aggregate and disaggregate level of sectors and
country level trade.
Policy implementations on the basis of insufficient evidence of a country’s response in
its macroeconomic indicators could generate significant drawbacks and doubts on such
policies. Recognition of this fact underlies the decision to provide short and long run
7
estimates of exchange rate and trade balance along with the test of the J-curve
phenomenon at aggregate, sectoral and bilateral level trade performances.
1.4 Theoretical framework and hypotheses 1.4.1 The trade balance and the real exchange rate: A theoretical background
The impact of currency devaluation on the trade balance of a devaluing country is often
analysed using the Marshall–Lerner (M–L, hereafter) condition. The M–L condition,
named after the work of Alfred Marshall and Abba Lerner in the early 1900s, provides
precise explanation under which currency devaluation or depreciation is expected to
improve a country’s trade balance. This condition in its brief description states that since
a country’s trade balance is determined by its value of exports less its value of imports, a
decline in the value of a country’s currency increases the exportable quantities and
reduces importable quantities. However, the M–L condition states that the resulting
impact on trade balance would ultimately depend on how much the quantity of exports
and imports responds to changes in the relative price level.
The relationship of currency devaluation or depreciation to trade and output is analysed
through various channels in the literature. It is worth noting that among several scholars
Magee (1973), Himarios (1985), Bahmani-Oskooee and Miteza (2003) have documented
several channels through which devaluation might have favourable impact on trade
balance. On the other hand, Cooper (1971), Krugman and Taylor (1978), Edwards
(1986), Acar (2000) and Bahmani-Oskooee and Miteza (2003) are some of the studies
that have outlined several demand and supply side channels that may create reduction in
employment, trade performance and output after devaluation.
The conventional idea behind devaluation being expansionary is based on the
proposition that devaluation leads to expenditure switching from import oriented
consumption to consumption of domestically produced goods. It is anticipated that as a
country devalues its currency, domestic exports become cheaper relative to goods
produced by trading partners’. This results in an increase in domestic quantity demanded
8
by foreigners. On the other hand, the reduction in the value of the domestic currency
makes the price of imports rise soon after devaluation. This causes domestic buyers to
pay more for the imports making them less attractive. This anticipated response by
domestic and foreign buyers is expected to raise domestic demand for exports and
reduce demand for imported items. The increasing demand for exportable commodities
induces the domestic firms to expand their production, employ more resources and
improve productivity. As a result, it helps stimulate employment, improves trade balance
and consequently derives economic gains in the economy.
In contrast to the idea presented above, there are equally strong arguments opposing this
view. Krugman and Taylor (1978) argue that via the import cost channel if a country has
an initial trade deficit, the effect of devaluation on aggregate demand will be negative.
They argue that following devaluation, imports become expensive in terms of domestic
currency which requires more domestic resources to meet the import costs. This results
in an increasing trade deficit even further. Acar (2000) argues that as a consequence of
devaluation, prices to be paid for traded goods will increase relative to non-traded goods.
He stresses that this ultimately leads to an increase in general price level and emphasises
the fact that the larger the share of traded goods in consumption, the greater is the impact
of price rise.
Additionally, countries with large amounts of debt are forced to accept an increased
burden of total debt and service payments that are denominated in foreign currency
(Cooper, 1971). The repayment of debt after devaluation requires more domestic
currency to meet the debt obligation. This means that the public and private sectors need
extra resources for debt servicing that could have been used in domestic consumption
and capital expenditure. This results in net reduction of total output (Bahmani-Oskooee
and Miteza, 2003). Krugman and Taylor (1978) and Edwards (1986) also argue that for
most of the developing countries in the world, the production process generally includes
raw materials, capital and intermediate goods which are by and large imported.
Thereafter, a rise in price of imported production components leads to the overall
9
increase in production costs. This translates in contraction of total supply and rising
unemployment (Bruno (1979) and Hanson (1983)).
Even if exports increase gradually in the short to medium run after devaluation, this will
increase income resulting in increase in imports. Hence, imports might accelerate for an
import dependent economy at a faster pace than exports resulting in widening gap of the
trade deficit. Branson (1986) and Van Wijnbergen (1986) also add that devaluation may
lead to decline in investment in the economy. They argue that since most of the time in
developing countries new investments consist of imported capital goods, a real
devaluation would make capital investments more costly. This in turn might reduce new
investments and aggregate demand in the economy.
This backdrop on the theoretical background on devaluation reaffirms the notion that
there is no general consensus regarding the impact of devaluation on trade performance
and output. The theory makes it clear that the supply and demand side channels in a
devaluing economy play an important role in determining the ultimate impact of
devaluation on the economy.
1.4.2 Theory of the J-curve phenomenon
Devaluation, which in many developing economies is generally implemented as a policy
instrument to improve trade balance, by and large requires a time delay to show signs of
improvements, if any. Bahmani-Oskooee and Ratha (2004) note that because of the time
lag structure, devaluation initially worsens trade balance before improving it at a later
stage. This results in a pattern that is similar to the letter J and is known in the
economics literature as the J-curve phenomenon6.
Since the initial work by Magee (1973) on the J-curve phenomenon of devaluation,
several studies have attempted with different techniques, models and countries to
validate the presence of the phenomenon. One explanation for the J-curve phenomenon
6 See Bahmani-Oskooee and Ratha (2004) for a detailed review of the studies validating the J-curve phenomenon of currency depreciation or devaluation in developed and developing countries.
10
is that the prices of imports rise soon after devaluation and depreciation but quantities
take time to adjust to the new condition. This is because the current imports and exports
are based on older orders placed some time previously. Junz and Rhomberg (1973) have
identified five lag periods- recognition, decision, delivery, replacement and production
lag- which leads to the J-curve phenomenon on trade balance after devaluation.
Following devaluation, the traders take time to realise the fact that there is a change in
market competitiveness and this may take even longer to be recognised in international
trading markets. This is known as the recognition lag. Then, after realising that the initial
situation has changed, there is some time taken to make decisions on the possibilities of
opening new business ventures and entering into new contracts. This is known as the
decision lag. Thereafter, it takes time before new payments are made for orders that
were placed soon after the change in relative prices. This is known as the delivery lag.
Purchase of new inventories and materials may also be delayed to allow previously
bought materials to be used, in the replacement lag. As a final stage, there is a
production lag after which the producers become convinced that the existing market
conditions will provide profitable outcomes. Figure 1.1 shows a theoretical behaviour of
net exports due to these adjustment lags by plotting the evolution of trade balance
against time in response to currency depreciation or devaluation in an economy.
Figure 1.1 The J-curve phenomenon
11
Specifically, in the first few periods following currency devaluation, the effect of the
devaluation is likely to be reflected more in price changes than in the changes of export
and import quantities (Junz and Rhomberg, 1973). As a result, imports become
expensive while the exports become cheaper for foreigners. However, the quantities of
imports and exports are likely to adjust slowly given the adjustment lags as discussed
earlier. As a result, the depreciation of the currency may lead to an initial deterioration
of the trade balance as neither exports nor imports adjust much at the initial periods.
This is represented by the movement of the trade balance from its pre-devaluation level
of OA to OB .
As time passes by, the effects of the change in the relative prices of both exports and
imports become stronger and the consumers and producers end up realising the effects of
change in the exchange rate. As a result, it is anticipated that domestic consumers start
demanding less of costly imports while foreigners begin to demand more of cheap
domestic exports. As a consequence, the quantitative response of exports and imports
tend to outweigh the adverse price effect on trade. Eventually, the result is an
improvement of the trade balance from B to C and beyond. This adjustment process
with bit of imagination resembles letter J showing initial decrease followed by an
increase in the trade balance.
1.4.3 Hypotheses Fiji, on various occasions has implemented currency devaluation as a policy tool to
improve its trade position with the rest of the world. As briefly discussed earlier, prior
similar studies in this context have found mixed results on the impact of currency
devaluation on trade and output performance in Fiji. Traditionally, it is expected that
devaluation of the currency will encourage exports and discourage imports, leading to
improvements in aggregate trade balance. However, it is also possible that devaluation
may actually worsen trade balance due to the resultant impact on domestic inflation.
Devaluation might also not be effective due to the lack of capacity in the economy to
12
meet additional export demand, heavy dependence on imports and lack of import
substitutability.
On the disaggregated goods and services sector, it is likely that goods sector would have
high import substitutability properties than the services after devaluation. As import
prices of commodities such as food and basic manufacturing items rise after devaluation,
some of these products can be produced and substituted for domestically produced
products. Additionally, demand for export of local food products, sugar and fish is also
expected to increase in the trading partner markets due to fall in relative prices. This can
result in favourable impact of devaluation on the goods trade balance in Fiji.
On the other hand, the services sector, particularly tourism services which makes up the
largest component of services trade in Fiji, might respond differently. Despite being the
largest foreign exchange earner in the economy, the tourism sector might not reap the
full benefits of devaluation. Even though the sector continues to experience increase in
tourist arrival numbers7 over the years, the real earnings from export of tourism services8
has increased only to some extent in the country. This is largely because the sale of
tourism services in Pacific countries including Fiji is often dependent on tourism
packages sold by tourism agents in overseas markets. Many a times, the prices of
package vacations are set in foreign currency by tour operators which results in the
tourists not directly benefiting from currency devaluation in the country (Culiuc, 2014).
The holiday and tour packages are also often discounted which coupled with devaluation
results in the decline in real services export receipts.
Additionally, travel by locals for educational and medical purposes to overseas is not
expected to decline significantly despite the increase in air fares after devaluation. As a
result, import of services has experienced steeper increases than services exports over
the last three decades9. Therefore, this can result in adverse consequences for the
services trade performance in Fiji. On the aggregate front, depending on the impact of 7 See Appendix A for detailed breakdown of tourist arrival numbers by major countries over the years. 8 See Figure 5.7 for real earnings on export of travel services. 9 See Figure 5.7 for real earnings on import of travel services.
13
devaluation on disaggregated goods and services sectors and its share on total trade, this
will determine the ultimate impact on the aggregate trade performance.
Moreover, the currency devaluation will make domestic currency weaker, thereby
pushing the price of imports higher. This for an import dependent economy will require
more to be paid for imports due to relative price changes. Though an equally possible
argument is that due to substitution of imports with local production, an increase in total
exports would translate in to increased income in the economy. Thus, growth in exports
accompanied by continuing dependence on imports can over time cause deterioration of
the trade performance in the economy.
On the basis of the brief arguments presented in this section, it is hypothesised that:
i. The effect of currency devaluation in Fiji has favourable impact on the trade balance
mainly because of the domestic substitutability in the goods sectors which is not
necessarily achieved in the service sectors.
ii. The favourable effect depends on the relative expansion of exportable compared to
importable as the imports continue to rise because of the growing basic consumption
needs in the country.
iii. The effect of devaluation on the trade balance does not necessarily reveal strong
evidence of the J-curve phenomenon because of increased imports with the growth
of exports and income overtime for heavily import dependent economy.
1.5 Potential contribution of this research Previous studies on the issue of devaluation and its impact on trade performance in the
context of Fiji are few and relatively narrow in scope. Although, some existing studies
have attempted to analyse the impact of devaluation on a range of macroeconomic
indicators (inflation, trade balance and economic growth), no study as at yet has
incorporated a recent dataset to include all of Fiji’s devaluation episodes as part of their
analysis at least over the last three decades. Nor has any particular study yet attempted to
14
do a detailed analysis of the impact of Fiji’s currency devaluation on its trade
performance disaggregated by sectoral and bilateral trade.
This study attempts to contribute to the body of knowledge on exchange rate and trade
balance literature in the context of Fiji. More specifically, it draws a special line to
distinguish this current work by incorporating data for analysis that include all the four
devaluation episodes in Fiji as well as extending beyond 2009. The current study
analyses how the major trade components at aggregate, sectoral and bilateral level
respond to each devaluation episode in years immediate and later to the devaluation. The
study also disaggregates total trade into goods and services sectors along with
disaggregation into major sectoral trade.
Along with contributing to the literature at aggregate and sectoral trade level, impact of
devaluation on the bilateral trade with Fiji’s major and emerging Asian trade partner
countries are also undertaken. These trading partners include Australia, China, Hong
Kong, India, Japan, Malaysia, New Zealand, Singapore, the UK and the USA.
Additionally, the important contribution also lies in validating the presence of the J-
curve phenomenon using three methods of analysis in one study at aggregate, sectoral
and bilateral trade balance for a developing Pacific island country.
1.6 Significance of the study It is ascertained that the contribution of this study would be useful in formulating
exchange rate based policies to enhance trade performance in Fiji. At the same time, the
study would enable better understanding of the relationship between exchange rate and
trade flows in the context of yet another developing country. Subsequently, it is
anticipated that the research will bring to the surface some important policy
recommendations for the Reserve Bank of Fiji and the Fiji Government to assess their
stand on devaluation in the country.
15
The study would also give an indication to existing and potential investors as to which
particular sectors are deriving benefits after devaluation in the country. More
specifically, the study will identify the particular industries in which opportunity lies for
more investment to derive increased gains. While the empirical findings will directly
benefit the Fiji’s economy under study, the methodology adopted can be usefully applied
to understand the same for other developing countries including those of the Pacific
region and other small states.
1.7 Thesis organisation This thesis is organised in six chapters. Following this introductory chapter, the
description of methodology and the database is in Chapter 2. Chapter 3 presents a brief
but well-rounded overview of Fiji’s economy and its recent economic performance, with
particular reference to trade and devaluation in Fiji. Chapters 4 and 5 presents the
literature reviews on the J-curve phenomenon and the relationship of exchange rate with
aggregate, sectoral and bilateral trade performance. Empirical estimations and
robustness tests are also performed and reported in these two chapters. A detailed
summary of findings, recommendations, policy implications and limitations of the study
are discussed in the final Chapter 6.
1.8 Concluding comments This chapter is an introductory chapter of the thesis. It highlights the important issue of
currency devaluation and how it plays a role in the trade performance of a country. After
providing a theoretical background on the issue and the objectives of the study, it
outlines the direction and scope of the present study. The following chapter discusses the
application of the methodology and database employed in the study.
16
CHAPTER 2
METHODOLOGY AND DATABASE
2.1 Introduction
The data quality and appropriateness of the methodology used in a study play a
significant role in determining its final outcomes and policy recommendations. This
chapter of the thesis describes the variables used and their sources followed by
methodological approach.
This chapter is organised as follows. Section 2.2 examines different approaches used in
the literature to investigate how exchange rate affects economy’s trade balance and its
output. Various approaches widely used to validate the presence of the J-curve
phenomenon are examined in Section 2.3. Section 2.4 of the chapter presents derivation
of the trade balance model. Section 2.5 presents and discusses the empirical
specification and various statistical tests performed for the robustness of the results.
Econometric technique and the data employed in the study are introduced in the next
two sections. The final section concludes.
2.2 The evolution of methodological approaches
A plethora of studies in economics and international trade literature analyse the
exchange rate and its relationship with various macroeconomic indicators in an
economy. Bahmani-Oskooee and Miteza (2003) in their attempt to assemble the
literature on the relationship between exchange rate and trade balance classify studies in
four different categories. The four categories are ‘before-after’ approach, ‘control group’
approach, ‘macro-simulation’ approach and ‘econometric’ approach. The earlier studies
fall into ‘before-after’ approach which compares the economic growth performance of
countries before and after the currency devaluation. The ‘control group’ approach
comprises of those studies which compare the output performance of devaluing
17
countries with the performance of that of a non-devaluing country (the control group).
The ‘macro-simulation’ approach uses simulation models to compare simulated output
performance after a hypothetical devaluation in a country. The ‘econometric’ approach
uses empirical models to understand the impact of devaluation on trade and output
performance. Hence, this study also reviews and organises selected literature into the
above four approaches in an attempt to select the best approach to be adopted for
analysis in this study.
i. Before-After approach
Among the influential studies falling under this approach is the work of Diaz-Alejandro
(1965). He examines Argentina’s 1959 devaluation between periods 1955-1961 by
analysing the growth rates before and after the 1959 devaluation. He finds that the
devaluation of the peso had contractionary effect on the economy’s real output. The
author argues that the unfavorable impact on output is as a result of a shift in income
distribution from high-propensity consumers to high-propensity savers that ultimately
reduced national consumption levels. In another influential study, Cooper (1971) uses the same approach to investigate the
impact of currency devaluation in 19 developing countries for the period 1953-1966. He
finds that while devaluation improves the external position of a country, it can also fuel
inflation and reduce economic activity. Moreover, he adds that developing countries
with relatively large amounts of debt would need excess funds to meet their debt
obligations in local currency after devaluation. This would put added pressure on the
economy to use additional domestic resources for repayment which otherwise could
have been used in expanding domestic production.
Similarly, Edwards (1989a) investigates the impact of 39 devaluation episodes on
aggregate output in developing countries and finds that in many cases real growth rates
often start falling prior to devaluation and tends to continue on that path. Hence, he
concludes that devaluations are historically contractionary. On the other hand, Krueger
18
(1978) finds that devaluation is followed by output improvements in 19 out of the 22
devaluing countries between 1951-1970 periods.
Although, studies using before-after approach produce some insights on devaluation and
its impact on the devaluing countries, Bahmani-Oskooee and Miteza (2003) point out
that these studies suffer from limitations. The major limitation is that this approach is not
strictly based on the ceteris paribus assumption. As a result, the approach fails to
independently assess the impact of currency devaluation on output performance because
in reality other macroeconomic variables interact and change over time. Therefore,
studies using this approach may not truly reflect the impact of currency devaluation on
output.
ii. Control group approach
The ‘control group’ approach categorises studies that compare output performance of a
devaluing country with that of a non-devaluing country (the control group). These
studies assume that both countries are influenced by the same set of external factors and
therefore the difference in trade and output performance would entirely reflect by the
effects of devaluation.
This approach largely includes studies involving countries that undertook currency
devaluations as part of IMF stabilisation programs especially in the 1970s and 1980s.
Donovan (1982) investigates 78 devaluations carried out under IMF supported programs
during the 1971 to 1980 period. The control group to make the comparisons was chosen
based on a pool of countries which are not oil-based economies and had not gone
through IMF stabilization program. The study finds that there was a relative
improvement of trade flows in the program countries during the study period. However,
output performance in program countries were estimated to decline more in the short–
run than in the long–run in comparison to the control group.
Kamin (1988) also utilise the control group approach to examine 107 devaluations
during the 1953-1983 period. The study focused on analysing the performance of some
the macroeconomic variables by evaluating the statistical significance of the difference
19
between those variables in the devaluing country and the control group. The findings
suggest that most of the devaluing countries achieve improved growth rates at least a
year after the devaluation.
Edwards (1989b) also observe the performance of several macroeconomic variables for
18 devaluations over the 1962-1982 periods in Latin American economies. In order to
analyse the impact of the devaluation on growth performance, the study selects a set of
24 developing countries that had kept their nominal exchange rates fixed as the control
group. Hence, the study concludes that the Latin American economies experience
declining growth rates immediately in the years after the devaluation. He further argues
that this may be as a result of policies and restrictions that may have complemented
devaluations in those countries.
Though studies under the control group approach have generally been superior to those
in the before-after approach, Bahmani-Oskooee and Miteza (2003) highlight that studies
under this approach suffer from a country selection biasness problem. They argue that
since most studies under this approach were based on analysing countries on IMF
stabilisation program, this would result in significant difference when comparing with
non-program countries. This is because these two sets of countries share very different
characteristics at the start of the program, which makes it very difficult to split the
impact on aggregate output stemming from devaluation.
iii. Macro-simulation approach
Studies classified as part of the macro-simulation approach rely on simulation of
economic models to make assertions on the output performance of an economy after a
hypothetical devaluation in an economy.
Gylfason and Schmid (1983) develop a model incorporating the cost of intermediate
goods in an open economy macroeconomic model. The model shows that devaluation
expands output via the demand channel while it reduces supply via its effects on the
increased costs of imported intermediate inputs. By testing their model on a sample of
ten countries using the ‘calibration’ method, they find support for expansionary effects
20
of devaluation in eight out of the ten countries. They infer that devaluation plays a huge
part in stimulating expenditure switching than causing contractionary effects on output.
Branson (1986) using the similar approach focused on imported intermediate goods and
wage indexation in a two-sector model for Kenya. The study argues that most of the
capital investment in developing countries is often in terms of imported capital goods
which face a higher price after devaluation. As a result, the rising cost of imported
capital goods would lead to less creation of new investments which would result in
reduced output. Hence, in the case of Kenya, Branson finds that devaluation results in
increasing price level which leads to reduced output in the economy.
Lizondo and Montiel (1989) develop a comprehensive general analytical framework to
study the relationship between real output and exchange rate. The study models a small
open economy with identical labour demand and specialised capital that uses imported
inputs. Given the different size of the impact on favorable demand and adverse supply
side factors, the study concluded that it was unclear to determine the effectiveness of
devaluation in the country.
Moreover, Gylfason and Radetzki (1991) developed a macro-simulation model to study
the short and medium term effects of devaluation on macroeconomic indicators in 12
least developed countries. The model captures the effect of devaluation on output via
export and import changes and through cost of the imported inputs. They estimated that
due to rising price of imported inputs in production, devaluation in most of these
countries supported the contractionary hypothesis of devaluation. Taye (1999) also using
a similar kind of model, finds contractionary effects of devaluation in Ethiopia.
Using another macro-simulation method by adopting IS-LM and Cobb-Douglas
framework, Kandil (2004) studies the impact of exchange rate fluctuations on economic
activity in 22 developing countries. Theoretically, he proves that the effect of currency
depreciation is contractionary via the supply side effects while the impact on demand
side is indecisive. The study hence, reinforces the arguments of Krugman and Taylor
(1978) and Edwards (1986) on the contractionary effects of depreciation on the
21
economy. The study also adds that the capacity of the devaluing economy to meet
additional demand and the response of the demand and supply side channels are
important in determining the net impact of depreciation on price and output
performance.
Whilst the macro-simulation method also suffers from shortfalls, Bahmani-Oskooee and
Miteza (2003) acknowledge that studies utilising this approach have provided significant
contribution in the literature dealing with effects of exchange rate on output
performance. The major drawback of this approach lies in the selection and use of the
parameters in the models that are assumed to be unchanged for countries across different
policy simulation. This assumption is actually unrealistic. They add that this kind of
shortfall can actually be overcome by employing econometric techniques to establish the
relationship between exchange rate, trade and output performance.
iv. Econometric approach
The fourth and the most popular category of studies as identified in Bahmani-Oskooee
and Miteza (2003) are the ones based on econometric approach to establish a
relationship between devaluation, trade and growth performance. Studies under this
approach have contributed enormously to the literature with the application of new and
advanced econometrics techniques over time.
Himarios (1985) using the OLS methodology employ annual time series data for 10
countries over the 1956-1972 period to establish relationship between devaluation and
trade balance. The econometric model includes variables such as trade balance, real
exchange rate, opportunity cost of money, money supply, government expenditure,
domestic income and foreign income. The study finds that in nine out of the ten
countries, real devaluation improves trade balance in the long–run. The major
contribution of the study is that it highlights that the domestic and foreign variables have
different impacts on trade balance outcome. It also stresses that it is the real exchange
rate rather than the nominal exchange rate that affects the trade balance in an economy.
22
Edwards (1986) estimates a fixed effects model of real output based on pooled data for
12 developing countries over the 1965-1980 period. His econometric model incorporates
variables such as government expenditure, monetary surprise variable, terms of trade
and real exchange rate with lagged values to capture the short- and long–run effects on
real output. The study estimates that the monetary surprise term and the government
expenditure has a positive impact on output. However, the study finds that currency
devaluation reduces output in the first year but has a positive impact in the following
year. As a result of the different lagged effects of devaluation in the first and second
years, it resulted in no effect on real output in the long–run.
Bahmani-Oskooee and Rhee (1997) apply Johansens’s cointegration and error correction
technique to model quarterly data over the 1971-1994 period for Korea. Their model
included variables such as real effective exchange rate, terms of trade, government
expenditure to GDP ratio and money supply. Unlike Edwards (1986), they confirm the
presence of a long–run relationship between money supply, real exchange rate and real
output in Korea. The study argues that the greatest impact of devaluation on output was
being felt at approximately three lags after devaluation in the economy.
However, in the case of Mexico, Kamin and Rogers (1997) find that devaluation has
contractionary effect on output. They argue that devaluation leads to inflation and
reduced government spending in the economy. On the other hand, Bahmani-Oskooee
(1998) does not find evidence of a long–run relationship between exchange rate and real
output for most of the 23 LDCs over the 1973-1988 period using the error-correction
techniques.
In a comprehensive study, Kamin and Klau (1998) examine the effects of devaluation on
output for 27 countries, including 8 Latin American, 6 Asian and 13 industrialised
countries over the 1970-1996 periods. In order to capture the short- and long–run
impacts of devaluation on output, they incorporate variables such as GDP, real exchange
rate and the output gap ratio in their error correction model. The authors find that the
currency appreciation leads economic growth whereas the depreciation lowers it. They
23
also find support for the view that while devaluation is contractionary in the short–run, it
has no effect on output in the long–run. The study also showed that in comparison to the
developing countries, industrialised countries were often faced with more contractionary
effects of devaluation in the long–run.
Bahmani-Oskooee (2001) studies the relationship of nominal and real effective
exchange rates with the trade performances of 7 Middle Eastern countries over the 1971-
1994 periods. With the application of the Johansen and Juseliu’s cointegration
technique, the study uses the reduced form the trade balance model for Bahrain, Egypt,
Jordan, Morocco, Syria, Tunisia and Turkey. The reduced form of the trade balance uses
variables such as domestic income, foreign income and real effective exchange rate to
establish association with the trade balance of a country. The study finds that real
depreciation in most of the non–oil exporting Middle Eastern countries have favourable
effects on the trade balance in the long–run.
The major contribution by Bahmani-Oskooee (2001) is that it argues in favour of using
the real effective exchange rates rather than the nominal effective exchange rates to
analyse the impact of devaluation or depreciation in the country. He argues that the
inflationary effects of nominal devaluation can take away the favourable effects it has on
a country’s trade balance. Therefore, when assessing the international competitiveness
of an economy, the study strongly suggests the use of the real effective exchange rate in
order to adjust for the inflationary effects of devaluation in the country.
Lal and Lowinger (2002) also use Johansen’s cointegration error-correction technique to
model trade balance for seven East Asian countries over the 1980Q1-1998Q4 periods.
The trade balance is modelled as a reduced form which includes domestic income,
foreign income and real effective exchange rate variable. The study finds presence of a
long–run relationship between real effective exchange rate and trade balance in all the
countries under study.
Similarly, Bahmani-Oskooee and others (2002) with particular focus on major countries
in the Asian region study the same for Indonesia, Korea, Malaysia, the Philippines and
24
Thailand. The study finds that the results are not same for all countries. It is noted that
real depreciation in the Philippines and Thailand lead to expansion in output while it
results in declining output for Indonesia and Malaysia. In the case of Korea, the results
do not indicate significant effect of devaluation on the country’s output.
Kalyoncu and others (2008) carry out a similar study using the error correction model
for 23 OECD countries generally over the 1980Q1 to 2005Q4 periods. Like some
previous studies, this study also finds mixed results. It establishes the presence of a
long–run relationship between real exchange rate and output in 9 out of the 23 countries
in study. However, from these 9 countries, only in 3 cases currency devaluation was
estimated to result in improvement in output performance.
Similarly, Galebotswe and Andrias (2011) analyse the effect of currency devaluation on
output performance for Botswana over the 1993Q1 to 2010Q4 period. Using VECM, the
study models real GDP as a function of real exchange rate, domestic interest rate,
foreign interest rate, government expenditure and foreign income. Hence, the study finds
that though currency devaluation has expansionary effects in the short–run, it often
results in an adverse impact on the output in the long–run. Similar contractionary effects
of devaluation are also found by Ayen (2014) in the case of Ethiopia in the long–run.
Recently, Bahmani-Oskooee and Gelan (2012) use the reduced form of the trade balance
model to study the relationship of exchange rate with trade balance in 9 African
countries10 over the 1971Q1-2008Q4 periods. The study using the VECM techniques
shows that in the long–run, real depreciation only has favourable effect on the output
performance of Egypt, Nigeria and South Africa.
This section on the evolution of various approaches to analyse the relationship among
exchange rate, trade and output performance highlights that the use of econometrics
approach with new and advanced techniques is recent and continues to gain popularity.
Hence, the widely used econometric model employed by many scholars is the reduced
10 The nine countries in the study include Burundi, Egypt, Kenya, Mauritius, Morocco, Nigeria, Sierra Leone, South Africa and Tanzania.
25
form of the trade balance model to investigate the impact of depreciation or devaluation
on the trade balance. This model measures trade balance as a function of real exchange
rate, domestic income and foreign income.
However, as introduced in Chapter 1, the trend in the literature has moved from trying
only to investigate the link between exchange rate, trade balance and output, to include
as well validation of the presence of the J-curve phenomenon on the trade balance. There
is again a plethora of studies on the validation of the phenomenon at various
disaggregated levels of trade balance following currency devaluation and depreciation in
the economy. Yet, there is no general consensus on the presence of this phenomenon.
Therefore, a detailed review of the literature associated with the test of the J-curve
phenomenon is carried out in Chapters 4 and 5. Nonetheless, it is important to highlight
the three methods of assessing the validity of the J-curve phenomenon in this
methodological chapter.
2.3 Testing the J-curve phenomenon on the trade balance
The assessment of the J-curve phenomenon in economics literature is categorised in
three different approaches. The phenomenon is often validated by one or other of the
three methods that are known in the literature as (i) the traditional definition (or the old
method), (ii) the new definition and (iii) the impulse response function (IRF) analysis.
However, in order to assess the phenomenon using the three approaches, econometric
techniques are required to estimate the trade balance models. In what follows, the
application of each of the approach is described in detail.
i. The traditional definition (old)
Using the traditional or the old definition, the presence of the J-curve phenomenon is
confirmed by analysing the short–run coefficients of the real exchange rate variable at
different lag lengths in a trade balance model. Therefore, negative real exchange rate
coefficients at shorter lags followed by positive coefficients at longer lag lengths
26
confirm the J-curve phenomenon under this method. The negative coefficients followed
by positive ones suggest that currency depreciation or devaluation reduces trade balance
at initial stages (negative coefficient) followed by improvements (positive coefficient) in
the trade balance at longer lags. Studies which have attempted to validate the presence of
the J-curve phenomenon using this approach include Shirvani and Wilbratte (1997),
Bahmani-Oskooee and Brooks (1999), Arora and others (2003), Bahmani-Oskooee and
Ratha (2004), and Wang and others (2012).
ii. The new definition
Following the work of Rose and Yellen (1989), another way of investigating the J-curve
phenomenon is by analysing the short- and long–run coefficients of real exchange rate
variable. For the J-curve phenomenon to hold valid, negative effect of exchange rate on
the trade balance in the short–run must be combined with significant positive impact in
the long–run. This approach suggests that currency depreciation or devaluation reduces
trade balance in the short–run (negative coefficient) followed by improvements (positive
coefficient) in the long–run. Studies which have attempted to validate the presence of
the J-curve phenomenon using this approach include Arora and others (2003), Bahmani-
Oskooee and others (2005) and Bahmani-Oskooee and Harvey (2012).
iii. The Impulse Response Function (IRF) analysis
Studies such as Lal and Lowinger (2002), Narayan (2006), Rahman and Islam (2006)
and Yazici and Klasra (2010), among others, did not rely much on the coefficient results
but rather focused on the impulse response function (IRF) to analyse the J-curve
phenomenon. The IRF analysis shows the time path of the trade balance to a shock
(innovation) in the real exchange rate. Hence, the phenomenon is confirmed if the path
the trade balance follows after the shock resembles a J-shape. Though there is more than
one method of carrying out IRF analysis, the use of the generalized IRF of Pesaran and
Shin (1998) has been popular in the literature. This is because the generalized IRF
controls for the intermediate effects and is insensitive to the ordering of the variables in
the model. Therefore, the use of generalized IRF is used to assess the graphical pattern
of the trade balance variable to a shock (innovation) in the real exchange rate variable.
27
Having discussed the various methodological approaches and the approaches to identify
the J-curve phenomenon, the literature is being followed closely to present the trade
balance model. The model being discussed in the next section would be used to estimate
the exchange rate and trade balance relationship along with the test for the J-curve
phenomenon in this thesis.
2.4 The trade balance model
Given the objective of this study to investigate the impact of the real exchange rate on
the trade balance and to validate the J-curve phenomenon, the trade balance model
originates from the open economy standard aggregate demand model. The model also
closely follows the pioneering works of Rose and Yellen (1989). In an open economy,
the standard aggregate demand model is made up of total private demand for goods and
services, total public demand for goods and services and the net exports, defined as:
TBGDY ��� (1.1)
where Y is the real GDP, D is the total private demand for goods and services, G is the
public demand for goods and services and TB represents the net exports or the trade
balance, defined as exports minus imports. The focus of this study is on the trade
balance and hence, special emphasis is placed on its relationship with the exchange rate.
The important determinant of net exports is the real exchange rate rE , defined as:
PEPE
fr � (1.2)
where fP is the price of foreign goods denominated in foreign currency, and fEP is the
price of foreign goods measured in domestic currency. P represents the domestic price
level. The real exchange rate is simply the relative price of foreign goods, indicating the
number of units of the domestic good that must be given up to acquire one unit of the
foreign good. It is also a measure of the international competitiveness of domestic
producers, implying that the higher the real exchange rate, the cheaper are domestic
goods relative to goods produced abroad.
28
All the components of demand on the right-hand side of (1.1) including the trade balance
are measured in units of the domestic good. This goes on to say that the domestic good
is used as the numeraire good. If the quantity of imported foreign goods is M , the
nominal value of imports measured in domestic currency units will be MEP f . Given
that one unit of the domestic good sells at the price P , the volume of imports measured
in units of the domestic good will then be MEPMEP rf �/ . If the quantity of domestic
goods that are exported is denoted by X , then the trade balance equation is defined as:
MEXTB r�� (1.3)
It is therefore, reasonable to assume that the volume of exports, X depends positively
on the international competitiveness of domestic producers, measured by the real
exchange rate rE . Theoretically, exports also depend positively on total output ( fY ) in
the rest of the world. It is argued that higher economic activity in trading partner
economies provides a larger export market for domestic producers. Hence, the export
equation is expressed as, ),( fr YEXX � .
The total demand in the economy is met from domestic production and imports. This in
an economy is influenced by the level of domestic income, Y . In addition, the quantity
of imported foreign goods depends negatively on the real exchange rate. The higher
relative price of foreign goods reduces consumer demand for foreign goods, partly
because of income and substitution effect. A rise in the price of imported goods reduces
the purchasing power of domestic nominal incomes (known as the income effect) which
induces consumers to substitute for domestically produced goods (known as the
substitution effect). Hence, the import equation is expressed as, ),( YEMM r� .
Based on these export and import model specifications, it follows from (1.3) that net
exports are given by:
),(),(����
�� YEMEYEXTB rrfr (1.4)
where the signs below the variables indicate the signs of partial derivatives. A rise in fY is expected to result in an increase in the level of exports as it is expected that rising
29
demand in the trading partner countries would translate in increased demand for
domestically produced goods. Similarly, a rise in Y is expected to result in an increase
in the level of imports as an increase in domestic income will stimulate total private
demand including the import of goods and services in the economy.
However, a crucial question is how a change in the real exchange rate will affect the
trade balance. To examine this, partial derivative of the net export function is calculated
for equation 1.4 with respect to rE :
MEME
EX
ETB
rr
rr ���
���
��� . (1.5)
Without using simplification, the units of measurement is chosen such that the initial
value of the real exchange rate is unity, 10 �rE . As a benchmark case, the trade balance
is initially assumed to be in equilibrium so that 000 MEX r� . Using these relationships,
equation (1.5) is rewritten as:
)1..(0
0
0
00 �
��
���
���
ME
EM
XE
EXM
ETB r
r
r
rr � (1.6)
)1(0 �����
Mxr METB �� , 0. �
��
�XE
EX r
rx� , 0. ���
��ME
EM r
rM�
where x� is the elasticity of exports with respect to the real exchange rate, and M� is the
elasticity of imports with respect to the real exchange rate. Equation (1.6) shows that a
real depreciation of the domestic currency (that is, a rise in the real exchange rate, rE )
will improve the trade balance provided the sum of the relative price elasticities of
export and import demand is greater than one )1( �� Mx �� . This result is also called
the Marshall-Lerner condition.
Similarly, following equation 1.4, in order to analyse the direct impact of exchange rate
on trade balance and to validate the J-curve phenomenon, the net effect of export and
30
import relative price elasticities is expressed in one equation in the following model
specification as:
),,( fr YYEfTB � (1.7) However, equation 1.7 does not include second round effect of income changes on the
changes of the trade balance. The improved trade balance raises income and the
increased income, thereby, limit the favourable effect of trade balance in the medium–
to long–run. This is demonstrated with simple algebra without bringing the complex
time path, for simplicity.
The first round effect is derived assuming )(TBYY � . Since M depends on Y , the
favourable effect of trade expansion seems to be limiting it by raising imports.
Therefore, we can write:
dt
dTBYdtdY
TB.� . Or dt
dTBYMEMEMXdt
dTBTBY
rrEr ... �
����
Rewriting the expression, the total effect can be presented as follows:
�
��
�
���
�
���
� r
TBY
rE E
YMMEMX
dtdTB r
1 (1.8)
It is well documented in the literature that when net export increases, it influences Y
and thereby M . This limits the effect of devaluation after a while and tends to be
showing cyclical movement on TB in response to devaluation particularly for an import
dependent economy. In other words, when devaluation starts improving trade balance
and thereby income, then the second round effect will come to effect in the opposite
direction (i.e., trade-induced income effect).
2.5 The empirical model
The key aspect in this study is the attempt to determine empirically the impact of
currency devaluation on aggregate, bilateral and sectoral level trade balance in the
context of a Pacific islands developing economy, Fiji. The study also attempts to
validate the presence of the J-curve phenomenon in each case.
31
The empirical technique applied in this study is different from the existing study in Fiji.
In this study, vector error correction model (VECM) is used to estimate the short- and
long–run effect of devaluation on the trade balance models in the economy. The model
specification as shown in Section 2.4 is based on the work of Rose and Yellen (1989).
The same model specification has also been widely used to empirically investigate the
presence of the J-curve phenomenon in the literature. This model specification is also
similar to the one employed previously by Narayan and Narayan (2004a) in the case of
Fiji.
The reduced form of the trade balance model continues to be used in many studies (such
as Bahmani and Gelan (2012), Wang and others (2012) and Wijeweera and Dollery
(2013)) to estimate the impact of currency devaluation or depreciation on aggregate,
bilateral and sectoral level trade performance. Hence, the trade balance model employed
in this study directly relates a measure of the trade balance to domestic real income,
foreign real income and real effective exchange rate as expressed in equation 1.7:
),,( fr YYEfTB �
where TB represents the trade balance measure, being defined as the ratio of the
domestic country’s exports over her imports. This measure of trade balance as argued by
Bahmani-Oskooee and Brooks (1999) makes the trade balance unit free, reflects either
real or nominal trade balance and also ensures it to be specified in log-linear form.
Moreover, this measure of the trade balance is now being widely used in the J-curve
literature. Furthermore, Y relates to domestic income for Fiji while fY measures the
real income of the trading partner economies. rE represents the real exchange rate of
Fiji with its trading partners. This is defined in such a way that the increase reflects a
real devaluation of the Fijian dollar against the currency of its trading partners.
The model for the empirical analysis being expressed in log-linear form is as follows:
������ fr YdYcEbaTB lnlnlnln (1.9)
As far as the signs of the variables are concerned, it has been well discussed in the
hypothesis section in Chapter 1. It is expected that currency devaluation would make
32
exports cheaper in the foreign market and make imports expensive in the domestic
economy. This will encourage exports and discourage imports resulting in a positive
sign for b . However, for many developing economies, its production processes are
heavily import–intensive and as a consequence, to maintain the level of production given
the relative price rise, the total import bill is expected to increase after devaluation. This
leads to an overall increase in production costs and ultimately in the price of exportables.
The gradual increase in the price of exports can over time discourage domestic exports
in the foreign markets. There might also be consequences of the resultant inflationary
pressure, economy’s lack of domestic capacity to meet additional export demand, heavy
dependence on imports and the lack of import substitutability in the country. If this
argument holds true, the sign of b can be negative. This would imply that devaluation
actually worsens trade balance in Fiji.
Moreover, a rise in the economy’s income level more often results in the increase in
consumption demand for goods and services. The total demand as expressed in equation
1.1 is met by the importables as well. Hence, a rise in demand would mean increase in
imports which subsequently results in the deterioration of the trade balance. If these
arguments hold true, the sign of c is expected to be negative. However, an equally
possible contrasting argument is that if the increase in Y is due to an increase in the
production of import-substitute goods, then the domestic economy may import less as
income increases. This can result in a positive sign for c .
On the other hand, if the demand for the Fiji’s exports is highly income elastic, it is
expected that rising income in the trading partner countries would result in an increase in
domestic exports. This would lead to an improvement in the trade balance which could
result in a positive sign for d . However, in a similar argument to that presented for
domestic income, the same can be applied for the effect of trading partner income on
Fiji’s trade performance. If the increase in fY is due to the production of those goods
typically produced and exported by Fiji, this would result in a negative sign for d .
Hence, based on the arguments presented, it is clear that there is no general agreement
on the response of the variables employed in the empirical model.
33
At the outset, it is also important to mention that a similar trade balance model
specification (Equation 1.7) is used for the analysis at different levels of trade. This
includes the analysis at aggregate, bilateral and sectoral level trade. The difference also
lies in the data requirements which are specific to aggregate, bilateral and sectoral trade
analysis. The bilateral real exchange rate is also used for bilateral trade analysis where
the specific bilateral exchange rates were available.
The respective export and import demand equations in each of the trade balance models
are also analysed in the study. This is done in order to investigate the underlying sources
of impact on the respective trade balance. These export and import demand equations
are consistent with the trade balance model in equation 1.7. They are modelled as
follows:
),( fr YEfEXP � (1.10)
),( YEfIMP r� (1.11)
where EXP represents total real exports of Fiji while IMP represents total imports of
Fiji. The exports demand equation is modelled as a function of real exchange rate, rE
and foreign income, fY while the imports demand equation is modelled as a function of
real exchange rate, rE and domestic income, Y .
The above two equations for the empirical analysis in a similar manner to the trade
balance model is expressed in a log-linear form as follows:
1111 lnlnln ����� fr YcEbaEXP (1.12)
2222 lnlnln ����� YcEbaIMP r (1.13)
The signs of the variables in export and import demand equations respond in a similar
way as discussed for the respective variables in the trade balance model.
Therefore, having discussed the possible long–run relationship of the variables in the
trade balance model, the next section outlines the series of empirical steps undertaken to
incorporate short–run dynamics into the trade equations. Hence, describes below is the
application of the system equation based VECM technique for analysis in this study.
34
2.6 The VECM technique
Hereafter, as part of the empirical procedure, the VECM is proposed to be used to
estimate the various equations. The use of VECM is justified on the purpose to evaluate
both short and long–run impact of variables in the models. The testing procedure
involves three steps; testing for the existence of unit root, cointegration test followed by
estimating the short and long–run relationship among the variables in trade balance and
output model specification.
2.6.1 Unit root tests
In time–series econometric analysis, it is very important as a first step to test for the
presence of unit root in the variables being employed in the analysis. To test for unit
root, the Augmented Dickey–Fuller (ADF) test based on the following auxiliary
regressions was utilised:
��
�� ������k
jtjtjtt ydyky
11 �� (1.14)
��
�� �������k
jtjtjtt ydtyky
11 ��� (1.15)
The test for a unit root in ty , where y refers to the variable in question, is a Tt ,...1�
index of time, jty �� is the lagged differences to accommodate serial correlation in the
errors, t� . The equation (1.13) tests the null of a unit root against a mean stationary
alternative in ty while equation (1.14) tests the null of a unit root against a trend
stationary alternative. Inclusion of intercept and time trend is based on the observation
whether the series has a drift or time trend. Evidence of unit root for each variable is
found if the null hypothesis of 0�� is not rejected. Otherwise, the evidence obtained is
that y is stationary, i.e. I(0). If y is non-stationary, the presence of unit root is tested
in first difference of y . Hence, it is only said to be integrated of order one, i.e. I(1) if
y� becomes stationary.
35
2.6.2 Cointegration test
If all the variables in the model are found to be integrated of order one, the next step is
to investigate whether there is a long–run relationship among those variables through the
cointegration test. The approach proposed by Johansen (1988) and Johansen and Juselius
(1990) is applied using the maximum likelihood procedure to determine the presence of
cointegration vectors. This procedure is based on the vector autoregressive (VAR)
model as follows:
ttt
k
iit YYCY ��������� ��
�� 11
1 (1.16)
where, tY in equation 1.15 is a vector of I(1), non-stationary in level form variables and
C is a constant. The information on the coefficient matrix is decomposed as ���� ,
where the relevant elements of the matrix are the adjustment coefficients and the �
matrix contains the cointegrating vectors. Johansen and Juselius (1990) recommend the
use of the maximum eigenvalue and the trace statistics to determine the number of
cointegrating vectors.
2.6.3 Granger causality
However, if any evidence for cointegration is not found, then the specification of the
granger causality test will be a VAR in first difference form. This would measure only
the short–run interactions among the variables in the specified model. However, if
evidence for cointegration is found then there is a need to augment the granger-type
causality test model with a one-period lagged error correction term. Equations (1.17) and
(1.18) illustrate this technique using two variables, X andY .
ttt
p
iit
p
iit ECTXYvY 1111
11
1lnlnln ���� �������� ��
��
��� (1.17)
ttt
p
iit
p
iit ECTYXvX 2121
1
11
1
1lnlnln ���� �������� ��
�
��
�
��� (1.18)
Besides the variables defined above, 1�tECT is the lagged error-correction term derived
from the long–run cointegrating relationship and t1� and t2� are serially independent
random errors with mean zero and finite covariance matrix. In each case the dependent
36
variable is regressed against past values of itself and past values of other variables. The
optimal lag length p in equation (1.17) and (1.18) are selected using the Schwarz Info
criterion (SIC).
Accordingly, based on the trade balance model for the four variables case with one
cointegrated relationship, the system approach based VECM which allow for the
feedback effect as discussed in Section 2.4 will have four equations. This results in the
specification of VECM of the trade balance function of the following forms:
ttf
tttr
tt ECTYYETBTB ������� ������������ ����� 15141312110 lnlnlnlnln (1.19)
ttf
ttrtt
rt ECTYYETBE ������� ������������ ����� 15141312110 lnlnlnlnln (1.20)
ttf
tttr
tt ECTYYETBY ������� ������������ ����� 15141312110 lnlnlnlnln (1.21)
ttf
tttr
tf
t ECTYYETBY ������� ������������ ����� 15141312110 lnlnlnlnln (1.22)
The signs and sizes of the ECT will reflect the direction and speed of adjustment in the
dependent variable to deviations from the linear long–run relationship.
2.7 Sources of data
The data used in the study are annual data series covering the 1975–2012 periods. All
the variables are transformed into their log-linear forms to allow the coefficients from
the regression results to be interpreted as elasticities. The data used in the regression
models are compiled from several publicly available sources including:
a. International Monetary Fund’s International Financial Statistics (IFS) online
database.
b. World Bank’s World Development Indicators (WDI) online database.
37
c. Fiji Bureau of Statistics, Key Statistics (various issues).
d. Fiji Bureau of Statistics, Overseas Merchandise Trade Statistics (various issues).
e. Reserve Bank of Fiji, Quarterly Review (various issues).
f. UNCOMTRADE online database.
However, it is important to note that every attempt has been made to use the data as
accurately as possible. This is because more than often, different data sources report
different figures for the same data of interest. Therefore, to be consistent, it has been
ensured that most data come from one particular source unless otherwise stated. Hence,
the sources from which data were compiled and how some of the required variables in
the study were constructed, is discussed in detail. Following the literature and the
specifications of the models, the variables employed in the study are as follows:
i. Price deflator data )(P
Since all the values used in the study are taken in terms of US dollars, GDP price
deflator used to deflate nominal values into real values are derived by using current and
constant GDP for Fiji measured in terms of US dollars at 2005 prices. This data is
obtained from source (b). All the other relevant indexes are also expressed at year 2005.
Therefore, using this price deflator all the following variables unless otherwise stated are
converted to real values before proceeding with the regression estimation.
ii. Export data )(EXP
Data on aggregate, bilateral and sectoral level export of goods and services are obtained
from source (d). On aggregate analysis, Fiji’s total exports for goods and services with
the rest of the world are considered. For the bilateral trade performance analysis, Fiji’s
exports with its ten trading partners are considered: Australia, Japan, New Zealand, the
USA, the UK, China, Hong Kong, India, Malaysia and Singapore are considered.
On the sectoral level, goods exports for food, sugar, fish and gold, along with services
export of travel and transportation sectors, are considered. However, data for food export
are only available from 1979 while for the transportation sector it is only available from
38
1980. The rest of the data cover the entire study period. Since all the data are in terms of
the Fiji dollar, these are converted to US dollar using the period end US and Fiji
exchange rate obtained from source (a). These values are then using variable (i)
converted to real values before proceeding with the analysis.
iii. Import data )(IMP
Data on aggregate, bilateral and sectoral level imports of goods and services are
obtained from source (d). Similarly, on aggregate analysis, Fiji’s total imports of goods
and services from the rest of the world are considered. For the bilateral trade
performance analysis, Fiji’s imports with the same ten trading partners as for exports are
considered.
On the sectoral level, goods imports of food, fuel, manufactured goods, crude oil,
textiles, machinery and transport equipment, tobacco and beverage, chemicals, oil and
fats and miscellaneous manufactured goods along with services imports of travel and
transportation sectors are considered. However, data for import of food go back only till
1979 while for textile and transportation, they are available only from 1980. The rest of
the data cover the entire study period. Again, since all the data are in terms of Fiji dollar,
these are converted to US dollars using the period end US and Fiji exchange rate
obtained from source (a) before using variable (i) to convert to real values.
iv. Trade balance )(TB
Data on trade balance are derived from the export (ii) and import (iii) data for the
respective aggregate, bilateral and sectoral trade performance based on trade flows for
which both export and import data are available. The measurement of trade balance11
follows the generally used method in the literature to define trade balance as exports of
11 The variable representing trade balance, TB is expressed as ratio of export/import. In log linear transformation it has been taken as ln(export/import) which is also the same as ln(exports) – ln(imports). Hence, an increase in this ratio, TB would mean either an increase in exports or a decline in imports. An increase in TB could also result when both exports and imports are increasing but exports are increasing at a faster rate or with a higher magnitude than imports. Similarly, an increase in TB could arise when both exports and imports are decreasing but imports are decreasing at a faster rate or with a higher magnitude than exports.
39
goods and services divided by imports of goods and services. This measure effectively
makes the measure unit free.
v. Fiji’s income )(Y
Measure of Fiji’s income is represented by the GDP values of Fiji obtained from source
(b). The variable is expressed in terms of constant US dollars based on 2005 prices.
vi. Trading partner income )( fY
The trade weighted trading partner income variable employed in the trade balance model
has been calculated based on the formula used in Bahmani-Oskooee (1986). The
variable construction hence follows the formula specification:
�� iif YPaY where � � ,1ia
where ia is the weight of trading partner i in Fiji’s total trade and iYP is the GDP of the
respective trading partner. The trading partners considered in the variable construction
are based on those against which Fiji’s fixed exchange rate is pegged. These include
Australia, New Zealand, Japan, the USA and the UK. The weights of each trading
partner in Fiji’s trade have been obtained from source (d) while the real GDP of
respective trading partner is sourced from source (b).
For bilateral trade equations, respective trading partner’s GDPs have been used from
source (b), which is expressed in terms of constant US dollars based on 2005 prices.
vii. Real Exchange Rate )( rE
The measure of currency devaluation in Fiji is captured by the real exchange rate. The
formula used for constructing real exchange rate as discussed earlier is:
PEPE
fr �
where; rE is the real exchange rate calculated by multiplying )(E which is the nominal
exchange rate, defined as the domestic currency price of the foreign currency variable to
40
the price index in the trading partner economy )( fP . For this, trading partner consumer
price index is used as a proxy. The product is then divided by domestic price level which
uses Fiji’s consumer price index )(P as its proxy. As such, this is defined in a way that
an increase in the index reflects devaluation of Fiji’s currency.
Besides calculating and using bilateral real exchange rate for modelling Fiji’s trade
performance with Australia, New Zealand, Japan and the USA, for the rest of the
models, real effective exchange rate for Fiji is been used. This is obtained from source
(a). This is because these were the only four Fiji’s bilateral trade partners for which
bilateral nominal exchange rate was available. For the data on real effective exchange
rate from source (a), no construction was done as this was already available. The only
task undertaken on the real effective exchange rate was to divide the rate by one to
ensure it represents domestic currency price of foreign currency consistent with the
specification. This is because source (a) reports real effective exchange rate in such a
way that an increase reflects currency appreciation.
For the bilateral exchange rate with the four trading partners mentioned, data
construction was undertaken as per the formula. Fiji’s nominal exchange rate with each
of these trading partners was obtained from source (e) with the trading partners and
Fiji’s consumer price index obtained from source (b).
viii. Coup )(COUP
The COUP is taken as a dummy variable to capture the impact of political instability
(coup) in 1987, 2000 and 2006 in Fiji. This is denoted by value 1 in the year of a coup
with the rest of the years taking a value of 0. Many scholars (for example Gounder,
2001; Kumar and Prasad, 2002; Narayan and Prasad, 2003; and Prasad, 2012a) argue
that political instability has had adverse impacts on economic growth in Fiji. Hence, this
variable has often been included as a dummy variable while trying to predict output
components. Gounder (2001) while investigating the impact of development assistance
on economic growth in Fiji incorporates a measure for political instability and finds
adverse impact of political instability on output performance.
41
Recently, Gong and Rao (2014) find that the prolonged political instability has resulted
in a decline in the economic growth rate of 3.2% annually in Fiji. Based on this
evidence, a measure for political instability to control for the effect of political
disturbance on yet another output indicator is incorporated, represented by trade balance
in this case. Similarly, as adverse impact is found on the output performance, negative
impact of political instability on exports and trade balance is expected in Fiji.
2.8 Concluding comments
This chapter describing how the entire study will unfold is one of the most important
chapters in this study. Hence, the core of this chapter lies in the explanation of the
various approaches used in analysis and the justification of the methodology adopted for
the study. Furthermore, the derivation of the trade balance model to be employed in the
study is presented. The trade balance model attempts to establish the relationship of
exchange rate, domestic income and foreign income in the economy. The details of the
regression technique using VECM along with the sources of data are also discussed in
detail. In what follows in the next chapter, a brief but well-rounded overview of Fiji’s
economy and its recent economic performance is presented, with particular reference to
trade and devaluation in Fiji.
42
CHAPTER 3
FIJI ECONOMY OVERVIEW
3.1 Introduction
Although some economists may argue that the developed and other developing
economies have nothing much to learn from the Pacific experience, it is surmised that
the experience from the Pacific region, including Fiji, can provide some useful lessons to
other developing and emerging economies. This chapter is organised as follows. Section
3.2 provides an overview of Fiji’s economic growth performance. This includes the
discussion on the economic performance of some of the important sectors along with
key discussions on the monetary and fiscal policies issues in the economy. Section 3.3
discusses on the trade policy issues in the economy. Included in the discussion are the
important issues surrounding the currency devaluation in Fiji. The final section
concludes.
3.2 Overview of Fiji’s economic performance12 Fiji, in the twenty-first century is a multi-ethnic small island nation located in the
tropical Pacific region, with diverse landscape, climate and people. The country’s central
geographical location in the region has made it the hub of the Pacific as it acts as a
crossroad for airline and shipping services to the other Pacific nations. Even though the
country has more than 300 islands, only about one-third are inhabited. The majority of
the people live in the two main islands of Viti Levu and Vanua Levu, which together
account for slightly more than 95% of the total population. The World Bank estimates
that the population as at the end of 2013 has been little less than 900,000 in the country.
The World Bank system, classifies Fiji as an upper middle income country with a GNI
12 For an in-depth survey of Fiji’s economy see Prasad and Narayan (2008); Prasad (2010); Prasad (2012a), Gounder and Prasad (2013)
43
per capita (using the World Bank Atlas method) of US$4,430. The economy has
achieved socio-economic development showing a modest gradual increase in its overall
Human Development Index (HDI) and has a reported value of 0.724 in 2013 (UNDP,
2014).
Fiji, a British colony for nearly a century, became independent in 1970. The economic
growth rate in the 1970s peaked as high as 12.7% and achieved annual average growth
rates of 6.5%. However, since 1987, Fiji’s path to economic success has been disrupted
by political instability initiated by the first military coup that toppled a democratically
elected government. Following this, a coup culture tightened its grip a second time in
1987 and leading to other coups in 2000 and 2006.
The consequent political instability seriously eroded investor confidence creating
fluctuating foreign investments, an uncertain business climate, and issues with property
rights, as well as triggering a massive outflow of skilled labour. The sugarcane industry,
which had been the backbone of this country for years, is on a decline. Expiry of land
leases is also a hindrance to growth in agricultural production. The economy at present
boasts the tourism sector as one of the fastest growing, which is providing many
backward and forward economic linkages in the economy.
3.2.1 Economic growth performance
The overall economic growth performance in Fiji over the past four decades has been
poor with unstable growth patterns. The Fijian economy has recorded mixed pattern of
growth with episodes of high and low growth rates and at times recording negative
growth in output. Even though the growth rates have been volatile, the economy grew at
an annual average rate of 4% after independence and prior to the first coup in 1987.
However, after the political turmoil in 1987, Fiji’s real GDP growth rate has been
mediocre as the economy took a steep dive from a positive GDP growth rate of 7.7% in
1986 to a negative growth rate of 6.6% in 1987.
44
The economy began to improve late in the 1990s but the coups in 2000 and 2006 led to
immediate and long term deterioration of the economic growth rates. This has averaged
around 2.5% per annum in the last two and half decades. Recently, Gong and Rao
(2014) estimated that the significant adverse impact of the prolonged political instability
in Fiji has translated in to a decline in economic growth rate by 3.2% annually.
However, for the last five years, the economy has been achieving positive growth rates
and as a result, positive growth rates are projected for the next three years. The economy
recorded a 4.5% growth rate in 2014 and is projected to growth by another 4.3% in
2015. The recent holding of the general elections since 2006 seems to have augured well
resulting in obtaining re-entry to the Commonwealth nations on a full membership and
removal of sanctions placed by Australia and New Zealand governments on Fiji after the
2006 coup.
Figure 3.1 Fiji's real GDP growth rates, 1975–2015(p)
-10.0
-5.0
0.0
5.0
10.0
15.0
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
(p)
Perc
ent (
%)
Years
Notes: 1. Data for 1975–2007 are extracted from The World Bank’s World Development Indicators (online), 2015.
2. Data from 2008 to 2013 periods are sourced from the Ministry of Finance 2015 Budget Supplement. Latest revised data for 2014 and 2015 are incorporated from the RBF press release number 13/2015.
3. p- projected; f- forecasted Source(s): Data derived from The World Bank's World Development Indicators (online), 2015, Ministry
of Finance 2015 Budget Supplement and Reserve Bank of Fiji.
45
Growth performance for Fiji has generally outstripped that for other Pacific Island
Countries (PICs; see Table 3.1). However, its economic performance was not promising
during the 2007 to 2012 period against the other countries although it picked up
momentum since then. The PICs economies have been increasingly challenged by the
forces of globalization and the increased global integration has provided increased
opportunities as well as increased exposure to global risks and challenges for the island
countries.
Table 3.1 Real GDP growth rates in the PICs, 1990–2015(f)
PICs 1990s 2000–2005
2007 2009 2011 2012
2013
2014
2015 (p)
Federates States of Micronesia 3.2 1.7 -2.2 1.0 1.8 0.1 -4.0 0.1 0.3
Fiji 3.1 2.2 -0.9 -1.4 2.7 1.8 4.6 4.5 4.3
Kiribati 0.8 2.8 2.2 0.3 -0.2 3.4 2.4 3.8 2.9
Marshall Islands 2.5 1.3 3.8 -1.7 0.0 4.7 3.0 0.5 1.7
Palau 1.1 2.9 1.7 -10.7 5.2 5.5 -0.2 8.0 2.2
Papua New Guinea 4.3 1.0 7.2 6.1 10.7 8.1 5.5 5.8 19.3
Samoa 1.4 5.4 1.1 -6.4 6.2 1.2 -1.1 1.9 2.8
Solomon Islands 4.6 -1.4 6.4 -4.7 12.9 4.7 3.0 1.5 3.3
Tonga 2.4 2.7 -1.1 2.6 1.3 -1.1 -0.3 2.3 2.7
Tuvalu 3.8 0.0 6.4 -4.4 8.5 0.2 1.3 2.5 2.5
Vanuatu 4.0 1.8 5.2 3.3 1.2 1.8 2.0 2.9 -4.0
PICs Average 2.8 1.9 2.7 -1.5 4.6 2.8 1.5 3.1 3.5
Notes: 1. Data from 1990–2005 are extracted from World Bank’s online version of World Development Indicators, 2014.
2. Data from 2007–2015 are extracted from IMF’s World Economic Outlook (April, 2015), except for Fiji which is sourced from Ministry of Finance 2015 Budget Supplement and RBF press release number 13/2015.
3. p- projected.
Source(s): Data derived from the World Bank's World Development Indicators (online), 2014, Ministry of Finance 2015 Budget Supplement and IMF’s World Economic Outlook (April, 2015).
46
The pressing global issues of surging food and fuel prices along with the GEC have also
seen heightened economic difficulties for the PICs13. Being heavily dependent on
remittances and tourism earning, these countries are often susceptible to global
downturn. The PICs in general, with their natural geographical constraints of smallness,
population size, isolation from major markets and impact of natural disasters, have had
many difficulties in integrating fully into the world market (Singh and Prasad, 2008).
3.2.2 Key sectoral performances Fiji’s economy has traditionally been dominated by primary industries: agriculture,
especially sugarcane; mining; and fisheries. However, the service sector, particularly,
the tourism industry, has become the largest foreign income earner in the recent past.
While the agricultural sector has played a key role in the Fijian economy by providing
employment, food security and foreign exchange earnings ((Ministry of Finance, 2014),
its contribution has been declining significantly over the last four decades (Table 3.2).
Its susceptibility to natural disasters, the expiry of land leases and declining preferential
trade prices have undercut its contribution to GDP from 25% in the 1970s to close to
13% in 2012, a drop by almost 50%.
Table 3.2 Sectoral contribution to real GDP in Fiji, 1970–2013
Sectors 1970s 1980s 1990s 2000s 2010 2011 2012 2013
Agriculture 24.7 20.1 19.5 14.6 12.2 13.4 12.7 12.2
Industry 21.9 21.0 24.5 20.6 18.8 18.7 18.2 20.2
Service 53.4 58.9 56.0 64.8 69.0 67.9 69.1 67.6
Notes: 1970s represents average of 1970-1979, 1980s represent average of 1980-1989, 1990s represent average of 1990-1999 while 2000s represent average of 2000-2009.
Source: Calculations based on data from the World Bank's World Development Indicators (online), 2015.
On the other hand, the services sector in Fiji, which includes sectors such as tourism,
transportation, financial services, and telecommunication and information services, has 13 See Feeny (2010) and Jayaraman (2011) for detailed discussion on the economic impact of the GEC in the PICs.
47
shown resistance to negative internal and external pressures and has continued to
contribute extensively, contributing close to 70% of GDP in recent years. The
contribution of the industrial sector has remained fairly stable at around 20% of GDP in
the last four decades.
3.2.2.1 Sugar
The sugar industry remains one of the largest contributors to goods export earnings and
supports the livelihood of close to 200,000 Fijians in the country (Ministry of Finance,
2013). Its key export markets are the USA, the UK and the EU. Nevertheless, the
industry which has been in existence and thriving for over a century has been on the
brink of collapse for the last decade or more. Among others, studies by Kumar and
Prasad (2002) and Prasad and Narayan (2008) have pointed out issues such as expiry of
land leases, milling inefficiencies, declining preferential prices, natural disasters and
withdrawal of huge government support through deregulation policies as some of the
contributory factors towards its decline.
Table 3.3 Exports of Fiji’s key activities
Years 1980s 1990s 2000s 2010 2011 2012
2013 (p)
2014 (f)
2015 (f)
Sugar production (000 tonnes) 418 407 246 96 122 138 144 171 175
Sugar exports (F$m) 151 245 195 70 127 175 142 183 192
Domestic Fish exports (F$m) 21 54 156 204 95 53 84 88 94
Garments exports (F$m) 29 188 146 99 90 89 107 109 113
Gold exports (F$m) 35 67 78 148 143 137 82 84 85
Tourism earnings (F$m) 163 401 941 1,194 1,286 1,300 1,318 1,392 1,434
Visitor arrivals (000) 214 322 551 632 675 661 658 693 701
Notes: 1.1980s represents average of 1980–1989, 1990s represent average of 1990–1999 while 2000s represent average of 2000–2009.
2. p = provisional; f = forecasted. 3. Data from 1975–2012 are derived from Fiji Bureau of Statistics (Key Statistics, March 2014
issue). 4. Data from 2013–2015 are extracted from Ministry of Finance 2015 Budget Supplement.
Source(s): Fiji Bureau of Statistics (Key Statistics, March 2014) and Ministry of Finance 2015 Budget Supplement.
48
As a result, production of sugar has declined from producing in excess of 400,000
tonnes in the 1980s to producing less than 200,000 tonnes as forecasted for 2015 (Table
3.3). This has also caused the inevitable decline in export revenue from processed sugar
in recent years. The removal of the preferential entry of local products to the EU markets
by 2017 is also expected to bring about huge challenges to the Fijian economy and in
particular to the sugar exports.
3.2.2.2 Fisheries
The fisheries sector in Fiji is also a major one. Most of the country’s fisheries products
are exported to key markets in New Zealand, Japan, China, Thailand and the USA. The
domestic sector in recent times has vastly increased. Its contribution to domestic export
earnings, with significant increases since the 1980s from around F$20 million to more
than F$90 million forecasted in 2015 (Table 3.4). The exports of Fiji’s fish also benefits
largely from the interim EPA trade agreement to which Fiji and PNG are signatories. Fiji
has also provided a draft Fisheries notification to the EU to lobby to gain access to
global sourcing of fish from any vessel regardless of the country of origin, process it
onshore and export to the EU as Fijian fish. This if successful is expected to provide
increased opportunities for further growth and development of the fisheries sector in Fiji.
3.2.2.3 Garment industry
The garment industry in Fiji reached its peak when the Tax Free Factory scheme was
introduced after 1987 to generate employment and income. The industry as such has
greatly benefited under the South Pacific Regional Trade and Economic Cooperation
Agreement (SPARTECA) which allows preferential market access for products from
PICs to enter Australian and New Zealand markets. Since then, the industry has
experienced significant increases in export earnings from an average of F$30 million in
the 1980s to almost F$200 million in the late 1990s (Table 3.3). However, the industry
has been declining since mid-2000. This is largely as a result of the loss of preferential
access in its major markets and the difficulty faced by local industries to compete in the
global market. On average, in recent times, it brings in around F$100 million in export
earnings annually.
49
3.2.2.4 Mining
The mining sector in Fiji is largely characterised by the export of gold. Other mining
activities in the country include extraction of silver, copper, zinc and bauxite. In recent
years, the mining sector has increased the issue of exploration licenses for new mineral
deposit sites (Ministry of Finance, 2013). The largest mineral deposit in Vatukoula has
contributed immensely to the production of gold and export earnings from the mineral
resource in Fiji. The export of gold contributes around 6% to total export earnings which
has significantly increased from an average of F$35 million in 1980s to around F$130
million in 2012 (Table 3.3). However, in the recent time, it is only bringing in around
F$80 million of export earnings. The fluctuating demand for gold in the global market
has resulted in volatile export earnings from the export of gold from Fiji.
3.2.2.5 Tourism
The largest service sector in Fiji is the tourism sector, which has become the leading
foreign exchange earner in the country since the late 1990s. This sector also continues to
be one of the key economic drivers at present. The export income from this sector has
significantly increased from an average of F$160 million in the 1980s to almost F$1,400
million in 2014 (Table 3.3). It is expected to even surpass this figure in 2015 reflecting a
greater contribution of this sector to the economy’s growth performance.
Being also highly vulnerable to global market conditions and the visiting countries’
domestic conditions, the industry has experienced significant movements in visitor
arrivals in times of political instability in the country. Significant reduction in visitor
arrival was noted during the time of 2000 coup where it recorded a fall in visitors from
410,000 in 1999 to 294,000 in 200014.
Major tourist source countries to Fiji over the last four decades have been from Australia
and New Zealand, which contribute around 70% of visitor arrival in Fiji (Table 3.4). In
fact, Australia alone boasts around half of annual visitors to Fiji. Other countries from
which Fiji attracts tourists include the UK, the USA, Japan, European countries and 14 See Appendix A for the trend in visitor arrivals in Fiji over the 1975–2014 periods.
50
from other PICs. The recent increase in visitor arrivals in excess of 690,000 in 2014 is
largely attributed to promotional campaigns undertaken by Tourism Fiji in its traditional
and emerging tourist markets. The sector alone contributes almost one–third to Fiji’s
total foreign earnings and employs close to some 30% of the total labour force.
Table 3.4 Visitor arrivals in Fiji by country, 1970–2014 (%)
Years 1970s 1980s 1990s 2000s 2010 2011 2012 2013 2014 Australia 35% 40% 28% 37% 50% 51% 51% 52% 50%
New Zealand 23% 12% 16% 18% 15% 15% 16% 16% 18%
USA 20% 18% 13% 13% 8% 8% 9% 8% 9%
UK 4% 3% 8% 8% 4% 4% 3% 3% 2%
PICs 6% 6% 6% 6% 6% 6% 6% 6% 6%
Others 13% 21% 29% 18% 16% 16% 16% 15% 15%
Total (000) 167.1 214.5 322.2 473.6 631.9 675.1 660.6 657.7 692.6
Notes: 1970s represents average of 1970–1979, 1980s represents average of 1980–1989, 1990s represent average of 1990–1999 while 2000s represent average of 2000–2009.
Source(s): RBF Quarterly review, various issues.
3.2.3 Investment and Savings
Fiji has experienced mixed levels of investment over the last four decades. The
investment climate in the economy has been severely affected by internal political crisis,
which resulted in an all-time low level of investment in 1992 at around 11.5% of GDP.
However, the level of investment peaked at 27% of GDP in 1998 but since then has
averaged around 20% of GDP (Table 3.5). Prasad (2012b) claims that the unfavourable
trend of investment is as a direct result of political instability in the economy.
Governments in the past have initiated various tax incentives and policies to promote
private and foreign investment, such as creation of tax free zones, double taxation
agreements and removal of remittances taxes, among others, to attract investment in the
economy. Despite these incentives and investment promotion efforts, the level of
51
investment has remained low. However, on the back of huge public investment on
infrastructure development and purchase of new aircrafts, level of investment reached an
all-time high of 29% of GDP in 2013. It is also forecasted to remain above the
government target of 25% of GDP in 2014.
The level of savings, which has important implications for the investment levels, is also
noted to remain at low levels (Table 3.5). This in most instances has remained below
investment levels, implying that foreign savings have largely been used to finance
investment in the country. However, the level of savings also reached its peak in 1998,
in the same year as the investment level was at its highest. Nonetheless, since then it has
significantly declined to average 10% of GDP and remains well below the 1980s level.
Table 3.5 Level of Investment and Savings in Fiji as a % of GDP, 1970–2014
Years 1970s 1980s 1990s 2000-08 2010 2012 2013 2014 Investment (% of GDP) 19.9 18.6 17.0 17.4 15.6 17.1 29.0 26.0
Savings (% of GDP) 17.3 19.0 16.9 9.5 13.2 n.a n.a n.a
Notes: 1.1970s represents average of 1970–1979, 1980s represent average of 1980–1989 while 1990s represent average of 1990–1999.
2. n.a – not available. Source(s): Reserve Bank of Fiji Key Statistics (various issues) and The World Bank's World
Development Indicators (online), 2014.
3.2.4 Monetary and Fiscal policy
The objectives of monetary policy pursued all over the world are the same regardless of
the size and structures; Fiji’s economy is no exception. Fiji’s monetary policy continues
to be set by the RBF with the principal objectives of safeguarding the country’s foreign
reserves, promoting a sound financial structure and maintaining stable prices desired to
create an environment conducive to growth and investment. As such, Fiji has a strong
and vibrant financial sector that accommodates one central bank, one superannuation
fund, five commercial banks, three credit institutions and nine insurance companies.
52
Lending interest rates have gradually declined in the last decade from a high of 11.2% in
1995 to a low of 5.8% in 2014, in an effort to boost investment in the economy (Table
3.6). Deposit interest rates have also declined accordingly and a significant decline has
been noted in the real interest rates. In particular, it has even gone into negative in recent
years where it recorded -0.04% and -0.5% in 2005 and 2010, respectively (Table 3.6).
This has been a deliberate policy by the central bank to discourage savings and
encourage borrowings to stimulate growth in the midst of GEC.
However, these historically low interest rates did not appear to be stimulating the
economy because while they encouraged individuals and businesses to borrow money,
they discouraged people and businesses from savings, and as a result savings have been
declining (Table 3.5). Hence, when the interest rates have become difficult to reduce
even further to exercise monetary policy, currency devaluation as a policy option has
been exercised in 2009 to safeguard reserves and give a boost to the local economy.
Table 3.6 Interest rates (%) in Fiji, 1995–2013
Years 1995 2000 2005 2008 2009 2010 2011 2012 2013 2014
Deposit interest rate (%) 6.8 3.1 1.8 2.8 4.9 5.4 3.7 2.4 2.1 1.9
Lending interest rate (%) 11.2 8.4 6.8 8.0 7.9 7.5 7.5 7.0 6.1 5.8
Real interest rate (%) 10.0 13.7 -0.04 6.6 6.4 3.1 1.7 3.4 6.6 n.a
Notes: n.a – not available. Source: The World Bank's World Development Indicators (online), 2015.
However, the overall monetary discipline has been well maintained by the RBF as
foreign reserves and inflation rates are being maintained at acceptable levels in recent
years despite political instability and expansionary fiscal policies. Inflation rates have
been significantly high in the late 1970s and 1980s reaching a high of 14.5% in 1980 but
thereafter have gradually declined with some exceptions (Table 3.7). A careful look at
the domestic inflation rates indicates that the economy has found it very difficult to
maintain inflation rates below their targeted level of 3% (Ministry of Finance, 2014).
53
Levels of foreign reserves have also been maintained at adequate levels over the years
and more recently have been well on target to cover at least 5 months of retained imports
and factor services (Table 3.7). Fiji’s adoption of a fixed exchange rate and periodic
currency devaluation15 seems to have served the country well in terms of maintaining
their level of foreign reserves. It has been noted that the level of foreign reserves
achieved a boost due to the series of devaluations in the economy.
Table 3.7 Inflation rates (%) and level of foreign reserves in Fiji, 1980–2014
Years 80s 90s 00-05 2007 2009 2011 2012 2013 2014
Inflation (year-on-year % change) 7.5 4.2 2.6 4.8 6.8 6.4 2.5 3.4 0.1
Foreign reserves (F$m) 179 532 758 805 1,091 1,513 1,636 1,778 1,811
Foreign reserves (months of imports) 3.0 3.6 3.3 2.7 3.8 5.0 5.2 4.9 4.7
Notes: 1. 80s represents average of 1980–1989 and 90s represent average of 1990–1999 while 00–05 represent average of 2000–2005.
2. Data on foreign reserves (F$m and months of imports) from 1980 to 1996 are extracted from Dulare (2005) while from 1997 to 2014 are extracted from RBF Quarterly review, various issues.
Source(s): The World Bank's World Development Indicators (online), 2015; Dulare (2005); Reserve Bank of Fiji Quarterly Review (various issues).
As such, after a lag of one year, foreign reserves picked up by 72% in 1988 after the
1987 devaluation. Thereafter, it increased by 37% after 1998 devaluation while it made a
massive jump after 2009 devaluation increasing from F$558.7 million in 2008 to F$
1,090.6 million in 2009. This resulted in an increase by 95% by the end of 2009. Hence,
since the last devaluation, foreign reserves continue to stay at stable levels.
Moreover, in the area of fiscal policy, government has opted for an expansionary fiscal
policy resulting in annual budget deficits for more than a decade now, with exception in
2008. The budget deficit has averaged around negative 2.5% of GDP since 2000 (Table
3.8). The government has also in recent years introduced substantial changes into the tax
structure of the economy, largely targeted towards broadening the tax base (Ministry of
Finance, 2014).
15 Further discussion on Fiji’s exchange rate system and on the issue of currency devaluation is found in Section 3.
54
Table 3.8 Budget deficits and debt levels in Fiji, 2000–2014(p)
Years (20xx) 00 02 04 06 08 10 12 13 14 (p)
Net Budget balance (% of GDP) -3.1 -5.6 -4.8 -2.9 0.5 -2.1 -1.1 -0.5 -2.0
Total Government debt (% of GDP) 41.0 47.8 51.0 53.3 50.5 54.7 53.4 51.4 49.7
External debt (% of total debt) 14.3 10.3 7.3 14.6 16.5 16.2 25.4 28.3 28.0
Domestic debt (% of total debt) 85.7 89.7 92.7 85.4 83.5 83.8 74.6 71.3 72.0
Note: p – provisional Source: Ministry of Finance Budget Supplement (various years)
Government debt as a percentage of GDP has increased from 41% in 2000 to around
50% in 2014 (Table 3.8) as governments fiscal position remains quite tight. Close to
90% of the total government debt until 2006 had been financed domestically. However,
given Fiji’s favourable credit rating in 2006, the government issued its first ever 5–year
international bond worth US$150 million, which was repaid in 2011 with a roll–over
global bond valued at US$250 million (Ministry of Finance, 2012). This has doubled the
share of external debt from 14% in 2000 to 28% in 2014.
3.3 Trade policy
In an effort to foster trade integration into the global economy16, Fiji has since 1989
become a party to several bilateral, regional and international trade agreements17. Fiji
has become a member of the United Nations, the Commonwealth, the Pacific
Community, and the Pacific Islands Forum to allow smoother flow of goods and
services across boundaries. Being centrally located with a relatively well developed
economy and infrastructure in the region, Fiji is the host to many regional and
international organizations.
16 A well-oriented overview of Fiji’s trade performance is provided in Section 4.3 of Chapter 4 and Section 5.3 of Chapter 5. 17 See Appendix B for an overview of some of the bilateral, regional and international trade agreement to which Fiji is a party to.
55
Fiji also became a member of the General Agreement on Tariffs and Trade (GATT) in
1993 and started to dismantle various tariff and non-tariff barriers since the mid-1980s.
It is also now a member of its successor organisation, the World Trade Organization
(WTO), formed just a year later. The countries in the region including Fiji have also
accelerated the pace of trade liberalisation for achieving greater economic outcomes.
With this aim, in early 2000, PICTA (Pacific Island Countries Trade Agreement) was
formed amongst the 14 countries in the region to accelerate trade by reducing trade
barriers amongst them. Another trade agreement, PACER (Pacific Agreement on Closer
Economic Relations), which also includes Australia and New Zealand, was formed for
facilitating such trade with the two most developed neighbouring countries in the region.
In 2007, regional economies also agreed to reduce barriers in trade in services.
Fiji thereafter has gradually put in efforts to remove trade barriers against its trading
partners but it still remains a long way from when the flow of goods and services in Fiji
is freed up. The government is also currently in the process of formulating its first Trade
Policy Framework which would provide a roadmap for devising current, inclusive and
beneficial trade policies for the country (Ministry of Finance, 2014). The framework is
expected to incorporate real issues and challenges faced by stakeholders and identify
new opportunities in trade negotiations.
Despite efforts at trade liberalisation over the years, Fiji continues to rely on trade taxes
as a source of government revenue. Therefore, efforts to open up the market completely
is taking a long time as huge concern is being raised towards the loss of tariff revenue in
the country. However, Gounder and Prasad (2012) strongly argue that Fiji should not be
part of any preferential trade agreements as it results in net loss. They recommend that
Fiji adopt unilateral trade liberalisation policies to realize maximum gains. They add that
PICs, to which Fiji is no exception, have shown commitment to liberalise regionally and
it is about time for them to free up their market to the rest of the world as well without
having to be committed to any preferential trade agreements.
56
The economy as well in an effort to foster export growth has been focusing on export
promotion and import substitution policies. Some of the policies currently in place
include the implementation of the ‘Buy Fiji Made’ campaign and programmes such as
‘National Export Strategy’, ‘Import Substitution and Export Finance Facility’ and the
‘Export Promotion and Food Security’ programmes.
Nonetheless, when the measure of globalisation is analysed, gradual integration of Fiji in
the world market is noticeable. Konjunkturforschungsstelle18 (KOF for short) shows that
the trade and capital restrictions in the economy have been gradually declining (Figure
3.2). This has resulted in a gradual increase in the flow of trade and foreign investment
in the country.
Figure 3.2 Fiji's KOF economic and trade index, 1970–2011
Notes: 1. Actual flows measure the flow of trade, foreign direct investment and portfolio investment in the country.
2. Restrictions measure the level of restriction imposed on trade and capital using hidden import barriers, mean tariff rates, taxes on international trade and an index of capital controls.
Source: KOF Index of Globalisation (2014).
Furthermore, Fiji has repeatedly opted for currency devaluation as a better manage way
of managing capital outflows, and foreign reserves and in an effort to achieve favourable
18 Is a global index that measures the three main dimensions of globalisation: economic, social and political. Data has been obtained from http://globalization.kof.ethz.ch/.
01020304050607080
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
Actual Flows Restrictions
57
gains on trade performance in the country. In the last three decades since 1987, Fiji for
these kinds of reasons has devalued its currency on four separate occasions.
3.3.1 Devaluation in Fiji
3.3.1.1 Exchange rate adjustments in Fiji – A brief history
In the early 1930s, the Fijian currency was measured in pounds and the pound was fixed
to the Pound Sterling from 1934 to 1969. On 19 November 1967, for instance, £1.00
Sterling was equivalent to FJ$2.22 (£1.11 Fijian pounds) which was equal to US$2.80
(FBOS, 1974). This meant that FJ$1.00 was the same as US$1.26. On 20 November
1967, the US dollar was realigned to be worth FJ$1.08 and £0.42 Sterling. Then on 28
November 1967, FJ$1.00 (£0.50 Fijian pound) was worth US$1.148. These exchange
rate realignments resulted in an overall net devaluation of the Fiji pound by
approximately 9% (FBOS, 1974).
Thereafter, the realignment of exchange rates in 1971 meant that Fiji’s currency value
remained intact with the pound sterling. This resulted in the appreciation of the Fijian
dollar by 8.57% against the US dollar. As at end of 1971 under the IMF membership
resolution, FJ$1.00 was equal to US$1.25. Subsequently, on 8 May 1972, the United
States changed its per value and then on 13 February 1973 it once again devalued its
dollar by 10% in per value from 0.92SDR19 to 0.83SDR. On the other hand on 23 June
1972, Britain advised IMF that the pound sterling would be allowed to float. This
impacted on Fiji dollar’s parity with the pound sterling, revaluing by 5.26% on 24
October 1972. This resulted in the Fijian dollar being pegged at FJ$1.98 to £1.00
Sterling. On 10 September 1973, FJ$1.88 was equivalent to £1.00 Sterling.
With effect from 25 February 1974, the Fiji dollar detached ties with pound sterling and
started using the US dollar as their central rate. The rate then was set at FJ$1.00 to
19 Special Drawings Rights (SDR) are the supplementary foreign exchange reserves maintained by the International Monetary Fund (IMF). Their value is based on a basket of key international currencies reviewed by IMF every five years.
58
US$1.25. However, since 7 April 1975, the Fiji dollar ceased links with US dollar and
thereafter the exchange rate for Fiji is determined on a daily basis based on the weighted
average of the currencies of Fiji’s major trading partners. More specifically, the currency
since then is being pegged to the currencies of Australia, New Zealand, Japan, the
United States and the United Kingdom. From early 2000, the UK pound was replaced by
the Euro in the currency basket. The weights used in the basket are based on a three–
year moving average of Fiji’s direction of trade and are reassessed annually but are not
published publicly.
3.3.1.2 Devaluation episodes in Fiji
The Reserve Bank of Fiji, which is tasked to maintain the stability of the domestic
currency, has devalued the Fijian currency by a total of 73% in its 28–year history.
Currency devaluation is a monetary policy option whereby the central monetary
authority of the country, with the approval of the Minister of Finance, officially reduces
the value of a country’s currency with respect to a foreign currency in a fixed exchange
rate regime to correct the exchange rate misalignment. The Fijian currency has
undergone four major currency devaluations, the first two occurring in 1987, the third
one in 1998 and most recently in 2009 (Table 3.9).
Table 3.9 Devaluation episodes in Fiji (1987–2009)
Year of devaluation Date of devaluation Rate of devaluation
1987 29 June 17.75 %
7 October 15.25 %
1998 20 January 20.00 %
2009 15 April 20.00 %
Source: Reserve Bank of Fiji Quarterly Review, various issues.
Fiji, after pegging its currency to a basket of currencies, undertook its first exchange rate
corrective measure through devaluation of its dollar by 17.75% on 29 June 1987. This,
as Jayaraman and Choong (2008) argue, was as a result of Fiji’s first ever military coup
on 14 May 1987, which triggered capital flight due to political imbalances in the
59
country. Similar arguments are also put forward by Chand (1998) and Jayaraman (1999)
who maintain that the devaluation in 1987 was as a result of coups and loss of
confidence in the economy and not as a result of any macroeconomic mismanagement.
To further discourage huge capital outflows and to protect foreign reserves, the RBF
once again devalued the currency by another 15.25% on 7 October in the same year.
Both the devaluations amounted in a net decline in the value of the domestic currency by
33% in 1987.
Furthermore, after almost one decade since the first devaluation, Fiji’s currency was
devalued again by a rate of 20% on 20 January 1998. Jayaraman (1999) and Jayaraman
and Choong (2008) argue that the 1998 devaluation by the RBF was a precautionary
measure in response to the East-Asian financial crisis, which was bound to affect Fiji’s
trade and capital flows. However, Chand (1998) argued that the flow on impact of
political uncertainty and the movement of financial ratios indicated the unsustainability
of the nominal exchange rate peg that triggered devaluation in 1998. His view is that the
East-Asian financial crisis just provided an opportune moment for the devaluation and
was not really the major reason for it.
Contrary to Chand (1998) and Jayaraman (1999), Dulare (2005) states that the severe
drought and the adverse weather conditions had a serious impact on sugar production
and exports, which created pressure on foreign reserves and ultimately led to
devaluation. This view was supported by Narayan and Narayan (2009) who also
maintained that the devaluation was carried as a measure to boost Fiji’s struggling
foreign reserve position in 1998.
The RBF undertook the fourth and most recent devaluation, by another 20% on 15 April
2009. The Fijian dollar was effectively reduced from being worth 57 cents before
devaluation to 45 cents after devaluation, against the US dollar. This in fact led to an
immediate drop in the value of the Fijian dollar by 12 cents. The RBF press release
explained the devaluation as a measure to cushion the severe effects of the GEC on the
domestic economy (RBF, 2009), adding that the devaluation will bring the Fijian dollar
60
in line with the major trading partner countries as it had significantly appreciated
unsustainably against them by around 20% since the 2007 and 2008 period.
Figure 3.3 Real effective exchange rate ( rE ) index in Fiji, 1975–2012
Source: IMF’s International Financial Statistics database, 2014
Figure 3.3 shows the movement of the trade weighted real effective exchange rate ( rE )
of Fiji’s currency against its major trading partner countries’ currencies. As is clearly
depicted, a sharp increase of the index in 1987, 1998 and 2009 points to the years of
devaluation in the country. One very important observation emerging clearly from
Figure 3.3 is that the Fijian currency immediately after devaluation stays devalued for
some time but starts to appreciate after a lapse of two to three years. This is noted for
1987 and 1998 devaluation; however, for the 2009 devaluation, currency starts to
appreciate after a lapse of just one year. This is argued to be possibly due to the effects
of the GEC and the surge in global oil price, which filters down to prices in the domestic
economy, resulting in an appreciation of the domestic currency.
Nonetheless, the varying impact of devaluation in each episode could also mean that
there would be more effects of devaluation on trade performance after 1987 and 1998
devaluations than after the 2009 devaluation episode. However, on the aggregate, the rE in general is on an increasing trend implying that the currency has been kept on a
60
70
80
90
100
110
12019
75
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2005
= 1
00
33% devaluation in 1987
20% devaluation in 2009
20% devaluation in 1998
61
devaluationary trend path. Hence, this trend of rE indicates that currency devaluation
would have a favourable impact on the overall trade performance in the economy.
However, it is argued that the short–run response to be analysed by the J-curve
phenomenon could vary and the impact be limited. This is because the impact would be
difficult to maintain in the long–run due to the accompanied appreciation of the
domestic exchange rate either through internal or external pressures.
3.3.1.3 Applicability of fixed exchange regime in Fiji Many scholars – de Brouwer (2000), Duncan (2005), Jayaraman (2001, 2004),
Jayaraman et al. (2007), Jayaraman and Choong (2009), Davies (2010) and Yang et al.
(2011) among others – have over time argued on the applicability of a fixed exchange
rate regime in the PICs and for possibilities of other exchange rate management options
available to the island countries. Yang and others (2011) argue that despite the fixed
exchange rate regimes adopted in most of the island countries in the PICs, exchange rate
policy still plays an important role in macroeconomic adjustment. They add that
although the flexible exchange rate acts as an automatic shock absorber of external
shocks, the flexibility of the fixed exchange regime either through one–off devaluation
or through movement within a narrow band provides the much needed stabilising effects
during times of adverse external shocks.
However, de Brouwer (2000) and Duncan (2005) argue that the PICs should abandon
their currencies and opt for either currency union or the adoption of the Australian
currency The choice of Australia is based which is one their major trading partner to
reduce administrative costs, manage liquidity in the foreign exchange markets and to
stabilize the exchange rate with the most important trading partner. However, studies by
Jayaraman (2001, 2004), Jayaraman et al. (2007) and Jayaraman and Choong (2009)
have argued against the idea of currency union including adopting the currencies of
either Australia or New Zealand (ANZ) because of several differences within the PICs
and with ANZ.
62
Similarly, Davies (2010) putting aside the idea of currency union, has argued in favour
of fixed exchange rates applicability for small open economies such as those on the
PICs. He contends that the use of a fixed exchange rate in the Pacific region reduces
exchange rate risks and reduces the need to establish complex institutions in situations of
scarce financial and human resources. Additionally, he debates against the idea of
having a flexible exchange rate in the region by arguing that the flexible exchange rate
system requires a complex monetary policy environment and greater institutional
infrastructure, among other important prerequisites. He adds that it is appropriate for
developing countries, as in the Pacific region with limited links to global financial
markets, less diversified production and export structure along with shallow financial
markets.
3.4 Concluding comments This chapter provides a brief discussion of Fijis historical growth trends and
performance of its key economic sectors. The issue of currency devaluation is also
discussed. It has been noted that Fiji has experienced episodes of low economic growth,
increasing debt levels, annual budget deficits along with political instability, which have
severely impacted its growth trend. The economy is found to be largely based on the
services sector, particularly tourism, which contributes significantly to foreign earning
and GDP. Nevertheless, the sugar industry along with garments, and fisheries continue
to be important industries in the country, providing export earnings and livelihood for
many in the country.
It is also noted that trade liberalisation continues to be an important agenda for Fiji’s
economic reforms and policies aimed at improving trade performance are gaining
dominance. Hence, when fiscal and monetary policies had become tight, one such policy
adopted to improve trade and growth performance is via currency devaluation. The next
two chapters present a well oriented survey of the literature on the impact of currency
devaluation on trade performance, along with estimating and discussing the empirical
results of the same for Fiji.
63
CHAPTER 4
AGGREGATE TRADE FLOWS IN RESPONSE
TO CURRENCY DEVALUATION
4.1 Introduction
Currency devaluation has been one of the widely studied concepts in the economic
literature, particularly because of its role in maintaining stable fixed exchange rates
regimes. Since long, many developing and emerging economies have been using
currency devaluation as an instrument to enhance their international trade
competitiveness. The dominant view up until the late 1970s was that devaluation
improves trade balance and consequently, enhances economy’s employment and output.
This view, however, was challenged by the end of 1970s (Acar, 2000). An alternative
approach emerged after the 1970s oil crisis which argued that devaluation could be
contractionary, especially in developing countries. This argument was based on the fact
that developing countries depend heavily on imports of capital and intermediate goods
and therefore, more likely to end up facing higher import bills after devaluation.
A plethora of studies in the context of developing and developed countries have
investigated the relationship among currency devaluation, trade performance and
economic growth. Some of the important research include Himarios (1985), Gupta-
Kapoor and Ramakrishnan (1999), Bahmani-Oskooee et al. (2002), Lal and Lowinger
(2002), Bahmani-Oskooee and Gelan (2012) and Ayen (2014). The focus of recent
research has also included the impact of currency devaluation or depreciation on sectoral
and bilateral trade flows (Yazici (2006), Narayan (2006), Bahmani-Oskooee and Wang
(2007), Wang et al. (2012), and Bahmani-Oskooee et al. (2014)).
However, as countries differ with respect to population size, geography, resource
endowment, economic and social structures along with other factors, the existing
64
literature provide contrary evidence on the effectiveness of devaluation on aggregate,
sectoral and bilateral level trade flows. Some studies suggest favourable impact of
devaluation on trade performance (Arora et al. (2003), Bahmani-Oskooee and Wang
(2007), Soleymani and Saboori (2012) and Bahmani-Oskooee and Zhang (2013)) while
some argue that devaluation further deteriorates trade balance position of an economy
(Yazici (2006), Narayan (2006) and Yazici and Klasra (2010)). Going by mixed
proposition, this chapter attempts to empirically investigate the impact of currency
devaluation on Fiji’s aggregate trade performances.
This chapter is organised as follows. Section 4.2 presents the review of the literature
focusing on the response of aggregate trade flows to currency devaluation or
depreciation. This section also includes studies that provide evidence on the hypothesis
of the J-curve phenomenon on aggregate trade balance. Relevant literature on this
subject in the context of Fiji is also reviewed. In Section 4.3, aggregate trade
performances in Fiji over the last three decades are discussed. Section 4.4 of this chapter
empirically examines long– and the short–run effects of devaluation on Fiji’s aggregate
trade performance. Estimates are provided and the results on the test of the J-curve
phenomenon on aggregate trade balance are also discussed. Section 4.5 concludes the
chapter.
4.2 Literature review: Aggregate trade flows This section reviews the relevant literature on the relationship between exchange rate
and the aggregate trade balance. As discussed in Section 2.2, the use of econometrics
techniques are found to be recent, advanced and have been proved to adequately address
the issues faced in other approaches. Therefore, the review is restricted to only those
studies that have examined the relationship between exchange rate and trade balance
using an econometric technique. Towards the end of this section, relevant literature in
the context of Fiji’s economy is also reviewed.
65
4.2.1 On the aggregate trade flows and the J-curve phenomenon In 1930s, world economy saw one of the largest currency devaluation episodes when at
least nine of the leading world economies (includes Australia, France, Italy, Japan and
the United States) devalued their domestic currencies. The great depression made these
countries vulnerable to external shocks and the abandonment of the gold standard was
decided. As a result, currencies were devalued by up to 40% to stabilize the currency
rates and to stimulate economic growth (Acar, 2000).
Again, following 1970s oil crisis, many industrial countries opted to abandon the fixed
exchange rate regime and implemented the floating rates (Bahmani-Oskooee and
Miteza, 2003). However, even after the mass implementation in industrial economies,
many developing countries around the globe maintained their fixed exchange rate
regimes (Bahmani-Oskooee and Miteza, 2003). The central monetary authorities in the
countries with fixed exchange rate regimes, though, resorted to currency devaluation,
often in response to strong external pressure and disturbances. Similarly, countries with
floating exchange rate regime were also often faced with currency depreciation as a
result of fluctuations in their domestic and external market conditions.
Magee (1973) is probably one of the pioneering scholars who attempted to examine the
presence of the J-curve phenomenon in the USA over the 1969–1973 periods. His
seminal paper highlights the theoretical consequences on the trade balance of adjustment
lags due to contractual agreements and quantity adjustments after devaluation. By
plotting the graphical response of the trade balance to devaluation in the USA, the study
does not find evidence of the J-curve phenomenon. However, the author argues that
devaluation will have favourable effect on the trade balance in the long–run.
Bahmani-Oskooee (1985, 1989) utilise quarterly data over the 1973–1980 period to test
for the presence of the J-curve phenomenon in Greece, India, Korea and Thailand. The
study models trade balance as a function of domestic income, foreign income, money
supply, foreign money supply and real exchange rate. The study uses traditional
definition to find evidence of the J-curve phenomenon for Greece and India only.
66
Similarly, Flemingham (1988) using quarterly data from 1965Q1–1985Q2 period for
Australia fails to find evidence of the J-curve phenomenon using the traditional
definition. However, he ascertains that there might have been some indications of a
delayed J-curve during the fixed exchange rate regime period prior to 1974 in Australia.
Some other studies on developed countries in the late 1980s (by Moffett (1989) on the
USA and Noland (1989) on Japan) did find evidence of the J-curve phenomenon. Both
of these studies used traditional definition to assess the presence of the J-curve
phenomenon.
Bahmani-Oskooee and Malixi (1992) use similar approach to Bahmani-Oskooee (1985)
to study the presence of the J-curve phenomenon using traditional definition. Using
quarterly data over the 1973–1985 periods for 13 LDCs, the study highlights that despite
most of the LDCs fixing their currencies to a major or to a basket of currencies;
exchange rate volatility cannot be completely avoided. This is because major currencies
tend to fluctuate against each other resulting in the exchange rate volatility. The study
finds that currency depreciation improves trade balance in the long–run for Brazil,
Greece, India and Korea. The study also highlights that the impact of currency
depreciation on trade balance in the short–run does not always follow the J-curve pattern
but may also provide evidence for patterns such as N-, M- and I-curves.
Moreover, with new developments in econometric techniques, Bahmani-Oskooee and
Alse (1994) employ the Engle–granger cointegration technique to study the impact of
real exchange rate on trade balance for 41 countries. In particular, the study included 19
developed and 22 less developed countries over the 1971Q1–1990Q4 period. The study
finds a long–run positive impact of devaluation on the trade balance for Brazil, Costa
Rica, the Netherlands, Singapore and Turkey while a negative impact on the trade
balance for Ireland. With regard to Canada, Denmark, Germany, Portugal, Spain, Sri
Lanka, the UK and the USA, no long–run relation of exchange rate with the trade
balance could be established. Hence, using the traditional definition, the J-curve
phenomenon is only valid for Costa Rica, Ireland, the Netherlands and Turkey.
67
Gupta-Kapoor and Ramakrishnan (1999) utilise the error-correction model to study
Japan’s trade balance over the 1975Q1–1996Q4 period using the reduced form of the
trade balance model. Using the Johansen’s likelihood cointegration test, the study finds
long–run favourable impact of exchange rate on the trade balance using both nominal
and real values in the model specification. Hence, using the IRF analysis, the study finds
support for the J-curve phenomenon on Japan’s trade balance. Recently, Ono and Baak
(2014) also provide evidence on the J-curve phenomenon on Japan’s trade balance.
Narayan (2004) using the reduced form of the trade balance model finds evidence of the
J-curve phenomenon using IRF analysis for New Zealand for the 1970–2000 period. The
analysis shows that New Zealand’s trade balance after currency depreciation deteriorates
for the first three years followed by positive impact at longer lags. Conversely, in the
context of another developed economy, Akbostanci (2004) fails to find evidence of the
J-curve phenomenon using traditional definition and IRF analysis for Turkey. Similarly,
Georgopoulos (2008) using IRF analysis does not find evidence of the J-curve
phenomenon on Canada’s trade balance.
Rahman and Islam (2006) study the same for Bangladesh using Engle-Granger
technique over the 1972–2003 periods. With the use of IRF analysis, the study finds
evidence of the J-curve phenomenon. The study emphasises that there are greater short–
run effects of exchange rate changes on the trade balance than in the long–run in
Bangladesh. Other studies, such as Rehman and Afzal (2003) for Pakistan and Gomes
and Paz (2005) for Brazil find evidence of the J-curve phenomenon using the same
approach.
In another comprehensive study, Bahmani-Oskooee and Kutan (2009) model the
reduced form of the trade balance for 11 emerging economies of Eastern Europe over
the 1990M1–2005M6 period. Using the bound testing approach to cointegration and
error correction modelling technique, the study using the new definition finds presence
of the J-curve phenomenon only in three countries: Bulgaria, Croatia and Russia.
68
In another study, Kalyoncu et al. (2009) using quarterly data for 4 Latin American
countries (Argentina, Brazil, Mexico and Peru) over the 1979–2005 periods, finds long–
run favourable effects of devaluation on the respective trade balances. However, the
evidence of the J-curve phenomenon is obtained only for two (Argentina and Peru) of
the four countries under study. Using the IRF analysis, the study establishes that in
Argentina, trade balance starts to improve after three to five quarters while in Peru it
starts improving after four quarters. However, in the case of Brazil and Mexico, the
study does not find a long–run association and therefore concludes non-presence of the
phenomenon in these two cases.
Yusoff (2010) studies the effect of the real exchange rate on the trade balance of
Malaysia over the 1977Q1–2001Q4 period. Using VECM, the results suggest that a real
depreciation of the ringgit leads to an improvement in the trade balance in the long–run.
The study further highlights that immediately after depreciation, trade balance tends to
improve, after which the trade balance and output start to fall. Hence, using the IRF
analysis, the study finds evidence of an inverted J-curve phenomenon. However, in the
long–run, both the trade balance and domestic output improves in the case of Malaysia.
Bahmani-Oskooee and Gelan (2012) also use the reduced form of the trade balance
model and VECM technique to validate the presence of the J-curve phenomenon for 9
African countries20 over the 1971Q1–2008Q4 period. Hence, using the traditional and
new definitions, the analysis does not find support in favour of the J-curve phenomenon.
However, the long–run effect of real depreciation was found to be favourable only in the
cases of Egypt, Nigeria and South Africa.
In a more recent study, Wijeweera and Dollery (2013) apply the same reduced form of
the trade balance model and disaggregate Australia’s aggregate trade balance into goods
and services sector to investigate the J-curve phenomenon. They argue that by
20 The 9 African countries included in the study are Burundi, Egypt, Kenya, Mauritius, Morocco, Nigeria, Sierra Leone, South Africa and Tanzania.
69
disaggregating the data into goods and services sectors, its will allow them to trace the
impact of depreciation on specific sectors. Using traditional definition, the analysis finds
evidence of the J-curve phenomenon for the services sector but not for the goods sector
in Australia. The authors maintain that currency depreciation in Australia helps support
the improvement of the services trade balance but not the goods sector trade in the long–
run.
Recently, Musawa (2014) follows a similar modelling strategy but does not find
evidence of the J-curve phenomenon for Zambia using the traditional definition. The
study, however, finds a long–run positive impact of currency devaluation on the trade
balance.
4.2.2 On the impact of devaluation in the context of Fiji’s economy
Devaluation is not a new word in Fiji’s economic history. Fiji’s currency has been
devalued four times in the last 28 years of its history, with the most recent devaluation,
by 20%, in the year 2009. The pattern seems to be emerging that Fiji devalues its
currency every ten years or so: twice in 1987, by a total of 33%, in 1998 by 20% and
then in 2009 by a further 20%. These devaluation events offer largely inconclusive
scholarly debate (see Sections 3.3.1.2 and 3.3.1.3). Though studies analysing the impact
of exchange rate on aggregate trade balance are relatively few, the general consensus is
that devaluation improves exports and reduces imports in both the short– and long–run.
Nonetheless, in an analysis on Fiji’s aggregate trade balance, Narayan and Narayan
(2004a) find evidence in favour of the J-curve phenomenon.
A mere handful of studies in Fiji have looked into the relationship of exchange rate with
trade related variables. Before, moving the focus to the literature that has attempted to
establish an exchange rate and trade balance relationship, review of other studies that
have analysed the relationship of exchange rate with export and import demand in the
economy is deemed to be relevant in the context of this study.
70
Reddy (1997) is one of the earlier attempts to estimate the export and import demand
equations for Fiji. The author develops an export and import demand model to estimate
the relative price and income elasticities using the OLS method over the 1970–1994
periods. The export model is developed as a function of Fiji’s relative price of exports to
the weighted average of trading partners’ export price and GDP of trading partner
countries. The partners considered in the model include Australia, Japan, New Zealand,
the UK and the USA. The variables in the import equations, on the other hand, include
Fiji’s real GDP, import price of Fiji relative to Fiji’s consumer price index and a lagged
dependent variable for imports. The results on the income elasticities were estimated at
0.76 and 2.39 for export and import demand equations, respectively. On the same note,
the relative price elasticities for the export demand equation were found at -0.78 while
the same for the import equation was estimated at -1.53.
Hence, based on the results, the study concludes that the M-L condition for Fiji is
satisfied in the long–run. However, the author argues that though devaluation would lead
to an improvement in the trade balance in the long–run, it can be costly in the short–run
due to resource constraints. He further deliberates that the trade balance would follow
the J-curve pattern but does this without any empirical estimation. It is argued that
though this study provides some meaningful insights, its results are to be taken with due
care. It is noted that the author, when dealing with time–series data, fails to test for the
unit root properties of the variables. This leads one to conclude that the results are
estimated with the presence of unit root and are deemed to be spurious. The author also
ignores the inclusion of the exchange rate variable in the model but makes
recommendations on the impact of devaluation in Fiji. This also creates serious doubts
on his claim on the M-L condition and the likeliness of the J-curve phenomenon in Fiji.
Asafu-Adjaye (1999) conducts an empirical analysis on the impact of exchange rate
variability on the exports sector in Fiji using monthly data over the 1981M1–1997M6
period. Using the error correction and cointegration techniques, the author models long–
run export demand as a function of Fiji’s real GDP, foreign income, Fiji’s export prices
relative to trade weighted foreign prices, real effective exchange rate and a variable
71
capturing exchange rate variability. Though it is not explicitly mentioned in the study,
the author addresses some of the limitations of the research done earlier by Reddy
(1997). As part of its analysis, the study undertakes unit root and cointegration tests and
subsequently predicts both long– and short–run dynamics of exports in Fiji. The study
concludes that in the long–run, for every 10% currency devaluation, exports in Fiji
increase by approximately 8.2%. There is also evidence that exchange rate reductions
have significant and positive impact on exports in the short–run as well.
Rogers (2000) on the other hand, examines Fiji’s import demand equation over the
1968–1998 periods using the error correction modelling technique. He models import
demand as a function of Fiji’s real GDP, price of imports, real effective exchange rate
and average tariff rate. The short–run result states that devaluation by 10% makes
imports relatively more expensive and reduces import volume by 8%. For the long–run
relationship, the study draws similar conclusions though, the estimated magnitude of the
change in the import demand is not mentioned anywhere in the study. The study may be
criticized for containing some element of ambiguity. The clarity on how the short–run
estimates and the cointegrating relationship have been obtained is missing. There is no
mention of an error correction term in their import model specification.
Similar to Rogers (2000), Prasad (2000) attempts to model export demand function for
Fiji over the 1968–1998 periods. The export demand is modelled as a function of real
effective exchange rate, trading partner income and sugarcane production per hectare.
The estimation results suggest that in the short–run, 10% devaluation leads to
improvement in the exports by 7.2% while the long–run income elasticity of exports was
estimated at 2.4. However, the study does not discuss the long–run impact of
devaluation on export volumes.
Narayan and Narayan (2004b) also develop an export demand model of Fiji to estimate
income and price elasticities using Autoregressive-Distributed Lag (ARDL), Fully
Modified Ordinary Least Squares (FMOLS) and Dynamic Ordinary Least Squares
(DOLS) approaches over the 1970–1999 periods. The model is estimated as a function
72
of the weighted average of trading partners’ real GDP, export price index and trading
partners’ export price index. The results show that trading partner export price elasticity
ranges from 2.09 to 2.23 while the trading partner income elasticity ranges from 0.70 to
0.81. On the other hand, Fiji’s export price elasticity ranges from -1.30 to -1.49. They
comment further that devaluation is an effective policy tool for improving export
performance in the long–run. Similar results are also obtained in the short–run.
In another study by the same authors (Narayan and Narayan, 2005), they model import
demand equation for Fiji. Following a similar model specification as in their previous
study, they find long–run domestic income elasticity for imports ranging from 1.05 to
1.90. However, the empirical evidence provided by the two studies raises important
concerns as the authors fail to include exchange rate as one of their variables in their
studies. This leads to misspecification of their relative price indicator. Their failure to
incorporate an exchange rate variable in their models also raises serious questions on the
implications made regarding the impact of devaluation on export and import volumes in
Fiji.
Singh (2006) is one of those studies that paid special attention to the relative price
specification and included exchange rate in their model to estimate export and import
demand equations for Fiji over the 1970–2002 period. The inclusion of the real
exchange rate in the specification of the export and import demand equations is
consistent with the trade literature (see, for example, Bernard and Jensen, 2004; Fang et
al., 2006; Kumar, 2009; Haddad and Pancaro, 2010; Freund and Pierola, 2010 and
Haider et al., 2011). Hence, the model specification used in the study takes the following
specification:
Export demand equation: ttFtt
tt YT
PEPX ���� ���
�� lnlnln 210
Import demand equation: ttFtt
tt Y
PEPRM ��� ���
�� lnlnln 210
73
where tX and tRM are the real exports and imports, tP is Fiji’s export price index, FtP
is the trading partners’ price index, tE represents nominal exchange rate while tY and
tYT denote Fiji’s and trading partners’ real GDP, respectively. The study uses the three
econometric methods of Johansen Maximum Likelihood (JML), LSE Hendry’s General
to Specific (GETS) and FMOLS to model both the equations.
For exports, the long–run income elasticity is estimated to be close to 1 while the
relative price elasticity is around -1.25. For imports, the long–run income elasticity and
price elasticity are estimated to be around 1.20 and 0.50, respectively. The study argues
that since the M–L condition is fulfilled, real devaluation helps to reduce trade deficits in
the long–run in Fiji. However, they caution that the inadequate capacity of the domestic
economy to meet the increased global demand could limit the possible gains. It is also
important to note that in an improved version of this study, Rao and Singh (2007) show
that the failure to incorporate exchange rate in the export demand equation overestimates
the income elasticity by 40–65%.
Kumar (2009) develops an export demand equation for developing countries (Fiji, Papua
New Guinea and Bangladesh) using annual data mainly over the 1970–2002 periods.
With the use of the GETS approach, the author follows a similar model specification as
proposed by Rao and Singh (2007). The results indicate that the income elasticity is
around 1.08 while the relative price elasticity is -0.83 for Fiji. This result consistent with
other similar studies in Fiji supports the idea that real devaluation helps to stimulate
export growth in Fiji’s economy.
The studies reviewed above have focused their attention to separately estimate export
and import demand equations for Fiji. The results obtained in all the studies largely
highlight that devaluation improves exports and reduces imports in the long–run in Fiji.
Though many of these studies do not include exchange rate when modelling exports and
imports (Reddy, 1997; Narayan and Narayan, 2004b; 2005), studies carried out since the
74
work of Singh (2006) make an attempt to include it in their analysis (Rao and Singh,
2007 and Kumar, 2009) when estimating export and import demand equations for Fiji.
However, in the interest of this study, it is noteworthy that a few other influential studies
have analysed the impact of devaluation either on the trade balance or on its contribution
to economic performance in Fiji (Narayan and Narayan, 2004a, 2007; Narayan, 2013).
Narayan and Narayan (2004a) in their attempt to contribute to the J-curve literature
adopt the widely used reduced form of the trade balance model to model Fiji’s trade
balance. In particular, following the works of Rose and Yellen (1989), the authors
develop the trade balance model for Fiji as follows:
ttttt GNIFYREERME ����� ����� lnlnlnln 321
where tMEln is the real imports to real exports ratio, tREERln is the trade weighted
real effective exchange rate, defined as the number of units of domestic currency per
unit of foreign currency, tFYln is the weighted average of trading partners’ real income
and tGNIln is the measure of Fiji’s real domestic income. All these variables are
transformed into its logarithmic form.
The study using ARDL, DOLS and FMOLS methodology use annual time–series data
over 1970–2000 periods to estimate short– and long–run relationships among the
variables in the model. The long–run elasticity of the ME ratio with respect to REER
is negative. This indicates that a real devaluation of the exchange rate results in an
improvement of the trade balance in Fiji. The elasticity ranges from 0.12 to 0.30 but is
found to be statistically insignificant. This means that though favourable, devaluation in
Fiji does not influence trade balance in the long–run.
The results on the long–run impact of domestic income indicate that increase in
domestic income leads to deterioration of the trade balance in Fiji. The estimates show
that a rise in income by 10% reduces the trade balance by around 7.8% to 12.3% in the
long–run. However, the impact of the trading partner income is estimated to show
favourable impact on the trade balance for Fiji in the long–run. It is estimated that a 10%
75
increase in trading partner income leads to an improvement of the trade balance by
approximately 2.8% to 4.1% in the long–run.
Moreover, to validate the presence of the J-curve phenomenon in Fiji, the study uses the
traditional definition and the IRF analysis to examine the phenomenon. The IRF analysis
shows that due to a devaluation shock on the real exchange rate, trade balance
deteriorates in the first two years followed by improvements in later periods. Hence, this
movement on the trade balance validates the J-curve phenomenon in Fiji. The results are
also supported by the traditional definition. The short–run estimates show initial
deterioration of REER at earlier lags followed by a favourable response at longer lag
lengths.
Nevertheless, though the study is commendable for its contribution to the J-curve
literature in the context of a developing Pacific island country, it has at least three
shortcomings. First, the study does not disclose information on the trading partners
being considered in constructing the trade weighted trading partner income. Secondly, it
is also important to note that since this particular study is considering only aggregate
trade data, there are high chances that the study might have suffered from aggregation
biasness21.
This biasness as pointed by Bahmani-Oskooee and Brooks (1999), Pattichis (2012) and
Bahmani-Oskooee and Xu (2012) raises the possibility that significant trade flows of a
devaluing country with its one partner could be offset by an insignificant relation that
may exist with another trading partner country. Moreover, a significant trade flow of one
or few major commodities could also be offset by insignificant trade relation that might
exist for other sectoral trade with the rest of the world. Therefore, it is argued that the
difference in the impact of devaluation on various sectors and trading partners of the
devaluing country would vary. In fact, this would mean that the results obtained at the
aggregate level would not necessarily be the same for different commodities and trade
partners of a country. 21 Further discussion on this aggregation biasness is found in Section 5.1 of Chapter 5.
76
Thirdly, the study is a decade old, covering the period only until year 2000. Since then,
many developments that have taken place in the J-curve literature demand rethinking in
the context of developing economies. Additionally, given the fact that Fiji’s economy
has undergone another round of devaluation in 2009, it is an opportune time to
investigate if the J-curve phenomenon on Fiji’s trade balance is still valid.
Narayan and Narayan (2007) utilise the ARDL methodology to investigate the impact of
devaluation on output performance in Fiji over the 1970–2000 period. Following the
works of Edwards (1986) and Bahmani-Oskooee et al. (2002), they model Fiji’s income
as a function of money supply, government expenditure, real effective exchange rate and
foreign income. The results indicate that devaluation has a positive and significant
impact on output. In particular, 10% devaluation leads to about 2.3% increase in output
in the short–run and a 3.3% increase in output in the long–run. Additionally, the study
finds that an increase in government spending, money supply and foreign income has a
positive impact on output in both the short– and in the long–run.
Based on these results the authors argue that the use of devaluation policy adopted for
Fiji’s economy helps to achieve an expansionary effect on output in both the short– and
the long–run. They claim that a fall in the value of the Fijian currency makes Fiji’s
exports more competitive and helps boost demand in the world market. The authors
suggest that though the expenditure on imports is bound to increase after devaluation,
the higher export demand outweighs any fall in output caused by high import prices.
They reason that Fiji’s export demand is relatively elastic while the import demand is
highly inelastic with respect to real effective exchange rates. This results in the
favourable outcome.
In another paper, Narayan et al. (2012) analyses Fiji’s monetary policy transmission
over the 1990–2006 period using a structural vector autoregressive (SVAR) model. The
objective of the study is to examine the relationship between GDP, CPI, money supply
(MS), interest rate (IR) and NEER. The results highlight that the appreciation of the
77
NEER has negative and statistically significant effect on real output. This means that a
devaluation of the currency leads to an improvement in output performance.
However, since the data coverage is only over 1990–2006 periods, only one devaluation
episode is being accounted for. The study also does not explicitly investigate how the
impact of exchange rate channels down to output performance in Fiji. It is argued that
analysis on one of the important components of output, namely the trade balance, has
been ignored. Detailed studies on how currency devaluation influences trade
components in Fiji appears not yet undertaken.
Recently, Narayan (2013) develops a four-dimensional SVAR model to investigate the
effects of fuel import and devaluation policy on Fiji’s current account deficit and
economic growth over the 1979–2007 periods. Unlike the previous study by Narayan
and Narayan (2007) on Fiji which finds an expansionary effect of devaluation on output,
this particular study, to the contrary, finds an immediate contractionary effect. The
author argues that this different result could be attributed to the different dataset and the
use of different methodology. Despite the unfavourable effects in the short–run, the
long–run effects of REER are favourable. In particular, it is estimated that in the short–
run, devaluation of the exchange rate leads to a deterioration of current account balance
in Fiji.
Although this recent investigation on the devaluation, current account and output nexus
in the context of Fiji is commendable, it has at least three shortcomings. Like other
previous studies, it has not included the entire four episodes of devaluation in Fiji to
investigate if this relationship still holds true. That would require extending the dataset
beyond 2009. The study also does not account for the impact of devaluation on trade
performance which makes up significant component of the current account in Fiji.
Though the study has raised some critical questions on the direction of the impact of
devaluation policy on output, they discuss only its short–run effects and fail to address
the long–run relationship. Nonetheless, based on these review it is quite clear that the
78
literature dealing with devaluation in Fiji, while not extensive, still provides conflicting
results.
4.2.3 Findings from the Literature review: Aggregate trade flows The literature review in this section has focused on analysing the impact of currency
depreciation and devaluation on trade components in the experience of a selection of
international countries as well as Fiji. The survey has also analysed studies that have
examined the validity of the J-curve phenomenon in response to currency devaluation or
depreciation on aggregate trade balance.
The empirical literature on the J-curve phenomenon is vast by any standards. However,
the general consensus highlighted in the literature is that the response of the trade
balance to currency devaluation or depreciation does not always follow the J-curve
pattern as described in the theory. Some experiences have found that currency
devaluation or depreciation leads to improvement in trade balances; some note that it
actually worsens trade balance, while some have not been able to establish any
relationship between the two variables. The results, however, are country specific. They
largely differ due to differences in the study period, due to the wide selection of
variables in the studies and also due to the use of different methodologies to investigate
the relationship.
Many studies have found evidence of the J-curve phenomenon on trade balance after
devaluation or depreciation (such as Noland, 1989; Gupta-Kapoor and Ramakrishnan,
1999; Rehman and Afzal, 2003; Narayan, 2004; Gomes and Paz, 2005; Ono and Baak,
2014) while some failed to find evidence of the same (such as Akbostanci, 2004;
Georgopoulos, 2008; Yusoff, 2010; Bahmani-Oskooee and Gelan, 2012; Musawa,
2014). Additionally, some studies which have found mixed results (such as Lal and
Lowinger, 2002; Bahmani-Oskooee and Kutan, 2009; Kalyoncu et al., 2009; Wijeweera
and Dollery, 2013). Nonetheless, the widely adopted trade balance model in many
studies estimates trade balance as a function of real effective exchange rate, domestic
income and foreign income in the economy.
79
The survey also analyses the impact of devaluation on trade components in the context
of Fiji. Similar to international experiences, some studies on Fiji have found that
devaluation helps to stimulate exports, reduce import and improve the overall trade
performance. On the other hand, some have argued that it has an unfavourable effect on
output growth. It is noted that some of the earlier studies identified in the survey have
issues with how they modelled export and import demand equations. Other studies that
did model it properly with the inclusion of the exchange rate variable did actually find
similar income and relative price elasticities. The general consensus in these studies has
been that devaluation improves exports and reduces imports in both the short– and in the
long–run.
However, for the Fiji economy that has undergone four episodes of devaluation in the
last 28 years of its history, no studies have been undertaken to investigate empirically
the relationship of devaluation and trade balance incorporating data beyond the 2009
devaluation. This opens up the prospect of analysing this dynamics in the midst of a lot
of recent developments in the domestic and global economy, which might have had
impact on the exchange rate and trade balance relationship in an economy. Therefore,
with rapid changes in the economies, these demands for re-examination of the exchange
rate and trade balance performance for a developing Pacific island country.
Consequently, in the sections ahead, modest attempt has been made to analyse Fiji’s
trend of aggregate trade performance in the last four decades. Thereafter, empirical
investigation on the impact of devaluation on aggregate goods and services trade has
been undertaken. This is further disaggregated into goods and services sectors. A modest
attempt has also been made to contribute to the literature on the J-curve phenomenon in
the context of a Pacific island developing economy, Fiji.
80
4.3 Trade in Fiji: Pace and Patterns International trade has long been part of development and growth strategies of countries
around the world. It has been found to be one of the important pillars in enhancing
growth as well as reducing poverty in developed and developing countries (see for
example; Bahmani-Oskooee and Alse, 1993; Dutt and Ghosh, 1994; Al-Yousif, 1997;
Giles and Williams 2000; Llyod and MacLaren, 2002; Berg and Krueger, 2003;
Siliverstovs and Herzer, 2006; Hasan et al., 2006 and Cain et al., 2012). Trade earnings
are a major source of foreign exchange for larger countries and as well as for small
developing nations (Katircioglu et al., 2010).
The developing PICs’ despite being relatively small and geographically remote from
many developed countries have become an integral part of the world trade. In particular,
PICs’ with its unique geographical characteristics depends heavily on trade in services,
in particular tourism. Trade of goods, however, is largely dependent on a few sectors
primarily dominated by fisheries, sugar, logging, garments, food products and mineral
products among a small number of others. Overall, PICs’ trade structure can be
summarized as having a consistently negative trade balance; Fiji is no exception. In this
section, Fiji’s trade patterns, the evolution of her trade balance and the composition of
goods and services sector trade performance in the economy are analysed.
4.3.1 Trends in Fiji’s trade balance Fiji, like any other open economy is an active player in the world trade market using
exports as an important engine for growth and development. Given a range of economic
problems of developing countries, Fiji faces considerable problems of its own:
vulnerability to natural disasters, political instability, limited economic activities and
heavy dependence on imports are among the difficulties the economy grapples with
(Kumar and Prasad, 2002; Prasad and Narayan, 2008 and Prasad, 2010; 2012a; 2012b).
With limited resources and production facilities, Fiji depends heavily on import of
intermediate and finished goods. Overall, on the trade front, Fiji since independence has
81
been troubled with negative trade balance on goods and services. This means that on
aggregate, Fiji’s import of merchandise, insurance, transport, communication, financial
and business services exceed its exports. Figure 4.1 gives an overview of the trend of
exports and imports over the last four decades.
Figure 4.1 Fiji’s exports and imports of goods and services, 1975–2012
Notes: Export and Import of goods and services include the value of merchandise, freight, insurance, transport, travel, royalties, license fees, and other services, such as communication, construction, financial, information, business, personal, and government services.
Source: World Bank’s World Development Indicators (online), 2014.
Trade balance in Fiji has largely been negative and recently the deficit been widening.
Some notable positive balances were seen in the late 1980s and the 1990s for a short
period of time. This was when the economy experienced growth in the manufacturing
sector such as sugar and garments as a result of preferential pricing systems under some
of Fiji’s trade agreements. Imports and exports have since shared a common trend path.
However, imports have been exceeding the export flows. The trend also shows a cyclical
path whereby the downward trend lasts for more years than the upward trend.
Since the beginning of the new millennium, exports and imports in Fiji are slightly on a
divergent path and the gap between them has been widening. Despite a rising share of
the trade balance to GDP ratio, more is contributed through the rise in the total imports
component. Specifically, imports in the last decade have averaged around 65% of GDP
while exports as a share of GDP have remained at around 51% of GDP. However, in
recent times, specifically in 2008, trade deficit hit an all-time high of approximately 21%
-40
-20
0
20
40
60
80
100
120
140
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
% of
GDP
Trade (exports+imports)Exports of goods and servicesImports of goods and servicesNet Exports (exports-imports)
82
of GDP with imports reaching up to 72% of GDP while exports remained at their usual
level of around 50% of GDP.
As a result of the continuous weak performance by the export sector, the 2009
devaluation was undertaken to encourage exports and discourage imports (RBF, 2009).
On the contrary, it also needs to be noted that the devaluations in 1987 and 1998 were
undertaken when the net exports were on the rise. This implies that the devaluation in
the country is not purely driven by the deterioration of the trade balance but for other
reasons as well. The other reasons, such as the effects of political instability, drought and
GEC, as stipulated by some scholars, are discussed in Section 3.3.1.
4.3.2 Relative contribution of goods and services trade in Fiji Trade in services around the globe is gaining momentum as it does also in Fiji. Fiji has
for a long period of time troubled with an increasing trade deficit but some marginal
improvement is noted in recent years (Figure 4.2). On the other hand, services trade in
Fiji has been the biggest strength in maintaining foreign reserves and has remained on a
stable trend with surplus balance for a relatively long period of time. The service trade
surplus however, has not been large enough to offset the deficit created by the goods
sector, resulting in the overall trade deficit in the economy.
Figure 4.2 Goods and Services trade balance in Fiji, 1975–2012
Notes: Net Goods/Services represents exports of goods/services minus imports of goods/services. Source: Fiji Bureau of Statistics, Key Statistics (various years).
-1400
-1200
-1000
-800
-600
-400
-200
0
200
400
600
800
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
USD
millio
ns (c
onsta
nt 200
5)
Net Goods
Net Services
Net Goods and Services balance
83
Furthermore, given the trend in overall trade balances, it becomes essential to
understand the relative importance of the goods and services sectors in the total exports
and imports of the country. It is clearly evident that the exports in the goods sector have
been continuously declining. It has declined from a high of 60%, which had been
recorded three decades ago, to a low of 50% in 2012 (Figure 4.3). Export of services
component, on the other hand, have been on the increasing trend. As at end of 2012,
both exports of goods and services are noted to contribute equally towards the total
exports of the economy.
Figure 4.3 Evolution of goods and services export shares in Fiji, 1975–2012
Source: Fiji Bureau of Statistics, Key Statistics (various years)
In contrast, import of goods and services in total imports of the economy are depicting a
different trend from those in the exports sector. While on one hand, export of goods are
declining, import of goods are trending upwards since the 1990s (Figure 4.4). Service
imports, however, after increasing from 1990 to 2000, have stabilised to pre-devaluation
level in 2012.
This trend in exports and imports in the goods and services sectors, by and large
explains the direction of the trend in the aggregate trade balance. In a nutshell, the goods
sector exhibits rising imports and declining exports while the services sector
characterises falling imports and rising exports. In particular, the major services sectors
54
6058 56
4650
4640
42 44
5450
0
10
20
30
40
50
60
70
1975 1980 1990 2000 2010 2012
% of
Tota
l Exp
orts
Goods Service Linear (Goods ) Linear (Service)
84
(which include travel, transportation and ICT services) perform better than the goods
sectors (which include food, manufactured goods, mineral fuel, chemicals, machinery
and transport equipment)22 in the economy.
Figure 4.4 Evolution of goods and services import shares in Fiji, 1975–2012
Source: Fiji Bureau of Statistics, Key Statistics (various years) .
Within the interest of this study, then, it is important to analyse the role currency
devaluation plays to determine the pace and pattern of goods and services trade in Fiji.
In what follows in the next section, a simple ‘before-after’ approach is adopted to
analyse the impact of devaluation on trade performance in Fiji.
4.3.3 Trade response to devaluation in Fiji – a simple ‘before-after’ approach This section of the chapter analyses how Fiji’s trade performance has responded to
currency devaluations in the economy. The analysis is carried out at the aggregate level
and more specifically, by categorising separately into goods and services sectors. For
preliminary analysis, a simple ‘before-after’ approach is adopted to analyse the impact
of devaluation on the trade performance in the country. The application of this approach
has been discussed in Section 2.2.
22 Further discussion on the types of tradable goods and services and their distribution are done in Section 5.3.2.1.
80 8074 72
80 80
20 2026 28
20 20
0
10
20
30
40
50
60
70
80
90
1975 1980 1990 2000 2010 2012
% of
Tot
al Im
ports
Goods Service Linear (Goods ) Linear (Service)
85
To further elaborate on the use of the method in the context of this study, this is
described in more detail. The average for the major trade components for the years
before devaluation and after devaluation has been calculated to analyse the impact of
devaluation on the trade performance. First, investigation in the very short term (1 year),
the medium term (3 years) and in the long term (5 years) is undertaken. For instance, in
order to understand the impact of the 1998 devaluation on total exports immediately
after devaluation, the 1997 balance (before devaluation) is compared with the 1998
balance (after devaluation). Similarly, when analysing what happens to various trade
components after three– and five–year periods, average of the trade component during
the same period before and after the devaluation year is compared. Thereafter, as part of
the analysis, the percentage change during this period in the trade component is
compared with the percentage change in the real exchange rate during the same period.
The response rate is also decomposed to present it in such a way as to show how much
the trade components have changed given a 1% devaluation of the Fijian currency in the
same period. It is also to be noted that since the study sample covers data only for four
years after the 2009 devaluation and therefore, the four–year post–devaluation period
average is compared with the five–year average of the trade components before the 2009
devaluation. For the other (1987 and 1998) devaluation, this is not an issue. As discussed
earlier, the real effective exchange rate is specified in such a way that an increase
denotes currency devaluation (improvement in international competitiveness) while a
decline in the real effective exchange rate indicates currency appreciation (loss in
international competitiveness).
Hence, denoting values in the negative implies decline whereas those in positive would
be interpreted as increase in the respective trade component. The major assumption used
in this analysis is that the changes in the trade components are only as a result of
currency devaluation and thus no other factors are accounted for. At the outset, it is
important to mention that these are just crude estimates and thus care must be taken in
their interpretation. Nevertheless, the following analysis is anticipated to provide with
86
some interesting insights into how devaluation has affected trade in Fiji. A more
rigorous analysis using the econometric techniques is done in the next section.
The two devaluations in 1987 resulted in the largest gain in international
competitiveness in comparison to the devaluations in 1998 and 2009 in Fiji. The results
show that the real effective exchange rate in the long–term after the 1987 devaluation
improved its competitiveness by about 43% after five periods (Table 4.1). It is also
evident that the ultimate response of the exchange rate is more than the 33% nominal
rate of devaluation in 1987. This signifies a larger than expected gain in international
competitiveness. However, for the 1998 and 2009 devaluations, the gain in the
competitiveness is less than the nominal rate of devaluation. For the 2009 devaluation,
the real effective exchange rate changed by only around 12.8% after one-year and
achieved even smaller gains in the five–year period after devaluation.
The less than expected changes in the real exchange rate in comparison to the changes in
the nominal rate indicate that the differences are due to differences in the domestic and
foreign inflation rates. By and large, the less than expected impact on the real exchange
rate indicates that the accompanying increase in domestic inflation rates erodes the
international competitiveness. Hence, given the level of response of the real effective
exchange rate after each devaluation episode, it is reasonable to assume that the potential
gains on trade from devaluations in 1987 are expected to be higher than devaluations in
1998 and 2009.
Based on the ‘before-after’ analysis, the decomposition results show that in response to
1% currency devaluation in the country, total exports experiences improvement in both
the short– and long terms in all three episodes. More specifically, after the 1987
devaluation, total export of goods increased by 1.2% after one period and showed even
more improvement after five periods. However, its response to 1998 and 2009
devaluations has been weak. It also shows that total exports immediately after the 2009
devaluation experienced a decline by around 1.3% and thereafter improved modestly.
The domestic exports component of total exports also depicts a similar trend. It shows
87
greater response to 1987 devaluations but smaller response to the other two
devaluations.
Table 4.1 Short–, medium– and long–run response of major trade indicators to 1% devaluation in Fiji
Devaluation Years
1987 1998 2009
1st 3rd 5th 1st 3rd 5th 1st 3rd 5th
Real Effective Exchange Rate 17.6 40.4 43.0 17.6 15.8 16.5 12.8 10.6 9.0
Goods Domestic Exports 1.6 1.9 2.3 0.6 0.4 0.7 -0.7 0.0 -0.6
Goods Re-exports -0.1 -0.1 0.0 -1.5 -0.2 0.9 -2.4 2.5 7.2
Goods Total Exports 1.2 1.4 1.7 0.3 0.3 0.7 -1.3 0.8 1.6
Goods Imports -0.7 0.3 0.5 -0.2 0.7 1.2 -1.7 -0.6 0.0
Goods Trade Balance 4.0 1.2 0.9 1.2 -1.5 -2.2 2.0 1.5 1.2
Service Exports -1.3 0.0 0.5 -0.2 0.0 0.0 -0.8 0.3 0.8
Service Imports 1.4 1.2 1.5 0.6 0.8 0.5 -0.7 -1.1 -1.4
Services Trade Balance -3.8 -1.0 -0.4 -1.3 -1.2 0.8 -1.0 2.2 4.2
Export of Goods and Services -0.1 0.6 1.1 0.0 0.2 0.4 -1.0 0.5 1.2
Import of Goods and Services -0.2 0.5 0.7 0.0 0.7 1.0 -1.5 -0.7 -0.3
Goods and Services Trade
Balance 13.5 3.2 2.5 0.7 -17.7 -12.4 2.6 3.5 4.1
Source: Author’s calculations.
The import of goods, on the other hand, has experienced mixed responses. For the first
two devaluation years, total imports have shown modest increases but for the
88
devaluation in 2009, it has declined. The 1998 devaluation shows that total imports after
three periods experienced smaller increase than after five periods. In comparison to how
the total exports and imports of goods have responded; the trade balance in the goods
sector shows modest improvement after the 1987 and 2009 devaluations.
Though the goods sector trade balance has always been in negative, the trade deficit has
achieved immediate improvement after all the devaluations. In particular, it shows
improvement in all the periods, except for some decline in three– and five–year periods
after the 1998 devaluation. The favourable response in the goods sector trade balance in
almost all periods after the devaluation indicates that devaluation helps to stimulate and
improve goods sector trade performance in the economy.
Fiji’s services sector, on the other hand, shows mixed response to different devaluation
episodes in the country. The export of services shows minimal response to all of the
devaluation episodes. The analysis shows that immediately after devaluation in 1987,
1998 and 2009, the export earnings from the services sector experienced corresponding
declines. However, no impact on export earnings is noted for three periods after the
1987 devaluation and for three and five periods after the 1998 devaluation. Only a
modest improvement in service export earnings is experienced for the five periods after
1987 and for three and five periods after the 2009 devaluation.
The import in the services sector shows that it increased in all periods after the 1987 and
1998 devaluations but declined at all periods after the 2009 one. However, despite the
positive services trade balance, it experienced immediate decline in the trade balance
after all the devaluation episodes. In particular, due to the subsequent increase in service
imports, services trade balance appear to have been adversely affected at all periods after
the 1987 and 1998 devaluation. Some signs of improvements are noted after the 2009
devaluation. Hence, these preliminary results point out that services sector has not really
achieved the desired favourable outcome in the economy.
89
Moreover, combining the goods and services trade in Fiji, export of goods and services
have, in most instances, responded positively to devaluation. The import of goods and
services, on the other hand, shows mixed results. It is noted to have increased after the
1987 and 1998 devaluations while it declined after the devaluation in 2009. As a result,
the combined goods and services sector trade balance shows improvement after the
devaluation in 1987 and 2009 but shows unfavourable effects after the 1998 devaluation.
However, the effects of devaluation on the combined goods and services trade balance is
largely similar to the effects it has on the goods sector trade balance. These early results
suggest that the impact of devaluation on the overall goods and services trade balance is
largely influenced by the goods sector rather than the services sector in Fiji.
In the next section, empirical results on the impact of devaluation on aggregate trade
balance, disaggregated into goods and services sector is presented. This is followed by
the test on the validity of the J-curve phenomenon at aggregate and disaggregates level
of trade balance in Fiji.
4.4 Empirical analysis: Impact of currency devaluation on the performance of the aggregate trade balance in Fiji
In this section, empirical analysis is undertaken to analyse and understand the impact of
currency devaluation on trade balance performance in Fiji over the 1975–2012 periods.
The trade balance model and the empirical methodology to be employed in this section
are discussed in Section 2.4 and 2.5.
The trade balance model as discussed earlier on is measured as a function of real
effective exchange rate, real domestic income and real foreign income. This is expressed
as below:
),,( fr YYEfTB �
This trade balance model is used for the empirical analysis for three separate trade
balance equations as follows:
),,( frj YYEfTB �
90
where j represents trade balance as one for goods sector, second for services sector and
the third for a combined goods and services sector. All the other variables of rE , Y and fY remain the same for the three equations.
Moreover, in order to provide empirical support on the findings of the trade balance
models, each trade sector’s export and import demand equations are also modelled as
follows:
),( frk YEfEXP �
),( YEfIMP rm �
where in respective export equations, k represents domestic exports of goods, total
export of goods, total export of services and total export of goods and services.
Similarly, for respective import equations, m represents total import of goods, total
import of services and total import of goods and services.
Since annual time–series data is used for all the models, empirical analysis begins from
testing the variables’ stationarity properties followed by cointegration tests. Thereafter,
long– and short–run responses of the variables are estimated using VECM. Results on
the test of the J-curve phenomenon using the three approaches to assess the phenomenon
are also presented. All the empirical analysis is carried out in the EViews 8 software.
4.4.1 Results of the unit root tests The empirical work begins with the test for the unit root properties of the variables used
in the models. As discussed in Section 2.6.1, the widely applied ADF test is used to
conduct the unit root tests. Before the tests are carried out, all variables are transformed
in their log-linear form. Hence, the unit root test is carried out by paying appropriate
attention to the correct specification to the ADF tests and using the SIC optimal lag
selection criterion to determine the optimal lag length in each case. To conserve space,
only the major findings from the test are reported here while the detailed test results are
presented in Appendix D.
91
The major finding from the test is that the unit root hypothesis for all the variables used
in the aggregate trade analysis is not rejected at the conventional level of significance.
However, when the variables are taken in first difference form, the unit root null
hypothesis is rejected. These results suggest that all the variables employed in the model
are integrated of order one, that is they are I(1) in nature.
4.4.2 Results of the cointegration tests Since all the variables to be employed in the model are integrated of order 1, this depicts
the possibility of long–run relationship among the variables in the respective trade
equations. Hence, using the Maximum-Eigenvalue method, cointegration tests are
carried out to identify presence of a long–run relationship. The major findings from the
cointegration tests are reported here while the detailed results from the Maximum-
Eigenvalue method are presented in Appendix E.
The major finding from the test reveals that at least one cointegrating equation does exist
in every model, suggesting the presence of co-movements among the variables and
indicating long–run stationarity in the models.
4.4.3 Estimates of the long–run elasticities After finding at least one cointegrating relationship among the variables in the models,
long–run coefficients using the VECM are estimated. The long–run elasticities based on
the system equations approach for the aggregate goods and services trade balance,
disaggregated into goods and services sectors, are reported in Table 4.2.
The coefficient of the main variable of interest, rE in the goods and services sectors
indicates contrasting long–run impacts. On the goods sector, the coefficient of rE is
found to be significantly positive while on the services sector it is found to be
significantly negative. However, when these two sectors are combined to model the
aggregate goods and services trade balance performance, the impact of rE is found to be
92
significantly positive in the long–run. These empirical results suggest that currency
devaluation in Fiji leads to significant improvement in the goods sector but worsens
services sector trade performance. Nevertheless, on the aggregate front, the impact of
currency devaluation is significantly favourable in the long–run.
Table 4.2 Estimates of long–run coefficients of the trade balance models
Trade Balance rE Y fY Constant Goods Trade Balance
TBG 2.963 (0.571)***
-1.809 (0.427)***
-0.820 (0.415)*
30.187
Services Trade Balance TBS -3.571
(0.812)*** 0.171 (1.117)
1.859 (0.861)**
-25.631
Goods and Services Trade Balance TBGS 0.548
(0.162)*** -1.423
(0.417)*** 0.013 (0.114)
0.018
Notes: 1. TBG stands for goods trade balance, TBS stands for services trade balance and TBGS stands for
goods and services trade balance. 2. Standard errors are given in parentheses. 3. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
The regression results suggest that ceteris paribus, a 10% devaluation of Fiji’s currency
leads to significant improvement in the goods trade balance by about 30% while the
same causes significant worsening of the services trade balance by 36% in the long–run.
On the aggregate trade balance model, ceteris paribus, the results indicate that 10%
currency devaluation significantly improves overall goods and services trade
performance by 5% in the long–run. Further analysis on the dynamics of the goods
sector reveals that the positive response to currency devaluation stems from its
favourable impact on the export of domestic goods (Table 4.3). The export of goods is
estimated to experience significant improvements as a result of devaluation in Fiji.
The import of the goods sector, on the other hand, responds positively to devaluation
and is significant at least at the 10% level of significance. This implies that currency
devaluation results in a significant increase in total imports in the country. The result on
93
the goods import demand, however, is in contrast to the findings of Rogers (2000) and
Singh (2006). They had concluded that devaluation leads to reduction in the import bill.
Table 4.3 Estimates of long–run coefficients of the export and import models
Trade rE Y fY ConstantGoods Trade
EXPG 2.651(0.360)***
-0.759(0.205)***
17.671
DOMEXPG 4.712(0.631)***
-1.890(0.349)***
32.265
IMPG 0.289(0.156)*
1.162(0.130)***
-4.438
Services TradeEXPS 0.456
(0.262)*0.793
(0.152)***-5.676
IMPS 3.850(0.762)***
-1.484(0.676)**
17.189
Goods and Services TradeEXPGS 1.332
(0.244)***0.133(0.140)
5.203
IMPGS 0.372(0.133)***
1.233(0.115)***
-5.557
Notes:1. EXPG stands for export of total goods, DOMEXPG stands for export of domestic goods, IMPG
stands for import of goods, EXPS stands for export of services, IMPS stands for import of services, EXPGS stands for export of goods and services and IMPGS stands for import of goods and services.
2. Standard errors are given in parentheses.3. (*), (**) and (***) denote significance at the 10%, 5% and 1% level, respectively.
Conversely, the difference in the results is argued to be due to two main reasons. One
being the inclusion of the recent dataset and secondly due to the fact that Fiji relies
heavily on import of goods such as food, machinery, transport equipment and other
imported intermediate goods. The devaluation of the Fijian dollar coupled with external
shocks such as the soaring global oil and commodity prices has put increased pressure
on the economy’s import bill. It is argued that in spite of the currency devaluation, Fiji
needs considerable amounts of imports to be used either as intermediate goods or as
final products. This has in turn led to a significant increase in goods imports.
94
However, these empirical results are found to be consistent with the earlier analysis
using the ‘before-after’ approach. It is argued that the positive and favourable impact of
devaluation on the goods sector results from the increased response of exportable
products and import substitutability in the economy. Government policies aimed at
improving domestic exportable capacities and import substitution policies such as tax
free zones, tax holidays and ‘Buy Fijian Made’23 campaigns, among other initiatives,
appear to be boosting domestic capacity to export.
Moreover, this study takes a first attempt to contribute to the literature on estimating
services trade equations for Fiji. The literature on the services sector is weak as there is
no evidence of empirically modelling service sector equations in the context of Fiji.
Hence, the results emanating from the analysis provides empirical grounding for some of
the arguments made in Yang et al. (2013) and Gounder and Prasad (2013). Besides other
conclusions, these studies claim that the 2009 devaluation in Fiji seems to have
benefited the service industry and specifically the tourism sector. This has been found to
be the case in this study as well using the ‘before-after’ approach for the 2009
devaluation.
However, after the 1987 and 1998 devaluations, services trade balance experienced
deterioration in the short to medium term. Therefore, it is argued that although the 2009
devaluation shows improvements, its long–run combined impact over the entire analysis
period outweighs the positive impact. It is argued that though the absolute tourism
numbers in Fiji have generally increased in recent years, the tourism related activities
such as food, tourist luxury products and recreational activities are also affected. Gibson
(as cited in Chen et al., 2014) estimates that approximately 70% of food products used in
the tourism industry in the Pacific are imported. Fiji provides no exception to this
pattern.
23 ‘Buy-Fijian-Made’ campaign is a recently launched campaign by the Ministry of Industry and Trade to promote Fijian made products and to create a Fijian brand to promote the sale of Fijian made products in both domestic and international markets.
95
The author also argues that for every one dollar of tourism expenditure in Fiji, only 44%
is retained in the economy. The rising cost of importable inputs has also affected
services receipts. This ultimately worsens the long–run services sector performance.
Additionally, Fiji holds little comparative advantage over other neighbouring tourism
oriented PICs24 as well. This is because the island countries boast similar climatic
conditions and compete for tourist numbers, mostly from the same tourism source
countries. The rising fuel costs coupled with devaluation has also increased costs for
transportation services. The combination of all these factors resulted in an overall
negative impact of devaluation on services trade performance.
The services sector imports such as travelling overseas for education or medical reasons,
among others appear not to have declined significantly after devaluation in the economy.
This resulted in an increase in import bills and subsequent adverse impact on services
trade balance performance. The findings from this study are also supported by the recent
IMF study by Culiuc (2014) who studies the determinants of tourism for a number of
countries. The study results suggest that tourism, which makes up a significant portion
of services trade in small island countries including Fiji, is less sensitive to changes in
the country’s real exchange rate. In fact, the effect is estimated to be approximately
close to zero. The study argues that this is due to higher import denominated food for
tourists than those which are locally produced food. The study also argues that because
most tourists to the island countries depend on packaged vacations, they often do not
directly benefit from the movements in the real exchange rates. This is because holiday
package prices are set in foreign currency by tour operators, mostly at pre-devaluation
prices.
Moreover, the impact of devaluation on the aggregate goods and services trade
performance is found to be significant and favourable in the long–run. It is argued that in
the long–run, the positive impact of devaluation on goods sector outweighs the adverse
impact on the services sector trade balance. This result is in the same direction to
24 The Pacific island countries of Palau, PNG, Samoa, Tonga and Vanuatu are also heavily tourism oriented economies.
96
Narayan and Narayan (2004a) who also found a positive but insignificant impact of
Fiji’s devaluation on its trade performance. It is argued that since this study includes
data till 2012, the 2009 devaluation appears to be favourably influencing many trade
sectors. This is ascertained to cause the earlier insignificant impact into a significantly
positive one. The regression results also show that the elasticity of the export of goods
and services is larger than the import of goods and services. This leads to the resultant
favourable impact on the aggregate trade balance. Therefore, devaluation as a policy tool
to boost overall trade performance in Fiji has been effective.
Furthermore, the impact of domestic income on the trade performance is found to be
significantly negative for goods and aggregate trade balance performance (Table 4.2).
However, it is insignificant for the services sector. The coefficient estimates suggest that
ceteris paribus a rise in domestic income by 10% leads to a significant increase in total
import demand for goods and services by 14% in the long–run. This ultimately translates
to an adverse impact on the aggregate trade performance. This result is also consistent
with the results of Narayan and Narayan (2004a) and Singh (2006) in the context of Fiji.
This result however, for the import demand for Fiji is not surprising. Increase in
economic activity means increase in production. Fiji being a highly import dependent
economy increases its demand for capital and intermediate goods both of which are
largely imported when production rises. Rising income levels would also mean increase
in consumption demand for importable commodities, thereby increasing the goods trade
deficit. On the other hand, the impact of domestic income on services trade performance
is insignificant. This suggests that rising income levels in the economy do not
necessarily mean that locals opt for services imports or for tourism activities outside of
the country. However, given the fact that import of goods makes up around 80% of total
imports, it is argued that the subsequent positive impact on the goods sector outweighs
its impact on the services sector. This consequently results in the positive impact on the
aggregate goods and services imports.
97
The impact of trading partners’ income, on the other hand, has a positive but
insignificant impact on the overall goods and services trade balance for Fiji in the long–
run (Table 4.2). The corresponding export of goods and services also reveal similar
insignificant relationship. However, the same variable is found to be significantly
positive for the services sector but is significantly negative for the goods sector in the
long–run (Table 4.2).
On the goods sector, ceteris paribus the results imply that a 10% rise in income for
trading partner countries leads to 19% and 8% decline in Fiji’s domestic exports and
total goods exports, respectively (Table 4.3). It is argued that the negative and
significant impact of foreign income on Fiji’s goods trade performance is due to the
nature of Fiji’s goods exports. The major goods exportable to these countries include
agricultural products which more than often have high degree of substitutability with
rising income in these countries. Thus, it is estimated that rising income in these trading
partner economies results in a reduction of export of agricultural products from Fiji.
The services sector, on the other hand, is significantly and positively related to trading
partner income. The results, suggest that rising income in major tourist source countries
results in an increase in flow of visitors to Fiji for tourism purposes. The coefficient
estimates, ceteris paribus suggests that a 10% increase in income in Fiji’s major trading
partner countries leads to significant improvement in export of Fiji’s services by 8% in
the long–run.
The different impact of trading partner income on the goods and services sectors results
in no significant impact on the overall trade performance in Fiji. Nonetheless, the
insignificant yet positive impact of trading partner income on export of goods and
services and the overall trade performance is supported by Asafu-Adjaye (1999),
Narayan and Narayan (2004a), Narayan and Narayan (2004b), Singh (2006) and Kumar
(2009) in the context of Fiji.
98
Therefore, based on the empirical evidence it is established that currency devaluation is
an effective tool to boost overall trade performance in Fiji. However, the impact of
devaluation on the goods sector is found to be significantly favourable but has adverse
impact on the services sector in the long–run. However, the long–run results in this
section do not shed any light on the short–run dynamics of the variables in the trade
balance models. Hence, the discussion of how the goods and services trade components
respond to currency devaluation in the short–run is the focus of the next section.
4.4.4 Estimates of the short–run elasticities: Results from error correction models The results in the previous section do not shed any light on the short–run dynamics of
the impact of currency devaluation on exports, imports and the resultant trade balance
models. As such, the short–run coefficients of the real exchange rate variable on the
trade components and the tests on the J-curve phenomenon are summarized and
presented in this section. At the outset, it is important to mention that all the short–run
error-correction models are statistically well behaved.
The error-correction term 1�tECT , which measures the speed of adjustment to restore
equilibrium in the dynamic model, has a negative sign, is statistically significant and
ranges between 0 and -1 in all the cases. This ensures that the series are non-explosive
and that the long–run equilibrium is attainable. The coefficient of 1�tECT in the goods
trade balance model is estimated at -0.53, in the service trade balance model at -0.15,
while for the aggregate goods and services trade balance model, it is estimated at -0.72.
Hence, this ensures that following a shock, convergence to equilibrium is relatively
swift.
The short–run regression results suggest that the impact of currency devaluation in Fiji is
again different for the two sectors, as evidenced earlier for the long–run results. The real
effective exchange rate variable on the goods sector, shows that in the short–run, the
impact is negative from lags 1 to 3 followed by positive impact at lag 4 (Table 4.4). This
99
implies that currency devaluation in Fiji causes short–run worsening of the goods sector
followed by improvements in the medium-term. These short–run results are also
supported by the short–run estimates of the exchange rates from the goods export and
import models (Table 4.5).
Table 4.4 Short–run coefficient estimates of real exchange rates in the trade balance models
Short–run results Goods sector
Services sector
Goods and Services sector
rtE 1ln �� -0.953
(0.940) 0.410 (0.512)
-0.050 (0.253)
rtE 2ln �� -2.392
(1.193)* 0.447 (0.454)
-0.165 (0.231)
rtE 3ln �� -1.209
(0.892)
rtE 4ln �� 0.088
(0.828)
COUP 0.117 (0.077)
-0.284 (0.074)***
-0.046 (0.038)
Diagnostics 1�tECT -0.527
(0.300)* -0.150 (0.081)*
-0.723 (0.360)*
2R 0.842 0.557 0.754
Adjusted 2R 0.640 0.373 0.648
! 0.135 0.107 0.074
NX 2
0.559 [0.756]
0.609 [0.737]
0.615 [0.735]
HetX 2
26.650 [0.483]
20.647 [0.357]
14.370 [0.762]
LM )(SCTest 13.246 [0.655]
23.392 [0.104]
20.282 [0.208]
AR roots graph stable stable stable Notes: 1. Standard errors are given in parentheses while p values are in square brackets. 2. (*), (**) and (***) denote significance at the 10%, 5% and 1% level, respectively. 3. ECTt-1 represents the error correction terms;! is the standard error of equation; diagnostics are
Jarque-Bera statistics for normality (X2N) and chi-squared for heteroskedasticity tests (X2Het), and LM Test statistics for serial correlation (SC) while AR roots graph tests for model stability.
With regard to the services sector, the regression results indicate that currency
devaluation, though insignificant, improves services trade performance at least in the
100
short–run (Table 4.4). This result, however, is in contrast to the long–run results.
Nonetheless, this is not surprising in the short–run. This is because currency devaluation
immediately does give an incentive to tourists to visit Fiji when the Fijian currency
becomes weak. It also creates a disincentive for locals to visit foreign countries for either
leisure or business purposes as the cost of travel immediately rises.
Table 4.5 Short–run coefficient estimates of real exchange rates in the export and import models
Short–run results
Goods export
Domestic Goods export
Service exports
Goods and Services export
Goods import
Service imports
Goods and
Services import
rtE 1ln �� -0.110
(0.585) 0.419 (0.534)
0.438 (0.278)
0.006 (0.295)
0.022 (0.461)
0.015 (0.319)
0.033 (0.399)
rtE 2ln �� -0.823
(0.724) -0.129
(0.332) -0.224 (0.384)
0.484 (0.329)
rtE 3ln �� -0.545
(0.577) 0.683
(0.328)** 0.098 (0.324)
COUP -0.056 (0.058)
0.022 (0.054)
-0.082 (0.040)**
-0.055 (0.033)
0.005 (0.044)
-0.036 (0.044)
-0.084 (0.041)*
Diagnostics 1�tECT -0.620
(0.212)*** -0.144
(0.066)** -0.939
(0.200)*** -0.818
(0.200)*** -0.632 (0.256)**
-0.104 (0.048)**
-0.890 (0.310)***
2R 0.577 0.268 0.682 0.681 0.314 0.240 0.561
Adjusted 2R 0.366 0.145 0.522 0.521 0.200 0.113 0.426 ! 0.116 0.144 0.081 0.070 0.114 0.091 0.081
NX 2
0.090 [0.956]
0.423 [0.809]
1.158 [0.561]
2.007 [0.367]
0.355 [0.837]
2.020 [0.364]
0.913 [0.633]
HetX 2
27.946 [0.142]
13.792 [0.130]
17.150 [0.702]
18.139 [0.640]
6.692 [0.669]
9.672 [0.378]
8.673 [0.894]
LM)(SCTest
9.102 [0.428]
10.350 [0.323]
16.079 [0.065]
9.583 [0.385]
6.827 [0.655]
2.242 [0.987]
4.275 [0.892]
AR rootsgraph
stable stable stable stable stable stable stable
Notes: 1. Standard errors are given in parentheses while p values are in square brackets. 2. (*), (**) and (***) denote significance at the 10%, 5% and 1% level, respectively. 3. ECTt-1 represents the error correction terms,! is the standard error of equation; diagnostics are
Jarque-Bera statistics for normality (X2N) and chi-squared for heteroskedasticity tests ((X2Het), and LM Test statistics for serial correlation (SC) while AR roots graph tests for model stability.
101
However, this is not expected to be the case for a very long period of time. Important
business travel and travel by individuals for education and health reasons are expected to
continue much as usual despite weakening of the domestic dollar. The estimates from
the services import model also support this idea (Table 4.5). The argument that tourists
get an incentive to visit Fiji when the Fijian dollar is devalued is supported by the short–
run results from the services export model. The results show favourable impact at lag 1
as well as significant and positive impact at lag 3 (Table 4.5). Service imports, on the
other hand, experiences modest increase but is insignificant.
Moreover, the short–run analysis at the aggregated goods and services sector reveals that
the impact of devaluation is negative but insignificant at all lags (Table 4.4). Fiji being a
highly import dependent economy continues to import as it has been doing prior to
devaluation despite the changes in the relative prices after devaluation. The lag effect
sets in for the case of Fiji. The exporters and importers take time to realise that
devaluation has taken place and that exports have become cheaper while imports have
become expensive. The consequent short–run worsening of the trade balance is followed
by improvements in the medium– to long–term.
This result also suggests that the goods sector, which dominates Fiji’s overall trade
balance, is largely influencing the ultimate impact on the aggregate goods and services
trade performance. The results on the disaggregated goods and services sectors suggest
unfavourable impact on the aggregate goods and services sector in the short–run. This is
similar to that obtained in the goods sector in the short–run. This argument is also
empirically supported by the short–run coefficients of the real exchange rate in the
goods and services export and import equations.
The model also incorporates a dummy variable COUP to capture the impact of political
instability on Fiji’s trade performance. The results show that political instability has a
significant and negative impact at least on the services sector in Fiji. It is ascertained that
the series of political instability creates loss of investment confidence and tends to affect
visitor arrivals in the country. As a result, it is likely to reduce investment levels in the
102
services oriented export sectors, mainly tourism. However, its impact on the goods
sector is estimated to be insignificant.
A number of diagnostic tests have also been applied to the models in the study to ensure
model appropriateness and stability. Based on the tests, all the models are noted to have
passed the diagnostic tests including tests of autocorrelation, normality and
heteroskedasticity along with tests for model stability. For all the models, the LM
diagnostic tests do not reveal any evidence of autocorrelation. The models also pass the
Jarque-Bera normality test, suggesting that the errors are normally distributed. The
models also do not show any evidence of the problem of heteroskedasticity. The AR
Roots graph shows that the lags chosen for the models are stable as the points lie within
the unit roots circle. The models also have a reasonably strong adjusted R-squared for
the goods trade balance model at 64%; for the service trade balance model, at 37% and
for the aggregate goods and services trade balance, at 65%. The other corresponding
exports and imports models are also well specified with reasonable goodness of fit.
4.4.5 Testing for the aggregate J-curve phenomenon
In this section, the three widely used methods: the traditional definition, new definition
and IRF analysis are used to assess the presence of the J-curve phenomenon for goods,
services and aggregate goods and services sector trade balance in Fiji.
1. Goods trade balance
Using the coefficient results under the traditional definition, the real exchange rate
variable is estimated to have unfavourable impact at shorter lags followed by favourable
response on the goods trade balance at longer lags (Table 4.3). This short–run
deterioration followed by improvement at longer lags, confirms the presence of the J-
curve phenomenon using the traditional definition.
Using the new definition, there is evidence of adverse impact of currency devaluation in
the short–run accompanied by positive and significant impact of currency devaluation on
103
the goods trade balance in the long–run (Tables 4.1 and 4.3). This also confirms the
presence of the J-curve phenomenon using the new definition.
The third method of assessing the J-curve phenomenon is using the IRF analysis. For
this, the generalized IRF is used to show the response of the goods trade balance to
generalized one standard deviation shock in the real exchange rate. The graphical
representation of the shock in the real exchange rate reveals that the goods trade balance
declines for the first 4 years followed by slight improvement till year 6 (Figure 4.5).
Thereafter, the goods trade balance continues to decline, with some improvement noted
after year 9. Hence, the graphical response pattern of the goods trade balance does not
depict the J-curve pattern. Therefore, there is no evidence of the J-curve phenomenon
using the IRF analysis. However, in consistent with the other two methods, the presence
of the J-curve phenomenon is confirmed for the goods sector in Fiji.
Figure 4.5 Response of the goods trade balance to generalised one standard deviation innovation in the real exchange rate
-.01
.00
.01
.02
.03
.04
.05
.06
1 2 3 4 5 6 7 8 9 10
Based on the IRF results, it is argued that the decline of the goods trade balance after
two periods indicates the continuing dependence on importable commodities in the
economy. As argued earlier, it is ascertained that the favourable impact of devaluation
on exportable commodities results in an increase in domestic income. This subsequently
results in an increase in imports for an import dependent economy like, Fiji. Even the
gradual appreciation of the domestic currency favours imports more than exports.
Consequently, this results in the deterioration of the trade balance in later periods.
Periods
104
However, currency devaluation after a lapse of ten years in the country appears to
correct for this and cause an improvement in the goods trade balance once again in the
long–run.
2. Services trade balance
The same procedures as undertaken for the goods trade balance are used to investigate
the presence of the J-curve phenomenon for services sector in Fiji. Using the traditional
definition, the coefficients of the real exchange rate in the services trade balance model
are estimated to be positive at all lags (Table 4.3). Hence, the evidence of short–run
improvement at all lags fails to show the presence of the J-curve phenomenon using the
traditional definition for the services sector.
Similarly, using the new definition, it is noted that the short–run coefficients of the real
exchange rate are positive, followed by negative and significant coefficients of the real
exchange rate in the long–run. This implies that the services sector experiences
immediate short–run improvements followed by deterioration in the long–run. This as a
result implies evidence of an inverse J-curve phenomenon for the services sector.
Figure 4.6 Response of the services trade balance to generalised one standard deviation innovation in the real exchange rate
-.09
-.08
-.07
-.06
-.05
-.04
-.03
1 2 3 4 5 6 7 8 9 10
The IRF analysis reveals that one standard deviation shock in the real exchange rate
results in an unchanged services trade balance for the first 2 years followed by slight
Periods
105
improvement in year 3 (Figure 4.6). Thereafter, the trade balance in the services sector
declines in the fourth and fifth periods. The results show modest improvement,
thereafter. Consistent with the new definition, the time trend path of services trade
balance shows evidence of an inverse J-curve phenomenon. Hence, there is no strict
presence of the J-curve phenomenon for services sector trade in Fiji.
3. Goods and Services trade balance
On the aggregate goods and services sector trade, the results using the traditional
definition show evidence of unfavourable impact on the trade balance at all lags. Hence,
presence of the J-curve phenomenon is rejected using the traditional definition.
However, in assessing the J-curve phenomenon using the new definition, short–run
deterioration is accompanied by long–run significant improvements in the trade balance.
This confirms the presence of the J-curve phenomenon for the aggregate goods and
services trade balance using the new definition.
Additionally, using the IRF analysis, the graphical pattern of the shock in the real
exchange rate reveals that the goods and services trade balance declines for the first 3
periods followed by fluctuations in later periods (Figure 4.7). Hence, the IRF analysis
does not show evidence of the J-curve phenomenon. The path it follows is somewhat
‘W’ shaped.
Figure 4.7 Response of the goods and services trade balance to generalised one standard deviation innovation in the real exchange rate
.008
.012
.016
.020
.024
.028
1 2 3 4 5 6 7 8 9 10
Periods
106
It is argued that a large decline followed by slightest increase in the trade balance using
the IRF analysis confirms that there is no strictly followed J-curve phenomenon in Fiji.
This is largely because over time, Fiji’s exchange rate does not stay devalued but starts
to appreciate soon after few periods from the devaluation year. Hence, this creates
doubts on the strictly followed J-curve phenomenon in Fiji. The results also create doubt
on the findings of Narayan and Narayan (2004a) as empirical evidence suggests that the
J-curve phenomenon is not strictly valid under all the methods of assessment.
The findings of this study are comprehensive as the phenomenon is tested using the new
definition, which had not previously been applied in the context of Fiji. The use of only
one of the available methods for assessing the effectiveness of devaluation appears to
cause biasness in the results. Hence, the use of all the three methods in one study ensures
robustness of the result as one method is not superior to the others. However, on the
basis of the three methods, the study does not find uniform evidence of the J-curve
phenomenon.
The use of the IRF analysis alone produces no evidence of the J-curve phenomenon in
any of the cases. The goods sector gains support using the traditional and new definition
while the aggregate goods and services trade balance shows the J-curve phenomenon
using the new definition only. Hence, for the goods sector it is supported by two
methods and for a combined goods and services sector, it is supported by only one
method. The services trade sector, on the other hand, does not reveal any evidence of the
J-curve phenomenon.
However, evidence of the inverse J-curve phenomenon in the services sector is visible.
More specifically, the results suggest that the support of the phenomenon in the overall
trade balance is largely contributed by the goods sector instead of the services sector in
Fiji. Nevertheless, the failure of the three methods to arrive at a consensus creates doubt
on the strong evidence of the J-curve phenomenon in Fiji. A summary of the results on
the tests using the three methods is provided in Table 4.6.
107
Table 4.6 Summary of the assessment on the J-curve phenomenon in Fiji on the aggregate trade performance
Trade Balances The J-curve phenomenon using: Traditional definition
New definition
Impulse Response Function Analysis
Goods Yes Yes No
Services No No No
Goods and Services No Yes No
4.4.6 Highlights of the empirical analysis at aggregate level trade balances In this section, the results of the empirical analysis to investigate the relationship of the
exchange rate, domestic income and foreign income to aggregate trade balances in Fiji is
presented. In particular, three separate trade equations, one for goods, one for services
and the third for an aggregate goods and services trade balance are modelled.
Additionally, the respective export and import demand models are also modelled to
ensure robustness of the results. This is also used to identify the sources of the impact of
each variable on the respective trade balance models.
The empirical evidence finds support for a favourable and significant impact of
devaluation on the aggregate goods and services trade balance. The same favourable
impact is also estimated for the disaggregated goods sector. However, the impact of
currency devaluation on the services sector is estimated to be significantly unfavourable.
Hence, it is argued that the favourable impact on the overall trade balance in Fiji is being
derived from the goods sector and not necessarily from the services sector. These results
are also supportive of the earlier findings from the ‘before-after’ analysis on the impact
of devaluation in Fiji.
Besides the exchange rate variable, the empirical evidence on the domestic income
suggests significant domestic income elasticity. The elasticity in the goods sector and the
aggregate goods and service sector, ranges from -1.4 to -1.8. Accordingly, it implies that
a rise in income for Fiji induces an increase in import demand, which as a result tends to
108
significantly worsen the trade balance. The empirical estimates for the trading partner
income are also found to be significantly positive for goods sector while it is
significantly negative for the services sector performance. However, when it is modelled
together, the impact appears to be insignificant.
Following the estimates of the trade balance model, the tests to validate the presence of
the J-curve phenomenon in Fiji are carried out. The findings are that there is evidence of
the J-curve phenomenon at the goods sector and at the aggregate goods and services
sector trade. However, the services sector does not reveal any evidence of the J-curve
phenomenon. But, the failure of the three methods to come to accord creates uncertainty
on the strong evidence of the J-curve phenomenon in Fiji.
4.5 Concluding comments
This chapter has analysed in detail the various methodologies employed by many
scholars around the globe to investigate the relationship of exchange rate with trade
performance at aggregate level. The major focus of the literature review has been on the
studies that have largely concentrated on investigating the response of the aggregate
trade flows to currency devaluation or depreciation using econometric techniques.
Included in the review are also studies that have attempted to validate the J-curve
phenomenon and the relevant studies in the context of Fiji. Hence, the empirical analysis
to estimate the long– and short–run effects of devaluation as well as validating the
presence of the J-curve phenomenon on the aggregate trade balance are among the
important contributions of this chapter.
Three separate trade equations: one for goods, one for services and one for an aggregate
goods and services trade balance are modelled as a function of real exchange rate,
domestic income and trading partner income. Along with the trade equations, respective
import and export demand equations are modelled to identify the sources of the impact
on the overall trade performance. The empirical results find support for a favourable and
significant impact of devaluation on the goods sector and the aggregate goods and
109
services sector trade balance in the long–run for Fiji. However, the services sector is
found to experience a significant adverse impact.
Empirical analysis on the test of the J-curve phenomenon finds support for the J-curve
phenomenon for the overall goods and services sector, which is largely contributed by
the goods sector instead of the services sector in Fiji. The services sector, in fact,
exhibits an inverse J-curve relationship. However, it is unable to negate the favourable
effect on the goods sector. This is because the goods sector exhibits characteristics of
high import substitution and capability to increase local production in order to meet
increased demand. The results also suggest that the high import content and the sale of
tourism packages at pre-devaluation prices in terms of foreign currency by tourism
agents negate the expected gains from devaluation on the services sector. Nonetheless,
currency devaluation on the aggregate front has been having favourable effects in Fiji
over the last three decades.
Hence, this particular chapter has provided empirical understanding into the relationship
of exchange rate with trade performance in Fiji. However, even after having analysed
the impact at aggregate trade balance level, the present analysis does not reveal anything
on which particular sectors in Fiji are benefiting after devaluation and which are not. In
a similar manner, it cannot as at yet be clearly outlined on how Fiji’s trade performance
responds with its trading partner countries after devaluation. It is ascertained that a
detailed analysis on the impact of devaluation on trade would allow better understanding
and well prescribed policy interventions.
This chapter provides a modest background for further analysis into the impact of
devaluation on sectoral and bilateral level trade analysis. The proposed further analysis
using a similar trade balance model as used in this chapter shall be the core component
of the next chapter.
110
CHAPTER 5
SECTORAL AND BILATERAL LEVEL TRADE FLOWS
IN RESPONSE TO CURRENCY DEVALUATION
5.1 Introduction
The trade literature on the effectiveness of currency devaluation along with the on the J-
curve phenomenon has evolved over time from aggregate to bilateral trade analysis.
Recently, it also includes studies at sectoral level trade performance. The aim was to
gain greater insights and policy prescriptions on the role of currency devaluation in
influencing trade performance in an economy. In Chapter 4, review of the empirical
literature on the aggregate goods and services trade performance are already presented.
However, results discussed and obtained in Chapter 4 might suffer from the aggregation
problem.
Literature points out that a significant trade flows of a devaluing country with its one
partner could be offset by an insignificant trade relation that might exist with another
trading partner country. This causes the problem of aggregation biasness (Bahmani-
Oskooee and Brooks (1999) and Bahmani-Oskooee and Xu (2012)). Moreover, a
significant trade flow of one or few major commodities could also be offset by
insignificant trade relation that might exist for other small sectoral trade with the rest of
the world.
Therefore, it is argued that the difference in the impact of devaluation on various sectors
and trading partners of the devaluing country could actually lead to biased results at the
aggregate level. In fact, this would mean that the results obtained at the aggregate level
would not necessarily be the same for different commodities and trade partners of a
country. This is because different sectors of an economy have different degrees of
openness and accordingly may react differently to the exchange rate shocks.
111
A few recent studies have also attempted to locate the effectiveness of currency
devaluation for different sectors along with various trade partners by using the sectoral
and bilateral trade data. These studies also test the J-curve phenomenon and claimed that
such arrangements will remove aggregation bias problem (Soleymani at al. (2011),
Pattichis (2012), Bahmani-Oskooee and Zhang (2013) and Bahmani-Oskooee et.al
(2013)).
Following the literature, this chapter aims to address the aggregation bias problem.
Consequently, this chapter attempts to understand and empirically analyse the response
of sectoral and bilateral trade flows to currency devaluation in Fiji. This is undertaken by
a modest attempt to review some of the existing J-curve literature on the sectoral trade
flows followed by the same on the bilateral trade performance in developed and
developing countries. A few existing relevant literature in the context of Fiji are also
reviewed.
5.2 Literature review: Sectoral and Bilateral trade flows This section reviews the relevant literature studying the relationship of exchange rate
with sectoral and bilateral trade performance in an economy. To ensure consistency with
the earlier review and the analysis on the aggregate trade performance, the focus in this
review is on the studies that have used recent advanced econometric techniques to
establish the relationship. In particular, included in these review are those studies that
have attempted to validate the J-curve phenomenon at the sectoral and bilateral level.
Towards the end of this section, relevant literature in the context of Fiji’s economy is
also reviewed.
5.2.1 On the sectoral trade flows and the J-curve phenomenon The literature review in the last chapter has looked into studies that attempt to establish
the relationship between exchange rate and aggregate trade performance. The test on the
112
J-curve phenomenon was also conducted at aggregate levels of goods and services trade
balance. The overall trade balance which is simply an aggregation of sectoral trade
balances tends to be influenced by the responses of various sectors to changes in
exchange rates in the economy. This distinction has been well recognised and argued in
the literature (Yazici and Klasra (2010), Pattichis (2012) and Bahmani-Oskooee and
Zhang (2014)). It is established that because different sectors respond differently to
different shocks in the economy, it has different levels of impact at the aggregate level.
Using Bahmani-Oskooee (1985) model, Yazici (2006) investigates the J-curve
hypothesis for the Turkish agricultural sector. The study uses quarterly data over the
1986Q1–1998Q3 period to examine the response of the agricultural sector to changes in
the exchange rate. Using the Almon lag structure on the exchange rate variable, the
study finds that the trade balance immediately improves with currency devaluation in the
economy. However, the trade balance shows signs of deterioration and improvements in
the later periods. Hence, using the traditional definition to assess the J-curve
phenomenon, the author is convinced that the J-curve phenomenon does not exist in the
Turkish agricultural sector. He adds that devaluation actually worsens the agricultural
sector’s trade balance in the long–run.
Bahmani-Oskooee and Wang (2007) disaggregate the Australian trade to the United
States into 108 industries to estimate the trade balance relationship for each sector.
Utilising the annual data over the 1962–2003 period, they adopt the model used by
Bahmani-Oskooee and others (2005). The study uses bounds testing approach to
cointegration and error correction modeling technique to establish industry level J-curve
phenomenon. Their version of the sectoral trade balance model takes the following
specification:
tttUStati REXYYTB �"��� ����� loglogloglog ,,,
where TB is defined as the ratio of Australia’s nominal exports of commodity i to the
US over her imports of commodity i from the US; aY is the Australian real GDP, USY is
113
the real GDP of the US while REX is the bilateral real exchange rate between the
Australian dollar and the US dollar.
The study finds that there are short–run significant effects of real depreciation in 64 of
the 108 industries being modelled in the study. Out of these 64 industries, only 44 of
them have significant impact in the long–run as well. Therefore, using the new
definition, the study provides support for the J-curve phenomenon in these 44 industries.
The scholars also add that not all the sectors respond in a similar pattern to currency
depreciation in the trade between the two countries.
Bahmani-Oskooee and Bolhasani (2008) use export and import data for 152 sectoral
trades between Canada and the USA over the 1962–2004 period. In order to assess the
impact of real depreciation of the Canadian dollar on the sectoral trade performance, the
study uses the bounds testing approach to cointegration and error-correction modelling
technique to estimate their model. Using the similar approach to Bahmani-Oskooee and
Wang (2007), the study finds that real depreciation has short–run effects on the trade
performance in two-thirds of the industries. However, out of these industries, only 50%
have their short–run effects lasting into long–run favourable gains. Hence, using the new
definition, the J-curve phenomenon is valid only for 85 industries.
Yazici and Klasra (2010) examine the J-curve phenomenon on Turkey’s manufacturing
and mining sector over the 1986Q1–1998Q3 periods. The study utilises the model
developed by Bahmani-Oskooee (1985) and uses the VECM technique and Almon lag
structure to establish the relationship. Hence, using the traditional definition and IRF
analysis, the study estimates that currency depreciation leads to initial improvement,
then deterioration followed by improvement in the trade balance for both the sectors. As
a result, this response pattern implies that the J-curve phenomenon does not exist in
either of the sectors in Turkey.
The study notes that the response of the two sectors after the initial improvement
appears to be different in the two sectors. It is observed that the violation of the J-curve
114
phenomenon persists more in the manufacturing sector than the mining sector. The study
argues that this is because the manufacturing sector has higher import content in its
exports than the mining sector. Nonetheless, the study finds that the M-L condition holds
true and hence, provides support for the favourable response of the two sectors to
currency devaluation in the long–run.
Yazici (2010) using similar methodological approach to Bahmani-Oskooee and Wang
(2007), studies the response of currency depreciation on services sector trade
performance in Turkey. The study finds that the services sector experiences initial
improvement, then worsening followed by improvement of trade balance in the short
run. Hence, using the traditional definition, the study does not find evidence of the J-
curve phenomenon in the service sector for Turkey.
Bahmani-Oskooee and Hegerty (2011) use annual trade data for 102 industries to model
sectoral trade balance between Mexico and USA over the 1962–2004 periods. Using the
model specification of Bahmani-Oskooee and Wang (2007), they use the bounds testing
approach to cointegration and error-correction modelling technique to examine the
relationship. The study finds that 59 industries depict significant relationship between
exchange rate and trade balance. However, a favourable long–run effect of the
depreciation of the peso is noted in only 24 out of the 102 industries in Mexico. Hence,
using the traditional and new definition of assessing the J-curve phenomenon, the
analysis shows presence of the J-curve phenomenon in only 4 out of the 102 industries.
The study also highlights that there is actually more evidence of the inverse J-curve
phenomenon which existed in 13 industries.
Additionally, Soleymani and others (2011) using traditional and new definitions finds
evidence of the J-curve phenomenon in 10 of the 53 sectoral trade of Malaysia with
China over the 1993Q1–2009Q4 period. Another study by Soleymani and Saboori
(2012) finds evidence of the J-curve phenomenon in 22 out of the 67 industrial trades
between Malaysia and Japan using the traditional and new definitions to assess the
phenomenon.
115
Bahmani-Oskooee and Xu (2012) apply the bounds testing approach to cointegration to
study the impact of exchange rate changes on 108 sectoral trades between the USA and
Hong Kong. Studying over the 1978–2006 period, they find long–run significant
relationship between real exchange rate and trade balance in at least 55 industries.
However, there are only 49 industries out of the 55 industries which indicate the J-curve
phenomenon using the new definition. Furthermore, the study argues that using
disaggregated trade data provides better understanding on the relationship between
exchange rate and trade balance between countries. They add that even when the total
trade balance does not react to changes in the exchange rates, there are certain industries
whose trade balance is sensitive and responds accordingly. Hence, they argue that the
use of aggregate data produces significant bias in the results.
Pattichis (2012) disaggregates services trade data into travel, passenger fares and other
transportation services to investigate the effects of currency depreciation between the
USA and the UK. Using data over the 1986Q1–2009Q4 periods, the study finds
evidence of a significant effect of exchange rate on all categories of services trade. In
particular, trade of travel services shows significant and negative impact in the short run
while in the long–run, it is found to cause significant improvements. With regard to
passenger fares trade balance, the study estimates positive effects in both the short and
the long–run. However, for other transportation trade balance, the empirical results show
initial negative impacts followed by a positive impact in the long–run. However, the
long–run impact of the exchange rate on other transportation services is insignificant.
This result further establishes that the J-curve phenomenon is valid using the traditional
and new definition only for the other transportation services trade between the USA and
the UK.
Similarly, Bahmani-Oskooee and others (2013) examine the response of 106 industries
to currency depreciation in the USA for its trade with Italy over the 1979–2010 periods.
Using the ARDL methodology, the analysis reveals that in almost half of the sectors in
the study, currency depreciation has both favourable and unfavourable impact in the
116
short–run. However, there are only 19 industries for which short–run improvement lasts
into the long–run improvement of the trade balance. Hence, using the old and the new
definition, the study finds evidence of the J-curve phenomenon in only 5 sectors. These
sectors are coffee; textile and leather machinery; jewellery and gold/silver smiths wares;
petroleum products and road motor vehicle trade balance.
In another study, Bahmani-Oskooee and Xu (2013) test the presence of the J-curve
phenomenon in 73 sectoral trades between Japan and China over the 1978–2008 periods.
Using the model specification of Bahmani-Oskooee and Wang (2007), the study finds
evidence of the J-curve phenomenon in the case of 24 industries. In another similar
study by Bahmani-Oskooee and Zhang (2013), they find evidence of the J-curve
phenomenon in 12 out of the 47 industries trade between the UK and China. In
particular, over the 1978-2010 periods, five sectors show presence of the phenomenon
using traditional definition and seven sectors hold it valid using the new definition.
Recently, Bahmani-Oskooee and others (2014) use quarterly data for ten industries for
Azerbaijan over the 2000–2009 periods to analyse the presence of the J-curve
phenomenon. The results show that the J-curve phenomenon is valid in most of the cases
but shows long–run improvements in 3 out of the 10 industrial trades only. In another
recent work, Bahmani-Oskooee and Zhang (2014) do not find significant effects of
depreciation on Korea’s aggregate trade with the rest of the world. However, suspecting
the issue of aggregation bias in the study, they disaggregate the trade flows into 148
sectoral trade balances. As a result, the study finds significant impact of currency
depreciation on the trade performances of 91 industries. However, the short–run effects
last into the long–run favourable gains in only 26 cases. Hence, using the new definition,
the study confirms the presence of the J-curve phenomenon for these 26 industries only.
The studies reviewed in this section try to remedy the aggregation bias problem which
might distort the findings on the J-curve phenomenon on the economy’s trade
performance. The results, however, like the aggregate trade analysis have been confined
117
to the country in study. The approaches for assessing the J-curve phenomenon in most
instances have been found to favour either the traditional or the new definition.
Moreover, the literature on the J-curve in recent years has not only analysed aggregate
and sectoral level trade performances but also at the bilateral level of trade. It has been
argued that using data at bilateral level would also help remedy the aggregation bias
problem (Narayan (2006), Bahmani-Oskooee and Harvey (2012), Wang et al. (2012)
and Dash (2013)). To this end, for the interest of this study, the next section highlights
some of the recent studies that investigate the impact of currency devaluation or
depreciation and attempt to validate the J-curve phenomenon at bilateral level.
5.2.2 On the bilateral trade flows and the J-curve phenomenon The studies reviewed in the previous chapter and in the previous section at the aggregate
and sectoral level trade, respectively fail to find overwhelming evidence of the J-curve
phenomenon. Among a number of scholars, Rose and Yellen (1989), Bahmani-Oskooee
and Brooks (1999) and Bahmani-Oskooee and others (2005) argue that despite the
overall response of currency devaluation or depreciation on the country’s overall trade
balance, the impact at the bilateral level trade could be very different. They reason that
this is because the changes in the exchange rate could cause deterioration in the bilateral
trade balance with one country and improvement with another.
In fact, the overall trade balance, which is simply an aggregation of bilateral trade
balances, tends to be influenced by changes in the bilateral trade performance. However,
due to heterogeneity among trading nations, most recent studies employ bilateral
exchange rate and bilateral trade data to analyse the impact of currency depreciation or
devaluation on bilateral trade performance. Hence, in this section, some of the studies
that have significantly contributed to the bilateral trade literature evolving around the J-
curve phenomenon are reviewed.
Rose and Yellen (1989) are probably the earlier scholars that attempted to analyse the J-
curve phenomenon using bilateral data for the USA with her six trading partners, namely
118
Canada, France, Germany, Italy, Japan and the UK over the 1960–1985 period. The
study argues that using bilateral trade data for analysis is more useful as it provides
better understanding into the dynamics of the exchange rate and trade balance
relationship. Using the conventional reduced form of the trade balance model, the
bilateral trade balance model developed in the study is as follows:
),,,( , jttUSjtjt YYREXfTB �
where jtTB is the US trade balance with country j in US real net exports; tUSY , is the
US real GDP and jtY is the real GDP in country j . jtREX represents the real exchange
rate of US$ vis-à-vis country j ’s currency defined as PEPx /* . E is the nominal
exchange rate defined as the number of domestic currency units per units of foreign
currency while *xP is the trading partner country’s price level and P represents the
domestic price level.
However, the study using the Engle-granger cointegration technique fails to find any
evidence of the long–run relationship among the variables in the bilateral trade models.
Hence, using the traditional definition, the study does not find support for the J-curve
phenomenon on the trade balance for USA with its major trade partners. Nonetheless,
the contribution of this study in its effort to disaggregate the trade data and establish its
relationship with the exchange rate is worth appreciation.
Bahmani-Oskooee and Brooks (1999) utilise a model similar to Rose and Yellen (1989)
and revisit the countries under study but with the use of a different technique. The study
uses the Johansen-Juselius Full Information Maximum Likelihood (FIML) estimation
technique on quarterly data over the 1973Q1–1996Q2 period. The estimation results do
not find specific short–run pattern and hence, no evidence of the J-curve phenomenon is
obtained using the traditional definition. However, the long–run results suggests that real
depreciation of the US dollar has a favourable effect on the USA’s trade balance with at
least four (France, Italy, Japan and the UK) of her trading partners.
119
However, the long run results obtained by Bahmani-Oskooee and Brooks (1999) are in
contrast to what was earlier concluded by Rose and Yellen (1989). Though Rose and
Yellen (1989) failed to find any short– or long–run relationship, the latter study
established significant long–run association of exchange rate of USA with its major
trade partners. The differences in the results we argue are largely because of the use of a
different econometric technique and using an extended time period for analysis.
However, Bahmani-Oskooee and Ratha (2004) using the same model specification
provide support for the J-curve phenomenon in the case of USA’s trade with the
Netherlands using the traditional definition. However, when the results of the other
eighteen25 trading partners are subjected to the new definition of assessing the J-curve
phenomenon, the study finds evidence of the phenomenon with ten of USA’s trading
partners over the 1975Q1–2000Q2 periods. These trade partner countries include
Austria, Denmark, France, Germany, Ireland, Italy, Japan, New Zealand, Sweden and
Switzerland. However, in another recent study, Bahmani-Oskooee and Xu (2012) do not
find the presence of the J-curve phenomenon in USA’s trade balance with Hong Kong
over the 1978–2006 periods.
Arora and others (2003) use the similar reduced form of the trade balance model and the
econometric technique to examine the presence of the J-curve phenomenon in India with
seven of its trading partner countries. The countries analysed in the study are Australia,
France, Germany, Italy, Japan, the UK and the USA over the 1977Q1–1998Q4 period.
The study using the traditional definition does not find any evidence of the J-curve
phenomenon. However, using the new definition, the study finds support for the bilateral
J-curve phenomenon in India with Australia, Germany, Italy and Japan.
Similarly, Onafowara (2003) examines the validity of the J-curve phenomenon in the
case of bilateral trade of Thailand, Indonesia and Malaysia with Japan and the USA over
the 1980–2004 periods. With the use of VECM and IRF analysis on the reduced form of 25 The eighteen trading partner countries of USA in the study by Bahmani-Oskooee and Ratha (2004) are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland and the UK.
120
the trade balance model, the study finds strong evidence of the J-curve phenomenon. In
particular, the study estimates significant and favourable impact of currency depreciation
on the trade balance for six country pairs in the long–run. Malaysia and Indonesia show
the J-curve phenomenon with USA and Japan while trade in Thailand suggests evidence
of the J-curve phenomenon with USA only.
Furthermore, Bahmani-Oskooee and others (2005) explore the bilateral trade
performance of Australia with its twenty-three trading partners using the quarterly data
series over the 1973Q1–2001Q3 periods. With the use of the ARDL technique and error
correction modelling, they utilise the similar trade balance model as used widely in the
literature. Their empirical analysis provides some evidence on the presence of the J-
curve phenomenon. Using the new definition, the study finds support for the J-curve
phenomenon for Australia’s trade with Denmark, Korea and New Zealand only.
Similarly, Bahmani-Oskooee and others (2006) studying the UK’s bilateral trade
performance with twenty of its trading partner, finds support of the J-curve phenomenon
with Canada and the USA only with the traditional definition. However, in the long–run
the bilateral exchange rate is estimated to have a significant favourable impact on the
UK’s trade balance with Australia, Austria, Greece, Singapore, Spain and South Africa.
Moreover, Narayan (2006) examines China’s bilateral trade relationship with the USA
using monthly data from November 1979 to September 2002. Using the ARDL
technique and the reduced form of the trade balance model, the study suggests that
currency depreciation in China improves its trade balance with the USA both in the
short– and in the long–run. Hence, the study shows no evidence of the J-curve
phenomenon using the new definition and IRF analysis. Even Bahmani-Oskooee and
Wang (2006) find similar results on the exchange rate and trade balance relationship of
China with the USA.
Celik and Kaya (2010) analyse Turkey’s bilateral trade dynamics with seven countries
using quarterly data for the period from 1985Q1–2006Q4. The study considers Turkey’s
trade relations with France, Germany, the Netherlands, Italy, Japan, the UK and the
121
USA using panel cointegration techniques. Following the literature, they adopt the
reduced form of the trade balance model and find that real depreciation improves
Turkey’s trade balance with at least three countries (Germany, Japan and the UK) in the
long–run. However, the study is unable to find support for the J-curve phenomenon
using IRF analysis but establishes that there are indications of the inverse J-curve
phenomenon in the case of Germany and the USA.
Bahmani-Oskooee and Harvey (2012) analyse the short– and long–run effects of real
exchange rate depreciation of the Singapore dollar on her trade balance with its
thirteen26 major trading partner countries over the 1973Q1–2009Q4 periods. Following
the model as widely discussed in the literature, they use the bounds testing approach to
cointegration and error-correction modelling to establish the relationship. The study uses
both the traditional and the new definition to assess the J-curve phenomenon. However,
using the traditional definition, the study fails to find any evidence of the J-curve
phenomenon but confirms the phenomenon using the new definition with China, the
Philippines, Saudi Arabia and the USA only.
Similarly, Wang and others (2012) investigate the J-curve hypothesis and the long–run
effect of the exchange rate on the trade balance of China with its eighteen27 major
trading partner countries over the months of August 2005 to September 2009. Utilising
the reduced form of the trade balance model, they adopt the panel FMOLS technique
and panel error correction model to test the J-curve phenomenon. Their empirical
analysis using the traditional definition does not provide support for the J-curve
phenomenon but rather support for the inverted J-curve hypothesis. In particular, it is
noted that real depreciation of the Chinese Renminbi (RMB) has a positive impact in
improving China’s trade flows with Japan, the UK and the USA only.
26 The thirteen trading partner countries of Singapore in study by Bahmani-Oskooee and Harvey (2012) are Australia, Canada, China, Hong Kong, India, Japan, Korea, Malaysia, the Philippines, Saudi Arabia, Thailand, the UK and the USA. 27 The eighteen trading partner countries of China in the study by Wang et.al (2012) are Australia, Brazil, Canada, France, Germany, India, Italy, Japan, Korea, Malaysia, the Netherlands, the Philippines, Taiwan, Thailand, Russia, the UK and the USA.
122
The evidence of the J-curve phenomenon with the USA is also supported by the earlier
findings of Bahmani-Oskooee and Wang (2006). Wang and others (2012) also find that
currency depreciation leads to deterioration of Chinese trade balance with Germany,
Italy and the Netherlands. In other cases apart from Brazil and Russia, they do not find
any significant relationship between bilateral exchange rate and trade balance in China.
Recently, Dash (2013) examines short– and long–run effects of real exchange rate
changes on India’s trade balance with its four trading partners, Germany, Japan, the UK
and the USA. The study with the use of the reduced form of the trade balance model and
IRF analysis finds evidence of the J-curve phenomenon on India’s trade balance with
Japan and Germany only over the 1991M1–2005M6 period. The study as such also
provides support to the earlier findings of Arora et al. (2003). Arora et al. (2003) had
also established evidence of the J-curve phenomenon for India’s trade with Japan and
Germany in their study. In another recent study, Bahmani-Oskooee and Xu (2013) fail to
find evidence of the J-curve phenomenon between Japan and China.
The studies reviewed in this section on the effectiveness of currency depreciation and
devaluation on the bilateral trade performance along with those on the validity of the J-
curve phenomenon try to remedy the aggregation bias problem. It is argued that
analysing the relationship of exchange rate with trade balance in the presence of this
biasness, misrepresents the role of exchange rates on the country’s trade performance.
The results, however, like the aggregate and sectoral trade analysis, have been confined
to the country context. Additionally, there have been no superior methods of assessing
the J-curve phenomenon and as a result all the three methods (traditional definition, new
definition, IRF analysis) have been used widely in the literature.
Hence, in what follows, relevant studies that have attempted to establish the relationship
of exchange rates with the sectoral and bilateral trade performance in the context of Fiji
are reviewed.
123
5.2.3 On the impact of devaluation on sectoral and bilateral trade flows in Fiji The literature on the response of exchange rate at sectoral and bilateral level trade
balance is limited in the context of Pacific island developing economies. The review
suggests that there is not yet any study that has attempted to investigate the response of
sectoral and bilateral level trade balance to currency devaluation in Fiji. Similarly, there
is no study on the J-curve phenomenon for sectoral and bilateral trade in Fiji. However,
the literature review shows that there is one study involving the sectoral trade and
exchange rate relationship by Murphy (as cited in Singh, 2006). Additionally, there is
also one study on the bilateral trade and exchange rate relationship by Kaufmann (2008).
Besides, these two studies, there are no other studies to the interest of this study.
Therefore, the following review presents and critically analyses the findings from these
two particular studies in the context of Fiji.
The study by Murphy (as cited in Singh, 2006) which is some two decades old is
probably the only study that has attempted to analyse three sectoral export demand
equations for Fiji over the 1974–1986 periods. The study in particular, disaggregates
exports into three major categories of sugar, travel and other goods. Using the Partial
Adjustment Model (PAM), the study incorporates exchange rate only to model tourism
export demand but not the other equations. Hence, disregarding the time series problem
associated with the use of very small time–series samples, the results suggest that a 10%
currency devaluation in Fiji leads to a 6.5% improvement in the tourism export revenue.
However, Singh (2006) argues that the failure to include exchange rate variable in other
equations, does not allow this study to make any conclusions on the effects of exchange
rate on the exports of the other two sectors.
Moreover, on the bilateral front, Kaufmann (2008) is the only study which attempts to
estimate export and import demand equations for Fiji with Australia, New Zealand and
the European countries28. The study follows the model specification of Rao and Singh
(2006) to estimate the trade equations for Fiji with its selected trading partner countries
28 The European countries included in the study are Belgium, Denmark (only import equation), France, Germany, Italy (only import equation), the Netherlands and the UK.
124
over the 1976–2006 periods. The study uses three econometric techniques for estimating
the models: Philip Hanson’s Fully Modified Ordinary Least Squares (FMOLS), LSE
Hendry’s General to Specific (GETS-NLLS) and Johansen Maximum Likelihood (JML)
method.
The study finds that the income elasticity of the demand for Fiji’s exports to Australia is
around unity and is estimated to be between 0.78 and 1.0. The relative elasticity for the
demand for exports is estimated to be approximately around -0.10. This indicates that
10% currency devaluation in Fiji improves exports to Australia by 1% in the long–run.
However, on the import front, the results indicate that a similar 10% devaluation results
in a decrease in imports from Australia by 4.4 to 5.5%.
With regard to the trade with New Zealand, the study finds that income elasticity for
exports and imports ranges between 0.73 to 0.75 and 0.75 to 1.14, respectively. In
particular, 10% currency devaluation is estimated to cause an improvement in the
exports by around 4.4 to 5.4% and subsequent decline in imports by 3.2 to 4.0%.
However, based on the estimates of the separate export and import demand, it is argued
that the M-L condition for Fiji’s trade with Australia and New Zealand is not been
achieved. The estimates show that the sum of the relative import and export price
elasticities is less than one, indicating that a real devaluation in Fiji worsens its trade
performance with Australia and New Zealand in the long–run.
Furthermore, with respect to Fiji’s trade with the EU countries, the export and import
demand equations show that besides the UK, Fiji has strong trade relationships with
Belgium, Denmark, France, Germany, Italy and the Netherlands. The UK, which is
Fiji’s major trading partner to the EU, however, shows a favourable but insignificant
impact of devaluation. Nonetheless, this study did not advance further to investigate the
J-curve phenomenon at the bilateral level. Though, the study is worth appreciation for its
contribution on estimating bilateral export and import demand equations for a
developing Pacific island country, Fiji.
125
As presented, the literature involving sectoral level trade and exchange rate relationship
is very thin in the context of PIC and for that matter in the context of Fiji’s economy.
The one and only study in the case of Fiji, though appreciative of its attempt, is clearly
outdated and very narrow. The study as discussed has only explored the relationship
between exchange rate and export of tourism services with no further analysis of the
other goods and services sector. The study, therefore, can be improved upon.
Firstly, the study is two decades old and requires new insights into the relationship of
exchange rate and export of tourism services. Secondly, since 1992, Fiji’s economy has
gone through two more devaluations in 1998 and 2009 by 20% in each case. This
definitely warrants further investigation of whether the relationship is still valid. Thirdly,
since only the tourism export demand equation was being investigated, the impact of
devaluation on other important sectors in the economy was ignored. It also does not
proceed further to investigate the sectoral level J-curve phenomenon. The study also
suffers from many specifications and econometrics issues which creates doubts on the
findings of the study.
Moreover, the inclusion of Australia, New Zealand and the EU countries in the only
study on Fiji’s bilateral trade analysis is commendable for its attempt. Nevertheless, the
study can be improved upon. Firstly, Fiji’s currency has been devalued by another 20%
in 2009 after the study was undertaken. Hence, this is very likely to have established
new bilateral trade and exchange rate relationship in the economy. Secondly, Fiji’s
bilateral trade flows with Japan and the USA could have also been included. This is
because as discussed in Section 3.3.1.1, Fiji’s fixed exchange rate is not only determined
based on the currencies of Australia, New Zealand and the UK but it also includes the
currencies of Japan and the USA. Thirdly, besides only these five major trade partners
and the EU countries, Fiji also engages in a significant amount of trade with some
emerging Asian trading partner countries. Bilateral trade analysis with Asian trade
partners such as China, Hong Kong, India, Malaysia and Singapore has not been
included in any other previous study in the context of Fiji.
126
Therefore, it is evident that the literature on estimating the relationship of exchange rate
with sectoral and bilateral level trade is insufficient in Fiji. In particular, a
comprehensive study on the impact of devaluation on sectoral and bilateral level trade
performance along with its associated test on the J-curve phenomenon in Fiji’s economy
has not yet been done. Therefore, it is the modest attempt in this chapter to add
significantly to this area of knowledge in the context of a Pacific island developing
economy, Fiji.
5.2.4 Findings from the Literature review: Sectoral and Bilateral trade flows The advancement in the econometric techniques and the demand for more disaggregated
analysis, have enticed scholars in recent times to analyse the impact of currency
devaluation or depreciation at sectoral and bilateral level trade as well. It has been well
established in the literature that further disaggregating the trade flows would help
remove the aggregation bias problem caused by only using aggregate level trade data.
The review as such shows that the various studies carried out around the globe have in
fact revealed some interesting country experiences. However, similar to the aggregate
trade outcomes, the results from the various country experiences on sectoral and bilateral
trade performance do not come to consensus.
The first part of the literature review has tried to assemble relevant literature dealing
with the impact of currency devaluation or depreciation on the performance of sectoral
level trade performance. Various studies in this context have tried to address the
aggregation bias problem which exists in the aggregate trade analysis. As such, many
scholars have attempted to model trade performance using the widely used reduced form
of the trade balance model for various sectors with its trading partner’s countries. While
modelling a number of sectoral trade balance performance, studies have on many
instances showed presence of the J-curve phenomenon. In particular, recent studies by
Bahmani-Oskooee and Wang (2007), Bahmani-Oskooee and Bolhasani (2008),
Soleymani and Saboori (2012), Bahmani-Oskooee et al. (2013), and Bahmani-Oskooee
and Zhang (2013) find evidence of the J-curve phenomenon at sectoral level while
127
Yazici (2006) and Yazici and Klasra (2010) do not find the J-curve phenomenon in their
analysis.
In the second part, the review focuses on some of the recent studies estimating long– and
short–run relationship of exchange rate with bilateral trade performance and testing the
presence of the J-curve phenomenon. Yet again, there have been studies with contrasting
conclusions. In particular, using any of the three methods of assessing the J-curve
phenomenon, studies of Rose and Yellen (1989), Shirvani and Wilbratte (1997),
Narayan (2006), Bahmani-Oskooee and Wang (2006) and Celik and Kaya (2010) fail to
find evidence of the bilateral J-curve phenomenon. On the other hand, Arora et al.
(2003), Bahmani-Oskooee and Ratha (2004), Bahmani-Oskooee et al. (2005) and
Bahmani-Oskooee and Harvey (2012) establish presence of the bilateral J-curve
phenomenon in their analysis. The review also highlights that most of the studies have
largely focused on sectoral and bilateral trade analysis for developed countries with little
emphasis on developing countries. As a result, it is evident that the scholars have not put
adequate emphasis for similar studies for the countries in the South Pacific region.
Furthermore, the relevant literature in the context of a Pacific island economy, Fiji, for
the purpose of this study, is very narrow. The only known study classified as part of the
sectoral level trade analysis is old and has only focused on one sectoral trade. Even the
only known study analysing the impact of currency devaluation on bilateral trade
performance in Fiji has its own shortcoming as discussed earlier. Therefore, it is clearly
evident that though the literature on establishing relationship of exchange rate with
sectoral and bilateral level trade balance along with its associated study on the J-curve
phenomenon has been growing, little focus has been given to the Pacific island
economies. As a result, there is no study as yet that has attempted to assess the J-curve
phenomenon at sectoral and bilateral level trade in the context of PICs.
Consequently, the growing body of the exchange rate and trade literature along with the
tests on the J-curve phenomenon at sectoral and bilateral level warrants some
experiences in the context of the South Pacific region as well. Hence, this study provides
128
a modest attempt to this body of knowledge by investigating the J-curve phenomenon at
sectoral and bilateral level trade in Fiji. Hence, in the sections ahead, a modest attempt is
made to analyse Fiji’s trend of sectoral and bilateral trade performance, after which the
empirical estimations are undertaken. Towards the end, the three approaches used to test
for the presence of the J-curve phenomenon are applied at the sectoral and bilateral level
trade estimates to contribute to the literature on the J-curve phenomenon.
5.3 Pace and patterns of Sectoral and Bilateral trade flows in Fiji Fiji, being centrally located in the South Pacific region, has utilised its geographic
advantage to build trade relations with many countries around the globe. It has also acted
in many instances as the link for facilitating trade from the rest of the world to other
PICs. In an effort to support trade liberalisation and to allow global economic
integration, Fiji since 1989 has paved the way for many trading agreements29.
Over the years, it has gradually opened up its market to global trade and as a result has
increased the number of trading partners in its trading list. Being a small developing
economy coupled with resource constraints, Fiji is also trapped into being a highly
import dependent economy. With limited domestic exportable commodities such as
sugar, garments, food products, gold and fish, it often ends up with large volumes of
importable commodities such as mineral fuels, chemicals, machinery and equipment,
among others.
Hence, apart from only studying how devaluation affects aggregate trade performance, it
becomes very important to analyse how it performs at sectoral and bilateral level as well.
However, before beginning with the analysis at sectoral and bilateral level, it is
important to mention that the analysis of Fiji’s bilateral trade performance has been
based on goods trade data only. This is because of the unavailability of consistent
bilateral services trade data for Fiji.
29 See Appendix B for a brief discussion on some of the important trade agreements in Fiji.
129
5.3.1 Sectoral trade trend and patterns
Pacific countries including Fiji are largely characterised as import dependent economies
with few exportable commodities. Though a detailed analysis on the various export and
import commodities over the years in Fiji would have been more interesting, the
discussion and analysis of major sectoral trade in 2012 is the emphasis of this section.
5.3.1.1 Sectoral performance The breakdown of Fiji’s goods export sector in 2012 reveals that food products account
for around 38% (US$225 million) of its total domestic exports (Figure 5.1). Among
these, sugar exports make up 17% (US$98 million) while 5% (US$30 million) is being
accounted for by export of fish. Beverages, spirits and tobacco contribute 16% (US$96
million) while exports of mineral products such as gold contribute another 13% (US$77
million). Export of garments contributes 8% (US$50 million) while crude materials add
up another 9% (US$54 million) of total domestic exports in 2012. Additionally, around
26% (US$150 million) is being contributed by other commodities (animal and vegetable
oil products, other food products, chemicals, machinery and transport equipment and
miscellaneous manufactured products) in small proportions.
Figure 5.1 Composition of Fiji’s goods exports sectors in 2012
Source: Fiji Bureau of Statistics, Key Statistics (June, 2014)
Sugar17%
Fish5%
Beverage and Tobacco16%
Crude Materials9%Manufactured Goods
5%
Garments9%
Gold13%
Others26%
130
It is however, also important to note that the recent shares of goods exports analysed in
Figure 5.1 have not necessarily had the same proportions in prior years. Its volume and
trend path has fluctuated over the years (Figure 5.2). In particular, export of sugar and
food products, has experienced gradual decline in real values while a slight increase is
noted for the export of fish and gold over the years. Fiji’s total exports have been largely
characterised by exports of sugar in the 1970s and 1980s which then accounted for
around 70% of total domestic exports. However, its share has gradually declined to
around 17% of total goods exports in 2012.
Prasad and Narayan (2008) argue that this trend is mainly due to the expiration of land
leases and the deregulation policies undertaken by previous governments. This led to the
withdrawal of support for developing the agricultural sector. In addition to this, the
expiration of the preferential treatment of sugar exports to the EU and declining
productivity in sugarcane industry has led to a reduction in its exports.
Figure 5.2 Fiji’s trade trend of her major goods exports sectors, 1975–2012
Source: Fiji Bureau of Statistics, Overseas Merchandise Trade Statistics (various years)
The goods import sector has for a long period of time been in excess of total goods
exports in Fiji. This has resulted in net goods trade deficit in the economy. The goods
imports sector is largely made up of imports of mineral fuel, manufactured goods,
machinery and related equipment, along with animal and vegetable products (Figure
0
1
2
3
4
5
6
7
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Real
2005
US$
m
Food Sugar Fish Gold
131
5.3). A simple breakdown of the imports sector in 2012 clearly points out that the
importation of mineral fuel, food products, machinery and transport equipment largely
dominates the imports sector. In total, these sectors account for the 68% of Fiji’s total
goods import. More specifically, import of mineral fuels alone accounts for around one-
third (US$678 million) of the total import bills in the country. The food import bill
accounts for 19% (US$430 million) while machinery and transport equipment makes up
another 19% (US$426 million) of total imports.
Figure 5.3 Composition of goods imports sectors in 2012
Source: Fiji Bureau of Statistics, Key Statistics (June, 2014)
Other important components of imports include manufactured goods (12%),
miscellaneous manufactured goods, which includes mattresses, heating and lighting
fixtures and fittings among others (8%) and chemicals (8%). Besides, the import of
mineral fuels, other components of imports have largely increased in absolute real terms
but their share in total imports has remained relatively the same over the years (Figure
5.4).
Food19%
Mineral Fuels30%
Chemicals8%
Manufactured Goods12%
Machinery and Transport
Equipment19%
Miscellaneous Manufacturd
Goods8%
Others4%
132
Figure 5.4 Fiji’s trade trend of her major goods imports sectors, 1975–2012
Source: Fiji Bureau of Statistics, Overseas Merchandise Trade Statistics (various years)
Furthermore, the services sector trade balance in Fiji has remained in surplus for quite a
long period of time. Fiji has largely been dependent on the accumulation of foreign
reserves from the services sector to meet its high import bills. The export of services in
the country is largely characterised by travel or tourism services along with
transportation service earnings. These two sectors together make up around 85% of
service exports while travel services on its own accounts for 60% (US$730 million) of
receipts in 2012 (Figure 5.5).
Figure 5.5 Composition of services exports sector in 2012
Source: Fiji Bureau of Statistics, Key Statistics (June, 2014)
0
1
2
3
4
5
6
7
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Real
2005
US$
m
Food Mineral Fuels Oils and FatsManufactured Goods Textiles Machinery and Transport EquipmentBeverage and Tobacco Crude Materials ChemicalsMiscellaneous Manufacturd Goods
Transportation25%
Telecommunications, computer, and
information services 2%
Travel60%
Other business services 3%
Government goods and services
9%
Others1%
133
Other sectors making up the total services exports in 2012 (US$1,222 million) include
telecommunication, computer and information services (US$20 million), government
goods and services (US$109 million), financial services (US$4 million) and other
business services (US$32 million). In particular, travel services revenue in the last
decade has risen significantly from US$184 in 2000 to US$730 in 2012. With a rising
tourism industry, Fiji attracts visitors each year from around the globe and in particular
from Australia (51%), New Zealand (16 %), the United States of America (9%), other
PICs (8%), Europe (4%), UK (3%) and Canada (2%) in 201230.
On the other hand, services imports are also largely characterised by the imports of
transportation and travel services as well. More specifically, transport services account
for 57% (US$325 million), travel services accounts for 16% (US$92 million), IT
services contribute 8% (US$48 million), insurance and pension makes up 7% (US$40
million) while other business services account for another 10% (US$49 million) of the
total services imports bill in 2012 (Figure 5.6).
Figure 5.6 Composition of services imports sector in 2012
Source: Fiji Bureau of Statistics, Key Statistics (June, 2014)
30 Appendix A shows annual visitor arrivals in Fiji over the 1970–2014 periods.
Transportation57%
Other business services 10%
Insurance and pension7%
Telecommunications, computer, and
information services 8%
Government goods and services
2%
Travel16%
134
The analysis of the major components of services trade in the last three decades
indicates that both exports and imports of services sector are gaining momentum (Figure
5.7). In particular, export of travel services, which includes earnings from tourism, has
been consistently increasing. Despite recording some slight decline in the year of coup,
it has been on the rise. Similarly, migration appears to be on the rise in the year of a
coup. This has resulted in the increase in import of travel services in those years.
Transportation services, on the other hand, have also experienced modest increase in its
exports and imports over the last three decades.
Figure 5.7 Fiji’s trade trend of her major services exports and imports sectors,
1975–2012
Source: Fiji Bureau of Statistics, Overseas Merchandise Trade Statistics (various years)
5.3.1.2 Sectoral trade response to devaluation in Fiji: A simple ‘before-after’ approach
Having analysed and gained a preliminary understanding on the major sectoral
components of goods and services exports and imports sectors in the economy, this
section attempts to analyse the major sectoral trade using the ‘before-after’ approach.
The application of this approach to evaluate the response of trade components to
currency devaluation in Fiji is discussed in Section 4.3.3.
0
1
2
3
4
5
6
7
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Real
2005
US$
m
Export of Travel services Import of Travel servicesExport of transportation services Import of Transportation services
135
As part of this sectoral analysis, only sectors that form a significant part of Fiji’s trade
and for which there is a complete dataset for at least the last three decades is included.
Hence, for the goods sector, included in the analysis are the exports of food, sugar, fish
and gold whereas the major importable products include food, mineral fuels,
manufactured goods, textiles, oils and fats, machinery and transport equipment, tobacco
and beverages, crude materials, chemicals and miscellaneous manufactured goods. For
the services sector analysis, only exports and imports of travel and transport sector
services are analysed.
Based on the analysis in Table 5.1, the goods sector exports appear to have been largely
benefited by the 1987 devaluation out of all the other devaluation years. In particular, all
the goods exports sectors including food, sugar, fish and gold responded positively to the
1987 devaluation. However, similar to the aggregate performance with regards to the
1998 devaluation, the export in goods sector experienced subsequent decline. However,
the 2009 devaluation shows mixed responses. It appears to have benefitted the export of
gold and fish while it has negatively affected exports of the food and sugar sectors. The
exports of sugar after a lapse of three years after the 2009 devaluation declined by 4.4%
for 1% currency reduction while the export of gold increased by 28.0% in the same
period.
Furthermore, the goods imports sector in response to the 1987 devaluation has
experienced a corresponding long–run increase in importable commodities, except for
the import of mineral fuels. Though in many instances, it is noted that immediately after
devaluation import of these commodities declined, when its impact is analysed for a
longer period of time, the response is different. In particular, it is noted that though the
import of food, manufactured goods, beverage and tobacco, chemical, miscellaneous
manufactured goods and machinery and transport equipment declined immediately after
the 1987 devaluation, it often increased in the longer term. This is because of the high
import dependence of the economy on these commodities. Additionally, mixed response
for the import of these commodities is noted after the 1998 devaluation.
136
Table 5.1 Response of major goods sectors trade performance to 1% devaluation in Fiji
Devaluation Years
1987 1998 2009
1st 3rd 5th 1st 3rd 5th 1st 3rd 5th EXPORTS
Food 1.6 1.0 0.9 0.1 -0.9 -0.6 -1.0 -1.4 -2.0
Sugar 1.7 1.0 0.7 0.4 -1.1 -1.2 -1.9 -4.4 -5.0
Fish 1.6 2.9 3.1 -0.5 -0.4 0.3 1.3 2.3 1.2
Gold 1.2 2.9 2.9 -0.6 -0.7 -0.5 4.2 28.0 12.4
IMPORTS Food -0.1 0.4 0.4 -0.1 -0.3 1.0 0.0 2.3 3.9
Mineral Fuels -0.8 -0.6 -0.5 -1.4 1.8 4.6 -3.2 -1.6 -0.4
Oils and Fats 1.5 -0.1 0.1 -0.6 -1.0 -0.7 -1.6 1.8 4.8
Manufactured Goods
-0.4 0.8 1.3 -0.3 0.4 0.7 -0.9 -1.1 -1.8
Textiles 0.3 1.9 2.9 0.3 1.4 1.8 -1.0 -0.7 -2.5
Machinery & Transport
Equipment -1.6 0.7 1.4 1.3 0.9 0.2 -1.3 -0.7 -0.6
Beverage & Tobacco
-0.4 0.3 0.5 -1.3 -1.4 -0.3 0.3 1.6 1.6
Crude Materials 0.2 0.8 0.5 -1.3 0.3 1.5 -1.5 -1.2 0.3
Chemicals -0.5 0.6 0.7 -1.3 -0.5 0.5 -1.4 0.2 0.8
Miscellaneous Manufactured
Goods -0.8 0.1 0.5 -0.5 1.2 1.6 -1.0 -0.6 -1.0
TRADE BALANCE
Food 3.0 1.5 1.4 0.4 -1.5 -2.3 -158.3 -63.9 -64.8
Source: Fiji Bureau of Statistics and Author’s calculation
The recent 2009 devaluation also shows some mixed responses. It is noted that though
the import of mineral fuels has increased after the 1998 devaluation, it has experienced
continuous decline after the 2009 devaluation. Additionally, the import of manufactured
goods, textiles, crude materials, miscellaneous manufactured goods, machinery and
137
transport equipment, which experienced long–run modest increase after the 1987 and
1998 devaluation, experienced long–run decline after the 2009 devaluation. Moreover,
on the overall trade front, food trade balance is noted to have improved only after the
1987 devaluation but experienced subsequent decline after the 1998 and 2009
devaluations. This is consistent with declining exports and increasing imports of food
products in those periods.
Moreover, the services sector has for a long time helped improve Fiji’s overall trade
balance position (Table 5.2). The two important components of services trade are the
travel and transport sector which together makes up more than 70% of services trade in
Fiji. The analysis finds that the devaluation has resulted in an increase of both exports
and imports of travel and transportation services after the 1987 devaluation. Since the
increase in travel imports after the 1987 devaluation seems to be higher than the increase
in exports, the ultimate impact on the travel trade balance is a decline. The transport
sector, on the other hand, experienced modest improvements.
Table 5.2 Response of major services sectors trade performance to 1% devaluation in Fiji
Devaluation Years
1987 1998 2009
1st 3rd 5th 1st 3rd 5th 1st 3rd 5th EXPORTS
Travel -1.4 0.0 0.3 0.0 -0.1 0.2 -0.4 1.5 2.1
Transport -0.5 1.2 2.5 -1.0 -1.1 -0.7 -2.8 -1.4 -0.3
IMPORTS Travel 7.1 3.1 3.1 -0.3 1.5 1.6 1.5 -0.2 -1.3
Transport -0.4 0.9 1.2 -1.1 -0.7 -0.2 -2.8 -1.3 -0.7
TRADE BALANCE Travel -2.9 -0.5 -0.1 0.1 -0.5 -0.3 -0.7 1.9 3.0
Transport 0.3 -0.4 0.8 0.4 -13.7 -17.9 -2.5 -1.1 5.4
Source: Author’s calculation
138
As far as the 1998 devaluation is concerned, the analysis shows that both the travel and
transport sectors trade balances have improved in the very short–run but deteriorated in
the longer term. However, after the 2009 devaluation, exports of travel services shows
modest improvements while the transportation exports sector experienced subsequent
decline. Though, both the sectors experienced decline in imports in the long–term, the
impact on the transport trade balance has been negative while the travel sector responded
positively.
The analysis on the impact devaluation on Fiji’s major sectoral trade does not reveal the
same level of responses on the various sectors in the economy. Most of the sectoral
exports are noted to experience improvements after the 1987 devaluation but mixed
responses after the 1998 and 2009 devaluations. The imports also show mixed responses
after devaluation in the country. In particular, the trade balance performance for the only
goods sector analysed shows that the food sector experiences deterioration in most
instances. The two components of the services sector trade balance (travel and
transportation sectors) also depict deteriorating performance of their respective trade
balances in most instances. This is consistent with our earlier findings on the aggregate
services trade performance. This also re-affirms our findings that the adverse impact of
devaluation on the services trade balance performances are due to the worsening effect
on its two major components.
Nevertheless, this section has provided a modest analysis on how the sectoral level trade
in Fiji has responded to devaluations in the country. However, the discussion and the
analysis in this section do not provide any understanding on how Fiji’s trade with its
major trading partner countries has responded to devaluations in the economy. This is
particularly important as this has an ultimate effect on the sectoral and aggregate level
trade performance. Hence, a similar analysis on the impact of devaluation on Fiji’s trade
performance with its major and emerging trade partner countries is presented in the next
section.
139
5.3.2 Bilateral trade trends and patterns
5.3.2.1 With whom does Fiji trade? More than half of Fiji’s domestic exports find their way to markets in Australia, the UK
and the USA. The largest share of Fiji’s domestic exports is to Australia which accounts
for around 28% of Fiji’s total domestic exports in 2012 (Table 5.3). The exports to
Australia largely include exports of gold and food products. The second most important
exporting destination is the USA. This accounts for another 18% and is dominated by
exports of mineral water. Closely following behind at 17% is trade to the UK which is
largely characterised by exports of sugar. Other important exporting destinations include
New Zealand (7%), Japan (4%), Vanuatu (3%) and Hong Kong (3%). It is also
important to highlight that trade with countries in Pacific region are also important as
Kiribati, PNG, Samoa and Vanuatu, are ranked among Fiji’s top ten exporting
destinations. These island countries together make up around 8% of Fiji’s exports.
Table 5.3 Fiji’s trading partners’ composition in 2012
Domestic Exports
Imports
Top 10 Countries Share (%) Top 10 Countries Share (%) Australia 27.7 Singapore 31.8
USA 17.5 Australia 18.3
UK 17.2 NZ 13.9
NZ 6.8 China 8.8
Japan 3.5 USA 3.0
Vanuatu 2.7 Japan 2.2
Hong Kong 2.6 Malaysia 2.2
Samoa 1.8 Thailand 2.2
PNG 1.7 Hong Kong 2.1
Kiribati 1.7 India 1.7
Source: Fiji Bureau of Statistics, Key Statistics (March, 2014)
140
Furthermore, Fiji’s imports mainly originate from Singapore, Australia, New Zealand
and China. The most important and obvious importing destination for Fiji is Singapore
which makes up around one third (32%) of its total imports. This mainly comprises the
imports of mineral fuels. In a similar manner to Fiji’s exports, Australia is also one of
the important partners for Fiji’s import demand. Fiji imports around 18% of its total
imports from Australia. The major commodities imported include food items, garments,
machinery, mechanical and electrical appliances. Fiji also imports around 14% of its
imports from New Zealand, which includes milk products and other food items.
Fiji, while also increasing its economic relations with one of the world’s fastest growing
economies, China, imports around 9% of its total imports from them. The imports from
China mostly include frozen fish, other food products and machinery items. Other
important importing sources in 2012 include USA (3%), Japan (2%), Malaysia (2%),
Thailand (2%), Hong Kong (2%) and India (2%). Fiji is also noted to be a net importer
with its two developed neighbouring countries of New Zealand and Australia. In
addition to this, it is also a net importer to most of its Asian trade partners. On the other
hand, it is a net exporter to a small number of countries mainly in the neighbouring PICs
but also including the UK and the USA.
5.3.2.2 Bilateral trade performance: Trade with major and emerging Asian trade partner countries
In this sub-section, brief discussions on the trend of Fiji’s trade performances with its
bilateral trade partners are presented. The countries considered as part of this analysis is
purely based on the availability of consistent time series bilateral trade data. These
countries include Australia, China, Hong Kong, India, Japan, Malaysia, New Zealand,
Singapore, the UK and the USA. The trading partner countries are categorised into two
different categories. Trade with ‘Major Trade Partners’ account for around 41% of total
trade while for those which are classified as ‘Emerging Asian Trade Partners’, it makes
up another 33% of total trade in 2012 (FBOS, 2013a).
141
Fiji’s trade performance with its major trade partners shows different trend paths. It is
noted that Fiji’s trade with the UK has, at least since 1975, been in trade surplus (Figure
5.8). Similarly for the USA, trade performance in recent years has experienced
improvements. Trade with Japan as well shows modest improvement over the years.
However, Fiji’s trade with Australia and New Zealand has always been in deficit. This is
largely because of Fiji’s high level of imports from these two neighbouring trading
partners be it either for capital or consumption goods.
Figure 5.8 Fiji’s trend of trade balance with her major trade partner countries, 1975–2012
Source: Fiji Bureau of Statistics, Overseas Merchandise Trade Statistics (various years)
Moreover, Fiji’s trend of trade performance with its emerging Asian trade partners,
except Singapore, shows similar trend paths (Figure 5.9). A fter the 2001–2002 oil crisis,
trade patterns have slowly started to differ. Since then, it is noted that Fiji’s trade with
these trade partners has been in continuous deficit. More noticeable is Fiji’s trade deficit
with Singapore. This is because Singapore is the major market for import of Fiji’s
mineral fuel supply. Since Fiji’s domestic exports to Singapore is relatively small, this
trade deficit has been a concern.
-500
-400
-300
-200
-100
0
100
200
300
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Real
US$m
Australia New Zealand Japan USA United Kingdom
Rea
l 200
5 U
S$m
142
Figure 5.9 Fiji’s trend of trade balance with her emerging Asian trade partner countries, 1975–2012
Source: Fiji Bureau of Statistics, Overseas Merchandise Trade Statistics (various years)
5.3.2.3 Bilateral trade response to devaluation in Fiji: A simple ‘before-after’ approach
In this section, similar ‘before-after’ approach is adopted as undertaken for the aggregate
and sectoral trade analysis to gain preliminary understanding on the bilateral trade
performance to devaluation in Fiji. The application of this approach follows the same
analytical procedures as discussed in Section 4.3.3.
The ten trading partner countries of Fiji that are being analysed have been classified as
‘Major Trade Partners’ and ‘Emerging Asian Trade Partners’. It is also important to
mention here that to capture the impact of devaluation on bilateral trade performance,
real effective exchange rate is used in all the cases except for Australia, Japan, New
Zealand and the USA. This is because as discussed in Section 2.7, consistent annual
nominal exchange rates of Fiji are available only for these four countries. This is used to
obtain real bilateral exchange rates for the analysis.
The real bilateral exchange rate of Fiji with its major trade partners has experienced
immediate improvement in international competiveness after each devaluation episode
-700
-600
-500
-400
-300
-200
-100
0
100
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
Real
US$m
Singapore China Malaysia India Hong Kong
Rea
l200
5U
S$m
143
(Table 5.4). However, after a lapse of three to five years, it has noted decline in
competitiveness.
Table 5.4 Response of Bilateral trade performance with major trade partner countries to 1% devaluation in Fiji
Devaluation Years
1987 1998 2009
1st 3rd 5th 1st 3rd 5th 1st 3rd 5th
AUSTRALIA Real Bilateral
Exchange Rate 39.9 46.5 38.9 14.5 9.6 9.2 38.9 28.1 26.8
Total Exports 0.7 2.5 3.4 -1.6 0.7 -1.7 0.2 1.1 0.1
Total Imports -0.6 -0.1 0.0 -1.4 0.7 -2.6 -0.3 -0.4 -0.6
Trade Balance 1.3 1.0 0.9 1.1 -0.6 -5.2 0.5 0.9 1.0
NEW ZEALAND Real Bilateral
Exchange Rate 72.9 85.9 70.5 11.1 1.1 4.3 34.6 13.9 10.4
Total Exports 0.1 2.0 2.4 -1.7 -6.7 -0.4 0.0 0.6 1.3
Total Imports -0.2 0.2 0.4 -1.0 5.0 4.2 -0.2 -0.6 -1.0
Trade Balance 0.2 0.3 0.1 0.8 -8.3 -5.4 0.2 0.8 1.3
USA Real Bilateral
Exchange Rate 23.5 18.4 17.9 23.2 35.5 39.4 5.1 0.1 -3.8
Total Exports 1.4 0.6 2.6 -0.8 0.8 1.3 -4.8 -176.8 4.8
Total Imports -0.1 1.3 5.7 2.3 -0.8 -1.0 -9.4 -76.0 0.2
Trade Balance 2.7 -4.6 -2588.0 -6.4 32.4 15.6 22.3 -620.2 12.3
UK Real Effective Exchange Rate 17.6 40.4 43.0 17.6 15.8 16.5 12.8 10.6 9.0
Total Exports 1.1 1.0 1.7 -3.6 -2.2 -1.3 -1.1 -3.3 -3.7
Total Imports -0.3 -0.2 -0.2 -0.8 -2.6 -2.8 -4.5 -1.9 -1.8
Trade Balance 1.4 1.4 2.4 -4.2 -2.1 -1.0 -0.5 -3.5 -3.9
JAPAN Real Bilateral
Exchange Rate 54.4 65.3 71.6 38.2 34.4 19.9 1.7 18.7 10.6
Total Exports 2.1 3.5 4.4 -0.9 -1.2 -1.4 17.5 2.7 5.2
Total Imports -0.5 -0.2 -0.1 -0.7 -0.3 -1.1 -5.9 -1.3 -3.1
Trade Balance 0.7 0.5 0.5 0.4 -1.1 0.8 89.6 9.1 14.8
Source: Author’s calculation
144
The most favourable impact on the competitiveness has been noted for the 1987
devaluation. This is because the gain in competitiveness has been higher than the rate of
devaluation in almost all the cases except with the USA. While comparing the 1998 and
2009 devaluation, the results show favourable gains in the real bilateral exchange rates
with Australia and New Zealand after the 2009 devaluation than the 1998 devaluation.
However, for Japan and the USA, there is greater favourable impact after the 1998
devaluation. On the contrary, the real bilateral exchange rate between Fiji and the USA
experienced currency appreciation few periods after the 2009 devaluation. It is also
interesting to note that the devaluation in 2009 appears to have had less impact than the
other devaluation episodes in improving international competitiveness in the economy.
Further analysis shows long–run improvement in the trade performance of Australia,
New Zealand and Japan after the 1987 and 2009 devaluations. The results show that this
is largely as a result of corresponding increase in exports and modest decline in imports.
However, the trade balances for these countries responded otherwise to the 1998
devaluation. For the USA, decline in the trade balances is noted after the 1987
devaluation while a modest improvement is noted after the 1998 and 2009 devaluations.
While most of the major trade partners experienced mixed responses on their trade
components, imports from Australia, Japan and the UK experienced decline after all the
different devaluation episodes at different time periods.
Fiji has also established and maintained its trade relations with some of the emerging
Asian trade partner countries. These emerging Asian counterparts as described earlier
include China, Hong Kong, India, Malaysia and Singapore. Based on the analysis, these
countries have experienced mixed results on their trade relations with Fiji after each
devaluation episodes except for China (Table 5.5). Fiji’s trade balance with China
experienced long–run continuous decline largely due to increase in imports. In
particular, after the 1987 devaluation, the trade balance with these countries experienced
deterioration, except for India and Malaysia. This impact has been largely due to heavy
dependence of importable commodities from these countries that results in an increase in
145
imports. Exports, on the other hand, after the 1998 devaluation experienced a slight
increase with Hong Kong and India.
Table 5.5 Response of bilateral trade performance with emerging Asian trade partner countries to 1% devaluation in Fiji
Devaluation Years
1987 1998 2009
1st 3rd 5th 1st 3rd 5th 1st 3rd 5th Real Effective Exchange Rate 17.6 40.4 43.0 17.6 15.8 16.5 12.8 10.6 9.0
SINGAPORE Total
Exports 2.6 0.2 -0.5 -4.3 -4.7 -1.4 -0.3 -2.0 -9.5
Total Imports 11.2 1.2 0.8 0.4 -0.1 0.2 -3.0 -1.6 0.1
Trade Balance -11.3 -1.2 -1.1 -0.7 -0.3 -0.3 3.0 1.6 -1.0
CHINA Total
Exports 5.6 2.3 0.4 82.9 -5.4 -5.1 20.4 16.8 22.7
Total Imports 0.3 1.0 1.4 -2.1 0.9 1.4 0.1 6.4 11.8
Trade Balance 6.9 -0.2 -2.6 2.3 -2.6 -2.5 0.4 -6.0 -11.3
MALAYSIA Total
Exports 2.2 0.3 0.9 -3.8 -5.9 -5.8 -1.4 17.6 58.4
Total Imports -1.4 2.3 4.3 -1.6 0.6 2.1 -0.7 1.6 4.0
Trade Balance 2.6 0.1 0.7 -2.6 -16.6 -12.8 0.7 -1.5 -3.7
INDIA Total
Exports 23.4 2.5 1.0 -0.6 4.2 4.6 -0.5 26.0 49.3
Total Imports -1.2 -0.4 0.1 -0.9 0.8 2.2 -4.3 -2.9 -1.9
Trade Balance 1.6 0.5 -0.1 0.9 -0.8 -2.2 4.4 3.2 2.5
HONG KONG Total
Exports 7.7 5.8 7.4 -1.2 1.9 5.9 2.0 5.3 6.2
Total Imports 0.1 2.2 2.9 -0.9 2.7 2.8 -1.3 0.0 1.1
Trade Balance 1.1 -1.7 -2.4 0.8 -2.9 -2.0 2.1 2.0 0.9
Source: Author’s calculation
146
The recent 2009 devaluation also shows mixed results as trade with countries such as
Hong Kong and Singapore experienced modest improvements. Import of mineral fuels
from Singapore has largely been the reason for increase in its imports and subsequent
decline in bilateral trade balance after the 1987 and 1998 devaluations. However,
imports after the 2009 devaluation declined resulting in modest improvement in trade
balance with Hong Kong, India and Singapore in the short–run.
Hence, the bilateral trade analysis reveals that following currency devaluation in Fiji,
trade balance performance with Australia, New Zealand, India, Japan and the UK seems
to be improving while it appears to be deteriorating with China, Hong Kong, Malaysia,
Singapore and the USA. However, these are just crude estimates and care needs to be
taken in their interpretation. A more detailed econometric analysis is carried out in the
next section to evaluate the effectiveness of devaluation at sectoral and bilateral level.
5.3.3 Highlights from sectoral and bilateral trade patterns
The preliminary analysis indicates that there are mixed responses of currency
devaluation on sectoral and bilateral levels of trade in Fiji. In particular, the 1987
devaluation appears to have resulted in an increase in both export and import of
commodities. However, some sectoral imports have declined after the 2009 devaluation.
Travel and transportation services sectors also appear to have experienced subsequent
decline after the 1987 and 1998 devaluation. However, a modest improvement is noted
for the 2009 devaluation. On the bilateral front, Fiji’s trade has largely benefited with
major trade partners while imports from emerging Asian trade partners continued to
increase.
Hence, analysis in this section has allowed some preliminary understanding on the
effects of devaluation on some of the important trade sectors in the economy. Therefore,
the next section employs econometric techniques to analyse the impact of devaluation on
sectoral and bilateral level trade performance in Fiji over the last three decades.
147
5.4 Empirical analysis: Sectoral and Bilateral trade
In this section, empirical estimation results on the impact of currency devaluation on the
sectoral and bilateral trade performance in Fiji are presented. This analysis is also
accompanied by test on the presence of the J-curve phenomenon at the two levels of
trade in the economy. For the analysis in this section, the same trade models as
developed for the aggregate trade analysis are used.
On the sectoral front, the inclusion of major export and import sectors for analysis are
identified based on the availability of a consistent dataset over a relatively long period of
time. However, due to the unavailability of consistent time series data on both exports
and imports for various sectors, analysing on the J-curve phenomenon has been
constrained to only three sectors. These are the food, travel services and transportation
services sectors. However, these three sectors approximately contribute around 43% of
total trade in 2012.
Nonetheless, included in the analysis are also other sectors for which at least exports or
imports trade data were available. These exportable sectors are for food, sugar, fish,
gold, travel services and transportation services. On the other hand, the importable
sectors include food, fuel, manufactured goods, crude oil, textile, machinery and
transport equipment, tobacco and beverages, chemicals, oil and fats, miscellaneous
manufactured goods, travel services and transportation services. Hence, for these
sectors, their respective export and import equations have been estimated. It is argued
that this particular empirical analysis will help to identify the response of major trade
sectors to devaluation in Fiji. It is also ascertained to help identify the direction set by
devaluation on the aggregate trade performance in the economy.
For the bilateral trade analysis, as discussed earlier, trading partners have been grouped
into ‘Major Trade Partners’ and ‘Emerging Asian Trade Partners’ and analysed over the
1975 to 2012 period. On the overall, trade with these countries account for
approximately three quarters of Fiji’s total trade. Utilising the trade balance model as
presented in section 2.4 and 2.5, the same models are used to carry out sectoral and
148
bilateral level trade analysis. In particular, to estimate for the impact of devaluation at
the sectoral trade analysis, real effective exchange rate is used as a proxy. This is
because the sectoral export and import data captures Fiji’s sectoral trade with the rest of
the world. However, for the bilateral trade models, bilateral real exchange rates and the
relevant trading partner country’s GDP are used for the analysis.
Hence, both the sectoral and bilateral trade balance equations are modelled as a function
of real exchange rate, domestic real income and trading partner real income as below:
),,( ,, jfjrj YYEfTB � (5.1)
where jTB represents trade balance at either sectoral or bilateral level. For sectoral trade
balance models, j represents the trade balance for food, travel services and
transportation services sectors. For bilateral trade balance models, j represents trade
balance for countries that are classified as ‘Major Trade Partners’ and ‘Emerging Asian
Trade Partners’. The trade balance is again expressed as the ratio of Fiji’s sectoral
exports over her total sectoral imports or its exports to its trading partner over her
imports from the same trading partner country. Y represents real GDP for Fiji while jfY , is a trade-weighted real income of trading partners’ for sectoral analysis or the
respective trading partners real income in case of bilateral trade analysis. jrE , for
bilateral analysis represents the bilateral real exchange rate between Fiji and its trading
partner country.
Additionally, to further analyse the sectoral and bilateral trade balance relationship,
respective export and import demand equations are modelled as follows:
),( ,, jfjrj YEfEXP � (5.2)
),( , YEfIMP jrj � (5.3)
where for sectoral trade, EXP and IMP represents exports and imports of sectors
which have been discussed earlier. The bilateral export and import equations represent
total goods export and imports, respectively between Fiji and its trading partner
149
countries, j . Similarly, all the other variables employed are the same as discussed
earlier for the models.
Since the annual time series data and VECM are used for analysis, the empirical tests
begin from testing the unit-root properties of the variables followed by cointegration
tests. Henceforth, the short and long–run relationships among the variables in the model
are established. Towards the end, results on the sectoral and bilateral J-curve
phenomenon are presented.
5.4.1 Results of the unit root tests
Using the method described in Section 2.6.1, the unit root test results suggest that all the
variables employed in the various sectoral and bilateral trade balance, export and import
models are integrated of order one, that is they are I(1) in nature. To conserve space, the
detailed test results are presented in Appendix D.
5.4.2 Results of the cointegration tests
Since all the variables to be employed in the model are integrated of order 1, the
cointegration tests using the Maximum-Eigenvalue method reveals at least one
cointegrating equation in every model. This suggests the presence of co-movements
among the variables and indicating long–run stationarity in the models. To conserve
space, the detailed results from the Maximum-Eigenvalue method are presented in
Appendix E.
5.4.3 Long–run elasticities
Since the long–run relationship is established by the cointegration tests, the long–run
relationship in the sectoral and bilateral trade models are then estimated using the
VECM. The results from the sectoral models are presented first followed by the results
on the bilateral models.
150
5.4.3.1 Sectoral trade analysis
As part of the sectoral trade analysis, three sectoral trade balance, six exportable sectors
and twelve sectoral imports are examined. The long–run coefficients of the sectoral trade
balance model are reported in Table 5.6 followed by sectoral exports and imports results
in Table 5.7 and 5.8, respectively.
Table 5.6 Estimates of long–run coefficients of major sectoral trade balance models Trade Sectors rE Y fY Constant
FOOD -3.880 (1.129)***
1.861 (1.702)
-1.462 (1.374)
22.103
TRAVEL -3.732 (0.144)***
1.630 (0.145)***
0.519 (0.127)***
-16.537
TRANSPORT -1.192 (0.670)*
-0.851 (1.049)
1.950 (0.826)***
-24.064
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
The main variable of interest, rE is noted to be significant for all the three sectors trade
balance. A closer look into the impact of devaluation on the three sectoral trade balance
reveals that none of the sectors experiences significant improvement. The empirical
results show that the performance of the food, travel services and transportation services
trade balance significantly worsens after currency devaluation in the long–run in Fiji.
The impact of devaluation on the food sector appears to be the highest. The regression
results suggest that, ceteris paribus, a 1% devaluation worsens the food trade balance by
approximately 3.9% in the long–run (Table 5.6). This is argued to be as a result of a
greater impact of devaluation on food imports than on its exports (Table 5.7 and 5.8).
Fiji, largely being an import dependent economy tends to import most of its food
products which as a result of devaluation is estimated to increase significantly. This
causes an ultimate negative impact on the overall food trade balance.
151
However, this result also suggests that the earlier estimated positive impact of
devaluation on the aggregate goods trade balance is not influenced by the food trade
sector but other sectors of the economy. It is argued that the positive impact on the
aggregate goods trade balance is largely as a result of the positive response of
devaluation on other exportable goods sectors. These exportable sectors include sugar,
fish and gold (Table 5.7). Moreover, the positive impact on the aggregate goods sector is
also as a result of the decline in the imports of textiles, tobacco and beverages (Table
5.8).
Similarly, with regard to the services sector trade, the estimation results suggest similar
response. The result on the travel and transportation sector trade balance suggests that
the performance of both the sectors significantly worsens due to devaluation in Fiji. The
regression coefficient implies that ceteris paribus, 1% devaluation in Fiji worsens the
trade balance in travel sector by 3.7% and transportation sector by 1.2% in the long–run.
It is argued that though the absolute tourism arrivals in the country have generally
increased in recent years, the sector’s real earnings have not kept up to pace31.
Additionally, most of the tourism related activities are highly import intensive such as
food and fuels for recreational activities. This results in an adverse impact on the travel
services trade balance.
Furthermore, results from the travel and transport sector export and import models
provide additional empirical support. The results on the transport sector suggest that the
increase in exports is relatively less than the increase in imports in the long run.
However, for the travel sector, the results suggest that services exports experience
significant decline while its imports experiences significant increase after devaluation
(Table 5.7). As discussed earlier in the empirical analysis in Chapter 4, these results are
supported by the findings of Culiuc (2014). The study finds that tourism in the Pacific
economies is less sensitive to changes in the country’s real exchange rate. The study
argues that this is as a result of higher import content in food products offered to tourists
rather than the locally produced food. They also add that since most of the tourists to the 31 See Figure 5.7.
152
Pacific countries depend on packaged vacations for which the prices are usually set in
foreign currency, the tourists often do not directly benefit from these real exchange rate
movements.
Table 5.7 Estimates of long–run coefficients of major services sector export and import models
Service Sectors rE Y fY Constant Services Export
TRAVEL -2.008 (0.681)***
2.454 (0.413)***
-30.778
TRANSPORT 1.331 (0.444)***
1.900 (0.277)***
-35.083
Services Import TRAVEL 5.998
(0.898)*** -2.868
(0.769)*** 26.065
TRANSPORT 1.819
(0.748)** 1.414
(0.592)** -17.174
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
This study also argues in the same direction. Since most of the holiday and tour
packages offered by hotels and resorts in Fiji are often discounted32, this coupled with
devaluation results in an eventual decline in the real tourism receipts. Additionally,
service sector imports such as travelling overseas for education, medical reasons, among
others after devaluation appears to have not significantly declined. This results in an
increase in total service import bill that has adverse impact on the services trade balance
in the long–run.
Nonetheless, this result is consistent with our earlier findings on the aggregate service
trade balance. The transportation sector is also estimated to be negatively impacted by
devaluation at least at 10% level of significance. It is also estimated that the overall
deterioration of the aggregate service trade balance is largely as a result of the adverse
32 The 2009 Fiji Government Budget Supplement (page 30) states that hotels and resorts in Fiji largely discount their holidays and tour packages to attract tourist market from Australia and New Zealand.
153
impact of devaluation on the travel sector trade balance. This is because trade of travel
services represents close to 50% of overall services trade in Fiji.
The impact of domestic income levels on the sectoral trade performances reveals that it
is only significant for the trade in travel services sector. For the other two sectors (food
and transportation services), the impact is found to be insignificant. The empirical
results suggest that, ceteris paribus 1% increase in Fiji’s GDP leads to significant
improvement in the trade balance of travel services by 1.6% in the long–run. It is also
noted that the impact of the domestic income on the travel services trade balance
corresponds well to the earlier estimated impact at the aggregate services trade balance.
Further, the regression results on the impact of trading partner income on these three
sectoral trade balance shows that it is significant only for the services sector trade.
However, for the food sector, the impact is insignificant. The travel and transportation
services sector show that a 10% rise in trading partner income, ceteris paribus, results in
a significant improvement in its trade balance by 5.2% and 19.5%, respectively. It is
argued that the positive contribution of the trading partner income on these sectoral trade
performances is derived from its positive contribution on the export components of these
sectors (Table 5.7).
The results in Table 5.7 show that the export in both the travel and transportation
services sector experience significant improvement in trade due to increase in trading
partner incomes. This result, in particular implies that increasing income level in trading
partner countries such as Australia, New Zealand, Japan, the US and the UK leads to
subsequent increase in demand for Fiji’s travel exports, which largely includes tourism.
This result is also consistent with our earlier findings that a rise in trading partner
incomes significantly improves services trade performance in the long–run in Fiji.
On the other hand, the goods sector which is represented by food sector indicates
significant and negative impact of the trading partner income on its trade balance. This
suggests that higher income level in these trading partner countries do not result in
154
increased demand for Fiji’s exportable food products. Hence, it is argued that this is
possibly due to the nature of Fiji’s goods exports which are largely agricultural products
to these trading partner economies. Therefore, with rising income in these countries, it is
argued that these countries tend to reduce consumption of traditional agricultural food
products from Fiji and switch to other cheaper trading partner countries. Nevertheless,
the result is consistent with our earlier finding on the adverse of trading partner income
on the aggregate goods trade balance performance.
Having analysed the impact of devaluation on the sectoral trade balance performance, in
what follows, other important exportable and importable commodities in Fiji are
analysed in Table 5.8. Since for these sectors, either the export or the import data is only
available, the models are analysed depending on data availability. The empirical results
on the export sectors estimate significant and positive impact of devaluation only on the
export of sugar. However, the other exportable sectors, which include fish and gold,
indicate a positive but insignificant impact of devaluation.
For export of sugar, the results show that 1% currency devaluation, ceteris paribus,
leads to a significant increase in sugar exports by around 6% in the long–run. However,
the challenge for this sector to derive gains from devaluation remains in its continued
effort to improve its productivity and export capacity to meet the increased demand.
With regard to the exports of fish, it is noted that devaluation does not have a long–run
significant impact on its export. It is argued that this is essentially because large amount
of fish exports from Fiji go to destinations such as China, Japan and the Euro area under
the interim EPA, for which prices have already been negotiated.
Moreover, for the importable commodities, it is noted that currency devaluation in Fiji
has a significant impact on almost all of the importable commodities. Out of these,
devaluation is noted to cause a significant increase in imports of food, fuel,
manufactured goods, crude oils, chemical, miscellaneous manufactured goods, oil and
fats along with machinery and transport equipment. On the other hand, it is found to
cause reduction in imports of textiles, tobacco and beverage.
155
Table 5.8 Estimates of the long–run coefficients of goods sector export and import models
Goods Sectors rE Y fY Constant
Goods Export
FOOD 0.980 (1.986)
-2.997 (1.261)**
72.465
SUGAR 5.944 (1.255)***
-4.226 (0.732)***
75.672
FISH 0.926 (0.655)
0.556 (0.386)
-5.536
GOLD 6.473 (6.698)
4.055 (4.367)
-105.746
Goods Import
FOOD 3.756 (1.214)***
-0.453 (1.114)
1.650
FUEL 5.023 (1.943)**
7.465 (1.739)***
-74.609
EDMANUFACTUR GOODS
3.519 (0.751)***
-1.897 (0.651)***
24.208
CRUDE OIL 1.296 (0.629)**
0.227 (0.529)
-0.154
TEXTILE -9.026 (3.907)**
5.609 (3.509)
-30.076
MACHINERY AND TRANSPORT
EQUIPMENT
1.486 (0.294)***
0.422 (0.247)*
-0.493
TOBACCO AND BEVERAGE
-2.803 (2.371)
3.936 (1.763)**
-36.011
CHEMICAL 0.685 (0.295)**
0.679 (0.248)***
-1.732
OIL AND FATS 2.217 (1.095)*
-1.343 (1.023)
19.082
OUSMISCELLANE EDMANUFACTUR
GOODS
1.309 (0.415)***
5.055 (1.015)***
0.094
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
156
Fiji, which is an import dependent economy for most of its raw materials, intermediate
and finished goods experiences significant increase in import bills as a result of price
rise after devaluation. This is largely because demand for Fiji’s imports is relatively
inelastic even though corresponding prices have increased. It is also important to note
that this result well complements our earlier findings estimating similar significant and
positive impact on aggregate goods imports in the long–run.
Analysing the impact of Fiji’s GDP on its sectoral imports reveals its significant impact
on the imports of fuel, manufactured goods, chemical, miscellaneous manufactured
goods, tobacco and beverage and machinery and transport equipment in the long–run.
As such, except for the import of manufactured goods, the other import sectors respond
positively to domestic income. This result is not surprising. The rise in income leads to
increase in economic activity, which causes increase in demand for fuel, manufactured
goods including the purchase of machinery and transport equipment to meet rising
demand and keep pace with growth performance. The positive impact on the other
importable commodities re-affirms the earlier notion that most of these goods that are
imported do not have immediate substitution in the domestic market. Hence, these
commodities continue to be imported in almost same quantities as prior to devaluation.
Hence, the varying impact of currency devaluation on various sectors helps to explain
how different sectors respond to devaluation and the resultant impact on the overall
goods and services trade performance in the economy. The disaggregated results re-
affirm the argument that aggregation biasness exists at aggregate trade analysis and this
exercise on sectoral trade performance to some extent helps to mitigate this issue. By
disaggregating trade data into sectors, it has allowed to trace the response of various
sectors to currency devaluation that has had its ultimate effect at the aggregate level.
The results in this section, however, do not shed light on how the impact of devaluation
on this sectoral trade influences the bilateral trade performance in the economy. Hence,
in what follows, the long–run estimates from the bilateral trade models are presented and
discussed.
157
5.4.3.2 Bilateral trade analysis The long–run elasticities of the ‘Major Trade Partners’ are discussed and reported first in
Table 5.9 followed by the results of the ‘Emerging Asian Trade Partners’ in Table 5.10.
The variable of interest, rE , which captures the impact of devaluation, is noted to have a
significant relation with trade balance for nine out of the ten bilateral trade partner
countries analysed in this study. The only exception with an insignificant relation is with
Japan. Out of these significant relations, only the exchange rate relationships with New
Zealand and the USA are found to be significantly positive. This implies that
devaluation only has significant and favourable impact on the trade balance for these
two countries.
Table 5.9 Estimates of long–run coefficients of trade balance models with major bilateral trade partners
Trading Partners rE Y fY Constant AUSTRALIA
-2.225 (1.260)*
-11.163 (2.147)***
1.919 (5.524)
-0.223
NEW ZEALAND 0.710 (0.190)***
-4.308 (1.088)***
1.823 (0.845)**
25.360
JAPAN -1.230 (0.779)
0.488 (1.780)
4.809 (1.589)***
-108.513
UNITED STATES
5.873 (0.902)***
-0.273 (2.504)
-0.007 (-1.771)
-22.334
UNITED KINGDOM -11.967 (2.615)***
5.497 (3.460)
4.810 (3.699)
-127.031
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
A closer look into the long–run coefficient reveals that out of the five major trading
partner countries, trade with the USA benefits the most from devaluation in Fiji. The
results suggest that, ceteris paribus, a 1% currency devaluation significantly improves
Fiji’s trade position with the USA by 6% in the long–run (Table 5.9). This is argued to
be as a result of greater influence of devaluation on Fiji’s exports to the USA than on its
bilateral imports (Table 5.10).
158
The results also suggests that Fiji’s traditional exports to the USA which primarily
consist of mineral and aerated bottled water, tuna, sugar, garments, lumber and
mahogany, appear to have benefitted in the midst of currency devaluation in Fiji.
However, there is also evidence that imports from the USA, which primarily include
petroleum bitumen used in road construction, medicaments and food products
experience modest increase. As modelled earlier for sectoral export analysis, the
significant response of devaluation for sugar exports suggests that it is one cause of a
significant and favourable impact for Fiji’s trade with the USA.
Additionally, trade with New Zealand also appears to be improving significantly as a
result of devaluation. The results suggest that ceteris paribus, a 1% devaluation in Fiji
significantly improves Fiji’s trade position with New Zealand by approximately 0.7% in
the long–run (Table 5.9). Additionally, Fiji experiences increases in exports and imports
from New Zealand in the long–run as a result of devaluation (Table 5.10). However, the
magnitude of the impact on the exports outweighs its influence on imports. This results
in the overall favourable impact on the bilateral trade balance in the long–run. Major
commodity exports to New Zealand such as sugar, garments, fish and other food
products experiences favourable increase after devaluation in Fiji. As evidenced earlier,
the favourable response of devaluation to sugar and fish exports is one reason for the
favourable impact for Fiji’s trade with New Zealand.
On the other hand, Fiji’s trade with Australia experiences deterioration in the bilateral
trade balance at least at the 10% level of significance. The results suggest that, ceteris
paribus, a 1% devaluation in Fiji reduces Fiji’s trade balance position with Australia by
2.2% in the long–run (Table 5.9). The results from the export model suggest that exports
from Fiji to Australia, which include gold, garments and food products, increases
significantly while imports from Australia to Fiji are not affected as a result of
devaluation (Table 5.10). This implies that changes in the real bilateral exchange rate do
not significantly alter import volumes from Australia to Fiji.
159
Table 5.10 Estimates of long–run coefficients of export and import models with major bilateral trade partners
Trading Partners rE Y fY Constant
Bilateral Exports AUSTRALIA 3.397
(1.494)** 0.116
(0.727) -5.968
NEW ZEALAND 0.811
(0.280)*** -0.468
(0.289) 15.570
JAPAN 0.331
(0.373) 1.782
(0.552)*** -30.601
UNITED STATES
5.215
(1.113)*** 0.364
(0.460) -20.868
UNITED KINGDOM -0.878
(0.765) -0.486
(0.516) 26.001
Bilateral Imports
AUSTRALIA -0.100 (0.424)
1.162 (0.283)***
-3.822
NEW ZEALAND 0.689 (0.149)***
0.757 (0.180)***
-2.027
JAPAN 0.372 (0.196)*
-2.325 (0.313)***
43.656
USA 1.858 (1.153)
0.613 (0.647)
-6.440
UNITED KINGDOM 1.893 (0.642)***
-4.491 (0.552)***
67.225
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
Hence, due to the inelastic nature of Fiji’s import demand from Australia, the continued
demand for imports such as food products, manufactured goods, chemicals, oil and fats
among others, results in an adverse impact on the bilateral trade balance with Australia.
The finding that devaluation worsens Fiji’s trade performance with Australia is also
consistent with the earlier findings of Kaufman (2008). Kaufman (2008) estimates that
the M-L condition does not hold true for Fiji’s trade performance with Australia. This
160
consequently implies that devaluation worsens the bilateral trade balance between the
two countries.
Similarly, it is found that devaluation significantly worsens Fiji’s bilateral trade balance
with the UK in the long–run. Fiji’s major export to the UK has been sugar, for which Fiji
for a long period of time has enjoyed preferential access to the EU markets and the UK.
Hence, it is argued that due to the agreed export prices of sugar to the UK, changes in
the exchange rate have not been beneficial. In fact, currency devaluation has not resulted
in a significant increase in sugar exports to the UK. However, this result also lends
support to the findings of Kaufman (2008). Study by Kaufman (2008) also finds an
insignificant impact of exchange rate on the Fiji–UK export relationship. Nonetheless,
imports such as chemicals and food products from the UK, experience significant
increases. This causes an ultimate adverse impact on the bilateral trade position.
However, this result is consistent with the sectoral level import analysis.
To the contrary, though, bilateral trade performance with Japan is found to be unaffected
by currency devaluation in Fiji (Table 5.9). This is attributable to the insignificant
increase in bilateral exports accompanied by a 10% significant increase in imports in the
long–run (Table 5.10). It is ascertained that since Fiji relies on imports of manufactured
goods, machinery and transport equipment from Japan, importation of these products
continue at a similar pace as before devaluation. This is because of its substantial use in
the domestic production processes and its lack of domestic substitutability.
Moreover, the empirical results of the trading partner countries classified as ‘Emerging
Asian Trade Partners’ in this study are presented in Table 5.11 and 5.12. Fiji’s trade with
emerging Asian economies such as with Singapore, China, Malaysia, India and Hong
Kong has been gaining increased attention over the last decade or so. Given the interest
of this study, the empirical results for this set of countries reveal that currency
devaluation in all the cases has an unfavourable impact on Fiji’s bilateral trade balance
with these countries (Table 5.11).
161
Table 5.11 Estimates of long–run coefficients of trade balance model with emerging Asian trade partners
Asian Trade Partners rE Y fY Constant
SINGAPORE -4.075 (1.937)**
6.877 (4.254)
-2.062 (1.428)
-48.538
CHINA -13.971 (5.320)**
2.370 (11.344)
1.383 (2.811)
-2.963
MALAYSIA -11.378 (3.473)***
-24.106 (9.179)**
6.704 (3.197)**
282.403
INDIA -11.718 (2.637)***
-9.377 (5.236)*
9.740 (2.156)***
-8.635
KONGHONG -7.984 (2.096)***
-8.284 (5.621)
-0.062 (2.558)
0.287
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
The results imply that Fiji’s exports to these Asian countries which largely involve
export of fish and other food products do not respond favourably to devaluation.
However, imports from these countries which are largely dominated by mineral fuel,
food products, manufactured goods, machinery and transport equipment, shows strong
positive association to devaluation (Table 5.12). This is largely because these importable
commodities are important components in local domestic production and consumption
processes. However, domestic substitution for these commodities is weak. Hence, this
results in continued import of these commodities despite devaluation in Fiji.
Henceforth, next discussed is the impact of domestic and foreign income on Fiji’s
bilateral trade performance. The impact of Fiji’s domestic income by and large has been
found to be negative and on most instances to be significantly influencing the bilateral
trade flows. This conventional argument is valid for almost half of the countries in the
study. This includes trade with Australia, New Zealand, Malaysia and India. The results
suggests that for a 1% increase in Fiji’s real GDP, ceteris paribus, Fiji’s bilateral trade
balance significantly decreases with Australia by 11%, New Zealand by 4%, Malaysia
by 24% and India by around 9% in the long–run (Table 5.11).
162
The results for these four trade partners suggest that a rise in Fiji’s GDP causes an
increase in demand for imports from these countries. Hence, this causes an adverse
impact on its bilateral trade balance. These results are also supported from each
country’s bilateral import demand models of Fiji (Table 5.12). Since Fiji’s GDP exhibits
negative and significant trade relationships for almost half of the trade partners, this also
provides support to our earlier similar findings on the aggregate goods trade balance. For
other trade partners, Fiji’s GDP does not have a significant impact on its bilateral trade
performance.
Table 5.12 Estimates of long–run coefficients of export and import models with emerging Asian trade partners
Asian Trade Partners rE Y fY Constant Bilateral Exports
SINGAPORE -4.432 (3.612)
1.657 (0.910)*
-1.303
CHINA 1.067 (4.134)
0.870 (0.613)
-14.946
MALAYSIA -17.240 (5.284)***
2.811 (1.529)*
35.083
INDIA -11.122 (1.812)***
6.078 (0.711)***
-66.056
KONGHONG 2.557 (0.669)***
-1.619 (0.901)*
0.144
Bilateral Imports SINGAPORE 16.304
(8.701)* -5.018 (7.607)
11.178
CHINA 4.787 (0.884)***
-1.122 (0.759)
5.238
MALAYSIA 8.152 (0.985)***
-1.221 (0.812)
-9.828
INDIA -1.679 (0.493)***
4.013 (0.376)***
-41.717
KONGHONG 4.403 (0.581)***
2.508 (1.373)*
-0.076
Notes: 1. Standard errors are given in parentheses. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
163
Furthermore, the long–run estimates of the trading partner income suggest that four of
the ten trading partner countries experience significant favourable relationship of the
bilateral trade performance with Fiji. These trading partners include New Zealand,
Japan, Malaysia and India. The results suggest that increase in income in these countries
translate into increase in Fiji’s exports to these countries. However, the income of other
trade partners does not significantly affect Fiji’s trade performance in the long–run. This
could be particularly true when Fiji’s exports are not among the most preferred
consumed goods in these countries and there are substitutes to Fiji’s exportable products.
The results are also supported by each country’s bilateral export demand models. It is
argued that since there is insignificant impact for most of the bilateral trade models, the
weak significant impact of foreign income on the goods sector and the insignificant
impact on the aggregate goods and services trade balance reinforce our earlier results.
Therefore, based on the empirical evidence it is established that currency devaluation is
not an effective tool to boost sectoral trade balance performance of food, travel and
transportation services sector in the long–run in Fiji. However, the impact of devaluation
on other exportable sectors such as on sugar and fish is favourable. Nonetheless, heavy
reliance on imports of fuel products, machinery, equipment, chemicals and
manufactured goods results in subsequent increases in the import bill due to devaluation
in Fiji. Additionally, at the bilateral front, the results suggest an unfavourable impact of
devaluation on bilateral trade balances with trading partners that supply significant
amounts of imports to Fiji.
The long–run results in this section, however, do not shed any light on the short–run
dynamics of the variables in the trade models in the study. Hence, in the next section,
short–run impacts of Fiji’s currency devaluation on the various sectoral and bilateral
trade components are discussed.
164
5.4.4 Short–run elasticities: Results from the error correction models
In this section, short–run impact of exchange rate on the sectoral and bilateral trade
balance models are presented and discussed. Hence, in order to test for the presence of
the J-curve phenomenon at both the levels of trade, short–run coefficient estimates of
only the real exchange rate is being discussed33. The results from the sectoral models are
presented first followed by those on the bilateral front.
5.4.4.1 Sectoral trade analysis
As part of the sector trade analysis in the short–run, the coefficients of the main variable
of interest; the real exchange rate is only discussed. This is to analyse and understand the
impact of currency devaluation in the short–run in Fiji.
The sectoral trade balance largely shows that currency devaluation has positive impact
on the three sectoral trade performances in the short–run in Fiji (Table 5.13). The food
sector, in particular shows a positive and significant impact at lag 1 whereas the
transportation services sector finds significant improvement at lag 3. However, for the
travel trade sector, significant improvement is noted from lags 1 to lag 4 but becomes
insignificant at lag 5. This indicates that at least in the short–run these three sectors
experience improvement in the trade balance as a result of devaluation in Fiji.
The impact on the trade balance is further supported by the evidence from these sectors’
export and import models34. The results from the export and import models indicate that
these sectors experience improvement in exports, which is complemented by decline in
imports in response to devaluation in the short–run. The positive short–run impacts on
travel and transportation services sectors are also supported by the earlier findings of
positive short–run impacts on services export and trade performances. This result,
however, is in contrast to the long–run results. However, this is not surprising in the
short–run, particularly when devaluation immediately gives an incentive to tourists to
33 Appendix F contains the entire VECM results for all the sectoral and bilateral equations being estimated. These include the export, import and trade balance models. 34 See Appendix F for detailed short–run results.
165
visit Fiji when the Fijian currency becomes weak. It also creates a disincentive for locals
to travel foreign countries as the cost of travel immediately rises. This, however, is not
expected to be the case indefinitely for a very long period of time.
Table 5.13 Short–run coefficient estimates of real exchange rates in the sectoral trade balance models
Short–run results Food sector
Travel services sector
Transportation services sector
rtE 1ln �� 1.599
(0.725)** 3.663
(2.011)* 0.868 (0.570)
rtE 2ln �� 0.405
(0.693) 7.444
(2.483)*** 0.998 (0.564)
rtE 3ln �� -0.268
(0.639) 6.174
(1.618)*** 1.362
(0.506)** rtE 4ln �� 2.994
(1.390)**
rtE 5ln �� 0.868
(1.251)
COUP -0.076 (0.091)
-0.447 (0.188)**
-0.097 (0.076)
Diagnostics
1�tECT -0.190 (0.078)**
-1.847 (0.596)***
-0.385 (0.178)**
2R 0.535 0.834 0.700 Adjusted 2R 0.101 0.429 0.399
! 0.136 0.220 0.127 NX 2
0.477 [0.788]
0.698 [0.730]
4.178 [0.124]
HetX 2 29.500
[0.337] 28.825 [0.369]
28.189 [0.401]
LM )(SCTest 28.268 [0.928]
11.010 [0.809]
16.714 [0.404]
AR roots graph stable stable stable Notes: 1. Standard errors are given in parentheses while p values are in brackets. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively. 3. ECTt-1 represents the error correction terms;! is the standard error of equation; diagnostics are
Jarque-Bera statistics for normality (X2N) and chi-squared for heteroskedasticity tests (X2Het), and LM Test statistics for serial correlation (SC) while AR roots graph tests for model stability.
166
As far as the other exportable sectors are concerned, the results show signs of significant
short–run improvements in the exports of gold while the export of sugar responds
otherwise. However, the results also indicate that the export of fish is not significantly
influenced by the exchange rate in the short to medium run. Since the exportable prices
for sugar are pre-determined under the trade agreements, devaluing the dollar translates
to reduced earnings. On the import front, the results show that devaluation in the short–
run leads to significant increases in imports of textiles, miscellaneous manufactured
goods, fuel, beverage and tobacco.
On the other hand, it causes significant decline in the imports of manufacturing goods,
oil and fats and machinery and transport equipment in the short–run. The other
importable commodities do not establish significant relationship in the short–run. The
significant impact of devaluation on import of fuel and textiles which are part of the
intermediate goods, are as expected. This is because these constitute a major component
of domestic production processes for which imports continue at the same pace in Fiji
despite the currency devaluation.
Additionally, all the sectoral models also incorporate the impact of political instability in
the economy. The results show that the occurrence of political instability causes
deterioration in the sectoral trade balances. However, the result is significantly negative
only for the travel sector. This indicates that the tourism services sector which is a major
component of the travel sector has been adversely affected by a series of coups in the
country. This ultimately causes decline in tourism earnings. The empirical evidence
suggests that as a result of political instability, visitor arrivals and tourism spending
declines in the country. This is also often coupled with an increase in overseas migration
due to political instability in the country. Other sectors that also experience negative and
significant impact are the transportation services sector, and gold exports along with the
import of textiles, miscellaneous manufactured goods, fuel, tobacco and beverage.
A number of diagnostic tests are also applied to the various models employed in the
study to ensure model appropriateness and stability. It is found that all the models pass
167
diagnostic tests including the tests of autocorrelation, normality and heteroskedasticity
along with tests for model stability. The R-squared showing the goodness of fit is also
relatively strong for the sectoral trade balance models ranging from 54% to 83% while
the adjusted R-squared is also deemed to be strong. The other related sectoral export and
import models are also well specified with reasonable goodness of fit. All the models are
also statistically well behaved. The error-correction term, which measures the speed of
adjustment to restore equilibrium, has negative sign and is statistically significant in all
the cases.
5.4.4.2 Bilateral trade analysis
As part of the bilateral trade analysis in the short–run, the coefficients of the main
variable of interest; the real exchange rate is only discussed. This is to analyse and
understand the short–run impact of the currency devaluation on the bilateral trade
balance models. Hence, the short–run results for the ‘Major Trade Partners’ are
presented in Table 5.14 followed by the results for the ‘Emerging Asian Trade Partners’
in Table 5.15.
The short–run impact of Fiji’s currency devaluation suggests that devaluation in Fiji
improves the bilateral trade position with Australia, Japan, the UK, Singapore, Malaysia
and Hong Kong at all lag lengths. Trade with New Zealand and the USA, on the other
hand, experiences short–run deterioration while trade with India and China experience
mixed responses. A closer look at the short–run elasticities for the trade partners reveal
that currency devaluation has insignificant relation at all the lags with New Zealand, the
USA, Singapore, Malaysia and China. Trade balance with Australia, on the other hand,
is estimated to improve significantly as a result of the decline in imports from Australia
at least in the short–run. Nevertheless, as the long–run results suggest, this does not last
for a long period of time. The inelastic nature of imports from Australia tends to
dominate in the long–run. India, Japan, Hong Kong and the UK also show similar
favourable impact of devaluation on its trade balance in the short–run.
168
Table 5.14 Short–run coefficient estimates of real exchange rates in the major trade partners’ trade balance models
Short–run results Australia New Zealand
Japan USA UK
rtE 1ln �� 2.115
(0.775)**-0.569(0.369)
0.745(0.595)
-1.527(0.943)
6.252(2.306)**
rtE 2ln �� 1.131
(0.784)-0.248(0.414)
1.932(0.592)***
5.854(2.351)**
rtE 3ln �� 0.648
(0.568)-0.135(0.301)
2.091(0.701)**
3.789(2.730)
rtE 4ln �� 9.161
(3.402)**
rtE 5ln �� 8.778
(3.575)**
COUP -0.251(0.112)
-0.009(0.082)
-0.236(0.188)
-0.415(0.167)**
0.073(0.277)
Diagnostics
1�tECT -0.937(0.295)***
-0.968(0.228)***
-0.899(0.211)***
-0.458(0.152)***
-0.604(0.183)***
2R 0.800 0.726 0.737 0.466 0.790
Adjusted 2R 0.646 0.524 0.544 0.356 0.278! 0.213 0.184 0.380 0.435 0.398
NX 2 0.214[0.899]
0.876[0.645]
0.036[0.982]
2.131[0.345]
3.441[0.179]
HetX 2 28.183[0.455]
32.380[0.218]
30.667[0.285]
11.395[0.411]
25.524[0.545]
LM )(SCTest 19.924[0.224]
21.120[0.174]
31.224[0.500]
10.850[0.819]
12.149[0.734]
AR roots graph stable stable stable stable stableNotes:1. Standard errors are given in parentheses while p values are in brackets.2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.3. ECTt-1 represents the error correction terms;! is the standard error of equation; diagnostics are
Jarque-Bera statistics for normality (X2N) and chi-squared for heteroskedasticity tests (X2Het), and LM Test statistics for serial correlation (SC) while AR roots graph tests for model stability.
169
Table 5.15 Short–run coefficient estimates of real exchange rates in emerging Asian trade partners’ trade balance models
Short–run results Singapore China Malaysia India Hong Kong
rtE 1ln �� 2.897
(4.475) 4.245 (8.110)
0.312 (3.394)
11.532 (3.644)***
7.414 (2.059)***
rtE 2ln �� 5.990
(4.471) -2.830 (8.571)
1.494 (3.059)
11.366 (4.606)**
4.172 (2.345)*
rtE 3ln �� -5.660
(8.850) 3.282
(3.799) 2.410 (1.984)
rtE 4ln �� -0.077
(3.420) 3.135
(1.758)* rtE 5ln �� 4.808
(2.838)*
coup -1.007 (0.720)
-1.142 (1.412)
-0.474 (0.538)
0.288 (0.603)
-0.087 (0.323)
Diagnostics
1�tECT -0.617 (0.326)*
-0.534 (0.228)**
-0.271 (0.139)*
-1.131 (0.223)***
-0.773 (0.279)***
2R 0.332 0.703 0.397 0.884 0.881
Adjusted 2R 0.053 0.484 0.145 0.601 0.717 ! 1.248 1.959 0.789 3.875 0.375
NX 2 1.217
[0.544] 1.908 [0.385]
0.367 [0.833]
0.009 [0.995]
0.317 [0.853]
HetX 2 17.149
[0.580] 29.079 [0.357]
22.417 [0.264]
27.203 [0.453]
13.690 [0.251]
LM )(SCTest 20.046 [0.218]
9.951 [0.869]
10.939 [0.813]
17.850 [0.333]
14.387 [0.570]
AR roots graph stable stable stable stable stable
Notes: 1. Standard errors are given in parentheses while p values are in brackets. 2. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively. 3. ECTt-1 represents the error correction terms;! is the standard error of equation; diagnostics are
Jarque-Bera statistics for normality (X2N) and chi-squared for heteroskedasticity tests (X2Het), and LM Test statistics for serial correlation (SC) while AR roots graph tests for model stability.
Additionally, in consistent with earlier findings, bilateral trade balance tends to
deteriorate as a results of political instability in the country. However, it is estimated to
be significant only in the case of bilateral trade balance with the USA. The diagnostic
tests shows that all the models pass tests of autocorrelation, normality and
170
heteroskedasticity along with the tests for model stability. The R-squared and the
adjusted R-squared is also relatively strong for all the bilateral trade balance models. All
the models are also statistically well behaved as the error-correction term has negative
sign and is statistically significant. This ensures that the series are non-explosive and
that long–run equilibrium is attainable.
5.4.5 Testing for the J-curve phenomenon
In this section, three methods (the traditional definition, new definition and the IRF
analysis) are used to analyse the presence of the J-curve phenomenon at the sectoral and
bilateral levels of trade.
5.4.5.1 Sectoral J-curve phenomenon
The J-curve phenomenon at sectoral level is analysed for the three sectoral trade
performances. These are for the sectoral trade of food, travel services and transportation
services. Based on the three methods of assessing the J-curve phenomenon, the results
are presented in a summarized form in Table 5.16. The analysis reveals very weak
evidence of the J-curve phenomenon at sectoral level of trade. In fact, the analysis shows
no evidence of the J-curve phenomenon using either traditional or new definition but
finds evidence of the phenomenon using IRF analysis, although in only one case.
Table 5.16 Results on the sectoral level J-curve phenomenon
Major Sectors The J-curve phenomenon using: Traditional definition
New definition
Impulse Response Function analysis
Goods Sector Food No No No
Services Sector Travel No No Yes
Transportation No No No
171
The goods trade sector which is represented by the food sector shows short–run positive
impact followed by long–run negative impact of devaluation on its trade balance. In
particular, using the traditional definition, the coefficients of the real exchange rate in
the food trade balance model are estimated to be positive at shorter lags followed by
negative impact in the long–run. Hence, the short–run improvement followed by
deterioration at longer lags rules out the presence of the J-curve phenomenon using the
traditional definition for the food sector.
Similarly, using the new definition, the results show a favourable impact in the short–run
followed by significant deterioration of the food trade balance in the long–run. Hence,
this method also rules out the presence of the J-curve phenomenon. Using the IRF
analysis, the graphical representation of the generalised one standard deviation shock in
the real exchange rate reveals that the food trade balance improves for the first two years
followed by deterioration at later periods (Figure 5.10). This again rules out the presence
of the J-curve phenomenon for the food sector in Fiji.
Figure 5.10 Response of the food sector trade balance to generalised one standard deviation innovation in the real exchange rate
-.06
-.04
-.02
.00
.02
.04
.06
.08
1 2 3 4 5 6 7 8 9 10
As argued for the aggregate goods sector, the non-presence of the J-curve phenomenon
for the food sector is largely attributed to the high dependence on importable food
products in the country. Even the modest increase in domestic exports causes an increase
in domestic income, which subsequently leads to an increase in imports for an import
Periods
172
dependent economy like, Fiji. This results in the deterioration of the food trade balance
in later periods. Hence, it is empirically evident that the J-curve phenomenon is not
strictly followed for the foods sector in Fiji.
However, this result is in contrast to the earlier findings on the aggregate goods sector.
The difference in the results is ascertained to be largely due to the fact that Fiji
consumes more of importable food products in which there is some short–run
substitution between local and importable food products. Nevertheless, this is not
necessarily the case for a very long time. Hence, it is argued that the aggregate J-curve
phenomenon at the goods sector is being supported by other goods sectors apart from the
food sector. The sectors contributing to the aggregate goods sector J-curve phenomenon
could possibly be gold, sugar, fisheries, textiles, clothing, tobacco and beverage, for
which specific trade balance equations could not be modelled due to insufficient data35.
Moreover, the travel sector reveals that the J-curve phenomenon is not valid using either
the traditional definition or the new definition. The estimates show short–run positive
effects followed by long–run negative impact of devaluation on the trade balance. This
effectively rules out the presence of the J-curve phenomenon using both the methods.
However, the IRF analysis gives some indication of the J-curve phenomenon (Figure
5.11). The impulse results show that due to a shock in the real exchange rate in the travel
sector model, the sectoral trade balance declines for the first two periods followed by
improvements at later stages. However, the trade balance momentarily declines between
period four to six. This creates doubt on the strict validity of the J-curve phenomenon on
the travel sector.
35 The lack of detailed trade data for specific sectoral level has been discussed in the Section 2.7 of Chapter 2. Apart from food, travel services and transportation services sectors, other sectors either have import or export data only and not the both.
173
Figure 5.11 Response of the travel services sector trade balance to generalised one standard deviation innovation in the real exchange rate
-.28
-.24
-.20
-.16
-.12
-.08
-.04
.00
1 2 3 4 5 6 7 8 9 10
Furthermore, with regard to the transportation sector, the analysis does not find support
for the J-curve phenomenon using either of the methods. The short–run favourable
effects followed by long–run significant and negative impact of the currency devaluation
on the transportation trade balance effectively rules out the presence of the J-curve
phenomenon using the traditional and the new definition. Even the IRF analysis does not
indicate the J-curve pattern (Figure 5.12). The impulse results show improvement of the
trade balance in the transportation sector for at least the first four periods. Thereafter, the
trade balance tends to deteriorate. A careful consideration shows that the IRF results on
the transportation sector exhibits an inverted U-shaped relationship in response to
devaluation. This suggests that a shock in the real exchange rate causes an improvement
of the transportation services sector in the short–run followed by deterioration in later
periods.
Figure 5.12 Response of the transportation services sector trade balance to
generalised one standard deviation innovation in the real exchange rate
-.04
-.02
.00
.02
.04
.06
.08
1 2 3 4 5 6 7 8 9 10
Period
Periods
174
The results on the sectoral services sectors are largely supportive of our earlier findings
on the J-curve phenomenon at the aggregate service sector trade balance. Hence, it is
ascertained that the reason for not being able to find the J-curve phenomenon at the
aggregate services trade balance using traditional and new definition is as a result of the
non-presence of the J-curve phenomenon in transport and travel trade sector using the
two methods. The only evidence of the J-curve phenomenon at the sectoral level trade is
obtained using IRF analysis for the travel sector.
5.4.5.2 Bilateral J-curve phenomenon
The bilateral level J-curve phenomenon is analysed for Fiji’s ten trading partner
countries in the study. Based on the three methods of assessing the J-curve phenomenon,
the results are summarised and presented in Table 5.17. The analysis reveals very weak
evidence of the J-curve phenomenon at the bilateral level of trade.
Table 5.17 Results on the bilateral level J-curve phenomenon
Trading Partner Countries
The J-curve phenomenon using: Traditional definition
New definition
Impulse Response Function Analysis
Major Trade Partners Australia No No No
New Zealand No Yes No
Japan No No Yes
United Kingdom No No Yes
Unites States of America No Yes Yes Emerging Asian Trade Partners
Singapore No No No
Malaysia No No No
India No No No
Hong Kong No No No
China No No No
175
In particular, the analysis shows no evidence of the J-curve phenomenon using the
traditional definition, in two cases using the new definition and in three cases using the
IRF analysis. The analysis using the traditional definition indicates that in no cases there
is a trade balance that experiences initial decline followed by improvement in response
to changes in the exchange rate in the short–run. However, trade with New Zealand and
the USA finds applicability of the J-curve phenomenon using the new definition.
Additionally, using the IRF analysis, the J-curve phenomenon is valid in the case of
Fiji’s trade with Japan, the UK and the USA.
Moreover, trade with Australia does not show presence of the J-curve phenomenon
using any of the methods. The results using traditional and new definition suggest short–
run improvement followed by long–run deterioration of the bilateral trade balance. The
IRF analysis also does not show the J-curve phenomenon (Figure 5.13). In fact, the
graphical representation of the shock in the real exchange rate to trade balance exhibits a
V-shaped relationship up until six periods after which it tends to decline. The response
pattern of trade balance between Fiji and Australia is largely consistent with our earlier
discussion and arguments. It is argued that despite some improvements in the trade
balance with Australia for some periods, the heavy reliance on importable commodities
causes deterioration in the trade performance in the later periods. Hence, the results are
consistent with the earlier estimated long–run unfavourable impact of devaluation on the
trade balance between the two countries.
Figure 5.13 Response of the trade balance between Fiji and Australia to generalised one standard deviation innovation in the real bilateral exchange rate
-.12
-.08
-.04
.00
.04
.08
1 2 3 4 5 6 7 8 9 10 Period
176
Fiji’s bilateral trade with New Zealand also shows no evidence of the J-curve
phenomenon using the traditional definition and the IRF analysis but confirms its
validity using the new definition. It is noted that the short–run effects of devaluation on
the trade balance between Fiji and New Zealand is unfavourable. However, using the
IRF analysis, the results reveal an inverted U-shaped relationship (Figure 5.14). In
particular, it reveals a decline in the bilateral trade balance four periods after the
devaluation year, after which it improves modestly.
Figure 5.14 Response of the trade balance between Fiji and New Zealand to generalised one standard deviation innovation in the real bilateral exchange rate
-.02
.00
.02
.04
.06
.08
.10
.12
.14
1 2 3 4 5 6 7 8 9 10
Japan and the UK, on the other hand, yield evidence of the J-curve phenomenon using
the IRF analysis only (Figure 5.15 and 5.16, respectively). Trade with Japan for
instance, shows that the bilateral trade balance immediately falls for at least two periods
accompanied by improvements between periods two and four. Bilateral trade with the
UK also falls for the first two periods followed by modest improvements till period five.
Nonetheless, the decline in Fiji’s trade with the UK in the later periods provides
evidence that the J-curve phenomenon for bilateral trade with the UK is not strictly
followed for a long period of time.
Periods
177
Figure 5.15 Response of the trade balance between Fiji and Japan to generalised one standard deviation innovation in the real bilateral exchange rate
-.25
-.20
-.15
-.10
-.05
.00
.05
.10
1 2 3 4 5 6 7 8 9 10
Figure 5.16 Response of the trade balance between Fiji and the UK to generalised one standard deviation innovation in the real bilateral exchange rate
-.6
-.5
-.4
-.3
-.2
-.1
.0
.1
.2
.3
1 2 3 4 5 6 7 8 9 10
With regard to Fiji’s trade with the USA, it gains support for the J-curve phenomenon
using the new definition and the IRF analysis (Figure 5.17). The results from the new
definition show short–run negative coefficients of the real bilateral exchange rate
followed by significant positive impact in the long–run. Additionally, the IRF analysis
shows that the trade balance with the USA falls for the first two periods followed by
modest improvements due to a shock in the real bilateral exchange rate. Hence, this
result locates the presence of the J-curve phenomenon at the aggregate goods sector to
be at the bilateral level with the USA.
Periods
Periods
178
Figure 5.17 Response of the trade balance between Fiji and the USA to generalised one standard deviation innovation in the real bilateral exchange rate
.10
.15
.20
.25
.30
.35
.40
1 2 3 4 5 6 7 8 9 10
Furthermore, tests on the presence of the bilateral J-curve phenomenon with Fiji’s
‘Emerging Asian Trade Partners’ suggests that the phenomenon is not valid for any of
the countries in this category. More specifically, using the traditional definition there is
evidence of an immediate favourable impact in the short–run in all the cases, followed
by instances of negative real exchange rate coefficients at longer periods in some cases.
These coefficients results do not provide any evidence of the J-curve phenomenon using
the traditional definition. Moreover, using the new definition, the short–run favourable
impact of the real exchange rate on the trade balances is followed by long–run
significant and negative impact on the same in all the cases. This again results in no
evidence of the J-curve phenomenon using the new definition.
Additionally, the IRF analysis also does not reveal the J-curve phenomenon in any of the
cases. In particular, the response of the bilateral trade balance with Singapore (Figure
5.18) exhibits an inverted U-shaped relationship while trade with Malaysia does not
depict any specific pattern (Figure 5.19). However, the response of the bilateral trade
balance with India, Hong Kong and China depicts a W-shaped relationship.
More specifically, the results show periods of immediate decline followed by
improvements combined with another round of decline and improvements. Nevertheless,
in the later periods, it follows a path of continuous decline. It is argued that the non-
Periods
179
presence of the J-curve phenomenon for this set of countries is largely because of Fiji
being a net importer and heavily dependent on importable commodities from these
countries. Hence, despite the currency devaluation, the bilateral trade balance situation
continues to deteriorate in the long–run with these countries.
Figure 5.18 Response of the trade balance between Fiji and Singapore to generalised one standard deviation innovation in the real bilateral exchange rate
-.5
-.4
-.3
-.2
-.1
.0
.1
.2
1 2 3 4 5 6 7 8 9 10
Figure 5.19 Response of the trade balance between Fiji and Malaysia to generalised one standard deviation innovation in the real bilateral exchange rate
-.9
-.8
-.7
-.6
-.5
-.4
-.3
-.2
-.1
.0
1 2 3 4 5 6 7 8 9 10
Periods
Periods
180
Figure 5.20 Response of the trade balance between Fiji and India to generalised one standard deviation innovation in the real bilateral exchange rate
-.8-.7-.6-.5-.4-.3-.2-.1.0.1
1 2 3 4 5 6 7 8 9 10
Figure 5.21 Response of the trade balance between Fiji and Hong Kong to
generalised one standard deviation innovation in the real bilateral exchange rate
-.4
-.3
-.2
-.1
.0
.1
.2
1 2 3 4 5 6 7 8 9 10
Figure 5.22 Response of the trade balance between Fiji and China to generalised
one standard deviation innovation in the real bilateral exchange rate
-.7-.6
-.5
-.4
-.3
-.2
-.1
.0
.1
.2
1 2 3 4 5 6 7 8 9 10
Periods
Periods
Periods
181
Based on the three methods to validate the J-curve phenomenon, the study finds
evidence of the phenomenon in four out of the ten trading partners considered in the
study. These include countries classified as part of the ‘Major Trade Partners’ only. The
countries with evidence of the J-curve phenomenon using any of the methods of
assessment include New Zealand, Japan, the UK and the USA. In particular, the study
finds that the J-curve phenomenon is more pronounced using either the new definition or
the IRF analysis for bilateral trade analysis. However, none of the trading partners
classified as part of the ‘Emerging Asian Trade Partners’ reveal evidence of the J-curve
phenomenon.
This result reinforces the argument that not all trade partners are affected in the same
manner as a result of devaluation. Moreover, this bilateral analysis on the J-curve
phenomenon lends support to the earlier findings on the J-curve phenomenon for the
aggregate goods sector. The goods trade balance shows strong support for the J-curve
phenomenon using all the methods and at the bilateral level strong support is gained
using the new definition and the IRF analysis. This suggests that the presence of the J-
curve phenomenon at the aggregate goods trade balance is contributed by the bilateral J-
curve phenomenon with New Zealand, Japan, the UK and largely by the USA.
5.4.6 Highlights from the empirical analysis at sectoral and bilateral levels The empirical analysis in this section makes up another important component of this
chapter and the overall thesis. In order to evaluate the effectiveness of devaluation and to
advance the aggregate trade analysis in Chapter 4, three sectoral and ten bilateral trade
balance equations were modelled. In this manner, the attempt has been to remove the
aggregation biasness from the aggregate trade analysis to locate the effectiveness of
devaluation in the country.
For the sectoral analysis, the long–run results indicate that the three sectors being
modelled are significantly and negatively influenced by currency devaluation in Fiji.
These results are also often backed by the respective export and import models. The
analysis also establishes that export of sugar is significantly and positively affected
182
while the other export commodities do not establish a significant relationship with
exchange rate. For the importable commodities, import of fuel, manufactured goods,
crude oil, machinery and transport equipment, chemical, oil and fat along with
miscellaneous manufactured goods are noted to significantly increase due to devaluation
in Fiji.
On the bilateral front, the long–run results note significant improvement in Fiji’s
bilateral trade balance with New Zealand and the USA while significant worsening of
bilateral goods trade performance is experienced in the cases of Australia, the UK,
Singapore, Malaysia, India, Hong Kong and China. In the short–run, the impact of
devaluation on is generally found to be favourable in most of the cases. Additionally, in
almost all the models, an increase in domestic income is noted to result in increase in
imports while a rise in trading partner income has a positive impact on the trade balance.
The impact of political instability has also been found to be affecting the trade
performance negatively in most of the cases in the economy.
Following the estimates of the sectoral and bilateral trade balance models, the study
finds evidence of the J-curve phenomenon for only one sector and in the case of four
trading partner countries. In particular, presence of the J-curve phenomenon is valid for
the travel services sector and for bilateral trade performance with New Zealand, Japan,
the UK and the USA.
5.5 Concluding comments This chapter has looked into the details of the exchange rate relationship with sectoral
and bilateral trade components in Fiji. More specifically, the major focus of the literature
review has been on the recent studies that have largely concentrated on investigating the
response of sectoral and bilateral trade flows to currency devaluation or depreciation in
the economy. It also included studies that have attempted to validate the J-curve
phenomenon. Included in the review are also relevant studies in the context of Fiji. The
significant contribution of this chapter lies in the empirical analysis to estimate the long–
run and the short–run effects of devaluation on Fiji’s disaggregated trade performance.
183
This also includes validating the presence of the J-curve phenomenon to locate the
effectiveness of currency devaluation in Fiji.
The sectoral analysis reveals that the trade balances of food, travel services and
transportation services sectors respond favourably to devaluation in the short–run but it
has an unfavourable impact in the long–run. Along with the three sectoral balances, an
additional three exportable and ten importable commodities trade equations were
modelled as well. Hence, imports of fuel, manufactured goods, crude oil, machinery and
transport equipment, chemical, oil and fat and miscellaneous manufactured goods are
found to experience significant increase due to devaluation in Fiji.
Similarly, on the bilateral front, ten trading partner countries of Fiji were analysed by
classifying them into ‘Major Trade Partners’ and ‘Emerging Asian Trade Partners’. The
findings from the bilateral trade analysis reveal long–run significant improvement in the
bilateral trade balance with New Zealand and the USA. On the contrary, significant
worsening of the bilateral trade performance is experienced with Australia, the UK,
Singapore, Malaysia, India, Hong Kong and China.
Following the estimates of the trade balance models, the study finds some evidence of
the J-curve phenomenon for only one sector and in the case of four trading partner
countries. In particular, some evidence on the J-curve phenomenon is obtained for the
travel services sector and for bilateral trade performance with New Zealand, Japan, the
UK and the USA. Hence, the results reinforce the argument that not all trade sectors and
partners are affected in the same manner as a result of devaluation in an economy.
Consequently, this particular chapter has attempted to fill the gap in the literature on the
nexus of exchange rate with sectoral and bilateral level trade performance. This analysis
to some extent has ensured that the analysis of devaluation on trade performance is as
comprehensive as possible in the context of a Pacific island developing economy, Fiji.
As a result, the analysis has allowed to identify how currency devaluation in Fiji has
impacted aggregate trade balance along with how Fiji’s sectoral and bilateral trade
performance responds to devaluation in the country.
184
CHAPTER 6
CONCLUSION AND POLICY RECOMMENDATIONS
6.1 Introduction This chapter highlights the major issue being investigated along with the major findings,
policy recommendations and limitations emanating from this thesis. With the use of
annual datasets over the 1975–2012 periods, the study has attempted to locate the
efficacy of the four episodes of currency devaluation carried out in Fiji. Hence, the
thesis had investigated empirically the impact of devaluation on the trade balance
disaggregated into sectoral and bilateral trade performance in the short– to long–run in
Fiji.
6.2 Major Findings Fiji is a small developing economy in the Pacific region which has experienced episodes
of low economic growth, annual budget deficits, trade deficits, increasing debt levels,
occurrence of severe natural disasters along with political instability in the country. Over
the last four decades since its independence, the economy has struggled to maintain
sustainable economic growth performances. The economy had historically been
dominated by agriculture. However, the services sector has become the largest foreign
exchange earner in the recent past.
The country has relatively small range of exportable commodities with limited
diversification in its exports. It mostly produces sugar, marine products, gold and
garments for exports while it offers tourism services to a large extent. With a relatively
small number of domestic economic activities, it largely depends on the import of
consumption and agricultural commodities. Hence, in the last three decades, the
185
economy has undergone four episodes of currency devaluation, primarily to boost trade
performance in the economy.
While most of the study finds a favourable impact of devaluation on trade components,
the current study finds only modest favourable impact on the trade performance. The
empirical results show presence of the J-curve phenomenon in the short to medium term.
However, it demonstrates a tendency to increasing trade deficit in the long–run. This is
because when the trade balance shows favourable effects after devaluation, income rises
and thus raises import demand in the economy. This results in the widening trade deficit
that puts the economy under continuous pressure to go for another round of devaluation.
Hence, it is argued that this often explains the four rounds of devaluation in the last three
decades in the economy. The existing studies have failed to elaborate that in a
convincing way. There seems to be a different argument from that provided by the
current literature, specifically for Fiji.
Therefore, this study follows the widely applied reduced form of the trade balance
model by Rose and Yellen (1989) and the VECM technique to investigate the impact of
devaluation on trade performance over the 1975–2012 periods in Fiji. Consequently, the
present study also incorporates the recent devaluation in 2009, which none of the earlier
studies had been able to incorporate as part of their analysis. Hence, the empirical results
support a favourable and significant impact of devaluation on the goods sector but not
on the services sector and the aggregate level (goods and services sector combined).
However, the services sector is found to have experienced a significantly adverse impact
in contradiction to other findings. In particular, the goods sector exhibits capacity to
increase local production to meet increased domestic demand more so than does the
services sector. The empirical evidence suggests that the export of local products such as
sugar, fish and gold largely experiences favourable gains. On the other hand, import of
textiles, beverage and tobacco experiences a corresponding decline in imports after
devaluation in the country. This results in an overall favourable impact on the goods
sector in the long–run.
186
In contrast, the tourism constitutes the largest share of the services sector trade, with a
limited possibility of substitution. As a result, the services sector only shows favourable
impact in the short–run but not in the long–run. It is argued that the relative reduction in
the price of services, particularly in the tourism sector, gives an immediate incentive to
tourists to increase their tourism activities to Fiji. Similarly, it also provides a
disincentive for locals to visit foreign countries either for leisure or business purposes as
the cost of travel immediately rises. This contributes to short–run favourable gains on
the services sector. However, travel plans by locals cannot be deferred for a very long
period of time as important business travel and travel by individuals for education and
health reasons are expected to continue despite weakening of the domestic dollar.
Moreover, tourists often do not directly benefit from the movements in the real exchange
rates as most tourists to the island countries depend on packaged vacations that are
quoted in foreign currency by tourism agents. This often negates the expected gains
from devaluation on the services sector in the medium– to long–run. Hence, it is
ascertained that the favourable impact on the overall trade balance in Fiji is particularly
being derived from the goods sector and not necessarily from the services sector.
Similarly, the evidence of the J-curve phenomenon at the aggregate goods and services
sector is also primarily derived from the validity of the same at the goods sector and not
at the services sector. Nevertheless, only aggregate analysis to investigate the impact of
devaluation on trade performance may not be able to explain clearly the detailed effect
of devaluation in the economy, due to the problem of aggregation biasness. Hence,
further analyses at sectoral and bilateral level of trade helps to locate the origin of the
effectiveness of the devaluation in Fiji.
At the sectoral level, the long–run results indicate that currency devaluation has
significant and adverse impact on the three sectors (food, travel services and
transportation services) trade balance in Fiji. The analysis also establishes that export of
187
sugar is significantly and positively affected while the other export commodities do not
establish a significant relationship with the exchange rate.
As far as the importable commodities are concerned, the import of fuel, manufactured
goods, crude oil, machinery and transport equipment, chemical, oil and fat along with
miscellaneous manufactured goods experience significant increase due to devaluation in
the country. This indicates that despite the currency devaluation, these importable
products are not effectively substituted by locally produced products. Due to the import
dependency of Fiji’s economy, these products continue to be imported. As a result, this
causes an adverse impact on the trade balance performance. However, there is evidence
of import substitution in the short–run for some of the importable goods commodities.
Empirical results show that the import of food, travel services, transportation services,
machinery and transport equipment, manufacturing goods, crude oil, oil and fats decline
as a result of devaluation in the short–run.
Furthermore, tests on the presence of the J-curve phenomenon at sectoral level is only
obtained for the travel services sector. Other sectoral trade balance for the food and
transportation services sector does not show indications of the J-curve phenomenon.
However, the result on the travel sector is only valid using the IRF analysis that reveals
devaluation helps to boost the travel sector in the short to medium term only. The results
on the food sector that shows relatively high dependence on importable food products
for domestic consumption experience an unfavourable impact in the long run. This
results in limited domestic substitution for locally produced products in the long–run.
Despite evidence of some short–run substitution between local and importable food
products, this is not necessarily the case over a long time. As a result, the tests do not
reveal presence of the J-curve phenomenon for the food sector in Fiji. This is also the
same for the transportation services sector.
At the bilateral level, Fiji’s trade partners are classified into ‘Major Trade Partners’ and
‘Emerging Asian Trade Partners’. The empirical results show a favourable impact of
devaluation on the bilateral goods trade performance with countries classified in the
188
‘Major Trade Partners’ only. Trade with New Zealand and the USA experiences
significant improvement in the trade balance in response to currency devaluation in Fiji.
However, all the countries classified as part of ‘Emerging Asian Trade Partners’ along
with Australia and the UK experience significant deterioration in bilateral trade
performance in response to devaluation in Fiji. The empirical results suggest that Fiji’s
exports to these destinations are relatively insensitive to changes in exchange rate while
imports from these countries continue to increase despite the devaluation. Hence, heavy
dependence on major importable commodities from these countries results in overall
trade deficits. However, evidence of improvement for most of the bilateral trade balance
performances is noted in the short–run.
Moreover, using either of the methods, the study finds evidence of the J-curve
phenomenon in only four out of the ten trading partner countries analysed in this study.
There are from the countries classified as the ‘Major Trade partners’ that include New
Zealand, Japan, the UK and the USA. However, none of the trading partners as part of
the ‘Emerging Asian Trade Partners’ show evidence of the J-curve phenomenon. This
result reinforces that by removing the aggregation biasness, evidence shows that not all
the trade partners are affected in the same manner as a result of devaluation. The results
show that the presence of the J-curve phenomenon at the aggregate goods trade balance
is contributed by the bilateral J-curve phenomenon with New Zealand, Japan, the UK
and the USA.
Further results on the impact of domestic income also indicate that rise in domestic
income has an adverse impact while increases in foreign income positively influences
trade balances in Fiji. The impact of political instability is also found to cause adverse
impacts on the trade performance in most of the cases. The impact is found to be larger
for exportable commodities and particularly on the export of tourism services in Fiji.
Hence, based on the analysis in this study, currency devaluation in Fiji has been found to
be partially effective due to the limited possibility of substitution in the goods sector, but
not in the services sector. However, this might not be a long–run remedy to solve the
189
problem of trade deficit in the long–run. This is particularly significant as most of Fiji’s
importable commodities are inelastic in nature and there is limited domestic substitution
in the economy. The rise in trade deficit is accelerated by the favourable impact of
devaluation on the trade balance through its income effect. Similarly, on the bilateral
front, trade balances with Fiji’s major trade partner, Australia, along with emerging
Asian counterparts, have limited gains from currency devaluation. Therefore, this study
has attempted to contribute and advance the literature on the relationship of currency
devaluation with the trade balance along with the assessment of the J-curve phenomenon
at the sectoral and bilateral level in the context of a Pacific island developing country.
6.3 Recommendations and policy implications From a policy perspective, the comprehensive analysis carried out at the aggregate,
sectoral and bilateral levels of trade performance on the impact of currency devaluation
indicates to the policy makers that devaluation has worked for Fiji in the short to
medium term to boost overall trade performance in the country. To overcome the
severity of crisis and external shocks, devaluation is an option in the short to medium
term but it is certainly not a permanent remedy.
It is also important to highlight that the favourable impact of devaluation is more
pronounced for the goods sector rather than the services sector. However, this can
actually make the situation worse for the service industry as it is the leading foreign
exchange earner in the economy. Though the impact of devaluation is favourable in the
short–run for the travel sector, its long–run impact is limited. Hence, maintaining the
international competitiveness of the domestic sectors is of paramount importance to
ensure sustained benefits from price adjustments due to deliberate currency devaluations
in the country. The productivity of the domestic export sectors needs to be improved or
else the unnecessary accompanying increase in import bills as a result of devaluation
will accelerate inflation rates and have severe repercussions in the economy.
Diversification of exports by tapping into the niche markets through further research and
development would help bring about sustained growth in export earnings.
190
Furthermore, increasing import penetration in the service industry particularly in the
tourism sector could prove to be counter-productive and negate the induced benefits
from rise in tourism numbers in the midst of devaluation in the country. Hence, it is
suggested that products that have large local content should be offered to tourists
whether it be in food, recreation or accommodation facilities.
With the center of global trade slowly shifting to Asian countries, Fiji is well situated to
benefit from increased goods export trade and particularly by tapping into the Asian
tourist market. An increasing export relationship with the Asian emerging markets
appears to bring about increasing opportunities as trade links through air and sea
transport are well established. Additionally, increased trade with New Zealand, Japan,
the UK and the USA are expected to bring benefit in the midst of devaluation in Fiji.
However, caution needs to be in place for increasing Fiji’s trade with Australia because
of unfavourable impacts observed on the bilateral trade balance. This is largely because
Fiji is dependent for most of its importable products from Australia, which pushes up the
import bill after devaluation. When Fiji, being a country with a fixed exchange regime,
heavily depends on the devaluation as a shock absorber during the crisis period, it is
noteworthy to mention that this cannot provide a permanent solution for the long-run.
Thus, we recommend carefully designed trade strategies to be an integral part to
minimise trade deficit problems in the economy. However, if it is not acted upon
carefully, this could result in one round of devaluation to compel for series of the same
in the immediate future. Hence, this could result in the economy falling into ‘the
exchange rate management crisis’.
6.4 Limitations One of the major limitations of carrying out comprehensive studies over a period of time
for Pacific island developing economies is the data availability. Due to the unavailability
of consistent data, Fiji’s trade relationship with other PICs in the region could not be
191
estimated. Though the trade share has historically been low, in the current times trade
among the countries in the region has gained momentum. Hence, evaluating the impact
of devaluation on the trade relationship with the neighbouring island countries would be
an interesting and useful exercise. It is suggested that for further research in this
direction, a wider consultation along with visitation to these countries’ statistical offices
and government agencies to collect further data from hard copy reports would be
needed.
Additionally, given the many obvious reasons as discussed in the thesis, this study has
only tested for the exchange rate and trade balance nexus along with the validity of the
J-curve phenomenon for Fiji. However, incorporating more countries from the region
would make this thesis overly optimistic. Secondly, the difficult tasks of gathering
complete and consistent data at disaggregate levels for other countries in the region are
already mentioned. Nonetheless, with a considerable amount of time and budget,
carrying out similar studies at aggregate, sectoral and bilateral level trade performance
for other PICs would be a useful addition to the literature.
It is also important to mention that the impact of currency devaluation on domestic price
level, debt levels, government budget balance and employment are beyond the scope of
the present study. Hence, these provide some additional research agendas which could
be explored in detail to understand further dynamics of devaluation in the country.
192
References
Acar, M. (2000). Devaluation in developing countries: expansionary or contractionary? Journal of Economic and Social Research, 2(1), 59-83.
Akbostanci, E. (2004). Dynamics of the trade balance: The Turkish experience.
Emerging Markets Finance and Trade, 40(5), 57-73. Al-Yousif, Y. (1997). Exports and economic growth: Some empirical evidence from the
Arab Gulf countries. Applied Economics, 29(6), 693-697. Arora, S., Bahmani-Oskooee, M., & Goswami, G. (2003). Bilateral J-curve between
India and her trading partners. Applied Economics, 35(9), 1037-1041. Asafu-Adjaye, J. (1999). Exchange rate variability and export growth in Fiji. Asia
Pacific School of Economics and Management Working Papers #99/4, Australian National University.
Ayen, Y. W. (2014). The effects of currency devaluation on output: The case of
Ethiopian economy. Journal of Economics and International Finance, 6(5), 103-111.
Bahmani-Oskooee, M. (1985). Devaluation and the J-curve: some evidence from LDCs.
The Review of Economics and Statistics, 67(3), 500-504. Bahmani-Oskooee, M. (1986). Determinants of international trade flows: The case of
developing countries. Journal of Development Economics, 20, 107-123. Bahmani-Oskooee, M. (1989). Devaluation and the J-curve: some evidence from LDCs:
errata. The Review of Economics and Statistics, 71(3), 553-554. Bahmani-Oskooee, M. (1998). Are devaluations contractionary in LDCs? Journal of
Economic Development, 23(1), 131-144. Bahmani-Oskooee, M. (2001). Nominal and real effective exchange rates of middle
eastern countries and their trade performance. Applied Economics, 33(1), 103-111.
Bahmani-Oskooee, M., & Alse, J. (1993). Export growth and economic growth: An
application of cointegration and error correction modeling. The Journal of Developing Areas, 27(4), 535-542.
Bahmani-Oskooee, M., & Alse, J. (1994). Short-run versus long-run effects of
devaluation: Error correction modeling and cointegration. Eastern Economic Journal, 20(4), 453-464.
193
Bahmani-Oskooee, M., & Bolhasani, M. (2008). The J-curve: Evidence from commodity trade between Canada and the U.S. Journal of Economics and Finance, 32(3), 207-225.
Bahmani-Oskooee, M., & Brooks, T. (1999). Cointegration approach to estimating
bilateral trade elasticities between U.S. and her trading partners. International Economic Journal, 13(4), 119-128.
Bahmani-Oskooee, M., & Gelan, A. (2012). Is there J-Curve effect in Africa?
International Review of Applied Economics, 26(1), 73-81. Bahmani-Oskooee, M., & Harvey, H. (2012). J-curve: Singapore versus her major
trading partners. Economic Papers, 31(4), 515-522. Bahmani-Oskooee, M., & Hegerty, S. (2011). The J-curve and NAFTA: evidence from
commodity trade between the US and Mexico. Applied Economics, 43, 1579-1593.
Bahmani-Oskooee, M., & Kutan, A. M. (2009). The J-curve in the emerging economies
of Eastern Europe. Applied Economics, 41(20), 2523-2532. Bahmani-Oskooee, M., & Malixi, M. (1992). More evidence on the J-curve from LDCs.
Journal of Policy Modeling, 14(5), 641-653. Bahmani-Oskooee, M., & Miteza, I. (2003). Are devaluation expansionary or
contractionary? A survey article. Economic Issues, 8(2), 1-28. Bahmani-Oskooee, M., & Ratha, A. (2004). The J-curve: A literature review. Applied
Economics, 36(3), 1377-1398. Bahmani-Oskooee, M., & Rhee, H.-J. (1997). Response of domestic production to
depreciation in Korea: An application of Johansen's cointegration methodology. International Economic Journal, 11(4), 103-112.
Bahmani-Oskooee, M., & Wang, Y. (2006). The J-curve: China versus her trading
partners. Bulletin of Economic Research, 58(4), 323-343. Bahmani-Oskooee, M., & Wang, Y. (2007). The J-curve at the industry level: Evidence
from trade between the US and Australia. Australian Economic Papers, 46(4), 315-328.
Bahmani-Oskooee, M., & Xu, J. (2012). Is there evidence of the J-curve in commodity
trade between the USA and Hong Kong? The Manchester School, 80(3), 295-320.
Bahmani-Oskooee, M., & Xu, J. (2013). The J-curve and Japan-China commodity trade. Journal of Chinese Economic and Business Studies, 11(1), 13-28.
194
Bahmani-Oskooee, M., & Zhang, R. (2013). The J-curve: Evidence from commodity trade between UK and China. Applied Economics, 45(31), 4369-4378.
Bahmani-Oskooee, M., & Zhang, R. (2014). Is there J-curve effect in the commodity
trade between Korea and rest of the world? Economic Change and Restructuring, 47(3), 227-250.
Bahmani-Oskooee, M., Chomsisengphet, S., & Kandil, M. (2002). Are devaluations
contractionary in Asia? Journal of Post Keynesian Economics, 25(1), 69-81. Bahmani-Oskooee, M., Economidou, C., & Goswami, G. (2006). Bilateral J-curve
between the UK vis-a-vis her major trading partners. Applied Economics, 38(8), 879–888.
Bahmani-Oskooee, M., Goswami, G., & Talukdar, B. (2005). The bilateral J-curve:
Australia versus her 23 trading partners. Australian Economic Papers, 44(2), 110-120.
Bahmani-Oskooee, M., Harvey, H., & Hegerty, S. (2013). Currency depreciations and
the U.S.–Italian trade balance: Industry-level estimates. Research in Economics, 67(3), 215-225.
Bahmani-Oskooee, M., Salman, H., & Rustam, J. (2014). Is there a J-curve for
Azerbaijan? New evidence from industry-level analysis. Macroeconomics & Finance in Emerging Market Economies, 7(1), 83-98.
Berg, A., & Krueger, A. (2003). Trade, growth and poverty: A selective study. IMF
Working Paper, WP/03/30, International Monetary Fund. Washington D.C. Bernard, A., & Jensen, B. (2004). Entry, expansion, and intensity in the US export
boom, 1987-1992. Review of International Economics, 12(4), 662-675. Branson, W. H. (1986). Stabilization, stagflation, and investment incentives: The case of
Kenya, 1979-1980. In S. Edwards, & L. Ahamed, Economic Adjustment and exchange rates in developing countries (pp. 267-293). Chicago: Chicago UP.
Bruno, M. (1979). Stabitization and stagflation in a semi-industrialized economy. In R.
Dornbusch, & J. A. Frenkel, International Economic Policy: Theory and Evidence (pp. 270-289). Baltimore: John Hopkins UP.
Cain, J., Hasan, R., & Mitra, D. (2012). Trade liberisation and poverty reduction: New
evidence from Indian States. In J. Bhagwati, & A. Panagariya, India's reforms: How they produced inclusive growth (pp. 91-185). New York: Oxford University Press.
Celik, S., & Kaya, H. (2010). Real exchange rates and bilateral trade dynamics of Turkey: Panel cointegration approach. Applied Economics Letters, 17(8), 791–795.
195
Chand, S. (1998). Exchange rate and financial management: Some lessons for and from Fiji. Pacific Economic Bulletin, 13(1), 36-46.
Chen, H., Rauqueue, L., Singh, S., Wu, Y., & Yang, Y. (2014). Pacific island countries:
In search of a trade strategy. IMF Working Paper #14/158, International Monetary Fund, Washington D.C.
Cooper, R. (1971). Currency devaluation in developing countries, Essays in
International Finance, no.86, Princeton University, International Finance Section, NJ.
Culiuc, A. (2014). Determinants of international tourism. IMF Working Paper 14/82,
International Monetary Fund, Washington D.C. Dash, A. (2013). Bilateral J-curve between India and her trading partners: A quantitative
perspective. Economic Analysis & Policy, 43(3), 315-338. Davies, M. (2010). Exchange rate in small open economies: PFTAC/CCBS Course
August, 2010. Retrieved April 16, 2014, from Pacific Financial Technical Assistance Centre: http://www.pftac.org/filemanager/files/Macro_Training/CCBS_2010/6_CCBS_Exchange_Rates_Presentation.ppt
de Brouwer, G. (2000). Should Pacific island countries adopt the Australian dollar?
Pacific Economic Bulletin, 15(2), 161-169. Diaz-Alejandro, C. (1965). Exchange rate devaluation in a semi industrialised country:
The experience of Argentina, 1955-1961. Cambridge MA: MIT Press. Donovan, D. J. (1982). Macroeconomic performance and adjustment under Fund-
supported programs: The experience of the seventies. IMF Staff Papers, 29, 171-203.
Dulare, C. (2005). Monetary and exchange rate policy in Fiji. Suva: Fiji Institute of
Applied Studies. Duncan, R. (2005). A common currency for the Pacific island economies. In S. Chand,
Pacific islands regional integration and governance (pp. 11-118). Canberra: The Australian National University.
Dutt, S. D., & Ghosh, D. (1994). An empirical investigation of the exportgrowth-
economic growth relationship. Applied Economics Letters, 1(3), 44-48. Edwards, S. (1986). Are devaluations contractionary? The Review of Economics and
Statistics, 68(3), 501-508. Edwards, S. (1989a). Real exchange rates, devaluation, and adjustment. Cambridge,
MA: MIT Press.
196
Edwards, S. (1989b). Exchange controls, devaluations, and real exchange rates: The Latin American experience. Economic Development and Cultural Change, 37(3), 457-494.
Fang, W., Lai, Y., & Miller, S. (2006). Export promotion through exchange rate
changes: Exchange rate depreciation or stabilization? Southern Economic Journal, 72(3), 611-626.
FBOS. (various issues(a)). Key Statistics. Suva: Fiji Bureau of Statistics. FBOS. (various issues(b)). Overseas Merchandise Trade Statistics. Suva: Fiji Bureau of
Statistics. Feeny, S. (2010). The impact of the global economic crisis on the Pacific region. Oxfam
Australia, Melbourne. Flemingham, B. (1988). Where is the Australian J-curve? Bulletin of Economic
Research, 40(1), 43-56. Freund, C., & Pierola, M. (2010). Export entrepreneurs: Evidence from Peru. Policy
Research Working Paper 5407, The World Bank. Galebotswe, O., & Andrias, T. (2011). Are devaluations contractionary in small import-
dependent economies? Evidence from Botswana. Botswana Journal of Economics, 8(12), 86-98.
Georgopoulos, G. (2008). The J-curve revisited: An empirical analysis for Canada.
Atlantic Economic Journal, 36(3), 315-332. Giles, J., & Williams, C. (2000). Export-led growth: A survey of the empirical literature
and some noncausality results, Part 1. The Journal of International Trade and Economic Development, 9(3), 261-337.
Gomes, F., & Paz, S. (2005). Can real exchange rate devaluation improve the trade
balance? The 1990–1998 Brazilian case. Applied Economics Letters, 12(9), 525-528.
Gong, X., & Rao, M. (2014). The economic impact of prolonged political instability: A
case study of Fiji. NATSEM Working Paper 14/26, National Centre for Social and Economic Modelling, Institute for Governance and Policy Analysis (IGPA), University of Canberra.
Gounder, N., & Prasad, B. (2013). Economic growth, investment, confidence and
poverty reduction in Fiji: semi-rational exuberance? Development Policy Centre Discussion Paper #31, Crawford School of Public Policy, The Australian National University, Canberra.
197
Gounder, N., & Prasad, B. (2012). Trading preferentially and protection- Is it good for Pacific island countries? The Journal of Pacific Studies, 32, 57-64.
Gounder, R. (2001). Aid-growth nexus: Empirical evidence from Fiji. Applied
Economics, 33(8), 1009-1019. Gupta-Kapoor, A., & Ramakrishnan, U. (1999). Is there a J-curve? A new estimation for
Japan. International Economic Journal, 13(4), 71-79. Gylfason, T., & Radetzki, M. (1991). Does devaluation make sense in the least
developed countries? Economic Development and Cultural Change, 40(1), 1-25. Gylfason, T., & Schmid, M. (1983). Does devaluation cause stagflation? The Canadian
Journal of Economics, 25(4), 37-64. Haddad, M., & Pancaro, C. (2010). Can real exchange rate undervaluation boost exports
and growth in developing countries? Yes, but not for long. Economic Premise No.20, The World Bank. Washington D.C.
Haider, J., Afzal, M., & Riaz, F. (2011). Estimation of import and export demadn
functions using bilateral trade data: The case of Pakistan. Business and Economic Horizons, 6(3), 40-53.
Hanson, J. A. (1983). Contractionary devaluation, substitution in production and
consumption and the role of the labour market. Journal of International Economics, 14(1-2), 179-189.
Hasan, R., Mitra, D., & Ural, B. P. (2006). Trade liberalization, labor-market
institutions, and poverty reduction: Evidence from Indian States. India Policy Forum, 3(1), 71-122.
Himarios, D. (1985). The effects of devaluation on the trade balance: a critical view and
reexamination of Miles’s (New Results). Journal of International Money and Finance, 4, 553-563.
IMF. (2014). International Financial Statistics (online database). Washington DC, USA.
Retrieved January 20, 2014, from International Monetary Fund: https://elibrary-data.imf.org/
IMF. (2015, April). World Economic Outlook (WEO). Retrieved May 1, 2015, from
International Monetary Fund: http://www.imf.org/external/pubs/ft/weo/2015/01/ Jayaraman, T. K (1999). Is there a case of freely floating exchange rate in Fiji? Pacific
Economic Bulletin, 14(1), 93-100. Jayaraman, T. K (2001). Prospects for a currency union in the Pacific: A preliminary
assessment. The Journal of Pacific Studies, 25, 173-201.
198
Jayaraman, T. K. (2004). A single currency for the Pacific island countries: A stepwise approach. Asia Pacific Development Journal, 11(1), 91-111.
Jayaraman, T. K. (2011). Global financial crisis and economic downturn: Challenges
and opportunities for the small island countries in the Pacific. Bank of Valletta Review, 43, 44-58.
Jayaraman, T. K., & Choong, C. (2008). Exchange market pressure in a small Pacific
Island country: A study of Fiji: 1975-2005. International Journal of Social Economics, 35(12), 985-1004.
Jayaraman, T. K., & Choong, C. (2009). A single currency for Pacific island countries:
A revisit. Journal of Economic Integration, 24(2), 321-342. Jayaraman, T. K., Ward, B. D., & Lu, Z. L. (2007). Are Pacific islands ready for a
currency union? An empirical study of degree of economic convergence. Journal of Asia and Pacific Economy, 12(4), 504-521.
Johansen, S. (1988). Statistical analysis of cointegration vectors. Journal of Economic
Dynamics and Control, 12(2-3), 231-254. Johansen, S., & Juselius, K. (1990). Maximum likelihood estimation and inferences on
cointegration-with applications to the demand for money. Oxford Bulletin of Economics, 52(2), 169-210.
Junz, H. B., & Rhomberg, R. R. (1973). Price-competitiveness in export trade among
industrial countries. American Economic Review, 63(2), 412-418. Kalyoncu, H., Artan, S., Tezekici, S., & Ozturk, I. (2008). Currency devaluation and
output growth: An empirical evidence from OECD countries. International Research Journal of Finance and Economics, 14, 232-238.
Kalyoncu, H., Ozturk, I., Artan, S., & Kalyoncu, K. (2009). Devaluation and trade
balance in Latin American countries. Zbornik radova Ekonomskog fakulteta u Rijeci/Proceedings of Rijeka Faculty of Economics, 27(1), 115-128.
Kamin, S. B. (1988). Devaluation, external balance, and macroeconomic performance in
developing countries: a look at the numbers. Princeton Essays in Internatioonal Finance, No. 62.
Kamin, S. B., & Klau, M. (1998). Some multi-country evidence on the effects of real
exchange rates on output. International Finance Discussion Papers, 611, Board of Governors of the Federal Reserve System, the Bank for International Settlements, Washington, D.C.
Kamin, S. B., & Rogers, J. H. (1997). Output and the real exchange rate in developing
countries: an application to Mexico. International Finance Discussion Paper,
199
580, Board of Governors of the Federal Reserve System, the Bank for International Settlements, Washington, D.C.
Kandil, M. (2004). Exchange rate fluctuations and economic activity in developing
countries: Theory and evidence. Journal of Economic Development, 29(1), 85-108.
Katircioglu, S., Eminer, F., Aga, M., & Ozyigit, A. (2010). Trade and growth in the
Pacific islands- Empirical evidence from the bounds tests to level relationships and granger causality tests. Romanian Journal of Economic Forecasting, 4, 88-101.
Kaufmann, U. (2008). Gains from trade versus adjustment costs: The Economic
Partnership Agreement and the Pacific Agreement on Closer Economic Relations. Unpublished Master of Commerce dissertation, The University of the South Pacific, Suva.
Krueger, A. (1978). Foreign Trade Regimes and Economic Development: Liberalization
Attempts and Consequences. Cambridge, MA:, Ballinger. Krugman, P., & Taylor, L. (1978). Contractionary effects of devaluation. Journal of
International Economics, 8(3), 445-456. Kumar, S. (2009). Estimating export equations for developing countries. The Icfai
University Journal of Applied Economics, 8(2), 17-28. Kumar, S., & Prasad, B. (2002). Fiji's economic woes: A nation in search of
development progress. Pacific Economic Bulletin, 17(1), 1-23. Lal, A., & Lowinger, T. (2002). The J-curve: Evidence from East Asia. Journal of
Economic Integration, 17, 397-415. Lizondo, S., & Montiel, P. J. (1989). Contractionary devaluation in developing
countries: An analytical overview. IMF Staff Papers, 36(1), 182-227. Llyod, P. J., & MacLaren, D. (2002). Measures of trade openness using CGE analysis.
Journal of Policy Modeling, 24(1), 67-81. Magee, S. P. (1973). Currency contracts, pass through and devaluation. Brooking Papers
on Economic Activity, 1, 303-325. Ministry of Finance. (2012). Investing in our Future. Economic and Fiscal Update:
Supplement to the 2013 Budget Address, Ministry of Finance, Government of Fiji, Suva.
Ministry of Finance. (2013). Building a Smarter Fiji. Economic and Fiscal Update:
Supplement to the 2014 Budget Address, Ministry of Finance, Government of Fiji, Suva.
200
Ministry of Finance. (2014). Turning Promised Into Deeds. Economic and Fiscal Update: Supplement to the 2015 Budget Address, Ministry of Finance, Government of Fiji, Suva.
Moffett, M. H. (1989). The J-curve revisited: An empirical examination for the United
States. Journal of International Money and Finance, 8(3), 425-444. Musawa, N. (2014). Relationship between Zambias exchange rates and the trade
balance- J-curve hypothesis. International Journal of Finance and Accounting, 3(3), 192-196.
Narayan, P., & Narayan, S. (2004a). The J-curve; evidence from Fiji. International
Review of Applied Economics, 18(3), 369-380. Narayan, P., & Narayan, S. (2004b). Determinants of demand for Fiji's exports: An
empirical estimation. The Developing Economies, 42(1), 95-112. Narayan, P., & Narayan, S. (2005). Estimating income and price elasticities of imports
for Fiji in a cointegration framework. Economic Modelling, 22(3), 423-438. Narayan, P., & Narayan, S. (2009). Understanding Fiji's declining foreign reserve
position. Pacific Economic Bulletin, 24(2), 95-112. Narayan, P., & Prasad, B. (2003). Fiji National Budget 2003: Building a resilient
economy? Pacific Economic Bulletin, 18(2), 91-106. Narayan, P. (2004). New Zealand's trade balance: evidence of tje J-curve and granger
causality. Applied Economics Letters, 11(6), 351-354. Narayan, P. (2006). Examining the relationship between trade balance and exchange
rate: the case of China's trade with the USA. Applied Economics Letters, 13(8), 507-510.
Narayan, P., & Narayan, S. (2007). Is devaluation expansionary or contractionary?
Empirical evidence from Fiji. Applied Economics, 2007, 39, 2589–2598, 39, 2589-2598.
Narayan, P., Narayan, S., Mishra, S., & Smyth, R. (2012). An analysis of Fiji's monetary
policy transmission. Studies in Economics and Finance, 29(1), 57-70. Narayan, S. (2013). A structural VAR model of the Fiji Islands. Economic Modelling,
31, 238–244. Noland, M. (1989). Japanese trade elasticities and the J-curve. The Review of Economics
and Statistics, 71(1), 175-179. Onafowara, O. (2003). Exchange rate and trade balance in East Asia: Is there a J-curve?
Economic Bulletin, 5, 1-13.
201
Ono, M., & Baak, S. (2014). Revisiting the J-curve for Japan. Modern Economy, 5, 32-
47. doi:http://dx.doi.org/10.4236/me.2014.51005 Pattichis, C. (2012). Exchange rate effects on trade in services. Journal of Economic
Studies, 39(6), 697-708. Pesaran, M., & Shin, Y. (1998). Generalised impulse response analysisin linear
multivariate models. Economic Bulletin, 58(1), 17-29. Prasad, B. (2012a). Fiji economy: Muddling through. The Round Table: The
Commonwealth Journal of International Affairs, 101(6), 557-573. Prasad, B. (2010). Global crisis, domestic crisis and crisis of confidence: Which way
forward for Fiji. Pacific Economic Bulletin, 25(2), 1-24. Prasad, B. (2012b). Why Fiji is not the “Mauritius” of the Pacific? Lessons for small
island nations in the Pacific. Development Policy Centre Discussion Paper #25, Crawford School of Public Policy, The Australian National University, Canberra.
Prasad, B., & Narayan, P. (2008). Reviving growth in the Fiji islands; Are we swimming
or sinking. Pacific Economic Bulletin, 23(2), 5-26. Prasad, S. (2000). Determinants of exports in Fiji. RBF Working Paper No. 2000/04,
Reserve Bank of Fiji, Suva. Rahman, M., & Islam, A. (2006). Taka-Dollar exchange rate and Bangladesh trade
balance: Evidence on J-curve or S-curve? Indian Journal of Economics and Business, 5, 279-288.
Rao, B. B., & Singh, R. (2007). Estimating export equations. Applied Economics Letters,
14(11), 799-802. RBF. (2009). Press Release No.16: Reserve Bank announces the following monetary
policy changes. Suva: Reserve Bank of Fiji. RBF. (various issues (a)). Quarterly Review. Suva: Reserve Bank of Fiji. Reddy, M. (1997). Devaluation and economic simulation: The Fiji economy post-coup.
Pacific Economic Bulletin, 12(2), 85-94. Rehman, H., & Afzal, M. (2003). The J curve phenomenon: An evidence from Pakistan.
Pakistan Economic and Social Review, 41(1&2), 85-94. Rogers, A. (2000). An analysis of the determinants of Fiji's imports. RBF Working
Paper No. 2000/03, Reserve Bank of Fiji, Suva. Rose, A., & Yellen, J. (1989). Is there a J-curve? Journal of Monetary Economics, 24(1),
53-68.
202
Shirvani, H., & Wilbratte, B. (1997). The relationship between the real exchange rate
and the trade balance: An empirical reassessment. International Economic Journal, 11(1), 39-50.
Siliverstovs, B., & Herzer, D. (2006). Export-led growth hypothesis: Evidence from
Chile. Applied Economics Letters, 13(5), 319-324. Singh, R. (2006). Cointegration tests on trade equation: Is devaluation an option for Fiji.
School of Economics Working paper series, The University of the South Pacific #2006/18.
Singh, R., & Prasad, B. C. (2008). Small states, big problems: Small solutions from big
countries. Journal of World Trade, 42(5), 905-926. Soleymani, A., & Saboori, B. (2012). The J-curve: Evidence from commodity trade
between Malaysia and Japan. The International Journal of Applied Economics and Finance, 6(2), 64-73.
Soleymani, A., Chua, S. Y., & Saboori, B. (2011). The J-curve at industry level:
Evidence from Malaysia-China trade. International Journal of Economics and Finance, 3(6), 66-78.
Taye, H. K. (1999). The impact of devaluation on macroeconomic performance: The
case of Ethiopia. Journal of Policy Modeling, 21(4), 481-496. The World Bank. (2014). World Development Indicators (online databse). Washington
DC, USA. Retrieved August 21, 2014, from The World Bank: http://databank.worldbank.org/data/views/variableSelection/selectvariables.aspx?source=world-development-indicators
The World Bank. (2015). World Development Indicators (online databse). Washington
DC, USA. Retrieved May 1, 2015, from The World Bank: http://data.worldbank.org/data-catalog/world-development-indicators
UNCOMTRADE. (2014). International Trade Statistics (online database). Retrieved
June 25, 2014, from UNCOMTRADE: http://unstats.un.org/unsd/snaama/dnlList.asp
UNDP. (2014). Human Development Reports. Retrieved September 7, 2014, from
United Nations Development Programme: http://hdr.undp.org/en/2014-report/download.
Van Wijnbergen, S. V. (1986). Exchange rate management and stabilization policies in
developing countries. Journal of Development Economics, 23(2), 227-247.
203
Wang, C.-H., Lin, C.-H. A., & Yang, C.-H. (2012). Short-run and long-run effects of exchange rate change on trade balance: Evidence from China and its trading partners. Japan and the World Economy, 24(4), 266-273.
Wijeweera, A., & Dollery, B. (2013). J-curve disparity between the goods sector and the
service sector: Evidence from Australia. Applied Economics Letters, 20(5), 452-456.
Yang, Y., Chen, H., Singh, S., & Singh, B. (2013). The Pacific speed of growth: How
fast can it be and what determines it? IMF Working Paper 13/104, International Monetary Fund, Washington D.C.
Yang, Y., Davies, M., Wang, S., Dunn, J., & Wu, Y. (2011) Monetary policy
transmission mechanisms in Pacific island countries. IMF Working Paper 11/96, International Monetary Fund, Washington D.C.
Yazici, M. (2006). Is the J-curve effect observable in Turkish Agricultural sector?
Journal of Central European Agriculture, 7(2), 319-322. Yazici, M. (2010). Is there a J-curve effect in Turkish services? Quality & Quantity, 44,
167-172. Yazici, M., & Klasra, M. (2010). Import-content of exports and J-curve effect. Applied
Economics, 42(6), 769-776. Yusoff, M. (2010). The effects of real exchange rate on trade balance and domestic
output: A case of Malaysia. The International Trade Journal, 24(2), 209-226.
204
Appendix A Visitor arrivals in Fiji by country of residence, 1975–2014
Source: Fiji Bureau of Statistics (various years)
Period Australia New Zealand
USA Canada UK Continental Europe
Japan PICs Others TOTAL
1975 58,482 42,242 25,270 10,685 4,711 5,533 9,266 5,518 161,7071976 66,591 37,601 28,193 11,262 4,545 4,788 9,311 6,373 168,6651977 63,748 41,338 28,171 11,555 4,311 5,763 8,703 9,403 173,0191978 74,922 40,825 29,449 9,803 4,177 6,052 5,709 8,957 4,169 184,0631979 73,191 44,434 27,336 10,085 4,768 7,624 5,665 11,851 3,786 188,7401980 72,260 36,389 30,137 13,386 5,244 8,990 7,141 10,137 6,312 189,9961981 80,912 32,490 24,164 15,708 4,814 6,921 10,606 9,617 4,703 189,9351982 95,455 28,304 23,211 13,698 4,332 6,111 18,029 9,970 4,526 203,6361983 85,027 24,048 25,636 13,037 5,888 8,330 14,401 10,588 4,661 191,6161984 101,403 26,803 37,285 16,522 8,569 11,283 14,864 13,178 5,320 235,2271985 89,459 19,542 49,557 18,908 7,707 12,667 12,602 11,936 5,797 228,1751986 86,297 22,720 69,732 23,651 9,972 15,088 11,801 12,815 5,748 257,8241987 65,382 16,197 47,037 16,819 8,511 14,726 5,487 11,217 4,490 189,8661988 75,264 21,507 42,144 16,883 8,464 20,498 3,425 14,219 5,751 208,1551989 96,992 28,128 34,425 16,536 11,404 23,916 13,840 18,064 7,260 250,5651990 103,535 29,432 36,928 18,438 16,773 27,211 21,619 17,528 7,532 278,9961991 86,625 30,631 31,842 15,242 16,555 27,802 26,265 16,227 8,161 259,3501992 87,395 37,227 34,802 12,602 16,795 29,513 35,960 15,627 8,613 278,5341993 77,609 40,778 42,557 12,447 20,223 29,786 38,203 16,985 8,864 287,462
1994 85,532 53,495 45,351 12,018 23,905 31,004 39,782 17,931 9,846 318,874
1995 78,503 59,019 39,736 10,412 24,409 30,968 45,300 17,461 12,687 318,495
1996 79,534 63,430 38,707 11,431 28,907 31,875 44,598 18,545 22,533 339,560
1997 80,351 68,116 44,376 13,359 35,019 32,806 44,783 20,381 20,250 359,441
1998 100,756 70,840 48,390 12,837 39,341 29,334 35,833 22,850 11,161 371,342
1999 118,272 72,156 62,131 13,552 40,316 28,371 37,930 26,090 11,137 409,955
2000 76,883 49,470 52,534 10,532 29,215 22,506 19,674 21,534 11,722 294,070
2001 98,213 66,472 57,711 10,752 30,508 20,917 20,411 23,608 19,422 348,014
2002 123,606 68,293 58,815 9,802 43,393 21,654 26,382 24,051 21,863 397,859
2003 141,873 75,016 58,323 10,990 49,794 21,847 23,464 28,167 21,326 430,800
2004 176,195 103,900 65,211 12,435 47,668 22,720 24,392 26,182 25,372 504,075
2005 203,250 112,932 62,640 12,625 44,472 25,123 22,304 28,476 33,323 545,145
2006 206,529 107,277 66,631 14,372 38,239 26,801 23,794 29,725 35,221 548,589
2007 207,001 99,744 64,687 16,992 34,785 26,311 22,800 34,221 33,340 539,881
2008 247,608 100,018 63,667 17,871 33,935 29,512 21,918 35,936 34,566 585,031
2009 248,589 90,898 51,592 13,452 26,213 28,926 14,975 35,078 32,463 542,186
2010 318,185 97,857 53,122 12,970 23,813 29,115 12,011 39,198 45,597 631,868
2011 344,829 103,181 55,086 14,090 24,054 32,354 9,616 38,823 53,005 675,050
2012 337,291 106,122 56,478 13,426 17,076 29,327 7,069 38,886 54,915 660,590
2013 340,151 108,239 55,385 13,052 17,209 28,905 7,314 39,450 48,002 657,707
2014 349,217 123,968 61,924 12,457 16,782 30,585 5,888 39,298 52,511 692,630
included in others section
205
Appendix B Major Trade Agreements in Fiji The Melanesian Spearhead Group Trade Agreement (MSGTA)
The Melanesian Spearhead Group (MSG) is formed with South Pacific countries of Papua New Guinea (PNG), Vanuatu, Solomon Islands and Fiji since 1986. Together they entered into a Melanesian Spearhead Group Trade Agreement (MSGTA) to establish a reciprocal free trade agreement (trade in goods) by adopting a negative list approach (Negative list approach requires countries to list products that they do not want to trade in, which may include sensitive products or products associated with infant industries. These products are exempted from the free trade agreement). Three countries, Fiji, PNG, and Vanuatu, have been trading under duty-free status since 2013; however, the Solomon Islands is expected to fully liberate their tariffs by 2017. A revised Rules of Origin Handbook for the MSGTA was developed in 2013. This has provided valuable information for Melanesian businesses to comply with customs requirements such as “rules of origin” conditions to qualify for duty free entry into MSG markets. The MSGTA is currently being reviewed by member countries to include new chapters on services, labour mobility, government procurement and investment.
The Pacific Island Countries Trade Agreement (PICTA)
PICTA is a free trade agreement amongst the 14 Forum Island Countries (FICs) that allows for duty free and quota free access to products within the region. Under this agreement, all barriers to merchandise trade, such as tariffs and quotas between FIC countries, are to be removed. However, the PICTA has only been ratified by 11 (Cook Islands, Fiji, Kiribati, Nauru, Niue, Papua New Guinea, Samoa, Solomon Islands, Tonga, Tuvalu and Vanuatu) out of the 14 FICs and only 7 have shown their readiness to trade, which presents substantial challenges in the implementation of the agreement. Additionally, just like the MSG trade agreement, PICTA also allows countries to maintain a negative list of products so as to maintain tariffs on selected items. Fiji, on the other hand, while demonstrating its commitment to regional integration, does not have any products on the negative list in the PICTA. Recently, the trading block has also successfully concluded the PICTA Trade in Services negotiations with almost half of the FICs signing the agreement except, Fiji and Papua New Guinea. Talks are also underway for labour mobility on the Temporary Movement of Natural Person scheme amongst FIC member countries. However, till to date Fiji has exported only a negligible level of its exports to the Pacific counterparts under the PICTA.
206
The Pacific Agreement on Closer Economic Relations Plus (PACER Plus)
In 2002, the Forum Trade Ministers entered into force with PACER, which allows FICs to move towards a more comprehensive framework for trade and economic cooperation between Australia and New Zealand. The PACER is a framework agreement to expand development of trade cooperation on a step by step basis and provides opportunities for technical assistance and capacity building in the area of trade facilitation from Australia and New Zealand to the Pacific countries. It was bound to start with free trade negotiations and later expand to services. However, Forum Island members’ engaging in trade negotiation with the European Union under the Economic Partnership Agreement (EPA) gave rise to PACER Plus negotiation. The PACER agreement obliged FICs along with Fiji to offer Australia and New Zealand the same trade preferences it provides to any other external trading partners. This effectively meant that all privileges given under the EPA to European Union must also be extended to PACER. However, it carried along with it higher possibility for revenue losses under PACER due to its voluminous trade with Australia and New Zealand than European Union.
The Interim Economic Partnership Agreement (I-EPA)
The European Council amended its Market Access Regulation in 2013 to exclude ACP countries from preferential access to the EU market. This adjustment had led to African, Caribbean and Pacific nations (ACP) countries to provisionally adopt the I-EPA or risk losing out on existing trade preferences. Consequently, in July 2014 Government was compelled to approve the provisional application of the I-EPA to prevent any disruption of exports to the European Market.
The Comprehensive Economic Partnership Agreement (Comprehensive EPA)
After the expiration of the Cotonou agreement which was the successor of the Lome IV convention in 2007, the negotiations into Economic Partnership Agreement took off between the European Union (EU) and the African, Caribbean and Pacific Group of States (ACP). This agreement is designed to effectively replace the non-reciprocal trade preferences of the previous agreements. The EPAs will be reciprocal trade agreements, with the consequence that the EU provides not only duty free access to its markets but also provide duty free entry to EU exports and companies into Pacific economies. This agreement was supposed to come into force by the end of 2008 but it is yet to be concluded. The full EPA is based on the provisions of development cooperation, Pacific related issues on fisheries and the social and environmental issues. However, the issue of including service trade, Temporary Movement of Natural Persons (TMNP) or labour mobility schemes,
207
competition policy and intellectual property rights remains open and requires further negotiations. At present, Fiji continues to negotiate development friendly and mutually beneficial provisions in the Comprehensive EPA (CEPA) with other Pacific ACP States.
Source: Ministry of Finance Budget Supplement (various years)
208
Appendix C Variable definitions
TBG measures Fiji’s goods trade balance defined as export of goods divided by import of goods in constant US dollars at 2005 prices
TBS measures Fiji’s services trade balance defined as export of services divided by import of services in constant US dollars at 2005 prices
TBGS measures Fiji’s goods and services trade balance defined as export of goods and services divided by import of goods and services in constant US dollars at 2005 prices
AUSTBG measures Fiji’s goods trade balance with Australia defined as Fiji’s export of goods to Australia divided by import of goods from Australia in constant US dollars at 2005 prices
NZTBG measures Fiji’s goods trade balance with New Zealand defined as Fiji’s export of goods to New Zealand divided by import of goods from New Zealand in constant US dollars at 2005 prices
JPNTBG measures Fiji’s goods trade balance with Japan defined as Fiji’s export of goods to Japan divided by import of goods from Japan in constant US dollars at 2005 prices
USATBG measures Fiji’s goods trade balance with USA defined as Fiji’s export of goods to USA divided by import of goods from USA in constant US dollars at 2005 prices
UKTBG measures Fiji’s goods trade balance with UK defined as Fiji’s export of goods to UK divided by import of goods from UK in constant US dollars at 2005 prices
SINGTBG measures Fiji’s goods trade balance with Singapore defined as Fiji’s export of goods to Singapore divided by import of goods from Singapore in constant US dollars at 2005 prices
CHNTBG measures Fiji’s goods trade balance with China defined as Fiji’s export of goods to China divided by import of goods from China in constant US dollars at 2005 prices
MALATBG measures Fiji’s goods trade balance with Malaysia defined as Fiji’s export of goods to Malaysia divided by import of goods from Malaysia in constant US dollars at 2005 prices
INDTBG measures Fiji’s goods trade balance with India defined as Fiji’s export of goods to India divided by import of goods from India in constant US dollars at 2005 prices
HKTBG measures Fiji’s goods trade balance with Hong Kong defined as Fiji’s export of goods to Hong Kong divided by import of goods from Hong Kong in constant US dollars at 2005 prices
FOODTB measures Fiji’s food sector trade balance defined as Fiji’s export of food divided by import of food in constant US dollars at 2005 prices
TRAVTB measures Fiji’s travel services sector trade balance defined as Fiji’s export of travel services divided by import of travel services in constant US dollars at 2005 prices
209
TRANTB measures Fiji’s transportation services sector trade balance defined as Fiji’s export of transportation services divided by import of transportation services in constant US dollars at 2005 prices
EXPG measures Fiji’s export of goods in constant US dollars at 2005 prices
DOMEXPG measures Fiji’s domestic export of goods in constant US dollars at 2005 prices
EXPS measures Fiji’s export of services in constant US dollars at 2005 prices
EXPGS measures Fiji’s export of goods and services in constant US dollars at 2005 prices. These include the value of all goods and other market services provided to the rest of the world. They include the value of merchandise, freight, insurance, transport, travel, royalties, license fees, and other services, such as communication, construction, financial, information, business, personal, and government services. They exclude compensation of employees and investment income and transfer payments.
AUSEXPG measures Fiji’s export of goods to Australia in constant US dollars at 2005 prices
NZEXPG measures Fiji’s export of goods to New Zealand in constant US dollars at 2005 prices
JPNEXPG measures Fiji’s export of goods to Japan in constant US dollars at 2005 prices
USAEXPG measures Fiji’s export of goods to USA in constant US dollars at 2005 prices
UKEXPG measures Fiji’s export of goods to UK in constant US dollars at 2005 prices
SINGEXPG measures Fiji’s export of goods to Singapore in constant US dollars at 2005 prices
CHNEXPG measures Fiji’s export of goods to China in constant US dollars at 2005 prices
MALAEXPG measures Fiji’s export of goods to Malaysia in constant US dollars at 2005 prices
INDEXPG measures Fiji’s export of goods to India in constant US dollars at 2005 prices
HKEXPG measures Fiji’s export of goods to Hong Kong in constant US dollars at 2005 prices
FOODEXP measures Fiji’s export of food in constant US dollars at 2005 prices
SUGEXP measures Fiji’s export of sugar in constant US dollars at 2005 prices
FISHEXP measures Fiji’s export of fish in constant US dollars at 2005 prices
GOLDEXP measures Fiji’s export of gold in constant US dollars at 2005 prices
210
TRAVEXP measures Fiji’s export of travel services in constant US dollars at 2005 prices
TRANEXP measures Fiji’s export of transportation services in constant US dollars at 2005 prices
IMPG measures Fiji’s import of goods in constant US dollars at 2005 prices
IMPS measures Fiji’s import of services in constant US dollars at 2005 prices
IMPGS measures Fiji’s import of goods and services in constant US dollars at 2005 prices. Imports of goods and services represent the value of all goods and other market services received from the rest of the world. They include the value of merchandise, freight, insurance, transport, travel, royalties, license fees, and other services, such as communication, construction, financial, information, business, personal, and government services. They exclude compensation of employees and investment income and transfer payments.
AUSIMPG measures Fiji’s import of goods from Australia in constant US dollars at 2005 prices
NZIMPG measures Fiji’s import of goods from New Zealand in constant US dollars at 2005 prices
JPNIMPG measures Fiji’s import of goods from Japan in constant US dollars at 2005 prices
USAIMPG measures Fiji’s import of goods from USA in constant US dollars at 2005 prices
UKIMPG measures Fiji’s import of goods from UK in constant US dollars at 2005 prices
SINGIMPG measures Fiji’s import of goods from Singapore in constant US dollars at 2005 prices
CHNIMPG measures Fiji’s import of goods from China in constant US dollars at 2005 prices
MALAIMPG measures Fiji’s import of goods from Malaysia in constant US dollars at 2005 prices
INDIMPG measures Fiji’s import of goods from India in constant US dollars at 2005 prices
HKIMPG measures Fiji’s import of goods from Hong Kong in constant US dollars at 2005 prices
FOODIMP measures Fiji’s import of food in constant US dollars at 2005 prices
FUELIMP measures Fiji’s import of fuel in constant US dollars at 2005 prices
MANUIMP measures Fiji’s import of manufactured goods in constant US dollars at 2005 prices
CRUDIMP measures Fiji’s import of crude oil in constant US dollars at 2005 prices
211
TEXTIMP measures Fiji’s import of textile in constant US dollars at 2005 prices
EQUIIMP measures Fiji’s import of machinery and transport equipment in constant US dollars at 2005 prices
TOBAIMP measures Fiji’s import of tobacco and beverage in constant US dollars at 2005 prices
CHEMIMP measures Fiji’s import of chemical in constant US dollars at 2005 prices
OILIMP measures Fiji’s import of oil and fats in constant US dollars at 2005 prices
MISCIMP measures Fiji’s import of miscellaneous manufactured goods in constant US dollars at 2005 prices
TRAVIMP measures Fiji’s import of travel services in constant US dollars at 2005 prices
TRANIMP measures Fiji’s import of transportation services in constant US dollars at 2005 prices
rE measures trade-weighted real effective exchange rate defined to show an increase as an devaluation of the Fijian currency expressed as an index of 2005=100
AUSrE , measures real exchange rate between Fiji and Australia defined to show an increase as an devaluation of the Fijian currency expressed as an index of 2005=100
NZrE , measures real exchange rate between Fiji and New Zealand defined to show an increase as an devaluation of the Fijian currency expressed as an index of 2005=100
JPNrE , measures real exchange rate between Fiji and Japan defined to show an increase as an devaluation of the Fijian currency expressed as an index of 2005=100
USArE , measures real exchange rate between Fiji and USA defined to show an increase as an devaluation of the Fijian currency expressed as an index of 2005=100
Y measures Fiji’s real GDP expressed in constant US dollars at 2005 prices
fY measures trade-weighted real GDP for Fiji’s trading partner countries expressed in constant US dollars at 2005 prices
AUSfY , measures real GDP for Australia expressed in constant US dollars at 2005 prices
NZfY , measures real GDP for New Zealand expressed in constant US dollars at 2005 prices
JPNfY , measures real GDP for Japan expressed in constant US dollars at 2005 prices
USAfY , measures real GDP for USA expressed in constant US dollars at 2005 prices
UKfY , measures real GDP for UK expressed in constant US dollars at 2005 prices
212
SINGfY , measures real GDP for Singapore expressed in constant US dollars at 2005 prices
CHNfY , measures real GDP for China expressed in constant US dollars at 2005 prices
MALAfY , measures real GDP for Malaysia expressed in constant US dollars at 2005 prices
INDfY , measures real GDP for India expressed in constant US dollars at 2005 prices
HKfY , measures real GDP for Hong Kong expressed in constant US dollars at 2005 prices
� represents variable taken in its first difference
COUP a dummy variable capturing the impact of political instability in 1987, 2000 and 2006 in Fiji. This is denoted by value 1 in the year of coups with the rest of the years taking a value of 0
213
Appendix D Results of unit root test for individual variables
Notes: 1. All variables are taken in its log-linear form.
2. Augmented Dickey-Fuller (ADF) test is applied for unit root testing
3. Appropriate attention is being made to the correct specification for the ADF tests in
terms of including in the test equation intercept, trend and intercept or none.
4. Schwarz Info Criterion (SIC) is being used for optimal lag selection. LL represents
optimal lag lengths for each variable included in the test are given in parenthesis.
5. The critical values for including intercept in the test equation are based on
MacKinnon (1996) which at 1%, 5% and 10% significance levels have values of -
3.627, -2.946 and -2.612 respectively. The null hypothesis for ADF tests is that a
series has a unit root (non-stationary).
6. Variables are taken in its change form for unit root test at first difference.
7. In summary, the major finding from the ADF test is that we are not able to reject the
unit root hypothesis at the conventional levels of significance; that is at 5%.
However, when the variables are taken in its first difference form, we are able to
reject the unit root null hypothesis at the 5% significance level. These results suggest
that all the variables employed in this study are integrated of order one, that is they
are I(1) in nature.
214
Variables ADF statistic p-valuesLevel (LL) First Difference
(LL)Level First Difference
TBG -1.944 (0) -8.126 (0) 0.309 0.000
TBS -2.737 (0) -5.593 (1) 0.078 0.000
TBGS -2.133 (0) -8.972 (0) 0.234 0.000
FOODTB 1.437 (2) -4.811 (1) 0.999 0.001
TRAVTB -2.498 (0) -6.880 (0) 0.327 0.000
TRANTB -2.288 (0) -5.600 (0) 0.182 0.000
AUSTBG -1.520 (0) -5.753 (0) 0.513 0.000
NZTBG -2.253 (1) -5.552 (0) 0.192 0.000
JPNTBG -2.326 (0) -6.461 (0) 0.169 0.000
USATBG -2.671 (1) -9.513 (0) 0.089 0.000
UKTBG -2.371 (0) -10.013 (0) 0.157 0.000
SINGTBG -0.823 (0) -7.185 (0) 0.353 0.000
CHNTBG -1.842 (1) -6.718 (1) 0.063 0.000
MALATBG -0.789 (0) -6.412 (0) 0.811 0.000
INDTBG -1.071 (1) -9.609 (0) 0.717 0.000
HKTBG -6.651 (2) -2.235 (0) 0.198 0.000
EXPG -1.557 (0) -6.691 (0) 0.494 0.000
DOMEXPG -1.758 (0) -6.511 (0) 0.395 0.000
EXPS -1.289 (0) -5.845 (1) 0.624 0.000
EXPGS -1.227 (0) -6.691 (0) 0.652 0.000
FOODEXP -1.740 (0) -5.691 (0) 0.402 0.000
SUGEXP 0.564 (2) -6.002 (1) 0.987 0.000
FISHEXP 0.859 (0) -4.931 (0) 0.891 0.000
GOLDEXP -2.868 (0) -7.350 (0) 0.184 0.000
215
TRAVEXP -0.832 (0) -7.573 (0) 0.798 0.000
TRANEXP -1.153 (0) -5.127 (0) 0.682 0.000
AUSEXPG -1.521 (0) -5.754 (0) 0.5122 0.000
NZEXPG -0.011 (2) -4.441 (1) 0.672 0.000
JPNEXPG 0.791 (0) -6.046 (1) 0.880 0.000
USAEXPG -2.959 (0) -5.689 (0) 0.157 0.000
UKEXPG -0.540 (0) -8.029 (0) 0.476 0.000
SINGEXPG -2.766 (0) -6.339 (0) 0.218 0.000
CHNEXPG 0.423 (2) -6.468 (1) 0.800 0.000
MALAEXPG -1.064 (0) -6.537 (0) 0.720 0.000
INDEXPG 0.018 (1) -9.891 (0) 0.954 0.000
HKEXPG -1.179 (0) -5.782 (1) 0.673 0.000
IMPG -1.202 (0) -7.010 (0) 0.663 0.000
IMPS -2.280 (0) -4.615 (0) 0.184 0.001
IMPGS -1.384 (0) -6.536 (0) 0.580 0.000
FOODIMP -0.151 (0) -6.020 (0) 0.936 0.000
FUELIMP -0.841 (0) -6.454 (0) 0.796 0.000
MANUIMP -2.023 (0) -5.405 (0) 0.276 0.000
CRUDIMP -1.506 (0) -8.872 (0) 0.520 0.000
TEXTIMP -1.661 (1) -4.015 (0) 0.440 0.004
EQUIIMP -2.015 (0) -5.557 (1) 0.279 0.000
TOBAIMP -1.507 (0) -7.732 (0) 0.519 0.000
CHEMIMP -1.255 (0) -5.287 (2) 0.640 0.000
OILIMP -2.503 (0) -6.029 (2) 0.123 0.000
MISCIMP -1.437 (0) -6.404 (0) 0.554 0.000
216
TRAVIMP -2.045 (0) -6.755 (0) 0.267 0.000
TRANIMP -0.655 (0) -5.823 (0) 0.844 0.000
AUSIMPG -2.024 (0) -5.646 (0) 0.276 0.000
NZIMPG -2.094 (0) -5.783 (0) 0.248 0.000
JPNIMPG -0.971 (0) -7.678 (0) 0.753 0.000
USAIMPG -2.568 (1) -10.292 (0) 0.109 0.000
UKIMPG -1.260 (1) -9.952 (0) 0.637 0.000
SINGIMPG -0.696 (0) -7.893 (0) 0.836 0.000
CHNIMPG -0.015 (1) -10.044 (0) 0.951 0.000
MALAIMPG -0.441 (2) -8.963 (0) 00891 0.000
INDIMPG -0.642 (1) -8.126 (0) 0.848 0.000
HKIMPG -2.012 (0) -7.539 (0) 0.281 0.000
rE -1.505 (1) -4.095 (0) 0.520 0.003
AUSrE , -1.381 (0) -6.136 (0) 0.581 0.000
NZrE , -1.774 (0) -6.448 (0) 0.387 0.000
JPNrE , -2.189 (1) -4.797 (0) 0.214 0.000
USArE , -2.110 (0) -5.703 (0) 0.242 0.000
Y -1.312 (1) -8.626 (0) 0.613 0.000fY -1.386 (1) -9.307 (0) 0.578 0.000
AUSfY , 0.203 (0) -5.467 (0) 0.969 0.000
NZfY , 1.020 (0) -5.807 (0) 0.996 0.000
JPNfY , -0.843 (0) -5.269 (0) 0.952 0.001
USAfY , -1.297 (1) -4.141 (0) 0.621 0.003
UKfY , -0.793 (1) -3.202 (0) 0.809 0.028
SINGfY , -1.826 (0) -5.051 (0) 0.363 0.000
217
CHNfY , 0.239 (6) -3.619 (3) 0.971 0.011
MALAfY , -1.801 (0) -4.933 (0) 0.374 0.000
INDfY , 2.343 (0) -5.755 (0) 0.998 0.000
HKfY , -2.997 (0) -5.262 (0) 0.147 0.001
218
Appendix E Results of cointegration test for various equations
Notes: 1. The unrestricted cointegration rank test are done for linear deterministic trend for
intercept (no trend) using the Maximum-Eigenvalue method.
2. For every respective trade balance models with four variables, the 5% critical Max-
Eigen Statistics at r = 0 is 27.584; r = 1 is 21.132; r=2 is 14.265; r=3 is 3.841.
3. For every respective export and import models with three variables, the 5% critical
Max-Eigen Statistics at r=0 is 21.132; r=1 is 14.265; r=2 is 3.841.
4. The values reported in the table are the observed Max-Eigen Statistics which are
compared with the 5% critical Max-Eigen Statistics to determine the number of
cointegration in each model.
5. P-values for the tests are obtained from MacKinnon-Haug-Michelis (1999) as
reported in the EViews 8 software. The observed P-value for each model is given in
parenthesis.
6. In summary, the major finding from the test reveals that there is existence of at least
one cointegrating equation, suggesting the presence of co-movements among the
variables and indicating long run stationarity in our models.
219
Null hypothesis 0�r (p-value)
1#r (p-value)
2#r (p-value)
Conclusion on
cointegration rank
Alternative hypothesis 1�r
2�r 3�r
Goods trade balance model 40.003 (0.001)
20.257 (0.066)
9.370 (0.257)
1
Services trade balance model
28.586 (0.037)
15.600 (0.249)
7.338 (0.450)
1
Goods and Services trade balance model
31.727 (0.014)
18.335 (0.118)
12.999 (0.079)
1
Fiji’s food sector trade balance model
39.102 (0.001)
11.764 (0.571)
1.706 (0.996)
1
Fiji’s travel services sector trade balance model
28.599 (0.037)
13.025 (0.450)
3.790 (0.881)
1
Fiji’s transportation services sector trade balance model
29.344 (0.029)
16.039 (0.223)
14.752 (0.042)
1
Fiji’s goods trade balance model with Australia
40.094 (0.001)
20.016 (0.071)
6.727 (0.522)
1
Fiji’s goods trade balance model with New Zealand
29.115 (0.032)
20.207 (0.067)
9.818 (0.224)
1
Fiji’s goods trade balance model with Japan
32.312 (0.011)
18.986 (0.097)
11.815 (0.118)
1
Fiji’s goods trade balance model with USA
27.852 (0.046)
16.377 (0.204)
6.031 (0.609)
1
Fiji’s goods trade balance model with UK
42.095 (0.000)
17.686 (0.142)
11.091 (0.150)
1
Fiji’s goods trade balance model with Singapore
39.759 (0.001)
20.643 (0.058)
9.522 (0.245)
1
Fiji’s goods trade balance model with China
35.673 (0.004)
16.084 (0.220)
9.949 (0.215)
1
Fiji’s goods trade balance model with Malaysia
34.741 (0.005)
20.301 (0.065)
15.502 (0.032)
1
Fiji’s goods trade balance model with India
34.629 (0.005)
19.826 (0.075)
15.541 (0.031)
1
220
Fiji’s goods trade balance model with Hong Kong
30.718 (0.019)
18.082 (0.127)
9.763 (0.228)
1
Goods export model
29.174 (0.003)
11.297 (0.140)
2.863 (0.091)
1
Domestic export model 30.599 (0.002)
8.146 (0.364)
1.211 (0.271)
1
Services export model 25.099 (0.013)
11.860 (0.116)
3.458 (0.063)
1
Goods and Services export model
24.735 (0.015)
13.979 (0.055)
3.545 (0.060)
1
Fiji’s export of food model 23.090 (0.026)
5.916 (0.624)
0.690 (0.406)
1
Fiji’s export of sugar model 29.947 (0.002)
5.717 (0.650)
0.881 (0.348)
1
Fiji’s export of fish model 24.312 (0.017)
7.930 (0.386)
0.013 (0.911)
1
Fiji’s export of gold model 26.589 (0.008)
9.819 (0.224)
4.385 (0.036)
1
Fiji’s export of travel services model
37.681 (0.000)
11.510 (0.130)
0.197 (0.657)
1
Fiji’s export of transportation services model
30.135 (0.002)
12.301 (0.100)
4.389 (0.036)
1
Fiji’s goods export model with Australia
33.012 (0.001)
10.462 (0.183)
0.284 (0.594)
1
Fiji’s goods export model with New Zealand
24.955 (0.014)
8.921 (0.293)
0.139 (0.709)
1
Fiji’s goods export model with Japan
40.353 (0.000)
6.931 (0.497)
6.108 (0.014)
1
Fiji’s goods export model with USA
50.774 (0.000)
7.589 (0.422)
0.278 (0.598)
1
Fiji’s goods export model with UK
24.064 (0.019)
10.561 (0.178)
0.456 (0.499)
1
Fiji’s goods export model with Singapore
24.442 7.441 2.118 1
221
(0.017) (0.438) (0.146)
Fiji’s goods export model with China
29.307 (0.003)
13.701 (0.061)
0.000 (0.999)
1
Fiji’s goods export model with Malaysia
23.049 (0.027)
12.089 (0.107)
1.102 (0.294)
1
Fiji’s goods export model with India
35.563 (0.000)
12.751 (0.086)
0.033 (0.856)
1
Fiji’s goods export model with Hong Kong
31.421 (0.001)
13.469 (0.067)
2.357 (0.125)
1
Goods import model 21.871 (0.039)
12.962 (0.080)
0.606 (0.436)
1
Services import model 53.425 (0.000)
10.060 (0.208)
6.953 (0.008)
1
Goods and Services import model
25.084 (0.013)
8.180 (0.361)
0.900 (0.343)
1
Fiji’s import of food model 23.055 (0.027)
11.258 (0.142)
6.083 (0.014)
1
Fiji’s import of fuel model 31.301 (0.001)
7.429 (0.440)
2.211 (0.137)
1
Fiji’s import of manufactured goods model
22.840 (0.029)
8.160 (0.363)
0.846 (0.358)
1
Fiji’s import of crude oil model
27.943 (0.005)
12.001 (0.111)
3.718 (0.054)
1
Fiji’s import of textiles model
23.070 (0.026)
4.707 (0.778)
3.266 (0.071)
1
Fiji’s import of machinery and transport equipment model
23.598 (0.022)
8.450 (0.335)
2.562 (0.109)
1
Fiji’s import of tobacco and beverage model
22.460 (0.032)
8.906 (0.294)
1.981 (0.159)
1
Fiji’s import of chemical model
29.723 (0.002)
5.643 (0.659)
0.970 (0.325)
1
Fiji’s import of oil and fats model
52.174 (0.000)
2.374 (0.980)
0.360 (0.548)
1
222
Fiji’s import of miscellaneous manufactured goods model
24.502 (0.016)
4.413 (0.813)
1.425 (0.233)
1
Fiji’s import of travel services model
23.133 (0.026)
13.720 (0.061)
5.724 (0.017)
1
Fiji’s import of transportation services model
22.901 (0.017)
8.776 (0.357)
2.738 (0.087)
1
Fiji’s goods import model with Australia
57.775 (0.000)
4.875 (0.758)
0.094 (0.759)
1
Fiji’s goods import model with New Zealand
29.869 (0.002)
11.909 (0.114)
2.785 (0.095)
1
Fiji’s goods import model with Japan
35.989 (0.000)
10.611 (0.175)
3.502 (0.061)
1
Fiji’s goods import model with USA
30.721 (0.002)
13.105 (0.076)
1.212 (0.271)
1
Fiji’s goods import model with UK
21.949 (0.038)
9.713 (0.231)
2.880 (0.090)
1
Fiji’s goods import model with Singapore
23.658 (0.022)
5.316 (0.702)
1.942 (0.164)
1
Fiji’s goods import model with China
28.536 (0.004)
5.869 (0.630)
0.614 (0.433)
1
Fiji’s goods import model with Malaysia
25.398 (0.012)
10.969 (0.156)
1.444 (0.230)
1
Fiji’s goods import model with India
25.987 (0.010)
10.524 (0.180)
2.501 (0.114)
1
Fiji’s goods import model with Hong Kong
37.803 (0.000)
11.944 (0.113)
1.454 (0.228)
1
223
Appendix F Results from Error Correction model (short-run dynamics)
Notes:
1. 1�tECT is the Error Correction term in the respective model.
2. � symbol in denotes the respective variable taken in its first difference.
3. Optimal lag lengths are chosen by SIC method.
4. Variables in the first column with numbers in subscript ( nt� ), shows the number of
lags taken in the respective model as suggested by the optimal lag selection method.
5. Standard errors are given in parenthesis (…).
6. (*), (**) and (***) denotes significance at the 10%, 5% and 1% level, respectively.
7. COUP is taken as dummy variable denoted by value 1 in the years of coup (1987,
2000 and 2006) capturing the impact of political instability with the rest of the years
taking a value of 0.
8. R2 shows the R-squared statistics to determine the how well model fits the data.
9. Adjusted R2 shows the adjusted R-squared statistics in a multiple regression analysis
by taking into account the number of explanatory variables in the model. In VECM, it
is the adjusted R2 which is generally used to describe the fit of the model.
10. ! represents the standard error of equation.
11. NX 2 shows the Jarque-Bera statistics for Normality tests.
12. HetX 2 shows the Chi-Squared statistics for Heteroskedasticity tests.
13. LM )(SCtest shows the LM statistics for Serial Correlation (SC) tests.
14. The p-values for the diagnostic tests ( NX 2, HetX 2
and LM )(SCtest ) are in
brackets […] and a value greater than 5% indicates, the model passing the particular
diagnostic test.
15. AR roots graph have been verified for test on model stability.
224
Part 1 – Summary Results Table SR1: Short–run coefficient estimates of real exchange rates in Fiji’s sectoral
exports model
Short-run results Food Sugar Fish Gold Travel Transport rtE 1ln �� -0.374
(0.485) -1.871 (0.973)*
0.890 (0.858)
10.543 (0.3.731)***
0.895 (0.499)*
0.620 (0.587)
rtE 2ln �� -0.220
(0.449) -0.062 (1.174)
1.679 (2.662)
0.925 (0.497)*
-0.119 (0.630)
rtE 3ln �� -0.196
(0.592) -1.977 (1.115)*
3.501 (3.004)
1.070 (0.587)*
1.220 (0.680)*
rtE 4ln �� -1.004
(0.780) -0.255
(3.545) 0.953
(0.536)* -1.938
(0.748)** rtE 5ln �� -2.127
(0.711)*** -2.554
(3.246) -0.253 (0.507)
rtE 6ln �� -0.658
(0.738) 5.740
(2.779)** 1.057
(0.449)**
rtE 7ln �� 6.181
(3.966)
COUP 0.172 (0.076)**
0.044 (0.129)
0.248 (0.103)**
-2.058 (0.574)***
-0.181 (0.071)**
-0.201 (0.103)*
Diagnostics
1�tECT -0.199 (0.053)***
-0.267 (0.141)*
-0.744 (0.117)***
0.450 (0.150)***
-0.440 (0.226)*
-1.044 (0.234)***
2R 0.884 0.437 0.642 0.928 0.865 0.842
Adjusted 2R 0.497 0.155 0.582 0.584 0.594 0.673
! 0.089 0.238 0.267 0.485 0.083 0.132
NX 2
0.803 [0.669]
3.046 [0.218]
1.922 [0.383]
0.0857 [0.958]
0.354 [0.838]
0.472 [0.790]
HetX 2
26.247 [0.505]
25.455 [0.228]
15.454 [0.079]
27.298 [0.502]
29.081 [0.357]
17.848 [0.659]
LM )(SCTest 12.504 [0.186]
14.065 [0.120]
9.072 [0.431]
6.225 [0.717]
12.083 [0.209]
4.791 [0.852]
AR roots graph stable stable stable stable stable stable
225
Table SR2: Short–run coefficient estimates of real exchange rates in Fiji’s sectoral imports model
Short-run results Food FuelManufactured
GoodsCrude
Oil Textiles Equipment
rtE 1ln �� -0.697
(0.521)5.485
(0.865)***-1.178(0.621)*
-0.400(0.787)
2.338(0.698)***
-1.178(0.621)*
rtE 2ln �� 0.144
(0.499)2.843
(1.328)**1.906
(0.682)***rtE 3ln �� 0.337
(0.456)0.213(1.239)
rtE 4ln �� -1.073
(0.481)**0.753(1.065)
rtE 5ln �� -0.470
(0.582)-2.468
(1.194)**rtE 6ln �� 0.688
(0.536)1.188(1.468)
rtE 7ln �� 2.099
(0.855)**-0.210(0.107)*
COUP 0.079(0.074)
-0.321(0.116)***
-0.095(0.102)
0.106(0.107)
-0.095(0.102)
Diagnostics
1�tECT -0.145(0.076)*
-0.827(0.336)**
-0.214(0.057)***
-0.363(0.154)**
-0.054(0.028)*
-0.943(0.180)***
2R 0.768 0.947 0.585 0.339 0.508 0.585
Adjusted 2R 0.304 0.744 0.515 0.229 0.321 0.515
! 0.094 0.114 0.160 0.223 0.170 0.160
NX 2 0.959[0.619]
2.224[0.329]
1.150[0.563]
1.386[0.500]
0.885[0.642]
1.401[0.496]
HetX 2 25.232[0.562]
29.342[0.345]
4.633[0.865]
6.022[0.738]
12.192[0.968]
8.644[0.471]
LM )(SCTest 10.849[0.286]
11.493[0.243]
7.751[0.560]
7.747[0.560]
5.745[0.765]
7.742[0.560]
AR roots graph stable stable stable stable stable stable
226
Table SR3: Short–run coefficient estimates of real exchange rates in Fiji’s sectoral imports model
Short-run results Tobacco and Beverage Chemical Oil and
Fats Miscellaneous
Goods Travel Transport
rtE 1ln �� 4.417
(0.680)*** 0.142 (0.370)
-1.642 (0.830)*
0.022 (0.615)
-0.928 (0.927)
0.290 (0.812)
rtE 2ln �� 0.618
(0.866) -0.511
(0.771) 1.381
(0.622)** -1.849 (0.969)*
-0.079 (0.764)
rtE 3ln �� 1.754
(1.646)*** 0.588
(0.722) 1.644
(0.519)*** -1.886
(0.860)** -1.444 (0.860)*
rtE 4ln �� 1.101
(0.640)* -2.080
(0.813)** -2.974
(0.924)*** -0.713 (0.685)
rtE 5ln �� 0.487
(0.582) -0.956
(0.698)
rtE 6ln �� 2.611
(0.596)***
rtE 7ln �� 1.201
(0.592)**
COUP -0.270 (0.114)**
0.036 (0.044)
0.173 (0.096)*
-0.196 (0.078)**
0.522 (0.128)***
-0.132 (0.083)
Diagnostics 1�tECT -0.581
(0.165)*** -0.492
(0.135)*** -0.542
(0.257)** -1.016
(0.401)** -0.335
(0.153)** -0.695
(0.208)*** 2R 0.931 0.517 0.650 0.759 0.784 0.623
Adjusted 2R 0.669 0.436 0.378 0.633 0.521 0.217
! 0.098 0.110 0.172 0.138 0.144 0.140 NX 2
1.043 [0.594]
5.038 [0.081]
0.486 [0.784]
1.836 [0.399]
0.225 [0.894]
2.153 [0.341]
HetX 2
28.511 [0.385]
5.437 [0.795]
20.460 [0.811]
31.858 [0.237]
31.834 [0.238]
24.760 [0.258]
LM )(SCTest 7.741 [0.560]
7.138 [0.623]
12.290 [0.198]
6.599 [0.679]
7.310 [0.605]
7.913 [0.543]
AR roots graph stable stable stable stable stable stable
227
Table SR4: Short–run coefficient estimates of real exchange rates in Fiji’s major bilateral trade partners export model
Short-run results Australia New Zealand
Japan USA UK
rtE 1ln �� -0.842
(0.794) -0.309 (0.351)
-0.777 (0.390)*
-1.781 (0.651)**
-0.579 (0.683)
rtE 2ln �� -1.908
(0.801)** 0.348 (0.303)
1.019 (0.379)**
1.249 (0.826)
rtE 3ln �� -1.968
(0.679)*** 0.382
(0.453) 0.870
(0.802) rtE 4ln �� -1.618
(0.612)** 2.129
(1.086) rtE 5ln �� 3.905
(1.076)*** COUP 0.165
(0.116) 0.002 (0.091)
-0.201 (0.172)
0.087 (0.196)
0.524 (0.129)
Diagnostics
1�tECT -0.308 (0.141)**
-0.790 (0.208)***
-0.913 (0.136)***
-0.369 (0.101)***
-0.771 (0.164)***
2R 0.536 0.625 0.867 0.363 0.856
Adjusted 2R 0.175 0.510 0.801 0.256 0.681
! 0.218 0.197 0.252 0.314 0.197
NX 2
0.388 [0.824]
2.691 [0.260]
0.252 [0.882]
0.232 [0.891]
4.336 [0.114]
HetX 2
29.999 [0.314]
12.510 [0.640]
23.275 [0.330]
12.045 [0.211]
10.254 [0.975]
LM )(SCTest 6.752 [0.663]
10.478 [0.313]
8.400 [0.494]
5.181 [0.818]
11.764 [0.227]
AR roots graph stable stable stable stable stable
228
Table SR5: Short–run coefficient estimates of real exchange rates in Fiji’s major bilateral trade partners import model
Short-run results Australia New Zealand
Japan USA UK
rtE 1ln �� 0.528
(0.315) -0.263 (0.153)*
-0.670 (0.235)***
-0.446 (0.950)
-3.367 (1.197)***
rtE 2ln �� 0.252
(0.333) -0.379
(0.154)** -0.173 (0.288)
-0.431 (0.895)
0.683 (1.291)
rtE 3ln �� -0.325
(0.359) 0.014 (0.135)
0.435 (1.129)
rtE 4ln �� -0.684
(0.468) -0.219 (0.140) -2.036
(1.309) rtE 5ln �� -0.017
(0.504) -0.597
(0.166)*** -1.482 (1.231)
COUP 0.178 (0.101)*
0.123 (0.036)
-0.024 (0.081)
-0.355 (0.251)
-0.075 (0.155)
Diagnostics
1�tECT -0.513 (0.239)**
-0.823 (0.161)***
-0.589 (0.161)***
-0.686 (0.231)***
-0.624 (0.307)*
2R 0.608 0.876 0.515 0.470 0.736 Adjusted 2R 0.132 0.725 0.365 0.307 0.414
! 0.150 0.065 0.153 0.451 0.238 NX 2
0.910 [0.635]
0.058 [0.971]
0.439 [0.803]
2.043 [0.360]
0.924 [0.630]
HetX 2
31.705 [0.243]
24.595 [0.597]
19.727 [0.183]
8.938 [0.881]
30.358 [0.298]
LM )(SCTest 4.929 [0.841]
6.628 [0.676]
4.302 [0.890]
13.393 [0.146]
6.904 [0.647]
AR roots graph stable stable stable stable stable
229
Table SR6: Short–run coefficient estimates of real exchange rates in Fiji’s emerging Asian bilateral trade partners export model
Short-run results Singapore China Malaysia India Hong Kong
rtE 1ln �� 2.939
(3.808) 3.581 (5.306)
0.811 (3.390)
3.723 (2.405)
2.091 (0.701)**
rtE 2ln �� 3.632
(5.086) 3.378 (3.047)
3.951 (2.595)
rtE 3ln �� 2.630
(5.790) 4.089 (2.602)
rtE 4ln �� 1.825
(6.303)
rtE 5ln �� 10.935
(6.460)*
rtE 6ln �� 7.587
(5.896)
COUP -1.223 (0.738)*
-0.383 (0.655)
-0.524 (0.478)
-0.158 (0.304)
0.300 (0.165)*
Diagnostics
1�tECT -0.307 (0.167)*
-0.790 (0.304)**
-0.140 (0.074)*
-0.644 (0.131)***
0.745 (0.595)
2R 0.216 0.872 0.408 0.696 0.696
Adjusted 2R 0.086 0.616 0.225 0.544 0.644
! 1.199 1.158 0.763 0.657 0.326
NX 2
1.561 [0.458]
5.567 [0.062]
0.420 [0.811]
1.159 [0.560]
1.349 [0.509]
HetX 2
11.080 [0.270]
27.966 [0.413]
13.867 [0.536]
17.434 [0.685]
23.615 [0.652]
LM )(SCTest 4.614 [0.867]
6.246 [0.715]
2.553 [0.979]
7.986 [0.536]
10.626 [0.302]
AR roots graph stable stable stable stable stable
230
Table SR7: Short–run coefficient estimates of real exchange rates in Fiji’s emerging Asian bilateral trade partners import model
Short-run results Singapore China Malaysia India Hong Kong
rtE 1ln �� -1.925
(2.056) -0.784 (1.730)
-4.796 (1.899)**
0.084 (1.016)
-4.962 (2.621)*
rtE 2ln �� -4.768
(2.354)** -1.114 (1.657)
-3.549 (1.681)**
3.449 (1.162)***
-2.252 (2.343)
rtE 3ln �� -4.470*
(2.312) -2.958 (1.658)*
-3.751 (1.277)***
0.521 (1.210)
0.511 (1.857)
rtE 4ln �� -2.522
(1.948) -0.497 (1.598)
-2.526 (1.152)**
-0.326 (1.080)
0.471 (1.188)
rtE 5ln �� 0.076
(1.936) -4.142
(1.661)** -4.243
(1.276)** rtE 6ln �� 1.166
(2.108) -1.074 (1.563)
-2.859 (1.220)**
rtE 7ln �� -1.945
(1.736)
COUP 0.123 (0.228)
-0.052 (0.117)
-0.088 (0.132)
-0.175 (0.152)
0.445 (0.187)**
Diagnostics
1�tECT -0.214 (0.104)**
-0.610 (0.321)
-0.756 (0.222)***
-0.786 (0.363)**
-1.637 (0.645)**
2R 0.838 0.865 0.818 0.529 0.875 Adjusted 2R 0.218 0.595 0.455 0.163 0.771
! 0.370 0.192 0.178 0.235 0.226 NX 2
0.028 [0.986]
0.468 [0.792]
0.825 [0.662]
1.393 [0.498]
0.096 [0.953]
HetX 2
27.327 [0.446]
30.602 [0.288]
24.964 [0.577]
30.208 [0.305]
30.021 [0.313]
LM )(SCTest 5.087 [0.827]
16.649 [0.055]
5.656 [0.774]
9.405 [0.401]
10.539 [0.309]
AR roots graph stable stable stable stable stable
231
Part 2 – Supplementary Appendices Table SA1: Short–run coefficient estimates of Fiji’s goods trade balance model, 1975–
2012
Short-run results TBGln� rEln� Yln� fYln�
1�tECT -0.527 (0.300)*
0.052 (0.112)
-0.021 (0.094)
-0.847 (0.202)***
1ln �� tTBG 0.447 (0.441)
-0.001 (0.164)
-0.116 (0.138)
0.669 (0.297)**
2ln �� tTBG 0.786 (0.457)*
0.366 (0.170)**
-0.065 (0.143)
0.349 (0.308)
3ln �� tTBG 0.260 (0.397)
0.249 (0.148)
-0.042 (0.124)
-0.144 (0.267)
4ln �� tTBG -0.058 (0.323)
0.066 (0.120)
0.039 (0.101)
-0.157 (0.218)
rtE 1ln �� -0.953
(0.940) 0.275 (0.350)
0.032 (0.293)
-2.051 (0.633)***
rtE 2ln �� -2.392
(1.193)* -0.600 (0.444)
0.338 (0.372)
-1.377 (0.803)*
rtE 3ln �� -1.209
(0.892) -0.682
(0.332)** 0.053 (0.278)
-0.152 (0.600)
rtE 4ln �� 0.088
(0.828) 0.367 (0.308)
-0.139 (0.258)
0.172 (0.558)
1ln �� tY 1.488 (0.946)
-0.003 (0.353)
-0.537 (0.295)*
1.175 (0.637)*
2ln �� tY -0.158 (0.934)
0.336 (0.348)
-0.036 (0.291)
0.704 (0.629)
3ln �� tY -1.296 (0.921)
-0.584 (0.343)*
0.127 (0.287)
-0.560 (0.620)
4ln �� tY -1.606 (0.842)*
-0.459 (0.313)
0.108 (0.263)
-0.272 (0.567)
ftY 1ln �� -0.215
(0.316) 0.012 (0.118)
0.125 (0.098)
-0.794 (0.212)***
ftY 2ln �� -0.505
(0.468) -0.179 (0.146)
0.111 (0.146)
-0.862 (0.315)**
ftY 3ln �� -0.422
(0.340) -0.157 (0.127)
0.035 (0.106)
-0.529 (0.229)**
ftY 4ln �� 0.091
(0.255) 0.037 (0.095)
-0.067 (0.079)
-0.142 (0.171)
C 0.082 (0.054)
0.020 (0.020)
0.025 (0.017)
0.073 (0.036)*
COUP 0.117 (0.077)
0.074 (0.029)**
-0.036 (0.024)
0.106 (0.052)**
Diagnostics 2R 0.842 0.712 0.528 0.842
232
Adjusted 2R 0.640 0.343 -0.078 0..640
! 0.135 0.050 0.042 0.091
NX 2
0.559 [0.756]
HetX 2
26.650 [0.483]
LM )(SCTest 13.246 [0.655]
AR roots graph stable
233
Table SA2: Short–run coefficient estimates of Fiji’s services trade balance model, 1975–2012
Short-run results TBSln� rEln� Yln� fYln�
1�tECT -0.150 (0.081)*
-0.036 (0.041)
0.060 (0.030)*
0.292 (0.079)***
1ln �� tTBS 0.062 (0.194)
-0.109 (0.098)
0.100 (0.072)
-0.094 (0.189)
2ln �� tTBS -0.281 (0.160)
0.036 (0.081)
-0.028 (0.059)
0.024 (0.156)
rtE 1ln �� 0.410
(0.512) 0.213 (0.259)
-0.163 (0.190)
-1.203 (0.499)**
rtE 2ln �� 0.447
(0.454) 0.111 (0.230)
-0.072 (0.168)
0.010 (0.443)
1ln �� tY -0.209 (0.598)
-0.579 (0.302)*
-0.542 (0.221)**
-0.540 (0.583)
2ln �� tY -0.390 (0.621)
-0.099 (0.314)
-0.075 (0.230)
-0.213 (0.605)
ftY 1ln �� -0.186
(0.174) 0.009 (0.088)
0.108 (0.064)
-0.355 (0.170)**
ftY 2ln �� 0.066
(0.164) 0.033 (0.083)
0.065 (0.061)
-0.132 (0.160)
C 0.030 (0.030)
0.010 (0.015)
0.030 (0.011)**
0.065 (0.030)**
COUP -0.284 (0.074)***
0.090 (0.038)**
-0.016 (0.028)
0.043 (0.072)
Diagnostics 2R 0.557 0.434 0.394 0.650
Adjusted 2R 0.373 0.198 0.142 0.504
! 0.107 0.054 0.040 0.104
NX 2
0.609 [0.737]
HetX 2
20.647 [0.357]
LM )(SCTest 23.392 [0.104]
AR roots graph stable Table SA3: Short–run coefficient estimates of Fiji’s goods and services trade balance
model, 1975–2012
234
Short-run results TBGSln� rEln� Yln� fYln�
1�tECT -0.723 (0.360)*
0.808 (0.313)**
-0.656 (0.193)***
-1.429 (0.912)
1ln �� tTBGS -0.498 (0.293)*
-0.651 (0.255)**
0.454 (0.157)***
0.635 (0.743)
2ln �� tTBGS -0.175 (0.194)
-0.236 (0.168)
0.302 (0.104)***
0.166 (0.491)
rtE 1ln �� -0.050
(0.253) -0.129 (0.220)
-0.137 (0.136)
-0.177 (0.642)
rtE 2ln �� -0.165
(0.231) -0.011 (0.200)
0.037 (0.124)
-0.127 (0.585)
1ln �� tY 0.854 (0.528)
-1.278 (0.458)***
-0.220 (0.283)
1.498 (1.337)
2ln �� tY 0.121 (0.439)
-0.495 (0.381)
0.076 (0.236)
1.030 (1.113)
ftY 1ln �� -0.112
(0.088) 0.089 (0.077)
-0.005 (0.047)
-0.944 (0.224)***
ftY 2ln �� -0.061
(0.096) 0.046 (0.083)
-0.020 (0.051)
-0.414 (0.242)*
C 0.012 (0.015)
-0.008 (0.013)
0.010 (0.008)
0.013 (0.037)
COUP -0.046 (0.038)
0.038 (0.033)
-0.065 (0.021)***
-0.084 (0.097)
Diagnostics 2R 0.754 0.425 0.789 0.563
Adjusted 2R 0.648 0.174 0.698 0.373
! 0.074 0.064 0.040 0.187
NX 2
0.615 [0.735]
HetX 2
14.370 [0.762]
LM )(SCTest 20.282 [0.208]
AR roots graph stable
235
Table SA4: Short–run coefficient estimates of Fiji’s food sector trade balance model, 1975–2012
Short-run results FOODTBln� rEln� Yln� fYln�
1�tECT -0.190 (0.078)**
-0.060 (0.036)
0.017 (0.021)
0.050 (0.077)
FOODtTB 1ln �� -0.316
(0.315) 0.020 (0.146)
-0.089 (0.084)
-0.085 (0.310)
FOODtTB 2ln �� -0.258
(0.219) 0.104 (0.101)
0.038 (0.058)
-0.149 (0.216)
FOODtTB 3ln �� -0.258
(0.188) 0.045 (0.087)
-0.076 (0.050)
0.051 (0.185)
rtE 1ln �� 1.599
(0.725)** 0.435 (0.335)
0.089 (0.193)
-0.982 (0.713)
rtE 2ln �� 0.405
(0.693) 0.179 (0.321)
0.094 (0.185)
0.438 (0.682)
rtE 3ln �� -0.268
(0.639) -0.203 (0.296)
-0.065 (0.170)
0.917 (0.629)
1ln �� tY 2.493 (1.162)**
0.230 (0.538)
-0.323 (0.310)
-0.914 (1.144)
2ln �� tY 3.049 (1.479)**
0.865 (0.684)
-0.044 (0.395)
-1.333 (1.456)
3ln �� tY 2.046 (1.031)*
0.445 (0.477)
0.001 (0.275)
-0.690 (1.015)
ftY 1ln �� 0.371
(0.302) 0.149 (0.140)
0.027 (0.081)
-0.800 (0.215)**
ftY 2ln �� 0.235
(0.301) 0.041 (0.139)
0.027 (0.081)
-0.210 (0.296)
ftY 3ln �� -0.182
(0.243) -0.089 (0.113)
0.021 (0.065)
-0.006 (0.240)
-0.228 (0.085)**
-0.013 (0.039)
0.021 (0.023)
0.063 (0.083)
COUP -0.076 (0.091)
0.003 (0.042)
-0.031 (0.024)
0.118 (0.090)
Diagnostics 0.535 0.478 0.543 0.606
0.101 -0.010 0.116 0.238 0.136 0.063 0.036 0.134
0.477 [0.788]
29.500 [0.337]
28.268 [0.928]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
236
Table SA5: Short–run coefficient estimates of Fiji’s travel services sector trade balance model, 1975–2012
Short-run results TRAVTBln� rEln� Yln� fYln�
1�tECT -1.847 (0.596)***
-0.046 (0.198)
-0.014 (0.119)
0.973 (0.278)***
TRAVtTB 1ln �� 1.155
(0.533)** -0.018 (0.177)
0.047 (0.107)
-0.799 (0.248)***
TRAVtTB 2ln �� 1.134
(0.487)** -0.047 (0.162)
0.027 (0.097)
-0.590 (0.227)**
TRAVtTB 3ln �� 0.781
(0.350)** -0.083 (0.116)
-0.079 (0.070)
-0.420 (0.163)**
TRAVtTB 4ln �� 0.821
(0.361)** -0.020 (0.120)
0.079 (0.072)
-0.377 (0.168)**
TRAVtTB 5ln �� 0.825
(0.308)** -0.073 (0.102)
-0.023 (0.062)
-0.276 (0.144)*
rtE 1ln �� 3.663
(2.011)* 0.387 (0.667)
0.202 (0.402)
-3.314 (0.937)***
rtE 2ln �� 7.444
(2.483)*** -0.275 (0.824)
0.313 (0.497)
-2.921 (1.156)**
rtE 3ln �� 6.174
(1.618)*** -0.428 (0.537)
-0.125 (0.324)
-1.544 (0.754)**
rtE 4ln �� 2.994
(1.390)** 0.004 (0.461)
0.042 (0.278)
-0.680 (0.647)
rtE 5ln �� 0.868
(1.251) 0.110 (0.415)
0.084 (0.250)
-0.565 (0.583)
1ln �� tY -5.752 (1.577)***
0.040 (0.523)
-0.315 (0.315)
0.400 (0.735)
2ln �� tY -2.041 (1.789)
0.281 (0.594)
0.282 (0.357)
-0.658 (0.833)
3ln �� tY 2.544 (1.960)
-0.279 (0.650)
0.185 (0.392)
-1.639 (0.913)*
4ln �� tY 3.682 (1.654)**
-0.589 (0.548)
-0.251 (0.331)
-1.079 (0.770)
5ln �� tY 0.576 (1.802)
-0.114 (0.598)
0.376 (0.360)
-0.148 (0.839)
ftY 1ln �� 0.329
(0.461) -0.013 (0.153)
0.148 (0.092)
-0.836 (0.215)***
ftY 2ln �� 1.327
(0.580)** -0.173 (0.193)
0.062 (0.116)
-0.610 (0.270)**
ftY 3ln �� 0.845
(0.558) -0.185 (0.185)
-0.037 (0.112)
-0.149 (0.260)
ftY 4ln �� 0.482
(0.553) -0.007 (0.183)
-0.002 (0.111)
0.179 (0.257)
ftY 5ln �� 1.068
(0.453)** -0.175 (0.150)
-0.036 (0.091)
0.247 (0.211)
237
C -0.183 (0.127)
0.041 (0.042)
0.014 (0.025)
0.083 (0.059)
COUP -0.447 (0.188)**
0.006 (0.062)
-0.036 (0.038)
0.336 (0.088)***
Diagnostics 2R 0.834 0.585 0.661 0.870
Adjusted 2R 0.429 -0.430 -0.169 0.554
! 0.220 0.073 0.044 0.103 NX 2
0.698 [0.730]
HetX 2
28.825 [0.369]
LM )(SCTest 11.010 [0.809]
AR roots graph stable
238
Table SA6: Short–run coefficient estimates of Fiji’s transport services sector trade balance model, 1975–2012
Short-run results TRANTBln� rEln� Yln� fYln�
1�tECT -0.385 (0.178)**
-0.033 (0.098)
0.067 (0.049)
0.256 (0.171)
TRANtTB 1ln �� 0.251
(0.211) 0.009 (0.116)
-0.014 (0.059)
-0.027 (0.203)
TRANtTB 2ln �� 0.083
(0.171) -0.076 (0.094)
-0.034 (0.048)
-0.200 (0.164)
TRANtTB 3ln �� 0.023
(0.178) 0.044 (0.098)
-0.058 (0.049)
0.111 (0.170)
rtE 1ln �� 0.868
(0.570) 0.309 (0.314)
-0.099 (0.158)
-0.946 (0.546)*
rtE 2ln �� 0.998
(0.564) -0.041 (0.311)
0.042 (0.157)
0.114 (0.540)
rtE 3ln �� 1.362
(0.506)** -0098 (0.279)
-0.033 (0.140)
0.848 (0.484)*
1ln �� tY 0.510 (0.905)
-0.170 (0.498)
-0.648 (0.251)**
-1.164 (0.867)
2ln �� tY 1.120 (1.101)
0.296 (0.607)
-0.093 (0.306)
-1.314 (0.055)
3ln �� tY 2.954 (0.802)***
-0.041 (0.442)
0.789 (0.223)
-0.719 (0.768)
ftY 1ln �� -0.654
(0.328)** -0.001 (0.091)
0.093 (0.091)
-0.415 (0.314)
ftY 2ln �� -0.388
(0.290) -0.051 (0.159)
0.074 (0.080)
-0.100 (0.277)
ftY 3ln �� -0.363
(0.213)* -0.123 (0.117)
0.041 (0.059)
0.021 (0.204)
-0.072 (0.047)
0.009 (0.026)
0.038 (0.013)***
0.081 (0.045)*
COUP -0.097 (0.076)
0.034 (0.042)
-0.037 (0.021)*
0.091 (0.073)
Diagnostics 0.700 0.393 0.559 0.694
0.399 -0.213 0.118 0.389
0.127 0.070 0.035 0.122
4.178 [0.124]
28.189 [0.401]
16.714 [0.404]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
239
Table SA7: Short–run coefficient estimates of Fiji’s Goods Trade Balance model with Australia, 1975–2012
Short-run results AUSTBGln� AUSrE ,ln� Yln� AUSfY ,ln�
1�tECT -0.937 (0.295)***
-0.089 (0.142)
-0.066 (0.045)
0.006 (0.023)
AUStTBG 1ln �� -0.166
(0.232) 0.265
(0.112)** 0.004 (0.043)
-0.009 (0.018)
AUStTBG 2ln �� -0.250
(0.234) 0.275
(0.113)** 0.013 (0.043)
0.029 (1.569)
AUStTBG 3ln ��
-0.397 (0.180)**
0.033 (0.087)
0.026 (0.033)
0.014 (0.014)
AUSrt
E ,1
ln�
� 2.115 (0.775)**
-0.941 (0.374)**
0.189 (0.143)
0.004 (0.062)
AUSrt
E ,2
ln�
� 1.131 (0.784)
-0.938 (0.379)**
0.197 (0.145)
-0.031 (0.062)
AUSrt
E ,3
ln�
� 0.648 (0.568)
-0.448 (0.274)
0.096 (0.105)
-0.054 (0.019)
1ln �� tY 6.924 (2.974)**
-0.190 (1.436)
-0.098 (0.549)
0.033 (0.237)
2ln �� tY 2.598 (2.476)
-0.828 (1.195)
-0.076 (0.457)
-0.134 (0.197)
3ln �� tY 1.125 (1.250)
-0.475 (0.603)
0.030 (0.231)
-0.086 (0.010)
AUSftY ,
1ln �� 2.223 (2.930)
2.994 (1.415)**
-1.000 (0.541)*
-0.537 (0.234)**
AUSftY ,
2ln �� 1.062 (2.694)
4.057 (1.301)***
-0.048 (0.498)
-0.250 (0.215)
AUSftY ,
3ln �� 4.273 (2.631)
3.944 (1.270)***
-1.412 (0.486)***
-0.546 (0.210)**
C 0.078 (0.049)
-0.020 (0.024)
-0.002 (0.009)
-0.002 (0.004)
COUP -0.251 (0.112)
0.065 (0.054)
-0.009 (0.021)
0.006 (0.009)
Diagnostics 2R 0.800 0.739 0.828 0.632
Adjusted 2R 0.646 0.536 0.694 0.346 ! 0.213 0.103 0.039 0.017
NX 2
0.214 [0.899]
HetX 2
28.183 [0.455]
LM )(SCTest 19.924 [0.224]
AR roots graph stable
240
Table SA8: Short–run coefficient estimates of Fiji’s goods trade balance model with New Zealand, 1975–2012
Short-run results NZTBGln� NZrE ,ln� Yln� NZfY ,ln�
1�tECT -0.968 (0.228)***
-0.250 (0.151)
-0.024 (0.045)
0.056 (0.015)
NZtTBG 1ln �� 0.680
(0.246)*** 0.428
(0.164)** -0.017 (0.048)
0.077 (0.016)***
NZtTBG 2ln �� 0.400
(0.215) 0.060 (0.143)
-0.003 (0.042)
-0.072 (0.014)***
NZtTBG 3ln ��
0.199 (0.199)
0.098 (0.132)
-0.095 (0.039)**
-0.070 (0.013)***
NZrtE ,
1ln �� -0.569 (0.369)
-0.338 (0.245)
-0.040 (0.072)
0.098 (0.024)***
NZrtE ,
2ln �� -0.248 (0.414)
-0.425 (0.275)
0.056 (0.081)
0.084 (0.027)***
NZrtE ,
3ln �� -0.135 (0.301)
-0.555 (0.200)***
0.163 (0.059)***
0.063 (0.019)***
1ln �� tY 2.892 (1.268)**
1.148 (0.842)
-0.280 (0.248)
-0.078 (0.081)
2ln �� tY 0.759 (1.354)
0.013 (0.899)
0.240 (0.265)
-0.091 (0.087)
3ln �� tY 2.419 (1.134)**
1.290 (0.752)*
0.176 (0.222)
-0.040 (0.073)
NZftY ,
1ln �� 3.143 (2.998)
-1.180 (1.990)
-0.263 (0.586)
-0.393 (0.192)*
NZftY ,
2ln �� -2.030 (2.419)
0.442 (1.606)
-0.464 (0.473)
0.308 (0.155)*
NZftY ,
3ln �� -3.408 (2.135)
-0.487 (1.417)
-0.283 (0.417)
0.252 (0.137)*
C -0.042 (0.075)
0.040 (0.050)
0.030 (0.015)**
0.010 (0.005)**
COUP -0.009 (0.082)
0.003 (0.054)
-0.002 (0.016)
0.009 (0.005)*
Diagnostics 2R 0.726 0.525 0.605 0.781
Adjusted 2R 0.524 0.176 0.314 0.619 ! 0.184 0.122 0.036 0.012
NX 2
0.876 [0.645]
HetX 2
32.380 [0.218]
LM )(SCTest 21.120 [0.174]
AR roots graph stable
241
Table SA9: Short–run coefficient estimates of Fiji’s goods trade balance model with Japan, 1975–2012
Short-run results JPNTBGln� JPNrE ,ln� Yln� JPNfY ,ln�
1�tECT -0.899 (0.211)***
-0.091 (0.085)
-0.016 (0.023)
0.013 (0.012)
JPNtTBG 1ln �� 0.407
(0.214)* 0.013 (0.086)
0.002 (0.023)
-0.002 (0.012)
JPNtTBG 2ln �� -0.042
(0.159) 0.004 (0.064)
0.005 (0.017)
-0.004 (0.009)
JPNtTBG 3ln ��
0.233 (0.147)
0.036 (0.059)
-0.032 (0.016)*
-0.001 (0.008)
JPNrtE ,
1ln �� 0.745 (0.595)
0.125 (0.240)
-0.008 (0.065)
-0.047 (0.034)
JPNrtE ,
2ln �� 1.932 (0.592)***
-0.150 (0.238)
0.013 (0.065)
0.076 (0.034)
JPNrtE ,
3ln �� 2.091 (0.701)**
-0.136 (0.283)
-0.165 (0.077)**
-0.011 (0.040)
JPNrtE ,
2ln �� 3.364 (1.910)*
0.028 (0.770)
-0.398 (0.209)*
-0.035 (0.108)
2ln �� tY 2.737 (2.033)
-0.645 (0.820)
0.101 (0.222)
-0.041 (0.115)
3ln �� tY 3.868 (1.933)*
-0.296 (0.779)
0.273 (0.211)
-0.204 (0.110)*
JPNftY ,
1ln �� -4.381 (3.826)
9.500 (1.542)***
-0.316 (0.418)
0.220 (0.217)
JPNftY ,
2ln �� 2.388 (3.358)
0.546 (1.353)
0.391 (0.367)
0.289 (0.190)
JPNftY ,
3ln �� 0.802 (3.691)
1.302 (1.487)
-0.030 (0.404)
0.239 (0.209)
C -0.170 (0.128)
-0.031 (0.051)
0.027 (0.014)*
0.011 (0.007)
COUP -0.236 (0.188)
0.038 (0.076)
-0.041 (0.021)*
-0.001 (0.011)
Diagnostics 2R 0.737 0.349 0.473 0.578
Adjusted 2R 0.544 -0.131 0.084 0.268 ! 0.380 0.153 0.042 0.022
NX 2
0.036 [0.982]
HetX 2
30.667 [0.285]
LM )(SCTest 31.224 [0.500]
AR roots graph stable
242
Table SA10: Short–run coefficient estimates of Fiji’s goods trade balance model with the USA, 1975–2012
Short-run results USATBGln� USArE ,ln� Yln� USAfY ,ln�
1�tECT -0.458 (0.152)***
0.051 (0.029)*
-0.028 (0.128)**
-0.001 (0.006)
USAtTBG 1ln �� -0.295
(0.153)* 0.026 (0.029)
0.005 (0.013)
0.011 (0.006)*
USArt
E ,1
ln�
� -1.527 (0.943)
0.144 (0.176)
-0.077 (0.079)
-0.002 (0.036)
1ln �� tY 1.106 (1.930)
0.163 (0.361)
-0.590 (0.162)***
-0.202 (0.073)***
USAftY ,
1ln �� 1.473 (3.998)
1.430 (0.747)*
0.913 (0.335)**
0.514 (0.151)***
C 0.126 (0.137)
-0.048 (0.026)*
0.014 (0.011)
0.020 (0.005)***
COUP -0.415 (0.167)**
0.018 (0.031)
-0.025 (0.014)*
-0.012 (0.006)*
Diagnostics 2R 0.466 0.306 0.394 0.448
Adjusted 2R 0.356 0.162 0.269 0.333 ! 0.435 0.081 0.036 0.016
NX 2
2.131 [0.345]
HetX 2
11.395 [0.411]
LM )(SCTest 10.850 [0.819]
AR roots graph stable
243
Table SA11: Short–run coefficient estimates of Fiji’s goods trade balance model with the UK, 1975–2012
Short-run results UKTBGln� rEln� Yln� UKfY ,ln�
1�tECT -0.604 (0.183)***
0.014 (0.030)
0.021 (0.016)
0.010 (0.006)
UKtTBG 1ln �� -0.467
(0.272)* 0.056 (0.044)
-0.051 (0.023)**
0.027 (0.009)***
UKtTBG 2ln �� -0.224
(0.385) 0.003 (0.062)
-0.031 (0.033)
0.023 (0.013)
UKtTBG 3ln ��
-0.777 (0.442)*
0.024 (0.072)
-0.026 (0.038)
0.029 (0.015)*
UKtTBG 4ln ��
-0.494 (0.462)
0.126 (0.075)*
-0.008 (0.040)
0.027 (0.014)*
UKtTBG 5ln ��
0.106 (0.321)
-0.057 (0.052)
0.026 (0.027)
-0.003 (0.011)
rt
E1
ln�
� 6.252 (2.306)**
0.253 (0.375)
-0.351 (0.198)*
0.075 (0.080)
rt
E2
ln�
� 5.854 (2.351)**
-0.191 (0.382)
0.072 (0.201)
-0.129 (0.081)
rt
E3
ln�
� 3.789 (2.730)
-0.560 (0.443)
-0.102 (0.234)
-0.242 (0.094)**
rt
E4
ln�
� 9.161 (3.402)**
0.034 (0.553)
-0.106 (0.291)
-0.157 (0.117)
rt
E5
ln�
� 8.778 (3.575)**
-0.991 (0.581)*
0.037 (0.306)
-0.135 (0.123)
1ln �� tY 3.578 (3.503)
-0.551 (0.569)
-0.772 (0.280)**
0.157 (0.121)
2ln �� tY 3.291 (3.693)
0.261 (0.600)
-0.334 (0.316)
0.091 (0.127)
3ln �� tY 5.916 (3.786)
0.018 (0.615)
-0.065 (0.324)
-0.139 (0.130)
4ln �� tY 6.548 (4.317)
-1.328 (0.701)*
-0.134 (0.370)
-0.252 (0.149)
5ln �� tY 4.262 (2.743)
-0.379 (0.445)
-0.150 (0.235)
-0.031 (0.095)
UKftY ,
1ln �� 9.835 (7.587)
-0.149 (1.232)
-0.076 (0.650)
0.384 (0.261)
UKftY ,
2ln �� 11.430 (8.064)
0.714 (1.309)
0.646 (0.691)
-0.218 (0.278)
UKftY ,
3ln �� -2.629 (9.369)
-1.437 (1.521)
0.835 (0.802)
-0.129 (0.323)
UKftY ,
4ln �� -15.784 (10.342)
1.598 (1.697)
-1.033 (0.886)
0.231 (0.356)
UKftY ,
5ln �� -3.418 (9.809)
1.214 (1.593)
0.297 (0.840)
0.416 (0.338)
C -0.659 0.024 0.044 0.230
244
(0.287)** (0.047) (0.025)** (0.010)**COUP 0.073
(0.277)-0.040(0.045)
-0.015(0.024)
-0.023(0.010)**
Diagnostics2R 0.790 0.675 0.797 0.879
Adjusted 2R 0.278 -0.119 0.300 0.584! 0.398 0.065 0.034 0.014
NX 2 3.441[0.179]
HetX 2 25.524[0.545]
LM )(SCTest 12.149[0.734]
AR roots graph stable
245
Table SA12: Short–run coefficient estimates of Fiji’s goods trade balance model with Singapore, 1975–2012
Short-run results SINGTBGln� rEln� Yln� SINGfY ,ln�
1�tECT -0.617 (0.326)*
-0.037 (0.013)***
0.024 (0.010)**
0.009 (0.011)
SINGtTBG 1ln �� 0.000
(0.294) -0.020 (0.012)
-0.020 (0.009)**
-0.003 (0.009)
SINGtTBG 2ln �� 0.052
(0.228) 0.021
(0.009)** -0.012 (0.007)*
-0.006 (0.007)
rtE 1ln �� 2.897
(4.475) 0.116 (0.191)
0.080 (0.139)
0.313 (0.153)
rtE 2ln �� 5.990
(4.471) -0.078 (0.180)
0.320 (0.131)**
-0.049 (0.144)
1ln �� tY 1.366 (5.797)
-0.380 (0.233)
-0.265 (0.170)
0.240 (0.187)
2ln �� tY 3.423 (6.250)
0.041 (0.252)
0.190 (0.183)
0;182 (0.201)
SINGftY ,
1ln �� 1.885 (6.379)
-0.302 (0.257)
-0.176 (0.187)
0.180 (0.205)
SINGftY ,
2ln �� 1.259 (6.475)
-0.027 (0.261)
1.44 (0.190)
-0.184 (0.208)
C -0.316 (0.667)
0.035 (0.027)
0.022 (0.020)
0.056 (0.021)**
COUP -1.007 (0.720)
0.038 (0.029)
-0.041 (0.021)
-0.007 (0.023)
Diagnostics 2R 0.332 0.511 0.484 0.260
Adjusted 2R 0.053 0.307 0.268 -0.048 ! 1.248 0.050 0.037 0.040
NX 2
1.217 [0.544]
HetX 2
17.149 [0.580]
LM )(SCTest 20.046 [0.218]
AR roots graph stable
246
Table SA13: Short–run coefficient estimates of Fiji’s goods trade balance model with China, 1975–2012
Short-run results CHNTBGln� rEln� Yln� CHNfY ,ln�
1�tECT -0.534 (0.228)**
-0.001 (0.006)
0.007 (0.003)**
0.003 (0.002)*
CHNtTBG 1ln �� -0.548
(0.212)** -0.009 (0.005)*
3.680 (0.003)
-0.004 (0.002)**
CHNtTBG 2ln �� -0.337
(0.228) -0.000 (0.005)
-0.002 (0.003)
-0.001 (0.002)
CHNtTBG 3ln ��
-0.105 (0.183)
0.000 (0.005)
-0.006 (0.003)**
-0.003 (0.001)*
rtE 1ln �� 4.245
(8.110) 0.264 (0.208)
0.092 (0.116)
-0.040 (0.065)
rtE 2ln �� -2.830
(8.571) -0.165 (0.220)
-0.013 (0.123)
-0.277 (0.069)
rtE 3ln �� -5.660
(8.850) 0.031 (0.227)
-0.026 (0.127)
-0.110 (0.071)
1ln �� tY 12.643 (10.809)
-0.759 (0.278)**
-0.283 (0.155)**
0.022 (0.087)
2ln �� tY 2.210 (11.753)
0.226 (0.302)
0.249 (0.168)
-0.148 (0.094)
3ln �� tY -3.213 (11.924)
-0.015 (0.306)
-0.142 (0.170)
-0.159 (0.096)*
CHNftY ,
1ln �� -19.639 (25.837)
-0.204 (0.664)
-0.574 (0.369)
0.598 (0.207)***
CHNftY ,
2ln �� -7.320 (26.843)
0.567 (0.690)
1.326 (0.384)***
-0.549 (0.215)**
CHNftY ,
3ln �� 2.800 (19.136)
0.461 (0.492)
-1.284 (0.274)***
0.108 (0.153)
C 2.571 (2.502)
-0.063 (0.064)
0.070 (0.036)*
0.090 (0.020)***
COUP -1.142 (1.412)
0.076 (0.036)**
-0.021 (0.020)
0.013 (0.011)
Diagnostics 2R 0.703 0.610 0.760 0.773
Adjusted 2R 0.484 0.323 0.583 0.605 ! 1.959 0.050 0.028 0.016
NX 2
1.908 [0.385]
HetX 2
29.079 [0.357]
LM )(SCTest 9.951 [0.869]
AR roots graph stable
247
Table SA14: Short–run coefficient estimates of Fiji’s goods trade balance model with Malaysia, 1975–2012
Short-run results MALATBGln� rEln� Yln� MALAfY ,ln�
1�tECT -0.271 (0.139)*
0.000 (0.010)
-0.001 (0.007)
0.012 (0.006)
MALAtTBG 1ln �� -0.061
(0.182) -0.006 (0.014)
-0.006 (0.009)
0.003 (0.008)
MALAtTBG 2ln �� 0.124
(0.193) 0.010 (0.014)
-0.011 (0.009)
0.012 (0.009)
rtE 1ln �� 0.312
(3.394) 0.105 (0.296)
-0.007 (0.191)
0.068 (0.173)
rtE 2ln �� 1.494
(3.059) 0.002 (0.230)
0.296 (0.149)*
0.055 (0.135)
1ln �� tY 6.912 (4.816)
-0.364 (0.362)
-0.310 (0.234)
0.000 (0.212)
2ln �� tY 2.104 (4.650)
0.028 (0.349)
0.075 (0.226)
0.062 (0.205)
MALAftY ,
1ln �� 9.078 (5.199)*
-0.307 (0.391)
-0.231 (0.253)
0.053 (0.229)
MALAftY ,
2ln �� -6.374 (5.422)
0.202 (0.408)
0.416 (0.264)
-0.151 (0.239)
C -0.398 (0.512)
0.010 (0.039)
0.011 (0.025)
0.064 (0.023)***
COUP -0.474 (0.538)
0.069 (0.040)*
-0.035 (0.026)
0.007 (0.024)
Diagnostics 2R 0.397 0.318 0.432 0.370
Adjusted 2R 0.145 0.034 0.196 0.107 ! 0.789 0.059 0.038 0.035
NX 2
0.367 [0.833]
HetX 2
22.417 [0.264]
LM )(SCTest 10.939 [0.813]
AR roots graph stable
248
Table SA15: Short–run coefficient estimates of Fiji’s goods trade balance model with India, 1975–2012
Short-run results INDTBGln� rEln� Yln� INDfY ,ln�
1�tECT -1.131 (0.223)***
0.020 (0.019)
-0.002 (0.015)
0.001 (0.009)
INDtTBG 1ln �� -0.079
(0.209) -0.006 (0.018)
-0.001 (0.014)
0.004 (0.008)
INDtTBG 2ln �� -0.029
(0.244) -0.010 (0.021)
0.009 (0.016)
0.008 (0.009)
INDtTBG 3ln ��
0.053 (0.251)
-0.019 (0.021)
0.014 (0.017)
0.004 (0.010)
INDtTBG 4ln ��
0.257 (0.215)
-0.007 (0.018)
-0.012 (0.014)
0.002 (0.008)
INDtTBG 5ln ��
0.462 (0.218)**
-0.029 (0.019)
0.002 (0.015)
-0.005 (0.008)
rtE 1ln �� 11.532
(3.644)*** 0.125 (0.310)
-0.078 (0.244)
0.170 (0.141)
rtE 2ln �� 11.366
(4.606)** -0.366 (0.392)
0.242 (0.309)
-0.156 (0.069)
rtE 3ln �� 3.282
(3.799) -0.049 (0.324)
0.024 (0.254)
-0.082 (0.147)
rtE 4ln �� -0.077
(3.420) -0.167 (0.291)
0.054 (0.229)
0.032 (0.132)
rtE 5ln �� 4.808
(2.838)* -0.569
(0.242)** 0.036 (0.190)
0.038 (0.110)
1ln �� tY 15.478 (7.053)**
-0.860 (0.601)
-0.040 (0.472)
-0.055 (0.272)
2ln �� tY 5.334 (4.866)
0.094 (0.415)
0.132 (0.326)
0.156 (0.188)
3ln �� tY -0.084 (4.545)
-0.152 (0.387)
-0.042 (0.304)
-0.098 (0.175)
4ln �� tY 2.279 (4.484)
-0.381 (0.382)
-0.132 (0.300)
0.121 (0.173)
5ln �� tY 3.875 (4.564)
-0.410 (0.389)
0.178 (0.306)
-0.007 (0.176)
INDftY ,
1ln �� -42.592 (12.240)***
-0.535 (1.043)
-0.139 (0.820)
0.306 (0.472)
INDftY ,
2ln �� -28.949 (7.610)***
0.509 (0.648)
-0.393 (0.510)
0.119 (0.294)
INDftY ,
3ln �� -34.641 (8.050)***
0.479 (0.686)
0.039 (0.539)
0.032 (0.311)
INDftY ,
4ln �� -24.198 (9.049)**
0.887 (0.771)
0.320 (0.606)
0.165 (0.349)
INDftY ,
5ln �� -26.718 (8.357)***
0.686 (0.712)
-0.388 (0.560)
0.039 (0.323)
249
C 8.188(1.692)***
-0.061(0.144)
0.470(0.113)
0.019(0.020)
COUP 0.288(0.603)
0.081(0.051)
-0.038(0.040)
-0.011(0.023)
Diagnostics2R 0.884 0.757 0.662 0.605
Adjusted 2R 0.601 0.163 -0.165 -0.361! 3.875 0.028 0.017 0.006
NX 2 0.009[0.995]
HetX 2 27.203[0.453]
LM )(SCTest 17.850[0.333]
AR roots graph stable
250
Table SA16: Short–run coefficient estimates of Fiji’s goods trade balance model with Hong Kong, 1975–2012
Short-run results HKTBGln� rEln� Yln� HKfY ,ln�
1�tECT -0.773 (0.279)***
-0.129 (0.052)**
0.024 (0.037)
-0.003 (0.029)
HKtTBG 1ln �� -0.738
(0.316)** 0.067 (0.059)
-0.053 (0.041)
0.003 (0.033)
HKtTBG 2ln �� -0.716
(0.325)** 0.067 (0.061)
-0.054 (0.043)
0.007 (0.034)
HKtTBG 3ln ��
-0.597 (0.303)*
0.052 (0.057)
-0.038 (0.040)
0.024 (0.032)
HKtTBG 4ln ��
-0.396 (0.261)
0.034 (0.049)
-0.020 (0.034)
-0.010 (0.027)
rtE 1ln �� 7.414
(2.059)*** 0.534 (0.386)
-0.104 (0.270)
0.131 (0.217)
rtE 2ln �� 4.172
(2.345)* 0.329 (0.440)
0.306 (0.307)
0.196 (0.247)
rtE 3ln �� 2.410
(1.984) 0.175 (0.372)
0.078 (0.260)
-0.294 (0.209)
rtE 4ln �� 3.135
(1.758)* 0.257 (0.330)
-0.110 (0.230)
-0.107 (0.185)
1ln �� tY 7.030 (3.831)*
0.862 (0.718)
-1.114 (0.502)**
0.219 (0.403)
2ln �� tY 11.125 (4.325)**
1.255 (0.811)
-0.494 (0.566)
0.638 (0.455)
3ln �� tY 8.525 (5.147)*
1.108 (0.965)
-0.031 (0.674)
0.246 (0.542)
4ln �� tY 4.453 (2.864)
0.530 (0.537)
-0.102 (0.375)
-0.199 (0.301)
HKftY ,
1ln �� 2.420 (4.450)
0.102 (0.834)
-0.174 (0.582)
-0.588 (0.468)
HKftY ,
2ln �� -3.340 (2.849)
-0.134 (0.534)
0.254 (0.373)
-0.681 (0.300)
HKftY ,
3ln �� 0.877 (3.601)
-0.023 (0.675)
-0.023 (0.675)
-0.157 (0.379)
HKftY ,
4ln �� -0.867 (2.456)
0.083 (0.461)
-0.024 (0.321)
-0.219 (0.258)
C 0.076 (0.103)
0.002 (0.019)
0.006 (0.014)
-0.006 (0.011)
COUP -0.087 (0.323)
0.021 (0.061)
-0.046 (0.042)
-0.014 (0.034)
Diagnostics 2R 0.881 0.602 0.784 0.727
Adjusted 2R 0.717 0.051 0.486 0.348 ! 0.375 0.070 0.049 0.039
251
NX 2 0.317[0.853]
HetX 2 13.690[0.251]
LM )(SCTest 14.387[0.570]
AR roots graph stable
252
Table SA17: Short–run coefficient estimates of Fiji’s goods export model, 1975–2012
Short-run results EXPGln� rEln� fYln�
1�tECT -0.620 (0.212)***
0.158 (0.103)
-0.650 (0.144)***
1ln �� tEXPG 0.368 (0.291)
-0.181 (0.141)
0.229 (0.197)
2ln �� tEXPG 0.270 (0.219)
0.115 (0.106)
-0.004 (0.149)
3ln �� tEXPG -0.224 (0.171)
-0.028 (0.083)
-0.130 (0.116)
rtE 1ln �� -0.110
(0.584) 0.630
(0.284)** -1.816
(0.397)*** rtE 2ln �� -0.823
(0.724) 0.305 (0.352)
-0.325 (0.491)
rtE 3ln �� -0.545
(0.577) -0.222 (0.280)
0.550 (0.392)
ftY 1ln �� -0.072
(0.229) 0.069 (0.111)
-0.079 (0.156)***
ftY 2ln �� -0.366
(0.221) -0.058 (0.107)
-0.337 (0.150)**
ftY 3ln �� -0.132
(0.194) -0.097 (0.094)
-0.124 (0.131)
0.058 (0.028)**
-0.002 (0.014)
0.066 (0.019)***
COUP -0.056 (0.058)
0.041 (0.028)
0.017 (0.040)
Diagnostics 0.577 0.436 0.815
0.366 0.153 0.723
0.116 0.056 0.079
0.090 [0.956]
27.946 [0.142]
9.102 [0.428]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
253
Table SA18: Short–run coefficient estimates of Fiji’s domestic export model, 1975–2012
Short-run results DOMEXPGln� rEln� fYln�
1�tECT -0.144 (0.067)**
0.518 (0.026)*
-0.245 (0.043)***
1ln �� tDOMEXPG -0.102 (0.179)
-0.029 (0.070)
0.168 (0.116)
rtE 1ln �� 0.419
(0.534) 0.464
(0.208)** -1.412
(0.346)*** 1ln �� t
fY 0.069 (0.189)
-0.041 (0.074)
-0.405 (0.122)***
C 0.008 (0.030)
-0.004 (0.012)
0.035 (0.019)*
COUP 0.022 (0.054)
0.029 (0.021)
0.061 (0.035)*
Diagnostics 2R 0.268 0.240 0.653
Adjusted 2R 0.145 0.113 0.595
! 0.144 0.056 0.093
NX 2
0.423 [0.809]
HetX 2
13.792 [0.130]
LM )(SCTest 10.350 [0.323]
AR roots graph stable
254
Table SA19: Short–run coefficient estimates of Fiji’s services export model, 1975–2012
Short-run results EXPSln� rEln� fYln�
1�tECT -0.939 (0.200)***
0.274 (0.150)**
0.180 (0.309)
1ln �� tEXPS 0.485 (0.204)**
-0.174 (0.153)
-0.211 (0.315)
2ln �� tEXPS -0.065 (0.173)
-0.123 (0.130)
0.002 (0.268)
2ln �� tEXPS 0.325 (0.151)**
-0.122 (0.113)
-0.109 (0.234)
rtE 1ln �� 0.438
(0.278) 0.320 (0.208)
-0.735 (0.430)*
rtE 2ln �� -0.129
(0.332) -0.097 (0.248)
0.941 (0.513)*
rtE 3ln �� 0.683
(0.328)** -0.116 (0.246)
0.685 (0.508)
ftY 1ln �� -0.743
(0.191)*** 0.177 (0.143)
-0.556 (0.295)*
ftY 2ln �� -0.500
(0.173)*** 0.104 (0.129)
-0.142 (0.267)
ftY 3ln �� -0.385
(0.128)*** -0.042 (0.096)
0.025 (0.199)
C 0.065 (0.022)***
0.009 (0.016)
0.031 (0.034)
COUP -0.082 (0.040)*
0.019 (0.030)
0.041 (0.063)
Diagnostics 2R 0.682 0.338 0.527
Adjusted 2R 0.522 0.006 0.290
! 0.081 0.061 0.126
NX 2
1.158 [0.561]
HetX 2
17.150 [0.702]
LM )(SCTest 16.079 [0.065]
AR roots graph stable
255
Table SA20: Short–run coefficient estimates of Fiji’s goods and services export model, 1975–2012
Short-run results EXPGSln� rEln� fYln�
1�tECT -0.818 (0.200)***
0.329 (0.167)*
-0.655 (0.300)**
1ln �� tEXPGS 0.264 (0.224)
-0.224 (0.187)
-0.075 (0.336)
2ln �� tEXPGS 0.043 (0.177)
0.071 (0.148)
0.051 (0.265)
3ln �� tEXPGS -0.072 (0.135)
-0.110 (0.113)
-0.136 (0.203)
rtE 1ln �� 0.006
(0.295) 0.630
(0.246)** -1.348
(0.442)*** rtE 2ln �� -0.224
(0.384) 0.255 (0.321)
0.614 (0.575)
rtE 3ln �� 0.098
(0.324) -0.230 (0.270)
0.701 (0.485)
ftY 1ln �� -0.329
(0.147)** 0.141 (0.123)
-0.783 (0.221)***
ftY 2ln �� -0.281
(0.143)* 0.047 (0.119)
-0.445 (0.214)**
ftY 3ln �� -0.228
(0.119)* -0.002 (0.014)
0.061 (0.026)
C 0.061 (0.017)***
-0.002 (0.014)
0.061 (0.026)**
COUP -0.055 (0.033)
0.023 (0.028)
0.017 (0.050)
Diagnostics 2R 0.681 0.391 0.672
Adjusted 2R 0.521 0.087 0.508
! 0.070 0.058 0.105
NX 2
2.007 [0.367]
HetX 2
18.139 [0.640]
LM )(SCTest 9.583 [0.385]
AR roots graph stable
256
Table SA21: Short–run coefficient estimates of Fiji’s export of food model, 1975–2012
Short-run results FOODEXPln� rEln� fYln�
1�tECT -0.199 (0.053)***
-0.084 (0.036)
-0.096 (0.073)
1ln �� tFOODEXP -0.654 (0.282)**
-0.150 (0.190)
-0.417 (0.384)
2ln �� tFOODEXP -0.485 (0.276)*
0.280 (0.186)
-0.440 (0.377)
3ln �� tFOODEXP -0.128 (0.334)
-0.086 (0.225)
0.518 (0.455)
4ln �� tFOODEXP 0.364 (0.409)
-0.033 (0.275)
0.373 (0.558)
5ln �� tFOODEXP 0.504 (0.339)
0.129 (0.228)
0.247 (0.462)
6ln �� tFOODEXP -0.209 (0.234)
-0.161 (0.158)
-0.146 (0.320)
rtE 1ln �� -0.374
(0.485) -0.245 (0.326)
-1.091 (0.661)
rtE 2ln �� -0.220
(0.449) -0.487 (0.302)
0.666 (0.612)
rtE 3ln �� -0.196
(0.592) -0.627 (0.398)
0.588 (0.806)
rtE 4ln �� -1.004
(0.780) 0.129 (0.525)
-1.080 (1.063)
rtE 5ln �� -2.127
(0.711)*** -0.253 (0.479)
-0.954 (0.969)
rtE 6ln �� -0.658
(0.738) -0.134 (0.497)
0.617 (1.006)
ftY 1ln �� 0.997
(0.327)*** 0.142 (0.220)
-0.347 (0.446)
ftY 2ln �� 1.370
(0.479)*** -0.238 (0.322)
0.675 (0.653)
ftY 3ln �� 1.154
(0.519)** -0.290 (0.349)
-0.491 (0.707)
ftY 4ln �� 0.521
(0.406) -0.093 (0.273)
-0.156 (0.553)
ftY 5ln �� 0.025
(0.261) -0.189 (0.176)
-0.139 (0.356)
ftY 6ln �� -0.043
(0.219) -0.065 (0.148)
-0.189 (0.299)
-0.106 (0.060)*
0.086 (0.041)**
-0.016 (0.082)
COUP 0.172 (0.076)**
-0.050 (0.051)
0.164 (0.103)
Diagnostics
C
257
0.884 0.809 0.863
0.497 0.171 0.407
0.089 0.060 0.121
0.803 [0.669]
26.247 [0.505]
12.504 [0.186]
AR roots graph stable
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
258
Table SA22: Short–run coefficient estimates of Fiji’s export of sugar model, 1975–2012
Short-run results SUGEXPln� rEln� fYln�
1�tECT -0.267 (0.141)*
0.044 (0.034)
-0.215 (0.060)***
SUGtEXP 1ln �� 0.108
(0.261) -0.012 (0.064)
0.202 (0.112)*
SUGtEXP 2ln �� -0.229
(0.241) 0.043 (0.059)
0.186 (0.103)*
SUGtEXP 3ln �� 0.320
(0.275) -0.015 (0.067)
0.024 (0.118)
rtE 1ln �� -1.871
(0.973)* 0.401 (0.237)
-1.516 (0.417)***
rtE 2ln �� -0.062
(1.174) 0.127 (0.286)
-0.452 (0.503)
rtE 3ln �� -1.977
(1.115)* 0.124 (0.271)
-0.198 (0.478)
ftY 1ln �� 1.080
(0.464)** -0.130 (0.113)
-0.289 (0.199)
ftY 2ln �� 0.163
(0.447) -0.119 (0.109)
-0.315 (0.192)
ftY 3ln �� -0.196
(0.370) -0.164 (0.090)*
-0.181 (0.159)
-0.017 (0.075)
0.008 (0.018)
0.090 (0.032)***
COUP 0.044 (0.129)
0.040 (0.031)
-0.000 (0.055)
Diagnostics 0.437 0.403 0.690
0.155 0.105 0.536
0.238 0.058 0.102
3.046 [0.218]
25.455 [0.228]
14.065 [0.120]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
259
Table SA23: Short–run coefficient estimates of Fiji’s export of fish model, 1975–2012
Short-run results FISHEXPln� rEln� fYln�
1�tECT -0.744 (0.117)***
-0.018 (0.026)
-0.031 (0.057)
FISHtEXP 1ln �� 0.035
(0.113) -0.008 (0.025)
0.040 (0.055)
rtE 1ln �� 0.890
(0.858) 0.292 (0.190)
-0.531 (0.420)
ftY 1ln �� -0.298
(0.318) -0.025 (0.070)
-0.505 (0.156)***
-0.027 (0.056)
-0.001 (0.012)
0.012 (0.027)
COUP 0.248 (0.103)**
0.027 (0.023)
0.100 (0.050)
Diagnostics 0.642 0.152 0.310
0.582 0.011 0.196 0.267 0.059 0.131
1.922 [0.383]
15.454 [0.079]
9.072 [0.431]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
260
Table SA24: Short–run coefficient estimates of Fiji’s export of gold model, 1975–2012
Short-run results GOLDEXPln� rEln� fYln�
1�tECT 0.450 (0.150)***
0.017 (0.017)
0.045 (0.041)
GOLDtEXP 1ln �� -1.838
(0.393)*** -0.073
(0.046)** -0.133 (0.107)
GOLDtEXP 2ln �� -1.402
(0.360)*** -0.070 (0.042)
-0.117 (0.098)
GOLDtEXP 3ln �� -1.475
(0.414)*** -0.049 (0.048)
-0.152 (0.113)
GOLDtEXP 4ln �� -1.494
(0.557)** -0.051 (0.065)
-0.181 (0.151)
GOLDtEXP 5ln �� -1.623
(0.624)** -0.055 (0.072)
-0.175 (0.170)
GOLDtEXP 6ln �� -2.160
(0.831)** 0.124 (0.096)
-0.034 (0.226)
GOLDtEXP 7ln �� -2.061
(0.983)** -0.141 (0.114)
0.092 (0.267)
1ln �� trE 10.543
(0.3731)*** 0.793
(0.433)* -0.530 (1.015)
2ln �� trE 1.679
(2.662) 0.015 (0.433)
0.796 (0.724)
3ln �� trE 3.501
(3.004) 0.197 (0.348)
1.763 (0.817)**
4ln �� trE -0.255
(3.545) -0.491 (0.411)
0.283 (1.964)
5ln �� trE -2.554
(3.246) -0.148 (0.376)
0.617 (0.883)
6ln �� trE 5.740
(2.779)** -0.470 (0.376)
0.735 (1.756)
7ln �� trE 6.181
(3.966) 0.171 (0.460)
0.562 (1.078)
1ln �� tY 1.855 (1.371)
0.165 (0.159)
-0.976 (0.373)**
2ln �� tY 1.987 (1.631)
0.119 (0.189)
-0.776 (0.443)*
3ln �� tY 0.600 (1.829)
0.225 (0.212)
-0.654 (0.497)
4ln �� tY -4.075 (1.866)**
0.201 (0.216)
-0.276 (0.507)
5ln �� tY -6.526 (2.235)***
-0.209 (0.259)
-0.164 (0.608)
6ln �� tY -6.069 (1.553)***
-0.308 (0.180)**
-0.208 (0.422)
7ln �� tY -4.665 (1.246)***
-0.276 (0.144)*
-0.192 (0.339)
261
0.795 (0.317)**
0.032 (0.037)
0.050 (0.086)
COUP -2.058 (0.574)***
-0.108 (0.067)
-0.013 (0.156)
Diagnostics 0.928 0.861 0.872
0.584 0.192 0.230 0.485 0.056 0.132
0.0857 [0.958]
27.298 [0.502]
6.225 [0.717]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
262
Table SA25: Short–run coefficient estimates of Fiji’s export of travel services model, 1975–2012
Short-run results TRAVEXPln� rEln� fYln�
1�tECT -0.440 (0.226)*
-0.041 (0.209)
0.092 (0.284)
1ln �� tTRAVEXP -0.062 (0.219)
-0.132 (0.202)
-0.088 (0.274)
2ln �� tTRAVEXP -0.229 (0.180)
-0.052 (0.166)
-0.109 (0.226)
3ln �� tTRAVEXP -0.215 (0.189)
-0.072 (0.175)
-0.188 (0.237)
4ln �� tTRAVEXP -0.167 (0.185)
0.021 (0.171)
0.104 (0.232)
5ln �� tTRAVEXP -0.033 (0.175)
-0.030 (0.162)
0.199 (0.220)
6ln �� tTRAVEXP -0.217 (0.159)
0.065 (0.147)
0.533 (0.199)**
rtE 1ln �� 0.895
(0.499)* 0.424 (0.462)
-0.700 (0.626)
rtE 2ln �� 0.925
(0.497)* -0.174 (0.460)
0.360 (0.624)
rtE 3ln �� 1.070
(0.587)* 0.017 (0.543)
0.814 (0.737)
rtE 4ln �� 0.953
(0.536)* 0.258 (0.496)
0.335 (0.673)
rtE 5ln �� -0.253
(0.507) -0.030 (0.469)
0.352 (0.636)
rtE 6ln �� 1.057
(0.449)** -0.327 (0.416)
0.998 (0.564)*
ftY 1ln �� -1.001
(0.507)* -0.046 (0.469)
-0.808 (0.637)
ftY 2ln �� -0.571
(0.418) -0.078 (0.386)
-0.546 (0.524)
ftY 3ln �� -0.299
(0.317) -0.038 (0.294)
-0.348 (0.398)
ftY 4ln �� -0.270
(0.252) 0.114 (0.233)
-0.296 (0.316)
ftY 5ln �� 0.324
(0.192)* -0.058 (0.177)
-0.072 (0.240)
ftY 6ln �� 0.061
(0.185) 0.106 (0.171)
-0.177 (0.233)
0.060 (0.035)*
0.010 (0.032)
0.035 (0.044)
COUP -0.181 (0.071)**
0.060 (0.066)
0.104 (0.089)
C
263
Diagnostics 0.865 0.479 0.850
0.594 -0.563 0.551
0.083 0.077 0.105
0.354 [0.838]
29.081 [0.357]
12.083 [0.209]
AR roots graph stable
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
264
Table SA26: Short–run coefficient estimates of Fiji’s export of transportation services model, 1975–2012
Short-run results TRANEXPln� rEln� fYln�
1�tECT -1.044 (0.234)***
0.209 (0.124)*
-0.115 (0.247)
1ln �� tTRANEXP 0.762 (0.191)***
-0.130 (0.101)
0.061 (0.202)
2ln �� tTRANEXP 0.274 (0.191)
-0.115 (0.101)
-0.077 (0.202)
3ln �� tTRANEXP 0.752 (0.199)***
-0.128 (0.105)
0.189 (0.209)
4ln �� tTRANEXP 0.232 (0.176)
-0.075 (0.093)
0.152 (0.186)
rtE 1ln �� 0.620
(0.587) 0.540
(0.310)* -0.567 (0.619)
rtE 2ln �� -0.119
(0.630) -0.103 (0.333)
0.538 (0.665)
rtE 3ln �� 1.220
(0.680)* -0.033 (0.360)
1.392 (0.717)*
rtE 4ln �� -1.938
(0.748)** 0.252 (0.396)
-0.187 (0.789)
ftY 1ln �� -1.726
(0.440)*** 0.319 (0.233)
-0.931 (0.465)*
ftY 2ln �� -0.833
(0.446)* 0.222 (0.236)
-0.371 (0.471)
ftY 3ln �� -0.736
(0.365)* 0.121 (0.193)
-0.254 (0.385)
ftY 4ln �� -0.325
(0.256) 0.154 (0.135)
-0.026 (0.270)
0.048 (0.044)
0.009 (0.023)
0.015 (0.047)
COUP -0.201 (0.103)*
0.044 (0.055)
0.074 (0.109)
Diagnostics 0.842 0.440 0.624
0.673 -0.163 0.218
0.132 0.070 0.139
0.472 [0.790]
17.848 [0.659]
4.791 [0.852]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
265
Table SA27: Short–run coefficient estimates of Fiji’s goods export model with Australia, 1975–2012
Short-run results AUSEXPGln� AUSrE ,ln� AUSfY ,ln�
1�tECT -0.308 (0.141)**
-0.124 (0.057)**
-0.021 (0.009)**
AUStEXPG 1ln �� 0.148
(0.226) 0.235
(0.092)** -0.021 (0.015)
AUStEXPG 2ln �� 0.455
(0.263)* 0.198* (0.107)
0.004 (0.018)
AUStEXPG 3ln �� 0.429
(0.221)* 0.096 (0.090)
0.004 (0.015)
AUStEXPG 4ln �� 0.342
(0.185)* -0.048 (0.075)
-0.002 (0.012)
AUSrt
E ,1
ln�
� -0.842 (0.794)
-0.899 (0.323)***
0.116 (0.005)**
AUSrt
E ,2
ln�
� -1.908 (0.801)**
-1.043 (0.326)***
0.060 (0.054)
AUSrt
E ,3
ln�
� -1.968 (0.679)***
-0.696** (0.277)
0.014 (0.046)
AUSrt
E ,4
ln�
� -1.618 (0.612)**
-0.304 (0.249)
-0.008 (0.041)
AUSftY ,
1ln �� -0.582 (3.396)
1.249 (1.383)
-0.289 (0.229)
AUSftY ,
2ln �� 0.597 (3.373)
3.204 (1.373)**
-0.542 (0.227)**
AUSftY ,
3ln �� -0.306 (3.584)
2.014 (1.459)
-0.573 (0.241)**
AUSftY ,
4ln �� 1.629 (3.688)
0.935 (1.501)
-0.220 (0.248)
C -0.016 (0.295)
-0.228 (0.120)**
0.077 (0.020)***
COUP 0.165 (0.116)
0.103 (0.047)**
0.011 (0.008)
Diagnostics
2R 0.536 0.602 0.508
Adjusted 2R 0.175 0.292 0.125
! 0.218 0.089 0.015 NX 2
0.388 [0.824]
HetX 2
29.999 [0.314]
LM )(SCTest 6.752 [0.663]
AR roots graph stable
266
Table SA28: Short–run coefficient estimates of Fiji’s goods export model with New Zealand, 1975–2012
Short-run results NZEXPGln� NZrE ,ln� NZfY ,ln�
1�tECT -0.790 (0.208)***
-0.226 (0.149)*
-0.006 (0.019)
NZtEXPG 1ln �� 0.552
(0.159)*** 0.188
(0.114)* -0.018 (0.014)
NZtEXPG 2ln �� 0.186
(0.178) 0.079 (0.127)
-0.013 (0.016)
NZrt
E ,1
ln�
� -0.309 (0.351)
-0.378 (0.252)
0.014 (0.031)
NZrtE ,
2ln �� 0.348 (0.303)
-0.191 (0.217)
-0.010 (0.027)
NZftY ,
1ln �� 0.677 (2.07)
-0.995 (1.484)
0.058 (0.186)
NZftY ,
2ln �� -3.407 (1.563)***
-0.278 (1.120)
0.231 (0.140)
C 0.044 (0.065)
0.057 (0.046)
0.018 (0.006)***
COUP 0.002 (0.091)
-0.008 (0.065)
-0.004 (0.008)
Diagnostics 2R 0.625 0.135 0.353
Adjusted 2R 0.510 -0.131 0.154
! 0.197 0.141 0.018
NX 2
2.691 [0.260]
HetX 2
12.510 [0.640]
LM )(SCTest 10.478 [0.313]
AR roots graph stable
267
Table SA29: Short–run coefficient estimates of Fiji’s goods export model with Japan, 1975–2012
Short-run results JPNEXPGln� JPNrE ,ln� JPNfY ,ln�
1�tECT -0.913 (0.136)***
-0.008 (0.079)
0.003 (0.012)
JPNtEXPG 1ln �� 0.668
(0.131)*** -0.075 (0.076)
0.001 (0.012)
JPNtEXPG 2ln �� -0.123
(0.090) -0.044 (0.052)
0.001 (0.008)
JPNtEXPG 3ln �� 0.437
(0.095)*** -0.026 (0.055)
-0.007 (0.008)
JPNrt
E ,1
ln�
� -0.777 (0.390)*
-0.012 (0.227)
-0.025 (0.035)
JPNrt
E ,2
ln�
� 1.019 (0.379)**
-0.187 (0.227)
0.071 (0.034)
JPNrt
E ,3
ln�
� 0.382 (0.453)
-0.055 (0.264)
0.027 (0.040)
JPNftY ,
1ln �� -3.969 (2.536)
0.221 (1.477)
0.127 (0.225)
JPNftY ,
2ln �� 2.790 (2.259)
0.857 (1.315)
0.281 (0.201)
JPNftY ,
3ln �� -0.051 (2.576)
1.459 (1.500)
0.258 (0.229)
C 0.024 (0.070)
-0.044 (0.040)
0.005 (0.006)
COUP -0.201 (0.172)
0.111 (0.100)
-0.001 (0.015)
Diagnostics 2R 0.867 0.308 0.472
Adjusted 2R 0.801 -0.038 0.209
! 0.252 0.147 0.022 NX 2
0.252 [0.882]
HetX 2
23.275 [0.330]
LM )(SCTest 8.400 [0.494]
AR roots graph stable
268
Table SA30: Short–run coefficient estimates of Fiji’s goods export model with the USA, 1975–2012
Short-run results USAEXPGln� USArE ,ln� USAfY ,ln�
1�tECT -0.369 (0.101)***
0.045 (0.025)*
-0.001 (0.006)
USAtEXPG 1ln �� 0.039
(0.154) 0.035 (0.039)
-0.015 (0.010)
USArt
E ,1
ln�
� -1.781 (0.651)**
0.096 (0.163)
0.012 (0.040)
USAftY ,
1ln �� 2.393 (2.755)
1.347 (0.692)*
0.424 (0.171)**
C -0.014 (0.091)
-0.047 (0.023)**
0.016 (0.006)***
COUP 0.087 (0.196)
0.112 (0.049)**
-0.000 (0.012)
Diagnostics 2R 0.363 0.324 0.197
Adjusted 2R 0.256 0.211 0.063
! 0.314 0.079 0.020 NX 2
0.232 [0.891]
HetX 2
12.045 [0.211]
LM )(SCTest 5.181 [0.818]
AR roots graph stable
269
Table SA31: Short–run coefficient estimates of Fiji’s goods export model with the UK, 1975–2012
Short-run results UKEXPGln� rEln� UKfY ,ln�
1�tECT -0.771 (0.164)***
-0.027 (0.058)
-0.004 (0.012)
UKtEXPG 1ln �� -0.476
(0.181)** 0.094 (0.063)
0.029 (0.013)**
UKtEXPG 2ln �� -0.179
(0.188) 0.052 (0.066)
0.007 (0.014)
UKtEXPG 3ln �� -0.302
(0.175)* 0.039 (0.061)
-0.005 (0.013)
UKtEXPG 4ln �� -0.033
(0.167) 0.043 (0.059)
0.001 (0.012)
UKtEXPG 5ln �� 0.476
(0.154)*** 0.015 (0.054)
-0.013 (0.011)
rt
E1
ln�
� -0.579 (0.683)
0.365 (0.239)
0.111 (0.050)**
rt
E2
ln�
� 1.249 (0.826)
-0.145 (0.289)
-0.029 (0.061)
rt
E3
ln�
� 0.870 (0.802)
-0.344 (0.281)
-0.150 (0.059)**
rt
E4
ln�
� 2.129 (1.086)
0.106 (0.357)
0.028 (0.075)
rt
E5
ln�
� 3.905 (1.076)***
-0.284 (0.377)
0.070 (0.079)
UKftY ,
1ln �� 0.686 (2.869)
-0.173 (1.005)
0.873 (0.211)***
UKftY ,
2ln �� 8.783 (3.950)**
0.889 (1.383)
-0.393 (0.290)
UKftY ,
3ln �� -4.409 (3.954)
-1.136 (1.385)
0.291 (0.290)
UKftY ,
4ln �� -1.496 (4.009)
1.665 (1.404)
-0.130 (0.294)
UKftY ,
5ln �� -7.587 (3.657)**
-0.618 (1.281)
0.141 (0.268)
C -0.175 (0.108)
0.009 (0.038)
0.013 (0.008)
COUP 0.524 (0.129)
-0.003 (0.045)
-0.018 (0.009)*
Diagnostics 2R 0.856 0.421 0.790
Adjusted 2R 0.681 -0.281 0.535 ! 0.197 0.069 0.014
NX 2
4.336 [0.114]
270
HetX 2 10.254[0.975]
LM )(SCTest 11.764[0.227]
AR roots graph stable
271
Table SA32: Short–run coefficient estimates of Fiji’s goods export model with Singapore, 1975–2012
Short-run results SINGEXPGln� rEln� SINGfY ,ln�
1�tECT -0.307 (0.167)*
-0.013 (0.007)*
0.010 (0.005)*
SINGtEXPG 1ln �� 0.053
(0.185) -0.003 (0.008)
-0.000 (0.006)
rtE 1ln �� 2.939
(3.808) 0.139 (0.168)
0.304 (0.113)**
SINGftY ,
1ln �� 2.609 (5.740)
-0.163 (0.253)
0.136 (0.170)
C -0.121 (0.455)
0.015 (0.020)
0.052 (0.014)***
COUP -1.223 (0.738)*
0.048 (0.032)
0.028 (0.022)
Diagnostics 2R 0.216 0.324 0.274
Adjusted 2R 0.086 0.211 0.153
! 1.199 0.053 0.036 NX 2
1.561 [0.458]
HetX 2
11.080 [0.270]
LM )(SCTest 4.614 [0.867]
AR roots graph stable
272
Table SA33: Short–run coefficient estimates of Fiji’s goods export model with China, 1975–2012
Short-run results CHNEXPGln� rEln� CHNfY ,ln�
1�tECT -0.790 (0.304)**
0.023 (0.017)
0.002 (0.012)
CHNtEXPG 1ln �� -0.343
(0.303) -0.024 (0.017)
5.460 (0.004)
CHNtEXPG 2ln �� -0.047
(0.321) -0.013 (0.018)
0.003 (0.004)
CHNtEXPG 3ln �� 0.458
(0.296) -0.001 (0.016)
0.001 (0.003)
CHNtEXPG 4ln �� 0.709
(0.270)** -0.004 (0.015)
0.002 (0.003)
CHNtEXPG 5ln �� 0.670
(0.225)*** 0.003 (0.012)
0.004 (0.003)
CHNtEXPG 6ln �� 0.382
(0.157)** 0.007 (0.009)
0.001 (0.002)
rtE 1ln �� 3.581
(5.306) -0.048 (0.292)
-0.042 (0.062)
rtE 2ln �� 3.632
(5.086) -0.465 (0.280)
-0.182 (0.059)***
rtE 3ln �� 2.630
(5.790) -0.558 (0.319)*
-0.133 (0.067)*
rtE 4ln �� 1.825
(6.303) -0.061 (0.347)
0.005 (0.073)
rtE 5ln �� 10.935
(6.460)* 0.038 (0.356)
0.097 (0.075)
rtE 6ln �� 7.587
(5.896) 0.012 (0.325)
0.069 (0.068)
CHNftY ,
1ln �� -17.765 (21.269)
-1.628 (1.173)
0.311 (0.247)
CHNftY ,
2ln �� 3.869 (25.333)
1.226 (1.397)
-0.297 (0.294)
CHNftY ,
3ln �� 7.044 (24.849)
-0.222 (1.370)
0.219 (0.288)
CHNftY ,
4ln �� 2.483 (25.456)
1.154 (1.403)
-0.001 (0.295)
CHNftY ,
5ln �� -7.781 (21.017)
-0.557 (1.159)
-0.195 (0.244)
CHNftY ,
6ln �� -23.391 (17.170)
1.013 (0.947)
0.261 (0.199)
C 2.772 (3.886)
-0.077 (0.214)
0.067 (0.045)
COUP -0.383 (0.655)
0.064 (0.036)*
0.012 (0.008)
Diagnostics
273
2R 0.872 0.643 0.898
Adjusted 2R 0.616 -0.072 0.695
! 1.158 0.064 0.013NX 2 5.567
[0.062]
HetX 2 27.966[0.413]
LM )(SCTest 6.246[0.715]
AR roots graph stable
274
Table SA34: Short–run coefficient estimates of Fiji’s goods export model with Malaysia, 1975–2012
Short-run results MALAEXPGln� rEln� MALAfY ,ln�
1�tECT -0.140 (0.074)*
0.001 (0.006)
0.007 (0.003)***
MALAtEXPG 1ln �� -0.055
(0.170) 0.004 (0.013)
0.000 (0.008)
MALAtEXPG 2ln �� 0.131
(0.179) 0.012 (0.014)
0.013 (0.008)
rtE 1ln �� 0.811
(3.390) 0.092 (0.264)
0.008 (0.151)
rtE 2ln �� 3.378
(3.047) 0.002 (0.237)
-0.070 (0.136)
MALAftY ,
1ln �� 11.410 (5.040)**
-0.424 (0.392)
0.124 (0.225)
MALAftY ,
2ln �� -6.069 (5.580)
0.142 (0.434)
-0.276 (0.249)
C -0.337 (0.515)
0.013 (0.040)
0.068 (0.023)***
COUP -0.524 (0.478)
0.073 (0.037)*
0.013 (0.021)
Diagnostics 2R 0.408 0.260 0.342
Adjusted 2R 0.225 0.032 0.140
! 0.763 0.059 0.034 NX 2
0.420 [0.811]
HetX 2
13.867 [0.536]
LM )(SCTest 2.553 [0.979]
AR roots graph stable
275
Table SA35: Short–run coefficient estimates of Fiji’s goods export model with India, 1975–2012
Short-run results INDEXPGln� rEln� INDfY ,ln�
1�tECT -0.644 (0.131)***
-0.008 (0.012)
-0.012 (0.006)
INDtEXPG 1ln �� -0.402
(0.159)** 0.021 (0.015)
0.004 (0.007)
INDtEXPG 2ln �� -0.135
(0.192) 0.004 (0.018)
0.004 (0.008)
INDtEXPG 3ln �� 0.028
(0.164) -0.017 (0.015)
0.000 (0.007)
rtE 1ln �� 3.723
(2.405) 0.398
(0.226)* 0.196 (0.101)
rtE 2ln �� 3.951
(2.595) 0.080 (0.244)
0.079 (0.109)
rtE 3ln �� 4.089
(2.602) -0.136 (0.245)
-0.053 (0.110)
INDftY ,
1ln �� -9.492 (5.612)*
-0.548 (0.528)
-0.281 (0.247)
INDftY ,
2ln �� -11.011 (5.476)*
-0.093 (0.514)
-0.234 (0.231)
INDftY ,
3ln �� -17.389 (5.250)***
0.073 (0.493)
0.030 (0.221)
C 2.175 (0.649)***
0.031 (0.061)
0.082 (0.027)***
COUP -0.158 (0.304)
0.023 (0.029)
-0.003 (0.013)
Diagnostics 2R 0.696 0.320 0.378
Adjusted 2R 0.544 -0.020 0.067
! 0.657 0.062 0.028
NX 2
1.159 [0.560]
HetX 2
17.434 [0.685]
LM )(SCTest 7.986 [0.536]
AR roots graph stable
276
Table SA36: Short–run coefficient estimates of Fiji’s goods export model with Hong Kong, 1975–2012
Short-run results HKEXPGln� rEln� HKfY ,ln�
1�tECT -1.647 (0.228)***
0.011 (0.050)
-0.042 (0.033)
HKtEXPG 1ln �� 0.419
(0.141) -0.006 (0.031)
0.017 (0.020)
rtE 1ln �� -2.010
(0.914)** -0.271 (0.202)
-0.030 (0.132)
HKftY ,
1ln �� 1.036 (1.384)
0.016 (0.306)
-0.239 (0.200)
C -0.090 (0.063)
-0.004 (0.014)
-0.006 (0.009)
COUP 0.300 (0.165)*
0.013 (0.036)
0.009 (0.024)
Diagnostics 2R 0.696 0.090 0.154
Adjusted 2R 0.644 -0.067 0.008
! 0.326 0.072 0.047
NX 2
1.349 [0.509]
HetX 2
23.615 [0.652]
LM )(SCTest 10.626 [0.302]
AR roots graph stable
277
Table SA37: Short–run coefficient estimates of Fiji’s goods import model, 1975–2012
Short-run results IMPGln� rEln� Yln�
1�tECT -0.632 (0.256)**
-0.229 (0.120)*
-0.149 (0.084)*
1ln �� tIMPG 0.172 (0.215)
0.052 (0.101)
0.187 (0.071)**
rtE 1ln �� 0.022
(0.461) -0.050 (0.216)
-0.116 (0.152)
1ln �� tY -0.724 (0.576)
-0.479 (0.270)*
-0.671 (0.190)***
C 0.039 (0.026)
0.009 (0.012)
0.037 (0.008)***
COUP 0.005 (0.044)
0.025 (0.021)
-0.024 (0.015)
Diagnostics 2R 0.314 0.303 0.331
Adjusted 2R 0.199 0.187 0.219
! 0.114 0.054 0.038 NX 2
0.355 [0.837]
HetX 2
6.692 [0.669]
LM )(SCTest 6.827 [0.655]
AR roots graph stable
278
Table SA38: Short–run coefficient estimates of Fiji’s services import model, 1975–2012
Short-run results IMPSln� rEln� Yln�
1�tECT -0.104 (0.048)**
0.057 (0.025)**
-0.059 (0.018)***
1ln �� tIMPS 0.164 (0.175)
0.248 (0.091)**
0.054 (0.065)
rtE 1ln �� 0.015
(0.319) 0.311
(0.165)* -0.144 (0.119)
1ln �� tY -0.034 (0.398)
-0.540 (0.205)**
-0.497 (0.148)***
C 0.033 (0.018)*
0.003 (0.009)
0.033 (0.007)***
COUP -0.036 (0.044)
0.033 (0.023)
-0.024 (0.016)
Diagnostics 2R 0.240 0.468 0.463
Adjusted 2R 0.113 0.379 0.373
! 0.091 0.047 0.034 NX 2
2.020 [0.364]
HetX 2
9.672 [0.378]
LM )(SCTest 2.242 [0.987]
AR roots graph stable
279
Table SA39: Short–run coefficient estimates of Fiji’s goods and services import model, 1975–2012
Short-run results IMPGSln� rEln� Yln�
1�tECT -0.890 (0.310)***
-0.326 (0.211)
-0.154 (0.148)
1ln �� tIMPGS 0.216 (0.258)
0.267 (0.175)
0.190 (0.123)
2ln �� tIMPGS 0.275 (0.206)
0.042 (0.140)
0.040 (0.098)
rtE 1ln �� 0.033
(0.399) -0.024 (0.271)
-0.523 (0.190)
rtE 2ln �� 0.484
(0.329) -0.333 (0.223)
0.030 (0.157)
1ln �� tY -0.468 (0.600)
-0.789 (0.407)*
-0.631 (0.286)**
2ln �� tY 0.363 (0.583)
-0.127 (0.396)
0.042 (0.278)
C 0.023 (0.022)
0.014 (0.015)
0.031 (0.010)***
COUP -0.084 (0.041)*
0.047 (0.028)
-0.044 (0.020)**
Diagnostics 2R 0.561 0.360 0.373
Adjusted 2R 0.426 0.163 0.180
! 0.081 0.055 0.039
NX 2
0.913 [0.633]
HetX 2
8.673 [0.894]
LM )(SCTest 4.275 [0.892]
AR roots graph stable
280
Table SA40: Short–run coefficient estimates of Fiji’s import of food model, 1975–2012
Short-run results FOODIMPln� rEln� Yln�
1�tECT -0.145 (0.076)*
0.027 (0.056)
-0.079 (0.028)
1ln �� tFOODIMP -0.159 (0.242)
-0.078 (0.177)
0.001 (0.089)***
2ln �� tFOODIMP 0.164 (0.257)
-0.022 (0.188)
-0.134 (0.095)
3ln �� tFOODIMP -0.255 (0.230)
-0.032 (0.167)
0.031 (0.085)
4ln �� tFOODIMP -0.274 (0.225)
-0.201 (0.164)
-0.103 (0.083)
5ln �� tFOODIMP -0.061 (0.204)
0.044 (0.148)
0.050 (0.075)
6ln �� tFOODIMP -0.261 (0.229)
-0.264 (0.167)
0.036 (0.084)
rtE 1ln �� -0.697
(0.521) 0.300 (0.380)
-0.125 (0.192)
rtE 2ln �� 0.144
(0.499) -0.251 (0.364)
-0.105 (0.184)
rtE 3ln �� 0.337
(0.456) 0.005 (0333)
-0.086 (0.168)
rtE 4ln �� -1.073
(0.481)** 0.201 (0.351)
0.045 (0.177)
rtE 5ln �� -0.470
(0.583) -0.432 (0.425)
-0.125 (0.215)
rtE 6ln �� 0.688
(0.536) 0.236 (0.391)
0.076 (0.198)
1ln �� tY -0.767 (0.623)
-0.051 (0.454)
-0.523 (0.230)**
2ln �� tY -0.420 (0.809)
0.048 (0.590)
-0.438 (0.298)
3ln �� tY -0.143 (0.658)
0.117 (0.480)
-0.337 (0.243)
4ln �� tY -0.444 (0.628)
-0.189 (0.458)
-0.315 (0.232)
5ln �� tY -0.720 (0.700)
-0.219 (0.510)
-0.189 (0.258)
6ln �� tY -0.415 (0.638)
0.006 (0.465)
0.062 (0.235)
C 0.101 (0.067)
0.022 (0.049)
0.064 (0.025)**
COUP 0.079 (0.074)
0.059 (0.054)
-0.053 (0.027)*
Diagnostics
281
2R 0.768 0.588 0.756
Adjusted 2R 0.304 -0.236 0.269! 0.094 0.069 0.035
NX 2 0.959[0.619]
HetX 2 25.232[0.562]
LM )(SCTest 10.849[0.286]
AR roots graph stable
282
Table SA41: Short–run coefficient estimates of Fiji’s import of fuel model, 1975–2012
Short-run results FUELIMPln� rEln� Yln�
1�tECT -0.827 (0.336)**
-0.119 (0.052)**
-0.069 (0.051)
FUELtIMP 1ln �� 0.319
(0.317) -0.086 (0.103)
0.085 (0.102)
FUELtIMP 2ln �� 0.913
(0.329)*** -0.244
(0.107)** 0.069 (0.106)
FUELtIMP 3ln �� 0.356
(0.266) -0.0247
(0.087)*** -0.099 (0.085)
FUELtIMP 4ln �� 0.890
(0.326)*** -0.316
(0.106)*** -0.133 (0.105)
FUELtIMP 5ln �� 0.632
(0.405) -0.226 (0.132)*
-0.073 (0.130)
FUELtIMP 6ln �� -0.433
(0.268) -0.026 (0.087)
-0.144 (0.086)
FUELtIMP 7ln �� 0.743
(0.231)*** -0.272
(0.075)*** 0.048 (0.074)
rtE 1ln �� 5.485
(0.865)*** -0.567 (0.282)*
0.341 (0.278)
rtE 2ln �� 2.843
(1.328)** -0.558 (0.433)
0.101 (0.427)
rtE 3ln �� 0.213
(1.239) -0.241 (0.404)
-0.135 (0.399)
rtE 4ln �� 0.753
(1.065) 0.288 (0.347)
0.520 (0.343)
rtE 5ln �� -2.468
(1.194)** 0.697
(0.389)* 0.463 (0.384)
rtE 6ln �� 1.188
(1.468) 0.168 (0.479)
0.530 (0.472)
rtE 7ln �� 2.099
(0.855)** -0.348 (0.279)
0.579 (0.275)**
1ln �� tY 5.281 (2.438)**
-2.113 (0.795)**
-1.340 (0.784)*
2ln �� tY 2.150 (2.709)
-0.830 (0.884)
-1.489 (0.872)*
3ln �� tY -0.187 (2.602)
-0.487 (0.849)
-1.049 (0.837)*
4ln �� tY -0.190 (1.918)
-0.129 (0.625)
-0.167 (0.617)
5ln �� tY -2.249 (1.854)
0.842 (0.605)
0.635 (0.597)
6ln �� tY 0.436 (1.916)
0.634 (0.625)
0.810 (0.616)
7ln �� tY 1.439 (1.225)
0.182 (0.399)
0.507 (0.394)
283
-0.244 (0.140)
0.052 (0.046)
0.026 (0.045)
COUP -0.321 (0.116)***
0.148 (0.038)***
0.011 (0.037)
Diagnostics 0.947 0.927 0.813
0.744 0.646 0.095
0.114 0.037 0.037
2.224 [0.329]
29.342 [0.345]
11.493 [0.243]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
284
Table SA42: Short–run coefficient estimates of Fiji’s import of manufactured goods model, 1975–2012
Short-run results MANUIMPln� rEln� Yln�
1�tECT -0.214 (0.057)***
0.017 (0.060)
-0.095 (0.039)**
MANUtIMP 1ln �� 0.272
(0.156)* 0.001 (0.052)
0.112 (0.034)***
rtE 1ln �� -1.178
(0.621)* -0.257 (0.201)
-0.229 (0.135)*
1ln �� tY -1.307 (0.829)
-0.563 (0.275)**
-0.576 (0.180)***
0.066 (0.031)**
0.011 (0.010)
0.032 (0.007)***
COUP -0.095 (0.102)
0.080 (0.034)**
-0.025 (0.022)
Diagnostics
0.585 0.317 0.429
0.515 0.203 0.334
0.160 0.053 0.035
1.150 [0.563]
4.633 [0.865]
7.751 [0.560]
AR roots graph stable
C
2RAdjusted 2R
!
NX 2
HetX 2
LM )(SCTest
285
Table SA43: Short–run coefficient estimates of Fiji’s import of crude oil model, 1975–2012
Short-run results CRUDIMPln� rEln� Yln�
1�tECT -0.363 (0.154)**
-0.037 (0.037)
-0.062 (0.024)**
CRUDtIMP 1ln �� -0.290
(0.167)* 0.009 (0.041)
0.059 (0.027)**
rtE 1ln �� -0.400
(0.787) 0.092 (0.191)
-0.090 (0.126)
1ln �� tY -0.338 (0.931)
-0.440 (0.226)*
-0.490 (0.149)***
0.041 (0.045)
0.009 (0.011)
0.036 (0.007)***
COUP 0.106 (0.107)
0.048 (0.026)*
-0.044 (0.017)**
Diagnostics 0.339 0.290 0.403
0.229 0.171 0.303
0.223 0.054 0.036
1.386 [0.500]
6.022 [0.738]
7.747 [0.560]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
286
Table SA44: Short–run coefficient estimates of Fiji’s import of textile model, 1975–2012
Short-run results TEXTIMPln� rEln� Yln�
1�tECT -0.054 (0.028)*
-0.019 (0.010)*
0.015 (0.005)***
TEXTtIMP 1ln �� -0.015
(0.196) 0.038 (0.067)
0.035 (0.037)
TEXTtIMP 2ln �� -0.267
(0.189) 0.066 (0.064)
-0.013 (0.036)
rtE 1ln �� 2.338
(0.698)*** 0.274 (0.237)
-0.110 (0.133)
rtE 2ln �� 1.906
(0.682)*** -0.078 (0.232)
-0.027 (0.130)
1ln �� tY 1.963 (1.075)*
-0.198 (0.366)
-0.543 (0.204)**
2ln �� tY 2.291 (1.097)**
0.234 (0.373)
-0.272 (0.208)
-0.079 (0.045)*
-0.010 (0.015)
0.036 (0.009)***
COUP -0.210 (0.107)*
0.050 (0.036)
-0.017 (0.020)
Diagnostics 0.508 0.386 0.496
0.321 0.152 0.304
0.170 0.058 0.032
0.885 [0.642]
12.192 [0.968]
5.745 [0.765]
AR roots graph stable
C
2RAdjusted 2R
!
NX 2
HetX 2
LM )(SCTest
287
Table SA45: Short–run coefficient estimates of Fiji’s import of machinery and transport equipment model, 1975–2012
Short-run results EQUIIMPln� rEln� fYln�
1�tECT -0.943 (0.180)***
0.017 (0.060)
-0.095 (0.039)**
EQUItIMP 1ln �� 0.272
(0.156)* 0.001 (0.052)
0.112 (0.034)***
rtE 1ln �� -1.178
(0.621)* -0.257 (0.201)
-0.229 (0.135)*
ftY 1ln �� -1.307
(0.829) -0.563
(0.275)** -0.576
(0.180)*** 0.066
(0.031)** 0.011 (0.010)
0.032 (0.007)***
COUP -0.095 (0.102)
0.080 (0.034)**
-0.025 (0.022)
Diagnostics 0.585 0.317 0.429
0.515 0.203 0.334 0.160 0.053 0.035
1.401 [0.496]
8.644 [0.471]
7.742 [0.560]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
288
Table SA46: Short–run coefficient estimates of Fiji’s import of tobacco and beverage model, 1975–2012
Short-run results TOBAIMPln� rEln� Yln�
1�tECT -0.581 (0.165)***
0.068 (0.095)
0.032 (0.058)
TOBAtIMP 1ln �� 0.287
(0.220) -0.181 (0.126)
0.062 (0.077)
TOBAtIMP 2ln �� 0.389
(0.221)* -0.261
(0.126)** -0.001 (0.077)
TOBAtIMP 3ln �� 0.051
(0.196) -0.062 (0.112)
0.012 (0.068)
TOBAtIMP 4ln �� -0.016
(0.169) -0.121 (0.097)
0.023 (0.059)
TOBAtIMP 5ln �� 0.055
(0.157) 0.059 (0.090)
0.075 (0.055)
TOBAtIMP 6ln �� -0.250
(0.168) -0.065 (0.096)
-0.110 (0.058)*
TOBAtIMP 7ln �� 0.384
(0.164)** -0.093 (0.094)
0.004 (0.057)
rtE 1ln �� 4.417
(0.680)*** -0.228 (0.389)
0.246 (0.237)
rtE 2ln �� 0.618
(0.866) -0.379 (0.495)
-0.159 (0.302)
rtE 3ln �� 1.754
(1.646)*** -0.130 (0.369)
0.218 (0.225)
rtE 4ln �� 1.101
(0.640)* -0.145 (0.366)
-0.064 (0.223)
rtE 5ln �� 0.487
(0.582) -0.141 (0.333)
-0.070 (0.203)
rtE 6ln �� 2.611
(0.596)*** -0.878
(0.340)** 0.120 (0.207)
rtE 7ln �� 1.201
(0.592)** 0.114 (0.338)
0.184 (0.206)
1ln �� tY 1.022 (1.008)
0.163 (0.576)
-0.366 (0.351)
2ln �� tY -0.161 (0.879)
0.893 (0.502)*
-0.113 (0.306)
3ln �� tY 0.234 (0.899)
-0.024 (0.514)
0.015 (0.313)
4ln �� tY 1.278 (0.824)
-0.278 (0.471)
0.019 (0.287)
5ln �� tY 1.282 (1.895)
-0.862 (0.511)*
0.114 (0.312)
6ln �� tY 1.988 (0.995)*
-0.731 (0.569)
0.516 (0.347)
289
7ln �� tY 0.607 (0.856)
-0.843 (0.489)*
0.183 (0.347)
-0.252 (0.097)**
0.089 (0.055)
0.001 (0.034)
COUP -0.270 (0.114)**
0.044 (0.065)
-0.012 (0.040)
Diagnostics 0.931 0.835 0.839
0.669 0.202 0.220
0.098 0.056 0.034
1.043 [0.594]
28.511 [0.385]
7.741 [0.560]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
290
Table SA47: Short–run coefficient estimates of Fiji’s import of chemical model, 1975–2012
Short-run results CHEMIMPln� rEln� Yln�
1�tECT -0.492 (0.135)***
-0.022 (0.069)
-0.099 (0.048)**
CHEMtIMP 1ln �� 0.017
(0.159) -0.015 (0.082)
0.071 (0.056)
rtE 1ln �� 0.142
(0.370) 0.196 (0.190)
-0.085 (0.132)
1ln �� tY -1.441 (0.517)***
-0.380 (0.266)
-0.504 (0.184)**
0.054 (0.025)**
0.006 (0.013)
0.034 (0.009)***
COUP 0.036 (0.044)
0.027 (0.022)
-0.016 (0.015)
Diagnostics 0.517 0.220 0.276
0.436 0.091 0.156
0.110 0.057 0.039
5.038 [0.081]
5.437 [0.795]
7.138 [0.623]
AR roots graph stable
C
2RAdjusted 2R
!
NX 2
HetX 2
LM )(SCTest
291
Table SA48: Short–run coefficient estimates of Fiji’s import of oil and fats model, 1975–2012
Short-run results OILIMPln� rEln� Yln�
1�tECT -0.542** (0.257)
0.156 (0.095)
-0.093* (0.051)
OILtIMP 1ln �� 0.023
(0.330) -0.162 (0.121)
0.088 (0.065)
OILtIMP 2ln �� 0.224
(0.364) -0.136 (0.134)
0.035 (0.072)
OILtIMP 3ln �� 0.554
(0.358) -0.073 (0.132)
-0.001 (0.071)
OILtIMP 4ln �� 0.590**
(0.270) -0.091 (0.099)
-0.042 (0.053)
rtE 1ln �� -1.642*
(0.830) 0.534* (0.305)
-0.070 (0.164)
rtE 2ln �� -0.511
(0.771) 0.055 (0.284)
0.064 (0.152)
rtE 3ln �� 0.588
(0.722) 0.018 (0.266)
-0.140 (0.142)
rtE 4ln �� -2.080**
(0.813) 0.282 (0.299)
-0.056 (0.160)
1ln �� tY -1.934** (1.903)
-0.176 (0.332)
-0.472** (0.178)
2ln �� tY -3.327*** (1.153)
0.267 (0.424)
-0.040 (0.227)
3ln �� tY -0.838 (1.140)
-0.089 (0.419)
-0.054 (0.225)
4ln �� tY 2.045* (1.023)
-0.142 (0.376)
-0.333 (0.202)
0.052 (0.074)
0.007 (0.027)
0.045*** (0.015)
COUP 0.173* (0.096)
0.019 (0.035)
-0.038* (0.019)
Diagnostics 0.650 0.411 0.604
0.378 -0.048 0.296
0.172 0.063 0.034
0.486 [0.784]
20.460 [0.811]
12.290 [0.198]
AR roots graph stable
C
2RAdjusted 2R
!
NX 2
HetX 2
LM )(SCTest
292
Table SA49: Short–run coefficient estimates of Fiji’s import of miscellaneous manufactured goods model, 1975–2012
Short-run results MISCIMPln� rEln� Yln�
1�tECT -1.016 (0.401)**
0.117 (0.215)
0.162 (0.130)
MISCtIMP 1ln �� -0.263
(0.333) -0.038 (0.178)
-0.137 (0.108)
MISCtIMP 2ln �� -0.091
(0.252) -0.037 (0.135)
-0.089 (0.082)
MISCtIMP 3ln �� 0.029
(0.174) -0.020 (0.093)
0.013 (0.057)
rtE 1ln �� 0.022
(0.615) -0.320 (0.329)
0.220 (0.199)
rtE 2ln �� 1.381
(0.622)** -0.389 (0.333)
0.430 (0.202)**
rtE 3ln �� 1.644
(0.519)*** -0.371 (0.278)
0.364 (0.168)**
1ln �� tY -3.971 (1.685)**
0.113 (0.902)
-0.351 (0.547)
2ln �� tY -2.510 (1.310)*
0.128 (0.701)
-0.045 (0.425)
3ln �� tY -1.335 (0.762)*
0.027 (0.408)
0.073 (0.247)
0.021 (0.028)
-0.008 (0.015)
0.005 (0.009)
COUP -0.196 (0.078)**
0.047 (0.042)
-0.048 (0.025)*
Diagnostics 0.759 0.304 0.740
0.633 -0.061 0.604
0.138 0.074 0.045
1.836 [0.399]
31.858 [0.237]
6.599 [0.679]
AR roots graph stable
C
2RAdjusted 2R
!
NX 2
HetX 2
LM )(SCTest
293
Table SA50: Short–run coefficient estimates of Fiji’s import of travel services model, 1975–2012
Short-run results TRAVIMPln� rEln� Yln�
1�tECT -0.335 (0.153)**
0.029 (0.059)
-0.101 (0.035)***
1ln �� tTRAVIMP -0.090 (0.213)
0.144 (0.083)*
0.021 (0.049)
2ln �� tTRAVIMP 0.029 (0.259)
0.084 (0.101)
0.003 (0.060)
3ln �� tTRAVIMP 0.315 (0.253)
0.093 (0.098)
0.072 (0.058)
4ln �� tTRAVIMP 0.657 (0.255)**
0.164 (0.099)
-0.024 (0.059)
5ln �� tTRAVIMP 0.658 (0.210)***
0.088 (0.082)
0.005 (0.049)
rtE 1ln �� -0.928
(0.927) -0.002 (0.361)
-0.275 (0.214)
rtE 2ln �� -1.849
(0.969)* -0.175 (0.377)
-0.309 (0.224)
rtE 3ln �� -1.886
(0.860)** -0.210 (0.335)
-0.278 (0.199)
rtE 4ln �� -2.974
(0.924)*** -0.085 (0.360)
-0.141 (0.214)
rtE 5ln �� -0.956
(0.698) -0.108 (0.272)
-0.273 (0.161)
1ln �� tY 0.110 (1.009)
-0.686 (0.393)*
-0.380 (0.233)
2ln �� tY -1.289 (1.160)
-0.435 (0.452)
0.027 (0.268)
3ln �� tY -1.345 (0.034)
-0.158 (0.403)
-0.207 (0.239)
4ln �� tY -2.766 (1.029)**
-0.324 (0.401)
-0.353 (0.238)
5ln �� tY -1.808 (1.063)*
-0.328 (0.414)
0.026 (0.246)
0.159 (0.076)**
0.028 (0.029)
0.045 (0.018)**
COUP 0.522 (0.128)***
0.074 (0.050)
-0.004 (0.030)
Diagnostics 0.784 0.617 0.696
0.521 0.153 0.328
0.144 0.056 0.033
0.225 [0.894]
C
2RAdjusted 2R
!NX 2
294
31.834[0.238]7.310[0.605]
AR roots graph stable
HetX 2
LM )(SCTest
295
Table SA51: Short–run coefficient estimates of Fiji’s import of transportation services model, 1975–2012
Short-run results TRANIMPln� rEln� Yln�
1�tECT -0.695 (0.208)***
0.230 (0.077)***
-0.091 (0.048)*
1ln �� tTRANIMP 0.434 (0.286)
-0.065 (0.106)
0.119 (0.066)*
2ln �� tTRANIMP 0.275044 (0.287)
-0.091 (0.106)
0.038 (0.066)
3ln �� tTRANIMP 0.144 (0.238)
-0.144 (0.088)
-0.002 (0.055)
4ln �� tTRANIMP 0.496 (0.278)*
-0.271 (0.103)**
0.115 (0.064)*
rtE 1ln �� 0.290
(0.812) 0.421 (0.300)
0.112 (0.188)
rtE 2ln �� -0.079
(0.764) 0.131 (0.282)
-0.027 (0.177)
rtE 3ln �� -1.444
(0.751)* 0.068 (0.277)
-0.187 (0.174)
rtE 4ln �� -0.713
(0.685) 0.333 (0.253)
0.012 (0.159)
1ln �� tY -1.317 (1.085)
-0.203 (0.401)
-0.538 (0.251)**
2ln �� tY -0.596 (1.361)
0.200 (0.502)
0.133 (0.315)
3ln �� tY -2.124 (0.132)*
0.344 (0.418)
-0.054 (0.262)
4�� tLY -1.696 (1.091)
0.628 (0.403)
-0.188 (0.252)
0.144 (0.069)**
0.002 (0.026)
0.028 (0.016)*
COUP -0.132 (0.083)
0.051 (0.030)
-0.047 (0.019)**
Diagnostics 0.623 0.694 0.622
0.217 0.364 0.216
0.140 0.051 0.032
2.153 [0.341]
24.760 [0.258]
7.913 [0.543]
AR roots graph stable
C
2RAdjusted 2R
!NX 2
HetX 2
LM )(SCTest
296
Table SA52: Short–run coefficient estimates of Fiji’s goods import model with Australia, 1975–2012
Short-run results AUSIMPGln� AUSrE ,ln� Yln�
1�tECT -0.513 (0.239)**
-0.239 (0.152)
-0.013 (0.079)
AUStIMPG 1ln �� 0.248
(0.295) 0.058 (0.187)
-0.013 (0.097)
AUStIMPG 2ln �� 0.620
(0.374)* 0.066 (0.237)
0.044 (0.123)
AUStIMPG 3ln �� 0.522
(0.325) 0.473
(0.206)** 0.105 (0.107)
AUStIMPG 4ln ��
0.268 (0.308)
0.244 (0.196)
0.010 (0.101)
AUStIMPG 5ln ��
0.217 (0.244)
-0.057 (0.155)
0.070 (0.080)
AUSrt
E ,1
ln�
� 0.528 (0.315)
-0.406 (0.200)**
0.023 (0.103)
AUSrt
E ,2
ln�
� 0.252 (0.333)
-0.548 (0.211)**
-0.029 (0.109)
AUSrt
E ,3
ln�
� -0.325 (0.359)
-0.363 (0.228)
0.065 (0.118)
AUSrt
E ,4
ln�
� -0.684 (0.468)
-0.404 (0.297)
-0.136 (0.154)
AUSrt
E ,5
ln�
� -0.017 (0.504)
-0.865 (0.320)**
0.002 (0.166)
1ln �� tY 0.837 (1.038)
-0.951 (0.658)
-0.337 (0.341)
2ln �� tY 0.724 (0.975)
-0.214 (0.618)
-0.070 (0.320)
3ln �� tY -1.022 (1.023)
-0.629 (0.649)
-0.238 (0.336)
4ln �� tY 0.096 (1.115)
-1.311 (0.707)
-0.256 (0.366)
5ln �� tY 0.831 (0.954)
-0.478 (0.605)
-0.066 (0.313)
C -0.120 (0.075)
0.056 (0.047)
0.026 (0.025)
COUP 0.178 (0.101)*
0.147 (0.064)**
0.010 (0.033)
Diagnostics 2R 0.608 0.643 0.338
Adjusted 2R 0.132 0.210 -0.465 ! 0.150 0.095 0.049
297
NX 2 0.910[0.635]
HetX 2 31.705[0.243]
LM )(SCTest 4.929[0.841]
AR roots graph stable
298
Table SA53: Short–run coefficient estimates of Fiji’s goods import model with New Zealand, 1975–2012
Short-run results NZIMPGln� NZrE ,ln� Yln�
1�tECT -0.823 (0.161)***
0.168 (0.359)
-0.171 (0.105)
NZtIMPG 1ln �� 0.442
(0.167)** -0.314 (0.374)
0.168 (0.109)
NZtIMPG 2ln �� 0.509
(0.180)*** 0.013 (0.401)
0.153 (0.117)
NZtIMPG 3ln �� 0.632
(0.151)*** -0.303 (0.338)
0.138 (0.099)
NZtIMPG 4ln ��
0.568 (0.159)***
-0.030 (0.354)
0.057 (0.104)
NZtIMPG 5ln ��
0.213 (0.129)
0.036 (0.288)
0.031 (0.084)
NZrtE ,
1ln �� -0.263 (0.153)*
0.074 (0.343)
-0.117 (0.100)
NZrtE ,
2ln �� -0.379
(0.154)** 0.001 (0.344)
-0.109 (0.100)
NZrtE ,
3ln �� 0.014 (0.135)
-0.309 (0.302)
-0.006 (0.088)
NZrtE ,
4ln �� -0.219 (0.140)
-0.020 (0.313)
-0.115 (0.092)
NZrtE ,
5ln �� -0.597 (0.166)***
-0.269 (0.370)
-0.081 (0.108)
1ln �� tY -0.700 (0.438)
0.748 (0.979)
-0.613 (0.286)
2ln �� tY -1.149 (0.450)**
0.238 (1.006)
-0.231 (0.294)
3ln �� tY -1.666 (0.427)
0.123 (0.954)
-0.206 (0.279)
4ln �� tY -0.357 (0.412)
-1.165 (0.921)
-0.280 (0.269)
5ln �� tY 0.113 (0.396)
-1.357 (0.884)
0.076 (0.259)
C 0.029 (0.027)
0.082 (0.060)
0.037 (0.018)
COUP 0.123 (0.036)
0.023 (0.081)
-0.022 (0.024)
Diagnostics 2R 0.876 0.502 0.507
Adjusted 2R 0.725 -0.103 -0.092 ! 0.065 0.146 0.043
NX 2
0.058 [0.971]
299
HetX 2 24.595[0.597]
LM )(SCTest 6.628[0.676]
AR roots graph stable
300
Table SA54: Short–run coefficient estimates of Fiji’s goods import model with Japan, 1975–2012
Short-run results JPNIMPGln� JPNrE ,ln� Yln�
1�tECT -0.589 (0.161)***
0.097 (0.162)
-0.072 (0.040)*
JPNtIMPG 1ln �� -0.132
(0.167) -0.054 (0.168)
0.075 (0.041)*
JPNtIMPG 2ln �� 0.149
(0.175) -0.049 (0.176)
-0.019 (0.043)
JPNrtE ,
1ln �� -0.670 (0.235)***
0.317 (0.236)
-0.011 (0.058)
JPNrtE ,
2ln �� -0.173 (0.288)
-0.135 (0.289)
-0.055 (0.071)
1ln �� tY -0.142 (0.818)
0.351 (0.822)
-0.280 (0.202)
2ln �� tY -0.264 (0.762)
-0.565 (0.766)
0.133 (0.188)
C -0.001 (0.040)
0.013 (0.040)
0.030 (0.010)***
COUP -0.024 (0.081)
-0.022 (0.081)
-0.027 (0.020)
Diagnostics 2R 0.515 0.121 0.403
Adjusted 2R 0.365 -0.149 0.219 ! 0.153 0.154 0.038
NX 2
0.439 [0.803]
HetX 2
19.727 [0.183]
LM )(SCTest 4.302 [0.890]
AR roots graph stable
301
Table SA55: Short–run coefficient estimates of Fiji’s goods import model with the USA, 1975–2012
Short-run results USAIMPGln� USArE ,ln� Yln�
1�tECT -0.686 (0.231)***
0.058 (0.046)
-0.022 (0.020)
USAtIMPG 1ln �� -0.298
(0.223) -0.087 (0.044)*
0.014 (0.020)
USAtIMPG 2ln �� -0.110
(0.205) -0.049 (0.041)
-0.008 (0.018)
USArt
E ,1
ln�
� -0.446 (0.950)
0.168 (0.189)
0.057 (0.083)
USArt
E ,2
ln�
� -0.431 (0.895)
-0.106 (0.178)
-0.071 (0.078)
1ln �� tY 1.790 (2.190)
0.449 (0.437)
-0.327 (0.192)*
2ln �� tY 2.519 (2.243)
-0.316 (0.447)
0.161 (0.197)
C 0.009 (0.105)
-0.005 (0.021)
0.031 (0.009)***
COUP -0.355 (0.251)
0.040 (0.050)
-0.056 (0.022)**
Diagnostics 2R 0.470 0.225 0.347
Adjusted 2R 0.307 -0.014 0.146 ! 0.451 0.090 0.039
NX 2
2.043 [0.360]
HetX 2
8.938 [0.881]
LM )(SCTest 13.393 [0.146]
AR roots graph stable
302
Table SA56: Short–run coefficient estimates of Fiji’s goods import model with the UK, 1975–2012
Short-run results UKIMPGln� rEln� Yln�
1�tECT -0.624* (0.307)
0.138 (0.076)
-0.123** (0.052)
UKtIMPG 1ln �� -0.307
(0.301) 0.043 (0.074)
0.101* (0.050)
UKtIMPG 2ln �� 0.066
(0.301) 0.078 (0.077)
0.068 (0.052)
UKtIMPG 3ln �� 0.216
(0.306) 0.046 (0.076)
0.049 (0.051)
UKtIMPG 4ln ��
0.159 (0.231)
-0.051 (0.057)
0.027 (0.039)
UKtIMPG 5ln ��
-0.192 (0.243)
0.040 (0.060)
0.001 (0.041)
rt
E1
ln�
� -3.367*** (1.197)
0.245 (0.296)
-0.223 (0.201)
rt
E2
ln�
� 0.683 (1.291)
-0.040 (0.319)
-0.081 (0.216)
rt
E3
ln�
� 0.435 (1.129)
-0.135 (0.279)
-0.170 (0.189)
rt
E4
ln�
� -2.036 (1.309)
0.008 (0.334)
-0.158 (0.220)
rt
E5
ln�
� -1.482 (1.231)
-0.399 (0.305)
-0.180 (0.207)
1ln �� tY -0.366 (1.540)
-0.271 (0.381)
-0.106 (0.258)
2ln �� tY 0.240 (1.694)
0.100 (0.419)
0.233 (0.284)
3ln �� tY 0.969 (1.495)
-0.026 (0.370)
0.200 (0.251)
4ln �� tY -0.480 (1.496)
-0.341 (0.370)
0.501 (0.251)
5ln �� tY -2.892**
(1.350) -0.359 (0.334)
0.137 (0.226)
C 0.064 (0.117)
0.037 (0.029)
0.036 (0.020)
COUP -0.075 (0.155)
0.042 (0.038)
-0.034 (0.026)
Diagnostics 2R 0.736 0.579 0.565
Adjusted 2R 0.414 0.069 0.037 ! 0.238 0.059 0.040
NX 2 0.924
303
[0.630]
HetX 2 30.358[0.298]
LM )(SCTest 6.904[0.647]
AR roots graph stable
304
Table SA57: Short–run coefficient estimates of Fiji’s goods import model with Singapore, 1975–2012
Short-run results SINGIMPGln� rEln� Yln�
1�tECT -0.214 (0.104)**
0.022 (0.017)
-0.025 (0.011)**
SINGtIMPG 1ln �� -0.051
(0.286) 0.007 (0.046)
-0.028 (0.029)
SINGtIMPG 2ln �� 0.236
(0.240) -0.090
(0.039)** 0.026 (0.024)
SINGtIMPG 3ln �� 0.267
(0.332) -0.124
(0.053)** 0.041 (0.034)
SINGtIMPG 4ln ��
0.140 (0.275)
-0.021 (0.044)
-0.003 (0.028)
SINGtIMPG 5ln ��
-0.383 (0.248)
-0.002 (0.040)
0.001 (0.025)
SINGtIMPG 6ln ��
-0.261 (0.232)
0.049 (0.037)
-0.026 (0.024)
SINGtIMPG 7ln ��
-0.050 (0.208)
-0.008 (0.033)
-0.031 (0.021)
rtE 1ln �� -1.925
(2.056) 0.188 (0.330)
-0.219 (0.208)
rtE 2ln �� -4.768
(2.354)** -0.054 (0.378)
-0.144 (0.238)
rtE 3ln �� -4.470
(2.312)* -0.086 (0.371)
-0.245 (0.234)
rtE 4ln �� -2.522
(1.948) 0.155 (0.313)
-0.094 (0.197)
rtE 5ln �� 0.076
(1.936) -0.094 (0.311)
-0.253 (0.196)
rtE 6ln �� 1.166
(2.108) -0.294 (0.338)
0.089 (0.214)
rtE 7ln �� -1.945
(1.736) -0.108 (0.279)
0.269 (0.176)
1ln �� tY 1.133 (2.692)
0.184 (0.432)
-0.893 (0.273)***
2ln �� tY -7.393 (3.594)**
0.985 (0.577)*
-0.641 (0.364)*
3ln �� tY -10.336 (4.974)**
1.141 (0.798)
-0.931 (0.504)*
4ln �� tY -9.259 (4.629)*
0.345 (0.743)
-0.767 (0.469)
5ln �� tY -2.829 (3.370)
-0.165 (0.541)
-0.285 (0.341)
6ln �� tY 1.873 (3.362)
-0.516 (0.540)
-0.073 (0.341)
305
7ln �� tY -3.447 (3.131)
-1.070 (0.503)**
0.363 (0.317)
C 0.778 (0.445)*
-0.000 (0.071)
0.091 (0.045)
COUP 0.123 (0.228)
0.067 (0.037)
-0.022 (0.023)
Diagnostics 2R 0.838 0.813 0.805
Adjusted 2R 0.218 0.098 0.056 ! 0.370 0.059 0.038
NX 2
0.028 [0.986]
HetX 2
27.327 [0.446]
LM )(SCTest 5.087 [0.827]
AR roots graph stable
306
Table SA58: Short–run coefficient estimates of Fiji’s goods import model with China, 1975–2012
Short-run results CHNIMPGln� rEln� Yln�
1�tECT -0.610 (0.321)*
0.189 (0.060)***
-0.189 (0.047)***
CHNtIMPG 1ln �� 0.062
(0.437) -0.099 (0.082)
0.162 (0.064)**
CHNtIMPG 2ln �� 1.080
(0.446)** -0.372
(0.083)*** 0.141
(0.066)** CHNtIMPG 3ln �� 1.022
(0.529)** -0.469
(0.099)*** 0.091 (0.078)
CHNtIMPG 4ln �� 0.776
(0.657) -0.455
(0.122)*** 0.110 (0.097)
CHNtIMPG 5ln ��
0.420 (0.557)
-0.261 (0.104)**
0.064 (0.082)
CHNtIMPG 6ln �� 0.212
(0.257) -0.121
(0.048)** 0.038 (0.038)
rtE 1ln �� -0.784
(1.730) 0.698
(0.322)** -0.833
(0.255)*** rtE 2ln �� -1.114
(1.657) -0.145 (0.309)
-0.392 (0.244)
rtE 3ln �� -2.958
(1.658)* 0.824
(0.309)** -0.644
(0.244)** rtE 4ln �� -0.497
(1.598) 0.645
(0.298)** -0.371 (0.235)
rtE 5ln �� -4.142
(1.661)** 0.619
(0.310)** -0.663
(0.245)** rtE 6ln �� -1.074
(1.563) 0.678
(0.291)** -0.159 (0.230)
1ln �� tY -1.605 (1.479)
-0.511 (0.276)**
-0.915 (0.219)***
2ln �� tY -5.478 (2.044)**
0.651 (0.381)*
-0.873 (0.301)***
3ln �� tY -1.886 (1.701)
0.493 (0.317)
-0.634 (0.250)**
4ln �� tY -0.807 (1.810)
0.019 (0.337)
-0.997 (0.267)***
5ln �� tY -2.294 (2.211)
-0.431 (0.412)
-0.665 (0.326)**
6ln �� tY -4.087 (1.958)**
0.060 (0.365)
-0.384 (0.288)
C 0.292 (0.222)
0.085 (0.041)**
0.109 (0.033)
307
COUP -0.052(0.117)
-0.008(0.022)
-0.041(0.017)**
Diagnostics2R 0.865 0.888 0.838
Adjusted 2R 0.595 0.663 0.514! 0.192 0.036 0.028
NX 2 0.468[0.792]
HetX 2 30.602[0.288]
LM )(SCTest 16.649[0.055]
AR roots graph stable
308
Table SA59: Short–run coefficient estimates of Fiji’s goods import model with Malaysia, 1975–2012
Short-run results MALAIMPGln� rEln� Yln�
1�tECT -0.756 (0.222)***
0.052 (0.072)
-0.119 (0.046)**
MALAtIMPG 1ln �� 0.006
(0.257) 0.086 (0.083)
0.083 (0.053)
MALAtIMPG 2ln �� -0.058
(0.268) 0.048 (0.087)
0.119 (0.056)**
MALAtIMPG 3ln �� 0.351
(0.300) 0.164 (0.097)
0.055 (0.062)
MALAtIMPG 4ln ��
0.505 (0.305)
0.129 (0.099)
0.080 (0.063)
MALAtIMPG 5ln ��
0.200 (0.258)
0.122 (0.084)
0.033 (0.054)
MALAtIMPG 6ln ��
-0.141 (0.213)
0.004 (0.069)
-0.003 (0.044)
rtE 1ln �� -4.796
(1.899)** 0.282 (0.617)
-0.894 (0.394)
rtE 2ln �� -3.549
(1.681)** -0.203 (0.546)
-0.519 (0.349)
rtE 3ln �� -3.751
(1.277)*** -0.074 (0.415)
-0.606 (0.265)**
rtE 4ln �� -2.526
(1.152)** 0.279 (0.374)
-0.366 (0.239)**
rtE 5ln �� -4.243
(1.276)** -0.307 (0.415)
-0.375 (0.265)
rtE 6ln �� -2.859
(1.220)** -0.165 (0.396)
-0.289 (0.253)
1ln �� tY -0.867 (1.490)
-0.575 (0.484)
-0.711 (0.309)
2ln �� tY -2.191 (1.953)
-0.110 (0.635)
-0.705 (0.405)
3ln �� tY -4.533 (1.600)***
-0.437 (0.520)
-0.480 (0.332)
4ln �� tY -3.018 (1.513)*
0.072 (0.492)
-0.574 (0.314)*
5ln �� tY -3.401 (1.394)**
-0.174 (0.453)
-0.086 (0.289)
6ln �� tY -3.480 (1.320)**
0.083 (0.428)
-0.122 (0.274)
C 0.649 (0.140)***
-0.027 (0.046)
0.066 (0.029)**
COUP -0.088 (0.132)
0.051 (0.043)
0.002 (0.027)
309
Diagnostics2R 0.818 0.707 0.724
Adjusted 2R 0.455 0.120 0.171! 0.178 0.058 0.037
NX 2 0.825[0.662]
HetX 2 24.964[0.577]
LM )(SCTest 5.656[0.774]
AR roots graph stable
310
Table SA60: Short–run coefficient estimates of Fiji’s goods import model with India, 1975–2012
Short-run results INDIMPGln� rEln� Yln�
1�tECT -0.786 (0.363)**
-0.127 (0.095)
0.023 (0.065)
INDtIMPG 1ln �� 0.179
(0.343) 0.106 (0.090)
0.025 (0.061)
INDtIMPG 2ln �� 0.676
(0.331)** 0.020 (0.086)
0.008 (0.059)
INDtIMPG 3ln �� 0.190
(0.299) 0.017 (0.078)
-0.004 (0.053)
INDtIMPG 4ln ��
0.345 (0.285)
-0.037 (0.075)
0.020 (0.051)
rtE 1ln �� 0.084
(1.016) 0.378 (0.266)
0.079 (0.181)
rtE 2ln �� 3.449
(1.162)*** -0.169 (0.304)
0.216 (0.207)
rtE 3ln �� 0.521
(1.210) -0.085 (0.316)
-0.070 (0.215)
rtE 4ln �� -0.326
(1.080) 0.133 (0.282)
-0.006 (0.192)
1ln �� tY -0.965 (1.334)
-0.509 (0.349)
-0.390 (0.238)
2ln �� tY -0.099 (1.494)
-0.005 (0.391)
0.110 (0.266)
3ln �� tY 1.603 (1.383)
-0.183 (0.362)
0.068 (0.246)
4ln �� tY 0.909 (1.258)
-0.149 (0.329)
-0.177 (0.224)
C -0.081 (0.080)
0.019 (0.021)
0.025 (0.014)
COUP -0.175 (0.152)
0.020 (0.040)
-0.038 (0.027)
Diagnostics 2R 0.529 0.447 0.400
Adjusted 2R 0.163 0.017 -0.067 ! 0.235 0.061 0.042
NX 2
1.393 [0.498]
HetX 2
30.208 [0.305]
LM )(SCTest 9.405 [0.401]
AR roots graph stable
311
Table SA61: Short–run coefficient estimates of Fiji’s goods import model with Hong Kong, 1975–2012
Short-run results HKIMPGln� rEln� Yln�
1�tECT -1.637 (0.645)**
0.328 (0.164)*
-0.072 (0.145)
HKtIMPG 1ln �� -0.049
(0.572) -0.237 (0.145)
0.030 (0.128)
HKtIMPG 2ln �� -0.321
(0.445) -0.213 (0.113)*
0.007 (0.010)
HKtIMPG 3ln �� -0.385
(0.283) -0.114 (0.072)
-0.011 (0.064)
HKtIMPG 4ln ��
-0.174 (0.168)
-0.100 (0.043)**
-0.006 (0.038)
rtE 1ln �� -4.962
(2.621)* 1.095 (0.665)
-0.297 (0.589)
rtE 2ln �� -2.252
(2.343) 0.485 (0.594)
-0.004 (0.526)
rtE 3ln �� 0.511
(1.857) 0.622 (0.471)
0.031 (0.417)
rtE 4ln �� 0.471
(1.188) 0.438 (0.301)
0.168 (0.267)
1ln �� tY -5.412 (1.921)***
0.434 (0.301)
0.168 (0.267)
2ln �� tY -5.130 (2.206)**
0.696 (0.560)
-0.855 (0.495)*
3ln �� tY -4.703 (1.999)**
0.452 (0.507)
-0.497 (0.449)
4ln �� tY -2.322 (1.197)*
0.446 (0.304)
-0.352 (0.269)
C -0.061 (0.046)
-0.005 (0.012)
-0.001 (0.010)
COUP 0.445 (0.187)**
0.077 (0.047)
-0.037 (0.042)
Diagnostics 2R 0.875 0.656 0.700
Adjusted 2R 0.771 0.372 0.453 ! 0.226 0.057 0.051
NX 2
0.096 [0.953]
HetX 2
30.021 [0.313]
LM )(SCTest 10.539 [0.309]
AR roots graph stable